diff --git "a/tmp-checkpoint-141142/trainer_state.json" "b/tmp-checkpoint-141142/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/tmp-checkpoint-141142/trainer_state.json"
@@ -0,0 +1,85045 @@
+{
+  "best_metric": 0.17507466673851013,
+  "best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT\\checkpoint-116235",
+  "epoch": 16.99993977717555,
+  "eval_steps": 500,
+  "global_step": 141142,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2045290291496025e-08,
+      "loss": 1.618,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.409058058299205e-08,
+      "loss": 1.643,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6135870874488076e-08,
+      "loss": 1.6249,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.81811611659841e-08,
+      "loss": 1.6316,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.022645145748012e-08,
+      "loss": 1.6223,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.227174174897615e-08,
+      "loss": 1.6194,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.431703204047217e-08,
+      "loss": 1.6223,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.63623223319682e-08,
+      "loss": 1.6369,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0840761262346423e-07,
+      "loss": 1.6251,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2045290291496024e-07,
+      "loss": 1.6344,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3249819320645627e-07,
+      "loss": 1.6342,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.445434834979523e-07,
+      "loss": 1.6207,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5658877378944834e-07,
+      "loss": 1.6193,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6863406408094434e-07,
+      "loss": 1.6204,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8067935437244037e-07,
+      "loss": 1.613,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.927246446639364e-07,
+      "loss": 1.612,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0476993495543244e-07,
+      "loss": 1.6379,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1681522524692847e-07,
+      "loss": 1.6225,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.288605155384245e-07,
+      "loss": 1.6313,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.409058058299205e-07,
+      "loss": 1.6287,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.529510961214165e-07,
+      "loss": 1.6071,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6499638641291254e-07,
+      "loss": 1.6256,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.770416767044086e-07,
+      "loss": 1.6243,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.890869669959046e-07,
+      "loss": 1.6133,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0113225728740064e-07,
+      "loss": 1.6262,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1317754757889667e-07,
+      "loss": 1.6096,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.252228378703927e-07,
+      "loss": 1.6154,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.372681281618887e-07,
+      "loss": 1.628,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.493134184533847e-07,
+      "loss": 1.6236,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6135870874488075e-07,
+      "loss": 1.5968,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.734039990363768e-07,
+      "loss": 1.6199,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.854492893278728e-07,
+      "loss": 1.6218,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.974945796193688e-07,
+      "loss": 1.6117,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.095398699108649e-07,
+      "loss": 1.6132,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2158516020236085e-07,
+      "loss": 1.624,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3363045049385694e-07,
+      "loss": 1.609,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.456757407853529e-07,
+      "loss": 1.614,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.57721031076849e-07,
+      "loss": 1.5852,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.69766321368345e-07,
+      "loss": 1.6097,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.81811611659841e-07,
+      "loss": 1.5951,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.938569019513371e-07,
+      "loss": 1.6238,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.05902192242833e-07,
+      "loss": 1.6078,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.179474825343292e-07,
+      "loss": 1.6084,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.299927728258251e-07,
+      "loss": 1.6042,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.420380631173212e-07,
+      "loss": 1.5997,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.540833534088172e-07,
+      "loss": 1.6163,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.661286437003132e-07,
+      "loss": 1.6056,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.781739339918092e-07,
+      "loss": 1.621,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.902192242833052e-07,
+      "loss": 1.6047,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.022645145748013e-07,
+      "loss": 1.6021,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.143098048662973e-07,
+      "loss": 1.6011,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.263550951577933e-07,
+      "loss": 1.6059,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.384003854492894e-07,
+      "loss": 1.6037,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.504456757407854e-07,
+      "loss": 1.6248,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.624909660322814e-07,
+      "loss": 1.6103,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.745362563237774e-07,
+      "loss": 1.6127,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.865815466152735e-07,
+      "loss": 1.602,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.986268369067694e-07,
+      "loss": 1.5854,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.106721271982656e-07,
+      "loss": 1.5853,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.227174174897615e-07,
+      "loss": 1.6071,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.347627077812576e-07,
+      "loss": 1.5947,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.468079980727536e-07,
+      "loss": 1.5873,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.588532883642496e-07,
+      "loss": 1.5872,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.708985786557456e-07,
+      "loss": 1.5932,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.829438689472417e-07,
+      "loss": 1.605,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.949891592387376e-07,
+      "loss": 1.5767,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.070344495302338e-07,
+      "loss": 1.5965,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.190797398217297e-07,
+      "loss": 1.5714,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.311250301132258e-07,
+      "loss": 1.5772,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.431703204047217e-07,
+      "loss": 1.5935,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.552156106962177e-07,
+      "loss": 1.5768,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.672609009877139e-07,
+      "loss": 1.5776,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.793061912792099e-07,
+      "loss": 1.5785,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.913514815707058e-07,
+      "loss": 1.5848,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.033967718622019e-07,
+      "loss": 1.5893,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.15442062153698e-07,
+      "loss": 1.5738,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.27487352445194e-07,
+      "loss": 1.5785,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.3953264273669e-07,
+      "loss": 1.5555,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.51577933028186e-07,
+      "loss": 1.566,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.63623223319682e-07,
+      "loss": 1.5961,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756685136111782e-07,
+      "loss": 1.5703,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877138039026742e-07,
+      "loss": 1.589,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9975909419417e-07,
+      "loss": 1.5582,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.011804384485666e-06,
+      "loss": 1.5772,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0238496747771623e-06,
+      "loss": 1.5611,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0358949650686583e-06,
+      "loss": 1.5549,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0479402553601541e-06,
+      "loss": 1.5564,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0599855456516502e-06,
+      "loss": 1.5875,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0720308359431462e-06,
+      "loss": 1.5517,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0840761262346425e-06,
+      "loss": 1.5553,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0961214165261383e-06,
+      "loss": 1.5471,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1081667068176343e-06,
+      "loss": 1.5581,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1202119971091303e-06,
+      "loss": 1.5512,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1322572874006264e-06,
+      "loss": 1.5387,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1443025776921224e-06,
+      "loss": 1.5541,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1563478679836184e-06,
+      "loss": 1.5618,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1683931582751145e-06,
+      "loss": 1.5539,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1804384485666105e-06,
+      "loss": 1.5651,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1924837388581065e-06,
+      "loss": 1.5555,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2045290291496026e-06,
+      "loss": 1.5345,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2165743194410986e-06,
+      "loss": 1.5386,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2286196097325946e-06,
+      "loss": 1.5398,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2406649000240907e-06,
+      "loss": 1.5354,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2527101903155867e-06,
+      "loss": 1.5375,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2647554806070827e-06,
+      "loss": 1.5325,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2768007708985788e-06,
+      "loss": 1.5159,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2888460611900748e-06,
+      "loss": 1.5329,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3008913514815708e-06,
+      "loss": 1.5263,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3129366417730668e-06,
+      "loss": 1.5456,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3249819320645629e-06,
+      "loss": 1.5279,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.337027222356059e-06,
+      "loss": 1.5281,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3490725126475547e-06,
+      "loss": 1.5259,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.361117802939051e-06,
+      "loss": 1.509,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.373163093230547e-06,
+      "loss": 1.5127,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.385208383522043e-06,
+      "loss": 1.5164,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3972536738135389e-06,
+      "loss": 1.5139,
+      "step": 1160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.409298964105035e-06,
+      "loss": 1.5099,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4213442543965311e-06,
+      "loss": 1.5127,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.433389544688027e-06,
+      "loss": 1.5041,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.445434834979523e-06,
+      "loss": 1.5022,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.457480125271019e-06,
+      "loss": 1.4873,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4695254155625153e-06,
+      "loss": 1.4851,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.481570705854011e-06,
+      "loss": 1.4827,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4936159961455071e-06,
+      "loss": 1.5158,
+      "step": 1240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5056612864370031e-06,
+      "loss": 1.4789,
+      "step": 1250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5177065767284992e-06,
+      "loss": 1.459,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5297518670199952e-06,
+      "loss": 1.4856,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5417971573114912e-06,
+      "loss": 1.4825,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5538424476029873e-06,
+      "loss": 1.4407,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5658877378944833e-06,
+      "loss": 1.4795,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5779330281859793e-06,
+      "loss": 1.4671,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5899783184774752e-06,
+      "loss": 1.4437,
+      "step": 1320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6020236087689714e-06,
+      "loss": 1.451,
+      "step": 1330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6140688990604676e-06,
+      "loss": 1.489,
+      "step": 1340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6261141893519635e-06,
+      "loss": 1.4714,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6381594796434595e-06,
+      "loss": 1.4717,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6502047699349553e-06,
+      "loss": 1.4306,
+      "step": 1370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6622500602264516e-06,
+      "loss": 1.4715,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6742953505179476e-06,
+      "loss": 1.4339,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6863406408094434e-06,
+      "loss": 1.4685,
+      "step": 1400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6983859311009397e-06,
+      "loss": 1.414,
+      "step": 1410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7104312213924355e-06,
+      "loss": 1.4217,
+      "step": 1420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7224765116839317e-06,
+      "loss": 1.4231,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7345218019754278e-06,
+      "loss": 1.42,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7465670922669236e-06,
+      "loss": 1.3899,
+      "step": 1450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7586123825584198e-06,
+      "loss": 1.3898,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7706576728499159e-06,
+      "loss": 1.3781,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7827029631414117e-06,
+      "loss": 1.4362,
+      "step": 1480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.794748253432908e-06,
+      "loss": 1.4133,
+      "step": 1490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8067935437244037e-06,
+      "loss": 1.3684,
+      "step": 1500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8188388340159e-06,
+      "loss": 1.4252,
+      "step": 1510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.830884124307396e-06,
+      "loss": 1.4306,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8429294145988918e-06,
+      "loss": 1.3647,
+      "step": 1530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.854974704890388e-06,
+      "loss": 1.3758,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867019995181884e-06,
+      "loss": 1.3738,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87906528547338e-06,
+      "loss": 1.3461,
+      "step": 1560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8911105757648762e-06,
+      "loss": 1.3683,
+      "step": 1570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.903155866056372e-06,
+      "loss": 1.3433,
+      "step": 1580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9152011563478682e-06,
+      "loss": 1.3271,
+      "step": 1590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.927246446639364e-06,
+      "loss": 1.3713,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9392917369308603e-06,
+      "loss": 1.3267,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9513370272223563e-06,
+      "loss": 1.3437,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.963382317513852e-06,
+      "loss": 1.3406,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9754276078053484e-06,
+      "loss": 1.3392,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.987472898096844e-06,
+      "loss": 1.3235,
+      "step": 1650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99951818838834e-06,
+      "loss": 1.3319,
+      "step": 1660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0115634786798365e-06,
+      "loss": 1.3191,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.023608768971332e-06,
+      "loss": 1.3068,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0356540592628286e-06,
+      "loss": 1.3093,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0476993495543246e-06,
+      "loss": 1.2408,
+      "step": 1700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.05974463984582e-06,
+      "loss": 1.264,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0717899301373166e-06,
+      "loss": 1.2897,
+      "step": 1720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0838352204288123e-06,
+      "loss": 1.2984,
+      "step": 1730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0958805107203083e-06,
+      "loss": 1.2369,
+      "step": 1740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1079258010118047e-06,
+      "loss": 1.208,
+      "step": 1750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1199710913033004e-06,
+      "loss": 1.3214,
+      "step": 1760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.132016381594797e-06,
+      "loss": 1.2363,
+      "step": 1770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1440616718862924e-06,
+      "loss": 1.2309,
+      "step": 1780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1561069621777884e-06,
+      "loss": 1.2572,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.168152252469285e-06,
+      "loss": 1.2339,
+      "step": 1800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1801975427607805e-06,
+      "loss": 1.2175,
+      "step": 1810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1922428330522765e-06,
+      "loss": 1.2457,
+      "step": 1820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2042881233437726e-06,
+      "loss": 1.2167,
+      "step": 1830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2163334136352686e-06,
+      "loss": 1.1986,
+      "step": 1840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.228378703926765e-06,
+      "loss": 1.2868,
+      "step": 1850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2404239942182607e-06,
+      "loss": 1.2269,
+      "step": 1860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2524692845097567e-06,
+      "loss": 1.1896,
+      "step": 1870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2645145748012527e-06,
+      "loss": 1.1821,
+      "step": 1880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2765598650927488e-06,
+      "loss": 1.1892,
+      "step": 1890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.288605155384245e-06,
+      "loss": 1.1755,
+      "step": 1900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.300650445675741e-06,
+      "loss": 1.1762,
+      "step": 1910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.312695735967237e-06,
+      "loss": 1.2152,
+      "step": 1920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.324741026258733e-06,
+      "loss": 1.1999,
+      "step": 1930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.336786316550229e-06,
+      "loss": 1.2199,
+      "step": 1940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.348831606841725e-06,
+      "loss": 1.1848,
+      "step": 1950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.360876897133221e-06,
+      "loss": 1.151,
+      "step": 1960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.372922187424717e-06,
+      "loss": 1.2209,
+      "step": 1970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.384967477716213e-06,
+      "loss": 1.088,
+      "step": 1980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.397012768007709e-06,
+      "loss": 1.1774,
+      "step": 1990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.409058058299205e-06,
+      "loss": 1.1574,
+      "step": 2000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.421103348590701e-06,
+      "loss": 1.1444,
+      "step": 2010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.433148638882197e-06,
+      "loss": 1.1154,
+      "step": 2020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4451939291736932e-06,
+      "loss": 1.0801,
+      "step": 2030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4572392194651892e-06,
+      "loss": 1.1164,
+      "step": 2040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4692845097566853e-06,
+      "loss": 1.0906,
+      "step": 2050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4813298000481813e-06,
+      "loss": 1.0527,
+      "step": 2060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4933750903396773e-06,
+      "loss": 1.0492,
+      "step": 2070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5054203806311734e-06,
+      "loss": 1.2051,
+      "step": 2080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5174656709226694e-06,
+      "loss": 1.0355,
+      "step": 2090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5295109612141654e-06,
+      "loss": 1.1023,
+      "step": 2100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5415562515056615e-06,
+      "loss": 1.0717,
+      "step": 2110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5536015417971575e-06,
+      "loss": 1.0252,
+      "step": 2120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5656468320886535e-06,
+      "loss": 1.0674,
+      "step": 2130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5776921223801496e-06,
+      "loss": 1.0206,
+      "step": 2140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5897374126716456e-06,
+      "loss": 1.1487,
+      "step": 2150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6017827029631416e-06,
+      "loss": 1.032,
+      "step": 2160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6138279932546377e-06,
+      "loss": 1.1007,
+      "step": 2170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6258732835461337e-06,
+      "loss": 1.103,
+      "step": 2180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6379185738376293e-06,
+      "loss": 1.1403,
+      "step": 2190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6499638641291258e-06,
+      "loss": 1.0392,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.662009154420622e-06,
+      "loss": 1.013,
+      "step": 2210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.674054444712118e-06,
+      "loss": 1.0057,
+      "step": 2220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.686099735003614e-06,
+      "loss": 1.0906,
+      "step": 2230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6981450252951095e-06,
+      "loss": 0.9256,
+      "step": 2240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.710190315586606e-06,
+      "loss": 1.0398,
+      "step": 2250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.722235605878102e-06,
+      "loss": 0.99,
+      "step": 2260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7342808961695976e-06,
+      "loss": 0.9632,
+      "step": 2270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.746326186461094e-06,
+      "loss": 0.8818,
+      "step": 2280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7583714767525896e-06,
+      "loss": 0.8905,
+      "step": 2290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.770416767044086e-06,
+      "loss": 0.9031,
+      "step": 2300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.782462057335582e-06,
+      "loss": 1.0191,
+      "step": 2310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7945073476270777e-06,
+      "loss": 0.9683,
+      "step": 2320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.806552637918574e-06,
+      "loss": 0.9268,
+      "step": 2330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.81859792821007e-06,
+      "loss": 0.8958,
+      "step": 2340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.830643218501566e-06,
+      "loss": 0.9567,
+      "step": 2350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8426885087930623e-06,
+      "loss": 0.9084,
+      "step": 2360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.854733799084558e-06,
+      "loss": 0.9237,
+      "step": 2370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.866779089376054e-06,
+      "loss": 0.8828,
+      "step": 2380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8788243796675504e-06,
+      "loss": 0.9673,
+      "step": 2390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.890869669959046e-06,
+      "loss": 0.9653,
+      "step": 2400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9029149602505424e-06,
+      "loss": 0.7813,
+      "step": 2410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.914960250542038e-06,
+      "loss": 0.9837,
+      "step": 2420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.927005540833534e-06,
+      "loss": 0.883,
+      "step": 2430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9390508311250305e-06,
+      "loss": 0.8632,
+      "step": 2440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.951096121416526e-06,
+      "loss": 1.018,
+      "step": 2450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.963141411708022e-06,
+      "loss": 0.8925,
+      "step": 2460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.975186701999518e-06,
+      "loss": 0.9369,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9872319922910142e-06,
+      "loss": 0.8835,
+      "step": 2480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9992772825825107e-06,
+      "loss": 0.7873,
+      "step": 2490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0113225728740063e-06,
+      "loss": 0.8504,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0233678631655023e-06,
+      "loss": 0.8252,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0354131534569984e-06,
+      "loss": 0.8785,
+      "step": 2520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0474584437484944e-06,
+      "loss": 0.8333,
+      "step": 2530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.0595037340399904e-06,
+      "loss": 0.8386,
+      "step": 2540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.0715490243314865e-06,
+      "loss": 1.0213,
+      "step": 2550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.0835943146229825e-06,
+      "loss": 1.0045,
+      "step": 2560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.095639604914479e-06,
+      "loss": 0.8696,
+      "step": 2570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1076848952059746e-06,
+      "loss": 0.8633,
+      "step": 2580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1197301854974706e-06,
+      "loss": 0.7537,
+      "step": 2590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1317754757889666e-06,
+      "loss": 0.9511,
+      "step": 2600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1438207660804622e-06,
+      "loss": 0.8493,
+      "step": 2610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1558660563719587e-06,
+      "loss": 0.8316,
+      "step": 2620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1679113466634547e-06,
+      "loss": 0.9185,
+      "step": 2630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1799566369549503e-06,
+      "loss": 0.825,
+      "step": 2640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.192001927246447e-06,
+      "loss": 0.8185,
+      "step": 2650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.204047217537943e-06,
+      "loss": 0.8666,
+      "step": 2660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.216092507829439e-06,
+      "loss": 0.6777,
+      "step": 2670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2281377981209353e-06,
+      "loss": 0.8687,
+      "step": 2680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.240183088412431e-06,
+      "loss": 0.7363,
+      "step": 2690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.252228378703927e-06,
+      "loss": 0.738,
+      "step": 2700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.2642736689954234e-06,
+      "loss": 0.8507,
+      "step": 2710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.276318959286919e-06,
+      "loss": 0.8058,
+      "step": 2720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.288364249578415e-06,
+      "loss": 0.8257,
+      "step": 2730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3004095398699106e-06,
+      "loss": 0.7094,
+      "step": 2740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.312454830161407e-06,
+      "loss": 0.8156,
+      "step": 2750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.324500120452903e-06,
+      "loss": 0.807,
+      "step": 2760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3365454107443987e-06,
+      "loss": 0.924,
+      "step": 2770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.348590701035895e-06,
+      "loss": 0.6514,
+      "step": 2780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3606359913273912e-06,
+      "loss": 0.8851,
+      "step": 2790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.372681281618887e-06,
+      "loss": 0.7191,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3847265719103837e-06,
+      "loss": 0.8087,
+      "step": 2810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3967718622018793e-06,
+      "loss": 0.7068,
+      "step": 2820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.408817152493375e-06,
+      "loss": 0.879,
+      "step": 2830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.420862442784871e-06,
+      "loss": 0.7523,
+      "step": 2840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4329077330763674e-06,
+      "loss": 0.6441,
+      "step": 2850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4449530233678634e-06,
+      "loss": 0.8603,
+      "step": 2860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.456998313659359e-06,
+      "loss": 0.7534,
+      "step": 2870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4690436039508555e-06,
+      "loss": 0.6736,
+      "step": 2880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4810888942423515e-06,
+      "loss": 0.6928,
+      "step": 2890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.493134184533847e-06,
+      "loss": 0.7565,
+      "step": 2900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5051794748253436e-06,
+      "loss": 0.7842,
+      "step": 2910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5172247651168396e-06,
+      "loss": 0.7485,
+      "step": 2920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5292700554083352e-06,
+      "loss": 0.6787,
+      "step": 2930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5413153456998317e-06,
+      "loss": 0.6819,
+      "step": 2940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5533606359913277e-06,
+      "loss": 0.8492,
+      "step": 2950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5654059262828233e-06,
+      "loss": 0.7842,
+      "step": 2960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5774512165743194e-06,
+      "loss": 0.7891,
+      "step": 2970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.589496506865816e-06,
+      "loss": 0.852,
+      "step": 2980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6015417971573114e-06,
+      "loss": 0.806,
+      "step": 2990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6135870874488075e-06,
+      "loss": 0.6639,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.625632377740304e-06,
+      "loss": 0.8274,
+      "step": 3010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6376776680318e-06,
+      "loss": 0.7737,
+      "step": 3020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6497229583232956e-06,
+      "loss": 0.7151,
+      "step": 3030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.661768248614792e-06,
+      "loss": 0.8096,
+      "step": 3040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.673813538906288e-06,
+      "loss": 0.6899,
+      "step": 3050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6858588291977837e-06,
+      "loss": 0.6569,
+      "step": 3060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6979041194892797e-06,
+      "loss": 0.7375,
+      "step": 3070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.709949409780776e-06,
+      "loss": 0.7564,
+      "step": 3080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7219947000722718e-06,
+      "loss": 0.7394,
+      "step": 3090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.734039990363768e-06,
+      "loss": 0.7201,
+      "step": 3100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7460852806552642e-06,
+      "loss": 0.7294,
+      "step": 3110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.75813057094676e-06,
+      "loss": 0.6986,
+      "step": 3120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.770175861238256e-06,
+      "loss": 0.6387,
+      "step": 3130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7822211515297523e-06,
+      "loss": 0.6523,
+      "step": 3140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.794266441821248e-06,
+      "loss": 0.7157,
+      "step": 3150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.806311732112744e-06,
+      "loss": 0.6255,
+      "step": 3160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8183570224042404e-06,
+      "loss": 0.6307,
+      "step": 3170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8304023126957365e-06,
+      "loss": 0.7539,
+      "step": 3180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8424476029872325e-06,
+      "loss": 0.761,
+      "step": 3190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.854492893278728e-06,
+      "loss": 0.7775,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8665381835702246e-06,
+      "loss": 0.6311,
+      "step": 3210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.878583473861721e-06,
+      "loss": 0.7344,
+      "step": 3220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.890628764153216e-06,
+      "loss": 0.6123,
+      "step": 3230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.902674054444713e-06,
+      "loss": 0.7842,
+      "step": 3240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.914719344736209e-06,
+      "loss": 0.6592,
+      "step": 3250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.926764635027704e-06,
+      "loss": 0.6351,
+      "step": 3260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.938809925319201e-06,
+      "loss": 0.6838,
+      "step": 3270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.950855215610697e-06,
+      "loss": 0.6787,
+      "step": 3280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.962900505902192e-06,
+      "loss": 0.6171,
+      "step": 3290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.974945796193688e-06,
+      "loss": 0.6119,
+      "step": 3300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986991086485185e-06,
+      "loss": 0.6064,
+      "step": 3310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.99903637677668e-06,
+      "loss": 0.8103,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.011081667068176e-06,
+      "loss": 0.7416,
+      "step": 3330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.023126957359673e-06,
+      "loss": 0.6061,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.035172247651168e-06,
+      "loss": 0.6428,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.047217537942664e-06,
+      "loss": 0.611,
+      "step": 3360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.059262828234161e-06,
+      "loss": 0.6872,
+      "step": 3370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.071308118525657e-06,
+      "loss": 0.7002,
+      "step": 3380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.083353408817152e-06,
+      "loss": 0.5585,
+      "step": 3390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.095398699108649e-06,
+      "loss": 0.7227,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.107443989400145e-06,
+      "loss": 0.7608,
+      "step": 3410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.11948927969164e-06,
+      "loss": 0.7097,
+      "step": 3420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.131534569983136e-06,
+      "loss": 0.5039,
+      "step": 3430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.143579860274633e-06,
+      "loss": 0.8008,
+      "step": 3440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1556251505661285e-06,
+      "loss": 0.8033,
+      "step": 3450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1676704408576245e-06,
+      "loss": 0.6777,
+      "step": 3460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.179715731149121e-06,
+      "loss": 0.5464,
+      "step": 3470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.191761021440617e-06,
+      "loss": 0.602,
+      "step": 3480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.203806311732113e-06,
+      "loss": 0.6271,
+      "step": 3490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2158516020236095e-06,
+      "loss": 0.5394,
+      "step": 3500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.227896892315105e-06,
+      "loss": 0.5083,
+      "step": 3510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.239942182606601e-06,
+      "loss": 0.7406,
+      "step": 3520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.251987472898097e-06,
+      "loss": 0.5001,
+      "step": 3530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.264032763189594e-06,
+      "loss": 0.581,
+      "step": 3540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.276078053481089e-06,
+      "loss": 0.5438,
+      "step": 3550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288123343772585e-06,
+      "loss": 0.6164,
+      "step": 3560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.300168634064082e-06,
+      "loss": 0.5689,
+      "step": 3570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.312213924355577e-06,
+      "loss": 0.6068,
+      "step": 3580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.324259214647073e-06,
+      "loss": 0.78,
+      "step": 3590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.33630450493857e-06,
+      "loss": 0.5941,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.348349795230065e-06,
+      "loss": 0.5974,
+      "step": 3610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.360395085521561e-06,
+      "loss": 0.6201,
+      "step": 3620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.372440375813058e-06,
+      "loss": 0.6388,
+      "step": 3630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.384485666104553e-06,
+      "loss": 0.6887,
+      "step": 3640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.396530956396049e-06,
+      "loss": 0.4635,
+      "step": 3650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.408576246687545e-06,
+      "loss": 0.5648,
+      "step": 3660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.420621536979041e-06,
+      "loss": 0.5182,
+      "step": 3670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.432666827270537e-06,
+      "loss": 0.5923,
+      "step": 3680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.444712117562033e-06,
+      "loss": 0.7088,
+      "step": 3690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.45675740785353e-06,
+      "loss": 0.4766,
+      "step": 3700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.468802698145025e-06,
+      "loss": 0.6942,
+      "step": 3710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.480847988436521e-06,
+      "loss": 0.5993,
+      "step": 3720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.492893278728018e-06,
+      "loss": 0.6405,
+      "step": 3730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.504938569019513e-06,
+      "loss": 0.5261,
+      "step": 3740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5169838593110094e-06,
+      "loss": 0.7343,
+      "step": 3750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5290291496025055e-06,
+      "loss": 0.5309,
+      "step": 3760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5410744398940015e-06,
+      "loss": 0.8686,
+      "step": 3770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5531197301854975e-06,
+      "loss": 0.4225,
+      "step": 3780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5651650204769936e-06,
+      "loss": 0.6005,
+      "step": 3790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.57721031076849e-06,
+      "loss": 0.5354,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.589255601059986e-06,
+      "loss": 0.6548,
+      "step": 3810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.601300891351482e-06,
+      "loss": 0.6499,
+      "step": 3820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.613346181642978e-06,
+      "loss": 0.7827,
+      "step": 3830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.625391471934474e-06,
+      "loss": 0.4766,
+      "step": 3840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.63743676222597e-06,
+      "loss": 0.485,
+      "step": 3850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.649482052517466e-06,
+      "loss": 0.6619,
+      "step": 3860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.661527342808962e-06,
+      "loss": 0.5379,
+      "step": 3870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.673572633100458e-06,
+      "loss": 0.5998,
+      "step": 3880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.685617923391954e-06,
+      "loss": 0.6462,
+      "step": 3890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.69766321368345e-06,
+      "loss": 0.5092,
+      "step": 3900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.709708503974946e-06,
+      "loss": 0.5216,
+      "step": 3910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.721753794266442e-06,
+      "loss": 0.5176,
+      "step": 3920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.733799084557938e-06,
+      "loss": 0.5415,
+      "step": 3930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.745844374849434e-06,
+      "loss": 0.4709,
+      "step": 3940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.75788966514093e-06,
+      "loss": 0.5058,
+      "step": 3950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.769934955432426e-06,
+      "loss": 0.4911,
+      "step": 3960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.781980245723922e-06,
+      "loss": 0.4754,
+      "step": 3970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.794025536015418e-06,
+      "loss": 0.655,
+      "step": 3980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.806070826306914e-06,
+      "loss": 0.6335,
+      "step": 3990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.81811611659841e-06,
+      "loss": 0.7247,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.830161406889906e-06,
+      "loss": 0.589,
+      "step": 4010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.842206697181402e-06,
+      "loss": 0.6547,
+      "step": 4020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.854251987472898e-06,
+      "loss": 0.5324,
+      "step": 4030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.866297277764394e-06,
+      "loss": 0.5176,
+      "step": 4040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.87834256805589e-06,
+      "loss": 0.4329,
+      "step": 4050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8903878583473864e-06,
+      "loss": 0.6436,
+      "step": 4060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9024331486388825e-06,
+      "loss": 0.7,
+      "step": 4070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9144784389303785e-06,
+      "loss": 0.6959,
+      "step": 4080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9265237292218745e-06,
+      "loss": 0.5458,
+      "step": 4090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9385690195133706e-06,
+      "loss": 0.5097,
+      "step": 4100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.950614309804867e-06,
+      "loss": 0.5525,
+      "step": 4110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.962659600096363e-06,
+      "loss": 0.545,
+      "step": 4120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.974704890387859e-06,
+      "loss": 0.6471,
+      "step": 4130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.986750180679355e-06,
+      "loss": 0.7276,
+      "step": 4140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.998795470970851e-06,
+      "loss": 0.5004,
+      "step": 4150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.010840761262347e-06,
+      "loss": 0.6134,
+      "step": 4160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.022886051553843e-06,
+      "loss": 0.4791,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.034931341845339e-06,
+      "loss": 0.6029,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.046976632136835e-06,
+      "loss": 0.5072,
+      "step": 4190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.059021922428331e-06,
+      "loss": 0.6683,
+      "step": 4200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.071067212719827e-06,
+      "loss": 0.4545,
+      "step": 4210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.083112503011323e-06,
+      "loss": 0.4516,
+      "step": 4220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.095157793302819e-06,
+      "loss": 0.6589,
+      "step": 4230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.107203083594315e-06,
+      "loss": 0.5102,
+      "step": 4240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.11924837388581e-06,
+      "loss": 0.5088,
+      "step": 4250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.131293664177307e-06,
+      "loss": 0.481,
+      "step": 4260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.143338954468803e-06,
+      "loss": 0.3794,
+      "step": 4270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.155384244760299e-06,
+      "loss": 0.4362,
+      "step": 4280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.167429535051795e-06,
+      "loss": 0.5157,
+      "step": 4290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.179474825343291e-06,
+      "loss": 0.4668,
+      "step": 4300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.191520115634787e-06,
+      "loss": 0.4992,
+      "step": 4310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.203565405926283e-06,
+      "loss": 0.5478,
+      "step": 4320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.215610696217779e-06,
+      "loss": 0.5219,
+      "step": 4330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.227655986509275e-06,
+      "loss": 0.5858,
+      "step": 4340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2397012768007705e-06,
+      "loss": 0.6463,
+      "step": 4350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.251746567092267e-06,
+      "loss": 0.5957,
+      "step": 4360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2637918573837634e-06,
+      "loss": 0.4581,
+      "step": 4370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.275837147675259e-06,
+      "loss": 0.5995,
+      "step": 4380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2878824379667555e-06,
+      "loss": 0.4992,
+      "step": 4390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2999277282582515e-06,
+      "loss": 0.5782,
+      "step": 4400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.311973018549747e-06,
+      "loss": 0.4754,
+      "step": 4410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.324018308841244e-06,
+      "loss": 0.5505,
+      "step": 4420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.33606359913274e-06,
+      "loss": 0.5724,
+      "step": 4430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.348108889424236e-06,
+      "loss": 0.5318,
+      "step": 4440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.360154179715732e-06,
+      "loss": 0.6368,
+      "step": 4450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.372199470007228e-06,
+      "loss": 0.6284,
+      "step": 4460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.384244760298724e-06,
+      "loss": 0.5533,
+      "step": 4470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.396290050590219e-06,
+      "loss": 0.4426,
+      "step": 4480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.408335340881716e-06,
+      "loss": 0.4107,
+      "step": 4490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.420380631173212e-06,
+      "loss": 0.6605,
+      "step": 4500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.432425921464707e-06,
+      "loss": 0.4764,
+      "step": 4510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.444471211756204e-06,
+      "loss": 0.4938,
+      "step": 4520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.4565165020477e-06,
+      "loss": 0.5035,
+      "step": 4530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.468561792339195e-06,
+      "loss": 0.4689,
+      "step": 4540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.480607082630692e-06,
+      "loss": 0.5122,
+      "step": 4550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.492652372922188e-06,
+      "loss": 0.4704,
+      "step": 4560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.504697663213683e-06,
+      "loss": 0.5194,
+      "step": 4570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.516742953505179e-06,
+      "loss": 0.6027,
+      "step": 4580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.528788243796676e-06,
+      "loss": 0.4998,
+      "step": 4590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.540833534088172e-06,
+      "loss": 0.5504,
+      "step": 4600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.552878824379667e-06,
+      "loss": 0.6342,
+      "step": 4610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.564924114671164e-06,
+      "loss": 0.5564,
+      "step": 4620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.57696940496266e-06,
+      "loss": 0.4296,
+      "step": 4630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.5890146952541554e-06,
+      "loss": 0.4676,
+      "step": 4640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.601059985545652e-06,
+      "loss": 0.7282,
+      "step": 4650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.613105275837148e-06,
+      "loss": 0.4806,
+      "step": 4660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.6251505661286435e-06,
+      "loss": 0.7047,
+      "step": 4670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.63719585642014e-06,
+      "loss": 0.3619,
+      "step": 4680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.6492411467116364e-06,
+      "loss": 0.5778,
+      "step": 4690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.661286437003132e-06,
+      "loss": 0.6242,
+      "step": 4700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.673331727294628e-06,
+      "loss": 0.4773,
+      "step": 4710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.6853770175861245e-06,
+      "loss": 0.5888,
+      "step": 4720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.69742230787762e-06,
+      "loss": 0.4765,
+      "step": 4730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.709467598169116e-06,
+      "loss": 0.39,
+      "step": 4740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.721512888460613e-06,
+      "loss": 0.4885,
+      "step": 4750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.733558178752108e-06,
+      "loss": 0.538,
+      "step": 4760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.745603469043604e-06,
+      "loss": 0.4018,
+      "step": 4770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.757648759335101e-06,
+      "loss": 0.6024,
+      "step": 4780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.769694049626597e-06,
+      "loss": 0.5773,
+      "step": 4790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.781739339918092e-06,
+      "loss": 0.4602,
+      "step": 4800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.793784630209588e-06,
+      "loss": 0.6262,
+      "step": 4810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.805829920501085e-06,
+      "loss": 0.4825,
+      "step": 4820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.81787521079258e-06,
+      "loss": 0.5615,
+      "step": 4830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.829920501084076e-06,
+      "loss": 0.612,
+      "step": 4840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.841965791375573e-06,
+      "loss": 0.2971,
+      "step": 4850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.854011081667068e-06,
+      "loss": 0.5283,
+      "step": 4860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.866056371958564e-06,
+      "loss": 0.4795,
+      "step": 4870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.878101662250061e-06,
+      "loss": 0.7299,
+      "step": 4880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.890146952541556e-06,
+      "loss": 0.6781,
+      "step": 4890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.902192242833052e-06,
+      "loss": 0.4355,
+      "step": 4900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.914237533124549e-06,
+      "loss": 0.602,
+      "step": 4910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.926282823416044e-06,
+      "loss": 0.5103,
+      "step": 4920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.93832811370754e-06,
+      "loss": 0.3305,
+      "step": 4930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.950373403999036e-06,
+      "loss": 0.5849,
+      "step": 4940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.962418694290533e-06,
+      "loss": 0.4632,
+      "step": 4950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9744639845820285e-06,
+      "loss": 0.504,
+      "step": 4960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9865092748735245e-06,
+      "loss": 0.4671,
+      "step": 4970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.998554565165021e-06,
+      "loss": 0.5343,
+      "step": 4980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0105998554565166e-06,
+      "loss": 0.4728,
+      "step": 4990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.022645145748013e-06,
+      "loss": 0.4928,
+      "step": 5000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0346904360395095e-06,
+      "loss": 0.4513,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.046735726331005e-06,
+      "loss": 0.5332,
+      "step": 5020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.058781016622501e-06,
+      "loss": 0.5641,
+      "step": 5030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.070826306913997e-06,
+      "loss": 0.4369,
+      "step": 5040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.082871597205493e-06,
+      "loss": 0.3886,
+      "step": 5050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.094916887496989e-06,
+      "loss": 0.3751,
+      "step": 5060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.106962177788485e-06,
+      "loss": 0.519,
+      "step": 5070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.119007468079981e-06,
+      "loss": 0.459,
+      "step": 5080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.131052758371477e-06,
+      "loss": 0.4521,
+      "step": 5090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.143098048662973e-06,
+      "loss": 0.4481,
+      "step": 5100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.15514333895447e-06,
+      "loss": 0.4205,
+      "step": 5110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.167188629245965e-06,
+      "loss": 0.421,
+      "step": 5120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.179233919537461e-06,
+      "loss": 0.4428,
+      "step": 5130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.191279209828958e-06,
+      "loss": 0.5141,
+      "step": 5140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.203324500120453e-06,
+      "loss": 0.3254,
+      "step": 5150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.215369790411949e-06,
+      "loss": 0.4218,
+      "step": 5160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.227415080703445e-06,
+      "loss": 0.4599,
+      "step": 5170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.239460370994941e-06,
+      "loss": 0.4596,
+      "step": 5180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.251505661286437e-06,
+      "loss": 0.5694,
+      "step": 5190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263550951577933e-06,
+      "loss": 0.3899,
+      "step": 5200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275596241869429e-06,
+      "loss": 0.3831,
+      "step": 5210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2876415321609244e-06,
+      "loss": 0.379,
+      "step": 5220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.299686822452422e-06,
+      "loss": 0.4253,
+      "step": 5230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311732112743917e-06,
+      "loss": 0.4373,
+      "step": 5240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323777403035413e-06,
+      "loss": 0.4003,
+      "step": 5250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335822693326909e-06,
+      "loss": 0.6563,
+      "step": 5260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3478679836184055e-06,
+      "loss": 0.5389,
+      "step": 5270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.359913273909901e-06,
+      "loss": 0.3611,
+      "step": 5280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.371958564201398e-06,
+      "loss": 0.4825,
+      "step": 5290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.384003854492894e-06,
+      "loss": 0.3614,
+      "step": 5300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.39604914478439e-06,
+      "loss": 0.4205,
+      "step": 5310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.408094435075886e-06,
+      "loss": 0.5031,
+      "step": 5320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.420139725367382e-06,
+      "loss": 0.4996,
+      "step": 5330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.432185015658878e-06,
+      "loss": 0.5412,
+      "step": 5340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.444230305950373e-06,
+      "loss": 0.5283,
+      "step": 5350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.456275596241871e-06,
+      "loss": 0.5612,
+      "step": 5360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.468320886533366e-06,
+      "loss": 0.5204,
+      "step": 5370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.480366176824862e-06,
+      "loss": 0.6218,
+      "step": 5380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.492411467116358e-06,
+      "loss": 0.5053,
+      "step": 5390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.504456757407854e-06,
+      "loss": 0.5456,
+      "step": 5400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.516502047699349e-06,
+      "loss": 0.6857,
+      "step": 5410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.528547337990847e-06,
+      "loss": 0.6966,
+      "step": 5420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.540592628282342e-06,
+      "loss": 0.3692,
+      "step": 5430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.552637918573838e-06,
+      "loss": 0.5127,
+      "step": 5440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.564683208865334e-06,
+      "loss": 0.538,
+      "step": 5450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.57672849915683e-06,
+      "loss": 0.4563,
+      "step": 5460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.588773789448325e-06,
+      "loss": 0.3748,
+      "step": 5470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.600819079739821e-06,
+      "loss": 0.4877,
+      "step": 5480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.612864370031319e-06,
+      "loss": 0.6102,
+      "step": 5490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.624909660322814e-06,
+      "loss": 0.753,
+      "step": 5500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.63695495061431e-06,
+      "loss": 0.5277,
+      "step": 5510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.649000240905806e-06,
+      "loss": 0.5599,
+      "step": 5520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.661045531197302e-06,
+      "loss": 0.4621,
+      "step": 5530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.6730908214887975e-06,
+      "loss": 0.3551,
+      "step": 5540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.685136111780295e-06,
+      "loss": 0.467,
+      "step": 5550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.69718140207179e-06,
+      "loss": 0.3381,
+      "step": 5560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.709226692363286e-06,
+      "loss": 0.5466,
+      "step": 5570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.7212719826547824e-06,
+      "loss": 0.4579,
+      "step": 5580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.7333172729462785e-06,
+      "loss": 0.407,
+      "step": 5590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.745362563237774e-06,
+      "loss": 0.4037,
+      "step": 5600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.75740785352927e-06,
+      "loss": 0.5914,
+      "step": 5610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.769453143820767e-06,
+      "loss": 0.4636,
+      "step": 5620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.781498434112263e-06,
+      "loss": 0.4421,
+      "step": 5630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.793543724403759e-06,
+      "loss": 0.4942,
+      "step": 5640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.805589014695255e-06,
+      "loss": 0.6613,
+      "step": 5650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.81763430498675e-06,
+      "loss": 0.3269,
+      "step": 5660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.829679595278246e-06,
+      "loss": 0.4642,
+      "step": 5670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.841724885569742e-06,
+      "loss": 0.3119,
+      "step": 5680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.853770175861239e-06,
+      "loss": 0.4078,
+      "step": 5690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.865815466152735e-06,
+      "loss": 0.4629,
+      "step": 5700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.877860756444231e-06,
+      "loss": 0.4968,
+      "step": 5710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.889906046735727e-06,
+      "loss": 0.6665,
+      "step": 5720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.901951337027222e-06,
+      "loss": 0.3961,
+      "step": 5730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.913996627318718e-06,
+      "loss": 0.3748,
+      "step": 5740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.926041917610215e-06,
+      "loss": 0.4046,
+      "step": 5750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.938087207901711e-06,
+      "loss": 0.4587,
+      "step": 5760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.950132498193207e-06,
+      "loss": 0.4221,
+      "step": 5770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.962177788484703e-06,
+      "loss": 0.5593,
+      "step": 5780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.974223078776198e-06,
+      "loss": 0.4419,
+      "step": 5790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.986268369067694e-06,
+      "loss": 0.4444,
+      "step": 5800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.99831365935919e-06,
+      "loss": 0.4194,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.010358949650687e-06,
+      "loss": 0.4815,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.022404239942183e-06,
+      "loss": 0.4017,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.034449530233679e-06,
+      "loss": 0.5477,
+      "step": 5840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.046494820525175e-06,
+      "loss": 0.4697,
+      "step": 5850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.0585401108166705e-06,
+      "loss": 0.3987,
+      "step": 5860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.0705854011081665e-06,
+      "loss": 0.519,
+      "step": 5870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.082630691399663e-06,
+      "loss": 0.4937,
+      "step": 5880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.0946759816911594e-06,
+      "loss": 0.4815,
+      "step": 5890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.1067212719826555e-06,
+      "loss": 0.2911,
+      "step": 5900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.1187665622741515e-06,
+      "loss": 0.4578,
+      "step": 5910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.130811852565647e-06,
+      "loss": 0.5965,
+      "step": 5920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.4649,
+      "step": 5930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.154902433148639e-06,
+      "loss": 0.2398,
+      "step": 5940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.166947723440136e-06,
+      "loss": 0.3879,
+      "step": 5950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.178993013731632e-06,
+      "loss": 0.4985,
+      "step": 5960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.191038304023128e-06,
+      "loss": 0.5634,
+      "step": 5970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.203083594314623e-06,
+      "loss": 0.5043,
+      "step": 5980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.215128884606119e-06,
+      "loss": 0.4324,
+      "step": 5990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.227174174897615e-06,
+      "loss": 0.4749,
+      "step": 6000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.239219465189112e-06,
+      "loss": 0.5217,
+      "step": 6010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.251264755480608e-06,
+      "loss": 0.4153,
+      "step": 6020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.263310045772104e-06,
+      "loss": 0.4597,
+      "step": 6030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.2753553360636e-06,
+      "loss": 0.5156,
+      "step": 6040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.287400626355095e-06,
+      "loss": 0.3854,
+      "step": 6050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.299445916646591e-06,
+      "loss": 0.4904,
+      "step": 6060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.311491206938087e-06,
+      "loss": 0.2574,
+      "step": 6070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.323536497229584e-06,
+      "loss": 0.361,
+      "step": 6080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.33558178752108e-06,
+      "loss": 0.5998,
+      "step": 6090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.347627077812576e-06,
+      "loss": 0.444,
+      "step": 6100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.359672368104071e-06,
+      "loss": 0.4503,
+      "step": 6110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.371717658395567e-06,
+      "loss": 0.5508,
+      "step": 6120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.383762948687063e-06,
+      "loss": 0.4293,
+      "step": 6130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.395808238978559e-06,
+      "loss": 0.5193,
+      "step": 6140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.407853529270056e-06,
+      "loss": 0.6465,
+      "step": 6150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.419898819561552e-06,
+      "loss": 0.3654,
+      "step": 6160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.4319441098530475e-06,
+      "loss": 0.5274,
+      "step": 6170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.4439894001445435e-06,
+      "loss": 0.5087,
+      "step": 6180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.4560346904360395e-06,
+      "loss": 0.4028,
+      "step": 6190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.468079980727536e-06,
+      "loss": 0.5037,
+      "step": 6200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.4801252710190325e-06,
+      "loss": 0.5378,
+      "step": 6210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.4921705613105285e-06,
+      "loss": 0.4784,
+      "step": 6220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5042158516020245e-06,
+      "loss": 0.5758,
+      "step": 6230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.51626114189352e-06,
+      "loss": 0.4719,
+      "step": 6240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.528306432185016e-06,
+      "loss": 0.4406,
+      "step": 6250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.540351722476512e-06,
+      "loss": 0.5498,
+      "step": 6260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.552397012768008e-06,
+      "loss": 0.3951,
+      "step": 6270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.564442303059505e-06,
+      "loss": 0.4104,
+      "step": 6280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.576487593351001e-06,
+      "loss": 0.41,
+      "step": 6290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.588532883642496e-06,
+      "loss": 0.4746,
+      "step": 6300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.600578173933992e-06,
+      "loss": 0.4351,
+      "step": 6310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.612623464225488e-06,
+      "loss": 0.4973,
+      "step": 6320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.624668754516984e-06,
+      "loss": 0.5322,
+      "step": 6330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.636714044808481e-06,
+      "loss": 0.5684,
+      "step": 6340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.648759335099976e-06,
+      "loss": 0.4497,
+      "step": 6350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.660804625391473e-06,
+      "loss": 0.5994,
+      "step": 6360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.672849915682968e-06,
+      "loss": 0.4334,
+      "step": 6370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.684895205974465e-06,
+      "loss": 0.5416,
+      "step": 6380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.69694049626596e-06,
+      "loss": 0.3739,
+      "step": 6390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.708985786557455e-06,
+      "loss": 0.4078,
+      "step": 6400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.721031076848952e-06,
+      "loss": 0.4728,
+      "step": 6410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.733076367140449e-06,
+      "loss": 0.4738,
+      "step": 6420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.745121657431944e-06,
+      "loss": 0.3702,
+      "step": 6430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.757166947723441e-06,
+      "loss": 0.3881,
+      "step": 6440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.769212238014936e-06,
+      "loss": 0.3192,
+      "step": 6450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.781257528306432e-06,
+      "loss": 0.3915,
+      "step": 6460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.793302818597928e-06,
+      "loss": 0.4486,
+      "step": 6470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.805348108889425e-06,
+      "loss": 0.4052,
+      "step": 6480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.81739339918092e-06,
+      "loss": 0.5315,
+      "step": 6490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.829438689472417e-06,
+      "loss": 0.4918,
+      "step": 6500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.841483979763913e-06,
+      "loss": 0.4985,
+      "step": 6510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.853529270055408e-06,
+      "loss": 0.412,
+      "step": 6520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.865574560346905e-06,
+      "loss": 0.4001,
+      "step": 6530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.877619850638402e-06,
+      "loss": 0.382,
+      "step": 6540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.889665140929897e-06,
+      "loss": 0.3336,
+      "step": 6550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.901710431221394e-06,
+      "loss": 0.4339,
+      "step": 6560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.913755721512889e-06,
+      "loss": 0.5157,
+      "step": 6570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.925801011804384e-06,
+      "loss": 0.5192,
+      "step": 6580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.93784630209588e-06,
+      "loss": 0.366,
+      "step": 6590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.949891592387376e-06,
+      "loss": 0.3981,
+      "step": 6600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.961936882678873e-06,
+      "loss": 0.4308,
+      "step": 6610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.97398217297037e-06,
+      "loss": 0.5161,
+      "step": 6620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.986027463261865e-06,
+      "loss": 0.2851,
+      "step": 6630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.99807275355336e-06,
+      "loss": 0.4614,
+      "step": 6640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.010118043844857e-06,
+      "loss": 0.5586,
+      "step": 6650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.022163334136352e-06,
+      "loss": 0.338,
+      "step": 6660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.034208624427849e-06,
+      "loss": 0.3199,
+      "step": 6670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.046253914719346e-06,
+      "loss": 0.2405,
+      "step": 6680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.058299205010841e-06,
+      "loss": 0.3322,
+      "step": 6690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.070344495302336e-06,
+      "loss": 0.4648,
+      "step": 6700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.082389785593833e-06,
+      "loss": 0.5113,
+      "step": 6710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.094435075885328e-06,
+      "loss": 0.6286,
+      "step": 6720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.106480366176825e-06,
+      "loss": 0.4101,
+      "step": 6730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.118525656468322e-06,
+      "loss": 0.5602,
+      "step": 6740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.130570946759817e-06,
+      "loss": 0.3852,
+      "step": 6750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.142616237051314e-06,
+      "loss": 0.4589,
+      "step": 6760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.15466152734281e-06,
+      "loss": 0.6245,
+      "step": 6770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.166706817634305e-06,
+      "loss": 0.5513,
+      "step": 6780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.178752107925801e-06,
+      "loss": 0.3232,
+      "step": 6790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.190797398217298e-06,
+      "loss": 0.4243,
+      "step": 6800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.202842688508794e-06,
+      "loss": 0.3588,
+      "step": 6810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.21488797880029e-06,
+      "loss": 0.4692,
+      "step": 6820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.226933269091786e-06,
+      "loss": 0.3904,
+      "step": 6830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.23897855938328e-06,
+      "loss": 0.2821,
+      "step": 6840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.251023849674778e-06,
+      "loss": 0.4965,
+      "step": 6850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.263069139966273e-06,
+      "loss": 0.3623,
+      "step": 6860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.27511443025777e-06,
+      "loss": 0.408,
+      "step": 6870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.287159720549267e-06,
+      "loss": 0.2602,
+      "step": 6880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.299205010840762e-06,
+      "loss": 0.412,
+      "step": 6890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.311250301132257e-06,
+      "loss": 0.5776,
+      "step": 6900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.323295591423754e-06,
+      "loss": 0.355,
+      "step": 6910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.335340881715249e-06,
+      "loss": 0.4796,
+      "step": 6920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.347386172006746e-06,
+      "loss": 0.4038,
+      "step": 6930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.359431462298243e-06,
+      "loss": 0.3637,
+      "step": 6940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.371476752589738e-06,
+      "loss": 0.5986,
+      "step": 6950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.383522042881233e-06,
+      "loss": 0.5413,
+      "step": 6960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.39556733317273e-06,
+      "loss": 0.5352,
+      "step": 6970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.407612623464225e-06,
+      "loss": 0.4325,
+      "step": 6980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.419657913755722e-06,
+      "loss": 0.2381,
+      "step": 6990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.431703204047219e-06,
+      "loss": 0.3547,
+      "step": 7000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.443748494338714e-06,
+      "loss": 0.4185,
+      "step": 7010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.45579378463021e-06,
+      "loss": 0.5981,
+      "step": 7020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.467839074921706e-06,
+      "loss": 0.3505,
+      "step": 7030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.479884365213201e-06,
+      "loss": 0.3978,
+      "step": 7040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.491929655504698e-06,
+      "loss": 0.6432,
+      "step": 7050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.503974945796193e-06,
+      "loss": 0.5061,
+      "step": 7060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.51602023608769e-06,
+      "loss": 0.4733,
+      "step": 7070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.528065526379187e-06,
+      "loss": 0.4144,
+      "step": 7080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.540110816670682e-06,
+      "loss": 0.3947,
+      "step": 7090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.552156106962178e-06,
+      "loss": 0.4079,
+      "step": 7100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.564201397253674e-06,
+      "loss": 0.3967,
+      "step": 7110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.57624668754517e-06,
+      "loss": 0.5846,
+      "step": 7120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.588291977836667e-06,
+      "loss": 0.4633,
+      "step": 7130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.600337268128163e-06,
+      "loss": 0.4765,
+      "step": 7140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.612382558419659e-06,
+      "loss": 0.4172,
+      "step": 7150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.624427848711154e-06,
+      "loss": 0.3337,
+      "step": 7160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.63647313900265e-06,
+      "loss": 0.356,
+      "step": 7170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.648518429294146e-06,
+      "loss": 0.3738,
+      "step": 7180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.660563719585641e-06,
+      "loss": 0.5098,
+      "step": 7190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.67260900987714e-06,
+      "loss": 0.3821,
+      "step": 7200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.684654300168635e-06,
+      "loss": 0.5555,
+      "step": 7210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.69669959046013e-06,
+      "loss": 0.6138,
+      "step": 7220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.708744880751627e-06,
+      "loss": 0.2638,
+      "step": 7230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.720790171043122e-06,
+      "loss": 0.3204,
+      "step": 7240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.732835461334617e-06,
+      "loss": 0.4521,
+      "step": 7250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.744880751626116e-06,
+      "loss": 0.3897,
+      "step": 7260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.756926041917611e-06,
+      "loss": 0.3247,
+      "step": 7270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.768971332209106e-06,
+      "loss": 0.3119,
+      "step": 7280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.781016622500603e-06,
+      "loss": 0.3051,
+      "step": 7290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.793061912792098e-06,
+      "loss": 0.4196,
+      "step": 7300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.805107203083595e-06,
+      "loss": 0.4143,
+      "step": 7310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.81715249337509e-06,
+      "loss": 0.4781,
+      "step": 7320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.829197783666587e-06,
+      "loss": 0.4575,
+      "step": 7330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.841243073958082e-06,
+      "loss": 0.4867,
+      "step": 7340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.85328836424958e-06,
+      "loss": 0.4684,
+      "step": 7350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.865333654541074e-06,
+      "loss": 0.3675,
+      "step": 7360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.877378944832571e-06,
+      "loss": 0.5028,
+      "step": 7370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.889424235124067e-06,
+      "loss": 0.398,
+      "step": 7380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.901469525415563e-06,
+      "loss": 0.5037,
+      "step": 7390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.91351481570706e-06,
+      "loss": 0.3101,
+      "step": 7400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.925560105998555e-06,
+      "loss": 0.4027,
+      "step": 7410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.93760539629005e-06,
+      "loss": 0.4091,
+      "step": 7420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.949650686581548e-06,
+      "loss": 0.3533,
+      "step": 7430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.961695976873043e-06,
+      "loss": 0.4538,
+      "step": 7440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.973741267164538e-06,
+      "loss": 0.5103,
+      "step": 7450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.985786557456036e-06,
+      "loss": 0.5349,
+      "step": 7460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.997831847747532e-06,
+      "loss": 0.3082,
+      "step": 7470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.009877138039027e-06,
+      "loss": 0.4585,
+      "step": 7480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.021922428330524e-06,
+      "loss": 0.5309,
+      "step": 7490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.033967718622019e-06,
+      "loss": 0.3636,
+      "step": 7500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.046013008913514e-06,
+      "loss": 0.4219,
+      "step": 7510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.058058299205011e-06,
+      "loss": 0.3823,
+      "step": 7520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.070103589496508e-06,
+      "loss": 0.3659,
+      "step": 7530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.082148879788003e-06,
+      "loss": 0.6386,
+      "step": 7540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.0941941700795e-06,
+      "loss": 0.3494,
+      "step": 7550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.106239460370995e-06,
+      "loss": 0.307,
+      "step": 7560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.11828475066249e-06,
+      "loss": 0.4388,
+      "step": 7570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.130330040953987e-06,
+      "loss": 0.3636,
+      "step": 7580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.142375331245484e-06,
+      "loss": 0.3525,
+      "step": 7590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.15442062153698e-06,
+      "loss": 0.5595,
+      "step": 7600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.166465911828476e-06,
+      "loss": 0.5447,
+      "step": 7610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.178511202119971e-06,
+      "loss": 0.4755,
+      "step": 7620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.190556492411468e-06,
+      "loss": 0.3651,
+      "step": 7630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.202601782702963e-06,
+      "loss": 0.3949,
+      "step": 7640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.214647072994459e-06,
+      "loss": 0.4126,
+      "step": 7650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.226692363285955e-06,
+      "loss": 0.5115,
+      "step": 7660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.238737653577452e-06,
+      "loss": 0.3829,
+      "step": 7670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.250782943868947e-06,
+      "loss": 0.4345,
+      "step": 7680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.262828234160444e-06,
+      "loss": 0.4506,
+      "step": 7690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.27487352445194e-06,
+      "loss": 0.5201,
+      "step": 7700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.286918814743435e-06,
+      "loss": 0.6001,
+      "step": 7710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.298964105034932e-06,
+      "loss": 0.3452,
+      "step": 7720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.311009395326428e-06,
+      "loss": 0.5694,
+      "step": 7730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.323054685617924e-06,
+      "loss": 0.4353,
+      "step": 7740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.33509997590942e-06,
+      "loss": 0.4726,
+      "step": 7750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.347145266200916e-06,
+      "loss": 0.4451,
+      "step": 7760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.359190556492411e-06,
+      "loss": 0.3735,
+      "step": 7770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.371235846783908e-06,
+      "loss": 0.51,
+      "step": 7780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.383281137075405e-06,
+      "loss": 0.5572,
+      "step": 7790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.3953264273669e-06,
+      "loss": 0.5198,
+      "step": 7800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.407371717658397e-06,
+      "loss": 0.4625,
+      "step": 7810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.419417007949892e-06,
+      "loss": 0.4026,
+      "step": 7820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.431462298241387e-06,
+      "loss": 0.4422,
+      "step": 7830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.443507588532884e-06,
+      "loss": 0.3775,
+      "step": 7840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.455552878824381e-06,
+      "loss": 0.395,
+      "step": 7850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.467598169115876e-06,
+      "loss": 0.3672,
+      "step": 7860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.479643459407373e-06,
+      "loss": 0.4422,
+      "step": 7870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.491688749698868e-06,
+      "loss": 0.2415,
+      "step": 7880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.503734039990363e-06,
+      "loss": 0.3443,
+      "step": 7890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.51577933028186e-06,
+      "loss": 0.2497,
+      "step": 7900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.527824620573355e-06,
+      "loss": 0.4515,
+      "step": 7910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.539869910864852e-06,
+      "loss": 0.3656,
+      "step": 7920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.551915201156349e-06,
+      "loss": 0.3656,
+      "step": 7930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.563960491447844e-06,
+      "loss": 0.3278,
+      "step": 7940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.57600578173934e-06,
+      "loss": 0.3987,
+      "step": 7950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.588051072030836e-06,
+      "loss": 0.4519,
+      "step": 7960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.600096362322332e-06,
+      "loss": 0.3105,
+      "step": 7970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.612141652613828e-06,
+      "loss": 0.395,
+      "step": 7980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.624186942905325e-06,
+      "loss": 0.3364,
+      "step": 7990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.63623223319682e-06,
+      "loss": 0.3145,
+      "step": 8000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.648277523488317e-06,
+      "loss": 0.4203,
+      "step": 8010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.660322813779813e-06,
+      "loss": 0.3641,
+      "step": 8020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.672368104071308e-06,
+      "loss": 0.2783,
+      "step": 8030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.684413394362805e-06,
+      "loss": 0.3757,
+      "step": 8040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.696458684654301e-06,
+      "loss": 0.2093,
+      "step": 8050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.708503974945797e-06,
+      "loss": 0.3385,
+      "step": 8060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.720549265237294e-06,
+      "loss": 0.3739,
+      "step": 8070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.732594555528789e-06,
+      "loss": 0.2254,
+      "step": 8080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.744639845820284e-06,
+      "loss": 0.5837,
+      "step": 8090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.75668513611178e-06,
+      "loss": 0.4332,
+      "step": 8100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.768730426403276e-06,
+      "loss": 0.4345,
+      "step": 8110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.780775716694773e-06,
+      "loss": 0.3919,
+      "step": 8120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.79282100698627e-06,
+      "loss": 0.3164,
+      "step": 8130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804866297277765e-06,
+      "loss": 0.6981,
+      "step": 8140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.81691158756926e-06,
+      "loss": 0.3197,
+      "step": 8150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.828956877860757e-06,
+      "loss": 0.4367,
+      "step": 8160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.841002168152252e-06,
+      "loss": 0.2865,
+      "step": 8170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.853047458443749e-06,
+      "loss": 0.5013,
+      "step": 8180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.865092748735246e-06,
+      "loss": 0.3213,
+      "step": 8190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.877138039026741e-06,
+      "loss": 0.4613,
+      "step": 8200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.889183329318236e-06,
+      "loss": 0.4298,
+      "step": 8210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.901228619609733e-06,
+      "loss": 0.3051,
+      "step": 8220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.913273909901228e-06,
+      "loss": 0.3338,
+      "step": 8230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.925319200192725e-06,
+      "loss": 0.4111,
+      "step": 8240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.937364490484222e-06,
+      "loss": 0.4244,
+      "step": 8250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.949409780775717e-06,
+      "loss": 0.313,
+      "step": 8260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.961455071067212e-06,
+      "loss": 0.1936,
+      "step": 8270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.97350036135871e-06,
+      "loss": 0.4937,
+      "step": 8280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.985545651650205e-06,
+      "loss": 0.3544,
+      "step": 8290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.997590941941701e-06,
+      "loss": 0.37,
+      "step": 8300
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": {
+        "accuracy": 0.8933453778982234
+      },
+      "eval_f1": {
+        "f1": 0.8942100052466936
+      },
+      "eval_loss": 0.34615740180015564,
+      "eval_precision": {
+        "precision": 0.8984250247518094
+      },
+      "eval_recall": {
+        "recall": 0.8931370564158605
+      },
+      "eval_runtime": 655.829,
+      "eval_samples_per_second": 25.319,
+      "eval_steps_per_second": 12.66,
+      "step": 8302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0009636232233198e-05,
+      "loss": 0.3992,
+      "step": 8310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0021681522524694e-05,
+      "loss": 0.4586,
+      "step": 8320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.003372681281619e-05,
+      "loss": 0.3856,
+      "step": 8330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0045772103107686e-05,
+      "loss": 0.6489,
+      "step": 8340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.005781739339918e-05,
+      "loss": 0.3425,
+      "step": 8350
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0069862683690678e-05,
+      "loss": 0.2965,
+      "step": 8360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0081907973982173e-05,
+      "loss": 0.4241,
+      "step": 8370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.009395326427367e-05,
+      "loss": 0.6012,
+      "step": 8380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0105998554565167e-05,
+      "loss": 0.2937,
+      "step": 8390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0118043844856662e-05,
+      "loss": 0.4138,
+      "step": 8400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0130089135148157e-05,
+      "loss": 0.449,
+      "step": 8410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0142134425439654e-05,
+      "loss": 0.5532,
+      "step": 8420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0154179715731149e-05,
+      "loss": 0.3685,
+      "step": 8430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0166225006022646e-05,
+      "loss": 0.3075,
+      "step": 8440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0178270296314143e-05,
+      "loss": 0.4835,
+      "step": 8450
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0190315586605638e-05,
+      "loss": 0.3404,
+      "step": 8460
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0202360876897133e-05,
+      "loss": 0.385,
+      "step": 8470
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.021440616718863e-05,
+      "loss": 0.4614,
+      "step": 8480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0226451457480125e-05,
+      "loss": 0.6181,
+      "step": 8490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.023849674777162e-05,
+      "loss": 0.3392,
+      "step": 8500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0250542038063119e-05,
+      "loss": 0.3725,
+      "step": 8510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0262587328354614e-05,
+      "loss": 0.4357,
+      "step": 8520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.027463261864611e-05,
+      "loss": 0.3347,
+      "step": 8530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0286677908937606e-05,
+      "loss": 0.3535,
+      "step": 8540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0298723199229101e-05,
+      "loss": 0.3758,
+      "step": 8550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0310768489520598e-05,
+      "loss": 0.3641,
+      "step": 8560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0322813779812093e-05,
+      "loss": 0.3299,
+      "step": 8570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.033485907010359e-05,
+      "loss": 0.4344,
+      "step": 8580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0346904360395086e-05,
+      "loss": 0.2793,
+      "step": 8590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0358949650686582e-05,
+      "loss": 0.4237,
+      "step": 8600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0370994940978078e-05,
+      "loss": 0.2914,
+      "step": 8610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0383040231269574e-05,
+      "loss": 0.3509,
+      "step": 8620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.039508552156107e-05,
+      "loss": 0.3521,
+      "step": 8630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0407130811852567e-05,
+      "loss": 0.4525,
+      "step": 8640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0419176102144063e-05,
+      "loss": 0.5443,
+      "step": 8650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0431221392435559e-05,
+      "loss": 0.4707,
+      "step": 8660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0443266682727054e-05,
+      "loss": 0.3739,
+      "step": 8670
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.045531197301855e-05,
+      "loss": 0.2813,
+      "step": 8680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0467357263310046e-05,
+      "loss": 0.4741,
+      "step": 8690
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0479402553601541e-05,
+      "loss": 0.2454,
+      "step": 8700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.049144784389304e-05,
+      "loss": 0.2791,
+      "step": 8710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0503493134184535e-05,
+      "loss": 0.2423,
+      "step": 8720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.051553842447603e-05,
+      "loss": 0.4175,
+      "step": 8730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0527583714767527e-05,
+      "loss": 0.4691,
+      "step": 8740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0539629005059022e-05,
+      "loss": 0.2991,
+      "step": 8750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0551674295350517e-05,
+      "loss": 0.4591,
+      "step": 8760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0563719585642016e-05,
+      "loss": 0.458,
+      "step": 8770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0575764875933511e-05,
+      "loss": 0.3077,
+      "step": 8780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0587810166225006e-05,
+      "loss": 0.3674,
+      "step": 8790
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0599855456516503e-05,
+      "loss": 0.2467,
+      "step": 8800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0611900746807998e-05,
+      "loss": 0.3926,
+      "step": 8810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0623946037099493e-05,
+      "loss": 0.5028,
+      "step": 8820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.063599132739099e-05,
+      "loss": 0.3821,
+      "step": 8830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0648036617682487e-05,
+      "loss": 0.3092,
+      "step": 8840
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0660081907973982e-05,
+      "loss": 0.5485,
+      "step": 8850
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.067212719826548e-05,
+      "loss": 0.4605,
+      "step": 8860
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0684172488556974e-05,
+      "loss": 0.3811,
+      "step": 8870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0696217778848471e-05,
+      "loss": 0.4402,
+      "step": 8880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0708263069139966e-05,
+      "loss": 0.2989,
+      "step": 8890
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0720308359431463e-05,
+      "loss": 0.2415,
+      "step": 8900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0732353649722959e-05,
+      "loss": 0.4552,
+      "step": 8910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0744398940014455e-05,
+      "loss": 0.2925,
+      "step": 8920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.075644423030595e-05,
+      "loss": 0.4753,
+      "step": 8930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0768489520597447e-05,
+      "loss": 0.4537,
+      "step": 8940
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0780534810888943e-05,
+      "loss": 0.3199,
+      "step": 8950
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0792580101180438e-05,
+      "loss": 0.3307,
+      "step": 8960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0804625391471936e-05,
+      "loss": 0.5302,
+      "step": 8970
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0816670681763432e-05,
+      "loss": 0.4864,
+      "step": 8980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0828715972054927e-05,
+      "loss": 0.3208,
+      "step": 8990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0840761262346424e-05,
+      "loss": 0.52,
+      "step": 9000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0852806552637919e-05,
+      "loss": 0.3761,
+      "step": 9010
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0864851842929414e-05,
+      "loss": 0.3495,
+      "step": 9020
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0876897133220911e-05,
+      "loss": 0.2601,
+      "step": 9030
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0888942423512408e-05,
+      "loss": 0.4162,
+      "step": 9040
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0900987713803903e-05,
+      "loss": 0.614,
+      "step": 9050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.09130330040954e-05,
+      "loss": 0.4303,
+      "step": 9060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0925078294386895e-05,
+      "loss": 0.4174,
+      "step": 9070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.093712358467839e-05,
+      "loss": 0.2943,
+      "step": 9080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0949168874969887e-05,
+      "loss": 0.3932,
+      "step": 9090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0961214165261384e-05,
+      "loss": 0.3248,
+      "step": 9100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.097325945555288e-05,
+      "loss": 0.2313,
+      "step": 9110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0985304745844376e-05,
+      "loss": 0.3972,
+      "step": 9120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0997350036135871e-05,
+      "loss": 0.5109,
+      "step": 9130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1009395326427366e-05,
+      "loss": 0.2809,
+      "step": 9140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1021440616718863e-05,
+      "loss": 0.4447,
+      "step": 9150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1033485907010358e-05,
+      "loss": 0.4777,
+      "step": 9160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1045531197301855e-05,
+      "loss": 0.3876,
+      "step": 9170
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1057576487593352e-05,
+      "loss": 0.2433,
+      "step": 9180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1069621777884847e-05,
+      "loss": 0.3221,
+      "step": 9190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1081667068176344e-05,
+      "loss": 0.5058,
+      "step": 9200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.109371235846784e-05,
+      "loss": 0.3444,
+      "step": 9210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1105757648759335e-05,
+      "loss": 0.3818,
+      "step": 9220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1117802939050832e-05,
+      "loss": 0.3573,
+      "step": 9230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1129848229342328e-05,
+      "loss": 0.3912,
+      "step": 9240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1141893519633824e-05,
+      "loss": 0.4674,
+      "step": 9250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.115393880992532e-05,
+      "loss": 0.2574,
+      "step": 9260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1165984100216816e-05,
+      "loss": 0.5234,
+      "step": 9270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1178029390508311e-05,
+      "loss": 0.4797,
+      "step": 9280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1190074680799808e-05,
+      "loss": 0.3532,
+      "step": 9290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1202119971091305e-05,
+      "loss": 0.2743,
+      "step": 9300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.12141652613828e-05,
+      "loss": 0.2132,
+      "step": 9310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1226210551674297e-05,
+      "loss": 0.356,
+      "step": 9320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1238255841965792e-05,
+      "loss": 0.3616,
+      "step": 9330
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1250301132257287e-05,
+      "loss": 0.3397,
+      "step": 9340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1262346422548784e-05,
+      "loss": 0.3918,
+      "step": 9350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.127439171284028e-05,
+      "loss": 0.3928,
+      "step": 9360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1286437003131776e-05,
+      "loss": 0.3664,
+      "step": 9370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1298482293423273e-05,
+      "loss": 0.517,
+      "step": 9380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1310527583714768e-05,
+      "loss": 0.4504,
+      "step": 9390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1322572874006263e-05,
+      "loss": 0.5106,
+      "step": 9400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.133461816429776e-05,
+      "loss": 0.4331,
+      "step": 9410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1346663454589255e-05,
+      "loss": 0.3698,
+      "step": 9420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1358708744880752e-05,
+      "loss": 0.4401,
+      "step": 9430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1370754035172249e-05,
+      "loss": 0.5724,
+      "step": 9440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1382799325463744e-05,
+      "loss": 0.3299,
+      "step": 9450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.139484461575524e-05,
+      "loss": 0.4635,
+      "step": 9460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1406889906046736e-05,
+      "loss": 0.4139,
+      "step": 9470
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1418935196338232e-05,
+      "loss": 0.3881,
+      "step": 9480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1430980486629728e-05,
+      "loss": 0.4271,
+      "step": 9490
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1443025776921225e-05,
+      "loss": 0.2873,
+      "step": 9500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.145507106721272e-05,
+      "loss": 0.3298,
+      "step": 9510
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1467116357504216e-05,
+      "loss": 0.2938,
+      "step": 9520
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1479161647795713e-05,
+      "loss": 0.6126,
+      "step": 9530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1491206938087208e-05,
+      "loss": 0.3208,
+      "step": 9540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1503252228378705e-05,
+      "loss": 0.3831,
+      "step": 9550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1515297518670201e-05,
+      "loss": 0.2818,
+      "step": 9560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1527342808961697e-05,
+      "loss": 0.5447,
+      "step": 9570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1539388099253194e-05,
+      "loss": 0.2883,
+      "step": 9580
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1551433389544689e-05,
+      "loss": 0.5855,
+      "step": 9590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1563478679836184e-05,
+      "loss": 0.5799,
+      "step": 9600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.157552397012768e-05,
+      "loss": 0.5013,
+      "step": 9610
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1587569260419176e-05,
+      "loss": 0.1882,
+      "step": 9620
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1599614550710673e-05,
+      "loss": 0.2029,
+      "step": 9630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.161165984100217e-05,
+      "loss": 0.4352,
+      "step": 9640
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1623705131293665e-05,
+      "loss": 0.4715,
+      "step": 9650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.163575042158516e-05,
+      "loss": 0.4523,
+      "step": 9660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1647795711876657e-05,
+      "loss": 0.3172,
+      "step": 9670
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1659841002168152e-05,
+      "loss": 0.3978,
+      "step": 9680
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1671886292459649e-05,
+      "loss": 0.4146,
+      "step": 9690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1683931582751146e-05,
+      "loss": 0.3802,
+      "step": 9700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1695976873042641e-05,
+      "loss": 0.3293,
+      "step": 9710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1708022163334136e-05,
+      "loss": 0.3704,
+      "step": 9720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1720067453625633e-05,
+      "loss": 0.2578,
+      "step": 9730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1732112743917128e-05,
+      "loss": 0.6322,
+      "step": 9740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1744158034208624e-05,
+      "loss": 0.2488,
+      "step": 9750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1756203324500122e-05,
+      "loss": 0.3944,
+      "step": 9760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1768248614791617e-05,
+      "loss": 0.5959,
+      "step": 9770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1780293905083112e-05,
+      "loss": 0.3774,
+      "step": 9780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.179233919537461e-05,
+      "loss": 0.3298,
+      "step": 9790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1804384485666105e-05,
+      "loss": 0.4073,
+      "step": 9800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1816429775957601e-05,
+      "loss": 0.3669,
+      "step": 9810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1828475066249098e-05,
+      "loss": 0.327,
+      "step": 9820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1840520356540593e-05,
+      "loss": 0.3494,
+      "step": 9830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1852565646832089e-05,
+      "loss": 0.5345,
+      "step": 9840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1864610937123586e-05,
+      "loss": 0.4357,
+      "step": 9850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.187665622741508e-05,
+      "loss": 0.2851,
+      "step": 9860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1888701517706578e-05,
+      "loss": 0.2626,
+      "step": 9870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1900746807998073e-05,
+      "loss": 0.4714,
+      "step": 9880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.191279209828957e-05,
+      "loss": 0.3791,
+      "step": 9890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1924837388581067e-05,
+      "loss": 0.4164,
+      "step": 9900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1936882678872562e-05,
+      "loss": 0.2851,
+      "step": 9910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1948927969164057e-05,
+      "loss": 0.3515,
+      "step": 9920
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1960973259455554e-05,
+      "loss": 0.3537,
+      "step": 9930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1973018549747049e-05,
+      "loss": 0.2608,
+      "step": 9940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1985063840038544e-05,
+      "loss": 0.2467,
+      "step": 9950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1997109130330043e-05,
+      "loss": 0.4219,
+      "step": 9960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2009154420621538e-05,
+      "loss": 0.2925,
+      "step": 9970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2021199710913033e-05,
+      "loss": 0.2669,
+      "step": 9980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.203324500120453e-05,
+      "loss": 0.5412,
+      "step": 9990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2045290291496025e-05,
+      "loss": 0.4498,
+      "step": 10000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.205733558178752e-05,
+      "loss": 0.2609,
+      "step": 10010
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2069380872079019e-05,
+      "loss": 0.2133,
+      "step": 10020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2081426162370514e-05,
+      "loss": 0.4027,
+      "step": 10030
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.209347145266201e-05,
+      "loss": 0.256,
+      "step": 10040
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2105516742953506e-05,
+      "loss": 0.4117,
+      "step": 10050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2117562033245001e-05,
+      "loss": 0.5944,
+      "step": 10060
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2129607323536497e-05,
+      "loss": 0.4816,
+      "step": 10070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2141652613827993e-05,
+      "loss": 0.301,
+      "step": 10080
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.215369790411949e-05,
+      "loss": 0.3282,
+      "step": 10090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2165743194410986e-05,
+      "loss": 0.4364,
+      "step": 10100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2177788484702482e-05,
+      "loss": 0.2965,
+      "step": 10110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2189833774993978e-05,
+      "loss": 0.548,
+      "step": 10120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2201879065285474e-05,
+      "loss": 0.1975,
+      "step": 10130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.221392435557697e-05,
+      "loss": 0.4656,
+      "step": 10140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2225969645868467e-05,
+      "loss": 0.3649,
+      "step": 10150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2238014936159962e-05,
+      "loss": 0.4439,
+      "step": 10160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2250060226451459e-05,
+      "loss": 0.4052,
+      "step": 10170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2262105516742954e-05,
+      "loss": 0.313,
+      "step": 10180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.227415080703445e-05,
+      "loss": 0.3402,
+      "step": 10190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2286196097325946e-05,
+      "loss": 0.6,
+      "step": 10200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2298241387617441e-05,
+      "loss": 0.3297,
+      "step": 10210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.231028667790894e-05,
+      "loss": 0.2274,
+      "step": 10220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2322331968200435e-05,
+      "loss": 0.3509,
+      "step": 10230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.233437725849193e-05,
+      "loss": 0.2626,
+      "step": 10240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2346422548783427e-05,
+      "loss": 0.4718,
+      "step": 10250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2358467839074922e-05,
+      "loss": 0.4903,
+      "step": 10260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2370513129366417e-05,
+      "loss": 0.1601,
+      "step": 10270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2382558419657916e-05,
+      "loss": 0.6041,
+      "step": 10280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2394603709949411e-05,
+      "loss": 0.3548,
+      "step": 10290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2406649000240906e-05,
+      "loss": 0.3877,
+      "step": 10300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2418694290532403e-05,
+      "loss": 0.3786,
+      "step": 10310
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2430739580823898e-05,
+      "loss": 0.2247,
+      "step": 10320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2442784871115393e-05,
+      "loss": 0.3873,
+      "step": 10330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.245483016140689e-05,
+      "loss": 0.3348,
+      "step": 10340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2466875451698387e-05,
+      "loss": 0.3212,
+      "step": 10350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2478920741989882e-05,
+      "loss": 0.4174,
+      "step": 10360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.249096603228138e-05,
+      "loss": 0.2686,
+      "step": 10370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2503011322572874e-05,
+      "loss": 0.2603,
+      "step": 10380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.251505661286437e-05,
+      "loss": 0.3362,
+      "step": 10390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2527101903155866e-05,
+      "loss": 0.2682,
+      "step": 10400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2539147193447362e-05,
+      "loss": 0.2858,
+      "step": 10410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2551192483738859e-05,
+      "loss": 0.3353,
+      "step": 10420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2563237774030354e-05,
+      "loss": 0.4343,
+      "step": 10430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2575283064321849e-05,
+      "loss": 0.8225,
+      "step": 10440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2587328354613347e-05,
+      "loss": 0.5498,
+      "step": 10450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2599373644904844e-05,
+      "loss": 0.4412,
+      "step": 10460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.261141893519634e-05,
+      "loss": 0.4852,
+      "step": 10470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2623464225487835e-05,
+      "loss": 0.2934,
+      "step": 10480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2635509515779332e-05,
+      "loss": 0.2178,
+      "step": 10490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2647554806070827e-05,
+      "loss": 0.3313,
+      "step": 10500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2659600096362324e-05,
+      "loss": 0.4309,
+      "step": 10510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2671645386653819e-05,
+      "loss": 0.4869,
+      "step": 10520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2683690676945314e-05,
+      "loss": 0.3481,
+      "step": 10530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2695735967236811e-05,
+      "loss": 0.5286,
+      "step": 10540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2707781257528306e-05,
+      "loss": 0.4182,
+      "step": 10550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2719826547819801e-05,
+      "loss": 0.3799,
+      "step": 10560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2731871838111298e-05,
+      "loss": 0.3756,
+      "step": 10570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2743917128402797e-05,
+      "loss": 0.3464,
+      "step": 10580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2755962418694292e-05,
+      "loss": 0.2329,
+      "step": 10590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2768007708985789e-05,
+      "loss": 0.4941,
+      "step": 10600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2780052999277284e-05,
+      "loss": 0.3596,
+      "step": 10610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279209828956878e-05,
+      "loss": 0.2119,
+      "step": 10620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2804143579860276e-05,
+      "loss": 0.4331,
+      "step": 10630
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2816188870151771e-05,
+      "loss": 0.3859,
+      "step": 10640
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2828234160443266e-05,
+      "loss": 0.3466,
+      "step": 10650
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2840279450734763e-05,
+      "loss": 0.2155,
+      "step": 10660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2852324741026258e-05,
+      "loss": 0.3596,
+      "step": 10670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2864370031317755e-05,
+      "loss": 0.5381,
+      "step": 10680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.287641532160925e-05,
+      "loss": 0.4123,
+      "step": 10690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2888460611900746e-05,
+      "loss": 0.3519,
+      "step": 10700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2900505902192244e-05,
+      "loss": 0.2735,
+      "step": 10710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2912551192483741e-05,
+      "loss": 0.3256,
+      "step": 10720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2924596482775236e-05,
+      "loss": 0.3526,
+      "step": 10730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2936641773066732e-05,
+      "loss": 0.3734,
+      "step": 10740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2948687063358228e-05,
+      "loss": 0.3678,
+      "step": 10750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2960732353649724e-05,
+      "loss": 0.4595,
+      "step": 10760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2972777643941219e-05,
+      "loss": 0.2911,
+      "step": 10770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2984822934232716e-05,
+      "loss": 0.3885,
+      "step": 10780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2996868224524211e-05,
+      "loss": 0.239,
+      "step": 10790
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3008913514815708e-05,
+      "loss": 0.3013,
+      "step": 10800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3020958805107203e-05,
+      "loss": 0.3748,
+      "step": 10810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3033004095398698e-05,
+      "loss": 0.2455,
+      "step": 10820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3045049385690195e-05,
+      "loss": 0.4706,
+      "step": 10830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3057094675981694e-05,
+      "loss": 0.4589,
+      "step": 10840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3069139966273189e-05,
+      "loss": 0.3049,
+      "step": 10850
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3081185256564684e-05,
+      "loss": 0.3244,
+      "step": 10860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.309323054685618e-05,
+      "loss": 0.4397,
+      "step": 10870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3105275837147676e-05,
+      "loss": 0.2534,
+      "step": 10880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3117321127439173e-05,
+      "loss": 0.1862,
+      "step": 10890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3129366417730668e-05,
+      "loss": 0.6529,
+      "step": 10900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3141411708022163e-05,
+      "loss": 0.361,
+      "step": 10910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.315345699831366e-05,
+      "loss": 0.4102,
+      "step": 10920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3165502288605155e-05,
+      "loss": 0.4684,
+      "step": 10930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.317754757889665e-05,
+      "loss": 0.3702,
+      "step": 10940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3189592869188147e-05,
+      "loss": 0.3239,
+      "step": 10950
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3201638159479643e-05,
+      "loss": 0.245,
+      "step": 10960
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3213683449771141e-05,
+      "loss": 0.284,
+      "step": 10970
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3225728740062638e-05,
+      "loss": 0.3126,
+      "step": 10980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3237774030354133e-05,
+      "loss": 0.4432,
+      "step": 10990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3249819320645628e-05,
+      "loss": 0.2476,
+      "step": 11000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3261864610937125e-05,
+      "loss": 0.3528,
+      "step": 11010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.327390990122862e-05,
+      "loss": 0.2475,
+      "step": 11020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3285955191520116e-05,
+      "loss": 0.3123,
+      "step": 11030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3298000481811613e-05,
+      "loss": 0.2975,
+      "step": 11040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3310045772103108e-05,
+      "loss": 0.2531,
+      "step": 11050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3322091062394605e-05,
+      "loss": 0.5253,
+      "step": 11060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.33341363526861e-05,
+      "loss": 0.2272,
+      "step": 11070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3346181642977595e-05,
+      "loss": 0.5105,
+      "step": 11080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3358226933269092e-05,
+      "loss": 0.4141,
+      "step": 11090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.337027222356059e-05,
+      "loss": 0.5504,
+      "step": 11100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3382317513852086e-05,
+      "loss": 0.3032,
+      "step": 11110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.339436280414358e-05,
+      "loss": 0.4668,
+      "step": 11120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3406408094435078e-05,
+      "loss": 0.306,
+      "step": 11130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3418453384726573e-05,
+      "loss": 0.4476,
+      "step": 11140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.343049867501807e-05,
+      "loss": 0.3189,
+      "step": 11150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3442543965309565e-05,
+      "loss": 0.2039,
+      "step": 11160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.345458925560106e-05,
+      "loss": 0.4322,
+      "step": 11170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3466634545892557e-05,
+      "loss": 0.3732,
+      "step": 11180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3478679836184052e-05,
+      "loss": 0.3319,
+      "step": 11190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3490725126475547e-05,
+      "loss": 0.47,
+      "step": 11200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3502770416767044e-05,
+      "loss": 0.3992,
+      "step": 11210
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.351481570705854e-05,
+      "loss": 0.5242,
+      "step": 11220
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3526860997350036e-05,
+      "loss": 0.2599,
+      "step": 11230
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3538906287641535e-05,
+      "loss": 0.4286,
+      "step": 11240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.355095157793303e-05,
+      "loss": 0.2825,
+      "step": 11250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3562996868224525e-05,
+      "loss": 0.5086,
+      "step": 11260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3575042158516022e-05,
+      "loss": 0.2748,
+      "step": 11270
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3587087448807517e-05,
+      "loss": 0.2844,
+      "step": 11280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3599132739099012e-05,
+      "loss": 0.3002,
+      "step": 11290
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.361117802939051e-05,
+      "loss": 0.5451,
+      "step": 11300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3623223319682005e-05,
+      "loss": 0.3293,
+      "step": 11310
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.36352686099735e-05,
+      "loss": 0.5627,
+      "step": 11320
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3647313900264997e-05,
+      "loss": 0.242,
+      "step": 11330
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3659359190556492e-05,
+      "loss": 0.5285,
+      "step": 11340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3671404480847989e-05,
+      "loss": 0.5053,
+      "step": 11350
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3683449771139484e-05,
+      "loss": 0.4217,
+      "step": 11360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3695495061430982e-05,
+      "loss": 0.4739,
+      "step": 11370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3707540351722478e-05,
+      "loss": 0.3974,
+      "step": 11380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3719585642013974e-05,
+      "loss": 0.4268,
+      "step": 11390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.373163093230547e-05,
+      "loss": 0.4897,
+      "step": 11400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3743676222596965e-05,
+      "loss": 0.2236,
+      "step": 11410
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3755721512888462e-05,
+      "loss": 0.2109,
+      "step": 11420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3767766803179957e-05,
+      "loss": 0.2655,
+      "step": 11430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3779812093471454e-05,
+      "loss": 0.2143,
+      "step": 11440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3791857383762949e-05,
+      "loss": 0.3833,
+      "step": 11450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3803902674054444e-05,
+      "loss": 0.3263,
+      "step": 11460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3815947964345941e-05,
+      "loss": 0.327,
+      "step": 11470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3827993254637436e-05,
+      "loss": 0.4004,
+      "step": 11480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3840038544928931e-05,
+      "loss": 0.4076,
+      "step": 11490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.385208383522043e-05,
+      "loss": 0.2048,
+      "step": 11500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3864129125511927e-05,
+      "loss": 0.269,
+      "step": 11510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3876174415803422e-05,
+      "loss": 0.4911,
+      "step": 11520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3888219706094919e-05,
+      "loss": 0.5099,
+      "step": 11530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3900264996386414e-05,
+      "loss": 0.4482,
+      "step": 11540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.391231028667791e-05,
+      "loss": 0.3959,
+      "step": 11550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3924355576969406e-05,
+      "loss": 0.279,
+      "step": 11560
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3936400867260901e-05,
+      "loss": 0.3413,
+      "step": 11570
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3948446157552397e-05,
+      "loss": 0.4297,
+      "step": 11580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3960491447843893e-05,
+      "loss": 0.2419,
+      "step": 11590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3972536738135389e-05,
+      "loss": 0.1612,
+      "step": 11600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3984582028426885e-05,
+      "loss": 0.2313,
+      "step": 11610
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.399662731871838e-05,
+      "loss": 0.3355,
+      "step": 11620
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.400867260900988e-05,
+      "loss": 0.4352,
+      "step": 11630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4020717899301374e-05,
+      "loss": 0.1389,
+      "step": 11640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4032763189592871e-05,
+      "loss": 0.4996,
+      "step": 11650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4044808479884366e-05,
+      "loss": 0.5073,
+      "step": 11660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4056853770175862e-05,
+      "loss": 0.3096,
+      "step": 11670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4068899060467359e-05,
+      "loss": 0.3518,
+      "step": 11680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4080944350758854e-05,
+      "loss": 0.3444,
+      "step": 11690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.409298964105035e-05,
+      "loss": 0.2343,
+      "step": 11700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4105034931341846e-05,
+      "loss": 0.2693,
+      "step": 11710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4117080221633341e-05,
+      "loss": 0.5802,
+      "step": 11720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4129125511924838e-05,
+      "loss": 0.1507,
+      "step": 11730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4141170802216333e-05,
+      "loss": 0.3518,
+      "step": 11740
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4153216092507828e-05,
+      "loss": 0.4109,
+      "step": 11750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4165261382799327e-05,
+      "loss": 0.4011,
+      "step": 11760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4177306673090824e-05,
+      "loss": 0.2582,
+      "step": 11770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4189351963382319e-05,
+      "loss": 0.2418,
+      "step": 11780
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4201397253673816e-05,
+      "loss": 0.3764,
+      "step": 11790
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4213442543965311e-05,
+      "loss": 0.4159,
+      "step": 11800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4225487834256806e-05,
+      "loss": 0.2722,
+      "step": 11810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4237533124548303e-05,
+      "loss": 0.4073,
+      "step": 11820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4249578414839798e-05,
+      "loss": 0.3489,
+      "step": 11830
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4261623705131293e-05,
+      "loss": 0.3851,
+      "step": 11840
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.427366899542279e-05,
+      "loss": 0.3512,
+      "step": 11850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.3986,
+      "step": 11860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.429775957600578e-05,
+      "loss": 0.3274,
+      "step": 11870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4309804866297277e-05,
+      "loss": 0.3681,
+      "step": 11880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4321850156588776e-05,
+      "loss": 0.5337,
+      "step": 11890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4333895446880271e-05,
+      "loss": 0.2649,
+      "step": 11900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4345940737171768e-05,
+      "loss": 0.3293,
+      "step": 11910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4357986027463263e-05,
+      "loss": 0.1857,
+      "step": 11920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4370031317754759e-05,
+      "loss": 0.2115,
+      "step": 11930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4382076608046255e-05,
+      "loss": 0.4037,
+      "step": 11940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.439412189833775e-05,
+      "loss": 0.4671,
+      "step": 11950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4406167188629246e-05,
+      "loss": 0.3055,
+      "step": 11960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4418212478920743e-05,
+      "loss": 0.1612,
+      "step": 11970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4430257769212238e-05,
+      "loss": 0.4918,
+      "step": 11980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4442303059503735e-05,
+      "loss": 0.3537,
+      "step": 11990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.445434834979523e-05,
+      "loss": 0.5459,
+      "step": 12000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4466393640086725e-05,
+      "loss": 0.309,
+      "step": 12010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4478438930378224e-05,
+      "loss": 0.4333,
+      "step": 12020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.449048422066972e-05,
+      "loss": 0.3093,
+      "step": 12030
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4502529510961216e-05,
+      "loss": 0.4552,
+      "step": 12040
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4514574801252711e-05,
+      "loss": 0.4958,
+      "step": 12050
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4526620091544208e-05,
+      "loss": 0.4371,
+      "step": 12060
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4538665381835703e-05,
+      "loss": 0.2332,
+      "step": 12070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.45507106721272e-05,
+      "loss": 0.3764,
+      "step": 12080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4562755962418695e-05,
+      "loss": 0.6118,
+      "step": 12090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.457480125271019e-05,
+      "loss": 0.2334,
+      "step": 12100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4586846543001687e-05,
+      "loss": 0.3318,
+      "step": 12110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4598891833293182e-05,
+      "loss": 0.3161,
+      "step": 12120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4610937123584677e-05,
+      "loss": 0.2787,
+      "step": 12130
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4622982413876174e-05,
+      "loss": 0.5164,
+      "step": 12140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.463502770416767e-05,
+      "loss": 0.5754,
+      "step": 12150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4647072994459168e-05,
+      "loss": 0.3001,
+      "step": 12160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4659118284750665e-05,
+      "loss": 0.4091,
+      "step": 12170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.467116357504216e-05,
+      "loss": 0.3288,
+      "step": 12180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4683208865333655e-05,
+      "loss": 0.2347,
+      "step": 12190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4695254155625152e-05,
+      "loss": 0.5855,
+      "step": 12200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4707299445916647e-05,
+      "loss": 0.3206,
+      "step": 12210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4719344736208143e-05,
+      "loss": 0.1844,
+      "step": 12220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.473139002649964e-05,
+      "loss": 0.3442,
+      "step": 12230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4743435316791135e-05,
+      "loss": 0.2072,
+      "step": 12240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4755480607082632e-05,
+      "loss": 0.2881,
+      "step": 12250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4767525897374127e-05,
+      "loss": 0.5588,
+      "step": 12260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4779571187665622e-05,
+      "loss": 0.2865,
+      "step": 12270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4791616477957119e-05,
+      "loss": 0.3952,
+      "step": 12280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4803661768248617e-05,
+      "loss": 0.2356,
+      "step": 12290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4815707058540113e-05,
+      "loss": 0.4747,
+      "step": 12300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4827752348831608e-05,
+      "loss": 0.5004,
+      "step": 12310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4839797639123105e-05,
+      "loss": 0.235,
+      "step": 12320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.48518429294146e-05,
+      "loss": 0.3126,
+      "step": 12330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4863888219706095e-05,
+      "loss": 0.381,
+      "step": 12340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4875933509997592e-05,
+      "loss": 0.4341,
+      "step": 12350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4887978800289087e-05,
+      "loss": 0.4029,
+      "step": 12360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4900024090580584e-05,
+      "loss": 0.293,
+      "step": 12370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4912069380872079e-05,
+      "loss": 0.3431,
+      "step": 12380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4924114671163574e-05,
+      "loss": 0.3762,
+      "step": 12390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4936159961455071e-05,
+      "loss": 0.3159,
+      "step": 12400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4948205251746566e-05,
+      "loss": 0.4036,
+      "step": 12410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4960250542038065e-05,
+      "loss": 0.3693,
+      "step": 12420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.497229583232956e-05,
+      "loss": 0.3619,
+      "step": 12430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4984341122621057e-05,
+      "loss": 0.4055,
+      "step": 12440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4996386412912552e-05,
+      "loss": 0.2692,
+      "step": 12450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5008431703204049e-05,
+      "loss": 0.3584,
+      "step": 12460
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5020476993495544e-05,
+      "loss": 0.2851,
+      "step": 12470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.503252228378704e-05,
+      "loss": 0.3085,
+      "step": 12480
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5044567574078536e-05,
+      "loss": 0.4008,
+      "step": 12490
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5056612864370031e-05,
+      "loss": 0.4373,
+      "step": 12500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5068658154661527e-05,
+      "loss": 0.1429,
+      "step": 12510
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5080703444953024e-05,
+      "loss": 0.2767,
+      "step": 12520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5092748735244519e-05,
+      "loss": 0.3339,
+      "step": 12530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5104794025536016e-05,
+      "loss": 0.4184,
+      "step": 12540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5116839315827514e-05,
+      "loss": 0.3385,
+      "step": 12550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.512888460611901e-05,
+      "loss": 0.2424,
+      "step": 12560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5140929896410505e-05,
+      "loss": 0.195,
+      "step": 12570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5152975186702001e-05,
+      "loss": 0.3646,
+      "step": 12580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5165020476993497e-05,
+      "loss": 0.2277,
+      "step": 12590
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5177065767284992e-05,
+      "loss": 0.1904,
+      "step": 12600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5189111057576489e-05,
+      "loss": 0.4617,
+      "step": 12610
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5201156347867984e-05,
+      "loss": 0.2848,
+      "step": 12620
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.521320163815948e-05,
+      "loss": 0.3798,
+      "step": 12630
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5225246928450976e-05,
+      "loss": 0.2946,
+      "step": 12640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5237292218742471e-05,
+      "loss": 0.3064,
+      "step": 12650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5249337509033968e-05,
+      "loss": 0.4657,
+      "step": 12660
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5261382799325463e-05,
+      "loss": 0.348,
+      "step": 12670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5273428089616962e-05,
+      "loss": 0.3528,
+      "step": 12680
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5285473379908457e-05,
+      "loss": 0.2054,
+      "step": 12690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5297518670199952e-05,
+      "loss": 0.5211,
+      "step": 12700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.530956396049145e-05,
+      "loss": 0.3843,
+      "step": 12710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5321609250782946e-05,
+      "loss": 0.5084,
+      "step": 12720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.533365454107444e-05,
+      "loss": 0.2415,
+      "step": 12730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5345699831365936e-05,
+      "loss": 0.4332,
+      "step": 12740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.535774512165743e-05,
+      "loss": 0.2622,
+      "step": 12750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.536979041194893e-05,
+      "loss": 0.2234,
+      "step": 12760
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5381835702240425e-05,
+      "loss": 0.4075,
+      "step": 12770
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.539388099253192e-05,
+      "loss": 0.3508,
+      "step": 12780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5405926282823416e-05,
+      "loss": 0.5162,
+      "step": 12790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.541797157311491e-05,
+      "loss": 0.4172,
+      "step": 12800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.543001686340641e-05,
+      "loss": 0.2841,
+      "step": 12810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5442062153697904e-05,
+      "loss": 0.4602,
+      "step": 12820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5454107443989403e-05,
+      "loss": 0.3063,
+      "step": 12830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5466152734280898e-05,
+      "loss": 0.3252,
+      "step": 12840
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5478198024572393e-05,
+      "loss": 0.3649,
+      "step": 12850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.549024331486389e-05,
+      "loss": 0.4884,
+      "step": 12860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5502288605155384e-05,
+      "loss": 0.3019,
+      "step": 12870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5514333895446882e-05,
+      "loss": 0.3976,
+      "step": 12880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5526379185738378e-05,
+      "loss": 0.2001,
+      "step": 12890
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5538424476029873e-05,
+      "loss": 0.291,
+      "step": 12900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5550469766321368e-05,
+      "loss": 0.1655,
+      "step": 12910
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.5562515056612863e-05,
+      "loss": 0.1467,
+      "step": 12920
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.557456034690436e-05,
+      "loss": 0.4834,
+      "step": 12930
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.5586605637195857e-05,
+      "loss": 0.3435,
+      "step": 12940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.5598650927487355e-05,
+      "loss": 0.5977,
+      "step": 12950
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.561069621777885e-05,
+      "loss": 0.2923,
+      "step": 12960
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.5622741508070346e-05,
+      "loss": 0.3485,
+      "step": 12970
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.563478679836184e-05,
+      "loss": 0.2219,
+      "step": 12980
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.5646832088653336e-05,
+      "loss": 0.2824,
+      "step": 12990
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5658877378944835e-05,
+      "loss": 0.2508,
+      "step": 13000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.567092266923633e-05,
+      "loss": 0.2949,
+      "step": 13010
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5682967959527825e-05,
+      "loss": 0.2749,
+      "step": 13020
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.569501324981932e-05,
+      "loss": 0.2485,
+      "step": 13030
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5707058540110816e-05,
+      "loss": 0.3601,
+      "step": 13040
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5719103830402314e-05,
+      "loss": 0.2874,
+      "step": 13050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.573114912069381e-05,
+      "loss": 0.4797,
+      "step": 13060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5743194410985304e-05,
+      "loss": 0.2689,
+      "step": 13070
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5755239701276803e-05,
+      "loss": 0.4906,
+      "step": 13080
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5767284991568298e-05,
+      "loss": 0.4924,
+      "step": 13090
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5779330281859793e-05,
+      "loss": 0.2001,
+      "step": 13100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.579137557215129e-05,
+      "loss": 0.4605,
+      "step": 13110
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5803420862442787e-05,
+      "loss": 0.3408,
+      "step": 13120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5815466152734282e-05,
+      "loss": 0.2144,
+      "step": 13130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5827511443025778e-05,
+      "loss": 0.6161,
+      "step": 13140
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5839556733317273e-05,
+      "loss": 0.3949,
+      "step": 13150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5851602023608768e-05,
+      "loss": 0.3768,
+      "step": 13160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5863647313900266e-05,
+      "loss": 0.2804,
+      "step": 13170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.587569260419176e-05,
+      "loss": 0.4068,
+      "step": 13180
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5887737894483257e-05,
+      "loss": 0.4433,
+      "step": 13190
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5899783184774752e-05,
+      "loss": 0.294,
+      "step": 13200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.591182847506625e-05,
+      "loss": 0.4707,
+      "step": 13210
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5923873765357746e-05,
+      "loss": 0.2421,
+      "step": 13220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5935919055649244e-05,
+      "loss": 0.2884,
+      "step": 13230
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594796434594074e-05,
+      "loss": 0.4972,
+      "step": 13240
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5960009636232235e-05,
+      "loss": 0.3178,
+      "step": 13250
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.597205492652373e-05,
+      "loss": 0.2397,
+      "step": 13260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5984100216815225e-05,
+      "loss": 0.3195,
+      "step": 13270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.599614550710672e-05,
+      "loss": 0.3819,
+      "step": 13280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.600819079739822e-05,
+      "loss": 0.3761,
+      "step": 13290
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6020236087689714e-05,
+      "loss": 0.2743,
+      "step": 13300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.603228137798121e-05,
+      "loss": 0.3008,
+      "step": 13310
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6044326668272704e-05,
+      "loss": 0.4692,
+      "step": 13320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.60563719585642e-05,
+      "loss": 0.4879,
+      "step": 13330
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6068417248855698e-05,
+      "loss": 0.4925,
+      "step": 13340
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6080462539147197e-05,
+      "loss": 0.2146,
+      "step": 13350
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6092507829438692e-05,
+      "loss": 0.2931,
+      "step": 13360
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6104553119730187e-05,
+      "loss": 0.3135,
+      "step": 13370
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6116598410021682e-05,
+      "loss": 0.4451,
+      "step": 13380
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6128643700313177e-05,
+      "loss": 0.5261,
+      "step": 13390
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6140688990604673e-05,
+      "loss": 0.3075,
+      "step": 13400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.615273428089617e-05,
+      "loss": 0.4711,
+      "step": 13410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6164779571187666e-05,
+      "loss": 0.4821,
+      "step": 13420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.617682486147916e-05,
+      "loss": 0.1759,
+      "step": 13430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6188870151770657e-05,
+      "loss": 0.2648,
+      "step": 13440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6200915442062152e-05,
+      "loss": 0.2824,
+      "step": 13450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.621296073235365e-05,
+      "loss": 0.4145,
+      "step": 13460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.622500602264515e-05,
+      "loss": 0.4369,
+      "step": 13470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6237051312936644e-05,
+      "loss": 0.3467,
+      "step": 13480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.624909660322814e-05,
+      "loss": 0.4327,
+      "step": 13490
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6261141893519635e-05,
+      "loss": 0.3503,
+      "step": 13500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.627318718381113e-05,
+      "loss": 0.3238,
+      "step": 13510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.628523247410263e-05,
+      "loss": 0.2933,
+      "step": 13520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6297277764394124e-05,
+      "loss": 0.1341,
+      "step": 13530
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.630932305468562e-05,
+      "loss": 0.3176,
+      "step": 13540
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6321368344977114e-05,
+      "loss": 0.3101,
+      "step": 13550
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.633341363526861e-05,
+      "loss": 0.3854,
+      "step": 13560
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6345458925560104e-05,
+      "loss": 0.5578,
+      "step": 13570
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6357504215851603e-05,
+      "loss": 0.2908,
+      "step": 13580
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6369549506143098e-05,
+      "loss": 0.2367,
+      "step": 13590
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6381594796434597e-05,
+      "loss": 0.2731,
+      "step": 13600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6393640086726092e-05,
+      "loss": 0.4618,
+      "step": 13610
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6405685377017587e-05,
+      "loss": 0.4616,
+      "step": 13620
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6417730667309082e-05,
+      "loss": 0.2046,
+      "step": 13630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.642977595760058e-05,
+      "loss": 0.4087,
+      "step": 13640
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6441821247892076e-05,
+      "loss": 0.1865,
+      "step": 13650
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.645386653818357e-05,
+      "loss": 0.2905,
+      "step": 13660
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6465911828475066e-05,
+      "loss": 0.1579,
+      "step": 13670
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.647795711876656e-05,
+      "loss": 0.3411,
+      "step": 13680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.649000240905806e-05,
+      "loss": 0.6369,
+      "step": 13690
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6502047699349555e-05,
+      "loss": 0.3428,
+      "step": 13700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.651409298964105e-05,
+      "loss": 0.4011,
+      "step": 13710
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6526138279932546e-05,
+      "loss": 0.3921,
+      "step": 13720
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6538183570224044e-05,
+      "loss": 0.3676,
+      "step": 13730
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.655022886051554e-05,
+      "loss": 0.307,
+      "step": 13740
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6562274150807035e-05,
+      "loss": 0.2852,
+      "step": 13750
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6574319441098533e-05,
+      "loss": 0.41,
+      "step": 13760
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.658636473139003e-05,
+      "loss": 0.076,
+      "step": 13770
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6598410021681524e-05,
+      "loss": 0.258,
+      "step": 13780
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.661045531197302e-05,
+      "loss": 0.0401,
+      "step": 13790
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6622500602264514e-05,
+      "loss": 0.2154,
+      "step": 13800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6634545892556013e-05,
+      "loss": 0.3682,
+      "step": 13810
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6646591182847508e-05,
+      "loss": 0.5444,
+      "step": 13820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6658636473139003e-05,
+      "loss": 0.4248,
+      "step": 13830
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6670681763430498e-05,
+      "loss": 0.4025,
+      "step": 13840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6682727053721993e-05,
+      "loss": 0.1882,
+      "step": 13850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6694772344013492e-05,
+      "loss": 0.2606,
+      "step": 13860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6706817634304987e-05,
+      "loss": 0.2605,
+      "step": 13870
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6718862924596486e-05,
+      "loss": 0.3162,
+      "step": 13880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.673090821488798e-05,
+      "loss": 0.3562,
+      "step": 13890
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6742953505179476e-05,
+      "loss": 0.2218,
+      "step": 13900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.675499879547097e-05,
+      "loss": 0.4068,
+      "step": 13910
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6767044085762466e-05,
+      "loss": 0.465,
+      "step": 13920
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6779089376053965e-05,
+      "loss": 0.2806,
+      "step": 13930
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.679113466634546e-05,
+      "loss": 0.4725,
+      "step": 13940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6803179956636955e-05,
+      "loss": 0.216,
+      "step": 13950
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.681522524692845e-05,
+      "loss": 0.2318,
+      "step": 13960
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6827270537219946e-05,
+      "loss": 0.1856,
+      "step": 13970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6839315827511444e-05,
+      "loss": 0.3509,
+      "step": 13980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6851361117802943e-05,
+      "loss": 0.4952,
+      "step": 13990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6863406408094438e-05,
+      "loss": 0.124,
+      "step": 14000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6875451698385933e-05,
+      "loss": 0.2282,
+      "step": 14010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.688749698867743e-05,
+      "loss": 0.3701,
+      "step": 14020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6899542278968924e-05,
+      "loss": 0.4223,
+      "step": 14030
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.691158756926042e-05,
+      "loss": 0.202,
+      "step": 14040
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6923632859551917e-05,
+      "loss": 0.3866,
+      "step": 14050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6935678149843412e-05,
+      "loss": 0.3391,
+      "step": 14060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6947723440134908e-05,
+      "loss": 0.3552,
+      "step": 14070
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6959768730426403e-05,
+      "loss": 0.2743,
+      "step": 14080
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6971814020717898e-05,
+      "loss": 0.1825,
+      "step": 14090
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6983859311009397e-05,
+      "loss": 0.454,
+      "step": 14100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6995904601300892e-05,
+      "loss": 0.2734,
+      "step": 14110
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7007949891592387e-05,
+      "loss": 0.5079,
+      "step": 14120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7019995181883886e-05,
+      "loss": 0.2319,
+      "step": 14130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.703204047217538e-05,
+      "loss": 0.56,
+      "step": 14140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7044085762466876e-05,
+      "loss": 0.3642,
+      "step": 14150
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.7056131052758374e-05,
+      "loss": 0.4052,
+      "step": 14160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.706817634304987e-05,
+      "loss": 0.3666,
+      "step": 14170
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.7080221633341365e-05,
+      "loss": 0.4243,
+      "step": 14180
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.709226692363286e-05,
+      "loss": 0.47,
+      "step": 14190
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.7104312213924355e-05,
+      "loss": 0.318,
+      "step": 14200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.711635750421585e-05,
+      "loss": 0.3056,
+      "step": 14210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.712840279450735e-05,
+      "loss": 0.2701,
+      "step": 14220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.7140448084798844e-05,
+      "loss": 0.4098,
+      "step": 14230
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.715249337509034e-05,
+      "loss": 0.3387,
+      "step": 14240
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.7164538665381835e-05,
+      "loss": 0.3452,
+      "step": 14250
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.7176583955673333e-05,
+      "loss": 0.3347,
+      "step": 14260
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.7188629245964828e-05,
+      "loss": 0.1865,
+      "step": 14270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.7200674536256327e-05,
+      "loss": 0.2718,
+      "step": 14280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.7212719826547822e-05,
+      "loss": 0.4819,
+      "step": 14290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.7224765116839317e-05,
+      "loss": 0.3818,
+      "step": 14300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.7236810407130812e-05,
+      "loss": 0.2123,
+      "step": 14310
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.7248855697422308e-05,
+      "loss": 0.2762,
+      "step": 14320
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.7260900987713806e-05,
+      "loss": 0.3271,
+      "step": 14330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.72729462780053e-05,
+      "loss": 0.2539,
+      "step": 14340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.7284991568296797e-05,
+      "loss": 0.2468,
+      "step": 14350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.7297036858588292e-05,
+      "loss": 0.3247,
+      "step": 14360
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.7309082148879787e-05,
+      "loss": 0.3987,
+      "step": 14370
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.7321127439171282e-05,
+      "loss": 0.2014,
+      "step": 14380
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.733317272946278e-05,
+      "loss": 0.3405,
+      "step": 14390
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.734521801975428e-05,
+      "loss": 0.3838,
+      "step": 14400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7357263310045774e-05,
+      "loss": 0.3863,
+      "step": 14410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.736930860033727e-05,
+      "loss": 0.5188,
+      "step": 14420
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7381353890628765e-05,
+      "loss": 0.1735,
+      "step": 14430
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.739339918092026e-05,
+      "loss": 0.2148,
+      "step": 14440
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.740544447121176e-05,
+      "loss": 0.4637,
+      "step": 14450
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7417489761503254e-05,
+      "loss": 0.2373,
+      "step": 14460
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.742953505179475e-05,
+      "loss": 0.2514,
+      "step": 14470
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7441580342086244e-05,
+      "loss": 0.3044,
+      "step": 14480
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.745362563237774e-05,
+      "loss": 0.3063,
+      "step": 14490
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7465670922669234e-05,
+      "loss": 0.2131,
+      "step": 14500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7477716212960733e-05,
+      "loss": 0.3743,
+      "step": 14510
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.748976150325223e-05,
+      "loss": 0.2936,
+      "step": 14520
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7501806793543727e-05,
+      "loss": 0.2125,
+      "step": 14530
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7513852083835222e-05,
+      "loss": 0.4889,
+      "step": 14540
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7525897374126717e-05,
+      "loss": 0.4442,
+      "step": 14550
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7537942664418212e-05,
+      "loss": 0.2729,
+      "step": 14560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.754998795470971e-05,
+      "loss": 0.3475,
+      "step": 14570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7562033245001206e-05,
+      "loss": 0.3476,
+      "step": 14580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.75740785352927e-05,
+      "loss": 0.3827,
+      "step": 14590
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7586123825584196e-05,
+      "loss": 0.309,
+      "step": 14600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.759816911587569e-05,
+      "loss": 0.2272,
+      "step": 14610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.761021440616719e-05,
+      "loss": 0.2344,
+      "step": 14620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7622259696458685e-05,
+      "loss": 0.1675,
+      "step": 14630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.763430498675018e-05,
+      "loss": 0.3641,
+      "step": 14640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.764635027704168e-05,
+      "loss": 0.2936,
+      "step": 14650
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7658395567333174e-05,
+      "loss": 0.411,
+      "step": 14660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.767044085762467e-05,
+      "loss": 0.4164,
+      "step": 14670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7682486147916165e-05,
+      "loss": 0.2581,
+      "step": 14680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7694531438207663e-05,
+      "loss": 0.1938,
+      "step": 14690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.770657672849916e-05,
+      "loss": 0.2734,
+      "step": 14700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7718622018790654e-05,
+      "loss": 0.4574,
+      "step": 14710
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.773066730908215e-05,
+      "loss": 0.3181,
+      "step": 14720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7742712599373644e-05,
+      "loss": 0.353,
+      "step": 14730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7754757889665143e-05,
+      "loss": 0.3924,
+      "step": 14740
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7766803179956638e-05,
+      "loss": 0.2645,
+      "step": 14750
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7778848470248133e-05,
+      "loss": 0.2565,
+      "step": 14760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7790893760539628e-05,
+      "loss": 0.3047,
+      "step": 14770
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7802939050831127e-05,
+      "loss": 0.2603,
+      "step": 14780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7814984341122622e-05,
+      "loss": 0.2049,
+      "step": 14790
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.782702963141412e-05,
+      "loss": 0.2337,
+      "step": 14800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7839074921705616e-05,
+      "loss": 0.2218,
+      "step": 14810
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.785112021199711e-05,
+      "loss": 0.3291,
+      "step": 14820
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7863165502288606e-05,
+      "loss": 0.2858,
+      "step": 14830
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.78752107925801e-05,
+      "loss": 0.1728,
+      "step": 14840
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7887256082871596e-05,
+      "loss": 0.2959,
+      "step": 14850
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7899301373163095e-05,
+      "loss": 0.3913,
+      "step": 14860
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.791134666345459e-05,
+      "loss": 0.2921,
+      "step": 14870
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7923391953746085e-05,
+      "loss": 0.296,
+      "step": 14880
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.793543724403758e-05,
+      "loss": 0.2029,
+      "step": 14890
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7947482534329076e-05,
+      "loss": 0.5612,
+      "step": 14900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7959527824620574e-05,
+      "loss": 0.3015,
+      "step": 14910
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7971573114912073e-05,
+      "loss": 0.2218,
+      "step": 14920
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7983618405203568e-05,
+      "loss": 0.1559,
+      "step": 14930
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7995663695495063e-05,
+      "loss": 0.2545,
+      "step": 14940
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.800770898578656e-05,
+      "loss": 0.3471,
+      "step": 14950
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8019754276078054e-05,
+      "loss": 0.2408,
+      "step": 14960
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.803179956636955e-05,
+      "loss": 0.3018,
+      "step": 14970
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8043844856661047e-05,
+      "loss": 0.2983,
+      "step": 14980
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8055890146952543e-05,
+      "loss": 0.2574,
+      "step": 14990
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8067935437244038e-05,
+      "loss": 0.2971,
+      "step": 15000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8079980727535533e-05,
+      "loss": 0.3266,
+      "step": 15010
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8092026017827028e-05,
+      "loss": 0.4204,
+      "step": 15020
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8104071308118527e-05,
+      "loss": 0.2603,
+      "step": 15030
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8116116598410022e-05,
+      "loss": 0.3498,
+      "step": 15040
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.812816188870152e-05,
+      "loss": 0.2594,
+      "step": 15050
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8140207178993016e-05,
+      "loss": 0.3597,
+      "step": 15060
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.815225246928451e-05,
+      "loss": 0.4581,
+      "step": 15070
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8164297759576006e-05,
+      "loss": 0.2563,
+      "step": 15080
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8176343049867505e-05,
+      "loss": 0.2272,
+      "step": 15090
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8188388340159e-05,
+      "loss": 0.3986,
+      "step": 15100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8200433630450495e-05,
+      "loss": 0.3184,
+      "step": 15110
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.821247892074199e-05,
+      "loss": 0.5686,
+      "step": 15120
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8224524211033485e-05,
+      "loss": 0.5083,
+      "step": 15130
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.823656950132498e-05,
+      "loss": 0.1569,
+      "step": 15140
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.824861479161648e-05,
+      "loss": 0.3912,
+      "step": 15150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8260660081907974e-05,
+      "loss": 0.4585,
+      "step": 15160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.827270537219947e-05,
+      "loss": 0.219,
+      "step": 15170
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8284750662490968e-05,
+      "loss": 0.2725,
+      "step": 15180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8296795952782463e-05,
+      "loss": 0.1995,
+      "step": 15190
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.830884124307396e-05,
+      "loss": 0.5374,
+      "step": 15200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8320886533365457e-05,
+      "loss": 0.2917,
+      "step": 15210
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8332931823656952e-05,
+      "loss": 0.2898,
+      "step": 15220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8344977113948447e-05,
+      "loss": 0.2882,
+      "step": 15230
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8357022404239943e-05,
+      "loss": 0.4494,
+      "step": 15240
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8369067694531438e-05,
+      "loss": 0.4427,
+      "step": 15250
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8381112984822936e-05,
+      "loss": 0.6007,
+      "step": 15260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.839315827511443e-05,
+      "loss": 0.18,
+      "step": 15270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8405203565405927e-05,
+      "loss": 0.4799,
+      "step": 15280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8417248855697422e-05,
+      "loss": 0.3997,
+      "step": 15290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8429294145988917e-05,
+      "loss": 0.2451,
+      "step": 15300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8441339436280416e-05,
+      "loss": 0.1734,
+      "step": 15310
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.845338472657191e-05,
+      "loss": 0.4535,
+      "step": 15320
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.846543001686341e-05,
+      "loss": 0.508,
+      "step": 15330
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8477475307154905e-05,
+      "loss": 0.2511,
+      "step": 15340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.84895205974464e-05,
+      "loss": 0.285,
+      "step": 15350
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8501565887737895e-05,
+      "loss": 0.4584,
+      "step": 15360
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.851361117802939e-05,
+      "loss": 0.2,
+      "step": 15370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.852565646832089e-05,
+      "loss": 0.3063,
+      "step": 15380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8537701758612384e-05,
+      "loss": 0.244,
+      "step": 15390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.854974704890388e-05,
+      "loss": 0.406,
+      "step": 15400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8561792339195374e-05,
+      "loss": 0.1243,
+      "step": 15410
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.857383762948687e-05,
+      "loss": 0.2677,
+      "step": 15420
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8585882919778365e-05,
+      "loss": 0.271,
+      "step": 15430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8597928210069863e-05,
+      "loss": 0.1796,
+      "step": 15440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8609973500361362e-05,
+      "loss": 0.4569,
+      "step": 15450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8622018790652857e-05,
+      "loss": 0.3373,
+      "step": 15460
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8634064080944352e-05,
+      "loss": 0.3345,
+      "step": 15470
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8646109371235847e-05,
+      "loss": 0.2536,
+      "step": 15480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8658154661527342e-05,
+      "loss": 0.3955,
+      "step": 15490
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.867019995181884e-05,
+      "loss": 0.3549,
+      "step": 15500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8682245242110336e-05,
+      "loss": 0.4603,
+      "step": 15510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.869429053240183e-05,
+      "loss": 0.3886,
+      "step": 15520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8706335822693327e-05,
+      "loss": 0.5313,
+      "step": 15530
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8718381112984822e-05,
+      "loss": 0.2165,
+      "step": 15540
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.873042640327632e-05,
+      "loss": 0.2749,
+      "step": 15550
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8742471693567816e-05,
+      "loss": 0.3809,
+      "step": 15560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8754516983859314e-05,
+      "loss": 0.3826,
+      "step": 15570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.876656227415081e-05,
+      "loss": 0.2446,
+      "step": 15580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8778607564442305e-05,
+      "loss": 0.396,
+      "step": 15590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.87906528547338e-05,
+      "loss": 0.2011,
+      "step": 15600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8802698145025295e-05,
+      "loss": 0.1042,
+      "step": 15610
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8814743435316793e-05,
+      "loss": 0.3822,
+      "step": 15620
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.882678872560829e-05,
+      "loss": 0.3027,
+      "step": 15630
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8838834015899784e-05,
+      "loss": 0.3612,
+      "step": 15640
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.885087930619128e-05,
+      "loss": 0.3436,
+      "step": 15650
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8862924596482774e-05,
+      "loss": 0.2867,
+      "step": 15660
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8874969886774273e-05,
+      "loss": 0.1464,
+      "step": 15670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8887015177065768e-05,
+      "loss": 0.3208,
+      "step": 15680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8899060467357263e-05,
+      "loss": 0.4024,
+      "step": 15690
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8911105757648762e-05,
+      "loss": 0.3799,
+      "step": 15700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8923151047940257e-05,
+      "loss": 0.4071,
+      "step": 15710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8935196338231752e-05,
+      "loss": 0.201,
+      "step": 15720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.894724162852325e-05,
+      "loss": 0.2279,
+      "step": 15730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8959286918814746e-05,
+      "loss": 0.3912,
+      "step": 15740
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.897133220910624e-05,
+      "loss": 0.3171,
+      "step": 15750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8983377499397736e-05,
+      "loss": 0.3496,
+      "step": 15760
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.899542278968923e-05,
+      "loss": 0.4952,
+      "step": 15770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9007468079980727e-05,
+      "loss": 0.4063,
+      "step": 15780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9019513370272225e-05,
+      "loss": 0.4042,
+      "step": 15790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.903155866056372e-05,
+      "loss": 0.2415,
+      "step": 15800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9043603950855216e-05,
+      "loss": 0.3367,
+      "step": 15810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.905564924114671e-05,
+      "loss": 0.2631,
+      "step": 15820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.906769453143821e-05,
+      "loss": 0.4188,
+      "step": 15830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9079739821729704e-05,
+      "loss": 0.3314,
+      "step": 15840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9091785112021203e-05,
+      "loss": 0.3671,
+      "step": 15850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9103830402312698e-05,
+      "loss": 0.5761,
+      "step": 15860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9115875692604193e-05,
+      "loss": 0.3058,
+      "step": 15870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.912792098289569e-05,
+      "loss": 0.2773,
+      "step": 15880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9139966273187184e-05,
+      "loss": 0.2895,
+      "step": 15890
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.915201156347868e-05,
+      "loss": 0.4341,
+      "step": 15900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9164056853770178e-05,
+      "loss": 0.3176,
+      "step": 15910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9176102144061673e-05,
+      "loss": 0.4531,
+      "step": 15920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9188147434353168e-05,
+      "loss": 0.3601,
+      "step": 15930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9200192724644663e-05,
+      "loss": 0.3109,
+      "step": 15940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9212238014936158e-05,
+      "loss": 0.3579,
+      "step": 15950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9224283305227657e-05,
+      "loss": 0.3883,
+      "step": 15960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9236328595519155e-05,
+      "loss": 0.2297,
+      "step": 15970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.924837388581065e-05,
+      "loss": 0.2892,
+      "step": 15980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9260419176102146e-05,
+      "loss": 0.1598,
+      "step": 15990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.927246446639364e-05,
+      "loss": 0.4829,
+      "step": 16000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9284509756685136e-05,
+      "loss": 0.4168,
+      "step": 16010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9296555046976635e-05,
+      "loss": 0.2772,
+      "step": 16020
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.930860033726813e-05,
+      "loss": 0.2266,
+      "step": 16030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9320645627559625e-05,
+      "loss": 0.3934,
+      "step": 16040
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.933269091785112e-05,
+      "loss": 0.3363,
+      "step": 16050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9344736208142615e-05,
+      "loss": 0.4,
+      "step": 16060
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935678149843411e-05,
+      "loss": 0.3144,
+      "step": 16070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.936882678872561e-05,
+      "loss": 0.3816,
+      "step": 16080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9380872079017104e-05,
+      "loss": 0.275,
+      "step": 16090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9392917369308603e-05,
+      "loss": 0.3182,
+      "step": 16100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9404962659600098e-05,
+      "loss": 0.2334,
+      "step": 16110
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9417007949891593e-05,
+      "loss": 0.4573,
+      "step": 16120
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.942905324018309e-05,
+      "loss": 0.4261,
+      "step": 16130
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9441098530474587e-05,
+      "loss": 0.2527,
+      "step": 16140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9453143820766082e-05,
+      "loss": 0.2289,
+      "step": 16150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9465189111057577e-05,
+      "loss": 0.4979,
+      "step": 16160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9477234401349073e-05,
+      "loss": 0.3014,
+      "step": 16170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9489279691640568e-05,
+      "loss": 0.2372,
+      "step": 16180
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9501324981932066e-05,
+      "loss": 0.2293,
+      "step": 16190
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.951337027222356e-05,
+      "loss": 0.514,
+      "step": 16200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9525415562515057e-05,
+      "loss": 0.2959,
+      "step": 16210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9537460852806552e-05,
+      "loss": 0.4496,
+      "step": 16220
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.954950614309805e-05,
+      "loss": 0.3025,
+      "step": 16230
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9561551433389546e-05,
+      "loss": 0.3887,
+      "step": 16240
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.957359672368104e-05,
+      "loss": 0.267,
+      "step": 16250
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.958564201397254e-05,
+      "loss": 0.4207,
+      "step": 16260
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9597687304264035e-05,
+      "loss": 0.3424,
+      "step": 16270
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.960973259455553e-05,
+      "loss": 0.2413,
+      "step": 16280
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9621777884847025e-05,
+      "loss": 0.1827,
+      "step": 16290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.963382317513852e-05,
+      "loss": 0.2709,
+      "step": 16300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.964586846543002e-05,
+      "loss": 0.2257,
+      "step": 16310
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9657913755721514e-05,
+      "loss": 0.324,
+      "step": 16320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.966995904601301e-05,
+      "loss": 0.1879,
+      "step": 16330
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9682004336304504e-05,
+      "loss": 0.2552,
+      "step": 16340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9694049626596e-05,
+      "loss": 0.2307,
+      "step": 16350
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9706094916887498e-05,
+      "loss": 0.3287,
+      "step": 16360
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9718140207178997e-05,
+      "loss": 0.4426,
+      "step": 16370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9730185497470492e-05,
+      "loss": 0.3599,
+      "step": 16380
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9742230787761987e-05,
+      "loss": 0.457,
+      "step": 16390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9754276078053482e-05,
+      "loss": 0.2401,
+      "step": 16400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9766321368344977e-05,
+      "loss": 0.3895,
+      "step": 16410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9778366658636473e-05,
+      "loss": 0.2891,
+      "step": 16420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.979041194892797e-05,
+      "loss": 0.3111,
+      "step": 16430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9802457239219466e-05,
+      "loss": 0.2576,
+      "step": 16440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.981450252951096e-05,
+      "loss": 0.2424,
+      "step": 16450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9826547819802457e-05,
+      "loss": 0.5225,
+      "step": 16460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9838593110093952e-05,
+      "loss": 0.3513,
+      "step": 16470
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.985063840038545e-05,
+      "loss": 0.1253,
+      "step": 16480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.986268369067695e-05,
+      "loss": 0.3113,
+      "step": 16490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9874728980968444e-05,
+      "loss": 0.2435,
+      "step": 16500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.988677427125994e-05,
+      "loss": 0.2349,
+      "step": 16510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9898819561551435e-05,
+      "loss": 0.3336,
+      "step": 16520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.991086485184293e-05,
+      "loss": 0.2742,
+      "step": 16530
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9922910142134425e-05,
+      "loss": 0.3472,
+      "step": 16540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9934955432425924e-05,
+      "loss": 0.31,
+      "step": 16550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.994700072271742e-05,
+      "loss": 0.3302,
+      "step": 16560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9959046013008914e-05,
+      "loss": 0.2803,
+      "step": 16570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.997109130330041e-05,
+      "loss": 0.1179,
+      "step": 16580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9983136593591904e-05,
+      "loss": 0.4523,
+      "step": 16590
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9995181883883403e-05,
+      "loss": 0.2375,
+      "step": 16600
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": {
+        "accuracy": 0.9053297199638664
+      },
+      "eval_f1": {
+        "f1": 0.9062005892826234
+      },
+      "eval_loss": 0.33534759283065796,
+      "eval_precision": {
+        "precision": 0.912717242831991
+      },
+      "eval_recall": {
+        "recall": 0.9052684275828231
+      },
+      "eval_runtime": 649.6653,
+      "eval_samples_per_second": 25.559,
+      "eval_steps_per_second": 12.78,
+      "step": 16605
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0007227174174898e-05,
+      "loss": 0.2914,
+      "step": 16610
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0019272464466397e-05,
+      "loss": 0.305,
+      "step": 16620
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0031317754757892e-05,
+      "loss": 0.2866,
+      "step": 16630
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0043363045049387e-05,
+      "loss": 0.3344,
+      "step": 16640
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0055408335340882e-05,
+      "loss": 0.1779,
+      "step": 16650
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.006745362563238e-05,
+      "loss": 0.3102,
+      "step": 16660
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0079498915923876e-05,
+      "loss": 0.357,
+      "step": 16670
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.009154420621537e-05,
+      "loss": 0.1414,
+      "step": 16680
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0103589496506866e-05,
+      "loss": 0.3453,
+      "step": 16690
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.011563478679836e-05,
+      "loss": 0.1864,
+      "step": 16700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0127680077089857e-05,
+      "loss": 0.2298,
+      "step": 16710
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0139725367381355e-05,
+      "loss": 0.2703,
+      "step": 16720
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.015177065767285e-05,
+      "loss": 0.3941,
+      "step": 16730
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0163815947964346e-05,
+      "loss": 0.0865,
+      "step": 16740
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0175861238255844e-05,
+      "loss": 0.1091,
+      "step": 16750
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.018790652854734e-05,
+      "loss": 0.2061,
+      "step": 16760
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0199951818838835e-05,
+      "loss": 0.4656,
+      "step": 16770
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0211997109130333e-05,
+      "loss": 0.4075,
+      "step": 16780
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.022404239942183e-05,
+      "loss": 0.3266,
+      "step": 16790
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0236087689713324e-05,
+      "loss": 0.3694,
+      "step": 16800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.024813298000482e-05,
+      "loss": 0.3631,
+      "step": 16810
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0260178270296314e-05,
+      "loss": 0.2557,
+      "step": 16820
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0272223560587812e-05,
+      "loss": 0.438,
+      "step": 16830
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0284268850879308e-05,
+      "loss": 0.2812,
+      "step": 16840
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0296314141170803e-05,
+      "loss": 0.2438,
+      "step": 16850
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0308359431462298e-05,
+      "loss": 0.1862,
+      "step": 16860
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0320404721753793e-05,
+      "loss": 0.3811,
+      "step": 16870
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0332450012045292e-05,
+      "loss": 0.2505,
+      "step": 16880
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0344495302336787e-05,
+      "loss": 0.4308,
+      "step": 16890
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.0356540592628286e-05,
+      "loss": 0.2882,
+      "step": 16900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.036858588291978e-05,
+      "loss": 0.3546,
+      "step": 16910
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.0380631173211276e-05,
+      "loss": 0.2764,
+      "step": 16920
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.039267646350277e-05,
+      "loss": 0.2269,
+      "step": 16930
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.0404721753794266e-05,
+      "loss": 0.2786,
+      "step": 16940
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.0416767044085765e-05,
+      "loss": 0.1561,
+      "step": 16950
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.042881233437726e-05,
+      "loss": 0.215,
+      "step": 16960
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.0440857624668755e-05,
+      "loss": 0.292,
+      "step": 16970
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.045290291496025e-05,
+      "loss": 0.3788,
+      "step": 16980
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.0464948205251746e-05,
+      "loss": 0.1888,
+      "step": 16990
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.047699349554324e-05,
+      "loss": 0.2586,
+      "step": 17000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.048903878583474e-05,
+      "loss": 0.2408,
+      "step": 17010
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.0501084076126238e-05,
+      "loss": 0.532,
+      "step": 17020
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.0513129366417733e-05,
+      "loss": 0.2302,
+      "step": 17030
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.0525174656709228e-05,
+      "loss": 0.375,
+      "step": 17040
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.0537219947000723e-05,
+      "loss": 0.1673,
+      "step": 17050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.054926523729222e-05,
+      "loss": 0.3951,
+      "step": 17060
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0561310527583717e-05,
+      "loss": 0.254,
+      "step": 17070
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0573355817875212e-05,
+      "loss": 0.3323,
+      "step": 17080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0585401108166708e-05,
+      "loss": 0.201,
+      "step": 17090
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0597446398458203e-05,
+      "loss": 0.25,
+      "step": 17100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0609491688749698e-05,
+      "loss": 0.3145,
+      "step": 17110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0621536979041197e-05,
+      "loss": 0.2449,
+      "step": 17120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0633582269332692e-05,
+      "loss": 0.1466,
+      "step": 17130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0645627559624187e-05,
+      "loss": 0.2614,
+      "step": 17140
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.0657672849915685e-05,
+      "loss": 0.2366,
+      "step": 17150
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.066971814020718e-05,
+      "loss": 0.333,
+      "step": 17160
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.0681763430498676e-05,
+      "loss": 0.6166,
+      "step": 17170
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.069380872079017e-05,
+      "loss": 0.2378,
+      "step": 17180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.070585401108167e-05,
+      "loss": 0.3643,
+      "step": 17190
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.0717899301373165e-05,
+      "loss": 0.3749,
+      "step": 17200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.072994459166466e-05,
+      "loss": 0.3215,
+      "step": 17210
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.0741989881956155e-05,
+      "loss": 0.249,
+      "step": 17220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.075403517224765e-05,
+      "loss": 0.2997,
+      "step": 17230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.076608046253915e-05,
+      "loss": 0.405,
+      "step": 17240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.0778125752830644e-05,
+      "loss": 0.3253,
+      "step": 17250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.079017104312214e-05,
+      "loss": 0.4503,
+      "step": 17260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.0802216333413634e-05,
+      "loss": 0.5839,
+      "step": 17270
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.0814261623705133e-05,
+      "loss": 0.2655,
+      "step": 17280
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.0826306913996628e-05,
+      "loss": 0.4254,
+      "step": 17290
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.0838352204288127e-05,
+      "loss": 0.23,
+      "step": 17300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.0850397494579622e-05,
+      "loss": 0.1886,
+      "step": 17310
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.0862442784871117e-05,
+      "loss": 0.242,
+      "step": 17320
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.0874488075162612e-05,
+      "loss": 0.2815,
+      "step": 17330
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.0886533365454108e-05,
+      "loss": 0.4134,
+      "step": 17340
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.0898578655745603e-05,
+      "loss": 0.2259,
+      "step": 17350
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.09106239460371e-05,
+      "loss": 0.3114,
+      "step": 17360
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.0922669236328596e-05,
+      "loss": 0.2106,
+      "step": 17370
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.093471452662009e-05,
+      "loss": 0.4336,
+      "step": 17380
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.0946759816911587e-05,
+      "loss": 0.1755,
+      "step": 17390
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.0958805107203082e-05,
+      "loss": 0.5467,
+      "step": 17400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.097085039749458e-05,
+      "loss": 0.3853,
+      "step": 17410
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.098289568778608e-05,
+      "loss": 0.2602,
+      "step": 17420
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.0994940978077574e-05,
+      "loss": 0.3967,
+      "step": 17430
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.100698626836907e-05,
+      "loss": 0.2118,
+      "step": 17440
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.1019031558660565e-05,
+      "loss": 0.4101,
+      "step": 17450
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.103107684895206e-05,
+      "loss": 0.2633,
+      "step": 17460
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.1043122139243555e-05,
+      "loss": 0.2408,
+      "step": 17470
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.1055167429535054e-05,
+      "loss": 0.448,
+      "step": 17480
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.106721271982655e-05,
+      "loss": 0.3208,
+      "step": 17490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.1079258010118044e-05,
+      "loss": 0.1705,
+      "step": 17500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.109130330040954e-05,
+      "loss": 0.1935,
+      "step": 17510
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.1103348590701034e-05,
+      "loss": 0.4944,
+      "step": 17520
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.1115393880992533e-05,
+      "loss": 0.1404,
+      "step": 17530
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.112743917128403e-05,
+      "loss": 0.3484,
+      "step": 17540
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.1139484461575527e-05,
+      "loss": 0.2847,
+      "step": 17550
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.1151529751867022e-05,
+      "loss": 0.3686,
+      "step": 17560
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.1163575042158517e-05,
+      "loss": 0.2192,
+      "step": 17570
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.1175620332450012e-05,
+      "loss": 0.3375,
+      "step": 17580
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.118766562274151e-05,
+      "loss": 0.2324,
+      "step": 17590
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.1199710913033006e-05,
+      "loss": 0.0811,
+      "step": 17600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.12117562033245e-05,
+      "loss": 0.2242,
+      "step": 17610
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.1223801493615996e-05,
+      "loss": 0.1555,
+      "step": 17620
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.123584678390749e-05,
+      "loss": 0.3489,
+      "step": 17630
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.1247892074198987e-05,
+      "loss": 0.3034,
+      "step": 17640
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.1259937364490485e-05,
+      "loss": 0.0997,
+      "step": 17650
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.127198265478198e-05,
+      "loss": 0.2187,
+      "step": 17660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.128402794507348e-05,
+      "loss": 0.1508,
+      "step": 17670
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.1296073235364974e-05,
+      "loss": 0.3971,
+      "step": 17680
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.130811852565647e-05,
+      "loss": 0.3516,
+      "step": 17690
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.1320163815947965e-05,
+      "loss": 0.2255,
+      "step": 17700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.1332209106239463e-05,
+      "loss": 0.1815,
+      "step": 17710
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.134425439653096e-05,
+      "loss": 0.1485,
+      "step": 17720
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.1356299686822454e-05,
+      "loss": 0.2289,
+      "step": 17730
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.136834497711395e-05,
+      "loss": 0.2994,
+      "step": 17740
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.1380390267405444e-05,
+      "loss": 0.4039,
+      "step": 17750
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.1392435557696943e-05,
+      "loss": 0.0527,
+      "step": 17760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.1404480847988438e-05,
+      "loss": 0.1493,
+      "step": 17770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.1416526138279933e-05,
+      "loss": 0.2762,
+      "step": 17780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.45,
+      "step": 17790
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.1440616718862927e-05,
+      "loss": 0.3264,
+      "step": 17800
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.1452662009154422e-05,
+      "loss": 0.2049,
+      "step": 17810
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.1464707299445917e-05,
+      "loss": 0.2005,
+      "step": 17820
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.1476752589737416e-05,
+      "loss": 0.3575,
+      "step": 17830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.148879788002891e-05,
+      "loss": 0.3329,
+      "step": 17840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.1500843170320406e-05,
+      "loss": 0.3524,
+      "step": 17850
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.15128884606119e-05,
+      "loss": 0.4009,
+      "step": 17860
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.1524933750903396e-05,
+      "loss": 0.3011,
+      "step": 17870
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.1536979041194895e-05,
+      "loss": 0.32,
+      "step": 17880
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.154902433148639e-05,
+      "loss": 0.2219,
+      "step": 17890
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1561069621777885e-05,
+      "loss": 0.3954,
+      "step": 17900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.157311491206938e-05,
+      "loss": 0.2671,
+      "step": 17910
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1585160202360876e-05,
+      "loss": 0.2958,
+      "step": 17920
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.159720549265237e-05,
+      "loss": 0.2239,
+      "step": 17930
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1609250782943873e-05,
+      "loss": 0.3014,
+      "step": 17940
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1621296073235368e-05,
+      "loss": 0.2026,
+      "step": 17950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1633341363526863e-05,
+      "loss": 0.2436,
+      "step": 17960
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.164538665381836e-05,
+      "loss": 0.3339,
+      "step": 17970
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1657431944109854e-05,
+      "loss": 0.3628,
+      "step": 17980
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.166947723440135e-05,
+      "loss": 0.2508,
+      "step": 17990
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1681522524692847e-05,
+      "loss": 0.459,
+      "step": 18000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1693567814984343e-05,
+      "loss": 0.3504,
+      "step": 18010
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1705613105275838e-05,
+      "loss": 0.2809,
+      "step": 18020
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1717658395567333e-05,
+      "loss": 0.2621,
+      "step": 18030
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1729703685858828e-05,
+      "loss": 0.28,
+      "step": 18040
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1741748976150327e-05,
+      "loss": 0.404,
+      "step": 18050
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1753794266441822e-05,
+      "loss": 0.3653,
+      "step": 18060
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.176583955673332e-05,
+      "loss": 0.2114,
+      "step": 18070
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1777884847024816e-05,
+      "loss": 0.6794,
+      "step": 18080
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.178993013731631e-05,
+      "loss": 0.2391,
+      "step": 18090
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1801975427607806e-05,
+      "loss": 0.3012,
+      "step": 18100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.18140207178993e-05,
+      "loss": 0.3213,
+      "step": 18110
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.18260660081908e-05,
+      "loss": 0.2734,
+      "step": 18120
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1838111298482295e-05,
+      "loss": 0.0894,
+      "step": 18130
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.185015658877379e-05,
+      "loss": 0.3062,
+      "step": 18140
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1862201879065285e-05,
+      "loss": 0.2842,
+      "step": 18150
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.187424716935678e-05,
+      "loss": 0.086,
+      "step": 18160
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.188629245964828e-05,
+      "loss": 0.353,
+      "step": 18170
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1898337749939774e-05,
+      "loss": 0.4306,
+      "step": 18180
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.191038304023127e-05,
+      "loss": 0.1276,
+      "step": 18190
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1922428330522768e-05,
+      "loss": 0.2943,
+      "step": 18200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1934473620814263e-05,
+      "loss": 0.2604,
+      "step": 18210
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.194651891110576e-05,
+      "loss": 0.3189,
+      "step": 18220
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.1958564201397257e-05,
+      "loss": 0.3042,
+      "step": 18230
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.1970609491688752e-05,
+      "loss": 0.2671,
+      "step": 18240
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.1982654781980247e-05,
+      "loss": 0.6304,
+      "step": 18250
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.1994700072271742e-05,
+      "loss": 0.2382,
+      "step": 18260
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2006745362563238e-05,
+      "loss": 0.3108,
+      "step": 18270
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2018790652854733e-05,
+      "loss": 0.2383,
+      "step": 18280
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.203083594314623e-05,
+      "loss": 0.2426,
+      "step": 18290
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2042881233437727e-05,
+      "loss": 0.4343,
+      "step": 18300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2054926523729222e-05,
+      "loss": 0.3187,
+      "step": 18310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2066971814020717e-05,
+      "loss": 0.3462,
+      "step": 18320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2079017104312216e-05,
+      "loss": 0.2321,
+      "step": 18330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.209106239460371e-05,
+      "loss": 0.229,
+      "step": 18340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.210310768489521e-05,
+      "loss": 0.3456,
+      "step": 18350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2115152975186705e-05,
+      "loss": 0.2307,
+      "step": 18360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.21271982654782e-05,
+      "loss": 0.3269,
+      "step": 18370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2139243555769695e-05,
+      "loss": 0.2373,
+      "step": 18380
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.215128884606119e-05,
+      "loss": 0.4131,
+      "step": 18390
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.216333413635269e-05,
+      "loss": 0.3287,
+      "step": 18400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2175379426644184e-05,
+      "loss": 0.3601,
+      "step": 18410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.218742471693568e-05,
+      "loss": 0.3011,
+      "step": 18420
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2199470007227174e-05,
+      "loss": 0.3371,
+      "step": 18430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.221151529751867e-05,
+      "loss": 0.432,
+      "step": 18440
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2223560587810165e-05,
+      "loss": 0.1302,
+      "step": 18450
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2235605878101663e-05,
+      "loss": 0.3449,
+      "step": 18460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2247651168393162e-05,
+      "loss": 0.3238,
+      "step": 18470
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2259696458684657e-05,
+      "loss": 0.3781,
+      "step": 18480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2271741748976152e-05,
+      "loss": 0.2215,
+      "step": 18490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2283787039267647e-05,
+      "loss": 0.2023,
+      "step": 18500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2295832329559142e-05,
+      "loss": 0.3182,
+      "step": 18510
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.230787761985064e-05,
+      "loss": 0.2086,
+      "step": 18520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2319922910142136e-05,
+      "loss": 0.333,
+      "step": 18530
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.233196820043363e-05,
+      "loss": 0.4096,
+      "step": 18540
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2344013490725127e-05,
+      "loss": 0.2103,
+      "step": 18550
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2356058781016622e-05,
+      "loss": 0.3476,
+      "step": 18560
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2368104071308117e-05,
+      "loss": 0.5966,
+      "step": 18570
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2380149361599616e-05,
+      "loss": 0.244,
+      "step": 18580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2392194651891114e-05,
+      "loss": 0.2959,
+      "step": 18590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.240423994218261e-05,
+      "loss": 0.3444,
+      "step": 18600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2416285232474104e-05,
+      "loss": 0.5162,
+      "step": 18610
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.24283305227656e-05,
+      "loss": 0.1157,
+      "step": 18620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2440375813057095e-05,
+      "loss": 0.3528,
+      "step": 18630
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2452421103348593e-05,
+      "loss": 0.1366,
+      "step": 18640
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.246446639364009e-05,
+      "loss": 0.2158,
+      "step": 18650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2476511683931584e-05,
+      "loss": 0.221,
+      "step": 18660
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.248855697422308e-05,
+      "loss": 0.2644,
+      "step": 18670
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2500602264514574e-05,
+      "loss": 0.2404,
+      "step": 18680
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2512647554806073e-05,
+      "loss": 0.3361,
+      "step": 18690
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2524692845097568e-05,
+      "loss": 0.2452,
+      "step": 18700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2536738135389063e-05,
+      "loss": 0.3258,
+      "step": 18710
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.254878342568056e-05,
+      "loss": 0.1848,
+      "step": 18720
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.2560828715972057e-05,
+      "loss": 0.3118,
+      "step": 18730
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.2572874006263552e-05,
+      "loss": 0.3791,
+      "step": 18740
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.2584919296555047e-05,
+      "loss": 0.6041,
+      "step": 18750
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.2596964586846546e-05,
+      "loss": 0.324,
+      "step": 18760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.260900987713804e-05,
+      "loss": 0.4027,
+      "step": 18770
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.2621055167429536e-05,
+      "loss": 0.1868,
+      "step": 18780
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.263310045772103e-05,
+      "loss": 0.195,
+      "step": 18790
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.2645145748012527e-05,
+      "loss": 0.3552,
+      "step": 18800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.2657191038304025e-05,
+      "loss": 0.3043,
+      "step": 18810
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.266923632859552e-05,
+      "loss": 0.4583,
+      "step": 18820
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.2681281618887015e-05,
+      "loss": 0.2351,
+      "step": 18830
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.269332690917851e-05,
+      "loss": 0.129,
+      "step": 18840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.2705372199470006e-05,
+      "loss": 0.456,
+      "step": 18850
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.2717417489761504e-05,
+      "loss": 0.4243,
+      "step": 18860
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.2729462780053003e-05,
+      "loss": 0.6836,
+      "step": 18870
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.2741508070344498e-05,
+      "loss": 0.4416,
+      "step": 18880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.2753553360635993e-05,
+      "loss": 0.1777,
+      "step": 18890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.276559865092749e-05,
+      "loss": 0.1317,
+      "step": 18900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.2777643941218984e-05,
+      "loss": 0.4661,
+      "step": 18910
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.278968923151048e-05,
+      "loss": 0.3253,
+      "step": 18920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.2801734521801977e-05,
+      "loss": 0.2412,
+      "step": 18930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.2813779812093473e-05,
+      "loss": 0.4389,
+      "step": 18940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.2825825102384968e-05,
+      "loss": 0.1474,
+      "step": 18950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.2837870392676463e-05,
+      "loss": 0.2352,
+      "step": 18960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.2849915682967958e-05,
+      "loss": 0.3675,
+      "step": 18970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.2861960973259457e-05,
+      "loss": 0.1691,
+      "step": 18980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.2874006263550955e-05,
+      "loss": 0.2047,
+      "step": 18990
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.288605155384245e-05,
+      "loss": 0.2093,
+      "step": 19000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.2898096844133946e-05,
+      "loss": 0.1675,
+      "step": 19010
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.291014213442544e-05,
+      "loss": 0.1145,
+      "step": 19020
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.2922187424716936e-05,
+      "loss": 0.2146,
+      "step": 19030
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.293423271500843e-05,
+      "loss": 0.2248,
+      "step": 19040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.294627800529993e-05,
+      "loss": 0.4549,
+      "step": 19050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2958323295591425e-05,
+      "loss": 0.3018,
+      "step": 19060
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.297036858588292e-05,
+      "loss": 0.2102,
+      "step": 19070
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2982413876174415e-05,
+      "loss": 0.3029,
+      "step": 19080
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.299445916646591e-05,
+      "loss": 0.1932,
+      "step": 19090
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.300650445675741e-05,
+      "loss": 0.3838,
+      "step": 19100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3018549747048904e-05,
+      "loss": 0.2002,
+      "step": 19110
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3030595037340403e-05,
+      "loss": 0.4463,
+      "step": 19120
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3042640327631898e-05,
+      "loss": 0.2369,
+      "step": 19130
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3054685617923393e-05,
+      "loss": 0.2805,
+      "step": 19140
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.306673090821489e-05,
+      "loss": 0.2233,
+      "step": 19150
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3078776198506387e-05,
+      "loss": 0.3876,
+      "step": 19160
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3090821488797882e-05,
+      "loss": 0.2707,
+      "step": 19170
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3102866779089377e-05,
+      "loss": 0.4088,
+      "step": 19180
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3114912069380873e-05,
+      "loss": 0.2962,
+      "step": 19190
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3126957359672368e-05,
+      "loss": 0.2919,
+      "step": 19200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3139002649963863e-05,
+      "loss": 0.2467,
+      "step": 19210
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.315104794025536e-05,
+      "loss": 0.2536,
+      "step": 19220
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3163093230546857e-05,
+      "loss": 0.1904,
+      "step": 19230
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3175138520838352e-05,
+      "loss": 0.2516,
+      "step": 19240
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.318718381112985e-05,
+      "loss": 0.3618,
+      "step": 19250
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3199229101421346e-05,
+      "loss": 0.1558,
+      "step": 19260
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.321127439171284e-05,
+      "loss": 0.2178,
+      "step": 19270
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.322331968200434e-05,
+      "loss": 0.1738,
+      "step": 19280
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3235364972295835e-05,
+      "loss": 0.3009,
+      "step": 19290
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.324741026258733e-05,
+      "loss": 0.2911,
+      "step": 19300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3259455552878825e-05,
+      "loss": 0.2195,
+      "step": 19310
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.327150084317032e-05,
+      "loss": 0.309,
+      "step": 19320
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.328354613346182e-05,
+      "loss": 0.296,
+      "step": 19330
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3295591423753314e-05,
+      "loss": 0.4469,
+      "step": 19340
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.330763671404481e-05,
+      "loss": 0.3804,
+      "step": 19350
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3319682004336304e-05,
+      "loss": 0.3234,
+      "step": 19360
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.33317272946278e-05,
+      "loss": 0.1655,
+      "step": 19370
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3343772584919298e-05,
+      "loss": 0.201,
+      "step": 19380
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3355817875210793e-05,
+      "loss": 0.3988,
+      "step": 19390
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3367863165502292e-05,
+      "loss": 0.2351,
+      "step": 19400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3379908455793787e-05,
+      "loss": 0.177,
+      "step": 19410
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3391953746085282e-05,
+      "loss": 0.3211,
+      "step": 19420
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3403999036376777e-05,
+      "loss": 0.4941,
+      "step": 19430
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3416044326668273e-05,
+      "loss": 0.2463,
+      "step": 19440
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.342808961695977e-05,
+      "loss": 0.3675,
+      "step": 19450
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3440134907251266e-05,
+      "loss": 0.1161,
+      "step": 19460
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.345218019754276e-05,
+      "loss": 0.2396,
+      "step": 19470
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3464225487834257e-05,
+      "loss": 0.2403,
+      "step": 19480
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3476270778125752e-05,
+      "loss": 0.3182,
+      "step": 19490
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3488316068417247e-05,
+      "loss": 0.4083,
+      "step": 19500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3500361358708746e-05,
+      "loss": 0.2613,
+      "step": 19510
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3512406649000244e-05,
+      "loss": 0.2057,
+      "step": 19520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.352445193929174e-05,
+      "loss": 0.3623,
+      "step": 19530
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3536497229583235e-05,
+      "loss": 0.3913,
+      "step": 19540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.354854251987473e-05,
+      "loss": 0.2372,
+      "step": 19550
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3560587810166225e-05,
+      "loss": 0.427,
+      "step": 19560
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3572633100457724e-05,
+      "loss": 0.5092,
+      "step": 19570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.358467839074922e-05,
+      "loss": 0.4897,
+      "step": 19580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3596723681040714e-05,
+      "loss": 0.449,
+      "step": 19590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.360876897133221e-05,
+      "loss": 0.2216,
+      "step": 19600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3620814261623704e-05,
+      "loss": 0.2319,
+      "step": 19610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3632859551915203e-05,
+      "loss": 0.3157,
+      "step": 19620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3644904842206698e-05,
+      "loss": 0.2008,
+      "step": 19630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3656950132498197e-05,
+      "loss": 0.2138,
+      "step": 19640
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3668995422789692e-05,
+      "loss": 0.2873,
+      "step": 19650
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3681040713081187e-05,
+      "loss": 0.1801,
+      "step": 19660
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3693086003372682e-05,
+      "loss": 0.116,
+      "step": 19670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3705131293664177e-05,
+      "loss": 0.4723,
+      "step": 19680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3717176583955676e-05,
+      "loss": 0.4795,
+      "step": 19690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.372922187424717e-05,
+      "loss": 0.2466,
+      "step": 19700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3741267164538666e-05,
+      "loss": 0.3965,
+      "step": 19710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.375331245483016e-05,
+      "loss": 0.1324,
+      "step": 19720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3765357745121657e-05,
+      "loss": 0.1808,
+      "step": 19730
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3777403035413155e-05,
+      "loss": 0.277,
+      "step": 19740
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.378944832570465e-05,
+      "loss": 0.3511,
+      "step": 19750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3801493615996146e-05,
+      "loss": 0.1964,
+      "step": 19760
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3813538906287644e-05,
+      "loss": 0.1887,
+      "step": 19770
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.382558419657914e-05,
+      "loss": 0.4115,
+      "step": 19780
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3837629486870635e-05,
+      "loss": 0.2176,
+      "step": 19790
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3849674777162133e-05,
+      "loss": 0.3131,
+      "step": 19800
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3861720067453628e-05,
+      "loss": 0.2453,
+      "step": 19810
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3873765357745123e-05,
+      "loss": 0.3278,
+      "step": 19820
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.388581064803662e-05,
+      "loss": 0.1994,
+      "step": 19830
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3897855938328114e-05,
+      "loss": 0.3938,
+      "step": 19840
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.390990122861961e-05,
+      "loss": 0.5587,
+      "step": 19850
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3921946518911108e-05,
+      "loss": 0.2673,
+      "step": 19860
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3933991809202603e-05,
+      "loss": 0.1382,
+      "step": 19870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3946037099494098e-05,
+      "loss": 0.2567,
+      "step": 19880
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3958082389785593e-05,
+      "loss": 0.2311,
+      "step": 19890
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.397012768007709e-05,
+      "loss": 0.1598,
+      "step": 19900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3982172970368587e-05,
+      "loss": 0.2893,
+      "step": 19910
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3994218260660085e-05,
+      "loss": 0.3223,
+      "step": 19920
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.400626355095158e-05,
+      "loss": 0.4007,
+      "step": 19930
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.4018308841243076e-05,
+      "loss": 0.3541,
+      "step": 19940
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.403035413153457e-05,
+      "loss": 0.5569,
+      "step": 19950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.4042399421826066e-05,
+      "loss": 0.1709,
+      "step": 19960
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.4054444712117565e-05,
+      "loss": 0.3553,
+      "step": 19970
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.406649000240906e-05,
+      "loss": 0.419,
+      "step": 19980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.4078535292700555e-05,
+      "loss": 0.223,
+      "step": 19990
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.409058058299205e-05,
+      "loss": 0.2623,
+      "step": 20000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.4102625873283546e-05,
+      "loss": 0.2627,
+      "step": 20010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.411467116357504e-05,
+      "loss": 0.3253,
+      "step": 20020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.412671645386654e-05,
+      "loss": 0.2556,
+      "step": 20030
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.4138761744158038e-05,
+      "loss": 0.0949,
+      "step": 20040
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.4150807034449533e-05,
+      "loss": 0.2357,
+      "step": 20050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.4162852324741028e-05,
+      "loss": 0.3031,
+      "step": 20060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.4174897615032523e-05,
+      "loss": 0.3499,
+      "step": 20070
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.418694290532402e-05,
+      "loss": 0.2368,
+      "step": 20080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.4198988195615517e-05,
+      "loss": 0.2478,
+      "step": 20090
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.4211033485907012e-05,
+      "loss": 0.2183,
+      "step": 20100
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.4223078776198508e-05,
+      "loss": 0.2982,
+      "step": 20110
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.4235124066490003e-05,
+      "loss": 0.3106,
+      "step": 20120
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.4247169356781498e-05,
+      "loss": 0.2505,
+      "step": 20130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4259214647072993e-05,
+      "loss": 0.4763,
+      "step": 20140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.427125993736449e-05,
+      "loss": 0.3278,
+      "step": 20150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4283305227655987e-05,
+      "loss": 0.244,
+      "step": 20160
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4295350517947485e-05,
+      "loss": 0.4553,
+      "step": 20170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.430739580823898e-05,
+      "loss": 0.4193,
+      "step": 20180
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4319441098530476e-05,
+      "loss": 0.2976,
+      "step": 20190
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.433148638882197e-05,
+      "loss": 0.5022,
+      "step": 20200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.434353167911347e-05,
+      "loss": 0.3763,
+      "step": 20210
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4355576969404965e-05,
+      "loss": 0.2371,
+      "step": 20220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.436762225969646e-05,
+      "loss": 0.3035,
+      "step": 20230
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4379667549987955e-05,
+      "loss": 0.2506,
+      "step": 20240
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.439171284027945e-05,
+      "loss": 0.1691,
+      "step": 20250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.440375813057095e-05,
+      "loss": 0.2119,
+      "step": 20260
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4415803420862444e-05,
+      "loss": 0.2203,
+      "step": 20270
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.442784871115394e-05,
+      "loss": 0.2862,
+      "step": 20280
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4439894001445434e-05,
+      "loss": 0.202,
+      "step": 20290
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4451939291736933e-05,
+      "loss": 0.512,
+      "step": 20300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4463984582028428e-05,
+      "loss": 0.4032,
+      "step": 20310
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4476029872319923e-05,
+      "loss": 0.4375,
+      "step": 20320
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4488075162611422e-05,
+      "loss": 0.4582,
+      "step": 20330
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4500120452902917e-05,
+      "loss": 0.3347,
+      "step": 20340
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4512165743194412e-05,
+      "loss": 0.2841,
+      "step": 20350
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4524211033485908e-05,
+      "loss": 0.4269,
+      "step": 20360
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4536256323777403e-05,
+      "loss": 0.2332,
+      "step": 20370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.45483016140689e-05,
+      "loss": 0.4699,
+      "step": 20380
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.4560346904360396e-05,
+      "loss": 0.2231,
+      "step": 20390
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.457239219465189e-05,
+      "loss": 0.3149,
+      "step": 20400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.4584437484943387e-05,
+      "loss": 0.2881,
+      "step": 20410
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.4596482775234882e-05,
+      "loss": 0.4093,
+      "step": 20420
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.460852806552638e-05,
+      "loss": 0.248,
+      "step": 20430
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.462057335581788e-05,
+      "loss": 0.3611,
+      "step": 20440
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.4632618646109374e-05,
+      "loss": 0.2147,
+      "step": 20450
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.464466393640087e-05,
+      "loss": 0.3283,
+      "step": 20460
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4656709226692365e-05,
+      "loss": 0.0916,
+      "step": 20470
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.466875451698386e-05,
+      "loss": 0.1911,
+      "step": 20480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4680799807275355e-05,
+      "loss": 0.3472,
+      "step": 20490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4692845097566854e-05,
+      "loss": 0.2637,
+      "step": 20500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.470489038785835e-05,
+      "loss": 0.1025,
+      "step": 20510
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4716935678149844e-05,
+      "loss": 0.377,
+      "step": 20520
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.472898096844134e-05,
+      "loss": 0.3237,
+      "step": 20530
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.4741026258732834e-05,
+      "loss": 0.2445,
+      "step": 20540
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4753071549024333e-05,
+      "loss": 0.3494,
+      "step": 20550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.476511683931583e-05,
+      "loss": 0.2109,
+      "step": 20560
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4777162129607327e-05,
+      "loss": 0.2756,
+      "step": 20570
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4789207419898822e-05,
+      "loss": 0.1373,
+      "step": 20580
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4801252710190317e-05,
+      "loss": 0.2192,
+      "step": 20590
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4813298000481812e-05,
+      "loss": 0.1628,
+      "step": 20600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4825343290773307e-05,
+      "loss": 0.1605,
+      "step": 20610
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.4837388581064806e-05,
+      "loss": 0.6493,
+      "step": 20620
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.48494338713563e-05,
+      "loss": 0.1737,
+      "step": 20630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.4861479161647796e-05,
+      "loss": 0.068,
+      "step": 20640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.487352445193929e-05,
+      "loss": 0.0615,
+      "step": 20650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.4885569742230787e-05,
+      "loss": 0.2926,
+      "step": 20660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.4897615032522285e-05,
+      "loss": 0.5142,
+      "step": 20670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.490966032281378e-05,
+      "loss": 0.5157,
+      "step": 20680
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.492170561310528e-05,
+      "loss": 0.2207,
+      "step": 20690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.4933750903396774e-05,
+      "loss": 0.259,
+      "step": 20700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.494579619368827e-05,
+      "loss": 0.2244,
+      "step": 20710
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.4957841483979765e-05,
+      "loss": 0.2879,
+      "step": 20720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.4969886774271263e-05,
+      "loss": 0.2618,
+      "step": 20730
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.498193206456276e-05,
+      "loss": 0.1757,
+      "step": 20740
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.4993977354854254e-05,
+      "loss": 0.2613,
+      "step": 20750
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.500602264514575e-05,
+      "loss": 0.2259,
+      "step": 20760
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5018067935437244e-05,
+      "loss": 0.1728,
+      "step": 20770
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.503011322572874e-05,
+      "loss": 0.2653,
+      "step": 20780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5042158516020238e-05,
+      "loss": 0.2818,
+      "step": 20790
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5054203806311733e-05,
+      "loss": 0.4594,
+      "step": 20800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5066249096603228e-05,
+      "loss": 0.2846,
+      "step": 20810
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5078294386894723e-05,
+      "loss": 0.2171,
+      "step": 20820
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.509033967718622e-05,
+      "loss": 0.3856,
+      "step": 20830
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5102384967477717e-05,
+      "loss": 0.2612,
+      "step": 20840
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5114430257769212e-05,
+      "loss": 0.4043,
+      "step": 20850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5126475548060707e-05,
+      "loss": 0.4001,
+      "step": 20860
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5138520838352203e-05,
+      "loss": 0.1688,
+      "step": 20870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5150566128643698e-05,
+      "loss": 0.2486,
+      "step": 20880
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.51626114189352e-05,
+      "loss": 0.4157,
+      "step": 20890
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5174656709226695e-05,
+      "loss": 0.1169,
+      "step": 20900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5186701999518194e-05,
+      "loss": 0.2877,
+      "step": 20910
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.519874728980969e-05,
+      "loss": 0.3148,
+      "step": 20920
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5210792580101184e-05,
+      "loss": 0.3448,
+      "step": 20930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.522283787039268e-05,
+      "loss": 0.4345,
+      "step": 20940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5234883160684174e-05,
+      "loss": 0.3308,
+      "step": 20950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.524692845097567e-05,
+      "loss": 0.3733,
+      "step": 20960
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5258973741267168e-05,
+      "loss": 0.2372,
+      "step": 20970
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5271019031558663e-05,
+      "loss": 0.3075,
+      "step": 20980
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.528306432185016e-05,
+      "loss": 0.3181,
+      "step": 20990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5295109612141654e-05,
+      "loss": 0.2813,
+      "step": 21000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.530715490243315e-05,
+      "loss": 0.5037,
+      "step": 21010
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5319200192724647e-05,
+      "loss": 0.1678,
+      "step": 21020
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5331245483016143e-05,
+      "loss": 0.2546,
+      "step": 21030
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5343290773307638e-05,
+      "loss": 0.258,
+      "step": 21040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5355336063599133e-05,
+      "loss": 0.4372,
+      "step": 21050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5367381353890628e-05,
+      "loss": 0.1636,
+      "step": 21060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5379426644182123e-05,
+      "loss": 0.2584,
+      "step": 21070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5391471934473622e-05,
+      "loss": 0.1621,
+      "step": 21080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5403517224765117e-05,
+      "loss": 0.2406,
+      "step": 21090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5415562515056612e-05,
+      "loss": 0.2137,
+      "step": 21100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5427607805348107e-05,
+      "loss": 0.3372,
+      "step": 21110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5439653095639603e-05,
+      "loss": 0.1659,
+      "step": 21120
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.54516983859311e-05,
+      "loss": 0.2762,
+      "step": 21130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5463743676222596e-05,
+      "loss": 0.309,
+      "step": 21140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5475788966514098e-05,
+      "loss": 0.2992,
+      "step": 21150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5487834256805593e-05,
+      "loss": 0.3256,
+      "step": 21160
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.549987954709709e-05,
+      "loss": 0.2369,
+      "step": 21170
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5511924837388584e-05,
+      "loss": 0.3958,
+      "step": 21180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.552397012768008e-05,
+      "loss": 0.3664,
+      "step": 21190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5536015417971578e-05,
+      "loss": 0.2497,
+      "step": 21200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5548060708263073e-05,
+      "loss": 0.1988,
+      "step": 21210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5560105998554568e-05,
+      "loss": 0.2674,
+      "step": 21220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5572151288846063e-05,
+      "loss": 0.3218,
+      "step": 21230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.558419657913756e-05,
+      "loss": 0.5888,
+      "step": 21240
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5596241869429054e-05,
+      "loss": 0.4599,
+      "step": 21250
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5608287159720552e-05,
+      "loss": 0.2878,
+      "step": 21260
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5620332450012047e-05,
+      "loss": 0.207,
+      "step": 21270
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5632377740303542e-05,
+      "loss": 0.2566,
+      "step": 21280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5644423030595038e-05,
+      "loss": 0.3024,
+      "step": 21290
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5656468320886533e-05,
+      "loss": 0.4768,
+      "step": 21300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.566851361117803e-05,
+      "loss": 0.1432,
+      "step": 21310
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5680558901469527e-05,
+      "loss": 0.3281,
+      "step": 21320
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5692604191761022e-05,
+      "loss": 0.3313,
+      "step": 21330
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5704649482052517e-05,
+      "loss": 0.2419,
+      "step": 21340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5716694772344012e-05,
+      "loss": 0.4625,
+      "step": 21350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.572874006263551e-05,
+      "loss": 0.2497,
+      "step": 21360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5740785352927006e-05,
+      "loss": 0.2966,
+      "step": 21370
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.57528306432185e-05,
+      "loss": 0.4213,
+      "step": 21380
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5764875933509996e-05,
+      "loss": 0.0535,
+      "step": 21390
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.577692122380149e-05,
+      "loss": 0.3164,
+      "step": 21400
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5788966514092987e-05,
+      "loss": 0.3348,
+      "step": 21410
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.580101180438449e-05,
+      "loss": 0.3176,
+      "step": 21420
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5813057094675984e-05,
+      "loss": 0.254,
+      "step": 21430
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5825102384967482e-05,
+      "loss": 0.2852,
+      "step": 21440
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5837147675258978e-05,
+      "loss": 0.4033,
+      "step": 21450
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5849192965550473e-05,
+      "loss": 0.409,
+      "step": 21460
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.5861238255841968e-05,
+      "loss": 0.7264,
+      "step": 21470
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.5873283546133463e-05,
+      "loss": 0.354,
+      "step": 21480
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.588532883642496e-05,
+      "loss": 0.2336,
+      "step": 21490
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.5897374126716457e-05,
+      "loss": 0.1809,
+      "step": 21500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.5909419417007952e-05,
+      "loss": 0.2313,
+      "step": 21510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.5921464707299447e-05,
+      "loss": 0.3703,
+      "step": 21520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.5933509997590942e-05,
+      "loss": 0.2312,
+      "step": 21530
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.5945555287882438e-05,
+      "loss": 0.1573,
+      "step": 21540
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.5957600578173936e-05,
+      "loss": 0.2092,
+      "step": 21550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.596964586846543e-05,
+      "loss": 0.2513,
+      "step": 21560
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.5981691158756927e-05,
+      "loss": 0.2505,
+      "step": 21570
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.5993736449048422e-05,
+      "loss": 0.3417,
+      "step": 21580
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6005781739339917e-05,
+      "loss": 0.3816,
+      "step": 21590
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6017827029631415e-05,
+      "loss": 0.1399,
+      "step": 21600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.602987231992291e-05,
+      "loss": 0.2792,
+      "step": 21610
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6041917610214406e-05,
+      "loss": 0.2773,
+      "step": 21620
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.60539629005059e-05,
+      "loss": 0.147,
+      "step": 21630
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6066008190797396e-05,
+      "loss": 0.2646,
+      "step": 21640
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6078053481088895e-05,
+      "loss": 0.1882,
+      "step": 21650
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.609009877138039e-05,
+      "loss": 0.1214,
+      "step": 21660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6102144061671885e-05,
+      "loss": 0.4093,
+      "step": 21670
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6114189351963387e-05,
+      "loss": 0.1405,
+      "step": 21680
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6126234642254882e-05,
+      "loss": 0.2164,
+      "step": 21690
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6138279932546377e-05,
+      "loss": 0.4823,
+      "step": 21700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6150325222837873e-05,
+      "loss": 0.2866,
+      "step": 21710
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6162370513129368e-05,
+      "loss": 0.2162,
+      "step": 21720
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6174415803420866e-05,
+      "loss": 0.2746,
+      "step": 21730
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.618646109371236e-05,
+      "loss": 0.2423,
+      "step": 21740
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6198506384003857e-05,
+      "loss": 0.2615,
+      "step": 21750
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6210551674295352e-05,
+      "loss": 0.3858,
+      "step": 21760
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6222596964586847e-05,
+      "loss": 0.2437,
+      "step": 21770
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6234642254878346e-05,
+      "loss": 0.1717,
+      "step": 21780
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.624668754516984e-05,
+      "loss": 0.2712,
+      "step": 21790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.6258732835461336e-05,
+      "loss": 0.1748,
+      "step": 21800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.627077812575283e-05,
+      "loss": 0.296,
+      "step": 21810
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.6282823416044326e-05,
+      "loss": 0.201,
+      "step": 21820
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.6294868706335825e-05,
+      "loss": 0.2119,
+      "step": 21830
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.630691399662732e-05,
+      "loss": 0.4415,
+      "step": 21840
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.6318959286918815e-05,
+      "loss": 0.3328,
+      "step": 21850
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.633100457721031e-05,
+      "loss": 0.156,
+      "step": 21860
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.6343049867501806e-05,
+      "loss": 0.3419,
+      "step": 21870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.63550951577933e-05,
+      "loss": 0.2542,
+      "step": 21880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.63671404480848e-05,
+      "loss": 0.1836,
+      "step": 21890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.6379185738376295e-05,
+      "loss": 0.3249,
+      "step": 21900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.639123102866779e-05,
+      "loss": 0.4331,
+      "step": 21910
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.6403276318959285e-05,
+      "loss": 0.2408,
+      "step": 21920
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.641532160925078e-05,
+      "loss": 0.317,
+      "step": 21930
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.6427366899542282e-05,
+      "loss": 0.3323,
+      "step": 21940
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.6439412189833777e-05,
+      "loss": 0.3771,
+      "step": 21950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.6451457480125276e-05,
+      "loss": 0.3542,
+      "step": 21960
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.646350277041677e-05,
+      "loss": 0.2114,
+      "step": 21970
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.6475548060708266e-05,
+      "loss": 0.2272,
+      "step": 21980
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.648759335099976e-05,
+      "loss": 0.3801,
+      "step": 21990
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.6499638641291257e-05,
+      "loss": 0.3304,
+      "step": 22000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.6511683931582755e-05,
+      "loss": 0.527,
+      "step": 22010
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.652372922187425e-05,
+      "loss": 0.3428,
+      "step": 22020
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.6535774512165746e-05,
+      "loss": 0.2774,
+      "step": 22030
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.654781980245724e-05,
+      "loss": 0.2585,
+      "step": 22040
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.6559865092748736e-05,
+      "loss": 0.3432,
+      "step": 22050
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.657191038304023e-05,
+      "loss": 0.3698,
+      "step": 22060
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.658395567333173e-05,
+      "loss": 0.169,
+      "step": 22070
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.6596000963623225e-05,
+      "loss": 0.3153,
+      "step": 22080
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.660804625391472e-05,
+      "loss": 0.4378,
+      "step": 22090
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.6620091544206215e-05,
+      "loss": 0.3538,
+      "step": 22100
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.663213683449771e-05,
+      "loss": 0.5005,
+      "step": 22110
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.664418212478921e-05,
+      "loss": 0.2574,
+      "step": 22120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.6656227415080704e-05,
+      "loss": 0.4529,
+      "step": 22130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.66682727053722e-05,
+      "loss": 0.3321,
+      "step": 22140
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.6680317995663695e-05,
+      "loss": 0.259,
+      "step": 22150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.669236328595519e-05,
+      "loss": 0.2119,
+      "step": 22160
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.6704408576246685e-05,
+      "loss": 0.4118,
+      "step": 22170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.6716453866538184e-05,
+      "loss": 0.2888,
+      "step": 22180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.672849915682968e-05,
+      "loss": 0.3953,
+      "step": 22190
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.674054444712118e-05,
+      "loss": 0.3275,
+      "step": 22200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.6752589737412676e-05,
+      "loss": 0.2427,
+      "step": 22210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.676463502770417e-05,
+      "loss": 0.1397,
+      "step": 22220
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.6776680317995666e-05,
+      "loss": 0.1517,
+      "step": 22230
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.678872560828716e-05,
+      "loss": 0.0927,
+      "step": 22240
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.680077089857866e-05,
+      "loss": 0.3317,
+      "step": 22250
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.6812816188870155e-05,
+      "loss": 0.3775,
+      "step": 22260
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.682486147916165e-05,
+      "loss": 0.1634,
+      "step": 22270
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.6836906769453146e-05,
+      "loss": 0.1831,
+      "step": 22280
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.684895205974464e-05,
+      "loss": 0.2767,
+      "step": 22290
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.686099735003614e-05,
+      "loss": 0.199,
+      "step": 22300
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.6873042640327635e-05,
+      "loss": 0.2277,
+      "step": 22310
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.688508793061913e-05,
+      "loss": 0.4781,
+      "step": 22320
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.6897133220910625e-05,
+      "loss": 0.1999,
+      "step": 22330
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.690917851120212e-05,
+      "loss": 0.4113,
+      "step": 22340
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.6921223801493615e-05,
+      "loss": 0.1297,
+      "step": 22350
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.6933269091785114e-05,
+      "loss": 0.5356,
+      "step": 22360
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.694531438207661e-05,
+      "loss": 0.1349,
+      "step": 22370
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.6957359672368104e-05,
+      "loss": 0.1879,
+      "step": 22380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.69694049626596e-05,
+      "loss": 0.2681,
+      "step": 22390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.6981450252951095e-05,
+      "loss": 0.2987,
+      "step": 22400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.6993495543242593e-05,
+      "loss": 0.278,
+      "step": 22410
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.700554083353409e-05,
+      "loss": 0.1949,
+      "step": 22420
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.7017586123825584e-05,
+      "loss": 0.2063,
+      "step": 22430
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.702963141411708e-05,
+      "loss": 0.5082,
+      "step": 22440
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.7041676704408574e-05,
+      "loss": 0.5487,
+      "step": 22450
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.7053721994700073e-05,
+      "loss": 0.2249,
+      "step": 22460
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.706576728499157e-05,
+      "loss": 0.2472,
+      "step": 22470
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.707781257528307e-05,
+      "loss": 0.2717,
+      "step": 22480
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.7089857865574565e-05,
+      "loss": 0.2413,
+      "step": 22490
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.710190315586606e-05,
+      "loss": 0.3517,
+      "step": 22500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.7113948446157555e-05,
+      "loss": 0.1782,
+      "step": 22510
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.712599373644905e-05,
+      "loss": 0.3212,
+      "step": 22520
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.7138039026740546e-05,
+      "loss": 0.4954,
+      "step": 22530
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.7150084317032044e-05,
+      "loss": 0.2772,
+      "step": 22540
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.716212960732354e-05,
+      "loss": 0.2065,
+      "step": 22550
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.7174174897615035e-05,
+      "loss": 0.3337,
+      "step": 22560
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.718622018790653e-05,
+      "loss": 0.2453,
+      "step": 22570
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.7198265478198025e-05,
+      "loss": 0.1573,
+      "step": 22580
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.7210310768489523e-05,
+      "loss": 0.138,
+      "step": 22590
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.722235605878102e-05,
+      "loss": 0.5015,
+      "step": 22600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.7234401349072514e-05,
+      "loss": 0.4585,
+      "step": 22610
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.724644663936401e-05,
+      "loss": 0.3155,
+      "step": 22620
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7258491929655504e-05,
+      "loss": 0.169,
+      "step": 22630
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7270537219947e-05,
+      "loss": 0.3197,
+      "step": 22640
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7282582510238498e-05,
+      "loss": 0.216,
+      "step": 22650
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7294627800529993e-05,
+      "loss": 0.3279,
+      "step": 22660
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.730667309082149e-05,
+      "loss": 0.2055,
+      "step": 22670
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7318718381112984e-05,
+      "loss": 0.2216,
+      "step": 22680
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.733076367140448e-05,
+      "loss": 0.332,
+      "step": 22690
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7342808961695977e-05,
+      "loss": 0.5532,
+      "step": 22700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7354854251987472e-05,
+      "loss": 0.4094,
+      "step": 22710
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7366899542278968e-05,
+      "loss": 0.1958,
+      "step": 22720
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.737894483257047e-05,
+      "loss": 0.2323,
+      "step": 22730
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7390990122861965e-05,
+      "loss": 0.4287,
+      "step": 22740
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.740303541315346e-05,
+      "loss": 0.4302,
+      "step": 22750
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7415080703444955e-05,
+      "loss": 0.2446,
+      "step": 22760
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7427125993736454e-05,
+      "loss": 0.4307,
+      "step": 22770
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.743917128402795e-05,
+      "loss": 0.2247,
+      "step": 22780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7451216574319444e-05,
+      "loss": 0.2067,
+      "step": 22790
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.746326186461094e-05,
+      "loss": 0.3797,
+      "step": 22800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7475307154902434e-05,
+      "loss": 0.2439,
+      "step": 22810
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.748735244519393e-05,
+      "loss": 0.2485,
+      "step": 22820
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7499397735485428e-05,
+      "loss": 0.4136,
+      "step": 22830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7511443025776923e-05,
+      "loss": 0.1345,
+      "step": 22840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.752348831606842e-05,
+      "loss": 0.3076,
+      "step": 22850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7535533606359914e-05,
+      "loss": 0.3709,
+      "step": 22860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.754757889665141e-05,
+      "loss": 0.2915,
+      "step": 22870
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7559624186942908e-05,
+      "loss": 0.226,
+      "step": 22880
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7571669477234403e-05,
+      "loss": 0.5767,
+      "step": 22890
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7583714767525898e-05,
+      "loss": 0.1771,
+      "step": 22900
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7595760057817393e-05,
+      "loss": 0.2592,
+      "step": 22910
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7607805348108888e-05,
+      "loss": 0.3167,
+      "step": 22920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7619850638400387e-05,
+      "loss": 0.3876,
+      "step": 22930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7631895928691882e-05,
+      "loss": 0.2299,
+      "step": 22940
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7643941218983377e-05,
+      "loss": 0.2058,
+      "step": 22950
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.7655986509274872e-05,
+      "loss": 0.3383,
+      "step": 22960
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.7668031799566368e-05,
+      "loss": 0.0507,
+      "step": 22970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.7680077089857863e-05,
+      "loss": 0.389,
+      "step": 22980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.7692122380149365e-05,
+      "loss": 0.3891,
+      "step": 22990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.770416767044086e-05,
+      "loss": 0.1275,
+      "step": 23000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.771621296073236e-05,
+      "loss": 0.2209,
+      "step": 23010
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.7728258251023854e-05,
+      "loss": 0.4105,
+      "step": 23020
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.774030354131535e-05,
+      "loss": 0.1448,
+      "step": 23030
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7752348831606844e-05,
+      "loss": 0.3497,
+      "step": 23040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.776439412189834e-05,
+      "loss": 0.4104,
+      "step": 23050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7776439412189838e-05,
+      "loss": 0.3292,
+      "step": 23060
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7788484702481333e-05,
+      "loss": 0.4962,
+      "step": 23070
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7800529992772828e-05,
+      "loss": 0.2761,
+      "step": 23080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7812575283064323e-05,
+      "loss": 0.2209,
+      "step": 23090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.782462057335582e-05,
+      "loss": 0.4074,
+      "step": 23100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7836665863647314e-05,
+      "loss": 0.368,
+      "step": 23110
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7848711153938812e-05,
+      "loss": 0.1944,
+      "step": 23120
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7860756444230308e-05,
+      "loss": 0.2146,
+      "step": 23130
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7872801734521803e-05,
+      "loss": 0.3026,
+      "step": 23140
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7884847024813298e-05,
+      "loss": 0.2399,
+      "step": 23150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7896892315104793e-05,
+      "loss": 0.2011,
+      "step": 23160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.790893760539629e-05,
+      "loss": 0.3207,
+      "step": 23170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7920982895687787e-05,
+      "loss": 0.3043,
+      "step": 23180
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7933028185979282e-05,
+      "loss": 0.1911,
+      "step": 23190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7945073476270777e-05,
+      "loss": 0.4578,
+      "step": 23200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.7957118766562272e-05,
+      "loss": 0.2688,
+      "step": 23210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.796916405685377e-05,
+      "loss": 0.197,
+      "step": 23220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.7981209347145266e-05,
+      "loss": 0.2193,
+      "step": 23230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.799325463743676e-05,
+      "loss": 0.2328,
+      "step": 23240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.8005299927728257e-05,
+      "loss": 0.2603,
+      "step": 23250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.801734521801976e-05,
+      "loss": 0.4764,
+      "step": 23260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.8029390508311254e-05,
+      "loss": 0.241,
+      "step": 23270
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.804143579860275e-05,
+      "loss": 0.2272,
+      "step": 23280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.8053481088894244e-05,
+      "loss": 0.3157,
+      "step": 23290
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.8065526379185743e-05,
+      "loss": 0.1755,
+      "step": 23300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.8077571669477238e-05,
+      "loss": 0.2747,
+      "step": 23310
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.8089616959768733e-05,
+      "loss": 0.3466,
+      "step": 23320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.8101662250060228e-05,
+      "loss": 0.3747,
+      "step": 23330
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.8113707540351723e-05,
+      "loss": 0.4825,
+      "step": 23340
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.8125752830643222e-05,
+      "loss": 0.2095,
+      "step": 23350
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.8137798120934717e-05,
+      "loss": 0.2131,
+      "step": 23360
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.8149843411226212e-05,
+      "loss": 0.444,
+      "step": 23370
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.8161888701517707e-05,
+      "loss": 0.248,
+      "step": 23380
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.8173933991809203e-05,
+      "loss": 0.2881,
+      "step": 23390
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.81859792821007e-05,
+      "loss": 0.3376,
+      "step": 23400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.8198024572392196e-05,
+      "loss": 0.3045,
+      "step": 23410
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.821006986268369e-05,
+      "loss": 0.2366,
+      "step": 23420
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.8222115152975187e-05,
+      "loss": 0.3722,
+      "step": 23430
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.8234160443266682e-05,
+      "loss": 0.1959,
+      "step": 23440
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.8246205733558177e-05,
+      "loss": 0.3344,
+      "step": 23450
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8258251023849676e-05,
+      "loss": 0.4158,
+      "step": 23460
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.827029631414117e-05,
+      "loss": 0.1405,
+      "step": 23470
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8282341604432666e-05,
+      "loss": 0.1122,
+      "step": 23480
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.829438689472416e-05,
+      "loss": 0.5645,
+      "step": 23490
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8306432185015656e-05,
+      "loss": 0.221,
+      "step": 23500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8318477475307155e-05,
+      "loss": 0.3729,
+      "step": 23510
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8330522765598654e-05,
+      "loss": 0.2177,
+      "step": 23520
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8342568055890152e-05,
+      "loss": 0.3059,
+      "step": 23530
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8354613346181647e-05,
+      "loss": 0.3359,
+      "step": 23540
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8366658636473143e-05,
+      "loss": 0.3591,
+      "step": 23550
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8378703926764638e-05,
+      "loss": 0.1704,
+      "step": 23560
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8390749217056133e-05,
+      "loss": 0.1867,
+      "step": 23570
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.840279450734763e-05,
+      "loss": 0.3245,
+      "step": 23580
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8414839797639127e-05,
+      "loss": 0.4672,
+      "step": 23590
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8426885087930622e-05,
+      "loss": 0.4372,
+      "step": 23600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8438930378222117e-05,
+      "loss": 0.3602,
+      "step": 23610
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8450975668513612e-05,
+      "loss": 0.2653,
+      "step": 23620
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8463020958805107e-05,
+      "loss": 0.323,
+      "step": 23630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8475066249096606e-05,
+      "loss": 0.3061,
+      "step": 23640
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.84871115393881e-05,
+      "loss": 0.3244,
+      "step": 23650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8499156829679596e-05,
+      "loss": 0.3713,
+      "step": 23660
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.851120211997109e-05,
+      "loss": 0.4599,
+      "step": 23670
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8523247410262587e-05,
+      "loss": 0.3721,
+      "step": 23680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8535292700554085e-05,
+      "loss": 0.3501,
+      "step": 23690
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.854733799084558e-05,
+      "loss": 0.2259,
+      "step": 23700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.8559383281137076e-05,
+      "loss": 0.2451,
+      "step": 23710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.2708,
+      "step": 23720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.8583473861720066e-05,
+      "loss": 0.1206,
+      "step": 23730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.859551915201156e-05,
+      "loss": 0.3059,
+      "step": 23740
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.860756444230306e-05,
+      "loss": 0.3561,
+      "step": 23750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.8619609732594555e-05,
+      "loss": 0.4388,
+      "step": 23760
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.863165502288605e-05,
+      "loss": 0.3176,
+      "step": 23770
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.8643700313177552e-05,
+      "loss": 0.3199,
+      "step": 23780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.8655745603469047e-05,
+      "loss": 0.2251,
+      "step": 23790
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.8667790893760543e-05,
+      "loss": 0.2392,
+      "step": 23800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.8679836184052038e-05,
+      "loss": 0.254,
+      "step": 23810
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.8691881474343536e-05,
+      "loss": 0.157,
+      "step": 23820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.870392676463503e-05,
+      "loss": 0.2289,
+      "step": 23830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.8715972054926527e-05,
+      "loss": 0.154,
+      "step": 23840
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.8728017345218022e-05,
+      "loss": 0.2883,
+      "step": 23850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.8740062635509517e-05,
+      "loss": 0.3291,
+      "step": 23860
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.8752107925801016e-05,
+      "loss": 0.3571,
+      "step": 23870
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.876415321609251e-05,
+      "loss": 0.1982,
+      "step": 23880
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.8776198506384006e-05,
+      "loss": 0.3095,
+      "step": 23890
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.87882437966755e-05,
+      "loss": 0.2637,
+      "step": 23900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.8800289086966996e-05,
+      "loss": 0.5958,
+      "step": 23910
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.881233437725849e-05,
+      "loss": 0.1342,
+      "step": 23920
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.882437966754999e-05,
+      "loss": 0.1502,
+      "step": 23930
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.8836424957841485e-05,
+      "loss": 0.341,
+      "step": 23940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.884847024813298e-05,
+      "loss": 0.3126,
+      "step": 23950
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.8860515538424476e-05,
+      "loss": 0.5042,
+      "step": 23960
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.887256082871597e-05,
+      "loss": 0.318,
+      "step": 23970
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.888460611900747e-05,
+      "loss": 0.3203,
+      "step": 23980
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.8896651409298965e-05,
+      "loss": 0.2613,
+      "step": 23990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.890869669959046e-05,
+      "loss": 0.4611,
+      "step": 24000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.8920741989881955e-05,
+      "loss": 0.3859,
+      "step": 24010
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.893278728017345e-05,
+      "loss": 0.2085,
+      "step": 24020
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.894483257046495e-05,
+      "loss": 0.2968,
+      "step": 24030
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.8956877860756447e-05,
+      "loss": 0.2486,
+      "step": 24040
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.8968923151047946e-05,
+      "loss": 0.2532,
+      "step": 24050
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.898096844133944e-05,
+      "loss": 0.363,
+      "step": 24060
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.8993013731630936e-05,
+      "loss": 0.2028,
+      "step": 24070
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.900505902192243e-05,
+      "loss": 0.159,
+      "step": 24080
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.9017104312213927e-05,
+      "loss": 0.156,
+      "step": 24090
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.9029149602505422e-05,
+      "loss": 0.2031,
+      "step": 24100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.904119489279692e-05,
+      "loss": 0.3288,
+      "step": 24110
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.9053240183088416e-05,
+      "loss": 0.3644,
+      "step": 24120
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.906528547337991e-05,
+      "loss": 0.1941,
+      "step": 24130
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.9077330763671406e-05,
+      "loss": 0.2012,
+      "step": 24140
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.90893760539629e-05,
+      "loss": 0.319,
+      "step": 24150
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.91014213442544e-05,
+      "loss": 0.3506,
+      "step": 24160
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.9113466634545895e-05,
+      "loss": 0.3147,
+      "step": 24170
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.912551192483739e-05,
+      "loss": 0.4684,
+      "step": 24180
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.9137557215128885e-05,
+      "loss": 0.0931,
+      "step": 24190
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.914960250542038e-05,
+      "loss": 0.3831,
+      "step": 24200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9161647795711876e-05,
+      "loss": 0.1619,
+      "step": 24210
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9173693086003374e-05,
+      "loss": 0.3145,
+      "step": 24220
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.918573837629487e-05,
+      "loss": 0.3001,
+      "step": 24230
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9197783666586365e-05,
+      "loss": 0.1481,
+      "step": 24240
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.920982895687786e-05,
+      "loss": 0.2407,
+      "step": 24250
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9221874247169355e-05,
+      "loss": 0.4357,
+      "step": 24260
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9233919537460853e-05,
+      "loss": 0.345,
+      "step": 24270
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.924596482775235e-05,
+      "loss": 0.1391,
+      "step": 24280
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9258010118043844e-05,
+      "loss": 0.1861,
+      "step": 24290
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.927005540833534e-05,
+      "loss": 0.2042,
+      "step": 24300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.928210069862684e-05,
+      "loss": 0.4824,
+      "step": 24310
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9294145988918336e-05,
+      "loss": 0.3685,
+      "step": 24320
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.930619127920983e-05,
+      "loss": 0.1789,
+      "step": 24330
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.931823656950133e-05,
+      "loss": 0.1879,
+      "step": 24340
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9330281859792825e-05,
+      "loss": 0.1064,
+      "step": 24350
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.934232715008432e-05,
+      "loss": 0.0305,
+      "step": 24360
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9354372440375815e-05,
+      "loss": 0.2212,
+      "step": 24370
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.936641773066731e-05,
+      "loss": 0.2533,
+      "step": 24380
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9378463020958806e-05,
+      "loss": 0.3723,
+      "step": 24390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9390508311250304e-05,
+      "loss": 0.3442,
+      "step": 24400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.94025536015418e-05,
+      "loss": 0.4811,
+      "step": 24410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9414598891833295e-05,
+      "loss": 0.2306,
+      "step": 24420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.942664418212479e-05,
+      "loss": 0.2211,
+      "step": 24430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9438689472416285e-05,
+      "loss": 0.169,
+      "step": 24440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9450734762707784e-05,
+      "loss": 0.1488,
+      "step": 24450
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.946278005299928e-05,
+      "loss": 0.0416,
+      "step": 24460
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9474825343290774e-05,
+      "loss": 0.2125,
+      "step": 24470
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.948687063358227e-05,
+      "loss": 0.2843,
+      "step": 24480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9498915923873764e-05,
+      "loss": 0.1981,
+      "step": 24490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9510961214165263e-05,
+      "loss": 0.2135,
+      "step": 24500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9523006504456758e-05,
+      "loss": 0.3554,
+      "step": 24510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9535051794748253e-05,
+      "loss": 0.2508,
+      "step": 24520
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.954709708503975e-05,
+      "loss": 0.1298,
+      "step": 24530
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9559142375331244e-05,
+      "loss": 0.1082,
+      "step": 24540
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.957118766562274e-05,
+      "loss": 0.4061,
+      "step": 24550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9583232955914238e-05,
+      "loss": 0.3149,
+      "step": 24560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9595278246205736e-05,
+      "loss": 0.2459,
+      "step": 24570
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9607323536497235e-05,
+      "loss": 0.2732,
+      "step": 24580
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.961936882678873e-05,
+      "loss": 0.2766,
+      "step": 24590
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9631414117080225e-05,
+      "loss": 0.4046,
+      "step": 24600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.964345940737172e-05,
+      "loss": 0.4159,
+      "step": 24610
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9655504697663215e-05,
+      "loss": 0.2421,
+      "step": 24620
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9667549987954714e-05,
+      "loss": 0.1116,
+      "step": 24630
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.967959527824621e-05,
+      "loss": 0.2036,
+      "step": 24640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9691640568537704e-05,
+      "loss": 0.1473,
+      "step": 24650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.97036858588292e-05,
+      "loss": 0.2274,
+      "step": 24660
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9715731149120695e-05,
+      "loss": 0.3042,
+      "step": 24670
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.972777643941219e-05,
+      "loss": 0.2341,
+      "step": 24680
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.973982172970369e-05,
+      "loss": 0.2255,
+      "step": 24690
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9751867019995184e-05,
+      "loss": 0.2568,
+      "step": 24700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.976391231028668e-05,
+      "loss": 0.2563,
+      "step": 24710
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9775957600578174e-05,
+      "loss": 0.215,
+      "step": 24720
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.978800289086967e-05,
+      "loss": 0.3708,
+      "step": 24730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9800048181161168e-05,
+      "loss": 0.1765,
+      "step": 24740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9812093471452663e-05,
+      "loss": 0.1991,
+      "step": 24750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9824138761744158e-05,
+      "loss": 0.398,
+      "step": 24760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9836184052035653e-05,
+      "loss": 0.2089,
+      "step": 24770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.984822934232715e-05,
+      "loss": 0.3183,
+      "step": 24780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9860274632618647e-05,
+      "loss": 0.2312,
+      "step": 24790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9872319922910142e-05,
+      "loss": 0.2768,
+      "step": 24800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9884365213201637e-05,
+      "loss": 0.236,
+      "step": 24810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9896410503493133e-05,
+      "loss": 0.1127,
+      "step": 24820
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9908455793784635e-05,
+      "loss": 0.2971,
+      "step": 24830
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.992050108407613e-05,
+      "loss": 0.1909,
+      "step": 24840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9932546374367625e-05,
+      "loss": 0.3927,
+      "step": 24850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.994459166465912e-05,
+      "loss": 0.2993,
+      "step": 24860
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.995663695495062e-05,
+      "loss": 0.1727,
+      "step": 24870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.9968682245242114e-05,
+      "loss": 0.336,
+      "step": 24880
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.998072753553361e-05,
+      "loss": 0.3784,
+      "step": 24890
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.9992772825825104e-05,
+      "loss": 0.5678,
+      "step": 24900
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": {
+        "accuracy": 0.9118940078289671
+      },
+      "eval_f1": {
+        "f1": 0.9116597109413729
+      },
+      "eval_loss": 0.3114425837993622,
+      "eval_precision": {
+        "precision": 0.9165908158739848
+      },
+      "eval_recall": {
+        "recall": 0.9116030141797212
+      },
+      "eval_runtime": 457.3219,
+      "eval_samples_per_second": 36.309,
+      "eval_steps_per_second": 18.156,
+      "step": 24907
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.00048181161166e-05,
+      "loss": 0.2599,
+      "step": 24910
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0016863406408098e-05,
+      "loss": 0.172,
+      "step": 24920
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0028908696699593e-05,
+      "loss": 0.2137,
+      "step": 24930
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.004095398699109e-05,
+      "loss": 0.4116,
+      "step": 24940
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0052999277282584e-05,
+      "loss": 0.3722,
+      "step": 24950
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.006504456757408e-05,
+      "loss": 0.1935,
+      "step": 24960
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0077089857865577e-05,
+      "loss": 0.2207,
+      "step": 24970
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0089135148157073e-05,
+      "loss": 0.4092,
+      "step": 24980
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0101180438448568e-05,
+      "loss": 0.3625,
+      "step": 24990
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0113225728740063e-05,
+      "loss": 0.2725,
+      "step": 25000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0125271019031558e-05,
+      "loss": 0.1696,
+      "step": 25010
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0137316309323053e-05,
+      "loss": 0.1985,
+      "step": 25020
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0149361599614552e-05,
+      "loss": 0.2278,
+      "step": 25030
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0161406889906047e-05,
+      "loss": 0.1984,
+      "step": 25040
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0173452180197542e-05,
+      "loss": 0.4152,
+      "step": 25050
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0185497470489037e-05,
+      "loss": 0.3541,
+      "step": 25060
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0197542760780533e-05,
+      "loss": 0.1387,
+      "step": 25070
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.020958805107203e-05,
+      "loss": 0.2596,
+      "step": 25080
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.022163334136353e-05,
+      "loss": 0.1854,
+      "step": 25090
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.023367863165503e-05,
+      "loss": 0.2492,
+      "step": 25100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0245723921946524e-05,
+      "loss": 0.2725,
+      "step": 25110
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.025776921223802e-05,
+      "loss": 0.2586,
+      "step": 25120
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.0269814502529514e-05,
+      "loss": 0.2255,
+      "step": 25130
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.028185979282101e-05,
+      "loss": 0.1305,
+      "step": 25140
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.0293905083112504e-05,
+      "loss": 0.3667,
+      "step": 25150
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.0305950373404003e-05,
+      "loss": 0.1553,
+      "step": 25160
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.0317995663695498e-05,
+      "loss": 0.1278,
+      "step": 25170
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.0330040953986993e-05,
+      "loss": 0.2466,
+      "step": 25180
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.034208624427849e-05,
+      "loss": 0.2929,
+      "step": 25190
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.0354131534569984e-05,
+      "loss": 0.2995,
+      "step": 25200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.0366176824861482e-05,
+      "loss": 0.2917,
+      "step": 25210
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.0378222115152977e-05,
+      "loss": 0.1022,
+      "step": 25220
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.0390267405444473e-05,
+      "loss": 0.3321,
+      "step": 25230
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.0402312695735968e-05,
+      "loss": 0.36,
+      "step": 25240
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.0414357986027463e-05,
+      "loss": 0.1658,
+      "step": 25250
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.042640327631896e-05,
+      "loss": 0.2034,
+      "step": 25260
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.0438448566610457e-05,
+      "loss": 0.1107,
+      "step": 25270
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.0450493856901952e-05,
+      "loss": 0.3025,
+      "step": 25280
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.0462539147193447e-05,
+      "loss": 0.2878,
+      "step": 25290
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.0474584437484942e-05,
+      "loss": 0.4318,
+      "step": 25300
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.0486629727776437e-05,
+      "loss": 0.351,
+      "step": 25310
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.0498675018067936e-05,
+      "loss": 0.266,
+      "step": 25320
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.051072030835943e-05,
+      "loss": 0.4601,
+      "step": 25330
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.0522765598650926e-05,
+      "loss": 0.2249,
+      "step": 25340
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.053481088894242e-05,
+      "loss": 0.2655,
+      "step": 25350
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.0546856179233923e-05,
+      "loss": 0.1564,
+      "step": 25360
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.055890146952542e-05,
+      "loss": 0.22,
+      "step": 25370
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.0570946759816914e-05,
+      "loss": 0.2266,
+      "step": 25380
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.058299205010841e-05,
+      "loss": 0.2388,
+      "step": 25390
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.0595037340399904e-05,
+      "loss": 0.5174,
+      "step": 25400
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.06070826306914e-05,
+      "loss": 0.2869,
+      "step": 25410
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.06191279209829e-05,
+      "loss": 0.1731,
+      "step": 25420
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.0631173211274397e-05,
+      "loss": 0.2053,
+      "step": 25430
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.064321850156589e-05,
+      "loss": 0.3669,
+      "step": 25440
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.065526379185739e-05,
+      "loss": 0.3826,
+      "step": 25450
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.066730908214888e-05,
+      "loss": 0.2083,
+      "step": 25460
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.067935437244038e-05,
+      "loss": 0.3435,
+      "step": 25470
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.069139966273187e-05,
+      "loss": 0.2688,
+      "step": 25480
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.070344495302337e-05,
+      "loss": 0.3383,
+      "step": 25490
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.071549024331486e-05,
+      "loss": 0.3739,
+      "step": 25500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.072753553360636e-05,
+      "loss": 0.2326,
+      "step": 25510
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.073958082389786e-05,
+      "loss": 0.2474,
+      "step": 25520
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.0751626114189355e-05,
+      "loss": 0.2917,
+      "step": 25530
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.076367140448085e-05,
+      "loss": 0.2756,
+      "step": 25540
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.0775716694772346e-05,
+      "loss": 0.2438,
+      "step": 25550
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.078776198506384e-05,
+      "loss": 0.168,
+      "step": 25560
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.0799807275355336e-05,
+      "loss": 0.2376,
+      "step": 25570
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.081185256564683e-05,
+      "loss": 0.2065,
+      "step": 25580
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.0823897855938326e-05,
+      "loss": 0.4056,
+      "step": 25590
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.083594314622982e-05,
+      "loss": 0.2064,
+      "step": 25600
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.084798843652132e-05,
+      "loss": 0.1649,
+      "step": 25610
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.086003372681282e-05,
+      "loss": 0.2391,
+      "step": 25620
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.0872079017104314e-05,
+      "loss": 0.1636,
+      "step": 25630
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.088412430739581e-05,
+      "loss": 0.4071,
+      "step": 25640
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.089616959768731e-05,
+      "loss": 0.0904,
+      "step": 25650
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.0908214887978806e-05,
+      "loss": 0.2451,
+      "step": 25660
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.09202601782703e-05,
+      "loss": 0.3351,
+      "step": 25670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.0932305468561797e-05,
+      "loss": 0.3635,
+      "step": 25680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.094435075885329e-05,
+      "loss": 0.3487,
+      "step": 25690
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.095639604914479e-05,
+      "loss": 0.4103,
+      "step": 25700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.096844133943628e-05,
+      "loss": 0.2533,
+      "step": 25710
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.098048662972778e-05,
+      "loss": 0.2963,
+      "step": 25720
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.099253192001927e-05,
+      "loss": 0.239,
+      "step": 25730
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.100457721031077e-05,
+      "loss": 0.1837,
+      "step": 25740
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.101662250060226e-05,
+      "loss": 0.3309,
+      "step": 25750
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.1028667790893765e-05,
+      "loss": 0.4147,
+      "step": 25760
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.104071308118526e-05,
+      "loss": 0.274,
+      "step": 25770
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1052758371476755e-05,
+      "loss": 0.3612,
+      "step": 25780
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.106480366176825e-05,
+      "loss": 0.2133,
+      "step": 25790
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1076848952059746e-05,
+      "loss": 0.0694,
+      "step": 25800
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.108889424235124e-05,
+      "loss": 0.3588,
+      "step": 25810
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1100939532642736e-05,
+      "loss": 0.2257,
+      "step": 25820
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.111298482293423e-05,
+      "loss": 0.2314,
+      "step": 25830
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.1125030113225726e-05,
+      "loss": 0.2875,
+      "step": 25840
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.113707540351722e-05,
+      "loss": 0.1578,
+      "step": 25850
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.114912069380872e-05,
+      "loss": 0.2069,
+      "step": 25860
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.116116598410022e-05,
+      "loss": 0.375,
+      "step": 25870
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.1173211274391714e-05,
+      "loss": 0.1165,
+      "step": 25880
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.1185256564683216e-05,
+      "loss": 0.2349,
+      "step": 25890
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.119730185497471e-05,
+      "loss": 0.2425,
+      "step": 25900
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.1209347145266206e-05,
+      "loss": 0.4854,
+      "step": 25910
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.12213924355577e-05,
+      "loss": 0.272,
+      "step": 25920
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.1233437725849196e-05,
+      "loss": 0.0181,
+      "step": 25930
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.124548301614069e-05,
+      "loss": 0.3008,
+      "step": 25940
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.125752830643219e-05,
+      "loss": 0.1838,
+      "step": 25950
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.126957359672368e-05,
+      "loss": 0.0766,
+      "step": 25960
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.128161888701518e-05,
+      "loss": 0.1983,
+      "step": 25970
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.129366417730667e-05,
+      "loss": 0.4983,
+      "step": 25980
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.1305709467598174e-05,
+      "loss": 0.4632,
+      "step": 25990
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.131775475788967e-05,
+      "loss": 0.3192,
+      "step": 26000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.1329800048181165e-05,
+      "loss": 0.3183,
+      "step": 26010
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.134184533847266e-05,
+      "loss": 0.3288,
+      "step": 26020
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.1353890628764155e-05,
+      "loss": 0.084,
+      "step": 26030
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.136593591905565e-05,
+      "loss": 0.3194,
+      "step": 26040
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.1377981209347145e-05,
+      "loss": 0.2384,
+      "step": 26050
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.139002649963864e-05,
+      "loss": 0.2473,
+      "step": 26060
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.1402071789930136e-05,
+      "loss": 0.1829,
+      "step": 26070
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.141411708022163e-05,
+      "loss": 0.2635,
+      "step": 26080
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.1426162370513126e-05,
+      "loss": 0.3213,
+      "step": 26090
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.143820766080463e-05,
+      "loss": 0.3113,
+      "step": 26100
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.145025295109612e-05,
+      "loss": 0.459,
+      "step": 26110
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.146229824138762e-05,
+      "loss": 0.2854,
+      "step": 26120
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.1474343531679114e-05,
+      "loss": 0.4353,
+      "step": 26130
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.148638882197061e-05,
+      "loss": 0.2821,
+      "step": 26140
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.149843411226211e-05,
+      "loss": 0.3122,
+      "step": 26150
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.1510479402553606e-05,
+      "loss": 0.2357,
+      "step": 26160
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.15225246928451e-05,
+      "loss": 0.4204,
+      "step": 26170
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.1534569983136596e-05,
+      "loss": 0.0572,
+      "step": 26180
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.154661527342809e-05,
+      "loss": 0.1571,
+      "step": 26190
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.155866056371959e-05,
+      "loss": 0.0867,
+      "step": 26200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.157070585401108e-05,
+      "loss": 0.3343,
+      "step": 26210
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.158275114430258e-05,
+      "loss": 0.3134,
+      "step": 26220
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.159479643459408e-05,
+      "loss": 0.1975,
+      "step": 26230
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.1606841724885574e-05,
+      "loss": 0.1028,
+      "step": 26240
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.161888701517707e-05,
+      "loss": 0.2928,
+      "step": 26250
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.1630932305468565e-05,
+      "loss": 0.3859,
+      "step": 26260
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.164297759576006e-05,
+      "loss": 0.3256,
+      "step": 26270
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1655022886051555e-05,
+      "loss": 0.2488,
+      "step": 26280
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.166706817634305e-05,
+      "loss": 0.1517,
+      "step": 26290
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1679113466634545e-05,
+      "loss": 0.1971,
+      "step": 26300
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.169115875692604e-05,
+      "loss": 0.2443,
+      "step": 26310
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1703204047217536e-05,
+      "loss": 0.3108,
+      "step": 26320
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.171524933750903e-05,
+      "loss": 0.3149,
+      "step": 26330
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.172729462780053e-05,
+      "loss": 0.2452,
+      "step": 26340
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.173933991809203e-05,
+      "loss": 0.2322,
+      "step": 26350
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.175138520838352e-05,
+      "loss": 0.2111,
+      "step": 26360
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.176343049867502e-05,
+      "loss": 0.155,
+      "step": 26370
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1775475788966514e-05,
+      "loss": 0.2402,
+      "step": 26380
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.178752107925801e-05,
+      "loss": 0.3156,
+      "step": 26390
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1799566369549504e-05,
+      "loss": 0.1624,
+      "step": 26400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1811611659841006e-05,
+      "loss": 0.2086,
+      "step": 26410
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.18236569501325e-05,
+      "loss": 0.134,
+      "step": 26420
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1835702240423996e-05,
+      "loss": 0.1561,
+      "step": 26430
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.184774753071549e-05,
+      "loss": 0.1846,
+      "step": 26440
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.185979282100699e-05,
+      "loss": 0.4856,
+      "step": 26450
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.187183811129849e-05,
+      "loss": 0.2791,
+      "step": 26460
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.1883883401589984e-05,
+      "loss": 0.3095,
+      "step": 26470
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.189592869188148e-05,
+      "loss": 0.2822,
+      "step": 26480
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.1907973982172974e-05,
+      "loss": 0.1731,
+      "step": 26490
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.192001927246447e-05,
+      "loss": 0.2702,
+      "step": 26500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.1932064562755965e-05,
+      "loss": 0.1237,
+      "step": 26510
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.194410985304746e-05,
+      "loss": 0.2848,
+      "step": 26520
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1956155143338955e-05,
+      "loss": 0.2309,
+      "step": 26530
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.196820043363045e-05,
+      "loss": 0.4189,
+      "step": 26540
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1980245723921945e-05,
+      "loss": 0.323,
+      "step": 26550
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.199229101421344e-05,
+      "loss": 0.3852,
+      "step": 26560
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.200433630450494e-05,
+      "loss": 0.0965,
+      "step": 26570
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.201638159479644e-05,
+      "loss": 0.15,
+      "step": 26580
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.202842688508793e-05,
+      "loss": 0.2596,
+      "step": 26590
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.204047217537943e-05,
+      "loss": 0.2513,
+      "step": 26600
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.205251746567092e-05,
+      "loss": 0.4017,
+      "step": 26610
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.206456275596242e-05,
+      "loss": 0.2846,
+      "step": 26620
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.2076608046253914e-05,
+      "loss": 0.1983,
+      "step": 26630
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.208865333654541e-05,
+      "loss": 0.1891,
+      "step": 26640
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.2100698626836904e-05,
+      "loss": 0.1694,
+      "step": 26650
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.21127439171284e-05,
+      "loss": 0.2307,
+      "step": 26660
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.21247892074199e-05,
+      "loss": 0.3475,
+      "step": 26670
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.2136834497711396e-05,
+      "loss": 0.2109,
+      "step": 26680
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.214887978800289e-05,
+      "loss": 0.1501,
+      "step": 26690
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.2160925078294393e-05,
+      "loss": 0.3511,
+      "step": 26700
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.217297036858589e-05,
+      "loss": 0.1537,
+      "step": 26710
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.2185015658877384e-05,
+      "loss": 0.4354,
+      "step": 26720
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.219706094916888e-05,
+      "loss": 0.2124,
+      "step": 26730
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.2209106239460374e-05,
+      "loss": 0.3478,
+      "step": 26740
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.222115152975187e-05,
+      "loss": 0.3381,
+      "step": 26750
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.2233196820043365e-05,
+      "loss": 0.0724,
+      "step": 26760
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.224524211033486e-05,
+      "loss": 0.317,
+      "step": 26770
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.2257287400626355e-05,
+      "loss": 0.5549,
+      "step": 26780
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.226933269091785e-05,
+      "loss": 0.2279,
+      "step": 26790
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.2281377981209345e-05,
+      "loss": 0.2565,
+      "step": 26800
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.229342327150085e-05,
+      "loss": 0.1639,
+      "step": 26810
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.230546856179234e-05,
+      "loss": 0.2193,
+      "step": 26820
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.231751385208384e-05,
+      "loss": 0.5281,
+      "step": 26830
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.232955914237533e-05,
+      "loss": 0.2905,
+      "step": 26840
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.234160443266683e-05,
+      "loss": 0.2982,
+      "step": 26850
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.235364972295832e-05,
+      "loss": 0.4296,
+      "step": 26860
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.236569501324982e-05,
+      "loss": 0.2324,
+      "step": 26870
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.2377740303541314e-05,
+      "loss": 0.2498,
+      "step": 26880
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.238978559383281e-05,
+      "loss": 0.4992,
+      "step": 26890
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.2401830884124304e-05,
+      "loss": 0.2781,
+      "step": 26900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.2413876174415806e-05,
+      "loss": 0.6001,
+      "step": 26910
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.24259214647073e-05,
+      "loss": 0.1741,
+      "step": 26920
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.24379667549988e-05,
+      "loss": 0.3282,
+      "step": 26930
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.24500120452903e-05,
+      "loss": 0.2494,
+      "step": 26940
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.2462057335581793e-05,
+      "loss": 0.3007,
+      "step": 26950
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.247410262587329e-05,
+      "loss": 0.2339,
+      "step": 26960
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.2486147916164784e-05,
+      "loss": 0.5343,
+      "step": 26970
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.249819320645628e-05,
+      "loss": 0.3372,
+      "step": 26980
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.2510238496747774e-05,
+      "loss": 0.333,
+      "step": 26990
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.252228378703927e-05,
+      "loss": 0.1817,
+      "step": 27000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.2534329077330765e-05,
+      "loss": 0.2449,
+      "step": 27010
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.254637436762226e-05,
+      "loss": 0.196,
+      "step": 27020
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.2558419657913755e-05,
+      "loss": 0.1557,
+      "step": 27030
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.257046494820526e-05,
+      "loss": 0.2312,
+      "step": 27040
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.258251023849675e-05,
+      "loss": 0.2446,
+      "step": 27050
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.259455552878825e-05,
+      "loss": 0.3231,
+      "step": 27060
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.260660081907974e-05,
+      "loss": 0.3948,
+      "step": 27070
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.261864610937124e-05,
+      "loss": 0.195,
+      "step": 27080
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.263069139966273e-05,
+      "loss": 0.1357,
+      "step": 27090
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.264273668995423e-05,
+      "loss": 0.3025,
+      "step": 27100
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.265478198024572e-05,
+      "loss": 0.1533,
+      "step": 27110
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.266682727053722e-05,
+      "loss": 0.1883,
+      "step": 27120
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.2678872560828714e-05,
+      "loss": 0.4334,
+      "step": 27130
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.269091785112021e-05,
+      "loss": 0.3685,
+      "step": 27140
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.270296314141171e-05,
+      "loss": 0.2211,
+      "step": 27150
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.2715008431703206e-05,
+      "loss": 0.3349,
+      "step": 27160
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.27270537219947e-05,
+      "loss": 0.2777,
+      "step": 27170
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.2739099012286196e-05,
+      "loss": 0.2564,
+      "step": 27180
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.275114430257769e-05,
+      "loss": 0.1833,
+      "step": 27190
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.276318959286919e-05,
+      "loss": 0.2043,
+      "step": 27200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.277523488316069e-05,
+      "loss": 0.1586,
+      "step": 27210
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.2787280173452184e-05,
+      "loss": 0.2566,
+      "step": 27220
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.279932546374368e-05,
+      "loss": 0.3144,
+      "step": 27230
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.2811370754035174e-05,
+      "loss": 0.3006,
+      "step": 27240
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.282341604432667e-05,
+      "loss": 0.1209,
+      "step": 27250
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.2835461334618164e-05,
+      "loss": 0.2154,
+      "step": 27260
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.284750662490966e-05,
+      "loss": 0.2126,
+      "step": 27270
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.285955191520116e-05,
+      "loss": 0.2153,
+      "step": 27280
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.287159720549266e-05,
+      "loss": 0.3973,
+      "step": 27290
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.288364249578415e-05,
+      "loss": 0.2594,
+      "step": 27300
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.289568778607565e-05,
+      "loss": 0.3626,
+      "step": 27310
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.290773307636714e-05,
+      "loss": 0.2522,
+      "step": 27320
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.291977836665864e-05,
+      "loss": 0.2456,
+      "step": 27330
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.293182365695013e-05,
+      "loss": 0.3565,
+      "step": 27340
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.294386894724163e-05,
+      "loss": 0.1523,
+      "step": 27350
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.295591423753312e-05,
+      "loss": 0.1415,
+      "step": 27360
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.296795952782462e-05,
+      "loss": 0.195,
+      "step": 27370
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.298000481811612e-05,
+      "loss": 0.4392,
+      "step": 27380
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2992050108407615e-05,
+      "loss": 0.2469,
+      "step": 27390
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.300409539869911e-05,
+      "loss": 0.2366,
+      "step": 27400
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3016140688990606e-05,
+      "loss": 0.2001,
+      "step": 27410
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.30281859792821e-05,
+      "loss": 0.3769,
+      "step": 27420
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3040231269573596e-05,
+      "loss": 0.2706,
+      "step": 27430
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.305227655986509e-05,
+      "loss": 0.1299,
+      "step": 27440
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3064321850156587e-05,
+      "loss": 0.3105,
+      "step": 27450
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.307636714044809e-05,
+      "loss": 0.3287,
+      "step": 27460
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3088412430739584e-05,
+      "loss": 0.2702,
+      "step": 27470
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.310045772103108e-05,
+      "loss": 0.3644,
+      "step": 27480
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3112503011322574e-05,
+      "loss": 0.1838,
+      "step": 27490
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.312454830161407e-05,
+      "loss": 0.2314,
+      "step": 27500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.313659359190557e-05,
+      "loss": 0.2208,
+      "step": 27510
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3148638882197066e-05,
+      "loss": 0.2811,
+      "step": 27520
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.316068417248856e-05,
+      "loss": 0.3014,
+      "step": 27530
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.317272946278006e-05,
+      "loss": 0.2912,
+      "step": 27540
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.318477475307155e-05,
+      "loss": 0.4619,
+      "step": 27550
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.319682004336305e-05,
+      "loss": 0.3145,
+      "step": 27560
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.320886533365454e-05,
+      "loss": 0.3264,
+      "step": 27570
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.322091062394604e-05,
+      "loss": 0.1826,
+      "step": 27580
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.323295591423753e-05,
+      "loss": 0.2336,
+      "step": 27590
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.324500120452903e-05,
+      "loss": 0.1922,
+      "step": 27600
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.325704649482052e-05,
+      "loss": 0.2336,
+      "step": 27610
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3269091785112025e-05,
+      "loss": 0.1899,
+      "step": 27620
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.328113707540352e-05,
+      "loss": 0.3204,
+      "step": 27630
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3293182365695015e-05,
+      "loss": 0.3708,
+      "step": 27640
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.330522765598651e-05,
+      "loss": 0.2505,
+      "step": 27650
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3317272946278006e-05,
+      "loss": 0.2448,
+      "step": 27660
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.33293182365695e-05,
+      "loss": 0.3365,
+      "step": 27670
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3341363526860996e-05,
+      "loss": 0.5215,
+      "step": 27680
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.335340881715249e-05,
+      "loss": 0.1472,
+      "step": 27690
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3365454107443986e-05,
+      "loss": 0.1775,
+      "step": 27700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.337749939773548e-05,
+      "loss": 0.2579,
+      "step": 27710
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3389544688026984e-05,
+      "loss": 0.1481,
+      "step": 27720
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.340158997831848e-05,
+      "loss": 0.3572,
+      "step": 27730
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3413635268609974e-05,
+      "loss": 0.2743,
+      "step": 27740
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3425680558901476e-05,
+      "loss": 0.1853,
+      "step": 27750
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.343772584919297e-05,
+      "loss": 0.163,
+      "step": 27760
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3449771139484466e-05,
+      "loss": 0.2779,
+      "step": 27770
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.346181642977596e-05,
+      "loss": 0.2821,
+      "step": 27780
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.347386172006746e-05,
+      "loss": 0.2313,
+      "step": 27790
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.348590701035895e-05,
+      "loss": 0.079,
+      "step": 27800
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.349795230065045e-05,
+      "loss": 0.1402,
+      "step": 27810
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.350999759094194e-05,
+      "loss": 0.315,
+      "step": 27820
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.352204288123344e-05,
+      "loss": 0.3868,
+      "step": 27830
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.353408817152493e-05,
+      "loss": 0.2866,
+      "step": 27840
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3546133461816435e-05,
+      "loss": 0.2879,
+      "step": 27850
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.355817875210793e-05,
+      "loss": 0.3291,
+      "step": 27860
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3570224042399425e-05,
+      "loss": 0.3301,
+      "step": 27870
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.358226933269092e-05,
+      "loss": 0.2164,
+      "step": 27880
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3594314622982415e-05,
+      "loss": 0.3495,
+      "step": 27890
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.360635991327391e-05,
+      "loss": 0.2831,
+      "step": 27900
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3618405203565406e-05,
+      "loss": 0.2365,
+      "step": 27910
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.36304504938569e-05,
+      "loss": 0.2632,
+      "step": 27920
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3642495784148396e-05,
+      "loss": 0.2833,
+      "step": 27930
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.365454107443989e-05,
+      "loss": 0.4187,
+      "step": 27940
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3666586364731386e-05,
+      "loss": 0.1767,
+      "step": 27950
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.367863165502289e-05,
+      "loss": 0.2488,
+      "step": 27960
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3690676945314384e-05,
+      "loss": 0.3396,
+      "step": 27970
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3702722235605886e-05,
+      "loss": 0.2273,
+      "step": 27980
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.371476752589738e-05,
+      "loss": 0.2951,
+      "step": 27990
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3726812816188876e-05,
+      "loss": 0.1065,
+      "step": 28000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.373885810648037e-05,
+      "loss": 0.3347,
+      "step": 28010
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3750903396771866e-05,
+      "loss": 0.2958,
+      "step": 28020
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.376294868706336e-05,
+      "loss": 0.4251,
+      "step": 28030
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.377499397735486e-05,
+      "loss": 0.3955,
+      "step": 28040
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.378703926764635e-05,
+      "loss": 0.1451,
+      "step": 28050
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.379908455793785e-05,
+      "loss": 0.3612,
+      "step": 28060
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.381112984822934e-05,
+      "loss": 0.212,
+      "step": 28070
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.382317513852084e-05,
+      "loss": 0.3269,
+      "step": 28080
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.383522042881234e-05,
+      "loss": 0.0841,
+      "step": 28090
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3847265719103835e-05,
+      "loss": 0.4192,
+      "step": 28100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.385931100939533e-05,
+      "loss": 0.2288,
+      "step": 28110
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3871356299686825e-05,
+      "loss": 0.2554,
+      "step": 28120
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.388340158997832e-05,
+      "loss": 0.2873,
+      "step": 28130
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3895446880269815e-05,
+      "loss": 0.4357,
+      "step": 28140
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.390749217056131e-05,
+      "loss": 0.4127,
+      "step": 28150
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3919537460852806e-05,
+      "loss": 0.17,
+      "step": 28160
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.39315827511443e-05,
+      "loss": 0.1578,
+      "step": 28170
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3943628041435796e-05,
+      "loss": 0.4238,
+      "step": 28180
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.395567333172729e-05,
+      "loss": 0.3858,
+      "step": 28190
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.396771862201879e-05,
+      "loss": 0.2309,
+      "step": 28200
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.397976391231029e-05,
+      "loss": 0.3369,
+      "step": 28210
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3991809202601784e-05,
+      "loss": 0.2925,
+      "step": 28220
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.400385449289328e-05,
+      "loss": 0.3534,
+      "step": 28230
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.4015899783184774e-05,
+      "loss": 0.1134,
+      "step": 28240
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.4027945073476276e-05,
+      "loss": 0.1432,
+      "step": 28250
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.403999036376777e-05,
+      "loss": 0.2639,
+      "step": 28260
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.4052035654059266e-05,
+      "loss": 0.3689,
+      "step": 28270
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.406408094435076e-05,
+      "loss": 0.2787,
+      "step": 28280
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.407612623464226e-05,
+      "loss": 0.2192,
+      "step": 28290
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.408817152493375e-05,
+      "loss": 0.4711,
+      "step": 28300
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.410021681522525e-05,
+      "loss": 0.2256,
+      "step": 28310
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.411226210551675e-05,
+      "loss": 0.2276,
+      "step": 28320
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.4124307395808244e-05,
+      "loss": 0.3108,
+      "step": 28330
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.413635268609974e-05,
+      "loss": 0.248,
+      "step": 28340
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.4148397976391235e-05,
+      "loss": 0.2026,
+      "step": 28350
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.416044326668273e-05,
+      "loss": 0.3229,
+      "step": 28360
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.4172488556974225e-05,
+      "loss": 0.1863,
+      "step": 28370
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.418453384726572e-05,
+      "loss": 0.1905,
+      "step": 28380
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.4196579137557215e-05,
+      "loss": 0.1956,
+      "step": 28390
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.420862442784871e-05,
+      "loss": 0.2965,
+      "step": 28400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.4220669718140206e-05,
+      "loss": 0.2003,
+      "step": 28410
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.42327150084317e-05,
+      "loss": 0.246,
+      "step": 28420
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.42447602987232e-05,
+      "loss": 0.2369,
+      "step": 28430
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.42568055890147e-05,
+      "loss": 0.1804,
+      "step": 28440
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.426885087930619e-05,
+      "loss": 0.2908,
+      "step": 28450
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.428089616959769e-05,
+      "loss": 0.2236,
+      "step": 28460
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.4292941459889184e-05,
+      "loss": 0.2773,
+      "step": 28470
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.430498675018068e-05,
+      "loss": 0.4637,
+      "step": 28480
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.4317032040472174e-05,
+      "loss": 0.2316,
+      "step": 28490
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.432907733076367e-05,
+      "loss": 0.2739,
+      "step": 28500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.434112262105517e-05,
+      "loss": 0.3636,
+      "step": 28510
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.4353167911346666e-05,
+      "loss": 0.2554,
+      "step": 28520
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.436521320163816e-05,
+      "loss": 0.2611,
+      "step": 28530
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.4377258491929657e-05,
+      "loss": 0.1838,
+      "step": 28540
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.438930378222115e-05,
+      "loss": 0.2369,
+      "step": 28550
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.4401349072512654e-05,
+      "loss": 0.1403,
+      "step": 28560
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.441339436280415e-05,
+      "loss": 0.1892,
+      "step": 28570
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.4425439653095644e-05,
+      "loss": 0.1982,
+      "step": 28580
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.443748494338714e-05,
+      "loss": 0.1611,
+      "step": 28590
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.4449530233678634e-05,
+      "loss": 0.1947,
+      "step": 28600
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.446157552397013e-05,
+      "loss": 0.3007,
+      "step": 28610
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.4473620814261625e-05,
+      "loss": 0.4272,
+      "step": 28620
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.448566610455312e-05,
+      "loss": 0.2701,
+      "step": 28630
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.4497711394844615e-05,
+      "loss": 0.3036,
+      "step": 28640
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.450975668513611e-05,
+      "loss": 0.5418,
+      "step": 28650
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.452180197542761e-05,
+      "loss": 0.2001,
+      "step": 28660
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.453384726571911e-05,
+      "loss": 0.3192,
+      "step": 28670
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.45458925560106e-05,
+      "loss": 0.2101,
+      "step": 28680
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.45579378463021e-05,
+      "loss": 0.1474,
+      "step": 28690
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.456998313659359e-05,
+      "loss": 0.2524,
+      "step": 28700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.458202842688509e-05,
+      "loss": 0.161,
+      "step": 28710
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.4594073717176583e-05,
+      "loss": 0.2108,
+      "step": 28720
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.460611900746808e-05,
+      "loss": 0.1493,
+      "step": 28730
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.4618164297759574e-05,
+      "loss": 0.4295,
+      "step": 28740
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.463020958805107e-05,
+      "loss": 0.4098,
+      "step": 28750
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.4642254878342564e-05,
+      "loss": 0.1491,
+      "step": 28760
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.4654300168634066e-05,
+      "loss": 0.2773,
+      "step": 28770
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.466634545892556e-05,
+      "loss": 0.3226,
+      "step": 28780
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.467839074921706e-05,
+      "loss": 0.3866,
+      "step": 28790
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.469043603950856e-05,
+      "loss": 0.2606,
+      "step": 28800
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.4702481329800054e-05,
+      "loss": 0.2448,
+      "step": 28810
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.471452662009155e-05,
+      "loss": 0.5117,
+      "step": 28820
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.4726571910383044e-05,
+      "loss": 0.2296,
+      "step": 28830
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.473861720067454e-05,
+      "loss": 0.221,
+      "step": 28840
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.4750662490966034e-05,
+      "loss": 0.1589,
+      "step": 28850
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.476270778125753e-05,
+      "loss": 0.1813,
+      "step": 28860
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.4774753071549025e-05,
+      "loss": 0.2542,
+      "step": 28870
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.478679836184052e-05,
+      "loss": 0.3602,
+      "step": 28880
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.4798843652132015e-05,
+      "loss": 0.5525,
+      "step": 28890
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.481088894242352e-05,
+      "loss": 0.3918,
+      "step": 28900
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.482293423271501e-05,
+      "loss": 0.426,
+      "step": 28910
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.483497952300651e-05,
+      "loss": 0.1926,
+      "step": 28920
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.4847024813298e-05,
+      "loss": 0.2345,
+      "step": 28930
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.48590701035895e-05,
+      "loss": 0.409,
+      "step": 28940
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.487111539388099e-05,
+      "loss": 0.4043,
+      "step": 28950
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.488316068417249e-05,
+      "loss": 0.0697,
+      "step": 28960
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.4895205974463983e-05,
+      "loss": 0.3538,
+      "step": 28970
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.490725126475548e-05,
+      "loss": 0.3459,
+      "step": 28980
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.4919296555046974e-05,
+      "loss": 0.3013,
+      "step": 28990
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.493134184533847e-05,
+      "loss": 0.3845,
+      "step": 29000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.494338713562997e-05,
+      "loss": 0.2159,
+      "step": 29010
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.4955432425921466e-05,
+      "loss": 0.1297,
+      "step": 29020
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.496747771621296e-05,
+      "loss": 0.0949,
+      "step": 29030
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.497952300650446e-05,
+      "loss": 0.1753,
+      "step": 29040
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.499156829679596e-05,
+      "loss": 0.1532,
+      "step": 29050
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.5003613587087454e-05,
+      "loss": 0.2305,
+      "step": 29060
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.501565887737895e-05,
+      "loss": 0.2291,
+      "step": 29070
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.5027704167670444e-05,
+      "loss": 0.2113,
+      "step": 29080
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.503974945796194e-05,
+      "loss": 0.377,
+      "step": 29090
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.5051794748253434e-05,
+      "loss": 0.2532,
+      "step": 29100
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.506384003854493e-05,
+      "loss": 0.1015,
+      "step": 29110
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.5075885328836425e-05,
+      "loss": 0.2461,
+      "step": 29120
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.508793061912793e-05,
+      "loss": 0.2566,
+      "step": 29130
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.509997590941942e-05,
+      "loss": 0.3046,
+      "step": 29140
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.511202119971092e-05,
+      "loss": 0.2855,
+      "step": 29150
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.512406649000241e-05,
+      "loss": 0.4425,
+      "step": 29160
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.513611178029391e-05,
+      "loss": 0.1984,
+      "step": 29170
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.51481570705854e-05,
+      "loss": 0.4098,
+      "step": 29180
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.51602023608769e-05,
+      "loss": 0.3805,
+      "step": 29190
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.517224765116839e-05,
+      "loss": 0.158,
+      "step": 29200
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.518429294145989e-05,
+      "loss": 0.3621,
+      "step": 29210
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.519633823175138e-05,
+      "loss": 0.3789,
+      "step": 29220
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.520838352204288e-05,
+      "loss": 0.3018,
+      "step": 29230
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.522042881233438e-05,
+      "loss": 0.2386,
+      "step": 29240
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.5232474102625876e-05,
+      "loss": 0.2217,
+      "step": 29250
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.524451939291737e-05,
+      "loss": 0.0859,
+      "step": 29260
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.5256564683208866e-05,
+      "loss": 0.2963,
+      "step": 29270
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.526860997350036e-05,
+      "loss": 0.1147,
+      "step": 29280
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.5280655263791856e-05,
+      "loss": 0.3466,
+      "step": 29290
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.529270055408336e-05,
+      "loss": 0.2558,
+      "step": 29300
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.5304745844374854e-05,
+      "loss": 0.1643,
+      "step": 29310
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.531679113466635e-05,
+      "loss": 0.2388,
+      "step": 29320
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.5328836424957844e-05,
+      "loss": 0.4024,
+      "step": 29330
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.534088171524934e-05,
+      "loss": 0.1953,
+      "step": 29340
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.5352927005540834e-05,
+      "loss": 0.2426,
+      "step": 29350
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.536497229583233e-05,
+      "loss": 0.3099,
+      "step": 29360
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.537701758612383e-05,
+      "loss": 0.1474,
+      "step": 29370
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.538906287641533e-05,
+      "loss": 0.332,
+      "step": 29380
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.540110816670682e-05,
+      "loss": 0.2334,
+      "step": 29390
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.541315345699832e-05,
+      "loss": 0.2899,
+      "step": 29400
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.542519874728981e-05,
+      "loss": 0.2405,
+      "step": 29410
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.543724403758131e-05,
+      "loss": 0.0439,
+      "step": 29420
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.54492893278728e-05,
+      "loss": 0.2572,
+      "step": 29430
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.54613346181643e-05,
+      "loss": 0.2024,
+      "step": 29440
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.547337990845579e-05,
+      "loss": 0.2211,
+      "step": 29450
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.548542519874729e-05,
+      "loss": 0.1838,
+      "step": 29460
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.549747048903878e-05,
+      "loss": 0.3312,
+      "step": 29470
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.5509515779330285e-05,
+      "loss": 0.2681,
+      "step": 29480
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.552156106962178e-05,
+      "loss": 0.3029,
+      "step": 29490
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.5533606359913276e-05,
+      "loss": 0.1685,
+      "step": 29500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.554565165020477e-05,
+      "loss": 0.1895,
+      "step": 29510
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.5557696940496266e-05,
+      "loss": 0.1539,
+      "step": 29520
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.556974223078776e-05,
+      "loss": 0.2321,
+      "step": 29530
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.5581787521079256e-05,
+      "loss": 0.396,
+      "step": 29540
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.559383281137075e-05,
+      "loss": 0.1805,
+      "step": 29550
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.5605878101662254e-05,
+      "loss": 0.0878,
+      "step": 29560
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.561792339195375e-05,
+      "loss": 0.3098,
+      "step": 29570
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.5629968682245244e-05,
+      "loss": 0.3556,
+      "step": 29580
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.564201397253674e-05,
+      "loss": 0.5092,
+      "step": 29590
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.565405926282824e-05,
+      "loss": 0.371,
+      "step": 29600
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.5666104553119736e-05,
+      "loss": 0.278,
+      "step": 29610
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.567814984341123e-05,
+      "loss": 0.2133,
+      "step": 29620
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.5690195133702727e-05,
+      "loss": 0.136,
+      "step": 29630
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.570224042399422e-05,
+      "loss": 0.2947,
+      "step": 29640
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.1043,
+      "step": 29650
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.572633100457721e-05,
+      "loss": 0.3384,
+      "step": 29660
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.573837629486871e-05,
+      "loss": 0.151,
+      "step": 29670
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.57504215851602e-05,
+      "loss": 0.3201,
+      "step": 29680
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.57624668754517e-05,
+      "loss": 0.3953,
+      "step": 29690
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.577451216574319e-05,
+      "loss": 0.1301,
+      "step": 29700
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.5786557456034695e-05,
+      "loss": 0.1684,
+      "step": 29710
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.579860274632619e-05,
+      "loss": 0.1028,
+      "step": 29720
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.5810648036617685e-05,
+      "loss": 0.1464,
+      "step": 29730
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.582269332690918e-05,
+      "loss": 0.5429,
+      "step": 29740
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.5834738617200676e-05,
+      "loss": 0.1933,
+      "step": 29750
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.584678390749217e-05,
+      "loss": 0.3122,
+      "step": 29760
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.5858829197783666e-05,
+      "loss": 0.2116,
+      "step": 29770
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.587087448807516e-05,
+      "loss": 0.349,
+      "step": 29780
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.5882919778366656e-05,
+      "loss": 0.3734,
+      "step": 29790
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.589496506865815e-05,
+      "loss": 0.036,
+      "step": 29800
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.590701035894965e-05,
+      "loss": 0.33,
+      "step": 29810
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.591905564924115e-05,
+      "loss": 0.2557,
+      "step": 29820
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.5931100939532644e-05,
+      "loss": 0.3023,
+      "step": 29830
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.5943146229824146e-05,
+      "loss": 0.4156,
+      "step": 29840
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.595519152011564e-05,
+      "loss": 0.332,
+      "step": 29850
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.5967236810407136e-05,
+      "loss": 0.0274,
+      "step": 29860
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.597928210069863e-05,
+      "loss": 0.1758,
+      "step": 29870
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.5991327390990127e-05,
+      "loss": 0.143,
+      "step": 29880
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.600337268128162e-05,
+      "loss": 0.1809,
+      "step": 29890
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.601541797157312e-05,
+      "loss": 0.1966,
+      "step": 29900
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.602746326186461e-05,
+      "loss": 0.2268,
+      "step": 29910
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.603950855215611e-05,
+      "loss": 0.2262,
+      "step": 29920
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.60515538424476e-05,
+      "loss": 0.2277,
+      "step": 29930
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.60635991327391e-05,
+      "loss": 0.3233,
+      "step": 29940
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.60756444230306e-05,
+      "loss": 0.2062,
+      "step": 29950
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.6087689713322095e-05,
+      "loss": 0.1506,
+      "step": 29960
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.609973500361359e-05,
+      "loss": 0.1795,
+      "step": 29970
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.6111780293905085e-05,
+      "loss": 0.324,
+      "step": 29980
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.612382558419658e-05,
+      "loss": 0.2705,
+      "step": 29990
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.6135870874488076e-05,
+      "loss": 0.2606,
+      "step": 30000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.614791616477957e-05,
+      "loss": 0.2471,
+      "step": 30010
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6159961455071066e-05,
+      "loss": 0.275,
+      "step": 30020
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.617200674536256e-05,
+      "loss": 0.2872,
+      "step": 30030
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6184052035654056e-05,
+      "loss": 0.1818,
+      "step": 30040
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.619609732594556e-05,
+      "loss": 0.289,
+      "step": 30050
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6208142616237053e-05,
+      "loss": 0.3251,
+      "step": 30060
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.622018790652855e-05,
+      "loss": 0.5788,
+      "step": 30070
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6232233196820044e-05,
+      "loss": 0.1556,
+      "step": 30080
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6244278487111546e-05,
+      "loss": 0.1435,
+      "step": 30090
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.625632377740304e-05,
+      "loss": 0.4737,
+      "step": 30100
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6268369067694536e-05,
+      "loss": 0.2713,
+      "step": 30110
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.628041435798603e-05,
+      "loss": 0.2977,
+      "step": 30120
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6292459648277526e-05,
+      "loss": 0.3495,
+      "step": 30130
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.630450493856902e-05,
+      "loss": 0.3357,
+      "step": 30140
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.631655022886052e-05,
+      "loss": 0.2426,
+      "step": 30150
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.632859551915201e-05,
+      "loss": 0.2325,
+      "step": 30160
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.634064080944351e-05,
+      "loss": 0.2328,
+      "step": 30170
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.635268609973501e-05,
+      "loss": 0.2438,
+      "step": 30180
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6364731390026504e-05,
+      "loss": 0.2695,
+      "step": 30190
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6376776680318e-05,
+      "loss": 0.2466,
+      "step": 30200
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6388821970609495e-05,
+      "loss": 0.3575,
+      "step": 30210
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.640086726090099e-05,
+      "loss": 0.2356,
+      "step": 30220
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6412912551192485e-05,
+      "loss": 0.2355,
+      "step": 30230
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.642495784148398e-05,
+      "loss": 0.2285,
+      "step": 30240
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6437003131775475e-05,
+      "loss": 0.1745,
+      "step": 30250
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.644904842206697e-05,
+      "loss": 0.3633,
+      "step": 30260
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6461093712358466e-05,
+      "loss": 0.0721,
+      "step": 30270
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.647313900264996e-05,
+      "loss": 0.1836,
+      "step": 30280
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.648518429294146e-05,
+      "loss": 0.2849,
+      "step": 30290
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.649722958323296e-05,
+      "loss": 0.3982,
+      "step": 30300
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.650927487352445e-05,
+      "loss": 0.3111,
+      "step": 30310
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.652132016381595e-05,
+      "loss": 0.3179,
+      "step": 30320
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6533365454107444e-05,
+      "loss": 0.2384,
+      "step": 30330
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.654541074439894e-05,
+      "loss": 0.1421,
+      "step": 30340
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.655745603469044e-05,
+      "loss": 0.3572,
+      "step": 30350
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6569501324981936e-05,
+      "loss": 0.2889,
+      "step": 30360
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.658154661527343e-05,
+      "loss": 0.4068,
+      "step": 30370
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6593591905564926e-05,
+      "loss": 0.2564,
+      "step": 30380
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.660563719585642e-05,
+      "loss": 0.3677,
+      "step": 30390
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.661768248614792e-05,
+      "loss": 0.2067,
+      "step": 30400
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.662972777643941e-05,
+      "loss": 0.2491,
+      "step": 30410
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6641773066730914e-05,
+      "loss": 0.1803,
+      "step": 30420
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.665381835702241e-05,
+      "loss": 0.301,
+      "step": 30430
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.6665863647313904e-05,
+      "loss": 0.3228,
+      "step": 30440
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.66779089376054e-05,
+      "loss": 0.3324,
+      "step": 30450
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.6689954227896895e-05,
+      "loss": 0.191,
+      "step": 30460
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.670199951818839e-05,
+      "loss": 0.1301,
+      "step": 30470
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.6714044808479885e-05,
+      "loss": 0.2735,
+      "step": 30480
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.672609009877138e-05,
+      "loss": 0.1805,
+      "step": 30490
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.6738135389062875e-05,
+      "loss": 0.2997,
+      "step": 30500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.675018067935437e-05,
+      "loss": 0.1745,
+      "step": 30510
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.676222596964587e-05,
+      "loss": 0.3216,
+      "step": 30520
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.677427125993737e-05,
+      "loss": 0.3147,
+      "step": 30530
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.678631655022886e-05,
+      "loss": 0.241,
+      "step": 30540
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.679836184052036e-05,
+      "loss": 0.1335,
+      "step": 30550
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.681040713081185e-05,
+      "loss": 0.1245,
+      "step": 30560
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.682245242110335e-05,
+      "loss": 0.2514,
+      "step": 30570
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.6834497711394844e-05,
+      "loss": 0.269,
+      "step": 30580
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.684654300168634e-05,
+      "loss": 0.2751,
+      "step": 30590
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.6858588291977834e-05,
+      "loss": 0.2758,
+      "step": 30600
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.6870633582269336e-05,
+      "loss": 0.2884,
+      "step": 30610
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.688267887256083e-05,
+      "loss": 0.4988,
+      "step": 30620
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.6894724162852326e-05,
+      "loss": 0.3114,
+      "step": 30630
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.690676945314382e-05,
+      "loss": 0.1234,
+      "step": 30640
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.6918814743435324e-05,
+      "loss": 0.3002,
+      "step": 30650
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.693086003372682e-05,
+      "loss": 0.2148,
+      "step": 30660
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.6942905324018314e-05,
+      "loss": 0.3237,
+      "step": 30670
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.695495061430981e-05,
+      "loss": 0.1988,
+      "step": 30680
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.6966995904601304e-05,
+      "loss": 0.3415,
+      "step": 30690
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.69790411948928e-05,
+      "loss": 0.1558,
+      "step": 30700
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.6991086485184295e-05,
+      "loss": 0.2451,
+      "step": 30710
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.700313177547579e-05,
+      "loss": 0.3001,
+      "step": 30720
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7015177065767285e-05,
+      "loss": 0.3589,
+      "step": 30730
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.702722235605878e-05,
+      "loss": 0.262,
+      "step": 30740
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7039267646350275e-05,
+      "loss": 0.21,
+      "step": 30750
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.705131293664178e-05,
+      "loss": 0.2191,
+      "step": 30760
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.706335822693327e-05,
+      "loss": 0.2786,
+      "step": 30770
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.707540351722477e-05,
+      "loss": 0.3213,
+      "step": 30780
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.708744880751626e-05,
+      "loss": 0.4135,
+      "step": 30790
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.709949409780776e-05,
+      "loss": 0.1375,
+      "step": 30800
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.711153938809925e-05,
+      "loss": 0.156,
+      "step": 30810
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.712358467839075e-05,
+      "loss": 0.2564,
+      "step": 30820
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.7135629968682244e-05,
+      "loss": 0.1209,
+      "step": 30830
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.714767525897374e-05,
+      "loss": 0.4383,
+      "step": 30840
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.7159720549265234e-05,
+      "loss": 0.231,
+      "step": 30850
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.717176583955673e-05,
+      "loss": 0.322,
+      "step": 30860
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.718381112984823e-05,
+      "loss": 0.5875,
+      "step": 30870
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.7195856420139726e-05,
+      "loss": 0.2961,
+      "step": 30880
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.720790171043123e-05,
+      "loss": 0.3705,
+      "step": 30890
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.7219947000722724e-05,
+      "loss": 0.3186,
+      "step": 30900
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.723199229101422e-05,
+      "loss": 0.1865,
+      "step": 30910
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.7244037581305714e-05,
+      "loss": 0.4117,
+      "step": 30920
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.725608287159721e-05,
+      "loss": 0.3478,
+      "step": 30930
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7268128161888704e-05,
+      "loss": 0.2881,
+      "step": 30940
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.72801734521802e-05,
+      "loss": 0.1268,
+      "step": 30950
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7292218742471695e-05,
+      "loss": 0.1755,
+      "step": 30960
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.730426403276319e-05,
+      "loss": 0.0614,
+      "step": 30970
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7316309323054685e-05,
+      "loss": 0.1117,
+      "step": 30980
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.732835461334619e-05,
+      "loss": 0.4885,
+      "step": 30990
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.734039990363768e-05,
+      "loss": 0.475,
+      "step": 31000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.735244519392918e-05,
+      "loss": 0.3494,
+      "step": 31010
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.736449048422067e-05,
+      "loss": 0.1653,
+      "step": 31020
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.737653577451217e-05,
+      "loss": 0.2319,
+      "step": 31030
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.738858106480366e-05,
+      "loss": 0.3022,
+      "step": 31040
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.740062635509516e-05,
+      "loss": 0.2353,
+      "step": 31050
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.741267164538665e-05,
+      "loss": 0.1884,
+      "step": 31060
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.742471693567815e-05,
+      "loss": 0.1914,
+      "step": 31070
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.7436762225969644e-05,
+      "loss": 0.1656,
+      "step": 31080
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.744880751626114e-05,
+      "loss": 0.1695,
+      "step": 31090
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.746085280655264e-05,
+      "loss": 0.2556,
+      "step": 31100
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.7472898096844136e-05,
+      "loss": 0.2783,
+      "step": 31110
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.748494338713563e-05,
+      "loss": 0.1514,
+      "step": 31120
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.7496988677427126e-05,
+      "loss": 0.1685,
+      "step": 31130
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.750903396771863e-05,
+      "loss": 0.1317,
+      "step": 31140
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.7521079258010123e-05,
+      "loss": 0.3363,
+      "step": 31150
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.753312454830162e-05,
+      "loss": 0.1205,
+      "step": 31160
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.7545169838593114e-05,
+      "loss": 0.3478,
+      "step": 31170
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.755721512888461e-05,
+      "loss": 0.2595,
+      "step": 31180
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.7569260419176104e-05,
+      "loss": 0.1813,
+      "step": 31190
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.75813057094676e-05,
+      "loss": 0.5207,
+      "step": 31200
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.7593350999759095e-05,
+      "loss": 0.2503,
+      "step": 31210
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.760539629005059e-05,
+      "loss": 0.1479,
+      "step": 31220
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.761744158034209e-05,
+      "loss": 0.1361,
+      "step": 31230
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.762948687063359e-05,
+      "loss": 0.4258,
+      "step": 31240
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.764153216092508e-05,
+      "loss": 0.2277,
+      "step": 31250
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.765357745121658e-05,
+      "loss": 0.294,
+      "step": 31260
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.766562274150807e-05,
+      "loss": 0.4002,
+      "step": 31270
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.767766803179957e-05,
+      "loss": 0.3469,
+      "step": 31280
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.768971332209106e-05,
+      "loss": 0.3815,
+      "step": 31290
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.770175861238256e-05,
+      "loss": 0.1233,
+      "step": 31300
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.771380390267405e-05,
+      "loss": 0.4479,
+      "step": 31310
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.772584919296555e-05,
+      "loss": 0.3598,
+      "step": 31320
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.7737894483257044e-05,
+      "loss": 0.2303,
+      "step": 31330
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.7749939773548546e-05,
+      "loss": 0.2205,
+      "step": 31340
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.776198506384004e-05,
+      "loss": 0.4345,
+      "step": 31350
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7774030354131536e-05,
+      "loss": 0.2429,
+      "step": 31360
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.778607564442303e-05,
+      "loss": 0.214,
+      "step": 31370
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7798120934714526e-05,
+      "loss": 0.1664,
+      "step": 31380
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.781016622500602e-05,
+      "loss": 0.2048,
+      "step": 31390
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7822211515297523e-05,
+      "loss": 0.3724,
+      "step": 31400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.783425680558902e-05,
+      "loss": 0.1379,
+      "step": 31410
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7846302095880514e-05,
+      "loss": 0.3881,
+      "step": 31420
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.785834738617201e-05,
+      "loss": 0.5259,
+      "step": 31430
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.7870392676463504e-05,
+      "loss": 0.1808,
+      "step": 31440
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.7882437966755e-05,
+      "loss": 0.2002,
+      "step": 31450
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.78944832570465e-05,
+      "loss": 0.1679,
+      "step": 31460
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.7906528547337996e-05,
+      "loss": 0.1672,
+      "step": 31470
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.791857383762949e-05,
+      "loss": 0.2711,
+      "step": 31480
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.793061912792099e-05,
+      "loss": 0.3224,
+      "step": 31490
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.794266441821248e-05,
+      "loss": 0.2515,
+      "step": 31500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.795470970850398e-05,
+      "loss": 0.194,
+      "step": 31510
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.796675499879547e-05,
+      "loss": 0.1304,
+      "step": 31520
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.797880028908697e-05,
+      "loss": 0.2688,
+      "step": 31530
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.799084557937846e-05,
+      "loss": 0.5799,
+      "step": 31540
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.800289086966996e-05,
+      "loss": 0.2252,
+      "step": 31550
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.801493615996145e-05,
+      "loss": 0.3822,
+      "step": 31560
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.8026981450252955e-05,
+      "loss": 0.2039,
+      "step": 31570
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.803902674054445e-05,
+      "loss": 0.4044,
+      "step": 31580
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.8051072030835945e-05,
+      "loss": 0.4511,
+      "step": 31590
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.806311732112744e-05,
+      "loss": 0.3262,
+      "step": 31600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.8075162611418936e-05,
+      "loss": 0.3387,
+      "step": 31610
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.808720790171043e-05,
+      "loss": 0.2547,
+      "step": 31620
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.8099253192001926e-05,
+      "loss": 0.3501,
+      "step": 31630
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.811129848229342e-05,
+      "loss": 0.2675,
+      "step": 31640
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.8123343772584917e-05,
+      "loss": 0.2941,
+      "step": 31650
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.813538906287642e-05,
+      "loss": 0.5954,
+      "step": 31660
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.8147434353167914e-05,
+      "loss": 0.2393,
+      "step": 31670
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.815947964345941e-05,
+      "loss": 0.2527,
+      "step": 31680
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.8171524933750904e-05,
+      "loss": 0.2628,
+      "step": 31690
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.8183570224042406e-05,
+      "loss": 0.4984,
+      "step": 31700
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.81956155143339e-05,
+      "loss": 0.3775,
+      "step": 31710
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.8207660804625396e-05,
+      "loss": 0.2264,
+      "step": 31720
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.821970609491689e-05,
+      "loss": 0.3325,
+      "step": 31730
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.823175138520839e-05,
+      "loss": 0.2153,
+      "step": 31740
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.824379667549988e-05,
+      "loss": 0.4226,
+      "step": 31750
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.825584196579138e-05,
+      "loss": 0.2325,
+      "step": 31760
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.826788725608287e-05,
+      "loss": 0.1181,
+      "step": 31770
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.827993254637437e-05,
+      "loss": 0.4905,
+      "step": 31780
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.829197783666586e-05,
+      "loss": 0.2614,
+      "step": 31790
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.830402312695736e-05,
+      "loss": 0.2512,
+      "step": 31800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.831606841724886e-05,
+      "loss": 0.3443,
+      "step": 31810
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.8328113707540355e-05,
+      "loss": 0.1375,
+      "step": 31820
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.834015899783185e-05,
+      "loss": 0.2822,
+      "step": 31830
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.8352204288123345e-05,
+      "loss": 0.3088,
+      "step": 31840
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.836424957841484e-05,
+      "loss": 0.2784,
+      "step": 31850
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8376294868706336e-05,
+      "loss": 0.3122,
+      "step": 31860
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.838834015899783e-05,
+      "loss": 0.1446,
+      "step": 31870
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8400385449289326e-05,
+      "loss": 0.3248,
+      "step": 31880
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.841243073958082e-05,
+      "loss": 0.3632,
+      "step": 31890
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8424476029872317e-05,
+      "loss": 0.2204,
+      "step": 31900
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.843652132016382e-05,
+      "loss": 0.2558,
+      "step": 31910
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8448566610455314e-05,
+      "loss": 0.214,
+      "step": 31920
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8460611900746816e-05,
+      "loss": 0.2772,
+      "step": 31930
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.847265719103831e-05,
+      "loss": 0.2405,
+      "step": 31940
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8484702481329806e-05,
+      "loss": 0.3423,
+      "step": 31950
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.84967477716213e-05,
+      "loss": 0.3093,
+      "step": 31960
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8508793061912796e-05,
+      "loss": 0.088,
+      "step": 31970
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.852083835220429e-05,
+      "loss": 0.4126,
+      "step": 31980
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.853288364249579e-05,
+      "loss": 0.3808,
+      "step": 31990
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.854492893278728e-05,
+      "loss": 0.1786,
+      "step": 32000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.855697422307878e-05,
+      "loss": 0.1148,
+      "step": 32010
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.856901951337027e-05,
+      "loss": 0.2674,
+      "step": 32020
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.858106480366177e-05,
+      "loss": 0.3931,
+      "step": 32030
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.859311009395327e-05,
+      "loss": 0.2202,
+      "step": 32040
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.8605155384244765e-05,
+      "loss": 0.1917,
+      "step": 32050
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.861720067453626e-05,
+      "loss": 0.2674,
+      "step": 32060
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.8629245964827755e-05,
+      "loss": 0.335,
+      "step": 32070
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.864129125511925e-05,
+      "loss": 0.3662,
+      "step": 32080
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.8653336545410745e-05,
+      "loss": 0.2781,
+      "step": 32090
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.866538183570224e-05,
+      "loss": 0.0613,
+      "step": 32100
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.8677427125993736e-05,
+      "loss": 0.197,
+      "step": 32110
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.868947241628523e-05,
+      "loss": 0.3122,
+      "step": 32120
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.8701517706576726e-05,
+      "loss": 0.289,
+      "step": 32130
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.871356299686822e-05,
+      "loss": 0.167,
+      "step": 32140
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.872560828715972e-05,
+      "loss": 0.4499,
+      "step": 32150
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.873765357745122e-05,
+      "loss": 0.2209,
+      "step": 32160
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.8749698867742714e-05,
+      "loss": 0.315,
+      "step": 32170
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.876174415803421e-05,
+      "loss": 0.2434,
+      "step": 32180
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.877378944832571e-05,
+      "loss": 0.2975,
+      "step": 32190
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.8785834738617206e-05,
+      "loss": 0.3039,
+      "step": 32200
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.87978800289087e-05,
+      "loss": 0.2617,
+      "step": 32210
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.8809925319200196e-05,
+      "loss": 0.0703,
+      "step": 32220
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.882197060949169e-05,
+      "loss": 0.2077,
+      "step": 32230
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.883401589978319e-05,
+      "loss": 0.4114,
+      "step": 32240
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.884606119007468e-05,
+      "loss": 0.2079,
+      "step": 32250
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.885810648036618e-05,
+      "loss": 0.2474,
+      "step": 32260
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.887015177065768e-05,
+      "loss": 0.2822,
+      "step": 32270
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.8882197060949174e-05,
+      "loss": 0.3251,
+      "step": 32280
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.889424235124067e-05,
+      "loss": 0.4296,
+      "step": 32290
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.8906287641532165e-05,
+      "loss": 0.4961,
+      "step": 32300
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.891833293182366e-05,
+      "loss": 0.1191,
+      "step": 32310
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.8930378222115155e-05,
+      "loss": 0.193,
+      "step": 32320
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.894242351240665e-05,
+      "loss": 0.1235,
+      "step": 32330
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.8954468802698145e-05,
+      "loss": 0.1821,
+      "step": 32340
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.896651409298964e-05,
+      "loss": 0.3404,
+      "step": 32350
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.8978559383281136e-05,
+      "loss": 0.2161,
+      "step": 32360
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.899060467357263e-05,
+      "loss": 0.4441,
+      "step": 32370
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.900264996386413e-05,
+      "loss": 0.4316,
+      "step": 32380
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.901469525415563e-05,
+      "loss": 0.2905,
+      "step": 32390
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.902674054444712e-05,
+      "loss": 0.159,
+      "step": 32400
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.903878583473862e-05,
+      "loss": 0.2364,
+      "step": 32410
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.9050831125030114e-05,
+      "loss": 0.1755,
+      "step": 32420
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.906287641532161e-05,
+      "loss": 0.3848,
+      "step": 32430
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.9074921705613104e-05,
+      "loss": 0.2705,
+      "step": 32440
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.9086966995904606e-05,
+      "loss": 0.2744,
+      "step": 32450
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.90990122861961e-05,
+      "loss": 0.4313,
+      "step": 32460
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.9111057576487596e-05,
+      "loss": 0.1795,
+      "step": 32470
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.912310286677909e-05,
+      "loss": 0.3415,
+      "step": 32480
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.913514815707059e-05,
+      "loss": 0.311,
+      "step": 32490
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.914719344736208e-05,
+      "loss": 0.2383,
+      "step": 32500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.9159238737653584e-05,
+      "loss": 0.194,
+      "step": 32510
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.917128402794508e-05,
+      "loss": 0.3744,
+      "step": 32520
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.9183329318236574e-05,
+      "loss": 0.2468,
+      "step": 32530
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.919537460852807e-05,
+      "loss": 0.1223,
+      "step": 32540
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.9207419898819565e-05,
+      "loss": 0.2399,
+      "step": 32550
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.921946518911106e-05,
+      "loss": 0.1196,
+      "step": 32560
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.9231510479402555e-05,
+      "loss": 0.2538,
+      "step": 32570
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.924355576969405e-05,
+      "loss": 0.3124,
+      "step": 32580
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.9255601059985545e-05,
+      "loss": 0.2125,
+      "step": 32590
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.926764635027704e-05,
+      "loss": 0.2427,
+      "step": 32600
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.9279691640568536e-05,
+      "loss": 0.1456,
+      "step": 32610
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.929173693086004e-05,
+      "loss": 0.1592,
+      "step": 32620
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.930378222115153e-05,
+      "loss": 0.3114,
+      "step": 32630
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.931582751144303e-05,
+      "loss": 0.1138,
+      "step": 32640
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.932787280173452e-05,
+      "loss": 0.3536,
+      "step": 32650
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.933991809202602e-05,
+      "loss": 0.2036,
+      "step": 32660
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.9351963382317514e-05,
+      "loss": 0.2999,
+      "step": 32670
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.936400867260901e-05,
+      "loss": 0.2174,
+      "step": 32680
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.9376053962900504e-05,
+      "loss": 0.1972,
+      "step": 32690
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.9388099253192e-05,
+      "loss": 0.2826,
+      "step": 32700
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.94001445434835e-05,
+      "loss": 0.3127,
+      "step": 32710
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.9412189833774996e-05,
+      "loss": 0.2374,
+      "step": 32720
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.942423512406649e-05,
+      "loss": 0.2799,
+      "step": 32730
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.943628041435799e-05,
+      "loss": 0.3392,
+      "step": 32740
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.944832570464949e-05,
+      "loss": 0.4945,
+      "step": 32750
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.9460370994940984e-05,
+      "loss": 0.3363,
+      "step": 32760
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.947241628523248e-05,
+      "loss": 0.3333,
+      "step": 32770
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.9484461575523974e-05,
+      "loss": 0.2672,
+      "step": 32780
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.949650686581547e-05,
+      "loss": 0.2658,
+      "step": 32790
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.9508552156106964e-05,
+      "loss": 0.421,
+      "step": 32800
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.952059744639846e-05,
+      "loss": 0.4374,
+      "step": 32810
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.9532642736689955e-05,
+      "loss": 0.3412,
+      "step": 32820
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.954468802698145e-05,
+      "loss": 0.4263,
+      "step": 32830
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.9556733317272945e-05,
+      "loss": 0.4438,
+      "step": 32840
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.956877860756445e-05,
+      "loss": 0.3669,
+      "step": 32850
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.958082389785594e-05,
+      "loss": 0.1728,
+      "step": 32860
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.959286918814744e-05,
+      "loss": 0.3205,
+      "step": 32870
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.960491447843893e-05,
+      "loss": 0.3602,
+      "step": 32880
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.961695976873043e-05,
+      "loss": 0.1639,
+      "step": 32890
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.962900505902192e-05,
+      "loss": 0.306,
+      "step": 32900
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.964105034931342e-05,
+      "loss": 0.1455,
+      "step": 32910
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.9653095639604913e-05,
+      "loss": 0.05,
+      "step": 32920
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.966514092989641e-05,
+      "loss": 0.1081,
+      "step": 32930
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.9677186220187904e-05,
+      "loss": 0.2159,
+      "step": 32940
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.96892315104794e-05,
+      "loss": 0.3165,
+      "step": 32950
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.97012768007709e-05,
+      "loss": 0.2959,
+      "step": 32960
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.9713322091062396e-05,
+      "loss": 0.0904,
+      "step": 32970
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.97253673813539e-05,
+      "loss": 0.176,
+      "step": 32980
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.973741267164539e-05,
+      "loss": 0.3635,
+      "step": 32990
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.974945796193689e-05,
+      "loss": 0.2314,
+      "step": 33000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.9761503252228384e-05,
+      "loss": 0.4163,
+      "step": 33010
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.977354854251988e-05,
+      "loss": 0.2576,
+      "step": 33020
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.9785593832811374e-05,
+      "loss": 0.3855,
+      "step": 33030
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.979763912310287e-05,
+      "loss": 0.2024,
+      "step": 33040
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.9809684413394364e-05,
+      "loss": 0.2654,
+      "step": 33050
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.982172970368586e-05,
+      "loss": 0.2568,
+      "step": 33060
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.9833774993977355e-05,
+      "loss": 0.2786,
+      "step": 33070
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.984582028426885e-05,
+      "loss": 0.1339,
+      "step": 33080
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.985786557456035e-05,
+      "loss": 0.3138,
+      "step": 33090
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.986991086485185e-05,
+      "loss": 0.2643,
+      "step": 33100
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.988195615514334e-05,
+      "loss": 0.217,
+      "step": 33110
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.989400144543484e-05,
+      "loss": 0.2583,
+      "step": 33120
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.990604673572633e-05,
+      "loss": 0.3285,
+      "step": 33130
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.991809202601783e-05,
+      "loss": 0.2812,
+      "step": 33140
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.993013731630932e-05,
+      "loss": 0.2848,
+      "step": 33150
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.994218260660082e-05,
+      "loss": 0.2948,
+      "step": 33160
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.9954227896892313e-05,
+      "loss": 0.293,
+      "step": 33170
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.996627318718381e-05,
+      "loss": 0.239,
+      "step": 33180
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.997831847747531e-05,
+      "loss": 0.1979,
+      "step": 33190
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.9990363767766806e-05,
+      "loss": 0.2219,
+      "step": 33200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.00024090580583e-05,
+      "loss": 0.09,
+      "step": 33210
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": {
+        "accuracy": 0.9270099367660344
+      },
+      "eval_f1": {
+        "f1": 0.9272025877193879
+      },
+      "eval_loss": 0.27678197622299194,
+      "eval_precision": {
+        "precision": 0.9305221603080029
+      },
+      "eval_recall": {
+        "recall": 0.9267551810236085
+      },
+      "eval_runtime": 429.0362,
+      "eval_samples_per_second": 38.703,
+      "eval_steps_per_second": 19.353,
+      "step": 33210
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0014454348349796e-05,
+      "loss": 0.2848,
+      "step": 33220
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.002649963864129e-05,
+      "loss": 0.2293,
+      "step": 33230
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.003854492893279e-05,
+      "loss": 0.1984,
+      "step": 33240
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.005059021922429e-05,
+      "loss": 0.1779,
+      "step": 33250
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0062635509515784e-05,
+      "loss": 0.3088,
+      "step": 33260
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.007468079980728e-05,
+      "loss": 0.2359,
+      "step": 33270
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0086726090098774e-05,
+      "loss": 0.1043,
+      "step": 33280
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.009877138039027e-05,
+      "loss": 0.2482,
+      "step": 33290
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0110816670681764e-05,
+      "loss": 0.1613,
+      "step": 33300
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.012286196097326e-05,
+      "loss": 0.2923,
+      "step": 33310
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.013490725126476e-05,
+      "loss": 0.2288,
+      "step": 33320
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.014695254155626e-05,
+      "loss": 0.3003,
+      "step": 33330
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.015899783184775e-05,
+      "loss": 0.067,
+      "step": 33340
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.017104312213925e-05,
+      "loss": 0.2218,
+      "step": 33350
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.018308841243074e-05,
+      "loss": 0.3513,
+      "step": 33360
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.019513370272224e-05,
+      "loss": 0.231,
+      "step": 33370
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.020717899301373e-05,
+      "loss": 0.1881,
+      "step": 33380
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.021922428330523e-05,
+      "loss": 0.1857,
+      "step": 33390
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.023126957359672e-05,
+      "loss": 0.2366,
+      "step": 33400
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.024331486388822e-05,
+      "loss": 0.3442,
+      "step": 33410
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.0255360154179713e-05,
+      "loss": 0.235,
+      "step": 33420
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.0267405444471215e-05,
+      "loss": 0.1487,
+      "step": 33430
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.027945073476271e-05,
+      "loss": 0.217,
+      "step": 33440
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.0291496025054206e-05,
+      "loss": 0.3399,
+      "step": 33450
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.03035413153457e-05,
+      "loss": 0.1086,
+      "step": 33460
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.0315586605637196e-05,
+      "loss": 0.1164,
+      "step": 33470
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.032763189592869e-05,
+      "loss": 0.431,
+      "step": 33480
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.0339677186220186e-05,
+      "loss": 0.1056,
+      "step": 33490
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.035172247651169e-05,
+      "loss": 0.4112,
+      "step": 33500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.0363767766803184e-05,
+      "loss": 0.2826,
+      "step": 33510
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.037581305709468e-05,
+      "loss": 0.3452,
+      "step": 33520
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.0387858347386174e-05,
+      "loss": 0.1911,
+      "step": 33530
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.039990363767767e-05,
+      "loss": 0.2294,
+      "step": 33540
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.0411948927969164e-05,
+      "loss": 0.2684,
+      "step": 33550
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.0423994218260666e-05,
+      "loss": 0.4026,
+      "step": 33560
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.043603950855216e-05,
+      "loss": 0.3656,
+      "step": 33570
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.044808479884366e-05,
+      "loss": 0.1778,
+      "step": 33580
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.046013008913515e-05,
+      "loss": 0.3002,
+      "step": 33590
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.047217537942665e-05,
+      "loss": 0.1779,
+      "step": 33600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.048422066971814e-05,
+      "loss": 0.1599,
+      "step": 33610
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.049626596000964e-05,
+      "loss": 0.2857,
+      "step": 33620
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.050831125030113e-05,
+      "loss": 0.2589,
+      "step": 33630
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.052035654059263e-05,
+      "loss": 0.1421,
+      "step": 33640
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.053240183088412e-05,
+      "loss": 0.3318,
+      "step": 33650
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.0544447121175625e-05,
+      "loss": 0.1712,
+      "step": 33660
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.055649241146712e-05,
+      "loss": 0.1877,
+      "step": 33670
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.0568537701758615e-05,
+      "loss": 0.1467,
+      "step": 33680
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.058058299205011e-05,
+      "loss": 0.1698,
+      "step": 33690
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.0592628282341606e-05,
+      "loss": 0.4227,
+      "step": 33700
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.06046735726331e-05,
+      "loss": 0.1164,
+      "step": 33710
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.0616718862924596e-05,
+      "loss": 0.2135,
+      "step": 33720
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.062876415321609e-05,
+      "loss": 0.3482,
+      "step": 33730
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.0640809443507586e-05,
+      "loss": 0.2802,
+      "step": 33740
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.065285473379908e-05,
+      "loss": 0.2943,
+      "step": 33750
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.0664900024090584e-05,
+      "loss": 0.1267,
+      "step": 33760
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.067694531438208e-05,
+      "loss": 0.3951,
+      "step": 33770
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.0688990604673574e-05,
+      "loss": 0.1685,
+      "step": 33780
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.0701035894965076e-05,
+      "loss": 0.2998,
+      "step": 33790
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.071308118525657e-05,
+      "loss": 0.3728,
+      "step": 33800
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.0725126475548066e-05,
+      "loss": 0.1915,
+      "step": 33810
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.073717176583956e-05,
+      "loss": 0.2399,
+      "step": 33820
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.074921705613106e-05,
+      "loss": 0.1379,
+      "step": 33830
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.076126234642255e-05,
+      "loss": 0.2221,
+      "step": 33840
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.077330763671405e-05,
+      "loss": 0.228,
+      "step": 33850
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.078535292700554e-05,
+      "loss": 0.4477,
+      "step": 33860
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.079739821729704e-05,
+      "loss": 0.2107,
+      "step": 33870
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.080944350758853e-05,
+      "loss": 0.4021,
+      "step": 33880
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.082148879788003e-05,
+      "loss": 0.2587,
+      "step": 33890
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.083353408817153e-05,
+      "loss": 0.1665,
+      "step": 33900
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.0845579378463025e-05,
+      "loss": 0.1526,
+      "step": 33910
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.085762466875452e-05,
+      "loss": 0.2276,
+      "step": 33920
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.0869669959046015e-05,
+      "loss": 0.1205,
+      "step": 33930
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.088171524933751e-05,
+      "loss": 0.1753,
+      "step": 33940
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.0893760539629006e-05,
+      "loss": 0.2157,
+      "step": 33950
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.09058058299205e-05,
+      "loss": 0.1346,
+      "step": 33960
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.0917851120211996e-05,
+      "loss": 0.2414,
+      "step": 33970
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.092989641050349e-05,
+      "loss": 0.3214,
+      "step": 33980
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.0941941700794986e-05,
+      "loss": 0.3031,
+      "step": 33990
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.095398699108648e-05,
+      "loss": 0.3728,
+      "step": 34000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.0966032281377984e-05,
+      "loss": 0.1784,
+      "step": 34010
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.097807757166948e-05,
+      "loss": 0.1688,
+      "step": 34020
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.099012286196098e-05,
+      "loss": 0.2658,
+      "step": 34030
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.1002168152252476e-05,
+      "loss": 0.1885,
+      "step": 34040
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.101421344254397e-05,
+      "loss": 0.0878,
+      "step": 34050
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.1026258732835466e-05,
+      "loss": 0.1709,
+      "step": 34060
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.103830402312696e-05,
+      "loss": 0.2661,
+      "step": 34070
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.1050349313418457e-05,
+      "loss": 0.1532,
+      "step": 34080
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.106239460370995e-05,
+      "loss": 0.2694,
+      "step": 34090
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.107443989400145e-05,
+      "loss": 0.2558,
+      "step": 34100
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.108648518429294e-05,
+      "loss": 0.1314,
+      "step": 34110
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.109853047458444e-05,
+      "loss": 0.2068,
+      "step": 34120
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.111057576487594e-05,
+      "loss": 0.0472,
+      "step": 34130
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.1122621055167434e-05,
+      "loss": 0.2623,
+      "step": 34140
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.113466634545893e-05,
+      "loss": 0.3258,
+      "step": 34150
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.1146711635750425e-05,
+      "loss": 0.1509,
+      "step": 34160
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.115875692604192e-05,
+      "loss": 0.3047,
+      "step": 34170
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.1170802216333415e-05,
+      "loss": 0.3142,
+      "step": 34180
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.118284750662491e-05,
+      "loss": 0.3367,
+      "step": 34190
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.1194892796916406e-05,
+      "loss": 0.2741,
+      "step": 34200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.12069380872079e-05,
+      "loss": 0.1541,
+      "step": 34210
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.1218983377499396e-05,
+      "loss": 0.3188,
+      "step": 34220
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.123102866779089e-05,
+      "loss": 0.0998,
+      "step": 34230
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.124307395808239e-05,
+      "loss": 0.4016,
+      "step": 34240
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.125511924837389e-05,
+      "loss": 0.2447,
+      "step": 34250
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.1267164538665383e-05,
+      "loss": 0.2129,
+      "step": 34260
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.127920982895688e-05,
+      "loss": 0.2318,
+      "step": 34270
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.1291255119248374e-05,
+      "loss": 0.1819,
+      "step": 34280
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.1303300409539876e-05,
+      "loss": 0.2334,
+      "step": 34290
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.131534569983137e-05,
+      "loss": 0.1018,
+      "step": 34300
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.1327390990122866e-05,
+      "loss": 0.2236,
+      "step": 34310
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.133943628041436e-05,
+      "loss": 0.2631,
+      "step": 34320
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.1351481570705857e-05,
+      "loss": 0.3035,
+      "step": 34330
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.136352686099735e-05,
+      "loss": 0.1854,
+      "step": 34340
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.137557215128885e-05,
+      "loss": 0.3571,
+      "step": 34350
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.138761744158034e-05,
+      "loss": 0.2795,
+      "step": 34360
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.1399662731871844e-05,
+      "loss": 0.5032,
+      "step": 34370
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.141170802216334e-05,
+      "loss": 0.2605,
+      "step": 34380
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.1423753312454834e-05,
+      "loss": 0.383,
+      "step": 34390
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.143579860274633e-05,
+      "loss": 0.193,
+      "step": 34400
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.1447843893037825e-05,
+      "loss": 0.226,
+      "step": 34410
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.145988918332932e-05,
+      "loss": 0.2275,
+      "step": 34420
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.1471934473620815e-05,
+      "loss": 0.344,
+      "step": 34430
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.148397976391231e-05,
+      "loss": 0.2784,
+      "step": 34440
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.1496025054203806e-05,
+      "loss": 0.1532,
+      "step": 34450
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.15080703444953e-05,
+      "loss": 0.1556,
+      "step": 34460
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.1520115634786796e-05,
+      "loss": 0.2741,
+      "step": 34470
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.15321609250783e-05,
+      "loss": 0.2575,
+      "step": 34480
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.154420621536979e-05,
+      "loss": 0.2658,
+      "step": 34490
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.155625150566129e-05,
+      "loss": 0.1676,
+      "step": 34500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.1568296795952783e-05,
+      "loss": 0.2814,
+      "step": 34510
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.158034208624428e-05,
+      "loss": 0.1741,
+      "step": 34520
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.1592387376535774e-05,
+      "loss": 0.2742,
+      "step": 34530
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.160443266682727e-05,
+      "loss": 0.1246,
+      "step": 34540
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.161647795711877e-05,
+      "loss": 0.2753,
+      "step": 34550
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.1628523247410266e-05,
+      "loss": 0.2132,
+      "step": 34560
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.164056853770176e-05,
+      "loss": 0.1867,
+      "step": 34570
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.1652613827993256e-05,
+      "loss": 0.0775,
+      "step": 34580
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.166465911828475e-05,
+      "loss": 0.4127,
+      "step": 34590
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.1676704408576254e-05,
+      "loss": 0.2134,
+      "step": 34600
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.168874969886775e-05,
+      "loss": 0.1471,
+      "step": 34610
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.1700794989159244e-05,
+      "loss": 0.273,
+      "step": 34620
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.171284027945074e-05,
+      "loss": 0.3621,
+      "step": 34630
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.1724885569742234e-05,
+      "loss": 0.3377,
+      "step": 34640
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.173693086003373e-05,
+      "loss": 0.2689,
+      "step": 34650
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.1748976150325225e-05,
+      "loss": 0.1868,
+      "step": 34660
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.176102144061672e-05,
+      "loss": 0.2096,
+      "step": 34670
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.1773066730908215e-05,
+      "loss": 0.3064,
+      "step": 34680
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.178511202119971e-05,
+      "loss": 0.1884,
+      "step": 34690
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.1797157311491205e-05,
+      "loss": 0.1012,
+      "step": 34700
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.180920260178271e-05,
+      "loss": 0.287,
+      "step": 34710
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.18212478920742e-05,
+      "loss": 0.2554,
+      "step": 34720
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.18332931823657e-05,
+      "loss": 0.2213,
+      "step": 34730
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.184533847265719e-05,
+      "loss": 0.2224,
+      "step": 34740
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.185738376294869e-05,
+      "loss": 0.4572,
+      "step": 34750
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.186942905324018e-05,
+      "loss": 0.3385,
+      "step": 34760
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.188147434353168e-05,
+      "loss": 0.3803,
+      "step": 34770
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.1893519633823174e-05,
+      "loss": 0.2067,
+      "step": 34780
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.190556492411467e-05,
+      "loss": 0.2881,
+      "step": 34790
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.1917610214406164e-05,
+      "loss": 0.2447,
+      "step": 34800
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.192965550469766e-05,
+      "loss": 0.1968,
+      "step": 34810
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.194170079498916e-05,
+      "loss": 0.4183,
+      "step": 34820
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.1953746085280656e-05,
+      "loss": 0.1916,
+      "step": 34830
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.196579137557216e-05,
+      "loss": 0.1375,
+      "step": 34840
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.1977836665863654e-05,
+      "loss": 0.2471,
+      "step": 34850
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.198988195615515e-05,
+      "loss": 0.1431,
+      "step": 34860
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.2001927246446644e-05,
+      "loss": 0.1863,
+      "step": 34870
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.201397253673814e-05,
+      "loss": 0.3069,
+      "step": 34880
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.2026017827029634e-05,
+      "loss": 0.2522,
+      "step": 34890
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.203806311732113e-05,
+      "loss": 0.2719,
+      "step": 34900
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.2050108407612625e-05,
+      "loss": 0.3427,
+      "step": 34910
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.206215369790412e-05,
+      "loss": 0.2643,
+      "step": 34920
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.2074198988195615e-05,
+      "loss": 0.4355,
+      "step": 34930
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.208624427848711e-05,
+      "loss": 0.3746,
+      "step": 34940
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.209828956877861e-05,
+      "loss": 0.2947,
+      "step": 34950
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.211033485907011e-05,
+      "loss": 0.1173,
+      "step": 34960
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.21223801493616e-05,
+      "loss": 0.2151,
+      "step": 34970
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.21344254396531e-05,
+      "loss": 0.2599,
+      "step": 34980
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.214647072994459e-05,
+      "loss": 0.1904,
+      "step": 34990
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.215851602023609e-05,
+      "loss": 0.404,
+      "step": 35000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.217056131052758e-05,
+      "loss": 0.33,
+      "step": 35010
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.218260660081908e-05,
+      "loss": 0.2585,
+      "step": 35020
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.2194651891110574e-05,
+      "loss": 0.3242,
+      "step": 35030
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.220669718140207e-05,
+      "loss": 0.1587,
+      "step": 35040
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.221874247169357e-05,
+      "loss": 0.3058,
+      "step": 35050
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.2230787761985066e-05,
+      "loss": 0.2632,
+      "step": 35060
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.224283305227656e-05,
+      "loss": 0.315,
+      "step": 35070
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.225487834256806e-05,
+      "loss": 0.1706,
+      "step": 35080
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.226692363285956e-05,
+      "loss": 0.3263,
+      "step": 35090
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2278968923151054e-05,
+      "loss": 0.4642,
+      "step": 35100
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.229101421344255e-05,
+      "loss": 0.3717,
+      "step": 35110
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2303059503734044e-05,
+      "loss": 0.1077,
+      "step": 35120
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.231510479402554e-05,
+      "loss": 0.2378,
+      "step": 35130
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2327150084317034e-05,
+      "loss": 0.3399,
+      "step": 35140
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.233919537460853e-05,
+      "loss": 0.2906,
+      "step": 35150
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2351240664900025e-05,
+      "loss": 0.3618,
+      "step": 35160
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.236328595519152e-05,
+      "loss": 0.2747,
+      "step": 35170
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.237533124548302e-05,
+      "loss": 0.3716,
+      "step": 35180
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.238737653577452e-05,
+      "loss": 0.1685,
+      "step": 35190
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.239942182606601e-05,
+      "loss": 0.1075,
+      "step": 35200
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.241146711635751e-05,
+      "loss": 0.2023,
+      "step": 35210
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2423512406649e-05,
+      "loss": 0.1142,
+      "step": 35220
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.24355576969405e-05,
+      "loss": 0.1607,
+      "step": 35230
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.244760298723199e-05,
+      "loss": 0.3939,
+      "step": 35240
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.245964827752349e-05,
+      "loss": 0.2126,
+      "step": 35250
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.247169356781498e-05,
+      "loss": 0.2509,
+      "step": 35260
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.248373885810648e-05,
+      "loss": 0.1977,
+      "step": 35270
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2495784148397974e-05,
+      "loss": 0.3365,
+      "step": 35280
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2507829438689476e-05,
+      "loss": 0.0871,
+      "step": 35290
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.251987472898097e-05,
+      "loss": 0.593,
+      "step": 35300
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2531920019272466e-05,
+      "loss": 0.1882,
+      "step": 35310
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.254396530956396e-05,
+      "loss": 0.268,
+      "step": 35320
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2556010599855456e-05,
+      "loss": 0.1368,
+      "step": 35330
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.256805589014696e-05,
+      "loss": 0.0972,
+      "step": 35340
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2580101180438453e-05,
+      "loss": 0.3301,
+      "step": 35350
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.259214647072995e-05,
+      "loss": 0.3782,
+      "step": 35360
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2604191761021444e-05,
+      "loss": 0.3369,
+      "step": 35370
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.261623705131294e-05,
+      "loss": 0.2774,
+      "step": 35380
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2628282341604434e-05,
+      "loss": 0.1598,
+      "step": 35390
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.264032763189593e-05,
+      "loss": 0.2128,
+      "step": 35400
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2652372922187425e-05,
+      "loss": 0.2064,
+      "step": 35410
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2664418212478927e-05,
+      "loss": 0.1388,
+      "step": 35420
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.267646350277042e-05,
+      "loss": 0.2025,
+      "step": 35430
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.268850879306192e-05,
+      "loss": 0.5525,
+      "step": 35440
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.270055408335341e-05,
+      "loss": 0.3266,
+      "step": 35450
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.271259937364491e-05,
+      "loss": 0.1617,
+      "step": 35460
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.27246446639364e-05,
+      "loss": 0.3875,
+      "step": 35470
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.27366899542279e-05,
+      "loss": 0.4432,
+      "step": 35480
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.274873524451939e-05,
+      "loss": 0.1232,
+      "step": 35490
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.276078053481089e-05,
+      "loss": 0.089,
+      "step": 35500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.277282582510238e-05,
+      "loss": 0.3685,
+      "step": 35510
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.2784871115393885e-05,
+      "loss": 0.3265,
+      "step": 35520
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.279691640568538e-05,
+      "loss": 0.086,
+      "step": 35530
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.2808961695976876e-05,
+      "loss": 0.2044,
+      "step": 35540
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.282100698626837e-05,
+      "loss": 0.0223,
+      "step": 35550
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.2833052276559866e-05,
+      "loss": 0.3998,
+      "step": 35560
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.284509756685136e-05,
+      "loss": 0.1732,
+      "step": 35570
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.2048,
+      "step": 35580
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.286918814743435e-05,
+      "loss": 0.3673,
+      "step": 35590
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2881233437725853e-05,
+      "loss": 0.3091,
+      "step": 35600
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.289327872801735e-05,
+      "loss": 0.1977,
+      "step": 35610
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2905324018308844e-05,
+      "loss": 0.099,
+      "step": 35620
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.291736930860034e-05,
+      "loss": 0.39,
+      "step": 35630
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2929414598891834e-05,
+      "loss": 0.3661,
+      "step": 35640
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2941459889183336e-05,
+      "loss": 0.5066,
+      "step": 35650
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.295350517947483e-05,
+      "loss": 0.2269,
+      "step": 35660
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.2965550469766327e-05,
+      "loss": 0.2134,
+      "step": 35670
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.297759576005782e-05,
+      "loss": 0.3613,
+      "step": 35680
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.298964105034932e-05,
+      "loss": 0.2355,
+      "step": 35690
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.300168634064081e-05,
+      "loss": 0.3032,
+      "step": 35700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.301373163093231e-05,
+      "loss": 0.2892,
+      "step": 35710
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.30257769212238e-05,
+      "loss": 0.392,
+      "step": 35720
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.30378222115153e-05,
+      "loss": 0.1878,
+      "step": 35730
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.304986750180679e-05,
+      "loss": 0.2843,
+      "step": 35740
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.306191279209829e-05,
+      "loss": 0.5649,
+      "step": 35750
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.307395808238979e-05,
+      "loss": 0.2119,
+      "step": 35760
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.3086003372681285e-05,
+      "loss": 0.2572,
+      "step": 35770
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.309804866297278e-05,
+      "loss": 0.2904,
+      "step": 35780
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.3110093953264276e-05,
+      "loss": 0.2794,
+      "step": 35790
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.312213924355577e-05,
+      "loss": 0.2807,
+      "step": 35800
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.3134184533847266e-05,
+      "loss": 0.2399,
+      "step": 35810
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.314622982413876e-05,
+      "loss": 0.1969,
+      "step": 35820
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.3158275114430256e-05,
+      "loss": 0.1774,
+      "step": 35830
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.317032040472175e-05,
+      "loss": 0.0564,
+      "step": 35840
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.318236569501325e-05,
+      "loss": 0.2285,
+      "step": 35850
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.319441098530474e-05,
+      "loss": 0.353,
+      "step": 35860
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.3206456275596244e-05,
+      "loss": 0.0899,
+      "step": 35870
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.3218501565887746e-05,
+      "loss": 0.3742,
+      "step": 35880
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.323054685617924e-05,
+      "loss": 0.0273,
+      "step": 35890
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.3242592146470736e-05,
+      "loss": 0.363,
+      "step": 35900
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.325463743676223e-05,
+      "loss": 0.2002,
+      "step": 35910
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.3266682727053726e-05,
+      "loss": 0.1672,
+      "step": 35920
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.327872801734522e-05,
+      "loss": 0.5501,
+      "step": 35930
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.329077330763672e-05,
+      "loss": 0.169,
+      "step": 35940
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.330281859792821e-05,
+      "loss": 0.2726,
+      "step": 35950
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.331486388821971e-05,
+      "loss": 0.2268,
+      "step": 35960
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.33269091785112e-05,
+      "loss": 0.2138,
+      "step": 35970
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.33389544688027e-05,
+      "loss": 0.1802,
+      "step": 35980
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.33509997590942e-05,
+      "loss": 0.2475,
+      "step": 35990
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.3363045049385695e-05,
+      "loss": 0.2541,
+      "step": 36000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.337509033967719e-05,
+      "loss": 0.1657,
+      "step": 36010
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.3387135629968685e-05,
+      "loss": 0.2851,
+      "step": 36020
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.339918092026018e-05,
+      "loss": 0.2052,
+      "step": 36030
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.3411226210551675e-05,
+      "loss": 0.0756,
+      "step": 36040
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.342327150084317e-05,
+      "loss": 0.1689,
+      "step": 36050
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.3435316791134666e-05,
+      "loss": 0.3548,
+      "step": 36060
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.344736208142616e-05,
+      "loss": 0.3876,
+      "step": 36070
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.3459407371717656e-05,
+      "loss": 0.2404,
+      "step": 36080
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.347145266200915e-05,
+      "loss": 0.2429,
+      "step": 36090
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.348349795230065e-05,
+      "loss": 0.1467,
+      "step": 36100
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.349554324259215e-05,
+      "loss": 0.1978,
+      "step": 36110
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.3507588532883644e-05,
+      "loss": 0.0458,
+      "step": 36120
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.3519633823175146e-05,
+      "loss": 0.2649,
+      "step": 36130
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.353167911346664e-05,
+      "loss": 0.2076,
+      "step": 36140
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.3543724403758136e-05,
+      "loss": 0.2791,
+      "step": 36150
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.355576969404963e-05,
+      "loss": 0.3706,
+      "step": 36160
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.3567814984341126e-05,
+      "loss": 0.075,
+      "step": 36170
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.357986027463262e-05,
+      "loss": 0.103,
+      "step": 36180
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.359190556492412e-05,
+      "loss": 0.2825,
+      "step": 36190
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.360395085521561e-05,
+      "loss": 0.2031,
+      "step": 36200
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.361599614550711e-05,
+      "loss": 0.1804,
+      "step": 36210
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.36280414357986e-05,
+      "loss": 0.0889,
+      "step": 36220
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.3640086726090104e-05,
+      "loss": 0.3436,
+      "step": 36230
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.36521320163816e-05,
+      "loss": 0.238,
+      "step": 36240
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.3664177306673095e-05,
+      "loss": 0.1423,
+      "step": 36250
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.367622259696459e-05,
+      "loss": 0.163,
+      "step": 36260
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.3688267887256085e-05,
+      "loss": 0.2892,
+      "step": 36270
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.370031317754758e-05,
+      "loss": 0.0762,
+      "step": 36280
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.3712358467839075e-05,
+      "loss": 0.0243,
+      "step": 36290
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.372440375813057e-05,
+      "loss": 0.0597,
+      "step": 36300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.3736449048422066e-05,
+      "loss": 0.1657,
+      "step": 36310
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.374849433871356e-05,
+      "loss": 0.3095,
+      "step": 36320
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.376053962900506e-05,
+      "loss": 0.1885,
+      "step": 36330
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.377258491929656e-05,
+      "loss": 0.4176,
+      "step": 36340
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.378463020958805e-05,
+      "loss": 0.5948,
+      "step": 36350
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.379667549987955e-05,
+      "loss": 0.2361,
+      "step": 36360
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.3808720790171044e-05,
+      "loss": 0.462,
+      "step": 36370
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.382076608046254e-05,
+      "loss": 0.1007,
+      "step": 36380
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.383281137075404e-05,
+      "loss": 0.2693,
+      "step": 36390
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.3844856661045536e-05,
+      "loss": 0.243,
+      "step": 36400
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.385690195133703e-05,
+      "loss": 0.2185,
+      "step": 36410
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.3868947241628526e-05,
+      "loss": 0.1573,
+      "step": 36420
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.388099253192002e-05,
+      "loss": 0.361,
+      "step": 36430
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.389303782221152e-05,
+      "loss": 0.3461,
+      "step": 36440
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.390508311250301e-05,
+      "loss": 0.2247,
+      "step": 36450
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.3917128402794514e-05,
+      "loss": 0.1405,
+      "step": 36460
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.392917369308601e-05,
+      "loss": 0.2074,
+      "step": 36470
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.3941218983377504e-05,
+      "loss": 0.2225,
+      "step": 36480
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.3953264273669e-05,
+      "loss": 0.116,
+      "step": 36490
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.3965309563960495e-05,
+      "loss": 0.4674,
+      "step": 36500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.397735485425199e-05,
+      "loss": 0.2041,
+      "step": 36510
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.3989400144543485e-05,
+      "loss": 0.2481,
+      "step": 36520
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.400144543483498e-05,
+      "loss": 0.1545,
+      "step": 36530
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.4013490725126475e-05,
+      "loss": 0.217,
+      "step": 36540
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.402553601541797e-05,
+      "loss": 0.2324,
+      "step": 36550
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.4037581305709466e-05,
+      "loss": 0.3543,
+      "step": 36560
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.404962659600097e-05,
+      "loss": 0.429,
+      "step": 36570
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.406167188629246e-05,
+      "loss": 0.3969,
+      "step": 36580
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.407371717658396e-05,
+      "loss": 0.1568,
+      "step": 36590
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.408576246687545e-05,
+      "loss": 0.3524,
+      "step": 36600
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.409780775716695e-05,
+      "loss": 0.2715,
+      "step": 36610
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.4109853047458444e-05,
+      "loss": 0.4011,
+      "step": 36620
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.412189833774994e-05,
+      "loss": 0.3286,
+      "step": 36630
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.4133943628041434e-05,
+      "loss": 0.2508,
+      "step": 36640
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.4145988918332936e-05,
+      "loss": 0.1096,
+      "step": 36650
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.415803420862443e-05,
+      "loss": 0.2304,
+      "step": 36660
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.4170079498915926e-05,
+      "loss": 0.3207,
+      "step": 36670
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.418212478920742e-05,
+      "loss": 0.3985,
+      "step": 36680
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.419417007949892e-05,
+      "loss": 0.1007,
+      "step": 36690
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.420621536979042e-05,
+      "loss": 0.2095,
+      "step": 36700
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.4218260660081914e-05,
+      "loss": 0.2333,
+      "step": 36710
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.423030595037341e-05,
+      "loss": 0.1411,
+      "step": 36720
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.4242351240664904e-05,
+      "loss": 0.1767,
+      "step": 36730
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.42543965309564e-05,
+      "loss": 0.321,
+      "step": 36740
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.4266441821247895e-05,
+      "loss": 0.3621,
+      "step": 36750
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.427848711153939e-05,
+      "loss": 0.2616,
+      "step": 36760
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.4290532401830885e-05,
+      "loss": 0.1627,
+      "step": 36770
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.430257769212238e-05,
+      "loss": 0.2903,
+      "step": 36780
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.4314622982413875e-05,
+      "loss": 0.4234,
+      "step": 36790
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.432666827270538e-05,
+      "loss": 0.2649,
+      "step": 36800
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.433871356299687e-05,
+      "loss": 0.4491,
+      "step": 36810
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.435075885328837e-05,
+      "loss": 0.4532,
+      "step": 36820
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.436280414357986e-05,
+      "loss": 0.1327,
+      "step": 36830
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.437484943387136e-05,
+      "loss": 0.3027,
+      "step": 36840
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.438689472416285e-05,
+      "loss": 0.1204,
+      "step": 36850
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.439894001445435e-05,
+      "loss": 0.1811,
+      "step": 36860
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.4410985304745844e-05,
+      "loss": 0.2029,
+      "step": 36870
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.442303059503734e-05,
+      "loss": 0.3184,
+      "step": 36880
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.4435075885328834e-05,
+      "loss": 0.1609,
+      "step": 36890
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.444712117562033e-05,
+      "loss": 0.2699,
+      "step": 36900
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.445916646591183e-05,
+      "loss": 0.354,
+      "step": 36910
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.4471211756203326e-05,
+      "loss": 0.1589,
+      "step": 36920
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.448325704649483e-05,
+      "loss": 0.3103,
+      "step": 36930
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.4495302336786323e-05,
+      "loss": 0.2306,
+      "step": 36940
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.450734762707782e-05,
+      "loss": 0.1168,
+      "step": 36950
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.4519392917369314e-05,
+      "loss": 0.3482,
+      "step": 36960
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.453143820766081e-05,
+      "loss": 0.3636,
+      "step": 36970
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.4543483497952304e-05,
+      "loss": 0.3062,
+      "step": 36980
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.45555287882438e-05,
+      "loss": 0.2258,
+      "step": 36990
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.4567574078535295e-05,
+      "loss": 0.1115,
+      "step": 37000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.457961936882679e-05,
+      "loss": 0.3165,
+      "step": 37010
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.4591664659118285e-05,
+      "loss": 0.0567,
+      "step": 37020
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.460370994940978e-05,
+      "loss": 0.2371,
+      "step": 37030
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.461575523970128e-05,
+      "loss": 0.2198,
+      "step": 37040
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.462780052999278e-05,
+      "loss": 0.1162,
+      "step": 37050
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.463984582028427e-05,
+      "loss": 0.4547,
+      "step": 37060
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.465189111057577e-05,
+      "loss": 0.3516,
+      "step": 37070
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.466393640086726e-05,
+      "loss": 0.2908,
+      "step": 37080
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.467598169115876e-05,
+      "loss": 0.3206,
+      "step": 37090
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.468802698145025e-05,
+      "loss": 0.315,
+      "step": 37100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.470007227174175e-05,
+      "loss": 0.2428,
+      "step": 37110
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.4712117562033244e-05,
+      "loss": 0.33,
+      "step": 37120
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.472416285232474e-05,
+      "loss": 0.2741,
+      "step": 37130
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.4736208142616234e-05,
+      "loss": 0.2572,
+      "step": 37140
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.4748253432907736e-05,
+      "loss": 0.3289,
+      "step": 37150
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.476029872319923e-05,
+      "loss": 0.1985,
+      "step": 37160
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.4772344013490726e-05,
+      "loss": 0.315,
+      "step": 37170
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.478438930378223e-05,
+      "loss": 0.4034,
+      "step": 37180
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.479643459407372e-05,
+      "loss": 0.3571,
+      "step": 37190
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.480847988436522e-05,
+      "loss": 0.0749,
+      "step": 37200
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.4820525174656714e-05,
+      "loss": 0.118,
+      "step": 37210
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.483257046494821e-05,
+      "loss": 0.3522,
+      "step": 37220
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.4844615755239704e-05,
+      "loss": 0.1337,
+      "step": 37230
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.48566610455312e-05,
+      "loss": 0.094,
+      "step": 37240
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.4868706335822694e-05,
+      "loss": 0.1713,
+      "step": 37250
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.488075162611419e-05,
+      "loss": 0.2844,
+      "step": 37260
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.489279691640569e-05,
+      "loss": 0.2115,
+      "step": 37270
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.490484220669719e-05,
+      "loss": 0.1393,
+      "step": 37280
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.491688749698868e-05,
+      "loss": 0.3343,
+      "step": 37290
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.492893278728018e-05,
+      "loss": 0.2221,
+      "step": 37300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.494097807757167e-05,
+      "loss": 0.1967,
+      "step": 37310
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.495302336786317e-05,
+      "loss": 0.3437,
+      "step": 37320
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.496506865815466e-05,
+      "loss": 0.3612,
+      "step": 37330
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.497711394844616e-05,
+      "loss": 0.0685,
+      "step": 37340
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.498915923873765e-05,
+      "loss": 0.3629,
+      "step": 37350
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.500120452902915e-05,
+      "loss": 0.2345,
+      "step": 37360
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.5013249819320643e-05,
+      "loss": 0.2778,
+      "step": 37370
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.5025295109612145e-05,
+      "loss": 0.2619,
+      "step": 37380
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.503734039990364e-05,
+      "loss": 0.335,
+      "step": 37390
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.5049385690195136e-05,
+      "loss": 0.0732,
+      "step": 37400
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.506143098048663e-05,
+      "loss": 0.3391,
+      "step": 37410
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.5073476270778126e-05,
+      "loss": 0.2749,
+      "step": 37420
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.508552156106962e-05,
+      "loss": 0.2605,
+      "step": 37430
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.509756685136112e-05,
+      "loss": 0.2746,
+      "step": 37440
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.510961214165262e-05,
+      "loss": 0.2604,
+      "step": 37450
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.5121657431944114e-05,
+      "loss": 0.1286,
+      "step": 37460
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.513370272223561e-05,
+      "loss": 0.2665,
+      "step": 37470
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.5145748012527104e-05,
+      "loss": 0.2659,
+      "step": 37480
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.51577933028186e-05,
+      "loss": 0.2172,
+      "step": 37490
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.5169838593110094e-05,
+      "loss": 0.0289,
+      "step": 37500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.5181883883401596e-05,
+      "loss": 0.3102,
+      "step": 37510
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.519392917369309e-05,
+      "loss": 0.2108,
+      "step": 37520
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.520597446398459e-05,
+      "loss": 0.1933,
+      "step": 37530
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.521801975427608e-05,
+      "loss": 0.0755,
+      "step": 37540
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.523006504456758e-05,
+      "loss": 0.3773,
+      "step": 37550
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.524211033485907e-05,
+      "loss": 0.1239,
+      "step": 37560
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.525415562515057e-05,
+      "loss": 0.1945,
+      "step": 37570
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.526620091544206e-05,
+      "loss": 0.4271,
+      "step": 37580
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.527824620573356e-05,
+      "loss": 0.3711,
+      "step": 37590
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.529029149602505e-05,
+      "loss": 0.3843,
+      "step": 37600
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.530233678631655e-05,
+      "loss": 0.2226,
+      "step": 37610
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.531438207660805e-05,
+      "loss": 0.1228,
+      "step": 37620
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5326427366899545e-05,
+      "loss": 0.1772,
+      "step": 37630
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.533847265719104e-05,
+      "loss": 0.2186,
+      "step": 37640
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5350517947482536e-05,
+      "loss": 0.2142,
+      "step": 37650
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.536256323777403e-05,
+      "loss": 0.302,
+      "step": 37660
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5374608528065526e-05,
+      "loss": 0.4555,
+      "step": 37670
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.538665381835702e-05,
+      "loss": 0.228,
+      "step": 37680
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5398699108648516e-05,
+      "loss": 0.2125,
+      "step": 37690
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.541074439894001e-05,
+      "loss": 0.216,
+      "step": 37700
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5422789689231514e-05,
+      "loss": 0.2761,
+      "step": 37710
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.543483497952301e-05,
+      "loss": 0.1678,
+      "step": 37720
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5446880269814504e-05,
+      "loss": 0.4802,
+      "step": 37730
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.5458925560106006e-05,
+      "loss": 0.2093,
+      "step": 37740
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.54709708503975e-05,
+      "loss": 0.1337,
+      "step": 37750
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.5483016140688996e-05,
+      "loss": 0.2551,
+      "step": 37760
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.549506143098049e-05,
+      "loss": 0.1871,
+      "step": 37770
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.550710672127199e-05,
+      "loss": 0.2901,
+      "step": 37780
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.551915201156348e-05,
+      "loss": 0.3214,
+      "step": 37790
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.553119730185498e-05,
+      "loss": 0.0989,
+      "step": 37800
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.554324259214647e-05,
+      "loss": 0.2702,
+      "step": 37810
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.555528788243797e-05,
+      "loss": 0.3401,
+      "step": 37820
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.556733317272946e-05,
+      "loss": 0.3097,
+      "step": 37830
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.557937846302096e-05,
+      "loss": 0.4212,
+      "step": 37840
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.559142375331246e-05,
+      "loss": 0.1121,
+      "step": 37850
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.5603469043603955e-05,
+      "loss": 0.2799,
+      "step": 37860
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.561551433389545e-05,
+      "loss": 0.1932,
+      "step": 37870
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.5627559624186945e-05,
+      "loss": 0.1169,
+      "step": 37880
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.563960491447844e-05,
+      "loss": 0.4246,
+      "step": 37890
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.5651650204769936e-05,
+      "loss": 0.1919,
+      "step": 37900
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.566369549506143e-05,
+      "loss": 0.3077,
+      "step": 37910
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.5675740785352926e-05,
+      "loss": 0.0646,
+      "step": 37920
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.568778607564442e-05,
+      "loss": 0.1999,
+      "step": 37930
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.5699831365935916e-05,
+      "loss": 0.2756,
+      "step": 37940
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.571187665622741e-05,
+      "loss": 0.2845,
+      "step": 37950
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.5723921946518914e-05,
+      "loss": 0.1914,
+      "step": 37960
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.573596723681041e-05,
+      "loss": 0.116,
+      "step": 37970
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.574801252710191e-05,
+      "loss": 0.323,
+      "step": 37980
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.5760057817393406e-05,
+      "loss": 0.2239,
+      "step": 37990
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.57721031076849e-05,
+      "loss": 0.2846,
+      "step": 38000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.5784148397976396e-05,
+      "loss": 0.2737,
+      "step": 38010
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.579619368826789e-05,
+      "loss": 0.2889,
+      "step": 38020
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.580823897855939e-05,
+      "loss": 0.072,
+      "step": 38030
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.582028426885088e-05,
+      "loss": 0.3168,
+      "step": 38040
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.583232955914238e-05,
+      "loss": 0.3255,
+      "step": 38050
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.584437484943387e-05,
+      "loss": 0.2995,
+      "step": 38060
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.585642013972537e-05,
+      "loss": 0.4198,
+      "step": 38070
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.586846543001686e-05,
+      "loss": 0.291,
+      "step": 38080
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.5880510720308365e-05,
+      "loss": 0.3397,
+      "step": 38090
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.589255601059986e-05,
+      "loss": 0.1235,
+      "step": 38100
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.5904601300891355e-05,
+      "loss": 0.2612,
+      "step": 38110
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.591664659118285e-05,
+      "loss": 0.4197,
+      "step": 38120
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.5928691881474345e-05,
+      "loss": 0.1137,
+      "step": 38130
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.594073717176584e-05,
+      "loss": 0.3811,
+      "step": 38140
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.5952782462057336e-05,
+      "loss": 0.3386,
+      "step": 38150
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.596482775234883e-05,
+      "loss": 0.3137,
+      "step": 38160
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.5976873042640326e-05,
+      "loss": 0.1106,
+      "step": 38170
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.598891833293182e-05,
+      "loss": 0.2022,
+      "step": 38180
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.600096362322332e-05,
+      "loss": 0.2309,
+      "step": 38190
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.601300891351482e-05,
+      "loss": 0.218,
+      "step": 38200
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.6025054203806314e-05,
+      "loss": 0.294,
+      "step": 38210
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.603709949409781e-05,
+      "loss": 0.2929,
+      "step": 38220
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.604914478438931e-05,
+      "loss": 0.2363,
+      "step": 38230
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.6061190074680806e-05,
+      "loss": 0.2947,
+      "step": 38240
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.60732353649723e-05,
+      "loss": 0.178,
+      "step": 38250
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.6085280655263796e-05,
+      "loss": 0.3071,
+      "step": 38260
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.609732594555529e-05,
+      "loss": 0.1764,
+      "step": 38270
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.610937123584679e-05,
+      "loss": 0.2749,
+      "step": 38280
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.612141652613828e-05,
+      "loss": 0.1701,
+      "step": 38290
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.613346181642978e-05,
+      "loss": 0.2564,
+      "step": 38300
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.614550710672127e-05,
+      "loss": 0.256,
+      "step": 38310
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.6157552397012774e-05,
+      "loss": 0.358,
+      "step": 38320
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.616959768730427e-05,
+      "loss": 0.1548,
+      "step": 38330
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.6181642977595765e-05,
+      "loss": 0.1264,
+      "step": 38340
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.619368826788726e-05,
+      "loss": 0.3961,
+      "step": 38350
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.6205733558178755e-05,
+      "loss": 0.2676,
+      "step": 38360
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.621777884847025e-05,
+      "loss": 0.3904,
+      "step": 38370
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.6229824138761745e-05,
+      "loss": 0.3816,
+      "step": 38380
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.624186942905324e-05,
+      "loss": 0.2795,
+      "step": 38390
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.6253914719344736e-05,
+      "loss": 0.1814,
+      "step": 38400
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.626596000963623e-05,
+      "loss": 0.3808,
+      "step": 38410
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.6278005299927726e-05,
+      "loss": 0.3767,
+      "step": 38420
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.629005059021923e-05,
+      "loss": 0.2624,
+      "step": 38430
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.630209588051072e-05,
+      "loss": 0.206,
+      "step": 38440
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.631414117080222e-05,
+      "loss": 0.1811,
+      "step": 38450
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.6326186461093713e-05,
+      "loss": 0.1415,
+      "step": 38460
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.633823175138521e-05,
+      "loss": 0.1309,
+      "step": 38470
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.6350277041676704e-05,
+      "loss": 0.1874,
+      "step": 38480
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.6362322331968206e-05,
+      "loss": 0.5427,
+      "step": 38490
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.63743676222597e-05,
+      "loss": 0.0691,
+      "step": 38500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.6386412912551196e-05,
+      "loss": 0.1615,
+      "step": 38510
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.639845820284269e-05,
+      "loss": 0.3116,
+      "step": 38520
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.6410503493134187e-05,
+      "loss": 0.2539,
+      "step": 38530
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.642254878342568e-05,
+      "loss": 0.2328,
+      "step": 38540
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.643459407371718e-05,
+      "loss": 0.1626,
+      "step": 38550
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.644663936400868e-05,
+      "loss": 0.1412,
+      "step": 38560
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.6458684654300174e-05,
+      "loss": 0.1373,
+      "step": 38570
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.647072994459167e-05,
+      "loss": 0.325,
+      "step": 38580
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.6482775234883164e-05,
+      "loss": 0.2564,
+      "step": 38590
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.649482052517466e-05,
+      "loss": 0.1454,
+      "step": 38600
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.6506865815466155e-05,
+      "loss": 0.3848,
+      "step": 38610
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.651891110575765e-05,
+      "loss": 0.0814,
+      "step": 38620
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.6530956396049145e-05,
+      "loss": 0.1807,
+      "step": 38630
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.654300168634064e-05,
+      "loss": 0.159,
+      "step": 38640
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.6555046976632136e-05,
+      "loss": 0.384,
+      "step": 38650
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.656709226692364e-05,
+      "loss": 0.2639,
+      "step": 38660
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.657913755721513e-05,
+      "loss": 0.2527,
+      "step": 38670
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.659118284750663e-05,
+      "loss": 0.1406,
+      "step": 38680
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.660322813779812e-05,
+      "loss": 0.2205,
+      "step": 38690
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.661527342808962e-05,
+      "loss": 0.5083,
+      "step": 38700
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.6627318718381113e-05,
+      "loss": 0.2509,
+      "step": 38710
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.663936400867261e-05,
+      "loss": 0.3641,
+      "step": 38720
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.6651409298964104e-05,
+      "loss": 0.3631,
+      "step": 38730
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.66634545892556e-05,
+      "loss": 0.2473,
+      "step": 38740
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.6675499879547094e-05,
+      "loss": 0.381,
+      "step": 38750
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.6687545169838596e-05,
+      "loss": 0.1776,
+      "step": 38760
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.669959046013009e-05,
+      "loss": 0.3014,
+      "step": 38770
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.6711635750421587e-05,
+      "loss": 0.2411,
+      "step": 38780
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.672368104071309e-05,
+      "loss": 0.2537,
+      "step": 38790
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.6735726331004584e-05,
+      "loss": 0.2217,
+      "step": 38800
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.674777162129608e-05,
+      "loss": 0.1692,
+      "step": 38810
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.6759816911587574e-05,
+      "loss": 0.3222,
+      "step": 38820
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.677186220187907e-05,
+      "loss": 0.2248,
+      "step": 38830
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.6783907492170564e-05,
+      "loss": 0.4162,
+      "step": 38840
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.679595278246206e-05,
+      "loss": 0.1155,
+      "step": 38850
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.6807998072753555e-05,
+      "loss": 0.3284,
+      "step": 38860
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.682004336304505e-05,
+      "loss": 0.4412,
+      "step": 38870
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.6832088653336545e-05,
+      "loss": 0.2969,
+      "step": 38880
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.684413394362804e-05,
+      "loss": 0.3358,
+      "step": 38890
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.685617923391954e-05,
+      "loss": 0.2485,
+      "step": 38900
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.686822452421104e-05,
+      "loss": 0.2555,
+      "step": 38910
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.688026981450253e-05,
+      "loss": 0.1909,
+      "step": 38920
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.689231510479403e-05,
+      "loss": 0.2342,
+      "step": 38930
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.690436039508552e-05,
+      "loss": 0.1791,
+      "step": 38940
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.691640568537702e-05,
+      "loss": 0.2582,
+      "step": 38950
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.6928450975668513e-05,
+      "loss": 0.114,
+      "step": 38960
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.694049626596001e-05,
+      "loss": 0.4115,
+      "step": 38970
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.6952541556251504e-05,
+      "loss": 0.3574,
+      "step": 38980
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.6964586846543e-05,
+      "loss": 0.2493,
+      "step": 38990
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.6976632136834494e-05,
+      "loss": 0.0805,
+      "step": 39000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.6988677427125996e-05,
+      "loss": 0.3367,
+      "step": 39010
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.700072271741749e-05,
+      "loss": 0.4513,
+      "step": 39020
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.701276800770899e-05,
+      "loss": 0.3846,
+      "step": 39030
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.702481329800049e-05,
+      "loss": 0.1616,
+      "step": 39040
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7036858588291984e-05,
+      "loss": 0.3118,
+      "step": 39050
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.704890387858348e-05,
+      "loss": 0.1291,
+      "step": 39060
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7060949168874974e-05,
+      "loss": 0.5088,
+      "step": 39070
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.707299445916647e-05,
+      "loss": 0.2192,
+      "step": 39080
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7085039749457964e-05,
+      "loss": 0.3907,
+      "step": 39090
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.709708503974946e-05,
+      "loss": 0.1095,
+      "step": 39100
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7109130330040955e-05,
+      "loss": 0.3246,
+      "step": 39110
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.712117562033245e-05,
+      "loss": 0.2343,
+      "step": 39120
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.713322091062395e-05,
+      "loss": 0.1757,
+      "step": 39130
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.714526620091545e-05,
+      "loss": 0.1963,
+      "step": 39140
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.715731149120694e-05,
+      "loss": 0.3622,
+      "step": 39150
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.716935678149844e-05,
+      "loss": 0.3445,
+      "step": 39160
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.718140207178993e-05,
+      "loss": 0.1758,
+      "step": 39170
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.719344736208143e-05,
+      "loss": 0.3365,
+      "step": 39180
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.720549265237292e-05,
+      "loss": 0.2765,
+      "step": 39190
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.721753794266442e-05,
+      "loss": 0.2606,
+      "step": 39200
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.722958323295591e-05,
+      "loss": 0.1623,
+      "step": 39210
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.724162852324741e-05,
+      "loss": 0.2381,
+      "step": 39220
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.7253673813538904e-05,
+      "loss": 0.1543,
+      "step": 39230
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.7265719103830406e-05,
+      "loss": 0.2891,
+      "step": 39240
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.72777643941219e-05,
+      "loss": 0.3069,
+      "step": 39250
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.7289809684413396e-05,
+      "loss": 0.0911,
+      "step": 39260
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.730185497470489e-05,
+      "loss": 0.3672,
+      "step": 39270
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.731390026499639e-05,
+      "loss": 0.4102,
+      "step": 39280
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.732594555528789e-05,
+      "loss": 0.1742,
+      "step": 39290
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.7337990845579384e-05,
+      "loss": 0.4201,
+      "step": 39300
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.735003613587088e-05,
+      "loss": 0.4159,
+      "step": 39310
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.7362081426162374e-05,
+      "loss": 0.2523,
+      "step": 39320
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.737412671645387e-05,
+      "loss": 0.2951,
+      "step": 39330
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.7386172006745364e-05,
+      "loss": 0.1453,
+      "step": 39340
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.739821729703686e-05,
+      "loss": 0.188,
+      "step": 39350
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.7410262587328355e-05,
+      "loss": 0.2381,
+      "step": 39360
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.742230787761986e-05,
+      "loss": 0.3507,
+      "step": 39370
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.743435316791135e-05,
+      "loss": 0.2198,
+      "step": 39380
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.744639845820285e-05,
+      "loss": 0.2701,
+      "step": 39390
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.745844374849434e-05,
+      "loss": 0.0925,
+      "step": 39400
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.747048903878584e-05,
+      "loss": 0.17,
+      "step": 39410
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.748253432907733e-05,
+      "loss": 0.404,
+      "step": 39420
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.749457961936883e-05,
+      "loss": 0.3197,
+      "step": 39430
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.750662490966032e-05,
+      "loss": 0.4805,
+      "step": 39440
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.751867019995182e-05,
+      "loss": 0.2292,
+      "step": 39450
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.753071549024331e-05,
+      "loss": 0.1112,
+      "step": 39460
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.754276078053481e-05,
+      "loss": 0.1421,
+      "step": 39470
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.755480607082631e-05,
+      "loss": 0.2773,
+      "step": 39480
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.7566851361117806e-05,
+      "loss": 0.238,
+      "step": 39490
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.75788966514093e-05,
+      "loss": 0.4375,
+      "step": 39500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.7590941941700796e-05,
+      "loss": 0.2647,
+      "step": 39510
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.760298723199229e-05,
+      "loss": 0.3542,
+      "step": 39520
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.7615032522283786e-05,
+      "loss": 0.2566,
+      "step": 39530
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.762707781257529e-05,
+      "loss": 0.0921,
+      "step": 39540
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.7639123102866784e-05,
+      "loss": 0.0273,
+      "step": 39550
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.765116839315828e-05,
+      "loss": 0.1625,
+      "step": 39560
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.7663213683449774e-05,
+      "loss": 0.1007,
+      "step": 39570
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.767525897374127e-05,
+      "loss": 0.2033,
+      "step": 39580
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.7687304264032764e-05,
+      "loss": 0.2559,
+      "step": 39590
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.7699349554324266e-05,
+      "loss": 0.0627,
+      "step": 39600
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.771139484461576e-05,
+      "loss": 0.1555,
+      "step": 39610
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.7723440134907257e-05,
+      "loss": 0.284,
+      "step": 39620
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.773548542519875e-05,
+      "loss": 0.2138,
+      "step": 39630
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.774753071549025e-05,
+      "loss": 0.3323,
+      "step": 39640
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.775957600578174e-05,
+      "loss": 0.2948,
+      "step": 39650
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.777162129607324e-05,
+      "loss": 0.3255,
+      "step": 39660
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.778366658636473e-05,
+      "loss": 0.3457,
+      "step": 39670
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.779571187665623e-05,
+      "loss": 0.1822,
+      "step": 39680
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.780775716694772e-05,
+      "loss": 0.3229,
+      "step": 39690
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.781980245723922e-05,
+      "loss": 0.1927,
+      "step": 39700
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.783184774753072e-05,
+      "loss": 0.1177,
+      "step": 39710
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.7843893037822215e-05,
+      "loss": 0.1644,
+      "step": 39720
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.785593832811371e-05,
+      "loss": 0.3014,
+      "step": 39730
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.7867983618405206e-05,
+      "loss": 0.2359,
+      "step": 39740
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.78800289086967e-05,
+      "loss": 0.3562,
+      "step": 39750
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.7892074198988196e-05,
+      "loss": 0.4367,
+      "step": 39760
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.790411948927969e-05,
+      "loss": 0.4191,
+      "step": 39770
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.7916164779571186e-05,
+      "loss": 0.2267,
+      "step": 39780
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.792821006986268e-05,
+      "loss": 0.1633,
+      "step": 39790
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.794025536015418e-05,
+      "loss": 0.1223,
+      "step": 39800
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.795230065044568e-05,
+      "loss": 0.2787,
+      "step": 39810
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.7964345940737174e-05,
+      "loss": 0.2278,
+      "step": 39820
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.797639123102867e-05,
+      "loss": 0.1735,
+      "step": 39830
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.798843652132017e-05,
+      "loss": 0.1615,
+      "step": 39840
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.8000481811611666e-05,
+      "loss": 0.3231,
+      "step": 39850
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.801252710190316e-05,
+      "loss": 0.2421,
+      "step": 39860
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.8024572392194657e-05,
+      "loss": 0.1981,
+      "step": 39870
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.803661768248615e-05,
+      "loss": 0.2372,
+      "step": 39880
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.804866297277765e-05,
+      "loss": 0.3957,
+      "step": 39890
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.806070826306914e-05,
+      "loss": 0.1762,
+      "step": 39900
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.807275355336064e-05,
+      "loss": 0.2971,
+      "step": 39910
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.808479884365213e-05,
+      "loss": 0.3588,
+      "step": 39920
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.809684413394363e-05,
+      "loss": 0.4163,
+      "step": 39930
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.810888942423513e-05,
+      "loss": 0.2094,
+      "step": 39940
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.8120934714526625e-05,
+      "loss": 0.4248,
+      "step": 39950
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.813298000481812e-05,
+      "loss": 0.3472,
+      "step": 39960
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.8145025295109615e-05,
+      "loss": 0.3238,
+      "step": 39970
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.815707058540111e-05,
+      "loss": 0.1751,
+      "step": 39980
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.8169115875692606e-05,
+      "loss": 0.21,
+      "step": 39990
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.81811611659841e-05,
+      "loss": 0.2921,
+      "step": 40000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.8193206456275596e-05,
+      "loss": 0.3349,
+      "step": 40010
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.820525174656709e-05,
+      "loss": 0.2892,
+      "step": 40020
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.8217297036858586e-05,
+      "loss": 0.1223,
+      "step": 40030
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.822934232715008e-05,
+      "loss": 0.3061,
+      "step": 40040
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.8241387617441583e-05,
+      "loss": 0.1247,
+      "step": 40050
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.825343290773308e-05,
+      "loss": 0.1978,
+      "step": 40060
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.826547819802458e-05,
+      "loss": 0.3971,
+      "step": 40070
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.8277523488316076e-05,
+      "loss": 0.3181,
+      "step": 40080
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.828956877860757e-05,
+      "loss": 0.2885,
+      "step": 40090
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.8301614068899066e-05,
+      "loss": 0.3062,
+      "step": 40100
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.831365935919056e-05,
+      "loss": 0.1994,
+      "step": 40110
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.8325704649482056e-05,
+      "loss": 0.1717,
+      "step": 40120
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.833774993977355e-05,
+      "loss": 0.3691,
+      "step": 40130
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.834979523006505e-05,
+      "loss": 0.2178,
+      "step": 40140
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.836184052035654e-05,
+      "loss": 0.5002,
+      "step": 40150
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.837388581064804e-05,
+      "loss": 0.1747,
+      "step": 40160
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.838593110093953e-05,
+      "loss": 0.2908,
+      "step": 40170
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.8397976391231034e-05,
+      "loss": 0.0878,
+      "step": 40180
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.841002168152253e-05,
+      "loss": 0.218,
+      "step": 40190
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.8422066971814025e-05,
+      "loss": 0.1294,
+      "step": 40200
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.843411226210552e-05,
+      "loss": 0.2163,
+      "step": 40210
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.8446157552397015e-05,
+      "loss": 0.0881,
+      "step": 40220
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.845820284268851e-05,
+      "loss": 0.2322,
+      "step": 40230
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8470248132980005e-05,
+      "loss": 0.144,
+      "step": 40240
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.84822934232715e-05,
+      "loss": 0.232,
+      "step": 40250
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8494338713562996e-05,
+      "loss": 0.323,
+      "step": 40260
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.850638400385449e-05,
+      "loss": 0.2229,
+      "step": 40270
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8518429294145986e-05,
+      "loss": 0.2213,
+      "step": 40280
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.853047458443749e-05,
+      "loss": 0.1944,
+      "step": 40290
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.854251987472898e-05,
+      "loss": 0.3094,
+      "step": 40300
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.855456516502048e-05,
+      "loss": 0.3185,
+      "step": 40310
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.8566610455311974e-05,
+      "loss": 0.3551,
+      "step": 40320
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.8578655745603476e-05,
+      "loss": 0.2517,
+      "step": 40330
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.859070103589497e-05,
+      "loss": 0.0906,
+      "step": 40340
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.8602746326186466e-05,
+      "loss": 0.1214,
+      "step": 40350
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.861479161647796e-05,
+      "loss": 0.253,
+      "step": 40360
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.8626836906769456e-05,
+      "loss": 0.2106,
+      "step": 40370
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.863888219706095e-05,
+      "loss": 0.3937,
+      "step": 40380
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.865092748735245e-05,
+      "loss": 0.2631,
+      "step": 40390
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.866297277764394e-05,
+      "loss": 0.3241,
+      "step": 40400
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.8675018067935444e-05,
+      "loss": 0.3431,
+      "step": 40410
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.868706335822694e-05,
+      "loss": 0.2578,
+      "step": 40420
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.8699108648518434e-05,
+      "loss": 0.3506,
+      "step": 40430
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.871115393880993e-05,
+      "loss": 0.1138,
+      "step": 40440
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.8723199229101425e-05,
+      "loss": 0.489,
+      "step": 40450
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.873524451939292e-05,
+      "loss": 0.2473,
+      "step": 40460
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.8747289809684415e-05,
+      "loss": 0.1872,
+      "step": 40470
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.875933509997591e-05,
+      "loss": 0.098,
+      "step": 40480
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.8771380390267405e-05,
+      "loss": 0.2768,
+      "step": 40490
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.87834256805589e-05,
+      "loss": 0.4325,
+      "step": 40500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.8795470970850396e-05,
+      "loss": 0.242,
+      "step": 40510
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.88075162611419e-05,
+      "loss": 0.1788,
+      "step": 40520
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.881956155143339e-05,
+      "loss": 0.1835,
+      "step": 40530
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.883160684172489e-05,
+      "loss": 0.1613,
+      "step": 40540
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.884365213201638e-05,
+      "loss": 0.2692,
+      "step": 40550
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.885569742230788e-05,
+      "loss": 0.2826,
+      "step": 40560
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.8867742712599374e-05,
+      "loss": 0.1606,
+      "step": 40570
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.887978800289087e-05,
+      "loss": 0.2468,
+      "step": 40580
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.8891833293182364e-05,
+      "loss": 0.4099,
+      "step": 40590
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.8903878583473866e-05,
+      "loss": 0.1925,
+      "step": 40600
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.891592387376536e-05,
+      "loss": 0.3035,
+      "step": 40610
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.8927969164056856e-05,
+      "loss": 0.1511,
+      "step": 40620
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.894001445434835e-05,
+      "loss": 0.2644,
+      "step": 40630
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.895205974463985e-05,
+      "loss": 0.2407,
+      "step": 40640
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.896410503493135e-05,
+      "loss": 0.1441,
+      "step": 40650
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.8976150325222844e-05,
+      "loss": 0.3296,
+      "step": 40660
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.898819561551434e-05,
+      "loss": 0.2568,
+      "step": 40670
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.9000240905805834e-05,
+      "loss": 0.4292,
+      "step": 40680
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.901228619609733e-05,
+      "loss": 0.1919,
+      "step": 40690
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.9024331486388825e-05,
+      "loss": 0.3391,
+      "step": 40700
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.903637677668032e-05,
+      "loss": 0.3385,
+      "step": 40710
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.9048422066971815e-05,
+      "loss": 0.2143,
+      "step": 40720
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.906046735726331e-05,
+      "loss": 0.1824,
+      "step": 40730
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.9072512647554805e-05,
+      "loss": 0.3566,
+      "step": 40740
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.90845579378463e-05,
+      "loss": 0.3458,
+      "step": 40750
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.90966032281378e-05,
+      "loss": 0.3154,
+      "step": 40760
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.91086485184293e-05,
+      "loss": 0.3923,
+      "step": 40770
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.912069380872079e-05,
+      "loss": 0.1826,
+      "step": 40780
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.913273909901229e-05,
+      "loss": 0.1384,
+      "step": 40790
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.914478438930378e-05,
+      "loss": 0.2136,
+      "step": 40800
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.915682967959528e-05,
+      "loss": 0.2585,
+      "step": 40810
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.9168874969886774e-05,
+      "loss": 0.2256,
+      "step": 40820
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.918092026017827e-05,
+      "loss": 0.1375,
+      "step": 40830
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.9192965550469764e-05,
+      "loss": 0.0845,
+      "step": 40840
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.920501084076126e-05,
+      "loss": 0.1365,
+      "step": 40850
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.921705613105276e-05,
+      "loss": 0.1562,
+      "step": 40860
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.9229101421344256e-05,
+      "loss": 0.2378,
+      "step": 40870
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.924114671163576e-05,
+      "loss": 0.2446,
+      "step": 40880
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.9253192001927253e-05,
+      "loss": 0.3895,
+      "step": 40890
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.926523729221875e-05,
+      "loss": 0.2501,
+      "step": 40900
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.9277282582510244e-05,
+      "loss": 0.1819,
+      "step": 40910
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.928932787280174e-05,
+      "loss": 0.1762,
+      "step": 40920
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.9301373163093234e-05,
+      "loss": 0.1746,
+      "step": 40930
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.931341845338473e-05,
+      "loss": 0.2707,
+      "step": 40940
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.9325463743676225e-05,
+      "loss": 0.2249,
+      "step": 40950
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.933750903396772e-05,
+      "loss": 0.2353,
+      "step": 40960
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.9349554324259215e-05,
+      "loss": 0.2236,
+      "step": 40970
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.936159961455071e-05,
+      "loss": 0.1565,
+      "step": 40980
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.937364490484221e-05,
+      "loss": 0.0992,
+      "step": 40990
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.938569019513371e-05,
+      "loss": 0.0727,
+      "step": 41000
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.93977354854252e-05,
+      "loss": 0.1844,
+      "step": 41010
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.94097807757167e-05,
+      "loss": 0.3889,
+      "step": 41020
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.942182606600819e-05,
+      "loss": 0.3507,
+      "step": 41030
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.943387135629969e-05,
+      "loss": 0.2408,
+      "step": 41040
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.944591664659118e-05,
+      "loss": 0.2676,
+      "step": 41050
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.945796193688268e-05,
+      "loss": 0.1112,
+      "step": 41060
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.9470007227174174e-05,
+      "loss": 0.2481,
+      "step": 41070
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.948205251746567e-05,
+      "loss": 0.2062,
+      "step": 41080
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.9494097807757164e-05,
+      "loss": 0.2198,
+      "step": 41090
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.9506143098048666e-05,
+      "loss": 0.5001,
+      "step": 41100
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.951818838834016e-05,
+      "loss": 0.4397,
+      "step": 41110
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.953023367863166e-05,
+      "loss": 0.3584,
+      "step": 41120
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.954227896892316e-05,
+      "loss": 0.2333,
+      "step": 41130
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.9554324259214653e-05,
+      "loss": 0.476,
+      "step": 41140
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.956636954950615e-05,
+      "loss": 0.2551,
+      "step": 41150
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.9578414839797644e-05,
+      "loss": 0.2103,
+      "step": 41160
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.959046013008914e-05,
+      "loss": 0.1598,
+      "step": 41170
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.9602505420380634e-05,
+      "loss": 0.2675,
+      "step": 41180
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.961455071067213e-05,
+      "loss": 0.5322,
+      "step": 41190
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.9626596000963625e-05,
+      "loss": 0.159,
+      "step": 41200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.963864129125512e-05,
+      "loss": 0.1916,
+      "step": 41210
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.9650686581546615e-05,
+      "loss": 0.1815,
+      "step": 41220
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.966273187183812e-05,
+      "loss": 0.2658,
+      "step": 41230
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.967477716212961e-05,
+      "loss": 0.3224,
+      "step": 41240
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.968682245242111e-05,
+      "loss": 0.3146,
+      "step": 41250
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.96988677427126e-05,
+      "loss": 0.1385,
+      "step": 41260
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.97109130330041e-05,
+      "loss": 0.1515,
+      "step": 41270
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.972295832329559e-05,
+      "loss": 0.2369,
+      "step": 41280
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.973500361358709e-05,
+      "loss": 0.2079,
+      "step": 41290
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.974704890387858e-05,
+      "loss": 0.1965,
+      "step": 41300
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.975909419417008e-05,
+      "loss": 0.1529,
+      "step": 41310
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.9771139484461574e-05,
+      "loss": 0.14,
+      "step": 41320
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.9783184774753076e-05,
+      "loss": 0.3655,
+      "step": 41330
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.979523006504457e-05,
+      "loss": 0.314,
+      "step": 41340
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.9807275355336066e-05,
+      "loss": 0.1959,
+      "step": 41350
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.981932064562756e-05,
+      "loss": 0.1931,
+      "step": 41360
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.9831365935919056e-05,
+      "loss": 0.3248,
+      "step": 41370
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.984341122621056e-05,
+      "loss": 0.2557,
+      "step": 41380
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.9855456516502053e-05,
+      "loss": 0.0866,
+      "step": 41390
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.986750180679355e-05,
+      "loss": 0.2265,
+      "step": 41400
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.9879547097085044e-05,
+      "loss": 0.5567,
+      "step": 41410
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.989159238737654e-05,
+      "loss": 0.347,
+      "step": 41420
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.9903637677668034e-05,
+      "loss": 0.1481,
+      "step": 41430
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.991568296795953e-05,
+      "loss": 0.1131,
+      "step": 41440
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.9927728258251025e-05,
+      "loss": 0.2222,
+      "step": 41450
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.9939773548542526e-05,
+      "loss": 0.2748,
+      "step": 41460
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.995181883883402e-05,
+      "loss": 0.4202,
+      "step": 41470
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.996386412912552e-05,
+      "loss": 0.3751,
+      "step": 41480
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.997590941941701e-05,
+      "loss": 0.3678,
+      "step": 41490
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.998795470970851e-05,
+      "loss": 0.1462,
+      "step": 41500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5e-05,
+      "loss": 0.266,
+      "step": 41510
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": {
+        "accuracy": 0.9312857573020175
+      },
+      "eval_f1": {
+        "f1": 0.9313123811138734
+      },
+      "eval_loss": 0.25953179597854614,
+      "eval_precision": {
+        "precision": 0.9327488749607135
+      },
+      "eval_recall": {
+        "recall": 0.931043574955592
+      },
+      "eval_runtime": 430.4725,
+      "eval_samples_per_second": 38.574,
+      "eval_steps_per_second": 19.288,
+      "step": 41512
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.999866163441206e-05,
+      "loss": 0.3017,
+      "step": 41520
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.999732326882411e-05,
+      "loss": 0.1724,
+      "step": 41530
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.999598490323617e-05,
+      "loss": 0.1451,
+      "step": 41540
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.9994646537648224e-05,
+      "loss": 0.2853,
+      "step": 41550
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.9993308172060286e-05,
+      "loss": 0.2682,
+      "step": 41560
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.999196980647234e-05,
+      "loss": 0.1388,
+      "step": 41570
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.9990631440884397e-05,
+      "loss": 0.2557,
+      "step": 41580
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.998929307529645e-05,
+      "loss": 0.2791,
+      "step": 41590
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.998795470970851e-05,
+      "loss": 0.2096,
+      "step": 41600
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.998661634412056e-05,
+      "loss": 0.2384,
+      "step": 41610
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.998527797853262e-05,
+      "loss": 0.2401,
+      "step": 41620
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.998393961294467e-05,
+      "loss": 0.2301,
+      "step": 41630
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.998260124735673e-05,
+      "loss": 0.4035,
+      "step": 41640
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.9981262881768784e-05,
+      "loss": 0.3383,
+      "step": 41650
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.9979924516180846e-05,
+      "loss": 0.2818,
+      "step": 41660
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.99785861505929e-05,
+      "loss": 0.1835,
+      "step": 41670
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.997724778500496e-05,
+      "loss": 0.2208,
+      "step": 41680
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.997590941941701e-05,
+      "loss": 0.211,
+      "step": 41690
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.997457105382907e-05,
+      "loss": 0.3801,
+      "step": 41700
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.997323268824112e-05,
+      "loss": 0.175,
+      "step": 41710
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.997189432265318e-05,
+      "loss": 0.3386,
+      "step": 41720
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.997055595706523e-05,
+      "loss": 0.1076,
+      "step": 41730
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.996921759147729e-05,
+      "loss": 0.2278,
+      "step": 41740
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.9967879225889344e-05,
+      "loss": 0.1523,
+      "step": 41750
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.9966540860301406e-05,
+      "loss": 0.2594,
+      "step": 41760
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.996520249471346e-05,
+      "loss": 0.3224,
+      "step": 41770
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.996386412912552e-05,
+      "loss": 0.3366,
+      "step": 41780
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.996252576353757e-05,
+      "loss": 0.3783,
+      "step": 41790
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.996118739794963e-05,
+      "loss": 0.1699,
+      "step": 41800
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.995984903236168e-05,
+      "loss": 0.16,
+      "step": 41810
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.995851066677374e-05,
+      "loss": 0.2051,
+      "step": 41820
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.9957172301185794e-05,
+      "loss": 0.135,
+      "step": 41830
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.995583393559785e-05,
+      "loss": 0.1992,
+      "step": 41840
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.9954495570009904e-05,
+      "loss": 0.2746,
+      "step": 41850
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.9953157204421966e-05,
+      "loss": 0.2226,
+      "step": 41860
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.995181883883402e-05,
+      "loss": 0.3651,
+      "step": 41870
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.995048047324608e-05,
+      "loss": 0.3959,
+      "step": 41880
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.994914210765813e-05,
+      "loss": 0.4023,
+      "step": 41890
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.994780374207019e-05,
+      "loss": 0.3724,
+      "step": 41900
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.994646537648224e-05,
+      "loss": 0.2924,
+      "step": 41910
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.99451270108943e-05,
+      "loss": 0.2436,
+      "step": 41920
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.9943788645306354e-05,
+      "loss": 0.0884,
+      "step": 41930
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.994245027971841e-05,
+      "loss": 0.1604,
+      "step": 41940
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.9941111914130464e-05,
+      "loss": 0.1719,
+      "step": 41950
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.9939773548542526e-05,
+      "loss": 0.2053,
+      "step": 41960
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.993843518295458e-05,
+      "loss": 0.2277,
+      "step": 41970
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.993709681736664e-05,
+      "loss": 0.1463,
+      "step": 41980
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.993575845177869e-05,
+      "loss": 0.2948,
+      "step": 41990
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.993442008619075e-05,
+      "loss": 0.1681,
+      "step": 42000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.99330817206028e-05,
+      "loss": 0.4403,
+      "step": 42010
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.993174335501486e-05,
+      "loss": 0.2463,
+      "step": 42020
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.9930404989426914e-05,
+      "loss": 0.1511,
+      "step": 42030
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.992906662383897e-05,
+      "loss": 0.225,
+      "step": 42040
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.9927728258251025e-05,
+      "loss": 0.1585,
+      "step": 42050
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.992638989266309e-05,
+      "loss": 0.1427,
+      "step": 42060
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.992505152707514e-05,
+      "loss": 0.2699,
+      "step": 42070
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.99237131614872e-05,
+      "loss": 0.1546,
+      "step": 42080
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.992237479589925e-05,
+      "loss": 0.2653,
+      "step": 42090
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.992103643031131e-05,
+      "loss": 0.2753,
+      "step": 42100
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.991969806472336e-05,
+      "loss": 0.3881,
+      "step": 42110
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.991835969913542e-05,
+      "loss": 0.3334,
+      "step": 42120
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.9917021333547474e-05,
+      "loss": 0.2683,
+      "step": 42130
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.991568296795953e-05,
+      "loss": 0.2863,
+      "step": 42140
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.9914344602371585e-05,
+      "loss": 0.2589,
+      "step": 42150
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.991300623678365e-05,
+      "loss": 0.2299,
+      "step": 42160
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.99116678711957e-05,
+      "loss": 0.1728,
+      "step": 42170
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.991032950560776e-05,
+      "loss": 0.2295,
+      "step": 42180
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.990899114001981e-05,
+      "loss": 0.1978,
+      "step": 42190
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.990765277443187e-05,
+      "loss": 0.3154,
+      "step": 42200
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.9906314408843923e-05,
+      "loss": 0.207,
+      "step": 42210
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.990497604325598e-05,
+      "loss": 0.1865,
+      "step": 42220
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.9903637677668034e-05,
+      "loss": 0.2286,
+      "step": 42230
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.990229931208009e-05,
+      "loss": 0.312,
+      "step": 42240
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.9900960946492145e-05,
+      "loss": 0.1874,
+      "step": 42250
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.989962258090421e-05,
+      "loss": 0.0783,
+      "step": 42260
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.989828421531626e-05,
+      "loss": 0.2597,
+      "step": 42270
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.989694584972832e-05,
+      "loss": 0.4188,
+      "step": 42280
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.989560748414037e-05,
+      "loss": 0.1574,
+      "step": 42290
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.989426911855243e-05,
+      "loss": 0.1663,
+      "step": 42300
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.9892930752964484e-05,
+      "loss": 0.5736,
+      "step": 42310
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.989159238737654e-05,
+      "loss": 0.1851,
+      "step": 42320
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.9890254021788594e-05,
+      "loss": 0.2128,
+      "step": 42330
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.988891565620065e-05,
+      "loss": 0.211,
+      "step": 42340
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.9887577290612705e-05,
+      "loss": 0.2659,
+      "step": 42350
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.988623892502477e-05,
+      "loss": 0.2758,
+      "step": 42360
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.988490055943682e-05,
+      "loss": 0.1701,
+      "step": 42370
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.988356219384888e-05,
+      "loss": 0.1665,
+      "step": 42380
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.988222382826093e-05,
+      "loss": 0.0777,
+      "step": 42390
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.988088546267299e-05,
+      "loss": 0.3692,
+      "step": 42400
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.9879547097085044e-05,
+      "loss": 0.272,
+      "step": 42410
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.98782087314971e-05,
+      "loss": 0.2405,
+      "step": 42420
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.9876870365909154e-05,
+      "loss": 0.3501,
+      "step": 42430
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.987553200032121e-05,
+      "loss": 0.4044,
+      "step": 42440
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.9874193634733265e-05,
+      "loss": 0.1627,
+      "step": 42450
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.987285526914533e-05,
+      "loss": 0.2406,
+      "step": 42460
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.987151690355738e-05,
+      "loss": 0.135,
+      "step": 42470
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.987017853796944e-05,
+      "loss": 0.281,
+      "step": 42480
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.986884017238149e-05,
+      "loss": 0.2488,
+      "step": 42490
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.986750180679355e-05,
+      "loss": 0.181,
+      "step": 42500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.9866163441205604e-05,
+      "loss": 0.1972,
+      "step": 42510
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.986482507561766e-05,
+      "loss": 0.2808,
+      "step": 42520
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.9863486710029715e-05,
+      "loss": 0.4412,
+      "step": 42530
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.986214834444177e-05,
+      "loss": 0.2544,
+      "step": 42540
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.9860809978853825e-05,
+      "loss": 0.3033,
+      "step": 42550
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.985947161326589e-05,
+      "loss": 0.173,
+      "step": 42560
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.985813324767794e-05,
+      "loss": 0.1182,
+      "step": 42570
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.985679488209e-05,
+      "loss": 0.2272,
+      "step": 42580
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.9855456516502053e-05,
+      "loss": 0.3546,
+      "step": 42590
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.985411815091411e-05,
+      "loss": 0.369,
+      "step": 42600
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.9852779785326164e-05,
+      "loss": 0.2267,
+      "step": 42610
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.985144141973822e-05,
+      "loss": 0.1877,
+      "step": 42620
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.9850103054150275e-05,
+      "loss": 0.3955,
+      "step": 42630
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.984876468856233e-05,
+      "loss": 0.3645,
+      "step": 42640
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.9847426322974385e-05,
+      "loss": 0.1928,
+      "step": 42650
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.984608795738645e-05,
+      "loss": 0.3008,
+      "step": 42660
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.98447495917985e-05,
+      "loss": 0.227,
+      "step": 42670
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.984341122621056e-05,
+      "loss": 0.1904,
+      "step": 42680
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.9842072860622614e-05,
+      "loss": 0.4912,
+      "step": 42690
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.984073449503466e-05,
+      "loss": 0.2463,
+      "step": 42700
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.983939612944672e-05,
+      "loss": 0.2742,
+      "step": 42710
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.983805776385877e-05,
+      "loss": 0.162,
+      "step": 42720
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.983671939827083e-05,
+      "loss": 0.3381,
+      "step": 42730
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.983538103268289e-05,
+      "loss": 0.4368,
+      "step": 42740
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.9834042667094946e-05,
+      "loss": 0.2044,
+      "step": 42750
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.9832704301507e-05,
+      "loss": 0.1588,
+      "step": 42760
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.9831365935919056e-05,
+      "loss": 0.3325,
+      "step": 42770
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.983002757033111e-05,
+      "loss": 0.2708,
+      "step": 42780
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.982868920474317e-05,
+      "loss": 0.1685,
+      "step": 42790
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.982735083915522e-05,
+      "loss": 0.3301,
+      "step": 42800
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.982601247356728e-05,
+      "loss": 0.2275,
+      "step": 42810
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.982467410797933e-05,
+      "loss": 0.3058,
+      "step": 42820
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.982333574239139e-05,
+      "loss": 0.3612,
+      "step": 42830
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.982199737680345e-05,
+      "loss": 0.2206,
+      "step": 42840
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.9820659011215506e-05,
+      "loss": 0.251,
+      "step": 42850
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.981932064562756e-05,
+      "loss": 0.2851,
+      "step": 42860
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.9817982280039616e-05,
+      "loss": 0.3296,
+      "step": 42870
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.981664391445167e-05,
+      "loss": 0.3282,
+      "step": 42880
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.981530554886373e-05,
+      "loss": 0.192,
+      "step": 42890
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.981396718327578e-05,
+      "loss": 0.5699,
+      "step": 42900
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.981262881768784e-05,
+      "loss": 0.3394,
+      "step": 42910
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.981129045209989e-05,
+      "loss": 0.187,
+      "step": 42920
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.980995208651195e-05,
+      "loss": 0.2669,
+      "step": 42930
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.980861372092401e-05,
+      "loss": 0.2954,
+      "step": 42940
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.9807275355336066e-05,
+      "loss": 0.2722,
+      "step": 42950
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.980593698974812e-05,
+      "loss": 0.2474,
+      "step": 42960
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.9804598624160177e-05,
+      "loss": 0.3461,
+      "step": 42970
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.980326025857223e-05,
+      "loss": 0.2926,
+      "step": 42980
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.980192189298429e-05,
+      "loss": 0.1808,
+      "step": 42990
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.980058352739634e-05,
+      "loss": 0.0718,
+      "step": 43000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.97992451618084e-05,
+      "loss": 0.1958,
+      "step": 43010
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.979790679622045e-05,
+      "loss": 0.2365,
+      "step": 43020
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.9796568430632515e-05,
+      "loss": 0.0837,
+      "step": 43030
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.979523006504457e-05,
+      "loss": 0.1016,
+      "step": 43040
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.9793891699456626e-05,
+      "loss": 0.1111,
+      "step": 43050
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.979255333386868e-05,
+      "loss": 0.223,
+      "step": 43060
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.979121496828074e-05,
+      "loss": 0.1334,
+      "step": 43070
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.978987660269279e-05,
+      "loss": 0.2716,
+      "step": 43080
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.978853823710485e-05,
+      "loss": 0.2173,
+      "step": 43090
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.97871998715169e-05,
+      "loss": 0.253,
+      "step": 43100
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.978586150592896e-05,
+      "loss": 0.3703,
+      "step": 43110
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.9784523140341013e-05,
+      "loss": 0.2122,
+      "step": 43120
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.9783184774753076e-05,
+      "loss": 0.2038,
+      "step": 43130
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.978184640916513e-05,
+      "loss": 0.1589,
+      "step": 43140
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.9780508043577186e-05,
+      "loss": 0.4794,
+      "step": 43150
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.977916967798924e-05,
+      "loss": 0.2206,
+      "step": 43160
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.97778313124013e-05,
+      "loss": 0.1371,
+      "step": 43170
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.977649294681335e-05,
+      "loss": 0.1965,
+      "step": 43180
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.977515458122541e-05,
+      "loss": 0.2022,
+      "step": 43190
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.977381621563746e-05,
+      "loss": 0.2536,
+      "step": 43200
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.977247785004952e-05,
+      "loss": 0.2762,
+      "step": 43210
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.9771139484461574e-05,
+      "loss": 0.2305,
+      "step": 43220
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.9769801118873636e-05,
+      "loss": 0.0563,
+      "step": 43230
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.976846275328569e-05,
+      "loss": 0.1715,
+      "step": 43240
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.9767124387697746e-05,
+      "loss": 0.3773,
+      "step": 43250
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.97657860221098e-05,
+      "loss": 0.1412,
+      "step": 43260
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.976444765652186e-05,
+      "loss": 0.3201,
+      "step": 43270
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.976310929093391e-05,
+      "loss": 0.1431,
+      "step": 43280
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.976177092534597e-05,
+      "loss": 0.2957,
+      "step": 43290
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.976043255975802e-05,
+      "loss": 0.4082,
+      "step": 43300
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.975909419417008e-05,
+      "loss": 0.0437,
+      "step": 43310
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.9757755828582134e-05,
+      "loss": 0.1685,
+      "step": 43320
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.9756417462994196e-05,
+      "loss": 0.4189,
+      "step": 43330
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.975507909740625e-05,
+      "loss": 0.0249,
+      "step": 43340
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.9753740731818306e-05,
+      "loss": 0.1949,
+      "step": 43350
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.975240236623036e-05,
+      "loss": 0.2728,
+      "step": 43360
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.975106400064242e-05,
+      "loss": 0.156,
+      "step": 43370
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.974972563505447e-05,
+      "loss": 0.2292,
+      "step": 43380
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.974838726946653e-05,
+      "loss": 0.3266,
+      "step": 43390
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.974704890387858e-05,
+      "loss": 0.371,
+      "step": 43400
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.974571053829064e-05,
+      "loss": 0.3543,
+      "step": 43410
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.9744372172702694e-05,
+      "loss": 0.1284,
+      "step": 43420
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.9743033807114756e-05,
+      "loss": 0.4461,
+      "step": 43430
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.974169544152681e-05,
+      "loss": 0.2004,
+      "step": 43440
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.974035707593887e-05,
+      "loss": 0.3368,
+      "step": 43450
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.973901871035092e-05,
+      "loss": 0.1776,
+      "step": 43460
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.973768034476298e-05,
+      "loss": 0.2855,
+      "step": 43470
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.973634197917503e-05,
+      "loss": 0.1914,
+      "step": 43480
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.973500361358709e-05,
+      "loss": 0.1737,
+      "step": 43490
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.973366524799914e-05,
+      "loss": 0.0827,
+      "step": 43500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.97323268824112e-05,
+      "loss": 0.2871,
+      "step": 43510
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.9730988516823254e-05,
+      "loss": 0.0677,
+      "step": 43520
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.9729650151235316e-05,
+      "loss": 0.2453,
+      "step": 43530
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.972831178564737e-05,
+      "loss": 0.2733,
+      "step": 43540
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.972697342005943e-05,
+      "loss": 0.202,
+      "step": 43550
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.972563505447148e-05,
+      "loss": 0.2444,
+      "step": 43560
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.972429668888354e-05,
+      "loss": 0.197,
+      "step": 43570
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.972295832329559e-05,
+      "loss": 0.1157,
+      "step": 43580
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.972161995770765e-05,
+      "loss": 0.4619,
+      "step": 43590
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.9720281592119703e-05,
+      "loss": 0.106,
+      "step": 43600
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.971894322653176e-05,
+      "loss": 0.354,
+      "step": 43610
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.9717604860943814e-05,
+      "loss": 0.2992,
+      "step": 43620
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.9716266495355876e-05,
+      "loss": 0.1909,
+      "step": 43630
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.971492812976793e-05,
+      "loss": 0.2053,
+      "step": 43640
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.971358976417999e-05,
+      "loss": 0.1484,
+      "step": 43650
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.971225139859204e-05,
+      "loss": 0.2939,
+      "step": 43660
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.97109130330041e-05,
+      "loss": 0.1802,
+      "step": 43670
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.970957466741615e-05,
+      "loss": 0.2249,
+      "step": 43680
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.970823630182821e-05,
+      "loss": 0.1129,
+      "step": 43690
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.9706897936240264e-05,
+      "loss": 0.209,
+      "step": 43700
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.970555957065232e-05,
+      "loss": 0.299,
+      "step": 43710
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.9704221205064374e-05,
+      "loss": 0.1676,
+      "step": 43720
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.9702882839476436e-05,
+      "loss": 0.4741,
+      "step": 43730
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.970154447388849e-05,
+      "loss": 0.1136,
+      "step": 43740
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.970020610830055e-05,
+      "loss": 0.2917,
+      "step": 43750
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.96988677427126e-05,
+      "loss": 0.3694,
+      "step": 43760
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.969752937712466e-05,
+      "loss": 0.3379,
+      "step": 43770
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.969619101153671e-05,
+      "loss": 0.6246,
+      "step": 43780
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.969485264594877e-05,
+      "loss": 0.2748,
+      "step": 43790
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.9693514280360824e-05,
+      "loss": 0.1138,
+      "step": 43800
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.969217591477288e-05,
+      "loss": 0.2417,
+      "step": 43810
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.9690837549184934e-05,
+      "loss": 0.0456,
+      "step": 43820
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.9689499183596997e-05,
+      "loss": 0.1032,
+      "step": 43830
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.968816081800905e-05,
+      "loss": 0.2441,
+      "step": 43840
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.968682245242111e-05,
+      "loss": 0.1786,
+      "step": 43850
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.968548408683316e-05,
+      "loss": 0.6289,
+      "step": 43860
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.968414572124522e-05,
+      "loss": 0.1546,
+      "step": 43870
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.968280735565727e-05,
+      "loss": 0.1412,
+      "step": 43880
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.968146899006933e-05,
+      "loss": 0.1919,
+      "step": 43890
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.9680130624481384e-05,
+      "loss": 0.08,
+      "step": 43900
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.967879225889344e-05,
+      "loss": 0.2416,
+      "step": 43910
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.9677453893305495e-05,
+      "loss": 0.479,
+      "step": 43920
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.967611552771756e-05,
+      "loss": 0.3577,
+      "step": 43930
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.967477716212961e-05,
+      "loss": 0.3684,
+      "step": 43940
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.967343879654167e-05,
+      "loss": 0.2898,
+      "step": 43950
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.967210043095372e-05,
+      "loss": 0.2968,
+      "step": 43960
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.967076206536578e-05,
+      "loss": 0.1957,
+      "step": 43970
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.9669423699777833e-05,
+      "loss": 0.237,
+      "step": 43980
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.966808533418989e-05,
+      "loss": 0.3329,
+      "step": 43990
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.9666746968601944e-05,
+      "loss": 0.201,
+      "step": 44000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.9665408603014e-05,
+      "loss": 0.0443,
+      "step": 44010
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.9664070237426055e-05,
+      "loss": 0.1521,
+      "step": 44020
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.966273187183812e-05,
+      "loss": 0.1908,
+      "step": 44030
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.966139350625017e-05,
+      "loss": 0.1899,
+      "step": 44040
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.966005514066223e-05,
+      "loss": 0.1384,
+      "step": 44050
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.965871677507428e-05,
+      "loss": 0.2534,
+      "step": 44060
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.965737840948634e-05,
+      "loss": 0.2311,
+      "step": 44070
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.9656040043898394e-05,
+      "loss": 0.3097,
+      "step": 44080
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.965470167831045e-05,
+      "loss": 0.3044,
+      "step": 44090
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.9653363312722504e-05,
+      "loss": 0.0626,
+      "step": 44100
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.965202494713456e-05,
+      "loss": 0.2287,
+      "step": 44110
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.9650686581546615e-05,
+      "loss": 0.3615,
+      "step": 44120
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.964934821595868e-05,
+      "loss": 0.21,
+      "step": 44130
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.964800985037073e-05,
+      "loss": 0.2151,
+      "step": 44140
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.964667148478279e-05,
+      "loss": 0.2643,
+      "step": 44150
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.964533311919484e-05,
+      "loss": 0.2061,
+      "step": 44160
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.96439947536069e-05,
+      "loss": 0.1495,
+      "step": 44170
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.9642656388018954e-05,
+      "loss": 0.2768,
+      "step": 44180
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.964131802243101e-05,
+      "loss": 0.2313,
+      "step": 44190
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.9639979656843064e-05,
+      "loss": 0.1116,
+      "step": 44200
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.963864129125512e-05,
+      "loss": 0.4357,
+      "step": 44210
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.9637302925667175e-05,
+      "loss": 0.1507,
+      "step": 44220
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.963596456007924e-05,
+      "loss": 0.2036,
+      "step": 44230
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.963462619449129e-05,
+      "loss": 0.1201,
+      "step": 44240
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.963328782890335e-05,
+      "loss": 0.4468,
+      "step": 44250
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.96319494633154e-05,
+      "loss": 0.4002,
+      "step": 44260
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.963061109772746e-05,
+      "loss": 0.1404,
+      "step": 44270
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.9629272732139514e-05,
+      "loss": 0.2335,
+      "step": 44280
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.962793436655157e-05,
+      "loss": 0.1505,
+      "step": 44290
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.9626596000963625e-05,
+      "loss": 0.2121,
+      "step": 44300
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.962525763537568e-05,
+      "loss": 0.2708,
+      "step": 44310
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.962391926978774e-05,
+      "loss": 0.5263,
+      "step": 44320
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.96225809041998e-05,
+      "loss": 0.4033,
+      "step": 44330
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.962124253861185e-05,
+      "loss": 0.2793,
+      "step": 44340
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.961990417302391e-05,
+      "loss": 0.2688,
+      "step": 44350
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.961856580743596e-05,
+      "loss": 0.3312,
+      "step": 44360
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.961722744184802e-05,
+      "loss": 0.3113,
+      "step": 44370
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.9615889076260074e-05,
+      "loss": 0.1149,
+      "step": 44380
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.961455071067213e-05,
+      "loss": 0.0891,
+      "step": 44390
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.9613212345084185e-05,
+      "loss": 0.3949,
+      "step": 44400
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.961187397949624e-05,
+      "loss": 0.181,
+      "step": 44410
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.96105356139083e-05,
+      "loss": 0.4605,
+      "step": 44420
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.960919724832036e-05,
+      "loss": 0.2374,
+      "step": 44430
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.960785888273241e-05,
+      "loss": 0.254,
+      "step": 44440
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.960652051714447e-05,
+      "loss": 0.1397,
+      "step": 44450
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.9605182151556524e-05,
+      "loss": 0.3702,
+      "step": 44460
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.960384378596858e-05,
+      "loss": 0.2286,
+      "step": 44470
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.9602505420380634e-05,
+      "loss": 0.2978,
+      "step": 44480
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.960116705479269e-05,
+      "loss": 0.407,
+      "step": 44490
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.9599828689204745e-05,
+      "loss": 0.1548,
+      "step": 44500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.95984903236168e-05,
+      "loss": 0.2538,
+      "step": 44510
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.959715195802886e-05,
+      "loss": 0.2944,
+      "step": 44520
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.959581359244092e-05,
+      "loss": 0.2202,
+      "step": 44530
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.959447522685297e-05,
+      "loss": 0.1002,
+      "step": 44540
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.959313686126503e-05,
+      "loss": 0.2403,
+      "step": 44550
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.9591798495677084e-05,
+      "loss": 0.1906,
+      "step": 44560
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.959046013008914e-05,
+      "loss": 0.114,
+      "step": 44570
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.9589121764501194e-05,
+      "loss": 0.267,
+      "step": 44580
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.958778339891325e-05,
+      "loss": 0.2161,
+      "step": 44590
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.9586445033325305e-05,
+      "loss": 0.1968,
+      "step": 44600
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.958510666773736e-05,
+      "loss": 0.2138,
+      "step": 44610
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.958376830214942e-05,
+      "loss": 0.2731,
+      "step": 44620
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.958242993656148e-05,
+      "loss": 0.142,
+      "step": 44630
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.958109157097353e-05,
+      "loss": 0.253,
+      "step": 44640
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.957975320538559e-05,
+      "loss": 0.0788,
+      "step": 44650
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.9578414839797644e-05,
+      "loss": 0.227,
+      "step": 44660
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.95770764742097e-05,
+      "loss": 0.1396,
+      "step": 44670
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.9575738108621754e-05,
+      "loss": 0.252,
+      "step": 44680
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.957439974303381e-05,
+      "loss": 0.1693,
+      "step": 44690
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.9573061377445865e-05,
+      "loss": 0.3988,
+      "step": 44700
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.957172301185792e-05,
+      "loss": 0.2847,
+      "step": 44710
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.957038464626998e-05,
+      "loss": 0.2936,
+      "step": 44720
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.956904628068204e-05,
+      "loss": 0.2201,
+      "step": 44730
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.956770791509409e-05,
+      "loss": 0.2105,
+      "step": 44740
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.956636954950615e-05,
+      "loss": 0.5137,
+      "step": 44750
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.9565031183918204e-05,
+      "loss": 0.1895,
+      "step": 44760
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.956369281833026e-05,
+      "loss": 0.3309,
+      "step": 44770
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.9562354452742315e-05,
+      "loss": 0.204,
+      "step": 44780
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.956101608715437e-05,
+      "loss": 0.2338,
+      "step": 44790
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.9559677721566425e-05,
+      "loss": 0.1928,
+      "step": 44800
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.955833935597848e-05,
+      "loss": 0.1448,
+      "step": 44810
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.955700099039054e-05,
+      "loss": 0.2326,
+      "step": 44820
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.95556626248026e-05,
+      "loss": 0.1757,
+      "step": 44830
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.9554324259214653e-05,
+      "loss": 0.3521,
+      "step": 44840
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.955298589362671e-05,
+      "loss": 0.3675,
+      "step": 44850
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.9551647528038764e-05,
+      "loss": 0.0497,
+      "step": 44860
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.955030916245082e-05,
+      "loss": 0.1279,
+      "step": 44870
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.9548970796862875e-05,
+      "loss": 0.2118,
+      "step": 44880
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.954763243127493e-05,
+      "loss": 0.0594,
+      "step": 44890
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.9546294065686985e-05,
+      "loss": 0.1621,
+      "step": 44900
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.954495570009904e-05,
+      "loss": 0.2124,
+      "step": 44910
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.95436173345111e-05,
+      "loss": 0.2697,
+      "step": 44920
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.954227896892316e-05,
+      "loss": 0.2073,
+      "step": 44930
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.9540940603335214e-05,
+      "loss": 0.2672,
+      "step": 44940
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.953960223774727e-05,
+      "loss": 0.2704,
+      "step": 44950
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.9538263872159324e-05,
+      "loss": 0.2959,
+      "step": 44960
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.953692550657138e-05,
+      "loss": 0.2693,
+      "step": 44970
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.9535587140983435e-05,
+      "loss": 0.3016,
+      "step": 44980
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.953424877539549e-05,
+      "loss": 0.1847,
+      "step": 44990
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.9532910409807546e-05,
+      "loss": 0.2438,
+      "step": 45000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.95315720442196e-05,
+      "loss": 0.0719,
+      "step": 45010
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.953023367863166e-05,
+      "loss": 0.3637,
+      "step": 45020
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.952889531304372e-05,
+      "loss": 0.261,
+      "step": 45030
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.9527556947455774e-05,
+      "loss": 0.3891,
+      "step": 45040
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.952621858186783e-05,
+      "loss": 0.1041,
+      "step": 45050
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.952488021627988e-05,
+      "loss": 0.3411,
+      "step": 45060
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.952354185069193e-05,
+      "loss": 0.1063,
+      "step": 45070
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.952220348510399e-05,
+      "loss": 0.1468,
+      "step": 45080
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.9520865119516044e-05,
+      "loss": 0.1539,
+      "step": 45090
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.9519526753928106e-05,
+      "loss": 0.3416,
+      "step": 45100
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.951818838834016e-05,
+      "loss": 0.155,
+      "step": 45110
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.9516850022752216e-05,
+      "loss": 0.2344,
+      "step": 45120
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.951551165716427e-05,
+      "loss": 0.2218,
+      "step": 45130
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.951417329157633e-05,
+      "loss": 0.2317,
+      "step": 45140
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.951283492598838e-05,
+      "loss": 0.2533,
+      "step": 45150
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.951149656040044e-05,
+      "loss": 0.1694,
+      "step": 45160
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.951015819481249e-05,
+      "loss": 0.1635,
+      "step": 45170
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.950881982922455e-05,
+      "loss": 0.3704,
+      "step": 45180
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.9507481463636604e-05,
+      "loss": 0.1211,
+      "step": 45190
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.9506143098048666e-05,
+      "loss": 0.1419,
+      "step": 45200
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.950480473246072e-05,
+      "loss": 0.3293,
+      "step": 45210
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.9503466366872777e-05,
+      "loss": 0.3032,
+      "step": 45220
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.950212800128483e-05,
+      "loss": 0.1373,
+      "step": 45230
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.950078963569689e-05,
+      "loss": 0.3126,
+      "step": 45240
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.949945127010894e-05,
+      "loss": 0.3367,
+      "step": 45250
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.9498112904521e-05,
+      "loss": 0.1237,
+      "step": 45260
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.949677453893305e-05,
+      "loss": 0.4488,
+      "step": 45270
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.949543617334511e-05,
+      "loss": 0.1613,
+      "step": 45280
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.9494097807757164e-05,
+      "loss": 0.1826,
+      "step": 45290
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.9492759442169226e-05,
+      "loss": 0.1389,
+      "step": 45300
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.949142107658128e-05,
+      "loss": 0.0857,
+      "step": 45310
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.949008271099334e-05,
+      "loss": 0.1694,
+      "step": 45320
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.948874434540539e-05,
+      "loss": 0.1662,
+      "step": 45330
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.948740597981745e-05,
+      "loss": 0.2189,
+      "step": 45340
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.94860676142295e-05,
+      "loss": 0.3363,
+      "step": 45350
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.948472924864156e-05,
+      "loss": 0.3852,
+      "step": 45360
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.9483390883053613e-05,
+      "loss": 0.3049,
+      "step": 45370
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.948205251746567e-05,
+      "loss": 0.2857,
+      "step": 45380
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.9480714151877724e-05,
+      "loss": 0.0128,
+      "step": 45390
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.9479375786289786e-05,
+      "loss": 0.296,
+      "step": 45400
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.947803742070184e-05,
+      "loss": 0.152,
+      "step": 45410
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.94766990551139e-05,
+      "loss": 0.0557,
+      "step": 45420
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.947536068952595e-05,
+      "loss": 0.3424,
+      "step": 45430
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.947402232393801e-05,
+      "loss": 0.2413,
+      "step": 45440
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.947268395835006e-05,
+      "loss": 0.1528,
+      "step": 45450
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.947134559276212e-05,
+      "loss": 0.3689,
+      "step": 45460
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.9470007227174174e-05,
+      "loss": 0.3637,
+      "step": 45470
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.946866886158623e-05,
+      "loss": 0.3493,
+      "step": 45480
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.9467330495998284e-05,
+      "loss": 0.3423,
+      "step": 45490
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.9465992130410346e-05,
+      "loss": 0.3443,
+      "step": 45500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.94646537648224e-05,
+      "loss": 0.2259,
+      "step": 45510
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.946331539923446e-05,
+      "loss": 0.4392,
+      "step": 45520
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.946197703364651e-05,
+      "loss": 0.2396,
+      "step": 45530
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.946063866805857e-05,
+      "loss": 0.2289,
+      "step": 45540
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.945930030247062e-05,
+      "loss": 0.2546,
+      "step": 45550
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.945796193688268e-05,
+      "loss": 0.2164,
+      "step": 45560
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.9456623571294734e-05,
+      "loss": 0.478,
+      "step": 45570
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.945528520570679e-05,
+      "loss": 0.289,
+      "step": 45580
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.9453946840118844e-05,
+      "loss": 0.1253,
+      "step": 45590
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.9452608474530907e-05,
+      "loss": 0.2898,
+      "step": 45600
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.945127010894296e-05,
+      "loss": 0.1831,
+      "step": 45610
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.944993174335502e-05,
+      "loss": 0.0958,
+      "step": 45620
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.944859337776707e-05,
+      "loss": 0.391,
+      "step": 45630
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.944725501217913e-05,
+      "loss": 0.2172,
+      "step": 45640
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.944591664659118e-05,
+      "loss": 0.1988,
+      "step": 45650
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.944457828100324e-05,
+      "loss": 0.1534,
+      "step": 45660
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.9443239915415294e-05,
+      "loss": 0.5528,
+      "step": 45670
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.944190154982735e-05,
+      "loss": 0.1822,
+      "step": 45680
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.9440563184239405e-05,
+      "loss": 0.2195,
+      "step": 45690
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.943922481865147e-05,
+      "loss": 0.3285,
+      "step": 45700
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.943788645306352e-05,
+      "loss": 0.2429,
+      "step": 45710
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.943654808747558e-05,
+      "loss": 0.2071,
+      "step": 45720
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.943520972188763e-05,
+      "loss": 0.2451,
+      "step": 45730
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.943387135629969e-05,
+      "loss": 0.1461,
+      "step": 45740
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.943253299071174e-05,
+      "loss": 0.2164,
+      "step": 45750
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.94311946251238e-05,
+      "loss": 0.1361,
+      "step": 45760
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.9429856259535854e-05,
+      "loss": 0.1823,
+      "step": 45770
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.942851789394791e-05,
+      "loss": 0.1815,
+      "step": 45780
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.9427179528359965e-05,
+      "loss": 0.516,
+      "step": 45790
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.942584116277203e-05,
+      "loss": 0.3174,
+      "step": 45800
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.942450279718408e-05,
+      "loss": 0.3558,
+      "step": 45810
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.942316443159614e-05,
+      "loss": 0.2505,
+      "step": 45820
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.942182606600819e-05,
+      "loss": 0.3011,
+      "step": 45830
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.942048770042025e-05,
+      "loss": 0.109,
+      "step": 45840
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.9419149334832304e-05,
+      "loss": 0.2381,
+      "step": 45850
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.941781096924436e-05,
+      "loss": 0.3127,
+      "step": 45860
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.9416472603656414e-05,
+      "loss": 0.4408,
+      "step": 45870
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.941513423806847e-05,
+      "loss": 0.0954,
+      "step": 45880
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.941379587248053e-05,
+      "loss": 0.2438,
+      "step": 45890
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.941245750689259e-05,
+      "loss": 0.2439,
+      "step": 45900
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.941111914130464e-05,
+      "loss": 0.1653,
+      "step": 45910
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.94097807757167e-05,
+      "loss": 0.3457,
+      "step": 45920
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.940844241012875e-05,
+      "loss": 0.144,
+      "step": 45930
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.940710404454081e-05,
+      "loss": 0.0494,
+      "step": 45940
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.9405765678952864e-05,
+      "loss": 0.1829,
+      "step": 45950
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.940442731336492e-05,
+      "loss": 0.1582,
+      "step": 45960
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.9403088947776974e-05,
+      "loss": 0.1981,
+      "step": 45970
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.940175058218903e-05,
+      "loss": 0.149,
+      "step": 45980
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.940041221660109e-05,
+      "loss": 0.1982,
+      "step": 45990
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.939907385101315e-05,
+      "loss": 0.4342,
+      "step": 46000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.93977354854252e-05,
+      "loss": 0.167,
+      "step": 46010
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.939639711983726e-05,
+      "loss": 0.2786,
+      "step": 46020
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.939505875424931e-05,
+      "loss": 0.2082,
+      "step": 46030
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.939372038866137e-05,
+      "loss": 0.2961,
+      "step": 46040
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.9392382023073424e-05,
+      "loss": 0.1233,
+      "step": 46050
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.939104365748548e-05,
+      "loss": 0.2545,
+      "step": 46060
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.9389705291897535e-05,
+      "loss": 0.1707,
+      "step": 46070
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.938836692630959e-05,
+      "loss": 0.2512,
+      "step": 46080
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.938702856072165e-05,
+      "loss": 0.1307,
+      "step": 46090
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.938569019513371e-05,
+      "loss": 0.0947,
+      "step": 46100
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.938435182954576e-05,
+      "loss": 0.278,
+      "step": 46110
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.938301346395782e-05,
+      "loss": 0.1453,
+      "step": 46120
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.938167509836987e-05,
+      "loss": 0.1837,
+      "step": 46130
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.938033673278193e-05,
+      "loss": 0.3246,
+      "step": 46140
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.9378998367193984e-05,
+      "loss": 0.1634,
+      "step": 46150
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.937766000160604e-05,
+      "loss": 0.0648,
+      "step": 46160
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.9376321636018095e-05,
+      "loss": 0.112,
+      "step": 46170
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.937498327043015e-05,
+      "loss": 0.1356,
+      "step": 46180
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.937364490484221e-05,
+      "loss": 0.1351,
+      "step": 46190
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.937230653925427e-05,
+      "loss": 0.1381,
+      "step": 46200
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.937096817366632e-05,
+      "loss": 0.362,
+      "step": 46210
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.936962980807838e-05,
+      "loss": 0.3223,
+      "step": 46220
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.9368291442490433e-05,
+      "loss": 0.2692,
+      "step": 46230
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.936695307690249e-05,
+      "loss": 0.119,
+      "step": 46240
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.9365614711314544e-05,
+      "loss": 0.0863,
+      "step": 46250
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.93642763457266e-05,
+      "loss": 0.1871,
+      "step": 46260
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.9362937980138655e-05,
+      "loss": 0.2031,
+      "step": 46270
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.936159961455071e-05,
+      "loss": 0.25,
+      "step": 46280
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.936026124896277e-05,
+      "loss": 0.1835,
+      "step": 46290
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.935892288337483e-05,
+      "loss": 0.3723,
+      "step": 46300
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.935758451778688e-05,
+      "loss": 0.1971,
+      "step": 46310
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.935624615219894e-05,
+      "loss": 0.149,
+      "step": 46320
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.9354907786610994e-05,
+      "loss": 0.3157,
+      "step": 46330
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.935356942102305e-05,
+      "loss": 0.2636,
+      "step": 46340
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.9352231055435104e-05,
+      "loss": 0.113,
+      "step": 46350
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.935089268984716e-05,
+      "loss": 0.1854,
+      "step": 46360
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.9349554324259215e-05,
+      "loss": 0.1742,
+      "step": 46370
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.934821595867127e-05,
+      "loss": 0.2161,
+      "step": 46380
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.934687759308333e-05,
+      "loss": 0.3479,
+      "step": 46390
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.934553922749539e-05,
+      "loss": 0.2152,
+      "step": 46400
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.934420086190744e-05,
+      "loss": 0.2263,
+      "step": 46410
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.93428624963195e-05,
+      "loss": 0.0776,
+      "step": 46420
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.9341524130731554e-05,
+      "loss": 0.1969,
+      "step": 46430
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.934018576514361e-05,
+      "loss": 0.2142,
+      "step": 46440
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.9338847399555664e-05,
+      "loss": 0.5397,
+      "step": 46450
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.933750903396772e-05,
+      "loss": 0.2318,
+      "step": 46460
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.9336170668379775e-05,
+      "loss": 0.2552,
+      "step": 46470
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.933483230279183e-05,
+      "loss": 0.2649,
+      "step": 46480
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.933349393720389e-05,
+      "loss": 0.3351,
+      "step": 46490
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.933215557161595e-05,
+      "loss": 0.2814,
+      "step": 46500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.9330817206028e-05,
+      "loss": 0.1237,
+      "step": 46510
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.932947884044006e-05,
+      "loss": 0.3198,
+      "step": 46520
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.9328140474852114e-05,
+      "loss": 0.169,
+      "step": 46530
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.932680210926417e-05,
+      "loss": 0.3054,
+      "step": 46540
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.9325463743676225e-05,
+      "loss": 0.1614,
+      "step": 46550
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.932412537808828e-05,
+      "loss": 0.3384,
+      "step": 46560
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.9322787012500335e-05,
+      "loss": 0.2409,
+      "step": 46570
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.932144864691239e-05,
+      "loss": 0.3429,
+      "step": 46580
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.932011028132445e-05,
+      "loss": 0.1329,
+      "step": 46590
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.931877191573651e-05,
+      "loss": 0.1839,
+      "step": 46600
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.9317433550148563e-05,
+      "loss": 0.1742,
+      "step": 46610
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.931609518456062e-05,
+      "loss": 0.2875,
+      "step": 46620
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.9314756818972674e-05,
+      "loss": 0.1713,
+      "step": 46630
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.931341845338473e-05,
+      "loss": 0.2474,
+      "step": 46640
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.9312080087796785e-05,
+      "loss": 0.2127,
+      "step": 46650
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.931074172220884e-05,
+      "loss": 0.203,
+      "step": 46660
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.9309403356620895e-05,
+      "loss": 0.3678,
+      "step": 46670
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.930806499103295e-05,
+      "loss": 0.3352,
+      "step": 46680
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.930672662544501e-05,
+      "loss": 0.1615,
+      "step": 46690
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.930538825985707e-05,
+      "loss": 0.1065,
+      "step": 46700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.9304049894269124e-05,
+      "loss": 0.2398,
+      "step": 46710
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.930271152868118e-05,
+      "loss": 0.4163,
+      "step": 46720
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.9301373163093234e-05,
+      "loss": 0.2028,
+      "step": 46730
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.930003479750529e-05,
+      "loss": 0.3249,
+      "step": 46740
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.9298696431917345e-05,
+      "loss": 0.1544,
+      "step": 46750
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.92973580663294e-05,
+      "loss": 0.1263,
+      "step": 46760
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.9296019700741456e-05,
+      "loss": 0.3755,
+      "step": 46770
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.929468133515351e-05,
+      "loss": 0.1954,
+      "step": 46780
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.929334296956557e-05,
+      "loss": 0.1671,
+      "step": 46790
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.929200460397763e-05,
+      "loss": 0.0922,
+      "step": 46800
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.9290666238389684e-05,
+      "loss": 0.2435,
+      "step": 46810
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.928932787280174e-05,
+      "loss": 0.3912,
+      "step": 46820
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.9287989507213794e-05,
+      "loss": 0.1564,
+      "step": 46830
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.928665114162585e-05,
+      "loss": 0.325,
+      "step": 46840
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.9285312776037905e-05,
+      "loss": 0.1965,
+      "step": 46850
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.928397441044996e-05,
+      "loss": 0.2212,
+      "step": 46860
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.9282636044862016e-05,
+      "loss": 0.3512,
+      "step": 46870
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.928129767927407e-05,
+      "loss": 0.1773,
+      "step": 46880
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.927995931368613e-05,
+      "loss": 0.2093,
+      "step": 46890
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.927862094809819e-05,
+      "loss": 0.2068,
+      "step": 46900
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.9277282582510244e-05,
+      "loss": 0.1718,
+      "step": 46910
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.92759442169223e-05,
+      "loss": 0.258,
+      "step": 46920
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.9274605851334355e-05,
+      "loss": 0.2965,
+      "step": 46930
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.927326748574641e-05,
+      "loss": 0.2049,
+      "step": 46940
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.9271929120158465e-05,
+      "loss": 0.0638,
+      "step": 46950
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.927059075457052e-05,
+      "loss": 0.237,
+      "step": 46960
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.9269252388982576e-05,
+      "loss": 0.267,
+      "step": 46970
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.926791402339463e-05,
+      "loss": 0.1932,
+      "step": 46980
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.926657565780669e-05,
+      "loss": 0.1409,
+      "step": 46990
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.926523729221875e-05,
+      "loss": 0.2008,
+      "step": 47000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.9263898926630804e-05,
+      "loss": 0.323,
+      "step": 47010
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.926256056104286e-05,
+      "loss": 0.0951,
+      "step": 47020
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.9261222195454915e-05,
+      "loss": 0.2585,
+      "step": 47030
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.925988382986697e-05,
+      "loss": 0.1785,
+      "step": 47040
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.9258545464279025e-05,
+      "loss": 0.163,
+      "step": 47050
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.925720709869108e-05,
+      "loss": 0.2629,
+      "step": 47060
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.9255868733103136e-05,
+      "loss": 0.1731,
+      "step": 47070
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.92545303675152e-05,
+      "loss": 0.4734,
+      "step": 47080
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.9253192001927253e-05,
+      "loss": 0.2939,
+      "step": 47090
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.925185363633931e-05,
+      "loss": 0.1222,
+      "step": 47100
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.9250515270751364e-05,
+      "loss": 0.1701,
+      "step": 47110
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.924917690516342e-05,
+      "loss": 0.2427,
+      "step": 47120
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.9247838539575475e-05,
+      "loss": 0.2869,
+      "step": 47130
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.924650017398753e-05,
+      "loss": 0.3036,
+      "step": 47140
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.9245161808399586e-05,
+      "loss": 0.1743,
+      "step": 47150
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.924382344281164e-05,
+      "loss": 0.2841,
+      "step": 47160
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.9242485077223696e-05,
+      "loss": 0.2372,
+      "step": 47170
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.924114671163576e-05,
+      "loss": 0.3055,
+      "step": 47180
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.9239808346047814e-05,
+      "loss": 0.3005,
+      "step": 47190
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.923846998045987e-05,
+      "loss": 0.1755,
+      "step": 47200
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.9237131614871924e-05,
+      "loss": 0.2279,
+      "step": 47210
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.923579324928398e-05,
+      "loss": 0.2099,
+      "step": 47220
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.9234454883696035e-05,
+      "loss": 0.1167,
+      "step": 47230
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.923311651810809e-05,
+      "loss": 0.1616,
+      "step": 47240
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.9231778152520146e-05,
+      "loss": 0.2994,
+      "step": 47250
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.92304397869322e-05,
+      "loss": 0.2657,
+      "step": 47260
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.9229101421344256e-05,
+      "loss": 0.1861,
+      "step": 47270
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.922776305575632e-05,
+      "loss": 0.1928,
+      "step": 47280
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.9226424690168374e-05,
+      "loss": 0.2101,
+      "step": 47290
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.922508632458043e-05,
+      "loss": 0.2698,
+      "step": 47300
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.9223747958992484e-05,
+      "loss": 0.2053,
+      "step": 47310
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.922240959340454e-05,
+      "loss": 0.2356,
+      "step": 47320
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.9221071227816595e-05,
+      "loss": 0.2105,
+      "step": 47330
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.921973286222865e-05,
+      "loss": 0.1568,
+      "step": 47340
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.9218394496640706e-05,
+      "loss": 0.1563,
+      "step": 47350
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.921705613105276e-05,
+      "loss": 0.1736,
+      "step": 47360
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.9215717765464817e-05,
+      "loss": 0.3678,
+      "step": 47370
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.921437939987688e-05,
+      "loss": 0.301,
+      "step": 47380
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.9213041034288934e-05,
+      "loss": 0.3947,
+      "step": 47390
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.921170266870099e-05,
+      "loss": 0.1688,
+      "step": 47400
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.9210364303113045e-05,
+      "loss": 0.2551,
+      "step": 47410
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.92090259375251e-05,
+      "loss": 0.2274,
+      "step": 47420
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.920768757193715e-05,
+      "loss": 0.1456,
+      "step": 47430
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.9206349206349204e-05,
+      "loss": 0.19,
+      "step": 47440
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.920501084076126e-05,
+      "loss": 0.2713,
+      "step": 47450
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.920367247517332e-05,
+      "loss": 0.2376,
+      "step": 47460
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.920233410958538e-05,
+      "loss": 0.3711,
+      "step": 47470
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.920099574399743e-05,
+      "loss": 0.1569,
+      "step": 47480
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.919965737840949e-05,
+      "loss": 0.2243,
+      "step": 47490
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.919831901282154e-05,
+      "loss": 0.3369,
+      "step": 47500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.91969806472336e-05,
+      "loss": 0.2871,
+      "step": 47510
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.919564228164565e-05,
+      "loss": 0.0965,
+      "step": 47520
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.919430391605771e-05,
+      "loss": 0.2455,
+      "step": 47530
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.9192965550469764e-05,
+      "loss": 0.1302,
+      "step": 47540
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.919162718488182e-05,
+      "loss": 0.327,
+      "step": 47550
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.919028881929388e-05,
+      "loss": 0.2843,
+      "step": 47560
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.918895045370594e-05,
+      "loss": 0.0979,
+      "step": 47570
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.918761208811799e-05,
+      "loss": 0.3093,
+      "step": 47580
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.918627372253005e-05,
+      "loss": 0.3501,
+      "step": 47590
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.91849353569421e-05,
+      "loss": 0.2235,
+      "step": 47600
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.918359699135416e-05,
+      "loss": 0.0633,
+      "step": 47610
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.9182258625766213e-05,
+      "loss": 0.227,
+      "step": 47620
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.918092026017827e-05,
+      "loss": 0.3618,
+      "step": 47630
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.9179581894590324e-05,
+      "loss": 0.2205,
+      "step": 47640
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.917824352900238e-05,
+      "loss": 0.1865,
+      "step": 47650
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.917690516341444e-05,
+      "loss": 0.1626,
+      "step": 47660
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.91755667978265e-05,
+      "loss": 0.0962,
+      "step": 47670
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.917422843223855e-05,
+      "loss": 0.1808,
+      "step": 47680
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.917289006665061e-05,
+      "loss": 0.1054,
+      "step": 47690
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.917155170106266e-05,
+      "loss": 0.1881,
+      "step": 47700
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.917021333547472e-05,
+      "loss": 0.1868,
+      "step": 47710
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.9168874969886774e-05,
+      "loss": 0.1949,
+      "step": 47720
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.916753660429883e-05,
+      "loss": 0.1288,
+      "step": 47730
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.9166198238710884e-05,
+      "loss": 0.2217,
+      "step": 47740
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.916485987312294e-05,
+      "loss": 0.1624,
+      "step": 47750
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.9163521507535e-05,
+      "loss": 0.1917,
+      "step": 47760
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.916218314194706e-05,
+      "loss": 0.3734,
+      "step": 47770
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.916084477635911e-05,
+      "loss": 0.133,
+      "step": 47780
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.915950641077117e-05,
+      "loss": 0.2893,
+      "step": 47790
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.915816804518322e-05,
+      "loss": 0.1296,
+      "step": 47800
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.915682967959528e-05,
+      "loss": 0.3763,
+      "step": 47810
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.9155491314007334e-05,
+      "loss": 0.3981,
+      "step": 47820
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.915415294841939e-05,
+      "loss": 0.182,
+      "step": 47830
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.9152814582831444e-05,
+      "loss": 0.1432,
+      "step": 47840
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.91514762172435e-05,
+      "loss": 0.2854,
+      "step": 47850
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.915013785165556e-05,
+      "loss": 0.2147,
+      "step": 47860
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.914879948606762e-05,
+      "loss": 0.1694,
+      "step": 47870
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.914746112047967e-05,
+      "loss": 0.3376,
+      "step": 47880
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.914612275489173e-05,
+      "loss": 0.2131,
+      "step": 47890
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.914478438930378e-05,
+      "loss": 0.2766,
+      "step": 47900
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.914344602371584e-05,
+      "loss": 0.2046,
+      "step": 47910
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.9142107658127894e-05,
+      "loss": 0.1945,
+      "step": 47920
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.914076929253995e-05,
+      "loss": 0.1715,
+      "step": 47930
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.9139430926952005e-05,
+      "loss": 0.2744,
+      "step": 47940
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.913809256136406e-05,
+      "loss": 0.2418,
+      "step": 47950
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.913675419577612e-05,
+      "loss": 0.3087,
+      "step": 47960
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.913541583018818e-05,
+      "loss": 0.2029,
+      "step": 47970
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.913407746460023e-05,
+      "loss": 0.1562,
+      "step": 47980
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.913273909901229e-05,
+      "loss": 0.3404,
+      "step": 47990
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.9131400733424343e-05,
+      "loss": 0.2296,
+      "step": 48000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.91300623678364e-05,
+      "loss": 0.3627,
+      "step": 48010
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.9128724002248454e-05,
+      "loss": 0.1289,
+      "step": 48020
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.912738563666051e-05,
+      "loss": 0.0739,
+      "step": 48030
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.9126047271072565e-05,
+      "loss": 0.2385,
+      "step": 48040
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.912470890548462e-05,
+      "loss": 0.1769,
+      "step": 48050
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.912337053989668e-05,
+      "loss": 0.5136,
+      "step": 48060
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.912203217430874e-05,
+      "loss": 0.154,
+      "step": 48070
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.912069380872079e-05,
+      "loss": 0.1843,
+      "step": 48080
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.911935544313285e-05,
+      "loss": 0.2804,
+      "step": 48090
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.9118017077544904e-05,
+      "loss": 0.3619,
+      "step": 48100
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.911667871195696e-05,
+      "loss": 0.3321,
+      "step": 48110
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.9115340346369014e-05,
+      "loss": 0.1615,
+      "step": 48120
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.911400198078107e-05,
+      "loss": 0.2532,
+      "step": 48130
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.9112663615193125e-05,
+      "loss": 0.2121,
+      "step": 48140
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.911132524960518e-05,
+      "loss": 0.2101,
+      "step": 48150
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.910998688401724e-05,
+      "loss": 0.0714,
+      "step": 48160
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.91086485184293e-05,
+      "loss": 0.1956,
+      "step": 48170
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.910731015284135e-05,
+      "loss": 0.2286,
+      "step": 48180
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.910597178725341e-05,
+      "loss": 0.2121,
+      "step": 48190
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.9104633421665464e-05,
+      "loss": 0.0894,
+      "step": 48200
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.910329505607752e-05,
+      "loss": 0.1815,
+      "step": 48210
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.9101956690489574e-05,
+      "loss": 0.3176,
+      "step": 48220
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.910061832490163e-05,
+      "loss": 0.2154,
+      "step": 48230
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.9099279959313685e-05,
+      "loss": 0.3393,
+      "step": 48240
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.909794159372574e-05,
+      "loss": 0.2094,
+      "step": 48250
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.90966032281378e-05,
+      "loss": 0.1904,
+      "step": 48260
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.909526486254986e-05,
+      "loss": 0.3708,
+      "step": 48270
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.909392649696191e-05,
+      "loss": 0.0382,
+      "step": 48280
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.909258813137397e-05,
+      "loss": 0.2781,
+      "step": 48290
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.9091249765786024e-05,
+      "loss": 0.2988,
+      "step": 48300
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.908991140019808e-05,
+      "loss": 0.3947,
+      "step": 48310
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.9088573034610135e-05,
+      "loss": 0.1276,
+      "step": 48320
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.908723466902219e-05,
+      "loss": 0.1691,
+      "step": 48330
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.9085896303434245e-05,
+      "loss": 0.1772,
+      "step": 48340
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.90845579378463e-05,
+      "loss": 0.0315,
+      "step": 48350
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.908321957225836e-05,
+      "loss": 0.2317,
+      "step": 48360
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.908188120667042e-05,
+      "loss": 0.3495,
+      "step": 48370
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.908054284108247e-05,
+      "loss": 0.3902,
+      "step": 48380
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.907920447549453e-05,
+      "loss": 0.2281,
+      "step": 48390
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.9077866109906584e-05,
+      "loss": 0.1642,
+      "step": 48400
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.907652774431864e-05,
+      "loss": 0.1109,
+      "step": 48410
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.9075189378730695e-05,
+      "loss": 0.1042,
+      "step": 48420
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.907385101314275e-05,
+      "loss": 0.2399,
+      "step": 48430
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.9072512647554805e-05,
+      "loss": 0.2738,
+      "step": 48440
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.907117428196686e-05,
+      "loss": 0.2338,
+      "step": 48450
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.906983591637892e-05,
+      "loss": 0.211,
+      "step": 48460
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.906849755079098e-05,
+      "loss": 0.1639,
+      "step": 48470
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.9067159185203034e-05,
+      "loss": 0.2135,
+      "step": 48480
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.906582081961509e-05,
+      "loss": 0.2391,
+      "step": 48490
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.9064482454027144e-05,
+      "loss": 0.2237,
+      "step": 48500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.90631440884392e-05,
+      "loss": 0.3012,
+      "step": 48510
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.9061805722851255e-05,
+      "loss": 0.3573,
+      "step": 48520
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.906046735726331e-05,
+      "loss": 0.3434,
+      "step": 48530
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.9059128991675366e-05,
+      "loss": 0.2299,
+      "step": 48540
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.905779062608742e-05,
+      "loss": 0.471,
+      "step": 48550
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.905645226049948e-05,
+      "loss": 0.2092,
+      "step": 48560
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.905511389491154e-05,
+      "loss": 0.1644,
+      "step": 48570
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.9053775529323594e-05,
+      "loss": 0.268,
+      "step": 48580
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.905243716373565e-05,
+      "loss": 0.397,
+      "step": 48590
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.9051098798147704e-05,
+      "loss": 0.2509,
+      "step": 48600
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.904976043255976e-05,
+      "loss": 0.0939,
+      "step": 48610
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.9048422066971815e-05,
+      "loss": 0.2963,
+      "step": 48620
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.904708370138387e-05,
+      "loss": 0.1174,
+      "step": 48630
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.9045745335795926e-05,
+      "loss": 0.225,
+      "step": 48640
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.904440697020799e-05,
+      "loss": 0.4745,
+      "step": 48650
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.904306860462004e-05,
+      "loss": 0.1528,
+      "step": 48660
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.90417302390321e-05,
+      "loss": 0.1071,
+      "step": 48670
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.9040391873444154e-05,
+      "loss": 0.2581,
+      "step": 48680
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.903905350785621e-05,
+      "loss": 0.0537,
+      "step": 48690
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.9037715142268265e-05,
+      "loss": 0.3696,
+      "step": 48700
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.903637677668032e-05,
+      "loss": 0.3345,
+      "step": 48710
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.9035038411092375e-05,
+      "loss": 0.504,
+      "step": 48720
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.903370004550443e-05,
+      "loss": 0.2693,
+      "step": 48730
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.9032361679916486e-05,
+      "loss": 0.1666,
+      "step": 48740
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.903102331432855e-05,
+      "loss": 0.2554,
+      "step": 48750
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.90296849487406e-05,
+      "loss": 0.389,
+      "step": 48760
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.902834658315266e-05,
+      "loss": 0.1989,
+      "step": 48770
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.9027008217564714e-05,
+      "loss": 0.0745,
+      "step": 48780
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.902566985197677e-05,
+      "loss": 0.373,
+      "step": 48790
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.9024331486388825e-05,
+      "loss": 0.1204,
+      "step": 48800
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.902299312080088e-05,
+      "loss": 0.2938,
+      "step": 48810
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.9021654755212935e-05,
+      "loss": 0.0894,
+      "step": 48820
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.902031638962499e-05,
+      "loss": 0.3802,
+      "step": 48830
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.9018978024037046e-05,
+      "loss": 0.0581,
+      "step": 48840
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.901763965844911e-05,
+      "loss": 0.1453,
+      "step": 48850
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.9016301292861163e-05,
+      "loss": 0.1935,
+      "step": 48860
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.901496292727322e-05,
+      "loss": 0.2438,
+      "step": 48870
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.9013624561685274e-05,
+      "loss": 0.1811,
+      "step": 48880
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.901228619609733e-05,
+      "loss": 0.1799,
+      "step": 48890
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.9010947830509385e-05,
+      "loss": 0.3365,
+      "step": 48900
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.900960946492144e-05,
+      "loss": 0.167,
+      "step": 48910
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.9008271099333495e-05,
+      "loss": 0.0753,
+      "step": 48920
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.900693273374555e-05,
+      "loss": 0.2713,
+      "step": 48930
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.9005594368157606e-05,
+      "loss": 0.474,
+      "step": 48940
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.900425600256967e-05,
+      "loss": 0.227,
+      "step": 48950
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.9002917636981724e-05,
+      "loss": 0.1176,
+      "step": 48960
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.900157927139378e-05,
+      "loss": 0.2557,
+      "step": 48970
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.9000240905805834e-05,
+      "loss": 0.1327,
+      "step": 48980
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.899890254021789e-05,
+      "loss": 0.2626,
+      "step": 48990
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.8997564174629945e-05,
+      "loss": 0.1328,
+      "step": 49000
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.8996225809042e-05,
+      "loss": 0.1483,
+      "step": 49010
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.8994887443454056e-05,
+      "loss": 0.5059,
+      "step": 49020
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.899354907786611e-05,
+      "loss": 0.1259,
+      "step": 49030
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.8992210712278166e-05,
+      "loss": 0.0668,
+      "step": 49040
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.899087234669023e-05,
+      "loss": 0.3025,
+      "step": 49050
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.8989533981102284e-05,
+      "loss": 0.2718,
+      "step": 49060
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.898819561551434e-05,
+      "loss": 0.2299,
+      "step": 49070
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.8986857249926394e-05,
+      "loss": 0.2505,
+      "step": 49080
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.898551888433845e-05,
+      "loss": 0.0686,
+      "step": 49090
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.8984180518750505e-05,
+      "loss": 0.2298,
+      "step": 49100
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.898284215316256e-05,
+      "loss": 0.4846,
+      "step": 49110
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.8981503787574616e-05,
+      "loss": 0.282,
+      "step": 49120
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.898016542198667e-05,
+      "loss": 0.2371,
+      "step": 49130
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.8978827056398726e-05,
+      "loss": 0.2378,
+      "step": 49140
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.897748869081079e-05,
+      "loss": 0.1516,
+      "step": 49150
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.8976150325222844e-05,
+      "loss": 0.2241,
+      "step": 49160
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.89748119596349e-05,
+      "loss": 0.4359,
+      "step": 49170
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.8973473594046955e-05,
+      "loss": 0.2601,
+      "step": 49180
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.897213522845901e-05,
+      "loss": 0.1292,
+      "step": 49190
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.8970796862871065e-05,
+      "loss": 0.1419,
+      "step": 49200
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.896945849728312e-05,
+      "loss": 0.1661,
+      "step": 49210
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.8968120131695176e-05,
+      "loss": 0.174,
+      "step": 49220
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.896678176610723e-05,
+      "loss": 0.1567,
+      "step": 49230
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.896544340051929e-05,
+      "loss": 0.1434,
+      "step": 49240
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.896410503493135e-05,
+      "loss": 0.082,
+      "step": 49250
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.8962766669343404e-05,
+      "loss": 0.2942,
+      "step": 49260
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.896142830375546e-05,
+      "loss": 0.1709,
+      "step": 49270
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.8960089938167515e-05,
+      "loss": 0.0834,
+      "step": 49280
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.895875157257957e-05,
+      "loss": 0.286,
+      "step": 49290
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.8957413206991625e-05,
+      "loss": 0.121,
+      "step": 49300
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.895607484140368e-05,
+      "loss": 0.1741,
+      "step": 49310
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.8954736475815736e-05,
+      "loss": 0.2094,
+      "step": 49320
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.895339811022779e-05,
+      "loss": 0.3082,
+      "step": 49330
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.895205974463985e-05,
+      "loss": 0.1644,
+      "step": 49340
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.895072137905191e-05,
+      "loss": 0.2321,
+      "step": 49350
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.8949383013463964e-05,
+      "loss": 0.2243,
+      "step": 49360
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.894804464787602e-05,
+      "loss": 0.2535,
+      "step": 49370
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.8946706282288075e-05,
+      "loss": 0.2429,
+      "step": 49380
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.894536791670013e-05,
+      "loss": 0.313,
+      "step": 49390
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.8944029551112186e-05,
+      "loss": 0.0981,
+      "step": 49400
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.894269118552424e-05,
+      "loss": 0.2704,
+      "step": 49410
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.8941352819936296e-05,
+      "loss": 0.2271,
+      "step": 49420
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.894001445434835e-05,
+      "loss": 0.1949,
+      "step": 49430
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.893867608876041e-05,
+      "loss": 0.1266,
+      "step": 49440
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.893733772317247e-05,
+      "loss": 0.3553,
+      "step": 49450
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.8935999357584524e-05,
+      "loss": 0.2543,
+      "step": 49460
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.893466099199658e-05,
+      "loss": 0.1566,
+      "step": 49470
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.8933322626408635e-05,
+      "loss": 0.3251,
+      "step": 49480
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.893198426082069e-05,
+      "loss": 0.3797,
+      "step": 49490
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.8930645895232746e-05,
+      "loss": 0.2327,
+      "step": 49500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.89293075296448e-05,
+      "loss": 0.2089,
+      "step": 49510
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.8927969164056856e-05,
+      "loss": 0.292,
+      "step": 49520
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.892663079846891e-05,
+      "loss": 0.1882,
+      "step": 49530
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.892529243288097e-05,
+      "loss": 0.1839,
+      "step": 49540
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.892395406729303e-05,
+      "loss": 0.4091,
+      "step": 49550
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.8922615701705085e-05,
+      "loss": 0.1854,
+      "step": 49560
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.892127733611714e-05,
+      "loss": 0.1292,
+      "step": 49570
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.8919938970529195e-05,
+      "loss": 0.1669,
+      "step": 49580
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.891860060494125e-05,
+      "loss": 0.1394,
+      "step": 49590
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.8917262239353306e-05,
+      "loss": 0.3699,
+      "step": 49600
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.891592387376536e-05,
+      "loss": 0.3723,
+      "step": 49610
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.8914585508177417e-05,
+      "loss": 0.2065,
+      "step": 49620
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.891324714258947e-05,
+      "loss": 0.2741,
+      "step": 49630
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.891190877700153e-05,
+      "loss": 0.4537,
+      "step": 49640
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.891057041141359e-05,
+      "loss": 0.2712,
+      "step": 49650
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.8909232045825645e-05,
+      "loss": 0.2977,
+      "step": 49660
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.89078936802377e-05,
+      "loss": 0.3531,
+      "step": 49670
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.8906555314649755e-05,
+      "loss": 0.1962,
+      "step": 49680
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.890521694906181e-05,
+      "loss": 0.2031,
+      "step": 49690
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.8903878583473866e-05,
+      "loss": 0.2072,
+      "step": 49700
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.890254021788592e-05,
+      "loss": 0.128,
+      "step": 49710
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.890120185229798e-05,
+      "loss": 0.249,
+      "step": 49720
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.889986348671003e-05,
+      "loss": 0.159,
+      "step": 49730
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.889852512112209e-05,
+      "loss": 0.2392,
+      "step": 49740
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.889718675553415e-05,
+      "loss": 0.0726,
+      "step": 49750
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.8895848389946205e-05,
+      "loss": 0.2796,
+      "step": 49760
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.889451002435826e-05,
+      "loss": 0.1245,
+      "step": 49770
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.8893171658770316e-05,
+      "loss": 0.1795,
+      "step": 49780
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.8891833293182364e-05,
+      "loss": 0.1571,
+      "step": 49790
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.889049492759442e-05,
+      "loss": 0.0835,
+      "step": 49800
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.8889156562006475e-05,
+      "loss": 0.2037,
+      "step": 49810
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": {
+        "accuracy": 0.9431496537187594
+      },
+      "eval_f1": {
+        "f1": 0.9428749572352995
+      },
+      "eval_loss": 0.2122853547334671,
+      "eval_precision": {
+        "precision": 0.9435965528419799
+      },
+      "eval_recall": {
+        "recall": 0.9429052318485974
+      },
+      "eval_runtime": 429.4215,
+      "eval_samples_per_second": 38.668,
+      "eval_steps_per_second": 19.335,
+      "step": 49815
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.888781819641853e-05,
+      "loss": 0.2541,
+      "step": 49820
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.888647983083059e-05,
+      "loss": 0.1736,
+      "step": 49830
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.888514146524265e-05,
+      "loss": 0.1454,
+      "step": 49840
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.88838030996547e-05,
+      "loss": 0.3127,
+      "step": 49850
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.888246473406676e-05,
+      "loss": 0.1076,
+      "step": 49860
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.8881126368478814e-05,
+      "loss": 0.2481,
+      "step": 49870
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.887978800289087e-05,
+      "loss": 0.2135,
+      "step": 49880
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.8878449637302924e-05,
+      "loss": 0.3033,
+      "step": 49890
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.887711127171498e-05,
+      "loss": 0.291,
+      "step": 49900
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.8875772906127035e-05,
+      "loss": 0.2638,
+      "step": 49910
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.887443454053909e-05,
+      "loss": 0.2007,
+      "step": 49920
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.887309617495115e-05,
+      "loss": 0.211,
+      "step": 49930
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.887175780936321e-05,
+      "loss": 0.3178,
+      "step": 49940
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.887041944377526e-05,
+      "loss": 0.166,
+      "step": 49950
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.886908107818732e-05,
+      "loss": 0.0669,
+      "step": 49960
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.8867742712599374e-05,
+      "loss": 0.0808,
+      "step": 49970
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.886640434701143e-05,
+      "loss": 0.126,
+      "step": 49980
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.8865065981423484e-05,
+      "loss": 0.26,
+      "step": 49990
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.886372761583554e-05,
+      "loss": 0.256,
+      "step": 50000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.8862389250247595e-05,
+      "loss": 0.2429,
+      "step": 50010
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.886105088465965e-05,
+      "loss": 0.311,
+      "step": 50020
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.885971251907171e-05,
+      "loss": 0.2152,
+      "step": 50030
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.885837415348377e-05,
+      "loss": 0.1005,
+      "step": 50040
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.885703578789582e-05,
+      "loss": 0.2445,
+      "step": 50050
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.885569742230788e-05,
+      "loss": 0.1649,
+      "step": 50060
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.8854359056719934e-05,
+      "loss": 0.1053,
+      "step": 50070
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.885302069113199e-05,
+      "loss": 0.3097,
+      "step": 50080
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.8851682325544045e-05,
+      "loss": 0.3131,
+      "step": 50090
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.88503439599561e-05,
+      "loss": 0.0764,
+      "step": 50100
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.8849005594368155e-05,
+      "loss": 0.2004,
+      "step": 50110
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.884766722878022e-05,
+      "loss": 0.1384,
+      "step": 50120
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.884632886319227e-05,
+      "loss": 0.1538,
+      "step": 50130
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.884499049760433e-05,
+      "loss": 0.3229,
+      "step": 50140
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.884365213201638e-05,
+      "loss": 0.3102,
+      "step": 50150
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.884231376642844e-05,
+      "loss": 0.2166,
+      "step": 50160
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.8840975400840494e-05,
+      "loss": 0.4435,
+      "step": 50170
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.883963703525255e-05,
+      "loss": 0.2628,
+      "step": 50180
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.8838298669664605e-05,
+      "loss": 0.1762,
+      "step": 50190
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.883696030407666e-05,
+      "loss": 0.2595,
+      "step": 50200
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.8835621938488715e-05,
+      "loss": 0.1848,
+      "step": 50210
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.883428357290078e-05,
+      "loss": 0.2922,
+      "step": 50220
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.883294520731283e-05,
+      "loss": 0.3401,
+      "step": 50230
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.883160684172489e-05,
+      "loss": 0.1195,
+      "step": 50240
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.8830268476136943e-05,
+      "loss": 0.3083,
+      "step": 50250
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.8828930110549e-05,
+      "loss": 0.2391,
+      "step": 50260
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.8827591744961054e-05,
+      "loss": 0.3924,
+      "step": 50270
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.882625337937311e-05,
+      "loss": 0.274,
+      "step": 50280
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.8824915013785165e-05,
+      "loss": 0.0843,
+      "step": 50290
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.882357664819722e-05,
+      "loss": 0.1395,
+      "step": 50300
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.8822238282609276e-05,
+      "loss": 0.0671,
+      "step": 50310
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.882089991702134e-05,
+      "loss": 0.1927,
+      "step": 50320
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.881956155143339e-05,
+      "loss": 0.2928,
+      "step": 50330
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.881822318584545e-05,
+      "loss": 0.1639,
+      "step": 50340
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.8816884820257504e-05,
+      "loss": 0.2159,
+      "step": 50350
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.881554645466956e-05,
+      "loss": 0.2892,
+      "step": 50360
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.8814208089081614e-05,
+      "loss": 0.1974,
+      "step": 50370
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.881286972349367e-05,
+      "loss": 0.2068,
+      "step": 50380
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.8811531357905725e-05,
+      "loss": 0.1965,
+      "step": 50390
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.881019299231778e-05,
+      "loss": 0.0669,
+      "step": 50400
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.8808854626729836e-05,
+      "loss": 0.2858,
+      "step": 50410
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.88075162611419e-05,
+      "loss": 0.1847,
+      "step": 50420
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.880617789555395e-05,
+      "loss": 0.0746,
+      "step": 50430
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.880483952996601e-05,
+      "loss": 0.3721,
+      "step": 50440
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.8803501164378064e-05,
+      "loss": 0.2055,
+      "step": 50450
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.880216279879012e-05,
+      "loss": 0.3856,
+      "step": 50460
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.8800824433202174e-05,
+      "loss": 0.3542,
+      "step": 50470
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.879948606761423e-05,
+      "loss": 0.2947,
+      "step": 50480
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.8798147702026285e-05,
+      "loss": 0.1796,
+      "step": 50490
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.879680933643834e-05,
+      "loss": 0.2391,
+      "step": 50500
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.8795470970850396e-05,
+      "loss": 0.0475,
+      "step": 50510
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.879413260526246e-05,
+      "loss": 0.182,
+      "step": 50520
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.879279423967451e-05,
+      "loss": 0.0941,
+      "step": 50530
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.879145587408657e-05,
+      "loss": 0.1752,
+      "step": 50540
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.8790117508498624e-05,
+      "loss": 0.1553,
+      "step": 50550
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.878877914291068e-05,
+      "loss": 0.2338,
+      "step": 50560
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.8787440777322735e-05,
+      "loss": 0.1576,
+      "step": 50570
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.878610241173479e-05,
+      "loss": 0.2143,
+      "step": 50580
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.8784764046146845e-05,
+      "loss": 0.1475,
+      "step": 50590
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.87834256805589e-05,
+      "loss": 0.1553,
+      "step": 50600
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.8782087314970956e-05,
+      "loss": 0.2403,
+      "step": 50610
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.878074894938302e-05,
+      "loss": 0.2661,
+      "step": 50620
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.8779410583795073e-05,
+      "loss": 0.2005,
+      "step": 50630
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.877807221820713e-05,
+      "loss": 0.2595,
+      "step": 50640
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.8776733852619184e-05,
+      "loss": 0.1674,
+      "step": 50650
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.877539548703124e-05,
+      "loss": 0.2424,
+      "step": 50660
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.8774057121443295e-05,
+      "loss": 0.165,
+      "step": 50670
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.877271875585535e-05,
+      "loss": 0.1534,
+      "step": 50680
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.8771380390267405e-05,
+      "loss": 0.231,
+      "step": 50690
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.877004202467946e-05,
+      "loss": 0.1239,
+      "step": 50700
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.8768703659091516e-05,
+      "loss": 0.2121,
+      "step": 50710
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.876736529350358e-05,
+      "loss": 0.339,
+      "step": 50720
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.8766026927915634e-05,
+      "loss": 0.1943,
+      "step": 50730
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.876468856232769e-05,
+      "loss": 0.2178,
+      "step": 50740
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.8763350196739744e-05,
+      "loss": 0.1022,
+      "step": 50750
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.87620118311518e-05,
+      "loss": 0.3314,
+      "step": 50760
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.8760673465563855e-05,
+      "loss": 0.2577,
+      "step": 50770
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.875933509997591e-05,
+      "loss": 0.0599,
+      "step": 50780
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.8757996734387966e-05,
+      "loss": 0.2537,
+      "step": 50790
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.875665836880002e-05,
+      "loss": 0.1527,
+      "step": 50800
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.8755320003212076e-05,
+      "loss": 0.1638,
+      "step": 50810
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.875398163762414e-05,
+      "loss": 0.2025,
+      "step": 50820
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.8752643272036194e-05,
+      "loss": 0.2411,
+      "step": 50830
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.875130490644825e-05,
+      "loss": 0.2086,
+      "step": 50840
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.8749966540860304e-05,
+      "loss": 0.1491,
+      "step": 50850
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.874862817527236e-05,
+      "loss": 0.1303,
+      "step": 50860
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.8747289809684415e-05,
+      "loss": 0.2692,
+      "step": 50870
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.874595144409647e-05,
+      "loss": 0.1479,
+      "step": 50880
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.8744613078508526e-05,
+      "loss": 0.2428,
+      "step": 50890
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.874327471292058e-05,
+      "loss": 0.2395,
+      "step": 50900
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.8741936347332636e-05,
+      "loss": 0.13,
+      "step": 50910
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.87405979817447e-05,
+      "loss": 0.2514,
+      "step": 50920
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.8739259616156754e-05,
+      "loss": 0.207,
+      "step": 50930
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.873792125056881e-05,
+      "loss": 0.0985,
+      "step": 50940
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.8736582884980865e-05,
+      "loss": 0.1778,
+      "step": 50950
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.873524451939292e-05,
+      "loss": 0.2916,
+      "step": 50960
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.8733906153804975e-05,
+      "loss": 0.2104,
+      "step": 50970
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.873256778821703e-05,
+      "loss": 0.4131,
+      "step": 50980
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.8731229422629086e-05,
+      "loss": 0.2777,
+      "step": 50990
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.872989105704114e-05,
+      "loss": 0.1819,
+      "step": 51000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.8728552691453197e-05,
+      "loss": 0.2289,
+      "step": 51010
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.872721432586526e-05,
+      "loss": 0.4877,
+      "step": 51020
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.8725875960277314e-05,
+      "loss": 0.1488,
+      "step": 51030
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.872453759468937e-05,
+      "loss": 0.2299,
+      "step": 51040
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.8723199229101425e-05,
+      "loss": 0.1999,
+      "step": 51050
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.872186086351348e-05,
+      "loss": 0.0809,
+      "step": 51060
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.8720522497925535e-05,
+      "loss": 0.1132,
+      "step": 51070
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.871918413233759e-05,
+      "loss": 0.0757,
+      "step": 51080
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.8717845766749646e-05,
+      "loss": 0.2523,
+      "step": 51090
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.87165074011617e-05,
+      "loss": 0.1765,
+      "step": 51100
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.871516903557376e-05,
+      "loss": 0.3265,
+      "step": 51110
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.871383066998582e-05,
+      "loss": 0.2717,
+      "step": 51120
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.8712492304397874e-05,
+      "loss": 0.149,
+      "step": 51130
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.871115393880993e-05,
+      "loss": 0.1559,
+      "step": 51140
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.8709815573221985e-05,
+      "loss": 0.1291,
+      "step": 51150
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.870847720763404e-05,
+      "loss": 0.24,
+      "step": 51160
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.8707138842046096e-05,
+      "loss": 0.1067,
+      "step": 51170
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.870580047645815e-05,
+      "loss": 0.1667,
+      "step": 51180
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.8704462110870206e-05,
+      "loss": 0.1261,
+      "step": 51190
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.870312374528226e-05,
+      "loss": 0.2378,
+      "step": 51200
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.870178537969432e-05,
+      "loss": 0.4325,
+      "step": 51210
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.870044701410638e-05,
+      "loss": 0.2655,
+      "step": 51220
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.8699108648518434e-05,
+      "loss": 0.4878,
+      "step": 51230
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.869777028293049e-05,
+      "loss": 0.2246,
+      "step": 51240
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.8696431917342545e-05,
+      "loss": 0.0657,
+      "step": 51250
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.86950935517546e-05,
+      "loss": 0.2355,
+      "step": 51260
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.8693755186166656e-05,
+      "loss": 0.1952,
+      "step": 51270
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.869241682057871e-05,
+      "loss": 0.4113,
+      "step": 51280
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.8691078454990766e-05,
+      "loss": 0.1733,
+      "step": 51290
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.868974008940282e-05,
+      "loss": 0.2582,
+      "step": 51300
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.868840172381488e-05,
+      "loss": 0.3326,
+      "step": 51310
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.868706335822694e-05,
+      "loss": 0.1354,
+      "step": 51320
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.8685724992638994e-05,
+      "loss": 0.2665,
+      "step": 51330
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.868438662705105e-05,
+      "loss": 0.1525,
+      "step": 51340
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.8683048261463105e-05,
+      "loss": 0.2738,
+      "step": 51350
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.868170989587516e-05,
+      "loss": 0.1162,
+      "step": 51360
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.8680371530287216e-05,
+      "loss": 0.3386,
+      "step": 51370
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.867903316469927e-05,
+      "loss": 0.3314,
+      "step": 51380
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.8677694799111327e-05,
+      "loss": 0.2097,
+      "step": 51390
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.867635643352338e-05,
+      "loss": 0.3992,
+      "step": 51400
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.8675018067935444e-05,
+      "loss": 0.1603,
+      "step": 51410
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.86736797023475e-05,
+      "loss": 0.1467,
+      "step": 51420
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.8672341336759555e-05,
+      "loss": 0.0716,
+      "step": 51430
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.867100297117161e-05,
+      "loss": 0.1387,
+      "step": 51440
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.8669664605583665e-05,
+      "loss": 0.1522,
+      "step": 51450
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.866832623999572e-05,
+      "loss": 0.1705,
+      "step": 51460
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.8666987874407776e-05,
+      "loss": 0.2444,
+      "step": 51470
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.866564950881983e-05,
+      "loss": 0.0261,
+      "step": 51480
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.866431114323189e-05,
+      "loss": 0.1285,
+      "step": 51490
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.866297277764394e-05,
+      "loss": 0.1276,
+      "step": 51500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.8661634412056004e-05,
+      "loss": 0.1743,
+      "step": 51510
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.866029604646806e-05,
+      "loss": 0.3369,
+      "step": 51520
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.8658957680880115e-05,
+      "loss": 0.1718,
+      "step": 51530
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.865761931529217e-05,
+      "loss": 0.3083,
+      "step": 51540
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.8656280949704225e-05,
+      "loss": 0.2287,
+      "step": 51550
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.865494258411628e-05,
+      "loss": 0.3323,
+      "step": 51560
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.8653604218528336e-05,
+      "loss": 0.2433,
+      "step": 51570
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.865226585294039e-05,
+      "loss": 0.1067,
+      "step": 51580
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.865092748735245e-05,
+      "loss": 0.4664,
+      "step": 51590
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.86495891217645e-05,
+      "loss": 0.2586,
+      "step": 51600
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.8648250756176564e-05,
+      "loss": 0.2515,
+      "step": 51610
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.864691239058862e-05,
+      "loss": 0.1822,
+      "step": 51620
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.8645574025000675e-05,
+      "loss": 0.2806,
+      "step": 51630
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.864423565941273e-05,
+      "loss": 0.1306,
+      "step": 51640
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.8642897293824786e-05,
+      "loss": 0.1927,
+      "step": 51650
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.864155892823684e-05,
+      "loss": 0.3695,
+      "step": 51660
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.8640220562648896e-05,
+      "loss": 0.145,
+      "step": 51670
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.863888219706095e-05,
+      "loss": 0.1272,
+      "step": 51680
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.863754383147301e-05,
+      "loss": 0.1023,
+      "step": 51690
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.863620546588506e-05,
+      "loss": 0.1922,
+      "step": 51700
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.8634867100297124e-05,
+      "loss": 0.1734,
+      "step": 51710
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.863352873470918e-05,
+      "loss": 0.154,
+      "step": 51720
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.8632190369121235e-05,
+      "loss": 0.3525,
+      "step": 51730
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.863085200353329e-05,
+      "loss": 0.1661,
+      "step": 51740
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.8629513637945346e-05,
+      "loss": 0.2997,
+      "step": 51750
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.86281752723574e-05,
+      "loss": 0.4147,
+      "step": 51760
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.8626836906769456e-05,
+      "loss": 0.2132,
+      "step": 51770
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.862549854118151e-05,
+      "loss": 0.2624,
+      "step": 51780
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.862416017559357e-05,
+      "loss": 0.1523,
+      "step": 51790
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.862282181000562e-05,
+      "loss": 0.1773,
+      "step": 51800
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.8621483444417685e-05,
+      "loss": 0.458,
+      "step": 51810
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.862014507882974e-05,
+      "loss": 0.2304,
+      "step": 51820
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.8618806713241795e-05,
+      "loss": 0.2196,
+      "step": 51830
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.861746834765385e-05,
+      "loss": 0.1886,
+      "step": 51840
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.8616129982065906e-05,
+      "loss": 0.1309,
+      "step": 51850
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.861479161647796e-05,
+      "loss": 0.3642,
+      "step": 51860
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.8613453250890017e-05,
+      "loss": 0.119,
+      "step": 51870
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.861211488530207e-05,
+      "loss": 0.2117,
+      "step": 51880
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.861077651971413e-05,
+      "loss": 0.1769,
+      "step": 51890
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.860943815412618e-05,
+      "loss": 0.1428,
+      "step": 51900
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.8608099788538245e-05,
+      "loss": 0.2328,
+      "step": 51910
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.86067614229503e-05,
+      "loss": 0.1368,
+      "step": 51920
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.8605423057362355e-05,
+      "loss": 0.1856,
+      "step": 51930
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.860408469177441e-05,
+      "loss": 0.0949,
+      "step": 51940
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.8602746326186466e-05,
+      "loss": 0.2184,
+      "step": 51950
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.860140796059852e-05,
+      "loss": 0.1568,
+      "step": 51960
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.860006959501058e-05,
+      "loss": 0.3672,
+      "step": 51970
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.859873122942263e-05,
+      "loss": 0.4091,
+      "step": 51980
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.859739286383469e-05,
+      "loss": 0.2995,
+      "step": 51990
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.859605449824674e-05,
+      "loss": 0.1453,
+      "step": 52000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.8594716132658805e-05,
+      "loss": 0.164,
+      "step": 52010
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.859337776707086e-05,
+      "loss": 0.1737,
+      "step": 52020
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.8592039401482916e-05,
+      "loss": 0.1081,
+      "step": 52030
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.859070103589497e-05,
+      "loss": 0.2036,
+      "step": 52040
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.8589362670307026e-05,
+      "loss": 0.1283,
+      "step": 52050
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.858802430471908e-05,
+      "loss": 0.0878,
+      "step": 52060
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.858668593913114e-05,
+      "loss": 0.2083,
+      "step": 52070
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.858534757354319e-05,
+      "loss": 0.5313,
+      "step": 52080
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.858400920795525e-05,
+      "loss": 0.3381,
+      "step": 52090
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.85826708423673e-05,
+      "loss": 0.1743,
+      "step": 52100
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.8581332476779365e-05,
+      "loss": 0.4948,
+      "step": 52110
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.857999411119142e-05,
+      "loss": 0.0984,
+      "step": 52120
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.8578655745603476e-05,
+      "loss": 0.138,
+      "step": 52130
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.857731738001553e-05,
+      "loss": 0.1918,
+      "step": 52140
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.8575979014427586e-05,
+      "loss": 0.3109,
+      "step": 52150
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.8574640648839635e-05,
+      "loss": 0.2625,
+      "step": 52160
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.857330228325169e-05,
+      "loss": 0.1556,
+      "step": 52170
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.8571963917663746e-05,
+      "loss": 0.1876,
+      "step": 52180
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.857062555207581e-05,
+      "loss": 0.2828,
+      "step": 52190
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.856928718648786e-05,
+      "loss": 0.2004,
+      "step": 52200
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.856794882089992e-05,
+      "loss": 0.3141,
+      "step": 52210
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.8566610455311974e-05,
+      "loss": 0.1029,
+      "step": 52220
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.856527208972403e-05,
+      "loss": 0.2363,
+      "step": 52230
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.8563933724136084e-05,
+      "loss": 0.269,
+      "step": 52240
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.856259535854814e-05,
+      "loss": 0.1807,
+      "step": 52250
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.8561256992960195e-05,
+      "loss": 0.2594,
+      "step": 52260
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.855991862737225e-05,
+      "loss": 0.3095,
+      "step": 52270
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.8558580261784306e-05,
+      "loss": 0.3403,
+      "step": 52280
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.855724189619637e-05,
+      "loss": 0.0238,
+      "step": 52290
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.855590353060842e-05,
+      "loss": 0.0944,
+      "step": 52300
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.855456516502048e-05,
+      "loss": 0.1338,
+      "step": 52310
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.8553226799432534e-05,
+      "loss": 0.1456,
+      "step": 52320
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.855188843384459e-05,
+      "loss": 0.1942,
+      "step": 52330
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.8550550068256645e-05,
+      "loss": 0.2261,
+      "step": 52340
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.85492117026687e-05,
+      "loss": 0.0919,
+      "step": 52350
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.8547873337080755e-05,
+      "loss": 0.157,
+      "step": 52360
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.854653497149281e-05,
+      "loss": 0.1528,
+      "step": 52370
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.8545196605904866e-05,
+      "loss": 0.1025,
+      "step": 52380
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.854385824031693e-05,
+      "loss": 0.2865,
+      "step": 52390
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.854251987472898e-05,
+      "loss": 0.1094,
+      "step": 52400
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.854118150914104e-05,
+      "loss": 0.124,
+      "step": 52410
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.8539843143553094e-05,
+      "loss": 0.2986,
+      "step": 52420
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.853850477796515e-05,
+      "loss": 0.2572,
+      "step": 52430
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.8537166412377205e-05,
+      "loss": 0.0917,
+      "step": 52440
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.853582804678926e-05,
+      "loss": 0.1501,
+      "step": 52450
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.8534489681201315e-05,
+      "loss": 0.5314,
+      "step": 52460
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.853315131561337e-05,
+      "loss": 0.2718,
+      "step": 52470
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.8531812950025426e-05,
+      "loss": 0.1073,
+      "step": 52480
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.853047458443749e-05,
+      "loss": 0.3184,
+      "step": 52490
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.8529136218849544e-05,
+      "loss": 0.1081,
+      "step": 52500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.85277978532616e-05,
+      "loss": 0.2514,
+      "step": 52510
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.8526459487673654e-05,
+      "loss": 0.2867,
+      "step": 52520
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.852512112208571e-05,
+      "loss": 0.3708,
+      "step": 52530
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.8523782756497765e-05,
+      "loss": 0.1443,
+      "step": 52540
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.852244439090982e-05,
+      "loss": 0.2883,
+      "step": 52550
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.8521106025321876e-05,
+      "loss": 0.3747,
+      "step": 52560
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.851976765973393e-05,
+      "loss": 0.2479,
+      "step": 52570
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.8518429294145986e-05,
+      "loss": 0.2091,
+      "step": 52580
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.851709092855805e-05,
+      "loss": 0.372,
+      "step": 52590
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.8515752562970104e-05,
+      "loss": 0.1688,
+      "step": 52600
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.851441419738216e-05,
+      "loss": 0.0917,
+      "step": 52610
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.8513075831794214e-05,
+      "loss": 0.0677,
+      "step": 52620
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.851173746620627e-05,
+      "loss": 0.0878,
+      "step": 52630
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.8510399100618325e-05,
+      "loss": 0.1741,
+      "step": 52640
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.850906073503038e-05,
+      "loss": 0.2014,
+      "step": 52650
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.8507722369442436e-05,
+      "loss": 0.1995,
+      "step": 52660
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.850638400385449e-05,
+      "loss": 0.093,
+      "step": 52670
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.8505045638266546e-05,
+      "loss": 0.2725,
+      "step": 52680
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.850370727267861e-05,
+      "loss": 0.164,
+      "step": 52690
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.8502368907090664e-05,
+      "loss": 0.192,
+      "step": 52700
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.850103054150272e-05,
+      "loss": 0.1241,
+      "step": 52710
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.8499692175914775e-05,
+      "loss": 0.0789,
+      "step": 52720
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.849835381032683e-05,
+      "loss": 0.0635,
+      "step": 52730
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.8497015444738885e-05,
+      "loss": 0.1901,
+      "step": 52740
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.849567707915094e-05,
+      "loss": 0.1994,
+      "step": 52750
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.8494338713562996e-05,
+      "loss": 0.0274,
+      "step": 52760
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.849300034797505e-05,
+      "loss": 0.2237,
+      "step": 52770
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.8491661982387107e-05,
+      "loss": 0.1925,
+      "step": 52780
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.849032361679917e-05,
+      "loss": 0.1161,
+      "step": 52790
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.8488985251211224e-05,
+      "loss": 0.135,
+      "step": 52800
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.848764688562328e-05,
+      "loss": 0.0825,
+      "step": 52810
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.8486308520035335e-05,
+      "loss": 0.203,
+      "step": 52820
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.848497015444739e-05,
+      "loss": 0.2522,
+      "step": 52830
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.8483631788859445e-05,
+      "loss": 0.0652,
+      "step": 52840
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.84822934232715e-05,
+      "loss": 0.4203,
+      "step": 52850
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.8480955057683556e-05,
+      "loss": 0.2555,
+      "step": 52860
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.847961669209561e-05,
+      "loss": 0.2357,
+      "step": 52870
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.8478278326507673e-05,
+      "loss": 0.1741,
+      "step": 52880
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.847693996091973e-05,
+      "loss": 0.1885,
+      "step": 52890
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.8475601595331784e-05,
+      "loss": 0.3141,
+      "step": 52900
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.847426322974384e-05,
+      "loss": 0.0461,
+      "step": 52910
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.8472924864155895e-05,
+      "loss": 0.3681,
+      "step": 52920
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.847158649856795e-05,
+      "loss": 0.126,
+      "step": 52930
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.8470248132980005e-05,
+      "loss": 0.3572,
+      "step": 52940
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.846890976739206e-05,
+      "loss": 0.1428,
+      "step": 52950
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.8467571401804116e-05,
+      "loss": 0.3029,
+      "step": 52960
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.846623303621617e-05,
+      "loss": 0.2197,
+      "step": 52970
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.8464894670628234e-05,
+      "loss": 0.2177,
+      "step": 52980
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.846355630504029e-05,
+      "loss": 0.2907,
+      "step": 52990
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.8462217939452344e-05,
+      "loss": 0.0821,
+      "step": 53000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.84608795738644e-05,
+      "loss": 0.0993,
+      "step": 53010
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.8459541208276455e-05,
+      "loss": 0.156,
+      "step": 53020
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.845820284268851e-05,
+      "loss": 0.2594,
+      "step": 53030
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.8456864477100566e-05,
+      "loss": 0.309,
+      "step": 53040
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.845552611151262e-05,
+      "loss": 0.2917,
+      "step": 53050
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.8454187745924676e-05,
+      "loss": 0.1329,
+      "step": 53060
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.845284938033673e-05,
+      "loss": 0.1067,
+      "step": 53070
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.8451511014748794e-05,
+      "loss": 0.4407,
+      "step": 53080
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.845017264916085e-05,
+      "loss": 0.0693,
+      "step": 53090
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.8448834283572904e-05,
+      "loss": 0.0947,
+      "step": 53100
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.844749591798496e-05,
+      "loss": 0.1738,
+      "step": 53110
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.8446157552397015e-05,
+      "loss": 0.3084,
+      "step": 53120
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.844481918680907e-05,
+      "loss": 0.1641,
+      "step": 53130
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.8443480821221126e-05,
+      "loss": 0.1631,
+      "step": 53140
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.844214245563318e-05,
+      "loss": 0.0783,
+      "step": 53150
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.8440804090045236e-05,
+      "loss": 0.2637,
+      "step": 53160
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.843946572445729e-05,
+      "loss": 0.3085,
+      "step": 53170
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.8438127358869354e-05,
+      "loss": 0.2831,
+      "step": 53180
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.843678899328141e-05,
+      "loss": 0.4079,
+      "step": 53190
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.8435450627693465e-05,
+      "loss": 0.048,
+      "step": 53200
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.843411226210552e-05,
+      "loss": 0.1794,
+      "step": 53210
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.8432773896517575e-05,
+      "loss": 0.1698,
+      "step": 53220
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.843143553092963e-05,
+      "loss": 0.1521,
+      "step": 53230
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.8430097165341686e-05,
+      "loss": 0.1639,
+      "step": 53240
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.842875879975374e-05,
+      "loss": 0.0713,
+      "step": 53250
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.84274204341658e-05,
+      "loss": 0.33,
+      "step": 53260
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.842608206857785e-05,
+      "loss": 0.1966,
+      "step": 53270
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.8424743702989914e-05,
+      "loss": 0.3088,
+      "step": 53280
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.842340533740197e-05,
+      "loss": 0.2321,
+      "step": 53290
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.8422066971814025e-05,
+      "loss": 0.3732,
+      "step": 53300
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.842072860622608e-05,
+      "loss": 0.1511,
+      "step": 53310
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.8419390240638135e-05,
+      "loss": 0.257,
+      "step": 53320
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.841805187505019e-05,
+      "loss": 0.2414,
+      "step": 53330
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.8416713509462246e-05,
+      "loss": 0.2402,
+      "step": 53340
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.84153751438743e-05,
+      "loss": 0.2877,
+      "step": 53350
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.841403677828636e-05,
+      "loss": 0.3101,
+      "step": 53360
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.841269841269841e-05,
+      "loss": 0.0915,
+      "step": 53370
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.8411360047110474e-05,
+      "loss": 0.226,
+      "step": 53380
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.841002168152253e-05,
+      "loss": 0.0862,
+      "step": 53390
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.8408683315934585e-05,
+      "loss": 0.1972,
+      "step": 53400
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.840734495034664e-05,
+      "loss": 0.0772,
+      "step": 53410
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.8406006584758696e-05,
+      "loss": 0.2357,
+      "step": 53420
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.840466821917075e-05,
+      "loss": 0.3421,
+      "step": 53430
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.8403329853582806e-05,
+      "loss": 0.0545,
+      "step": 53440
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.840199148799486e-05,
+      "loss": 0.0836,
+      "step": 53450
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.840065312240692e-05,
+      "loss": 0.178,
+      "step": 53460
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.839931475681897e-05,
+      "loss": 0.1138,
+      "step": 53470
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.8397976391231034e-05,
+      "loss": 0.0905,
+      "step": 53480
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.839663802564309e-05,
+      "loss": 0.3885,
+      "step": 53490
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.8395299660055145e-05,
+      "loss": 0.1964,
+      "step": 53500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.83939612944672e-05,
+      "loss": 0.2607,
+      "step": 53510
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.8392622928879256e-05,
+      "loss": 0.1842,
+      "step": 53520
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.839128456329131e-05,
+      "loss": 0.2014,
+      "step": 53530
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.8389946197703366e-05,
+      "loss": 0.0566,
+      "step": 53540
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.838860783211542e-05,
+      "loss": 0.2624,
+      "step": 53550
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.838726946652748e-05,
+      "loss": 0.367,
+      "step": 53560
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.838593110093953e-05,
+      "loss": 0.3489,
+      "step": 53570
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.8384592735351595e-05,
+      "loss": 0.065,
+      "step": 53580
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.838325436976365e-05,
+      "loss": 0.2996,
+      "step": 53590
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.8381916004175705e-05,
+      "loss": 0.1529,
+      "step": 53600
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.838057763858776e-05,
+      "loss": 0.3455,
+      "step": 53610
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.8379239272999816e-05,
+      "loss": 0.4916,
+      "step": 53620
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.837790090741187e-05,
+      "loss": 0.0585,
+      "step": 53630
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.8376562541823927e-05,
+      "loss": 0.1728,
+      "step": 53640
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.837522417623598e-05,
+      "loss": 0.1882,
+      "step": 53650
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.837388581064804e-05,
+      "loss": 0.0536,
+      "step": 53660
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.837254744506009e-05,
+      "loss": 0.2588,
+      "step": 53670
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.8371209079472155e-05,
+      "loss": 0.2429,
+      "step": 53680
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.836987071388421e-05,
+      "loss": 0.1544,
+      "step": 53690
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.8368532348296265e-05,
+      "loss": 0.1546,
+      "step": 53700
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.836719398270832e-05,
+      "loss": 0.2205,
+      "step": 53710
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.8365855617120376e-05,
+      "loss": 0.1807,
+      "step": 53720
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.836451725153243e-05,
+      "loss": 0.4361,
+      "step": 53730
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.836317888594449e-05,
+      "loss": 0.3393,
+      "step": 53740
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.836184052035654e-05,
+      "loss": 0.1547,
+      "step": 53750
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.83605021547686e-05,
+      "loss": 0.1789,
+      "step": 53760
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.835916378918065e-05,
+      "loss": 0.1191,
+      "step": 53770
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.8357825423592715e-05,
+      "loss": 0.0969,
+      "step": 53780
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.835648705800477e-05,
+      "loss": 0.0698,
+      "step": 53790
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.8355148692416826e-05,
+      "loss": 0.2284,
+      "step": 53800
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.835381032682888e-05,
+      "loss": 0.3161,
+      "step": 53810
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.8352471961240936e-05,
+      "loss": 0.2259,
+      "step": 53820
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.835113359565299e-05,
+      "loss": 0.2072,
+      "step": 53830
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.834979523006505e-05,
+      "loss": 0.217,
+      "step": 53840
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.83484568644771e-05,
+      "loss": 0.4043,
+      "step": 53850
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.834711849888916e-05,
+      "loss": 0.285,
+      "step": 53860
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.834578013330121e-05,
+      "loss": 0.1112,
+      "step": 53870
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.8344441767713275e-05,
+      "loss": 0.1203,
+      "step": 53880
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.834310340212533e-05,
+      "loss": 0.2783,
+      "step": 53890
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.8341765036537386e-05,
+      "loss": 0.2627,
+      "step": 53900
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.834042667094944e-05,
+      "loss": 0.1803,
+      "step": 53910
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.8339088305361496e-05,
+      "loss": 0.0853,
+      "step": 53920
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.833774993977355e-05,
+      "loss": 0.207,
+      "step": 53930
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.833641157418561e-05,
+      "loss": 0.3161,
+      "step": 53940
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.833507320859766e-05,
+      "loss": 0.1781,
+      "step": 53950
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.833373484300972e-05,
+      "loss": 0.1125,
+      "step": 53960
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.833239647742177e-05,
+      "loss": 0.1271,
+      "step": 53970
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.8331058111833835e-05,
+      "loss": 0.1516,
+      "step": 53980
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.832971974624589e-05,
+      "loss": 0.1927,
+      "step": 53990
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.8328381380657946e-05,
+      "loss": 0.3408,
+      "step": 54000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.832704301507e-05,
+      "loss": 0.0859,
+      "step": 54010
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.8325704649482056e-05,
+      "loss": 0.1428,
+      "step": 54020
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.832436628389411e-05,
+      "loss": 0.3033,
+      "step": 54030
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.832302791830617e-05,
+      "loss": 0.3753,
+      "step": 54040
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.832168955271822e-05,
+      "loss": 0.1626,
+      "step": 54050
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.832035118713028e-05,
+      "loss": 0.1841,
+      "step": 54060
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.831901282154233e-05,
+      "loss": 0.3413,
+      "step": 54070
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.8317674455954395e-05,
+      "loss": 0.1902,
+      "step": 54080
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.831633609036645e-05,
+      "loss": 0.1558,
+      "step": 54090
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.8314997724778506e-05,
+      "loss": 0.1712,
+      "step": 54100
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.831365935919056e-05,
+      "loss": 0.1191,
+      "step": 54110
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.831232099360262e-05,
+      "loss": 0.1468,
+      "step": 54120
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.831098262801467e-05,
+      "loss": 0.1789,
+      "step": 54130
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.830964426242673e-05,
+      "loss": 0.1425,
+      "step": 54140
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.830830589683878e-05,
+      "loss": 0.5177,
+      "step": 54150
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.830696753125084e-05,
+      "loss": 0.3279,
+      "step": 54160
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.83056291656629e-05,
+      "loss": 0.1881,
+      "step": 54170
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.8304290800074955e-05,
+      "loss": 0.2972,
+      "step": 54180
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.830295243448701e-05,
+      "loss": 0.3208,
+      "step": 54190
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.8301614068899066e-05,
+      "loss": 0.3601,
+      "step": 54200
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.830027570331112e-05,
+      "loss": 0.1271,
+      "step": 54210
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.829893733772318e-05,
+      "loss": 0.2198,
+      "step": 54220
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.829759897213523e-05,
+      "loss": 0.0918,
+      "step": 54230
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.829626060654729e-05,
+      "loss": 0.1302,
+      "step": 54240
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.829492224095934e-05,
+      "loss": 0.1961,
+      "step": 54250
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.82935838753714e-05,
+      "loss": 0.207,
+      "step": 54260
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.829224550978346e-05,
+      "loss": 0.1686,
+      "step": 54270
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.8290907144195516e-05,
+      "loss": 0.0918,
+      "step": 54280
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.828956877860757e-05,
+      "loss": 0.2767,
+      "step": 54290
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.8288230413019626e-05,
+      "loss": 0.2348,
+      "step": 54300
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.828689204743168e-05,
+      "loss": 0.0937,
+      "step": 54310
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.828555368184374e-05,
+      "loss": 0.2525,
+      "step": 54320
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.828421531625579e-05,
+      "loss": 0.2283,
+      "step": 54330
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.828287695066785e-05,
+      "loss": 0.1044,
+      "step": 54340
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.82815385850799e-05,
+      "loss": 0.2909,
+      "step": 54350
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.828020021949196e-05,
+      "loss": 0.1841,
+      "step": 54360
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.827886185390402e-05,
+      "loss": 0.2098,
+      "step": 54370
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.8277523488316076e-05,
+      "loss": 0.2283,
+      "step": 54380
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.827618512272813e-05,
+      "loss": 0.2188,
+      "step": 54390
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.8274846757140186e-05,
+      "loss": 0.4155,
+      "step": 54400
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.827350839155224e-05,
+      "loss": 0.3383,
+      "step": 54410
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.82721700259643e-05,
+      "loss": 0.2426,
+      "step": 54420
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.827083166037635e-05,
+      "loss": 0.2832,
+      "step": 54430
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.826949329478841e-05,
+      "loss": 0.2556,
+      "step": 54440
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.826815492920046e-05,
+      "loss": 0.2258,
+      "step": 54450
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.826681656361252e-05,
+      "loss": 0.1827,
+      "step": 54460
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.826547819802458e-05,
+      "loss": 0.1301,
+      "step": 54470
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.8264139832436636e-05,
+      "loss": 0.0901,
+      "step": 54480
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.826280146684869e-05,
+      "loss": 0.2229,
+      "step": 54490
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.8261463101260747e-05,
+      "loss": 0.2244,
+      "step": 54500
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.82601247356728e-05,
+      "loss": 0.4895,
+      "step": 54510
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.825878637008485e-05,
+      "loss": 0.2587,
+      "step": 54520
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.8257448004496906e-05,
+      "loss": 0.2427,
+      "step": 54530
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.825610963890896e-05,
+      "loss": 0.1342,
+      "step": 54540
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.825477127332102e-05,
+      "loss": 0.321,
+      "step": 54550
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.825343290773308e-05,
+      "loss": 0.1728,
+      "step": 54560
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.8252094542145134e-05,
+      "loss": 0.1989,
+      "step": 54570
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.825075617655719e-05,
+      "loss": 0.0922,
+      "step": 54580
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.8249417810969245e-05,
+      "loss": 0.4039,
+      "step": 54590
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.82480794453813e-05,
+      "loss": 0.1607,
+      "step": 54600
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.8246741079793355e-05,
+      "loss": 0.2128,
+      "step": 54610
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.824540271420541e-05,
+      "loss": 0.1306,
+      "step": 54620
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.8244064348617466e-05,
+      "loss": 0.2306,
+      "step": 54630
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.824272598302952e-05,
+      "loss": 0.3451,
+      "step": 54640
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.8241387617441583e-05,
+      "loss": 0.253,
+      "step": 54650
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.824004925185364e-05,
+      "loss": 0.2067,
+      "step": 54660
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.8238710886265694e-05,
+      "loss": 0.3441,
+      "step": 54670
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.823737252067775e-05,
+      "loss": 0.26,
+      "step": 54680
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.8236034155089805e-05,
+      "loss": 0.3014,
+      "step": 54690
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.823469578950186e-05,
+      "loss": 0.1336,
+      "step": 54700
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.8233357423913915e-05,
+      "loss": 0.397,
+      "step": 54710
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.823201905832597e-05,
+      "loss": 0.0993,
+      "step": 54720
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.8230680692738026e-05,
+      "loss": 0.1688,
+      "step": 54730
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.822934232715008e-05,
+      "loss": 0.1307,
+      "step": 54740
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.8228003961562144e-05,
+      "loss": 0.0982,
+      "step": 54750
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.82266655959742e-05,
+      "loss": 0.16,
+      "step": 54760
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.8225327230386254e-05,
+      "loss": 0.0502,
+      "step": 54770
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.822398886479831e-05,
+      "loss": 0.2774,
+      "step": 54780
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.8222650499210365e-05,
+      "loss": 0.2399,
+      "step": 54790
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.822131213362242e-05,
+      "loss": 0.0938,
+      "step": 54800
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.8219973768034476e-05,
+      "loss": 0.1926,
+      "step": 54810
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.821863540244653e-05,
+      "loss": 0.1862,
+      "step": 54820
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.8217297036858586e-05,
+      "loss": 0.4533,
+      "step": 54830
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.821595867127064e-05,
+      "loss": 0.2483,
+      "step": 54840
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.8214620305682704e-05,
+      "loss": 0.1494,
+      "step": 54850
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.821328194009476e-05,
+      "loss": 0.3647,
+      "step": 54860
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.8211943574506814e-05,
+      "loss": 0.1409,
+      "step": 54870
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.821060520891887e-05,
+      "loss": 0.2331,
+      "step": 54880
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.8209266843330925e-05,
+      "loss": 0.0684,
+      "step": 54890
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.820792847774298e-05,
+      "loss": 0.2513,
+      "step": 54900
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.8206590112155036e-05,
+      "loss": 0.191,
+      "step": 54910
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.820525174656709e-05,
+      "loss": 0.1872,
+      "step": 54920
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.8203913380979146e-05,
+      "loss": 0.2663,
+      "step": 54930
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.82025750153912e-05,
+      "loss": 0.1746,
+      "step": 54940
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.8201236649803264e-05,
+      "loss": 0.3683,
+      "step": 54950
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.819989828421532e-05,
+      "loss": 0.1653,
+      "step": 54960
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.8198559918627375e-05,
+      "loss": 0.2299,
+      "step": 54970
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.819722155303943e-05,
+      "loss": 0.1423,
+      "step": 54980
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.8195883187451485e-05,
+      "loss": 0.1866,
+      "step": 54990
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.819454482186354e-05,
+      "loss": 0.1758,
+      "step": 55000
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.8193206456275596e-05,
+      "loss": 0.1764,
+      "step": 55010
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.819186809068765e-05,
+      "loss": 0.1963,
+      "step": 55020
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.8190529725099707e-05,
+      "loss": 0.1886,
+      "step": 55030
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.818919135951176e-05,
+      "loss": 0.1752,
+      "step": 55040
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.8187852993923824e-05,
+      "loss": 0.1387,
+      "step": 55050
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.818651462833588e-05,
+      "loss": 0.2159,
+      "step": 55060
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.8185176262747935e-05,
+      "loss": 0.0419,
+      "step": 55070
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.818383789715999e-05,
+      "loss": 0.2168,
+      "step": 55080
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.8182499531572045e-05,
+      "loss": 0.1774,
+      "step": 55090
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.81811611659841e-05,
+      "loss": 0.2958,
+      "step": 55100
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.8179822800396156e-05,
+      "loss": 0.2409,
+      "step": 55110
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.817848443480821e-05,
+      "loss": 0.3519,
+      "step": 55120
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.817714606922027e-05,
+      "loss": 0.1802,
+      "step": 55130
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.817580770363232e-05,
+      "loss": 0.2064,
+      "step": 55140
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.8174469338044384e-05,
+      "loss": 0.2836,
+      "step": 55150
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.817313097245644e-05,
+      "loss": 0.2466,
+      "step": 55160
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.8171792606868495e-05,
+      "loss": 0.1039,
+      "step": 55170
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.817045424128055e-05,
+      "loss": 0.0967,
+      "step": 55180
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.8169115875692606e-05,
+      "loss": 0.1818,
+      "step": 55190
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.816777751010466e-05,
+      "loss": 0.0917,
+      "step": 55200
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.8166439144516716e-05,
+      "loss": 0.0822,
+      "step": 55210
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.816510077892877e-05,
+      "loss": 0.1356,
+      "step": 55220
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.816376241334083e-05,
+      "loss": 0.2695,
+      "step": 55230
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.816242404775288e-05,
+      "loss": 0.1336,
+      "step": 55240
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.8161085682164944e-05,
+      "loss": 0.2801,
+      "step": 55250
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.8159747316577e-05,
+      "loss": 0.3947,
+      "step": 55260
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.8158408950989055e-05,
+      "loss": 0.181,
+      "step": 55270
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.815707058540111e-05,
+      "loss": 0.0483,
+      "step": 55280
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.8155732219813166e-05,
+      "loss": 0.2168,
+      "step": 55290
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.815439385422522e-05,
+      "loss": 0.0913,
+      "step": 55300
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.8153055488637276e-05,
+      "loss": 0.0834,
+      "step": 55310
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.815171712304933e-05,
+      "loss": 0.4142,
+      "step": 55320
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.815037875746139e-05,
+      "loss": 0.0703,
+      "step": 55330
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.814904039187344e-05,
+      "loss": 0.1953,
+      "step": 55340
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.8147702026285504e-05,
+      "loss": 0.2113,
+      "step": 55350
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.814636366069756e-05,
+      "loss": 0.0317,
+      "step": 55360
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.8145025295109615e-05,
+      "loss": 0.2226,
+      "step": 55370
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.814368692952167e-05,
+      "loss": 0.2013,
+      "step": 55380
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.8142348563933726e-05,
+      "loss": 0.3455,
+      "step": 55390
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.814101019834578e-05,
+      "loss": 0.1855,
+      "step": 55400
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.8139671832757837e-05,
+      "loss": 0.2367,
+      "step": 55410
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.813833346716989e-05,
+      "loss": 0.1598,
+      "step": 55420
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.813699510158195e-05,
+      "loss": 0.0881,
+      "step": 55430
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.8135656735994e-05,
+      "loss": 0.132,
+      "step": 55440
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.8134318370406065e-05,
+      "loss": 0.1298,
+      "step": 55450
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.813298000481812e-05,
+      "loss": 0.1626,
+      "step": 55460
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.8131641639230175e-05,
+      "loss": 0.1486,
+      "step": 55470
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.813030327364223e-05,
+      "loss": 0.3001,
+      "step": 55480
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.8128964908054286e-05,
+      "loss": 0.0568,
+      "step": 55490
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.812762654246634e-05,
+      "loss": 0.2111,
+      "step": 55500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.81262881768784e-05,
+      "loss": 0.3081,
+      "step": 55510
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.812494981129045e-05,
+      "loss": 0.1679,
+      "step": 55520
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.812361144570251e-05,
+      "loss": 0.2341,
+      "step": 55530
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.812227308011456e-05,
+      "loss": 0.1241,
+      "step": 55540
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.8120934714526625e-05,
+      "loss": 0.2732,
+      "step": 55550
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.811959634893868e-05,
+      "loss": 0.3778,
+      "step": 55560
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.8118257983350735e-05,
+      "loss": 0.5114,
+      "step": 55570
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.811691961776279e-05,
+      "loss": 0.0928,
+      "step": 55580
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.8115581252174846e-05,
+      "loss": 0.169,
+      "step": 55590
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.81142428865869e-05,
+      "loss": 0.1031,
+      "step": 55600
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.811290452099896e-05,
+      "loss": 0.2811,
+      "step": 55610
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.811156615541101e-05,
+      "loss": 0.1347,
+      "step": 55620
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.811022778982307e-05,
+      "loss": 0.2488,
+      "step": 55630
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.810888942423513e-05,
+      "loss": 0.0454,
+      "step": 55640
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.8107551058647185e-05,
+      "loss": 0.242,
+      "step": 55650
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.810621269305924e-05,
+      "loss": 0.176,
+      "step": 55660
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.8104874327471296e-05,
+      "loss": 0.2805,
+      "step": 55670
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.810353596188335e-05,
+      "loss": 0.2528,
+      "step": 55680
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.8102197596295406e-05,
+      "loss": 0.1574,
+      "step": 55690
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.810085923070746e-05,
+      "loss": 0.116,
+      "step": 55700
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.809952086511952e-05,
+      "loss": 0.2343,
+      "step": 55710
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.809818249953157e-05,
+      "loss": 0.1198,
+      "step": 55720
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.809684413394363e-05,
+      "loss": 0.1468,
+      "step": 55730
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.809550576835569e-05,
+      "loss": 0.346,
+      "step": 55740
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.8094167402767745e-05,
+      "loss": 0.1421,
+      "step": 55750
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.80928290371798e-05,
+      "loss": 0.2669,
+      "step": 55760
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.8091490671591856e-05,
+      "loss": 0.045,
+      "step": 55770
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.809015230600391e-05,
+      "loss": 0.1,
+      "step": 55780
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.8088813940415966e-05,
+      "loss": 0.1715,
+      "step": 55790
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.808747557482802e-05,
+      "loss": 0.1928,
+      "step": 55800
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.808613720924008e-05,
+      "loss": 0.2524,
+      "step": 55810
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.808479884365213e-05,
+      "loss": 0.0703,
+      "step": 55820
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.808346047806419e-05,
+      "loss": 0.1927,
+      "step": 55830
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.808212211247625e-05,
+      "loss": 0.0521,
+      "step": 55840
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.8080783746888305e-05,
+      "loss": 0.3583,
+      "step": 55850
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.807944538130036e-05,
+      "loss": 0.085,
+      "step": 55860
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.8078107015712416e-05,
+      "loss": 0.0952,
+      "step": 55870
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.807676865012447e-05,
+      "loss": 0.0701,
+      "step": 55880
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.8075430284536527e-05,
+      "loss": 0.2678,
+      "step": 55890
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.807409191894858e-05,
+      "loss": 0.2716,
+      "step": 55900
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.807275355336064e-05,
+      "loss": 0.1752,
+      "step": 55910
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.807141518777269e-05,
+      "loss": 0.171,
+      "step": 55920
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.807007682218475e-05,
+      "loss": 0.0094,
+      "step": 55930
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.806873845659681e-05,
+      "loss": 0.1774,
+      "step": 55940
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.8067400091008865e-05,
+      "loss": 0.2119,
+      "step": 55950
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.806606172542092e-05,
+      "loss": 0.1236,
+      "step": 55960
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.8064723359832976e-05,
+      "loss": 0.4123,
+      "step": 55970
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.806338499424503e-05,
+      "loss": 0.2073,
+      "step": 55980
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.806204662865709e-05,
+      "loss": 0.2449,
+      "step": 55990
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.806070826306914e-05,
+      "loss": 0.2237,
+      "step": 56000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.80593698974812e-05,
+      "loss": 0.187,
+      "step": 56010
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.805803153189325e-05,
+      "loss": 0.1999,
+      "step": 56020
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.805669316630531e-05,
+      "loss": 0.1438,
+      "step": 56030
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.805535480071737e-05,
+      "loss": 0.1825,
+      "step": 56040
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.8054016435129426e-05,
+      "loss": 0.4537,
+      "step": 56050
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.805267806954148e-05,
+      "loss": 0.1379,
+      "step": 56060
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.8051339703953536e-05,
+      "loss": 0.4305,
+      "step": 56070
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.805000133836559e-05,
+      "loss": 0.2456,
+      "step": 56080
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.804866297277765e-05,
+      "loss": 0.2045,
+      "step": 56090
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.80473246071897e-05,
+      "loss": 0.2485,
+      "step": 56100
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.804598624160176e-05,
+      "loss": 0.0695,
+      "step": 56110
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.804464787601381e-05,
+      "loss": 0.159,
+      "step": 56120
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.804330951042587e-05,
+      "loss": 0.4219,
+      "step": 56130
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.804197114483793e-05,
+      "loss": 0.4112,
+      "step": 56140
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.8040632779249986e-05,
+      "loss": 0.1701,
+      "step": 56150
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.803929441366204e-05,
+      "loss": 0.1282,
+      "step": 56160
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.8037956048074096e-05,
+      "loss": 0.0726,
+      "step": 56170
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.803661768248615e-05,
+      "loss": 0.2816,
+      "step": 56180
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.803527931689821e-05,
+      "loss": 0.3285,
+      "step": 56190
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.803394095131026e-05,
+      "loss": 0.2201,
+      "step": 56200
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.803260258572232e-05,
+      "loss": 0.2774,
+      "step": 56210
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.803126422013437e-05,
+      "loss": 0.1986,
+      "step": 56220
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.802992585454643e-05,
+      "loss": 0.2409,
+      "step": 56230
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.802858748895849e-05,
+      "loss": 0.1698,
+      "step": 56240
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.8027249123370546e-05,
+      "loss": 0.1306,
+      "step": 56250
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.80259107577826e-05,
+      "loss": 0.1613,
+      "step": 56260
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.8024572392194657e-05,
+      "loss": 0.0488,
+      "step": 56270
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.802323402660671e-05,
+      "loss": 0.3006,
+      "step": 56280
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.802189566101877e-05,
+      "loss": 0.2531,
+      "step": 56290
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.802055729543082e-05,
+      "loss": 0.2468,
+      "step": 56300
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.801921892984288e-05,
+      "loss": 0.1959,
+      "step": 56310
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.801788056425493e-05,
+      "loss": 0.2128,
+      "step": 56320
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.801654219866699e-05,
+      "loss": 0.3739,
+      "step": 56330
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.801520383307905e-05,
+      "loss": 0.135,
+      "step": 56340
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.8013865467491106e-05,
+      "loss": 0.1744,
+      "step": 56350
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.801252710190316e-05,
+      "loss": 0.1866,
+      "step": 56360
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.801118873631522e-05,
+      "loss": 0.3217,
+      "step": 56370
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.800985037072727e-05,
+      "loss": 0.3338,
+      "step": 56380
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.800851200513933e-05,
+      "loss": 0.3566,
+      "step": 56390
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.800717363955138e-05,
+      "loss": 0.2817,
+      "step": 56400
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.800583527396344e-05,
+      "loss": 0.2123,
+      "step": 56410
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.800449690837549e-05,
+      "loss": 0.1303,
+      "step": 56420
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.800315854278755e-05,
+      "loss": 0.1451,
+      "step": 56430
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.800182017719961e-05,
+      "loss": 0.2685,
+      "step": 56440
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.8000481811611666e-05,
+      "loss": 0.1833,
+      "step": 56450
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.799914344602372e-05,
+      "loss": 0.1062,
+      "step": 56460
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.799780508043578e-05,
+      "loss": 0.1,
+      "step": 56470
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.799646671484783e-05,
+      "loss": 0.2713,
+      "step": 56480
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.799512834925989e-05,
+      "loss": 0.185,
+      "step": 56490
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.799378998367194e-05,
+      "loss": 0.1494,
+      "step": 56500
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.7992451618084e-05,
+      "loss": 0.308,
+      "step": 56510
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.7991113252496054e-05,
+      "loss": 0.1408,
+      "step": 56520
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.798977488690811e-05,
+      "loss": 0.1904,
+      "step": 56530
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.798843652132017e-05,
+      "loss": 0.3307,
+      "step": 56540
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.7987098155732226e-05,
+      "loss": 0.1567,
+      "step": 56550
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.798575979014428e-05,
+      "loss": 0.1123,
+      "step": 56560
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.798442142455634e-05,
+      "loss": 0.1612,
+      "step": 56570
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.798308305896839e-05,
+      "loss": 0.4808,
+      "step": 56580
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.798174469338045e-05,
+      "loss": 0.1609,
+      "step": 56590
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.79804063277925e-05,
+      "loss": 0.2984,
+      "step": 56600
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.797906796220456e-05,
+      "loss": 0.1696,
+      "step": 56610
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.7977729596616614e-05,
+      "loss": 0.1045,
+      "step": 56620
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.797639123102867e-05,
+      "loss": 0.3706,
+      "step": 56630
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.797505286544073e-05,
+      "loss": 0.2965,
+      "step": 56640
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.7973714499852786e-05,
+      "loss": 0.1733,
+      "step": 56650
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.797237613426484e-05,
+      "loss": 0.1719,
+      "step": 56660
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.79710377686769e-05,
+      "loss": 0.1766,
+      "step": 56670
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.796969940308895e-05,
+      "loss": 0.2321,
+      "step": 56680
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.796836103750101e-05,
+      "loss": 0.1818,
+      "step": 56690
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.796702267191306e-05,
+      "loss": 0.0639,
+      "step": 56700
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.796568430632512e-05,
+      "loss": 0.1766,
+      "step": 56710
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.7964345940737174e-05,
+      "loss": 0.2013,
+      "step": 56720
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.796300757514923e-05,
+      "loss": 0.1621,
+      "step": 56730
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.796166920956129e-05,
+      "loss": 0.1689,
+      "step": 56740
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.796033084397335e-05,
+      "loss": 0.3047,
+      "step": 56750
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.79589924783854e-05,
+      "loss": 0.1696,
+      "step": 56760
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.795765411279746e-05,
+      "loss": 0.2111,
+      "step": 56770
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.795631574720951e-05,
+      "loss": 0.0338,
+      "step": 56780
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.795497738162157e-05,
+      "loss": 0.2028,
+      "step": 56790
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.795363901603362e-05,
+      "loss": 0.2564,
+      "step": 56800
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.795230065044568e-05,
+      "loss": 0.238,
+      "step": 56810
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.7950962284857734e-05,
+      "loss": 0.1432,
+      "step": 56820
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.794962391926979e-05,
+      "loss": 0.261,
+      "step": 56830
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.794828555368185e-05,
+      "loss": 0.1992,
+      "step": 56840
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.794694718809391e-05,
+      "loss": 0.1935,
+      "step": 56850
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.794560882250596e-05,
+      "loss": 0.4198,
+      "step": 56860
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.794427045691802e-05,
+      "loss": 0.1024,
+      "step": 56870
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.7942932091330066e-05,
+      "loss": 0.3487,
+      "step": 56880
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.794159372574212e-05,
+      "loss": 0.3125,
+      "step": 56890
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.794025536015418e-05,
+      "loss": 0.2662,
+      "step": 56900
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.793891699456623e-05,
+      "loss": 0.3549,
+      "step": 56910
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.7937578628978294e-05,
+      "loss": 0.0916,
+      "step": 56920
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.793624026339035e-05,
+      "loss": 0.2408,
+      "step": 56930
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.7934901897802405e-05,
+      "loss": 0.4397,
+      "step": 56940
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.793356353221446e-05,
+      "loss": 0.3361,
+      "step": 56950
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.7932225166626515e-05,
+      "loss": 0.3653,
+      "step": 56960
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.793088680103857e-05,
+      "loss": 0.1922,
+      "step": 56970
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.7929548435450626e-05,
+      "loss": 0.1929,
+      "step": 56980
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.792821006986268e-05,
+      "loss": 0.1287,
+      "step": 56990
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.792687170427474e-05,
+      "loss": 0.1785,
+      "step": 57000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.792553333868679e-05,
+      "loss": 0.1085,
+      "step": 57010
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.7924194973098854e-05,
+      "loss": 0.1584,
+      "step": 57020
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.792285660751091e-05,
+      "loss": 0.0461,
+      "step": 57030
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.7921518241922965e-05,
+      "loss": 0.1751,
+      "step": 57040
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.792017987633502e-05,
+      "loss": 0.0744,
+      "step": 57050
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.7918841510747076e-05,
+      "loss": 0.1368,
+      "step": 57060
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.791750314515913e-05,
+      "loss": 0.0263,
+      "step": 57070
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.7916164779571186e-05,
+      "loss": 0.1078,
+      "step": 57080
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.791482641398324e-05,
+      "loss": 0.2837,
+      "step": 57090
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.79134880483953e-05,
+      "loss": 0.1954,
+      "step": 57100
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.791214968280735e-05,
+      "loss": 0.1837,
+      "step": 57110
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.7910811317219414e-05,
+      "loss": 0.1541,
+      "step": 57120
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.790947295163147e-05,
+      "loss": 0.0906,
+      "step": 57130
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.7908134586043525e-05,
+      "loss": 0.264,
+      "step": 57140
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.790679622045558e-05,
+      "loss": 0.4572,
+      "step": 57150
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.7905457854867636e-05,
+      "loss": 0.2344,
+      "step": 57160
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.790411948927969e-05,
+      "loss": 0.2764,
+      "step": 57170
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.7902781123691746e-05,
+      "loss": 0.1674,
+      "step": 57180
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.79014427581038e-05,
+      "loss": 0.2668,
+      "step": 57190
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.790010439251586e-05,
+      "loss": 0.2328,
+      "step": 57200
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.789876602692792e-05,
+      "loss": 0.1513,
+      "step": 57210
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.7897427661339975e-05,
+      "loss": 0.1962,
+      "step": 57220
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.789608929575203e-05,
+      "loss": 0.2342,
+      "step": 57230
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.7894750930164085e-05,
+      "loss": 0.14,
+      "step": 57240
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.789341256457614e-05,
+      "loss": 0.1502,
+      "step": 57250
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.7892074198988196e-05,
+      "loss": 0.3552,
+      "step": 57260
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.789073583340025e-05,
+      "loss": 0.1488,
+      "step": 57270
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.788939746781231e-05,
+      "loss": 0.369,
+      "step": 57280
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.788805910222436e-05,
+      "loss": 0.2095,
+      "step": 57290
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.788672073663642e-05,
+      "loss": 0.2617,
+      "step": 57300
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.788538237104848e-05,
+      "loss": 0.2228,
+      "step": 57310
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.7884044005460535e-05,
+      "loss": 0.1586,
+      "step": 57320
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.788270563987259e-05,
+      "loss": 0.0678,
+      "step": 57330
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.7881367274284645e-05,
+      "loss": 0.128,
+      "step": 57340
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.78800289086967e-05,
+      "loss": 0.1405,
+      "step": 57350
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.7878690543108756e-05,
+      "loss": 0.2487,
+      "step": 57360
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.787735217752081e-05,
+      "loss": 0.0989,
+      "step": 57370
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.787601381193287e-05,
+      "loss": 0.1921,
+      "step": 57380
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.787467544634492e-05,
+      "loss": 0.3379,
+      "step": 57390
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.787333708075698e-05,
+      "loss": 0.129,
+      "step": 57400
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.787199871516904e-05,
+      "loss": 0.1318,
+      "step": 57410
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.7870660349581095e-05,
+      "loss": 0.0817,
+      "step": 57420
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.786932198399315e-05,
+      "loss": 0.1858,
+      "step": 57430
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.7867983618405206e-05,
+      "loss": 0.1864,
+      "step": 57440
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.786664525281726e-05,
+      "loss": 0.2651,
+      "step": 57450
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.7865306887229316e-05,
+      "loss": 0.1368,
+      "step": 57460
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.786396852164137e-05,
+      "loss": 0.1637,
+      "step": 57470
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.786263015605343e-05,
+      "loss": 0.1977,
+      "step": 57480
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.786129179046548e-05,
+      "loss": 0.1846,
+      "step": 57490
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.785995342487754e-05,
+      "loss": 0.3364,
+      "step": 57500
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.78586150592896e-05,
+      "loss": 0.2373,
+      "step": 57510
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.7857276693701655e-05,
+      "loss": 0.153,
+      "step": 57520
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.785593832811371e-05,
+      "loss": 0.0849,
+      "step": 57530
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.7854599962525766e-05,
+      "loss": 0.1902,
+      "step": 57540
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.785326159693782e-05,
+      "loss": 0.1242,
+      "step": 57550
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.7851923231349876e-05,
+      "loss": 0.256,
+      "step": 57560
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.785058486576193e-05,
+      "loss": 0.0648,
+      "step": 57570
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.784924650017399e-05,
+      "loss": 0.1639,
+      "step": 57580
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.784790813458604e-05,
+      "loss": 0.0552,
+      "step": 57590
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.78465697689981e-05,
+      "loss": 0.1335,
+      "step": 57600
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.784523140341016e-05,
+      "loss": 0.3262,
+      "step": 57610
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.7843893037822215e-05,
+      "loss": 0.1477,
+      "step": 57620
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.784255467223427e-05,
+      "loss": 0.2988,
+      "step": 57630
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.7841216306646326e-05,
+      "loss": 0.1455,
+      "step": 57640
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.783987794105838e-05,
+      "loss": 0.2417,
+      "step": 57650
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.7838539575470437e-05,
+      "loss": 0.1636,
+      "step": 57660
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.783720120988249e-05,
+      "loss": 0.3044,
+      "step": 57670
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.783586284429455e-05,
+      "loss": 0.2686,
+      "step": 57680
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.78345244787066e-05,
+      "loss": 0.3347,
+      "step": 57690
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.783318611311866e-05,
+      "loss": 0.3357,
+      "step": 57700
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.783184774753072e-05,
+      "loss": 0.1894,
+      "step": 57710
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.7830509381942775e-05,
+      "loss": 0.178,
+      "step": 57720
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.782917101635483e-05,
+      "loss": 0.1747,
+      "step": 57730
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.7827832650766886e-05,
+      "loss": 0.2189,
+      "step": 57740
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.782649428517894e-05,
+      "loss": 0.216,
+      "step": 57750
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.7825155919591e-05,
+      "loss": 0.178,
+      "step": 57760
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.782381755400305e-05,
+      "loss": 0.0968,
+      "step": 57770
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.782247918841511e-05,
+      "loss": 0.2054,
+      "step": 57780
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.782114082282716e-05,
+      "loss": 0.1479,
+      "step": 57790
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.781980245723922e-05,
+      "loss": 0.2708,
+      "step": 57800
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.781846409165128e-05,
+      "loss": 0.1063,
+      "step": 57810
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.7817125726063336e-05,
+      "loss": 0.1182,
+      "step": 57820
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.781578736047539e-05,
+      "loss": 0.1796,
+      "step": 57830
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.7814448994887446e-05,
+      "loss": 0.1386,
+      "step": 57840
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.78131106292995e-05,
+      "loss": 0.1042,
+      "step": 57850
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.781177226371156e-05,
+      "loss": 0.35,
+      "step": 57860
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.781043389812361e-05,
+      "loss": 0.1684,
+      "step": 57870
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.780909553253567e-05,
+      "loss": 0.2105,
+      "step": 57880
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.780775716694772e-05,
+      "loss": 0.1476,
+      "step": 57890
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.780641880135978e-05,
+      "loss": 0.2408,
+      "step": 57900
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.780508043577184e-05,
+      "loss": 0.3043,
+      "step": 57910
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.7803742070183896e-05,
+      "loss": 0.1337,
+      "step": 57920
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.780240370459595e-05,
+      "loss": 0.1511,
+      "step": 57930
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.7801065339008006e-05,
+      "loss": 0.1665,
+      "step": 57940
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.779972697342006e-05,
+      "loss": 0.136,
+      "step": 57950
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.779838860783212e-05,
+      "loss": 0.0952,
+      "step": 57960
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.779705024224417e-05,
+      "loss": 0.1083,
+      "step": 57970
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.779571187665623e-05,
+      "loss": 0.1011,
+      "step": 57980
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.779437351106828e-05,
+      "loss": 0.1684,
+      "step": 57990
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.779303514548034e-05,
+      "loss": 0.3432,
+      "step": 58000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.77916967798924e-05,
+      "loss": 0.1191,
+      "step": 58010
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.7790358414304456e-05,
+      "loss": 0.3114,
+      "step": 58020
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.778902004871651e-05,
+      "loss": 0.0847,
+      "step": 58030
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.7787681683128567e-05,
+      "loss": 0.2703,
+      "step": 58040
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.778634331754062e-05,
+      "loss": 0.2012,
+      "step": 58050
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.778500495195268e-05,
+      "loss": 0.2593,
+      "step": 58060
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.778366658636473e-05,
+      "loss": 0.1291,
+      "step": 58070
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.778232822077679e-05,
+      "loss": 0.3449,
+      "step": 58080
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.778098985518884e-05,
+      "loss": 0.2668,
+      "step": 58090
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.77796514896009e-05,
+      "loss": 0.0888,
+      "step": 58100
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.777831312401296e-05,
+      "loss": 0.1487,
+      "step": 58110
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": {
+        "accuracy": 0.9430292080698585
+      },
+      "eval_f1": {
+        "f1": 0.9430188942480495
+      },
+      "eval_loss": 0.22824625670909882,
+      "eval_precision": {
+        "precision": 0.9444609819488103
+      },
+      "eval_recall": {
+        "recall": 0.9428880066548226
+      },
+      "eval_runtime": 431.1779,
+      "eval_samples_per_second": 38.511,
+      "eval_steps_per_second": 19.257,
+      "step": 58117
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.7776974758425016e-05,
+      "loss": 0.0933,
+      "step": 58120
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.777563639283707e-05,
+      "loss": 0.2743,
+      "step": 58130
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.777429802724913e-05,
+      "loss": 0.2768,
+      "step": 58140
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.777295966166118e-05,
+      "loss": 0.1894,
+      "step": 58150
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.777162129607324e-05,
+      "loss": 0.0072,
+      "step": 58160
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.777028293048529e-05,
+      "loss": 0.1712,
+      "step": 58170
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.776894456489735e-05,
+      "loss": 0.1297,
+      "step": 58180
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.77676061993094e-05,
+      "loss": 0.0379,
+      "step": 58190
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.776626783372146e-05,
+      "loss": 0.1495,
+      "step": 58200
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.776492946813352e-05,
+      "loss": 0.2384,
+      "step": 58210
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.7763591102545576e-05,
+      "loss": 0.2333,
+      "step": 58220
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.776225273695763e-05,
+      "loss": 0.2506,
+      "step": 58230
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.776091437136969e-05,
+      "loss": 0.1343,
+      "step": 58240
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.775957600578174e-05,
+      "loss": 0.2154,
+      "step": 58250
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.77582376401938e-05,
+      "loss": 0.1593,
+      "step": 58260
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.775689927460585e-05,
+      "loss": 0.0933,
+      "step": 58270
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.775556090901791e-05,
+      "loss": 0.1318,
+      "step": 58280
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.7754222543429963e-05,
+      "loss": 0.4339,
+      "step": 58290
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.775288417784202e-05,
+      "loss": 0.1486,
+      "step": 58300
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.775154581225408e-05,
+      "loss": 0.1784,
+      "step": 58310
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.7750207446666136e-05,
+      "loss": 0.1917,
+      "step": 58320
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.774886908107819e-05,
+      "loss": 0.1314,
+      "step": 58330
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.774753071549025e-05,
+      "loss": 0.097,
+      "step": 58340
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.77461923499023e-05,
+      "loss": 0.2749,
+      "step": 58350
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.774485398431436e-05,
+      "loss": 0.1443,
+      "step": 58360
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.774351561872641e-05,
+      "loss": 0.2607,
+      "step": 58370
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.774217725313847e-05,
+      "loss": 0.3687,
+      "step": 58380
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.7740838887550524e-05,
+      "loss": 0.3187,
+      "step": 58390
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.7739500521962586e-05,
+      "loss": 0.0451,
+      "step": 58400
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.773816215637464e-05,
+      "loss": 0.3083,
+      "step": 58410
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.7736823790786696e-05,
+      "loss": 0.0652,
+      "step": 58420
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.773548542519875e-05,
+      "loss": 0.1223,
+      "step": 58430
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.773414705961081e-05,
+      "loss": 0.0977,
+      "step": 58440
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.773280869402286e-05,
+      "loss": 0.1005,
+      "step": 58450
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.773147032843492e-05,
+      "loss": 0.3525,
+      "step": 58460
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.773013196284697e-05,
+      "loss": 0.3032,
+      "step": 58470
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.772879359725903e-05,
+      "loss": 0.1912,
+      "step": 58480
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.7727455231671084e-05,
+      "loss": 0.3634,
+      "step": 58490
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.7726116866083146e-05,
+      "loss": 0.0532,
+      "step": 58500
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.77247785004952e-05,
+      "loss": 0.2376,
+      "step": 58510
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.7723440134907257e-05,
+      "loss": 0.3605,
+      "step": 58520
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.772210176931931e-05,
+      "loss": 0.2181,
+      "step": 58530
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.772076340373137e-05,
+      "loss": 0.2173,
+      "step": 58540
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.771942503814342e-05,
+      "loss": 0.2738,
+      "step": 58550
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.771808667255548e-05,
+      "loss": 0.1834,
+      "step": 58560
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.771674830696753e-05,
+      "loss": 0.0435,
+      "step": 58570
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.771540994137959e-05,
+      "loss": 0.33,
+      "step": 58580
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.7714071575791644e-05,
+      "loss": 0.1373,
+      "step": 58590
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.7712733210203706e-05,
+      "loss": 0.184,
+      "step": 58600
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.771139484461576e-05,
+      "loss": 0.225,
+      "step": 58610
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.771005647902782e-05,
+      "loss": 0.2341,
+      "step": 58620
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.770871811343987e-05,
+      "loss": 0.1742,
+      "step": 58630
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.770737974785193e-05,
+      "loss": 0.1891,
+      "step": 58640
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.770604138226398e-05,
+      "loss": 0.1545,
+      "step": 58650
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.770470301667604e-05,
+      "loss": 0.1465,
+      "step": 58660
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.7703364651088093e-05,
+      "loss": 0.1724,
+      "step": 58670
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.770202628550015e-05,
+      "loss": 0.1143,
+      "step": 58680
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.7700687919912204e-05,
+      "loss": 0.0864,
+      "step": 58690
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.7699349554324266e-05,
+      "loss": 0.1765,
+      "step": 58700
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.769801118873632e-05,
+      "loss": 0.1229,
+      "step": 58710
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.769667282314838e-05,
+      "loss": 0.2128,
+      "step": 58720
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.769533445756043e-05,
+      "loss": 0.1025,
+      "step": 58730
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.769399609197249e-05,
+      "loss": 0.3068,
+      "step": 58740
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.769265772638454e-05,
+      "loss": 0.1552,
+      "step": 58750
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.76913193607966e-05,
+      "loss": 0.3344,
+      "step": 58760
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.7689980995208654e-05,
+      "loss": 0.0295,
+      "step": 58770
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.768864262962071e-05,
+      "loss": 0.3137,
+      "step": 58780
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.7687304264032764e-05,
+      "loss": 0.0736,
+      "step": 58790
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.7685965898444826e-05,
+      "loss": 0.4902,
+      "step": 58800
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.768462753285688e-05,
+      "loss": 0.3625,
+      "step": 58810
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.768328916726894e-05,
+      "loss": 0.1856,
+      "step": 58820
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.768195080168099e-05,
+      "loss": 0.0723,
+      "step": 58830
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.768061243609305e-05,
+      "loss": 0.3347,
+      "step": 58840
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.76792740705051e-05,
+      "loss": 0.3037,
+      "step": 58850
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.767793570491716e-05,
+      "loss": 0.1877,
+      "step": 58860
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.7676597339329214e-05,
+      "loss": 0.1539,
+      "step": 58870
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.767525897374127e-05,
+      "loss": 0.379,
+      "step": 58880
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.7673920608153324e-05,
+      "loss": 0.1376,
+      "step": 58890
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.7672582242565387e-05,
+      "loss": 0.2263,
+      "step": 58900
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.767124387697744e-05,
+      "loss": 0.1799,
+      "step": 58910
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.76699055113895e-05,
+      "loss": 0.0599,
+      "step": 58920
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.766856714580155e-05,
+      "loss": 0.364,
+      "step": 58930
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.766722878021361e-05,
+      "loss": 0.2816,
+      "step": 58940
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.766589041462566e-05,
+      "loss": 0.2081,
+      "step": 58950
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.766455204903772e-05,
+      "loss": 0.1247,
+      "step": 58960
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.7663213683449774e-05,
+      "loss": 0.3227,
+      "step": 58970
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.766187531786183e-05,
+      "loss": 0.0496,
+      "step": 58980
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.7660536952273885e-05,
+      "loss": 0.1296,
+      "step": 58990
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.765919858668595e-05,
+      "loss": 0.1948,
+      "step": 59000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.7657860221098e-05,
+      "loss": 0.0317,
+      "step": 59010
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.765652185551006e-05,
+      "loss": 0.1803,
+      "step": 59020
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.765518348992211e-05,
+      "loss": 0.2699,
+      "step": 59030
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.765384512433417e-05,
+      "loss": 0.0598,
+      "step": 59040
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.765250675874622e-05,
+      "loss": 0.1492,
+      "step": 59050
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.765116839315828e-05,
+      "loss": 0.4859,
+      "step": 59060
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.7649830027570334e-05,
+      "loss": 0.2624,
+      "step": 59070
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.764849166198239e-05,
+      "loss": 0.095,
+      "step": 59080
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.7647153296394445e-05,
+      "loss": 0.1878,
+      "step": 59090
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.764581493080651e-05,
+      "loss": 0.1218,
+      "step": 59100
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.764447656521856e-05,
+      "loss": 0.2026,
+      "step": 59110
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.764313819963062e-05,
+      "loss": 0.211,
+      "step": 59120
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.764179983404267e-05,
+      "loss": 0.21,
+      "step": 59130
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.764046146845473e-05,
+      "loss": 0.2486,
+      "step": 59140
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.7639123102866784e-05,
+      "loss": 0.0842,
+      "step": 59150
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.763778473727884e-05,
+      "loss": 0.1854,
+      "step": 59160
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.7636446371690894e-05,
+      "loss": 0.1665,
+      "step": 59170
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.763510800610295e-05,
+      "loss": 0.1805,
+      "step": 59180
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.7633769640515005e-05,
+      "loss": 0.1106,
+      "step": 59190
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.763243127492707e-05,
+      "loss": 0.1387,
+      "step": 59200
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.763109290933912e-05,
+      "loss": 0.1636,
+      "step": 59210
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.762975454375118e-05,
+      "loss": 0.0825,
+      "step": 59220
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.762841617816323e-05,
+      "loss": 0.0389,
+      "step": 59230
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.762707781257529e-05,
+      "loss": 0.1627,
+      "step": 59240
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.762573944698734e-05,
+      "loss": 0.2165,
+      "step": 59250
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.762440108139939e-05,
+      "loss": 0.1731,
+      "step": 59260
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.762306271581145e-05,
+      "loss": 0.2533,
+      "step": 59270
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.762172435022351e-05,
+      "loss": 0.2505,
+      "step": 59280
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.7620385984635565e-05,
+      "loss": 0.1011,
+      "step": 59290
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.3029,
+      "step": 59300
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.7617709253459676e-05,
+      "loss": 0.1765,
+      "step": 59310
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.761637088787173e-05,
+      "loss": 0.301,
+      "step": 59320
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.7615032522283786e-05,
+      "loss": 0.165,
+      "step": 59330
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.761369415669584e-05,
+      "loss": 0.0619,
+      "step": 59340
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.76123557911079e-05,
+      "loss": 0.205,
+      "step": 59350
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.761101742551995e-05,
+      "loss": 0.0808,
+      "step": 59360
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.760967905993201e-05,
+      "loss": 0.2193,
+      "step": 59370
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.760834069434407e-05,
+      "loss": 0.122,
+      "step": 59380
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.7607002328756125e-05,
+      "loss": 0.1502,
+      "step": 59390
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.760566396316818e-05,
+      "loss": 0.1924,
+      "step": 59400
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.7604325597580236e-05,
+      "loss": 0.1187,
+      "step": 59410
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.760298723199229e-05,
+      "loss": 0.1011,
+      "step": 59420
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.7601648866404347e-05,
+      "loss": 0.1052,
+      "step": 59430
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.76003105008164e-05,
+      "loss": 0.0752,
+      "step": 59440
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.759897213522846e-05,
+      "loss": 0.3499,
+      "step": 59450
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.759763376964051e-05,
+      "loss": 0.1083,
+      "step": 59460
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.759629540405257e-05,
+      "loss": 0.2671,
+      "step": 59470
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.759495703846463e-05,
+      "loss": 0.175,
+      "step": 59480
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.7593618672876685e-05,
+      "loss": 0.1843,
+      "step": 59490
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.759228030728874e-05,
+      "loss": 0.0733,
+      "step": 59500
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.7590941941700796e-05,
+      "loss": 0.2167,
+      "step": 59510
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.758960357611285e-05,
+      "loss": 0.2304,
+      "step": 59520
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.758826521052491e-05,
+      "loss": 0.2052,
+      "step": 59530
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.758692684493696e-05,
+      "loss": 0.0271,
+      "step": 59540
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.758558847934902e-05,
+      "loss": 0.0378,
+      "step": 59550
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.758425011376107e-05,
+      "loss": 0.2994,
+      "step": 59560
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.758291174817313e-05,
+      "loss": 0.1119,
+      "step": 59570
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.758157338258519e-05,
+      "loss": 0.1693,
+      "step": 59580
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.7580235016997245e-05,
+      "loss": 0.2213,
+      "step": 59590
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.75788966514093e-05,
+      "loss": 0.1842,
+      "step": 59600
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.7577558285821356e-05,
+      "loss": 0.0897,
+      "step": 59610
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.757621992023341e-05,
+      "loss": 0.3072,
+      "step": 59620
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.757488155464547e-05,
+      "loss": 0.2901,
+      "step": 59630
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.757354318905752e-05,
+      "loss": 0.2243,
+      "step": 59640
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.757220482346958e-05,
+      "loss": 0.1003,
+      "step": 59650
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.757086645788163e-05,
+      "loss": 0.2216,
+      "step": 59660
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.756952809229369e-05,
+      "loss": 0.1888,
+      "step": 59670
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.756818972670575e-05,
+      "loss": 0.1081,
+      "step": 59680
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.7566851361117806e-05,
+      "loss": 0.1738,
+      "step": 59690
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.756551299552986e-05,
+      "loss": 0.1685,
+      "step": 59700
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.7564174629941916e-05,
+      "loss": 0.2631,
+      "step": 59710
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.756283626435397e-05,
+      "loss": 0.2008,
+      "step": 59720
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.756149789876603e-05,
+      "loss": 0.3036,
+      "step": 59730
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.756015953317808e-05,
+      "loss": 0.1012,
+      "step": 59740
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.755882116759014e-05,
+      "loss": 0.0631,
+      "step": 59750
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.755748280200219e-05,
+      "loss": 0.4144,
+      "step": 59760
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.755614443641425e-05,
+      "loss": 0.1208,
+      "step": 59770
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.755480607082631e-05,
+      "loss": 0.2542,
+      "step": 59780
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.7553467705238366e-05,
+      "loss": 0.0716,
+      "step": 59790
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.755212933965042e-05,
+      "loss": 0.3352,
+      "step": 59800
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.7550790974062476e-05,
+      "loss": 0.2177,
+      "step": 59810
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.754945260847453e-05,
+      "loss": 0.1265,
+      "step": 59820
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.754811424288659e-05,
+      "loss": 0.1283,
+      "step": 59830
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.754677587729864e-05,
+      "loss": 0.2471,
+      "step": 59840
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.75454375117107e-05,
+      "loss": 0.1236,
+      "step": 59850
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.754409914612275e-05,
+      "loss": 0.2795,
+      "step": 59860
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.754276078053481e-05,
+      "loss": 0.1681,
+      "step": 59870
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.754142241494687e-05,
+      "loss": 0.292,
+      "step": 59880
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.7540084049358926e-05,
+      "loss": 0.0437,
+      "step": 59890
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.753874568377098e-05,
+      "loss": 0.1947,
+      "step": 59900
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.753740731818304e-05,
+      "loss": 0.2258,
+      "step": 59910
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.753606895259509e-05,
+      "loss": 0.2764,
+      "step": 59920
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.753473058700715e-05,
+      "loss": 0.2367,
+      "step": 59930
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.75333922214192e-05,
+      "loss": 0.0857,
+      "step": 59940
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.753205385583126e-05,
+      "loss": 0.1726,
+      "step": 59950
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.753071549024331e-05,
+      "loss": 0.0826,
+      "step": 59960
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.7529377124655375e-05,
+      "loss": 0.1517,
+      "step": 59970
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.752803875906743e-05,
+      "loss": 0.2279,
+      "step": 59980
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.7526700393479486e-05,
+      "loss": 0.1745,
+      "step": 59990
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.752536202789154e-05,
+      "loss": 0.178,
+      "step": 60000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.75240236623036e-05,
+      "loss": 0.1202,
+      "step": 60010
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.752268529671565e-05,
+      "loss": 0.0807,
+      "step": 60020
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.752134693112771e-05,
+      "loss": 0.0739,
+      "step": 60030
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.752000856553976e-05,
+      "loss": 0.0532,
+      "step": 60040
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.751867019995182e-05,
+      "loss": 0.0987,
+      "step": 60050
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.7517331834363873e-05,
+      "loss": 0.0809,
+      "step": 60060
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.7515993468775936e-05,
+      "loss": 0.3535,
+      "step": 60070
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.751465510318799e-05,
+      "loss": 0.2739,
+      "step": 60080
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.7513316737600046e-05,
+      "loss": 0.2079,
+      "step": 60090
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.75119783720121e-05,
+      "loss": 0.3523,
+      "step": 60100
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.751064000642416e-05,
+      "loss": 0.2113,
+      "step": 60110
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.750930164083621e-05,
+      "loss": 0.1104,
+      "step": 60120
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.750796327524827e-05,
+      "loss": 0.0412,
+      "step": 60130
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.750662490966032e-05,
+      "loss": 0.1343,
+      "step": 60140
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.750528654407238e-05,
+      "loss": 0.2172,
+      "step": 60150
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.7503948178484434e-05,
+      "loss": 0.2611,
+      "step": 60160
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.7502609812896496e-05,
+      "loss": 0.3015,
+      "step": 60170
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.750127144730855e-05,
+      "loss": 0.2746,
+      "step": 60180
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.7499933081720606e-05,
+      "loss": 0.1153,
+      "step": 60190
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.749859471613266e-05,
+      "loss": 0.0254,
+      "step": 60200
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.749725635054472e-05,
+      "loss": 0.2297,
+      "step": 60210
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.749591798495677e-05,
+      "loss": 0.0498,
+      "step": 60220
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.749457961936883e-05,
+      "loss": 0.1095,
+      "step": 60230
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.749324125378088e-05,
+      "loss": 0.1741,
+      "step": 60240
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.749190288819294e-05,
+      "loss": 0.1365,
+      "step": 60250
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.7490564522604994e-05,
+      "loss": 0.1682,
+      "step": 60260
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.7489226157017056e-05,
+      "loss": 0.396,
+      "step": 60270
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.748788779142911e-05,
+      "loss": 0.4058,
+      "step": 60280
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.7486549425841167e-05,
+      "loss": 0.2047,
+      "step": 60290
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.748521106025322e-05,
+      "loss": 0.2338,
+      "step": 60300
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.748387269466528e-05,
+      "loss": 0.1895,
+      "step": 60310
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.748253432907733e-05,
+      "loss": 0.1922,
+      "step": 60320
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.748119596348939e-05,
+      "loss": 0.0995,
+      "step": 60330
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.747985759790144e-05,
+      "loss": 0.3175,
+      "step": 60340
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.74785192323135e-05,
+      "loss": 0.2265,
+      "step": 60350
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.7477180866725554e-05,
+      "loss": 0.0629,
+      "step": 60360
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.7475842501137616e-05,
+      "loss": 0.2669,
+      "step": 60370
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.747450413554967e-05,
+      "loss": 0.1304,
+      "step": 60380
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.747316576996173e-05,
+      "loss": 0.3105,
+      "step": 60390
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.747182740437378e-05,
+      "loss": 0.0844,
+      "step": 60400
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.747048903878584e-05,
+      "loss": 0.1986,
+      "step": 60410
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.746915067319789e-05,
+      "loss": 0.202,
+      "step": 60420
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.746781230760995e-05,
+      "loss": 0.139,
+      "step": 60430
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.7466473942022003e-05,
+      "loss": 0.1364,
+      "step": 60440
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.746513557643406e-05,
+      "loss": 0.06,
+      "step": 60450
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.7463797210846114e-05,
+      "loss": 0.4412,
+      "step": 60460
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.7462458845258176e-05,
+      "loss": 0.2928,
+      "step": 60470
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.746112047967023e-05,
+      "loss": 0.0432,
+      "step": 60480
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.745978211408229e-05,
+      "loss": 0.2121,
+      "step": 60490
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.745844374849434e-05,
+      "loss": 0.2287,
+      "step": 60500
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.74571053829064e-05,
+      "loss": 0.1641,
+      "step": 60510
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.745576701731845e-05,
+      "loss": 0.2721,
+      "step": 60520
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.745442865173051e-05,
+      "loss": 0.1361,
+      "step": 60530
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.7453090286142564e-05,
+      "loss": 0.0705,
+      "step": 60540
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.745175192055462e-05,
+      "loss": 0.1711,
+      "step": 60550
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.7450413554966674e-05,
+      "loss": 0.2024,
+      "step": 60560
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.7449075189378736e-05,
+      "loss": 0.3552,
+      "step": 60570
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.744773682379079e-05,
+      "loss": 0.2488,
+      "step": 60580
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.744639845820285e-05,
+      "loss": 0.2618,
+      "step": 60590
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.74450600926149e-05,
+      "loss": 0.3827,
+      "step": 60600
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.744372172702696e-05,
+      "loss": 0.1781,
+      "step": 60610
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.744238336143901e-05,
+      "loss": 0.1354,
+      "step": 60620
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.744104499585107e-05,
+      "loss": 0.0569,
+      "step": 60630
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.7439706630263124e-05,
+      "loss": 0.1267,
+      "step": 60640
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.743836826467518e-05,
+      "loss": 0.0789,
+      "step": 60650
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.7437029899087234e-05,
+      "loss": 0.2657,
+      "step": 60660
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.7435691533499296e-05,
+      "loss": 0.1242,
+      "step": 60670
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.743435316791135e-05,
+      "loss": 0.1773,
+      "step": 60680
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.743301480232341e-05,
+      "loss": 0.0811,
+      "step": 60690
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.743167643673546e-05,
+      "loss": 0.0683,
+      "step": 60700
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.743033807114752e-05,
+      "loss": 0.1126,
+      "step": 60710
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.742899970555957e-05,
+      "loss": 0.2233,
+      "step": 60720
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.742766133997163e-05,
+      "loss": 0.3883,
+      "step": 60730
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.7426322974383684e-05,
+      "loss": 0.3629,
+      "step": 60740
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.742498460879574e-05,
+      "loss": 0.0707,
+      "step": 60750
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.7423646243207795e-05,
+      "loss": 0.254,
+      "step": 60760
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.742230787761986e-05,
+      "loss": 0.1393,
+      "step": 60770
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.742096951203191e-05,
+      "loss": 0.2683,
+      "step": 60780
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.741963114644397e-05,
+      "loss": 0.1081,
+      "step": 60790
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.741829278085602e-05,
+      "loss": 0.1314,
+      "step": 60800
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.741695441526808e-05,
+      "loss": 0.2224,
+      "step": 60810
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.741561604968013e-05,
+      "loss": 0.2798,
+      "step": 60820
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.741427768409219e-05,
+      "loss": 0.1378,
+      "step": 60830
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.7412939318504244e-05,
+      "loss": 0.2095,
+      "step": 60840
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.74116009529163e-05,
+      "loss": 0.2823,
+      "step": 60850
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.7410262587328355e-05,
+      "loss": 0.1601,
+      "step": 60860
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.740892422174042e-05,
+      "loss": 0.1512,
+      "step": 60870
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.740758585615247e-05,
+      "loss": 0.3379,
+      "step": 60880
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.740624749056453e-05,
+      "loss": 0.1655,
+      "step": 60890
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.740490912497658e-05,
+      "loss": 0.2058,
+      "step": 60900
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.740357075938864e-05,
+      "loss": 0.1519,
+      "step": 60910
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.7402232393800693e-05,
+      "loss": 0.1848,
+      "step": 60920
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.740089402821275e-05,
+      "loss": 0.0355,
+      "step": 60930
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.7399555662624804e-05,
+      "loss": 0.2816,
+      "step": 60940
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.739821729703686e-05,
+      "loss": 0.355,
+      "step": 60950
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.7396878931448915e-05,
+      "loss": 0.268,
+      "step": 60960
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.739554056586098e-05,
+      "loss": 0.2579,
+      "step": 60970
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.739420220027303e-05,
+      "loss": 0.123,
+      "step": 60980
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.739286383468509e-05,
+      "loss": 0.0386,
+      "step": 60990
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.739152546909714e-05,
+      "loss": 0.0897,
+      "step": 61000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.73901871035092e-05,
+      "loss": 0.2312,
+      "step": 61010
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.7388848737921254e-05,
+      "loss": 0.0555,
+      "step": 61020
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.738751037233331e-05,
+      "loss": 0.1061,
+      "step": 61030
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.7386172006745364e-05,
+      "loss": 0.185,
+      "step": 61040
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.738483364115742e-05,
+      "loss": 0.2725,
+      "step": 61050
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.7383495275569475e-05,
+      "loss": 0.1025,
+      "step": 61060
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.738215690998154e-05,
+      "loss": 0.3046,
+      "step": 61070
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.738081854439359e-05,
+      "loss": 0.1199,
+      "step": 61080
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.737948017880565e-05,
+      "loss": 0.2168,
+      "step": 61090
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.73781418132177e-05,
+      "loss": 0.1712,
+      "step": 61100
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.737680344762976e-05,
+      "loss": 0.0282,
+      "step": 61110
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.7375465082041814e-05,
+      "loss": 0.1215,
+      "step": 61120
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.737412671645387e-05,
+      "loss": 0.146,
+      "step": 61130
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.7372788350865924e-05,
+      "loss": 0.3339,
+      "step": 61140
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.737144998527798e-05,
+      "loss": 0.1882,
+      "step": 61150
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.7370111619690035e-05,
+      "loss": 0.1085,
+      "step": 61160
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.73687732541021e-05,
+      "loss": 0.3074,
+      "step": 61170
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.736743488851415e-05,
+      "loss": 0.2179,
+      "step": 61180
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.736609652292621e-05,
+      "loss": 0.2202,
+      "step": 61190
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.736475815733826e-05,
+      "loss": 0.1426,
+      "step": 61200
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.736341979175032e-05,
+      "loss": 0.0764,
+      "step": 61210
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.7362081426162374e-05,
+      "loss": 0.2414,
+      "step": 61220
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.736074306057443e-05,
+      "loss": 0.2081,
+      "step": 61230
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.7359404694986485e-05,
+      "loss": 0.1353,
+      "step": 61240
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.735806632939854e-05,
+      "loss": 0.014,
+      "step": 61250
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.73567279638106e-05,
+      "loss": 0.2948,
+      "step": 61260
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.735538959822266e-05,
+      "loss": 0.2751,
+      "step": 61270
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.735405123263471e-05,
+      "loss": 0.2267,
+      "step": 61280
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.735271286704677e-05,
+      "loss": 0.051,
+      "step": 61290
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.7351374501458823e-05,
+      "loss": 0.2313,
+      "step": 61300
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.735003613587088e-05,
+      "loss": 0.0519,
+      "step": 61310
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.7348697770282934e-05,
+      "loss": 0.1181,
+      "step": 61320
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.734735940469499e-05,
+      "loss": 0.3338,
+      "step": 61330
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.7346021039107045e-05,
+      "loss": 0.1575,
+      "step": 61340
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.73446826735191e-05,
+      "loss": 0.0074,
+      "step": 61350
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.734334430793116e-05,
+      "loss": 0.2389,
+      "step": 61360
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.734200594234322e-05,
+      "loss": 0.21,
+      "step": 61370
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.734066757675527e-05,
+      "loss": 0.1351,
+      "step": 61380
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.733932921116733e-05,
+      "loss": 0.1611,
+      "step": 61390
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.7337990845579384e-05,
+      "loss": 0.0972,
+      "step": 61400
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.733665247999144e-05,
+      "loss": 0.1589,
+      "step": 61410
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.7335314114403494e-05,
+      "loss": 0.2572,
+      "step": 61420
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.733397574881555e-05,
+      "loss": 0.2089,
+      "step": 61430
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.7332637383227605e-05,
+      "loss": 0.1537,
+      "step": 61440
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.733129901763966e-05,
+      "loss": 0.3031,
+      "step": 61450
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.732996065205172e-05,
+      "loss": 0.2107,
+      "step": 61460
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.732862228646378e-05,
+      "loss": 0.185,
+      "step": 61470
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.732728392087583e-05,
+      "loss": 0.3555,
+      "step": 61480
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.732594555528789e-05,
+      "loss": 0.068,
+      "step": 61490
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.7324607189699944e-05,
+      "loss": 0.1388,
+      "step": 61500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.7323268824112e-05,
+      "loss": 0.0761,
+      "step": 61510
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.7321930458524054e-05,
+      "loss": 0.1599,
+      "step": 61520
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.732059209293611e-05,
+      "loss": 0.1233,
+      "step": 61530
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.7319253727348165e-05,
+      "loss": 0.0754,
+      "step": 61540
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.731791536176022e-05,
+      "loss": 0.0956,
+      "step": 61550
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.731657699617228e-05,
+      "loss": 0.1897,
+      "step": 61560
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.731523863058434e-05,
+      "loss": 0.0375,
+      "step": 61570
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.731390026499639e-05,
+      "loss": 0.1289,
+      "step": 61580
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.731256189940845e-05,
+      "loss": 0.1484,
+      "step": 61590
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.7311223533820504e-05,
+      "loss": 0.3203,
+      "step": 61600
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.730988516823255e-05,
+      "loss": 0.0882,
+      "step": 61610
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.730854680264461e-05,
+      "loss": 0.1842,
+      "step": 61620
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.730720843705666e-05,
+      "loss": 0.0808,
+      "step": 61630
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.7305870071468725e-05,
+      "loss": 0.2236,
+      "step": 61640
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.730453170588078e-05,
+      "loss": 0.072,
+      "step": 61650
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.7303193340292836e-05,
+      "loss": 0.2016,
+      "step": 61660
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.730185497470489e-05,
+      "loss": 0.3173,
+      "step": 61670
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.7300516609116947e-05,
+      "loss": 0.2098,
+      "step": 61680
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.7299178243529e-05,
+      "loss": 0.1371,
+      "step": 61690
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.729783987794106e-05,
+      "loss": 0.1662,
+      "step": 61700
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.729650151235311e-05,
+      "loss": 0.12,
+      "step": 61710
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.729516314676517e-05,
+      "loss": 0.0703,
+      "step": 61720
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.729382478117722e-05,
+      "loss": 0.2665,
+      "step": 61730
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.7292486415589285e-05,
+      "loss": 0.1379,
+      "step": 61740
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.729114805000134e-05,
+      "loss": 0.1409,
+      "step": 61750
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.7289809684413396e-05,
+      "loss": 0.0999,
+      "step": 61760
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.728847131882545e-05,
+      "loss": 0.2195,
+      "step": 61770
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.728713295323751e-05,
+      "loss": 0.2224,
+      "step": 61780
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.728579458764956e-05,
+      "loss": 0.1991,
+      "step": 61790
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.728445622206162e-05,
+      "loss": 0.198,
+      "step": 61800
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.728311785647367e-05,
+      "loss": 0.2006,
+      "step": 61810
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.728177949088573e-05,
+      "loss": 0.1405,
+      "step": 61820
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.7280441125297783e-05,
+      "loss": 0.109,
+      "step": 61830
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.7279102759709846e-05,
+      "loss": 0.163,
+      "step": 61840
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.72777643941219e-05,
+      "loss": 0.0066,
+      "step": 61850
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.7276426028533956e-05,
+      "loss": 0.1212,
+      "step": 61860
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.727508766294601e-05,
+      "loss": 0.2102,
+      "step": 61870
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.727374929735807e-05,
+      "loss": 0.267,
+      "step": 61880
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.727241093177012e-05,
+      "loss": 0.2705,
+      "step": 61890
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.727107256618218e-05,
+      "loss": 0.264,
+      "step": 61900
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.726973420059423e-05,
+      "loss": 0.0997,
+      "step": 61910
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.726839583500629e-05,
+      "loss": 0.2364,
+      "step": 61920
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.7267057469418344e-05,
+      "loss": 0.2827,
+      "step": 61930
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.7265719103830406e-05,
+      "loss": 0.154,
+      "step": 61940
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.726438073824246e-05,
+      "loss": 0.1908,
+      "step": 61950
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.7263042372654516e-05,
+      "loss": 0.1921,
+      "step": 61960
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.726170400706657e-05,
+      "loss": 0.2151,
+      "step": 61970
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.726036564147863e-05,
+      "loss": 0.143,
+      "step": 61980
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.725902727589068e-05,
+      "loss": 0.2695,
+      "step": 61990
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.725768891030274e-05,
+      "loss": 0.1029,
+      "step": 62000
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.725635054471479e-05,
+      "loss": 0.2432,
+      "step": 62010
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.725501217912685e-05,
+      "loss": 0.1504,
+      "step": 62020
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.7253673813538904e-05,
+      "loss": 0.1177,
+      "step": 62030
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.7252335447950966e-05,
+      "loss": 0.2198,
+      "step": 62040
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.725099708236302e-05,
+      "loss": 0.179,
+      "step": 62050
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.7249658716775077e-05,
+      "loss": 0.3286,
+      "step": 62060
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.724832035118713e-05,
+      "loss": 0.1807,
+      "step": 62070
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.724698198559919e-05,
+      "loss": 0.0962,
+      "step": 62080
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.724564362001124e-05,
+      "loss": 0.1768,
+      "step": 62090
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.72443052544233e-05,
+      "loss": 0.1966,
+      "step": 62100
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.724296688883535e-05,
+      "loss": 0.0946,
+      "step": 62110
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.724162852324741e-05,
+      "loss": 0.0881,
+      "step": 62120
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.7240290157659464e-05,
+      "loss": 0.2431,
+      "step": 62130
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.7238951792071526e-05,
+      "loss": 0.1886,
+      "step": 62140
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.723761342648358e-05,
+      "loss": 0.209,
+      "step": 62150
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.723627506089564e-05,
+      "loss": 0.1038,
+      "step": 62160
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.723493669530769e-05,
+      "loss": 0.1763,
+      "step": 62170
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.723359832971975e-05,
+      "loss": 0.1928,
+      "step": 62180
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.72322599641318e-05,
+      "loss": 0.2008,
+      "step": 62190
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.723092159854386e-05,
+      "loss": 0.3038,
+      "step": 62200
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.722958323295591e-05,
+      "loss": 0.2901,
+      "step": 62210
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.722824486736797e-05,
+      "loss": 0.0826,
+      "step": 62220
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.7226906501780024e-05,
+      "loss": 0.0876,
+      "step": 62230
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.7225568136192086e-05,
+      "loss": 0.3141,
+      "step": 62240
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.722422977060414e-05,
+      "loss": 0.2047,
+      "step": 62250
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.72228914050162e-05,
+      "loss": 0.1089,
+      "step": 62260
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.722155303942825e-05,
+      "loss": 0.2432,
+      "step": 62270
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.722021467384031e-05,
+      "loss": 0.1397,
+      "step": 62280
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.721887630825236e-05,
+      "loss": 0.0939,
+      "step": 62290
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.721753794266442e-05,
+      "loss": 0.4613,
+      "step": 62300
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.7216199577076474e-05,
+      "loss": 0.1228,
+      "step": 62310
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.721486121148853e-05,
+      "loss": 0.2661,
+      "step": 62320
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.7213522845900584e-05,
+      "loss": 0.0837,
+      "step": 62330
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.7212184480312646e-05,
+      "loss": 0.3193,
+      "step": 62340
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.72108461147247e-05,
+      "loss": 0.29,
+      "step": 62350
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.720950774913676e-05,
+      "loss": 0.1559,
+      "step": 62360
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.720816938354881e-05,
+      "loss": 0.1652,
+      "step": 62370
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.720683101796087e-05,
+      "loss": 0.0826,
+      "step": 62380
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.720549265237292e-05,
+      "loss": 0.2627,
+      "step": 62390
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.720415428678498e-05,
+      "loss": 0.1689,
+      "step": 62400
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.7202815921197034e-05,
+      "loss": 0.1855,
+      "step": 62410
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.720147755560909e-05,
+      "loss": 0.333,
+      "step": 62420
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.7200139190021144e-05,
+      "loss": 0.0639,
+      "step": 62430
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.7198800824433206e-05,
+      "loss": 0.1034,
+      "step": 62440
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.719746245884526e-05,
+      "loss": 0.1298,
+      "step": 62450
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.719612409325732e-05,
+      "loss": 0.1501,
+      "step": 62460
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.719478572766937e-05,
+      "loss": 0.2021,
+      "step": 62470
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.719344736208143e-05,
+      "loss": 0.162,
+      "step": 62480
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.719210899649348e-05,
+      "loss": 0.1183,
+      "step": 62490
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.719077063090554e-05,
+      "loss": 0.1353,
+      "step": 62500
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7189432265317594e-05,
+      "loss": 0.101,
+      "step": 62510
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.718809389972965e-05,
+      "loss": 0.1119,
+      "step": 62520
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7186755534141704e-05,
+      "loss": 0.1237,
+      "step": 62530
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7185417168553767e-05,
+      "loss": 0.2193,
+      "step": 62540
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.718407880296582e-05,
+      "loss": 0.2155,
+      "step": 62550
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.718274043737788e-05,
+      "loss": 0.1581,
+      "step": 62560
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.718140207178993e-05,
+      "loss": 0.2173,
+      "step": 62570
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.718006370620199e-05,
+      "loss": 0.1256,
+      "step": 62580
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.717872534061404e-05,
+      "loss": 0.3014,
+      "step": 62590
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.71773869750261e-05,
+      "loss": 0.2777,
+      "step": 62600
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.7176048609438154e-05,
+      "loss": 0.1349,
+      "step": 62610
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.717471024385021e-05,
+      "loss": 0.2123,
+      "step": 62620
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.7173371878262265e-05,
+      "loss": 0.1151,
+      "step": 62630
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.717203351267433e-05,
+      "loss": 0.1204,
+      "step": 62640
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.717069514708638e-05,
+      "loss": 0.0257,
+      "step": 62650
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.716935678149844e-05,
+      "loss": 0.1666,
+      "step": 62660
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.716801841591049e-05,
+      "loss": 0.316,
+      "step": 62670
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.716668005032255e-05,
+      "loss": 0.1429,
+      "step": 62680
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.7165341684734603e-05,
+      "loss": 0.2397,
+      "step": 62690
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.716400331914666e-05,
+      "loss": 0.2229,
+      "step": 62700
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.7162664953558714e-05,
+      "loss": 0.2414,
+      "step": 62710
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.716132658797077e-05,
+      "loss": 0.3661,
+      "step": 62720
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.715998822238283e-05,
+      "loss": 0.0462,
+      "step": 62730
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.715864985679489e-05,
+      "loss": 0.0486,
+      "step": 62740
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.715731149120694e-05,
+      "loss": 0.0922,
+      "step": 62750
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.7155973125619e-05,
+      "loss": 0.1549,
+      "step": 62760
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.715463476003105e-05,
+      "loss": 0.0835,
+      "step": 62770
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.715329639444311e-05,
+      "loss": 0.1715,
+      "step": 62780
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.7151958028855164e-05,
+      "loss": 0.292,
+      "step": 62790
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.715061966326722e-05,
+      "loss": 0.1742,
+      "step": 62800
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.7149281297679274e-05,
+      "loss": 0.1521,
+      "step": 62810
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.714794293209133e-05,
+      "loss": 0.1731,
+      "step": 62820
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.714660456650339e-05,
+      "loss": 0.1045,
+      "step": 62830
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.714526620091545e-05,
+      "loss": 0.1989,
+      "step": 62840
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.71439278353275e-05,
+      "loss": 0.2584,
+      "step": 62850
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.714258946973956e-05,
+      "loss": 0.2449,
+      "step": 62860
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.714125110415161e-05,
+      "loss": 0.2472,
+      "step": 62870
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.713991273856367e-05,
+      "loss": 0.2039,
+      "step": 62880
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.7138574372975724e-05,
+      "loss": 0.124,
+      "step": 62890
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.713723600738778e-05,
+      "loss": 0.2023,
+      "step": 62900
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.7135897641799834e-05,
+      "loss": 0.074,
+      "step": 62910
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.713455927621189e-05,
+      "loss": 0.5518,
+      "step": 62920
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.713322091062395e-05,
+      "loss": 0.1354,
+      "step": 62930
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.713188254503601e-05,
+      "loss": 0.2424,
+      "step": 62940
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.713054417944806e-05,
+      "loss": 0.0469,
+      "step": 62950
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.712920581386012e-05,
+      "loss": 0.0846,
+      "step": 62960
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.712786744827217e-05,
+      "loss": 0.0704,
+      "step": 62970
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.712652908268423e-05,
+      "loss": 0.0891,
+      "step": 62980
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.7125190717096284e-05,
+      "loss": 0.2002,
+      "step": 62990
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.712385235150834e-05,
+      "loss": 0.2258,
+      "step": 63000
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.7122513985920395e-05,
+      "loss": 0.3508,
+      "step": 63010
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.712117562033245e-05,
+      "loss": 0.1012,
+      "step": 63020
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.711983725474451e-05,
+      "loss": 0.1559,
+      "step": 63030
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.711849888915657e-05,
+      "loss": 0.113,
+      "step": 63040
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.711716052356862e-05,
+      "loss": 0.27,
+      "step": 63050
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.711582215798068e-05,
+      "loss": 0.1767,
+      "step": 63060
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.711448379239273e-05,
+      "loss": 0.1964,
+      "step": 63070
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.711314542680479e-05,
+      "loss": 0.1471,
+      "step": 63080
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.7111807061216844e-05,
+      "loss": 0.4882,
+      "step": 63090
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.71104686956289e-05,
+      "loss": 0.2593,
+      "step": 63100
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.7109130330040955e-05,
+      "loss": 0.2801,
+      "step": 63110
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.710779196445301e-05,
+      "loss": 0.1269,
+      "step": 63120
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.710645359886507e-05,
+      "loss": 0.0872,
+      "step": 63130
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.710511523327713e-05,
+      "loss": 0.1934,
+      "step": 63140
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.710377686768918e-05,
+      "loss": 0.1086,
+      "step": 63150
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.710243850210124e-05,
+      "loss": 0.1382,
+      "step": 63160
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.7101100136513294e-05,
+      "loss": 0.3235,
+      "step": 63170
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.709976177092535e-05,
+      "loss": 0.2956,
+      "step": 63180
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.7098423405337404e-05,
+      "loss": 0.2661,
+      "step": 63190
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.709708503974946e-05,
+      "loss": 0.4018,
+      "step": 63200
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.7095746674161515e-05,
+      "loss": 0.3288,
+      "step": 63210
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.709440830857357e-05,
+      "loss": 0.2497,
+      "step": 63220
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.709306994298563e-05,
+      "loss": 0.1038,
+      "step": 63230
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.709173157739769e-05,
+      "loss": 0.286,
+      "step": 63240
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.709039321180974e-05,
+      "loss": 0.11,
+      "step": 63250
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.70890548462218e-05,
+      "loss": 0.197,
+      "step": 63260
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.7087716480633854e-05,
+      "loss": 0.11,
+      "step": 63270
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.708637811504591e-05,
+      "loss": 0.1081,
+      "step": 63280
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.7085039749457964e-05,
+      "loss": 0.162,
+      "step": 63290
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.708370138387002e-05,
+      "loss": 0.225,
+      "step": 63300
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.7082363018282075e-05,
+      "loss": 0.2393,
+      "step": 63310
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.708102465269413e-05,
+      "loss": 0.4061,
+      "step": 63320
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.707968628710619e-05,
+      "loss": 0.0468,
+      "step": 63330
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.707834792151825e-05,
+      "loss": 0.1456,
+      "step": 63340
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.70770095559303e-05,
+      "loss": 0.2117,
+      "step": 63350
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.707567119034236e-05,
+      "loss": 0.2463,
+      "step": 63360
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.7074332824754414e-05,
+      "loss": 0.287,
+      "step": 63370
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.707299445916647e-05,
+      "loss": 0.094,
+      "step": 63380
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.7071656093578525e-05,
+      "loss": 0.1368,
+      "step": 63390
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.707031772799058e-05,
+      "loss": 0.2006,
+      "step": 63400
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.7068979362402635e-05,
+      "loss": 0.0655,
+      "step": 63410
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.706764099681469e-05,
+      "loss": 0.1929,
+      "step": 63420
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.706630263122675e-05,
+      "loss": 0.0677,
+      "step": 63430
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.706496426563881e-05,
+      "loss": 0.2467,
+      "step": 63440
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.706362590005086e-05,
+      "loss": 0.1612,
+      "step": 63450
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.706228753446292e-05,
+      "loss": 0.1243,
+      "step": 63460
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.7060949168874974e-05,
+      "loss": 0.0954,
+      "step": 63470
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.705961080328703e-05,
+      "loss": 0.2713,
+      "step": 63480
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.7058272437699085e-05,
+      "loss": 0.0477,
+      "step": 63490
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.705693407211114e-05,
+      "loss": 0.1436,
+      "step": 63500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.7055595706523195e-05,
+      "loss": 0.2215,
+      "step": 63510
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.705425734093525e-05,
+      "loss": 0.336,
+      "step": 63520
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.705291897534731e-05,
+      "loss": 0.0926,
+      "step": 63530
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.705158060975937e-05,
+      "loss": 0.231,
+      "step": 63540
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.7050242244171423e-05,
+      "loss": 0.1977,
+      "step": 63550
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.704890387858348e-05,
+      "loss": 0.2322,
+      "step": 63560
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.7047565512995534e-05,
+      "loss": 0.1166,
+      "step": 63570
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.704622714740759e-05,
+      "loss": 0.1879,
+      "step": 63580
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.7044888781819645e-05,
+      "loss": 0.1408,
+      "step": 63590
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.70435504162317e-05,
+      "loss": 0.0425,
+      "step": 63600
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.7042212050643755e-05,
+      "loss": 0.2232,
+      "step": 63610
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.704087368505581e-05,
+      "loss": 0.1812,
+      "step": 63620
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.703953531946787e-05,
+      "loss": 0.1537,
+      "step": 63630
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.703819695387993e-05,
+      "loss": 0.0827,
+      "step": 63640
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.7036858588291984e-05,
+      "loss": 0.2812,
+      "step": 63650
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.703552022270404e-05,
+      "loss": 0.1232,
+      "step": 63660
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.7034181857116094e-05,
+      "loss": 0.0404,
+      "step": 63670
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.703284349152815e-05,
+      "loss": 0.2348,
+      "step": 63680
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.7031505125940205e-05,
+      "loss": 0.2888,
+      "step": 63690
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.703016676035226e-05,
+      "loss": 0.1556,
+      "step": 63700
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.7028828394764316e-05,
+      "loss": 0.3252,
+      "step": 63710
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.702749002917637e-05,
+      "loss": 0.1281,
+      "step": 63720
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.702615166358843e-05,
+      "loss": 0.1646,
+      "step": 63730
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.702481329800049e-05,
+      "loss": 0.1736,
+      "step": 63740
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.7023474932412544e-05,
+      "loss": 0.2937,
+      "step": 63750
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.70221365668246e-05,
+      "loss": 0.1127,
+      "step": 63760
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.7020798201236654e-05,
+      "loss": 0.1622,
+      "step": 63770
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.701945983564871e-05,
+      "loss": 0.2862,
+      "step": 63780
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.7018121470060765e-05,
+      "loss": 0.3393,
+      "step": 63790
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.701678310447282e-05,
+      "loss": 0.0708,
+      "step": 63800
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.7015444738884876e-05,
+      "loss": 0.195,
+      "step": 63810
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.701410637329693e-05,
+      "loss": 0.2521,
+      "step": 63820
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.701276800770899e-05,
+      "loss": 0.1462,
+      "step": 63830
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.701142964212105e-05,
+      "loss": 0.0939,
+      "step": 63840
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.7010091276533104e-05,
+      "loss": 0.2582,
+      "step": 63850
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.700875291094516e-05,
+      "loss": 0.2436,
+      "step": 63860
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.7007414545357215e-05,
+      "loss": 0.1815,
+      "step": 63870
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.700607617976927e-05,
+      "loss": 0.2404,
+      "step": 63880
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.7004737814181325e-05,
+      "loss": 0.1619,
+      "step": 63890
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.700339944859338e-05,
+      "loss": 0.1194,
+      "step": 63900
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.7002061083005436e-05,
+      "loss": 0.3436,
+      "step": 63910
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.700072271741749e-05,
+      "loss": 0.1561,
+      "step": 63920
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.6999384351829553e-05,
+      "loss": 0.192,
+      "step": 63930
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.699804598624161e-05,
+      "loss": 0.3056,
+      "step": 63940
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.6996707620653664e-05,
+      "loss": 0.2567,
+      "step": 63950
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.699536925506572e-05,
+      "loss": 0.0818,
+      "step": 63960
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.6994030889477775e-05,
+      "loss": 0.0516,
+      "step": 63970
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.699269252388982e-05,
+      "loss": 0.1861,
+      "step": 63980
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.699135415830188e-05,
+      "loss": 0.1483,
+      "step": 63990
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.6990015792713934e-05,
+      "loss": 0.1311,
+      "step": 64000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.6988677427125996e-05,
+      "loss": 0.3123,
+      "step": 64010
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.698733906153805e-05,
+      "loss": 0.1223,
+      "step": 64020
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.698600069595011e-05,
+      "loss": 0.263,
+      "step": 64030
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.698466233036216e-05,
+      "loss": 0.3012,
+      "step": 64040
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.698332396477422e-05,
+      "loss": 0.1075,
+      "step": 64050
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.698198559918627e-05,
+      "loss": 0.2018,
+      "step": 64060
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.698064723359833e-05,
+      "loss": 0.2558,
+      "step": 64070
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.6979308868010383e-05,
+      "loss": 0.216,
+      "step": 64080
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.697797050242244e-05,
+      "loss": 0.2124,
+      "step": 64090
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.6976632136834494e-05,
+      "loss": 0.0144,
+      "step": 64100
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.6975293771246556e-05,
+      "loss": 0.4304,
+      "step": 64110
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.697395540565861e-05,
+      "loss": 0.2811,
+      "step": 64120
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.697261704007067e-05,
+      "loss": 0.132,
+      "step": 64130
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.697127867448272e-05,
+      "loss": 0.0901,
+      "step": 64140
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.696994030889478e-05,
+      "loss": 0.1716,
+      "step": 64150
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.696860194330683e-05,
+      "loss": 0.2337,
+      "step": 64160
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.696726357771889e-05,
+      "loss": 0.2204,
+      "step": 64170
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.6965925212130944e-05,
+      "loss": 0.1196,
+      "step": 64180
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.6964586846543e-05,
+      "loss": 0.2167,
+      "step": 64190
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.696324848095506e-05,
+      "loss": 0.1893,
+      "step": 64200
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.6961910115367116e-05,
+      "loss": 0.2807,
+      "step": 64210
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.696057174977917e-05,
+      "loss": 0.1386,
+      "step": 64220
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.695923338419123e-05,
+      "loss": 0.0766,
+      "step": 64230
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.695789501860328e-05,
+      "loss": 0.1105,
+      "step": 64240
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.695655665301534e-05,
+      "loss": 0.1182,
+      "step": 64250
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.695521828742739e-05,
+      "loss": 0.329,
+      "step": 64260
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.695387992183945e-05,
+      "loss": 0.259,
+      "step": 64270
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.6952541556251504e-05,
+      "loss": 0.3074,
+      "step": 64280
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.695120319066356e-05,
+      "loss": 0.2439,
+      "step": 64290
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.694986482507562e-05,
+      "loss": 0.1638,
+      "step": 64300
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.6948526459487677e-05,
+      "loss": 0.1972,
+      "step": 64310
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.694718809389973e-05,
+      "loss": 0.1955,
+      "step": 64320
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.694584972831179e-05,
+      "loss": 0.1415,
+      "step": 64330
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.694451136272384e-05,
+      "loss": 0.0734,
+      "step": 64340
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.69431729971359e-05,
+      "loss": 0.1802,
+      "step": 64350
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.694183463154795e-05,
+      "loss": 0.1151,
+      "step": 64360
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.694049626596001e-05,
+      "loss": 0.0695,
+      "step": 64370
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.6939157900372064e-05,
+      "loss": 0.3048,
+      "step": 64380
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.693781953478412e-05,
+      "loss": 0.1671,
+      "step": 64390
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.693648116919618e-05,
+      "loss": 0.1505,
+      "step": 64400
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.693514280360824e-05,
+      "loss": 0.2518,
+      "step": 64410
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.693380443802029e-05,
+      "loss": 0.1751,
+      "step": 64420
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.693246607243235e-05,
+      "loss": 0.212,
+      "step": 64430
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.69311277068444e-05,
+      "loss": 0.0842,
+      "step": 64440
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.692978934125646e-05,
+      "loss": 0.1391,
+      "step": 64450
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.6928450975668513e-05,
+      "loss": 0.3328,
+      "step": 64460
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.692711261008057e-05,
+      "loss": 0.1972,
+      "step": 64470
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.6925774244492624e-05,
+      "loss": 0.2011,
+      "step": 64480
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.692443587890468e-05,
+      "loss": 0.0876,
+      "step": 64490
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.692309751331674e-05,
+      "loss": 0.2259,
+      "step": 64500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.69217591477288e-05,
+      "loss": 0.0629,
+      "step": 64510
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.692042078214085e-05,
+      "loss": 0.054,
+      "step": 64520
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.691908241655291e-05,
+      "loss": 0.1505,
+      "step": 64530
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.691774405096496e-05,
+      "loss": 0.3416,
+      "step": 64540
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.691640568537702e-05,
+      "loss": 0.2233,
+      "step": 64550
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.6915067319789074e-05,
+      "loss": 0.1095,
+      "step": 64560
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.691372895420113e-05,
+      "loss": 0.1269,
+      "step": 64570
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.6912390588613184e-05,
+      "loss": 0.1596,
+      "step": 64580
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.691105222302524e-05,
+      "loss": 0.1551,
+      "step": 64590
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.69097138574373e-05,
+      "loss": 0.1301,
+      "step": 64600
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.690837549184936e-05,
+      "loss": 0.1551,
+      "step": 64610
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.690703712626141e-05,
+      "loss": 0.1848,
+      "step": 64620
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.690569876067347e-05,
+      "loss": 0.1002,
+      "step": 64630
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.690436039508552e-05,
+      "loss": 0.168,
+      "step": 64640
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.690302202949758e-05,
+      "loss": 0.3484,
+      "step": 64650
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.6901683663909634e-05,
+      "loss": 0.1204,
+      "step": 64660
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.690034529832169e-05,
+      "loss": 0.1499,
+      "step": 64670
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.6899006932733744e-05,
+      "loss": 0.0107,
+      "step": 64680
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.68976685671458e-05,
+      "loss": 0.1706,
+      "step": 64690
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.689633020155786e-05,
+      "loss": 0.1557,
+      "step": 64700
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.689499183596992e-05,
+      "loss": 0.247,
+      "step": 64710
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.689365347038197e-05,
+      "loss": 0.2983,
+      "step": 64720
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.689231510479403e-05,
+      "loss": 0.327,
+      "step": 64730
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.689097673920608e-05,
+      "loss": 0.242,
+      "step": 64740
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.688963837361814e-05,
+      "loss": 0.1971,
+      "step": 64750
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.6888300008030194e-05,
+      "loss": 0.2154,
+      "step": 64760
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.688696164244225e-05,
+      "loss": 0.1099,
+      "step": 64770
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.6885623276854305e-05,
+      "loss": 0.2786,
+      "step": 64780
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.688428491126636e-05,
+      "loss": 0.1392,
+      "step": 64790
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.688294654567842e-05,
+      "loss": 0.2047,
+      "step": 64800
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.688160818009048e-05,
+      "loss": 0.3671,
+      "step": 64810
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.688026981450253e-05,
+      "loss": 0.0685,
+      "step": 64820
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.687893144891459e-05,
+      "loss": 0.1195,
+      "step": 64830
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.687759308332664e-05,
+      "loss": 0.238,
+      "step": 64840
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.68762547177387e-05,
+      "loss": 0.2278,
+      "step": 64850
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.6874916352150754e-05,
+      "loss": 0.2644,
+      "step": 64860
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.687357798656281e-05,
+      "loss": 0.1881,
+      "step": 64870
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.6872239620974865e-05,
+      "loss": 0.2413,
+      "step": 64880
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.687090125538692e-05,
+      "loss": 0.081,
+      "step": 64890
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.686956288979898e-05,
+      "loss": 0.1938,
+      "step": 64900
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.686822452421104e-05,
+      "loss": 0.447,
+      "step": 64910
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.686688615862309e-05,
+      "loss": 0.2149,
+      "step": 64920
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.686554779303515e-05,
+      "loss": 0.0887,
+      "step": 64930
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.6864209427447203e-05,
+      "loss": 0.2064,
+      "step": 64940
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.686287106185926e-05,
+      "loss": 0.2319,
+      "step": 64950
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.6861532696271314e-05,
+      "loss": 0.2803,
+      "step": 64960
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.686019433068337e-05,
+      "loss": 0.2528,
+      "step": 64970
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.6858855965095425e-05,
+      "loss": 0.0666,
+      "step": 64980
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.685751759950748e-05,
+      "loss": 0.0914,
+      "step": 64990
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.685617923391954e-05,
+      "loss": 0.1864,
+      "step": 65000
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.68548408683316e-05,
+      "loss": 0.1189,
+      "step": 65010
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.685350250274365e-05,
+      "loss": 0.2673,
+      "step": 65020
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.685216413715571e-05,
+      "loss": 0.2049,
+      "step": 65030
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.6850825771567764e-05,
+      "loss": 0.1351,
+      "step": 65040
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.684948740597982e-05,
+      "loss": 0.2047,
+      "step": 65050
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.6848149040391874e-05,
+      "loss": 0.1544,
+      "step": 65060
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.684681067480393e-05,
+      "loss": 0.242,
+      "step": 65070
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.6845472309215985e-05,
+      "loss": 0.3464,
+      "step": 65080
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.684413394362804e-05,
+      "loss": 0.2611,
+      "step": 65090
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.68427955780401e-05,
+      "loss": 0.234,
+      "step": 65100
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.684145721245216e-05,
+      "loss": 0.1152,
+      "step": 65110
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.684011884686421e-05,
+      "loss": 0.161,
+      "step": 65120
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.683878048127627e-05,
+      "loss": 0.0592,
+      "step": 65130
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.6837442115688324e-05,
+      "loss": 0.1144,
+      "step": 65140
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.683610375010038e-05,
+      "loss": 0.1543,
+      "step": 65150
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.6834765384512434e-05,
+      "loss": 0.1369,
+      "step": 65160
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.683342701892449e-05,
+      "loss": 0.1746,
+      "step": 65170
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.6832088653336545e-05,
+      "loss": 0.3344,
+      "step": 65180
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.68307502877486e-05,
+      "loss": 0.0621,
+      "step": 65190
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.682941192216066e-05,
+      "loss": 0.242,
+      "step": 65200
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.682807355657272e-05,
+      "loss": 0.1854,
+      "step": 65210
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.682673519098477e-05,
+      "loss": 0.0241,
+      "step": 65220
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.682539682539683e-05,
+      "loss": 0.1836,
+      "step": 65230
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.6824058459808884e-05,
+      "loss": 0.2627,
+      "step": 65240
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.682272009422094e-05,
+      "loss": 0.2045,
+      "step": 65250
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.6821381728632995e-05,
+      "loss": 0.2712,
+      "step": 65260
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.682004336304505e-05,
+      "loss": 0.0607,
+      "step": 65270
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.6818704997457105e-05,
+      "loss": 0.0371,
+      "step": 65280
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.681736663186916e-05,
+      "loss": 0.1476,
+      "step": 65290
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.681602826628122e-05,
+      "loss": 0.0983,
+      "step": 65300
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.681468990069328e-05,
+      "loss": 0.372,
+      "step": 65310
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.6813351535105333e-05,
+      "loss": 0.1943,
+      "step": 65320
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.681201316951739e-05,
+      "loss": 0.2432,
+      "step": 65330
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.6810674803929444e-05,
+      "loss": 0.3785,
+      "step": 65340
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.68093364383415e-05,
+      "loss": 0.2276,
+      "step": 65350
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.6807998072753555e-05,
+      "loss": 0.3643,
+      "step": 65360
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.680665970716561e-05,
+      "loss": 0.274,
+      "step": 65370
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.6805321341577665e-05,
+      "loss": 0.2046,
+      "step": 65380
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.680398297598972e-05,
+      "loss": 0.0759,
+      "step": 65390
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.680264461040178e-05,
+      "loss": 0.0995,
+      "step": 65400
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.680130624481384e-05,
+      "loss": 0.2767,
+      "step": 65410
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.6799967879225894e-05,
+      "loss": 0.1704,
+      "step": 65420
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.679862951363795e-05,
+      "loss": 0.1715,
+      "step": 65430
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.6797291148050004e-05,
+      "loss": 0.2261,
+      "step": 65440
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.679595278246206e-05,
+      "loss": 0.2733,
+      "step": 65450
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.6794614416874115e-05,
+      "loss": 0.2726,
+      "step": 65460
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.679327605128617e-05,
+      "loss": 0.3651,
+      "step": 65470
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.6791937685698226e-05,
+      "loss": 0.0343,
+      "step": 65480
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.679059932011029e-05,
+      "loss": 0.205,
+      "step": 65490
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.678926095452234e-05,
+      "loss": 0.1595,
+      "step": 65500
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.67879225889344e-05,
+      "loss": 0.1571,
+      "step": 65510
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.6786584223346454e-05,
+      "loss": 0.1345,
+      "step": 65520
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.678524585775851e-05,
+      "loss": 0.2552,
+      "step": 65530
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.6783907492170564e-05,
+      "loss": 0.0889,
+      "step": 65540
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.678256912658262e-05,
+      "loss": 0.4631,
+      "step": 65550
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.6781230760994675e-05,
+      "loss": 0.1152,
+      "step": 65560
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.677989239540673e-05,
+      "loss": 0.115,
+      "step": 65570
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.6778554029818786e-05,
+      "loss": 0.0769,
+      "step": 65580
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.677721566423085e-05,
+      "loss": 0.1662,
+      "step": 65590
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.67758772986429e-05,
+      "loss": 0.0929,
+      "step": 65600
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.677453893305496e-05,
+      "loss": 0.188,
+      "step": 65610
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.6773200567467014e-05,
+      "loss": 0.1763,
+      "step": 65620
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.677186220187907e-05,
+      "loss": 0.199,
+      "step": 65630
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.6770523836291125e-05,
+      "loss": 0.1811,
+      "step": 65640
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.676918547070318e-05,
+      "loss": 0.1306,
+      "step": 65650
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.6767847105115235e-05,
+      "loss": 0.0873,
+      "step": 65660
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.676650873952729e-05,
+      "loss": 0.2489,
+      "step": 65670
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.6765170373939346e-05,
+      "loss": 0.213,
+      "step": 65680
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.676383200835141e-05,
+      "loss": 0.184,
+      "step": 65690
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.676249364276346e-05,
+      "loss": 0.2738,
+      "step": 65700
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.676115527717552e-05,
+      "loss": 0.0879,
+      "step": 65710
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.6759816911587574e-05,
+      "loss": 0.0984,
+      "step": 65720
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.675847854599963e-05,
+      "loss": 0.1426,
+      "step": 65730
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.6757140180411685e-05,
+      "loss": 0.2703,
+      "step": 65740
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.675580181482374e-05,
+      "loss": 0.2173,
+      "step": 65750
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.6754463449235795e-05,
+      "loss": 0.1141,
+      "step": 65760
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.675312508364785e-05,
+      "loss": 0.2385,
+      "step": 65770
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.6751786718059906e-05,
+      "loss": 0.129,
+      "step": 65780
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.675044835247197e-05,
+      "loss": 0.0849,
+      "step": 65790
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.6749109986884024e-05,
+      "loss": 0.0826,
+      "step": 65800
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.674777162129608e-05,
+      "loss": 0.1424,
+      "step": 65810
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.6746433255708134e-05,
+      "loss": 0.1994,
+      "step": 65820
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.674509489012019e-05,
+      "loss": 0.2652,
+      "step": 65830
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.6743756524532245e-05,
+      "loss": 0.2232,
+      "step": 65840
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.67424181589443e-05,
+      "loss": 0.2221,
+      "step": 65850
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.6741079793356356e-05,
+      "loss": 0.1852,
+      "step": 65860
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.673974142776841e-05,
+      "loss": 0.0819,
+      "step": 65870
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.6738403062180466e-05,
+      "loss": 0.1577,
+      "step": 65880
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.673706469659253e-05,
+      "loss": 0.0208,
+      "step": 65890
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.6735726331004584e-05,
+      "loss": 0.1652,
+      "step": 65900
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.673438796541664e-05,
+      "loss": 0.3469,
+      "step": 65910
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.6733049599828694e-05,
+      "loss": 0.0813,
+      "step": 65920
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.673171123424075e-05,
+      "loss": 0.0968,
+      "step": 65930
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.6730372868652805e-05,
+      "loss": 0.1641,
+      "step": 65940
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.672903450306486e-05,
+      "loss": 0.3207,
+      "step": 65950
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.6727696137476916e-05,
+      "loss": 0.3398,
+      "step": 65960
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.672635777188897e-05,
+      "loss": 0.2008,
+      "step": 65970
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.6725019406301026e-05,
+      "loss": 0.187,
+      "step": 65980
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.672368104071309e-05,
+      "loss": 0.1284,
+      "step": 65990
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.6722342675125144e-05,
+      "loss": 0.2762,
+      "step": 66000
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.67210043095372e-05,
+      "loss": 0.3353,
+      "step": 66010
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.6719665943949254e-05,
+      "loss": 0.1621,
+      "step": 66020
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.671832757836131e-05,
+      "loss": 0.1478,
+      "step": 66030
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.6716989212773365e-05,
+      "loss": 0.1341,
+      "step": 66040
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.671565084718542e-05,
+      "loss": 0.1763,
+      "step": 66050
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6714312481597476e-05,
+      "loss": 0.3299,
+      "step": 66060
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.671297411600953e-05,
+      "loss": 0.1071,
+      "step": 66070
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6711635750421587e-05,
+      "loss": 0.0404,
+      "step": 66080
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.671029738483365e-05,
+      "loss": 0.1106,
+      "step": 66090
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6708959019245704e-05,
+      "loss": 0.1733,
+      "step": 66100
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.670762065365776e-05,
+      "loss": 0.2206,
+      "step": 66110
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6706282288069815e-05,
+      "loss": 0.2115,
+      "step": 66120
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.670494392248187e-05,
+      "loss": 0.2237,
+      "step": 66130
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.6703605556893925e-05,
+      "loss": 0.2354,
+      "step": 66140
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.670226719130598e-05,
+      "loss": 0.1142,
+      "step": 66150
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.6700928825718036e-05,
+      "loss": 0.2923,
+      "step": 66160
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.669959046013009e-05,
+      "loss": 0.2017,
+      "step": 66170
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.669825209454215e-05,
+      "loss": 0.1796,
+      "step": 66180
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.669691372895421e-05,
+      "loss": 0.0746,
+      "step": 66190
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.6695575363366264e-05,
+      "loss": 0.1962,
+      "step": 66200
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.669423699777832e-05,
+      "loss": 0.1409,
+      "step": 66210
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.6692898632190375e-05,
+      "loss": 0.045,
+      "step": 66220
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.669156026660243e-05,
+      "loss": 0.0904,
+      "step": 66230
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.6690221901014485e-05,
+      "loss": 0.26,
+      "step": 66240
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.668888353542654e-05,
+      "loss": 0.0923,
+      "step": 66250
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.6687545169838596e-05,
+      "loss": 0.1181,
+      "step": 66260
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.668620680425065e-05,
+      "loss": 0.2383,
+      "step": 66270
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.668486843866271e-05,
+      "loss": 0.12,
+      "step": 66280
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.668353007307477e-05,
+      "loss": 0.0847,
+      "step": 66290
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.6682191707486824e-05,
+      "loss": 0.1622,
+      "step": 66300
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.668085334189888e-05,
+      "loss": 0.0847,
+      "step": 66310
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.6679514976310935e-05,
+      "loss": 0.1374,
+      "step": 66320
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.667817661072299e-05,
+      "loss": 0.1019,
+      "step": 66330
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.667683824513504e-05,
+      "loss": 0.2464,
+      "step": 66340
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.6675499879547094e-05,
+      "loss": 0.0805,
+      "step": 66350
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.667416151395915e-05,
+      "loss": 0.104,
+      "step": 66360
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.667282314837121e-05,
+      "loss": 0.3637,
+      "step": 66370
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.667148478278327e-05,
+      "loss": 0.0825,
+      "step": 66380
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.667014641719532e-05,
+      "loss": 0.165,
+      "step": 66390
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.666880805160738e-05,
+      "loss": 0.1835,
+      "step": 66400
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.666746968601943e-05,
+      "loss": 0.1888,
+      "step": 66410
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.666613132043149e-05,
+      "loss": 0.1405,
+      "step": 66420
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": {
+        "accuracy": 0.9454381210478772
+      },
+      "eval_f1": {
+        "f1": 0.9455191951029847
+      },
+      "eval_loss": 0.2439512014389038,
+      "eval_precision": {
+        "precision": 0.9467893516678145
+      },
+      "eval_recall": {
+        "recall": 0.9453224042508239
+      },
+      "eval_runtime": 430.1891,
+      "eval_samples_per_second": 38.599,
+      "eval_steps_per_second": 19.301,
+      "step": 66420
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.6664792954843544e-05,
+      "loss": 0.1013,
+      "step": 66430
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.66634545892556e-05,
+      "loss": 0.0717,
+      "step": 66440
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.6662116223667654e-05,
+      "loss": 0.1442,
+      "step": 66450
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.666077785807971e-05,
+      "loss": 0.1858,
+      "step": 66460
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.665943949249177e-05,
+      "loss": 0.2704,
+      "step": 66470
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.665810112690383e-05,
+      "loss": 0.2376,
+      "step": 66480
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.665676276131588e-05,
+      "loss": 0.1269,
+      "step": 66490
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.665542439572794e-05,
+      "loss": 0.0894,
+      "step": 66500
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.665408603013999e-05,
+      "loss": 0.2804,
+      "step": 66510
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.665274766455205e-05,
+      "loss": 0.1089,
+      "step": 66520
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.6651409298964104e-05,
+      "loss": 0.1751,
+      "step": 66530
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.665007093337616e-05,
+      "loss": 0.1103,
+      "step": 66540
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.6648732567788214e-05,
+      "loss": 0.0622,
+      "step": 66550
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.664739420220027e-05,
+      "loss": 0.1536,
+      "step": 66560
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.664605583661233e-05,
+      "loss": 0.0779,
+      "step": 66570
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.664471747102439e-05,
+      "loss": 0.1936,
+      "step": 66580
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.664337910543644e-05,
+      "loss": 0.1502,
+      "step": 66590
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.66420407398485e-05,
+      "loss": 0.1486,
+      "step": 66600
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.664070237426055e-05,
+      "loss": 0.1521,
+      "step": 66610
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.663936400867261e-05,
+      "loss": 0.2647,
+      "step": 66620
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.6638025643084664e-05,
+      "loss": 0.2466,
+      "step": 66630
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.663668727749672e-05,
+      "loss": 0.1184,
+      "step": 66640
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.6635348911908775e-05,
+      "loss": 0.3006,
+      "step": 66650
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.663401054632083e-05,
+      "loss": 0.1035,
+      "step": 66660
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.663267218073289e-05,
+      "loss": 0.0481,
+      "step": 66670
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.663133381514495e-05,
+      "loss": 0.1761,
+      "step": 66680
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.6629995449557e-05,
+      "loss": 0.2453,
+      "step": 66690
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.662865708396906e-05,
+      "loss": 0.2565,
+      "step": 66700
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.6627318718381113e-05,
+      "loss": 0.0775,
+      "step": 66710
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.662598035279317e-05,
+      "loss": 0.0653,
+      "step": 66720
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.6624641987205224e-05,
+      "loss": 0.4767,
+      "step": 66730
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.662330362161728e-05,
+      "loss": 0.2688,
+      "step": 66740
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.6621965256029335e-05,
+      "loss": 0.1405,
+      "step": 66750
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.662062689044139e-05,
+      "loss": 0.1233,
+      "step": 66760
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.661928852485345e-05,
+      "loss": 0.1684,
+      "step": 66770
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.661795015926551e-05,
+      "loss": 0.125,
+      "step": 66780
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.661661179367756e-05,
+      "loss": 0.2532,
+      "step": 66790
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.661527342808962e-05,
+      "loss": 0.3709,
+      "step": 66800
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.6613935062501674e-05,
+      "loss": 0.1946,
+      "step": 66810
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.661259669691373e-05,
+      "loss": 0.1472,
+      "step": 66820
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.6611258331325784e-05,
+      "loss": 0.4438,
+      "step": 66830
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.660991996573784e-05,
+      "loss": 0.2656,
+      "step": 66840
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.6608581600149895e-05,
+      "loss": 0.366,
+      "step": 66850
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.660724323456195e-05,
+      "loss": 0.0839,
+      "step": 66860
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.660590486897401e-05,
+      "loss": 0.1641,
+      "step": 66870
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.660456650338607e-05,
+      "loss": 0.1595,
+      "step": 66880
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.660322813779812e-05,
+      "loss": 0.3771,
+      "step": 66890
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.660188977221018e-05,
+      "loss": 0.0944,
+      "step": 66900
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.6600551406622234e-05,
+      "loss": 0.1539,
+      "step": 66910
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.659921304103429e-05,
+      "loss": 0.3343,
+      "step": 66920
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.6597874675446344e-05,
+      "loss": 0.1829,
+      "step": 66930
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.65965363098584e-05,
+      "loss": 0.1832,
+      "step": 66940
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.6595197944270455e-05,
+      "loss": 0.0742,
+      "step": 66950
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.659385957868252e-05,
+      "loss": 0.2326,
+      "step": 66960
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.659252121309457e-05,
+      "loss": 0.2162,
+      "step": 66970
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.659118284750663e-05,
+      "loss": 0.1541,
+      "step": 66980
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.658984448191868e-05,
+      "loss": 0.2201,
+      "step": 66990
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.658850611633074e-05,
+      "loss": 0.0854,
+      "step": 67000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.6587167750742794e-05,
+      "loss": 0.1537,
+      "step": 67010
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.658582938515485e-05,
+      "loss": 0.0788,
+      "step": 67020
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.6584491019566905e-05,
+      "loss": 0.176,
+      "step": 67030
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.658315265397896e-05,
+      "loss": 0.1048,
+      "step": 67040
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.6581814288391015e-05,
+      "loss": 0.1695,
+      "step": 67050
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.658047592280308e-05,
+      "loss": 0.1012,
+      "step": 67060
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.657913755721513e-05,
+      "loss": 0.1899,
+      "step": 67070
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.657779919162719e-05,
+      "loss": 0.2705,
+      "step": 67080
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.657646082603924e-05,
+      "loss": 0.3833,
+      "step": 67090
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.65751224604513e-05,
+      "loss": 0.3221,
+      "step": 67100
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.6573784094863354e-05,
+      "loss": 0.0614,
+      "step": 67110
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.657244572927541e-05,
+      "loss": 0.1941,
+      "step": 67120
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.6571107363687465e-05,
+      "loss": 0.0991,
+      "step": 67130
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.656976899809952e-05,
+      "loss": 0.101,
+      "step": 67140
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.6568430632511575e-05,
+      "loss": 0.1204,
+      "step": 67150
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.656709226692364e-05,
+      "loss": 0.1872,
+      "step": 67160
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.656575390133569e-05,
+      "loss": 0.1037,
+      "step": 67170
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.656441553574775e-05,
+      "loss": 0.2144,
+      "step": 67180
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.6563077170159804e-05,
+      "loss": 0.1993,
+      "step": 67190
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.656173880457186e-05,
+      "loss": 0.1081,
+      "step": 67200
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.6560400438983914e-05,
+      "loss": 0.0997,
+      "step": 67210
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.655906207339597e-05,
+      "loss": 0.0621,
+      "step": 67220
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.6557723707808025e-05,
+      "loss": 0.1911,
+      "step": 67230
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.655638534222008e-05,
+      "loss": 0.2288,
+      "step": 67240
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.6555046976632136e-05,
+      "loss": 0.2233,
+      "step": 67250
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.65537086110442e-05,
+      "loss": 0.1879,
+      "step": 67260
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.655237024545625e-05,
+      "loss": 0.3054,
+      "step": 67270
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.655103187986831e-05,
+      "loss": 0.1229,
+      "step": 67280
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.6549693514280364e-05,
+      "loss": 0.0481,
+      "step": 67290
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.654835514869242e-05,
+      "loss": 0.1173,
+      "step": 67300
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.6547016783104474e-05,
+      "loss": 0.3001,
+      "step": 67310
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.654567841751653e-05,
+      "loss": 0.1654,
+      "step": 67320
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.6544340051928585e-05,
+      "loss": 0.0409,
+      "step": 67330
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.654300168634064e-05,
+      "loss": 0.1545,
+      "step": 67340
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.6541663320752696e-05,
+      "loss": 0.1961,
+      "step": 67350
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.654032495516476e-05,
+      "loss": 0.165,
+      "step": 67360
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.653898658957681e-05,
+      "loss": 0.0995,
+      "step": 67370
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.653764822398887e-05,
+      "loss": 0.0779,
+      "step": 67380
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.6536309858400924e-05,
+      "loss": 0.3389,
+      "step": 67390
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.653497149281298e-05,
+      "loss": 0.1353,
+      "step": 67400
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.6533633127225035e-05,
+      "loss": 0.3118,
+      "step": 67410
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.653229476163709e-05,
+      "loss": 0.092,
+      "step": 67420
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.6530956396049145e-05,
+      "loss": 0.2164,
+      "step": 67430
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.65296180304612e-05,
+      "loss": 0.1533,
+      "step": 67440
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.6528279664873256e-05,
+      "loss": 0.3323,
+      "step": 67450
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.652694129928532e-05,
+      "loss": 0.1501,
+      "step": 67460
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.652560293369737e-05,
+      "loss": 0.1755,
+      "step": 67470
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.652426456810943e-05,
+      "loss": 0.0263,
+      "step": 67480
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.6522926202521484e-05,
+      "loss": 0.2484,
+      "step": 67490
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.652158783693354e-05,
+      "loss": 0.2664,
+      "step": 67500
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.6520249471345595e-05,
+      "loss": 0.1968,
+      "step": 67510
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.651891110575765e-05,
+      "loss": 0.159,
+      "step": 67520
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.6517572740169705e-05,
+      "loss": 0.1733,
+      "step": 67530
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.651623437458176e-05,
+      "loss": 0.2029,
+      "step": 67540
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.6514896008993816e-05,
+      "loss": 0.2409,
+      "step": 67550
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.651355764340588e-05,
+      "loss": 0.1484,
+      "step": 67560
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.6512219277817933e-05,
+      "loss": 0.0786,
+      "step": 67570
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.651088091222999e-05,
+      "loss": 0.0555,
+      "step": 67580
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.6509542546642044e-05,
+      "loss": 0.1426,
+      "step": 67590
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.65082041810541e-05,
+      "loss": 0.3282,
+      "step": 67600
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.6506865815466155e-05,
+      "loss": 0.2374,
+      "step": 67610
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.650552744987821e-05,
+      "loss": 0.0921,
+      "step": 67620
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.6504189084290265e-05,
+      "loss": 0.0282,
+      "step": 67630
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.650285071870232e-05,
+      "loss": 0.1941,
+      "step": 67640
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.6501512353114376e-05,
+      "loss": 0.204,
+      "step": 67650
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.650017398752644e-05,
+      "loss": 0.024,
+      "step": 67660
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.6498835621938494e-05,
+      "loss": 0.1472,
+      "step": 67670
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.649749725635055e-05,
+      "loss": 0.2015,
+      "step": 67680
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.6496158890762604e-05,
+      "loss": 0.2543,
+      "step": 67690
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.649482052517466e-05,
+      "loss": 0.1792,
+      "step": 67700
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.6493482159586715e-05,
+      "loss": 0.0791,
+      "step": 67710
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.649214379399877e-05,
+      "loss": 0.1967,
+      "step": 67720
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.6490805428410826e-05,
+      "loss": 0.195,
+      "step": 67730
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.648946706282288e-05,
+      "loss": 0.1086,
+      "step": 67740
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.6488128697234936e-05,
+      "loss": 0.173,
+      "step": 67750
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.6486790331647e-05,
+      "loss": 0.2307,
+      "step": 67760
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.6485451966059054e-05,
+      "loss": 0.0393,
+      "step": 67770
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.648411360047111e-05,
+      "loss": 0.0113,
+      "step": 67780
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.6482775234883164e-05,
+      "loss": 0.1914,
+      "step": 67790
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.648143686929522e-05,
+      "loss": 0.1509,
+      "step": 67800
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.6480098503707275e-05,
+      "loss": 0.1377,
+      "step": 67810
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.647876013811933e-05,
+      "loss": 0.1841,
+      "step": 67820
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.6477421772531386e-05,
+      "loss": 0.1709,
+      "step": 67830
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.647608340694344e-05,
+      "loss": 0.1329,
+      "step": 67840
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.6474745041355496e-05,
+      "loss": 0.1691,
+      "step": 67850
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.647340667576756e-05,
+      "loss": 0.0947,
+      "step": 67860
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.6472068310179614e-05,
+      "loss": 0.2083,
+      "step": 67870
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.647072994459167e-05,
+      "loss": 0.1785,
+      "step": 67880
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.6469391579003725e-05,
+      "loss": 0.2063,
+      "step": 67890
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.646805321341578e-05,
+      "loss": 0.1979,
+      "step": 67900
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.6466714847827835e-05,
+      "loss": 0.1329,
+      "step": 67910
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.646537648223989e-05,
+      "loss": 0.0584,
+      "step": 67920
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.6464038116651946e-05,
+      "loss": 0.188,
+      "step": 67930
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.6462699751064e-05,
+      "loss": 0.2401,
+      "step": 67940
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.646136138547606e-05,
+      "loss": 0.1208,
+      "step": 67950
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.646002301988812e-05,
+      "loss": 0.329,
+      "step": 67960
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.6458684654300174e-05,
+      "loss": 0.1398,
+      "step": 67970
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.645734628871223e-05,
+      "loss": 0.2187,
+      "step": 67980
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.6456007923124285e-05,
+      "loss": 0.1451,
+      "step": 67990
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.645466955753634e-05,
+      "loss": 0.2096,
+      "step": 68000
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.6453331191948395e-05,
+      "loss": 0.0814,
+      "step": 68010
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.645199282636045e-05,
+      "loss": 0.2539,
+      "step": 68020
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.6450654460772506e-05,
+      "loss": 0.2074,
+      "step": 68030
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.644931609518456e-05,
+      "loss": 0.1095,
+      "step": 68040
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.644797772959662e-05,
+      "loss": 0.2743,
+      "step": 68050
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.644663936400868e-05,
+      "loss": 0.2219,
+      "step": 68060
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.6445300998420734e-05,
+      "loss": 0.1338,
+      "step": 68070
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.644396263283279e-05,
+      "loss": 0.0616,
+      "step": 68080
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.6442624267244845e-05,
+      "loss": 0.26,
+      "step": 68090
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.64412859016569e-05,
+      "loss": 0.075,
+      "step": 68100
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.6439947536068956e-05,
+      "loss": 0.1584,
+      "step": 68110
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.643860917048101e-05,
+      "loss": 0.18,
+      "step": 68120
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.6437270804893066e-05,
+      "loss": 0.1154,
+      "step": 68130
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.643593243930512e-05,
+      "loss": 0.1811,
+      "step": 68140
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.643459407371718e-05,
+      "loss": 0.291,
+      "step": 68150
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.643325570812924e-05,
+      "loss": 0.1262,
+      "step": 68160
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.6431917342541294e-05,
+      "loss": 0.2152,
+      "step": 68170
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.643057897695335e-05,
+      "loss": 0.2397,
+      "step": 68180
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.6429240611365405e-05,
+      "loss": 0.1947,
+      "step": 68190
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.642790224577746e-05,
+      "loss": 0.1982,
+      "step": 68200
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.6426563880189516e-05,
+      "loss": 0.1526,
+      "step": 68210
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.642522551460157e-05,
+      "loss": 0.312,
+      "step": 68220
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.6423887149013626e-05,
+      "loss": 0.2526,
+      "step": 68230
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.642254878342568e-05,
+      "loss": 0.3377,
+      "step": 68240
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.6421210417837744e-05,
+      "loss": 0.3826,
+      "step": 68250
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.64198720522498e-05,
+      "loss": 0.2022,
+      "step": 68260
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.6418533686661855e-05,
+      "loss": 0.2078,
+      "step": 68270
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.641719532107391e-05,
+      "loss": 0.1647,
+      "step": 68280
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.6415856955485965e-05,
+      "loss": 0.1857,
+      "step": 68290
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.641451858989802e-05,
+      "loss": 0.1727,
+      "step": 68300
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.6413180224310076e-05,
+      "loss": 0.2345,
+      "step": 68310
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.641184185872213e-05,
+      "loss": 0.0906,
+      "step": 68320
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.6410503493134187e-05,
+      "loss": 0.2746,
+      "step": 68330
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.640916512754624e-05,
+      "loss": 0.1133,
+      "step": 68340
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.6407826761958304e-05,
+      "loss": 0.3049,
+      "step": 68350
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.640648839637036e-05,
+      "loss": 0.1858,
+      "step": 68360
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.6405150030782415e-05,
+      "loss": 0.0858,
+      "step": 68370
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.640381166519447e-05,
+      "loss": 0.0911,
+      "step": 68380
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.6402473299606525e-05,
+      "loss": 0.2389,
+      "step": 68390
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.640113493401858e-05,
+      "loss": 0.1294,
+      "step": 68400
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.6399796568430636e-05,
+      "loss": 0.0577,
+      "step": 68410
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.639845820284269e-05,
+      "loss": 0.1287,
+      "step": 68420
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.639711983725475e-05,
+      "loss": 0.3952,
+      "step": 68430
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.63957814716668e-05,
+      "loss": 0.1111,
+      "step": 68440
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.6394443106078864e-05,
+      "loss": 0.4654,
+      "step": 68450
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.639310474049092e-05,
+      "loss": 0.1978,
+      "step": 68460
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.6391766374902975e-05,
+      "loss": 0.1287,
+      "step": 68470
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.639042800931503e-05,
+      "loss": 0.0767,
+      "step": 68480
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.6389089643727086e-05,
+      "loss": 0.1097,
+      "step": 68490
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.638775127813914e-05,
+      "loss": 0.3123,
+      "step": 68500
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.6386412912551196e-05,
+      "loss": 0.1631,
+      "step": 68510
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.638507454696325e-05,
+      "loss": 0.2334,
+      "step": 68520
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.638373618137531e-05,
+      "loss": 0.1115,
+      "step": 68530
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.638239781578736e-05,
+      "loss": 0.0933,
+      "step": 68540
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.6381059450199424e-05,
+      "loss": 0.3912,
+      "step": 68550
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.637972108461148e-05,
+      "loss": 0.1146,
+      "step": 68560
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.6378382719023535e-05,
+      "loss": 0.1195,
+      "step": 68570
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.637704435343559e-05,
+      "loss": 0.0969,
+      "step": 68580
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.6375705987847646e-05,
+      "loss": 0.0858,
+      "step": 68590
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.63743676222597e-05,
+      "loss": 0.2408,
+      "step": 68600
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.6373029256671756e-05,
+      "loss": 0.2753,
+      "step": 68610
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.637169089108381e-05,
+      "loss": 0.1267,
+      "step": 68620
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.637035252549587e-05,
+      "loss": 0.2051,
+      "step": 68630
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.636901415990792e-05,
+      "loss": 0.113,
+      "step": 68640
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.6367675794319984e-05,
+      "loss": 0.2055,
+      "step": 68650
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.636633742873204e-05,
+      "loss": 0.2244,
+      "step": 68660
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.6364999063144095e-05,
+      "loss": 0.2101,
+      "step": 68670
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.636366069755615e-05,
+      "loss": 0.0957,
+      "step": 68680
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.6362322331968206e-05,
+      "loss": 0.2657,
+      "step": 68690
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.636098396638026e-05,
+      "loss": 0.1939,
+      "step": 68700
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.635964560079231e-05,
+      "loss": 0.0653,
+      "step": 68710
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.6358307235204365e-05,
+      "loss": 0.1503,
+      "step": 68720
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.635696886961643e-05,
+      "loss": 0.1011,
+      "step": 68730
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.635563050402848e-05,
+      "loss": 0.178,
+      "step": 68740
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.635429213844054e-05,
+      "loss": 0.2351,
+      "step": 68750
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.635295377285259e-05,
+      "loss": 0.0526,
+      "step": 68760
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.635161540726465e-05,
+      "loss": 0.3294,
+      "step": 68770
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.6350277041676704e-05,
+      "loss": 0.1687,
+      "step": 68780
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.634893867608876e-05,
+      "loss": 0.0276,
+      "step": 68790
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.6347600310500815e-05,
+      "loss": 0.1385,
+      "step": 68800
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.634626194491287e-05,
+      "loss": 0.2668,
+      "step": 68810
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.6344923579324925e-05,
+      "loss": 0.1224,
+      "step": 68820
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.634358521373699e-05,
+      "loss": 0.2755,
+      "step": 68830
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.634224684814904e-05,
+      "loss": 0.1767,
+      "step": 68840
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.63409084825611e-05,
+      "loss": 0.1822,
+      "step": 68850
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.633957011697315e-05,
+      "loss": 0.143,
+      "step": 68860
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.633823175138521e-05,
+      "loss": 0.1981,
+      "step": 68870
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.6336893385797264e-05,
+      "loss": 0.2558,
+      "step": 68880
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.633555502020932e-05,
+      "loss": 0.1533,
+      "step": 68890
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.6334216654621375e-05,
+      "loss": 0.3909,
+      "step": 68900
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.633287828903343e-05,
+      "loss": 0.2421,
+      "step": 68910
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.6331539923445485e-05,
+      "loss": 0.1462,
+      "step": 68920
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.633020155785755e-05,
+      "loss": 0.1329,
+      "step": 68930
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.63288631922696e-05,
+      "loss": 0.3006,
+      "step": 68940
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.632752482668166e-05,
+      "loss": 0.144,
+      "step": 68950
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.6326186461093713e-05,
+      "loss": 0.1417,
+      "step": 68960
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.632484809550577e-05,
+      "loss": 0.1369,
+      "step": 68970
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.6323509729917824e-05,
+      "loss": 0.1281,
+      "step": 68980
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.632217136432988e-05,
+      "loss": 0.1699,
+      "step": 68990
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.6320832998741935e-05,
+      "loss": 0.3067,
+      "step": 69000
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.631949463315399e-05,
+      "loss": 0.1467,
+      "step": 69010
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.6318156267566046e-05,
+      "loss": 0.2298,
+      "step": 69020
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.631681790197811e-05,
+      "loss": 0.0182,
+      "step": 69030
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.631547953639016e-05,
+      "loss": 0.2461,
+      "step": 69040
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.631414117080222e-05,
+      "loss": 0.1014,
+      "step": 69050
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.6312802805214274e-05,
+      "loss": 0.2602,
+      "step": 69060
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.631146443962633e-05,
+      "loss": 0.0643,
+      "step": 69070
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.6310126074038384e-05,
+      "loss": 0.1563,
+      "step": 69080
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.630878770845044e-05,
+      "loss": 0.1903,
+      "step": 69090
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.6307449342862495e-05,
+      "loss": 0.2029,
+      "step": 69100
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.630611097727455e-05,
+      "loss": 0.0556,
+      "step": 69110
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.6304772611686606e-05,
+      "loss": 0.0275,
+      "step": 69120
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.630343424609867e-05,
+      "loss": 0.1948,
+      "step": 69130
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.630209588051072e-05,
+      "loss": 0.0759,
+      "step": 69140
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.630075751492278e-05,
+      "loss": 0.21,
+      "step": 69150
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.6299419149334834e-05,
+      "loss": 0.1673,
+      "step": 69160
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.629808078374689e-05,
+      "loss": 0.0909,
+      "step": 69170
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.6296742418158944e-05,
+      "loss": 0.1704,
+      "step": 69180
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.6295404052571e-05,
+      "loss": 0.3113,
+      "step": 69190
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.6294065686983055e-05,
+      "loss": 0.0911,
+      "step": 69200
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.629272732139511e-05,
+      "loss": 0.2191,
+      "step": 69210
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.6291388955807166e-05,
+      "loss": 0.1735,
+      "step": 69220
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.629005059021923e-05,
+      "loss": 0.1513,
+      "step": 69230
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.628871222463128e-05,
+      "loss": 0.1002,
+      "step": 69240
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.628737385904334e-05,
+      "loss": 0.1958,
+      "step": 69250
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.6286035493455394e-05,
+      "loss": 0.2664,
+      "step": 69260
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.628469712786745e-05,
+      "loss": 0.289,
+      "step": 69270
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.6283358762279505e-05,
+      "loss": 0.2392,
+      "step": 69280
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.628202039669156e-05,
+      "loss": 0.0456,
+      "step": 69290
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.6280682031103615e-05,
+      "loss": 0.1097,
+      "step": 69300
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.627934366551567e-05,
+      "loss": 0.1789,
+      "step": 69310
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.6278005299927726e-05,
+      "loss": 0.108,
+      "step": 69320
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.627666693433979e-05,
+      "loss": 0.4006,
+      "step": 69330
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.6275328568751843e-05,
+      "loss": 0.2076,
+      "step": 69340
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.62739902031639e-05,
+      "loss": 0.1902,
+      "step": 69350
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.6272651837575954e-05,
+      "loss": 0.14,
+      "step": 69360
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.627131347198801e-05,
+      "loss": 0.0535,
+      "step": 69370
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.6269975106400065e-05,
+      "loss": 0.2331,
+      "step": 69380
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.626863674081212e-05,
+      "loss": 0.3097,
+      "step": 69390
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.6267298375224175e-05,
+      "loss": 0.0348,
+      "step": 69400
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.626596000963623e-05,
+      "loss": 0.2364,
+      "step": 69410
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.6264621644048286e-05,
+      "loss": 0.1637,
+      "step": 69420
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.626328327846035e-05,
+      "loss": 0.3969,
+      "step": 69430
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.6261944912872404e-05,
+      "loss": 0.2543,
+      "step": 69440
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.626060654728446e-05,
+      "loss": 0.117,
+      "step": 69450
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.6259268181696514e-05,
+      "loss": 0.1454,
+      "step": 69460
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.625792981610857e-05,
+      "loss": 0.1444,
+      "step": 69470
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.6256591450520625e-05,
+      "loss": 0.1957,
+      "step": 69480
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.625525308493268e-05,
+      "loss": 0.1141,
+      "step": 69490
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.6253914719344736e-05,
+      "loss": 0.4085,
+      "step": 69500
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.625257635375679e-05,
+      "loss": 0.1514,
+      "step": 69510
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.6251237988168846e-05,
+      "loss": 0.1878,
+      "step": 69520
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.624989962258091e-05,
+      "loss": 0.2004,
+      "step": 69530
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.6248561256992964e-05,
+      "loss": 0.245,
+      "step": 69540
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.624722289140502e-05,
+      "loss": 0.1946,
+      "step": 69550
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.6245884525817074e-05,
+      "loss": 0.0701,
+      "step": 69560
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.624454616022913e-05,
+      "loss": 0.1679,
+      "step": 69570
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.6243207794641185e-05,
+      "loss": 0.1199,
+      "step": 69580
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.624186942905324e-05,
+      "loss": 0.2343,
+      "step": 69590
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.6240531063465296e-05,
+      "loss": 0.1197,
+      "step": 69600
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.623919269787735e-05,
+      "loss": 0.2664,
+      "step": 69610
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.6237854332289406e-05,
+      "loss": 0.123,
+      "step": 69620
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.623651596670147e-05,
+      "loss": 0.0878,
+      "step": 69630
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.6235177601113524e-05,
+      "loss": 0.1604,
+      "step": 69640
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.623383923552558e-05,
+      "loss": 0.1624,
+      "step": 69650
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.6232500869937635e-05,
+      "loss": 0.1827,
+      "step": 69660
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.623116250434969e-05,
+      "loss": 0.2401,
+      "step": 69670
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.6229824138761745e-05,
+      "loss": 0.3947,
+      "step": 69680
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.62284857731738e-05,
+      "loss": 0.262,
+      "step": 69690
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.6227147407585856e-05,
+      "loss": 0.1469,
+      "step": 69700
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.622580904199791e-05,
+      "loss": 0.1463,
+      "step": 69710
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.6224470676409967e-05,
+      "loss": 0.4664,
+      "step": 69720
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.622313231082203e-05,
+      "loss": 0.0427,
+      "step": 69730
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.6221793945234084e-05,
+      "loss": 0.0867,
+      "step": 69740
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.622045557964614e-05,
+      "loss": 0.186,
+      "step": 69750
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.6219117214058195e-05,
+      "loss": 0.1139,
+      "step": 69760
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.621777884847025e-05,
+      "loss": 0.197,
+      "step": 69770
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.6216440482882305e-05,
+      "loss": 0.1477,
+      "step": 69780
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.621510211729436e-05,
+      "loss": 0.1097,
+      "step": 69790
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.6213763751706416e-05,
+      "loss": 0.3423,
+      "step": 69800
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.621242538611847e-05,
+      "loss": 0.1401,
+      "step": 69810
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.6211087020530534e-05,
+      "loss": 0.0619,
+      "step": 69820
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.620974865494259e-05,
+      "loss": 0.3209,
+      "step": 69830
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.6208410289354644e-05,
+      "loss": 0.13,
+      "step": 69840
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.62070719237667e-05,
+      "loss": 0.2728,
+      "step": 69850
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.6205733558178755e-05,
+      "loss": 0.1311,
+      "step": 69860
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.620439519259081e-05,
+      "loss": 0.111,
+      "step": 69870
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.6203056827002866e-05,
+      "loss": 0.1403,
+      "step": 69880
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.620171846141492e-05,
+      "loss": 0.1517,
+      "step": 69890
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.6200380095826976e-05,
+      "loss": 0.0932,
+      "step": 69900
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.619904173023903e-05,
+      "loss": 0.2533,
+      "step": 69910
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.6197703364651094e-05,
+      "loss": 0.1091,
+      "step": 69920
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.619636499906315e-05,
+      "loss": 0.3214,
+      "step": 69930
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.6195026633475204e-05,
+      "loss": 0.0797,
+      "step": 69940
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.619368826788726e-05,
+      "loss": 0.0863,
+      "step": 69950
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.6192349902299315e-05,
+      "loss": 0.1411,
+      "step": 69960
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.619101153671137e-05,
+      "loss": 0.2339,
+      "step": 69970
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.6189673171123426e-05,
+      "loss": 0.1135,
+      "step": 69980
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.618833480553548e-05,
+      "loss": 0.2999,
+      "step": 69990
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.6186996439947536e-05,
+      "loss": 0.1682,
+      "step": 70000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.618565807435959e-05,
+      "loss": 0.3456,
+      "step": 70010
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.6184319708771654e-05,
+      "loss": 0.3369,
+      "step": 70020
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.618298134318371e-05,
+      "loss": 0.1139,
+      "step": 70030
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.6181642977595765e-05,
+      "loss": 0.1683,
+      "step": 70040
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.618030461200782e-05,
+      "loss": 0.1401,
+      "step": 70050
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.6178966246419875e-05,
+      "loss": 0.2696,
+      "step": 70060
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.617762788083193e-05,
+      "loss": 0.3551,
+      "step": 70070
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.6176289515243986e-05,
+      "loss": 0.1034,
+      "step": 70080
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.617495114965604e-05,
+      "loss": 0.0717,
+      "step": 70090
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.6173612784068097e-05,
+      "loss": 0.167,
+      "step": 70100
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.617227441848015e-05,
+      "loss": 0.1878,
+      "step": 70110
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.6170936052892214e-05,
+      "loss": 0.1246,
+      "step": 70120
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.616959768730427e-05,
+      "loss": 0.1607,
+      "step": 70130
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.6168259321716325e-05,
+      "loss": 0.2561,
+      "step": 70140
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.616692095612838e-05,
+      "loss": 0.1012,
+      "step": 70150
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.6165582590540435e-05,
+      "loss": 0.0835,
+      "step": 70160
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.616424422495249e-05,
+      "loss": 0.1731,
+      "step": 70170
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.6162905859364546e-05,
+      "loss": 0.1451,
+      "step": 70180
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.61615674937766e-05,
+      "loss": 0.1664,
+      "step": 70190
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.616022912818866e-05,
+      "loss": 0.1182,
+      "step": 70200
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.615889076260071e-05,
+      "loss": 0.3024,
+      "step": 70210
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.6157552397012774e-05,
+      "loss": 0.1779,
+      "step": 70220
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.615621403142483e-05,
+      "loss": 0.2283,
+      "step": 70230
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.6154875665836885e-05,
+      "loss": 0.3128,
+      "step": 70240
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.615353730024894e-05,
+      "loss": 0.1846,
+      "step": 70250
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.6152198934660995e-05,
+      "loss": 0.1465,
+      "step": 70260
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.615086056907305e-05,
+      "loss": 0.141,
+      "step": 70270
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.6149522203485106e-05,
+      "loss": 0.0684,
+      "step": 70280
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.614818383789716e-05,
+      "loss": 0.1936,
+      "step": 70290
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.614684547230922e-05,
+      "loss": 0.2442,
+      "step": 70300
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.614550710672127e-05,
+      "loss": 0.2214,
+      "step": 70310
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.6144168741133334e-05,
+      "loss": 0.1785,
+      "step": 70320
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.614283037554539e-05,
+      "loss": 0.0548,
+      "step": 70330
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.6141492009957445e-05,
+      "loss": 0.0186,
+      "step": 70340
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.61401536443695e-05,
+      "loss": 0.2324,
+      "step": 70350
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.6138815278781556e-05,
+      "loss": 0.123,
+      "step": 70360
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.613747691319361e-05,
+      "loss": 0.2942,
+      "step": 70370
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.6136138547605666e-05,
+      "loss": 0.2652,
+      "step": 70380
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.613480018201772e-05,
+      "loss": 0.1607,
+      "step": 70390
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.613346181642978e-05,
+      "loss": 0.2919,
+      "step": 70400
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.613212345084183e-05,
+      "loss": 0.1369,
+      "step": 70410
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.6130785085253894e-05,
+      "loss": 0.0732,
+      "step": 70420
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.612944671966595e-05,
+      "loss": 0.372,
+      "step": 70430
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.6128108354078005e-05,
+      "loss": 0.2254,
+      "step": 70440
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.612676998849006e-05,
+      "loss": 0.3614,
+      "step": 70450
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.6125431622902116e-05,
+      "loss": 0.0786,
+      "step": 70460
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.612409325731417e-05,
+      "loss": 0.1762,
+      "step": 70470
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.6122754891726226e-05,
+      "loss": 0.2015,
+      "step": 70480
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.612141652613828e-05,
+      "loss": 0.1407,
+      "step": 70490
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.612007816055034e-05,
+      "loss": 0.4339,
+      "step": 70500
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.611873979496239e-05,
+      "loss": 0.3083,
+      "step": 70510
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.6117401429374455e-05,
+      "loss": 0.0984,
+      "step": 70520
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.611606306378651e-05,
+      "loss": 0.0867,
+      "step": 70530
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.6114724698198565e-05,
+      "loss": 0.1611,
+      "step": 70540
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.611338633261062e-05,
+      "loss": 0.0912,
+      "step": 70550
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.6112047967022676e-05,
+      "loss": 0.4026,
+      "step": 70560
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.611070960143473e-05,
+      "loss": 0.0946,
+      "step": 70570
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.610937123584679e-05,
+      "loss": 0.1834,
+      "step": 70580
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.610803287025884e-05,
+      "loss": 0.0805,
+      "step": 70590
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.61066945046709e-05,
+      "loss": 0.088,
+      "step": 70600
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.610535613908295e-05,
+      "loss": 0.0498,
+      "step": 70610
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.6104017773495015e-05,
+      "loss": 0.2662,
+      "step": 70620
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.610267940790707e-05,
+      "loss": 0.0684,
+      "step": 70630
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.6101341042319125e-05,
+      "loss": 0.2299,
+      "step": 70640
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.610000267673118e-05,
+      "loss": 0.2841,
+      "step": 70650
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.6098664311143236e-05,
+      "loss": 0.1395,
+      "step": 70660
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.609732594555529e-05,
+      "loss": 0.1523,
+      "step": 70670
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.609598757996735e-05,
+      "loss": 0.388,
+      "step": 70680
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.60946492143794e-05,
+      "loss": 0.145,
+      "step": 70690
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.609331084879146e-05,
+      "loss": 0.2014,
+      "step": 70700
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.609197248320351e-05,
+      "loss": 0.3565,
+      "step": 70710
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.6090634117615575e-05,
+      "loss": 0.1697,
+      "step": 70720
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.608929575202763e-05,
+      "loss": 0.2359,
+      "step": 70730
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.6087957386439686e-05,
+      "loss": 0.2552,
+      "step": 70740
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.608661902085174e-05,
+      "loss": 0.0956,
+      "step": 70750
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.6085280655263796e-05,
+      "loss": 0.1566,
+      "step": 70760
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.608394228967585e-05,
+      "loss": 0.2245,
+      "step": 70770
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.608260392408791e-05,
+      "loss": 0.1689,
+      "step": 70780
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.608126555849996e-05,
+      "loss": 0.0931,
+      "step": 70790
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.607992719291202e-05,
+      "loss": 0.1535,
+      "step": 70800
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.607858882732407e-05,
+      "loss": 0.0856,
+      "step": 70810
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.6077250461736135e-05,
+      "loss": 0.1462,
+      "step": 70820
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.607591209614819e-05,
+      "loss": 0.0296,
+      "step": 70830
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.6074573730560246e-05,
+      "loss": 0.1526,
+      "step": 70840
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.60732353649723e-05,
+      "loss": 0.1739,
+      "step": 70850
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.6071896999384356e-05,
+      "loss": 0.0413,
+      "step": 70860
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.607055863379641e-05,
+      "loss": 0.2349,
+      "step": 70870
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.606922026820847e-05,
+      "loss": 0.0728,
+      "step": 70880
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.606788190262052e-05,
+      "loss": 0.1891,
+      "step": 70890
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.606654353703258e-05,
+      "loss": 0.2609,
+      "step": 70900
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.606520517144463e-05,
+      "loss": 0.1991,
+      "step": 70910
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.6063866805856695e-05,
+      "loss": 0.2943,
+      "step": 70920
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.606252844026875e-05,
+      "loss": 0.2144,
+      "step": 70930
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.6061190074680806e-05,
+      "loss": 0.0829,
+      "step": 70940
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.605985170909286e-05,
+      "loss": 0.1203,
+      "step": 70950
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.6058513343504917e-05,
+      "loss": 0.0979,
+      "step": 70960
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.605717497791697e-05,
+      "loss": 0.1588,
+      "step": 70970
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.605583661232903e-05,
+      "loss": 0.0943,
+      "step": 70980
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.605449824674108e-05,
+      "loss": 0.0366,
+      "step": 70990
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.605315988115314e-05,
+      "loss": 0.2495,
+      "step": 71000
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.60518215155652e-05,
+      "loss": 0.1811,
+      "step": 71010
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.6050483149977255e-05,
+      "loss": 0.0732,
+      "step": 71020
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.604914478438931e-05,
+      "loss": 0.2345,
+      "step": 71030
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6047806418801366e-05,
+      "loss": 0.2981,
+      "step": 71040
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.604646805321342e-05,
+      "loss": 0.0834,
+      "step": 71050
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.604512968762548e-05,
+      "loss": 0.2627,
+      "step": 71060
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6043791322037525e-05,
+      "loss": 0.1491,
+      "step": 71070
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.604245295644958e-05,
+      "loss": 0.076,
+      "step": 71080
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6041114590861636e-05,
+      "loss": 0.1588,
+      "step": 71090
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.60397762252737e-05,
+      "loss": 0.074,
+      "step": 71100
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6038437859685753e-05,
+      "loss": 0.2761,
+      "step": 71110
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.603709949409781e-05,
+      "loss": 0.2038,
+      "step": 71120
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6035761128509864e-05,
+      "loss": 0.2696,
+      "step": 71130
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.603442276292192e-05,
+      "loss": 0.206,
+      "step": 71140
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6033084397333975e-05,
+      "loss": 0.2439,
+      "step": 71150
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.603174603174603e-05,
+      "loss": 0.2053,
+      "step": 71160
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6030407666158085e-05,
+      "loss": 0.153,
+      "step": 71170
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.602906930057014e-05,
+      "loss": 0.172,
+      "step": 71180
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6027730934982196e-05,
+      "loss": 0.0999,
+      "step": 71190
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.602639256939426e-05,
+      "loss": 0.1363,
+      "step": 71200
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6025054203806314e-05,
+      "loss": 0.14,
+      "step": 71210
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.602371583821837e-05,
+      "loss": 0.1421,
+      "step": 71220
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6022377472630424e-05,
+      "loss": 0.2012,
+      "step": 71230
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.602103910704248e-05,
+      "loss": 0.0682,
+      "step": 71240
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6019700741454535e-05,
+      "loss": 0.1163,
+      "step": 71250
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.601836237586659e-05,
+      "loss": 0.0749,
+      "step": 71260
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6017024010278646e-05,
+      "loss": 0.1829,
+      "step": 71270
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.60156856446907e-05,
+      "loss": 0.1458,
+      "step": 71280
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.601434727910276e-05,
+      "loss": 0.0784,
+      "step": 71290
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.601300891351482e-05,
+      "loss": 0.0612,
+      "step": 71300
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.6011670547926874e-05,
+      "loss": 0.2166,
+      "step": 71310
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.601033218233893e-05,
+      "loss": 0.2631,
+      "step": 71320
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.6008993816750984e-05,
+      "loss": 0.1407,
+      "step": 71330
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.600765545116304e-05,
+      "loss": 0.115,
+      "step": 71340
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.6006317085575095e-05,
+      "loss": 0.2005,
+      "step": 71350
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.600497871998715e-05,
+      "loss": 0.072,
+      "step": 71360
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.6003640354399206e-05,
+      "loss": 0.3297,
+      "step": 71370
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.600230198881126e-05,
+      "loss": 0.2865,
+      "step": 71380
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.600096362322332e-05,
+      "loss": 0.0865,
+      "step": 71390
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.599962525763538e-05,
+      "loss": 0.0178,
+      "step": 71400
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5998286892047434e-05,
+      "loss": 0.2498,
+      "step": 71410
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.599694852645949e-05,
+      "loss": 0.1779,
+      "step": 71420
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5995610160871545e-05,
+      "loss": 0.2234,
+      "step": 71430
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.59942717952836e-05,
+      "loss": 0.2448,
+      "step": 71440
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5992933429695655e-05,
+      "loss": 0.2516,
+      "step": 71450
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.599159506410771e-05,
+      "loss": 0.3396,
+      "step": 71460
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5990256698519766e-05,
+      "loss": 0.0835,
+      "step": 71470
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.598891833293182e-05,
+      "loss": 0.2973,
+      "step": 71480
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.598757996734388e-05,
+      "loss": 0.0863,
+      "step": 71490
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.598624160175594e-05,
+      "loss": 0.2141,
+      "step": 71500
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5984903236167994e-05,
+      "loss": 0.0903,
+      "step": 71510
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.598356487058005e-05,
+      "loss": 0.2004,
+      "step": 71520
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.5982226504992105e-05,
+      "loss": 0.1131,
+      "step": 71530
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.598088813940416e-05,
+      "loss": 0.1376,
+      "step": 71540
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.5979549773816215e-05,
+      "loss": 0.2283,
+      "step": 71550
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.597821140822827e-05,
+      "loss": 0.1679,
+      "step": 71560
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.5976873042640326e-05,
+      "loss": 0.1137,
+      "step": 71570
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.597553467705238e-05,
+      "loss": 0.2873,
+      "step": 71580
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.5974196311464443e-05,
+      "loss": 0.1746,
+      "step": 71590
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.59728579458765e-05,
+      "loss": 0.1369,
+      "step": 71600
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.5971519580288554e-05,
+      "loss": 0.136,
+      "step": 71610
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.597018121470061e-05,
+      "loss": 0.1854,
+      "step": 71620
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.5968842849112665e-05,
+      "loss": 0.1671,
+      "step": 71630
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.596750448352472e-05,
+      "loss": 0.1021,
+      "step": 71640
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.5966166117936776e-05,
+      "loss": 0.1448,
+      "step": 71650
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.596482775234883e-05,
+      "loss": 0.1983,
+      "step": 71660
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.5963489386760886e-05,
+      "loss": 0.3136,
+      "step": 71670
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.596215102117294e-05,
+      "loss": 0.1001,
+      "step": 71680
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.5960812655585004e-05,
+      "loss": 0.0565,
+      "step": 71690
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.595947428999706e-05,
+      "loss": 0.1453,
+      "step": 71700
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.5958135924409114e-05,
+      "loss": 0.0865,
+      "step": 71710
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.595679755882117e-05,
+      "loss": 0.3071,
+      "step": 71720
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.5955459193233225e-05,
+      "loss": 0.2055,
+      "step": 71730
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.595412082764528e-05,
+      "loss": 0.089,
+      "step": 71740
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.5952782462057336e-05,
+      "loss": 0.1431,
+      "step": 71750
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.595144409646939e-05,
+      "loss": 0.2364,
+      "step": 71760
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.5950105730881446e-05,
+      "loss": 0.266,
+      "step": 71770
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.59487673652935e-05,
+      "loss": 0.218,
+      "step": 71780
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.5947428999705564e-05,
+      "loss": 0.2494,
+      "step": 71790
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.594609063411762e-05,
+      "loss": 0.1765,
+      "step": 71800
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.5944752268529674e-05,
+      "loss": 0.0085,
+      "step": 71810
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.594341390294173e-05,
+      "loss": 0.2476,
+      "step": 71820
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.5942075537353785e-05,
+      "loss": 0.2137,
+      "step": 71830
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.594073717176584e-05,
+      "loss": 0.2389,
+      "step": 71840
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.5939398806177896e-05,
+      "loss": 0.0877,
+      "step": 71850
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.593806044058995e-05,
+      "loss": 0.0204,
+      "step": 71860
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.5936722075002006e-05,
+      "loss": 0.3214,
+      "step": 71870
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.593538370941406e-05,
+      "loss": 0.1868,
+      "step": 71880
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.5934045343826124e-05,
+      "loss": 0.3143,
+      "step": 71890
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.593270697823818e-05,
+      "loss": 0.0705,
+      "step": 71900
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.5931368612650235e-05,
+      "loss": 0.1506,
+      "step": 71910
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.593003024706229e-05,
+      "loss": 0.1467,
+      "step": 71920
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.5928691881474345e-05,
+      "loss": 0.1557,
+      "step": 71930
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.59273535158864e-05,
+      "loss": 0.0538,
+      "step": 71940
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.5926015150298456e-05,
+      "loss": 0.1142,
+      "step": 71950
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.592467678471051e-05,
+      "loss": 0.231,
+      "step": 71960
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.592333841912257e-05,
+      "loss": 0.2762,
+      "step": 71970
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.592200005353462e-05,
+      "loss": 0.2186,
+      "step": 71980
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.5920661687946684e-05,
+      "loss": 0.2443,
+      "step": 71990
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.591932332235874e-05,
+      "loss": 0.2764,
+      "step": 72000
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.5917984956770795e-05,
+      "loss": 0.1293,
+      "step": 72010
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.591664659118285e-05,
+      "loss": 0.2299,
+      "step": 72020
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.5915308225594905e-05,
+      "loss": 0.1996,
+      "step": 72030
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.591396986000696e-05,
+      "loss": 0.0679,
+      "step": 72040
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.5912631494419016e-05,
+      "loss": 0.1238,
+      "step": 72050
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.591129312883107e-05,
+      "loss": 0.3629,
+      "step": 72060
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.590995476324313e-05,
+      "loss": 0.01,
+      "step": 72070
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.590861639765518e-05,
+      "loss": 0.1194,
+      "step": 72080
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.5907278032067244e-05,
+      "loss": 0.1839,
+      "step": 72090
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.59059396664793e-05,
+      "loss": 0.2537,
+      "step": 72100
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.5904601300891355e-05,
+      "loss": 0.1069,
+      "step": 72110
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.590326293530341e-05,
+      "loss": 0.1892,
+      "step": 72120
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.5901924569715466e-05,
+      "loss": 0.1654,
+      "step": 72130
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.590058620412752e-05,
+      "loss": 0.2097,
+      "step": 72140
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.5899247838539576e-05,
+      "loss": 0.2564,
+      "step": 72150
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.589790947295163e-05,
+      "loss": 0.1218,
+      "step": 72160
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.589657110736369e-05,
+      "loss": 0.2029,
+      "step": 72170
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.589523274177574e-05,
+      "loss": 0.1851,
+      "step": 72180
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.5893894376187804e-05,
+      "loss": 0.1035,
+      "step": 72190
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.589255601059986e-05,
+      "loss": 0.2725,
+      "step": 72200
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.5891217645011915e-05,
+      "loss": 0.2107,
+      "step": 72210
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.588987927942397e-05,
+      "loss": 0.1389,
+      "step": 72220
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.5888540913836026e-05,
+      "loss": 0.1498,
+      "step": 72230
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.588720254824808e-05,
+      "loss": 0.1639,
+      "step": 72240
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.5885864182660136e-05,
+      "loss": 0.2916,
+      "step": 72250
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.588452581707219e-05,
+      "loss": 0.2528,
+      "step": 72260
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.588318745148425e-05,
+      "loss": 0.041,
+      "step": 72270
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.58818490858963e-05,
+      "loss": 0.094,
+      "step": 72280
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.5880510720308365e-05,
+      "loss": 0.1762,
+      "step": 72290
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.587917235472042e-05,
+      "loss": 0.1243,
+      "step": 72300
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.5877833989132475e-05,
+      "loss": 0.1997,
+      "step": 72310
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.587649562354453e-05,
+      "loss": 0.2156,
+      "step": 72320
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.5875157257956586e-05,
+      "loss": 0.0935,
+      "step": 72330
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.587381889236864e-05,
+      "loss": 0.2139,
+      "step": 72340
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.5872480526780697e-05,
+      "loss": 0.0639,
+      "step": 72350
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.587114216119275e-05,
+      "loss": 0.0806,
+      "step": 72360
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.586980379560481e-05,
+      "loss": 0.1907,
+      "step": 72370
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.586846543001686e-05,
+      "loss": 0.1264,
+      "step": 72380
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.5867127064428925e-05,
+      "loss": 0.0082,
+      "step": 72390
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.586578869884098e-05,
+      "loss": 0.0728,
+      "step": 72400
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.5864450333253035e-05,
+      "loss": 0.2172,
+      "step": 72410
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.586311196766509e-05,
+      "loss": 0.2414,
+      "step": 72420
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.5861773602077146e-05,
+      "loss": 0.1112,
+      "step": 72430
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.58604352364892e-05,
+      "loss": 0.0843,
+      "step": 72440
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.585909687090126e-05,
+      "loss": 0.1153,
+      "step": 72450
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.585775850531331e-05,
+      "loss": 0.1283,
+      "step": 72460
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.585642013972537e-05,
+      "loss": 0.2022,
+      "step": 72470
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.585508177413742e-05,
+      "loss": 0.0809,
+      "step": 72480
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.5853743408549485e-05,
+      "loss": 0.2314,
+      "step": 72490
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.585240504296154e-05,
+      "loss": 0.1808,
+      "step": 72500
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.5851066677373596e-05,
+      "loss": 0.1501,
+      "step": 72510
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.584972831178565e-05,
+      "loss": 0.1923,
+      "step": 72520
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.5848389946197706e-05,
+      "loss": 0.1977,
+      "step": 72530
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.584705158060976e-05,
+      "loss": 0.3253,
+      "step": 72540
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.584571321502182e-05,
+      "loss": 0.0523,
+      "step": 72550
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.584437484943387e-05,
+      "loss": 0.2279,
+      "step": 72560
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.584303648384593e-05,
+      "loss": 0.1924,
+      "step": 72570
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.584169811825799e-05,
+      "loss": 0.1424,
+      "step": 72580
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.5840359752670045e-05,
+      "loss": 0.1277,
+      "step": 72590
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.58390213870821e-05,
+      "loss": 0.1584,
+      "step": 72600
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.5837683021494156e-05,
+      "loss": 0.2858,
+      "step": 72610
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.583634465590621e-05,
+      "loss": 0.1065,
+      "step": 72620
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.5835006290318266e-05,
+      "loss": 0.1728,
+      "step": 72630
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.583366792473032e-05,
+      "loss": 0.4167,
+      "step": 72640
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.583232955914238e-05,
+      "loss": 0.2669,
+      "step": 72650
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.583099119355443e-05,
+      "loss": 0.1695,
+      "step": 72660
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.582965282796649e-05,
+      "loss": 0.1864,
+      "step": 72670
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.582831446237855e-05,
+      "loss": 0.0516,
+      "step": 72680
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.5826976096790605e-05,
+      "loss": 0.2191,
+      "step": 72690
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.582563773120266e-05,
+      "loss": 0.1923,
+      "step": 72700
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.5824299365614716e-05,
+      "loss": 0.2711,
+      "step": 72710
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.582296100002677e-05,
+      "loss": 0.0998,
+      "step": 72720
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.5821622634438827e-05,
+      "loss": 0.2024,
+      "step": 72730
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.582028426885088e-05,
+      "loss": 0.1659,
+      "step": 72740
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.581894590326294e-05,
+      "loss": 0.1906,
+      "step": 72750
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.581760753767499e-05,
+      "loss": 0.1145,
+      "step": 72760
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.581626917208705e-05,
+      "loss": 0.1916,
+      "step": 72770
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.581493080649911e-05,
+      "loss": 0.0956,
+      "step": 72780
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.5813592440911165e-05,
+      "loss": 0.1873,
+      "step": 72790
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.581225407532322e-05,
+      "loss": 0.2455,
+      "step": 72800
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.5810915709735276e-05,
+      "loss": 0.0973,
+      "step": 72810
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.580957734414733e-05,
+      "loss": 0.0908,
+      "step": 72820
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.580823897855939e-05,
+      "loss": 0.1078,
+      "step": 72830
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.580690061297144e-05,
+      "loss": 0.1353,
+      "step": 72840
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.58055622473835e-05,
+      "loss": 0.2347,
+      "step": 72850
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.580422388179555e-05,
+      "loss": 0.1474,
+      "step": 72860
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.580288551620761e-05,
+      "loss": 0.2357,
+      "step": 72870
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.580154715061967e-05,
+      "loss": 0.0254,
+      "step": 72880
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.5800208785031725e-05,
+      "loss": 0.0601,
+      "step": 72890
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.579887041944378e-05,
+      "loss": 0.1583,
+      "step": 72900
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.5797532053855836e-05,
+      "loss": 0.055,
+      "step": 72910
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.579619368826789e-05,
+      "loss": 0.1642,
+      "step": 72920
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.579485532267995e-05,
+      "loss": 0.2661,
+      "step": 72930
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.5793516957092e-05,
+      "loss": 0.112,
+      "step": 72940
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.579217859150406e-05,
+      "loss": 0.068,
+      "step": 72950
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.579084022591611e-05,
+      "loss": 0.1794,
+      "step": 72960
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.578950186032817e-05,
+      "loss": 0.0649,
+      "step": 72970
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.578816349474023e-05,
+      "loss": 0.1377,
+      "step": 72980
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.5786825129152286e-05,
+      "loss": 0.2299,
+      "step": 72990
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.578548676356434e-05,
+      "loss": 0.2696,
+      "step": 73000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.5784148397976396e-05,
+      "loss": 0.0409,
+      "step": 73010
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.578281003238845e-05,
+      "loss": 0.0737,
+      "step": 73020
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.578147166680051e-05,
+      "loss": 0.1091,
+      "step": 73030
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.578013330121256e-05,
+      "loss": 0.2121,
+      "step": 73040
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.577879493562462e-05,
+      "loss": 0.1033,
+      "step": 73050
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.577745657003667e-05,
+      "loss": 0.1034,
+      "step": 73060
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.577611820444873e-05,
+      "loss": 0.0146,
+      "step": 73070
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.577477983886079e-05,
+      "loss": 0.1302,
+      "step": 73080
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.5773441473272846e-05,
+      "loss": 0.1432,
+      "step": 73090
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.57721031076849e-05,
+      "loss": 0.3903,
+      "step": 73100
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.5770764742096956e-05,
+      "loss": 0.35,
+      "step": 73110
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.576942637650901e-05,
+      "loss": 0.1275,
+      "step": 73120
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.576808801092107e-05,
+      "loss": 0.0611,
+      "step": 73130
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.576674964533312e-05,
+      "loss": 0.2059,
+      "step": 73140
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.576541127974518e-05,
+      "loss": 0.2356,
+      "step": 73150
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.576407291415723e-05,
+      "loss": 0.2248,
+      "step": 73160
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.576273454856929e-05,
+      "loss": 0.1206,
+      "step": 73170
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.576139618298135e-05,
+      "loss": 0.0212,
+      "step": 73180
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.5760057817393406e-05,
+      "loss": 0.2994,
+      "step": 73190
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.575871945180546e-05,
+      "loss": 0.1027,
+      "step": 73200
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.5757381086217517e-05,
+      "loss": 0.2145,
+      "step": 73210
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.575604272062957e-05,
+      "loss": 0.1573,
+      "step": 73220
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.575470435504163e-05,
+      "loss": 0.2353,
+      "step": 73230
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.575336598945368e-05,
+      "loss": 0.2613,
+      "step": 73240
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.575202762386574e-05,
+      "loss": 0.1601,
+      "step": 73250
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.575068925827779e-05,
+      "loss": 0.1961,
+      "step": 73260
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.574935089268985e-05,
+      "loss": 0.1255,
+      "step": 73270
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.574801252710191e-05,
+      "loss": 0.1206,
+      "step": 73280
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.5746674161513966e-05,
+      "loss": 0.1948,
+      "step": 73290
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.574533579592602e-05,
+      "loss": 0.1783,
+      "step": 73300
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.574399743033808e-05,
+      "loss": 0.0911,
+      "step": 73310
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.574265906475013e-05,
+      "loss": 0.257,
+      "step": 73320
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.574132069916219e-05,
+      "loss": 0.1303,
+      "step": 73330
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.573998233357424e-05,
+      "loss": 0.1596,
+      "step": 73340
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.57386439679863e-05,
+      "loss": 0.1792,
+      "step": 73350
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.5737305602398353e-05,
+      "loss": 0.0764,
+      "step": 73360
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.573596723681041e-05,
+      "loss": 0.3042,
+      "step": 73370
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.573462887122247e-05,
+      "loss": 0.0708,
+      "step": 73380
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.5733290505634526e-05,
+      "loss": 0.1978,
+      "step": 73390
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.573195214004658e-05,
+      "loss": 0.1399,
+      "step": 73400
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.573061377445864e-05,
+      "loss": 0.0959,
+      "step": 73410
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.572927540887069e-05,
+      "loss": 0.1591,
+      "step": 73420
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.572793704328275e-05,
+      "loss": 0.1388,
+      "step": 73430
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.5726598677694796e-05,
+      "loss": 0.1272,
+      "step": 73440
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.572526031210685e-05,
+      "loss": 0.1246,
+      "step": 73450
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.5723921946518914e-05,
+      "loss": 0.3199,
+      "step": 73460
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.572258358093097e-05,
+      "loss": 0.157,
+      "step": 73470
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.5721245215343024e-05,
+      "loss": 0.048,
+      "step": 73480
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.571990684975508e-05,
+      "loss": 0.1497,
+      "step": 73490
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.5718568484167135e-05,
+      "loss": 0.0437,
+      "step": 73500
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.571723011857919e-05,
+      "loss": 0.1734,
+      "step": 73510
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.5715891752991246e-05,
+      "loss": 0.0421,
+      "step": 73520
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.57145533874033e-05,
+      "loss": 0.1265,
+      "step": 73530
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.5713215021815356e-05,
+      "loss": 0.2711,
+      "step": 73540
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.571187665622741e-05,
+      "loss": 0.1381,
+      "step": 73550
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.5710538290639474e-05,
+      "loss": 0.317,
+      "step": 73560
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.570919992505153e-05,
+      "loss": 0.163,
+      "step": 73570
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.5707861559463584e-05,
+      "loss": 0.105,
+      "step": 73580
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.570652319387564e-05,
+      "loss": 0.04,
+      "step": 73590
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.5705184828287695e-05,
+      "loss": 0.118,
+      "step": 73600
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.570384646269975e-05,
+      "loss": 0.0861,
+      "step": 73610
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.5702508097111806e-05,
+      "loss": 0.1897,
+      "step": 73620
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.570116973152386e-05,
+      "loss": 0.0546,
+      "step": 73630
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.5699831365935916e-05,
+      "loss": 0.0737,
+      "step": 73640
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.569849300034797e-05,
+      "loss": 0.1074,
+      "step": 73650
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.5697154634760034e-05,
+      "loss": 0.0383,
+      "step": 73660
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.569581626917209e-05,
+      "loss": 0.2801,
+      "step": 73670
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.5694477903584145e-05,
+      "loss": 0.1817,
+      "step": 73680
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.56931395379962e-05,
+      "loss": 0.0169,
+      "step": 73690
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.5691801172408255e-05,
+      "loss": 0.1702,
+      "step": 73700
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.569046280682031e-05,
+      "loss": 0.1392,
+      "step": 73710
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.5689124441232366e-05,
+      "loss": 0.3117,
+      "step": 73720
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.568778607564442e-05,
+      "loss": 0.0932,
+      "step": 73730
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.5686447710056477e-05,
+      "loss": 0.3675,
+      "step": 73740
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.568510934446853e-05,
+      "loss": 0.1932,
+      "step": 73750
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.5683770978880594e-05,
+      "loss": 0.2819,
+      "step": 73760
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.568243261329265e-05,
+      "loss": 0.1101,
+      "step": 73770
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.5681094247704705e-05,
+      "loss": 0.1654,
+      "step": 73780
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.567975588211676e-05,
+      "loss": 0.154,
+      "step": 73790
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.5678417516528815e-05,
+      "loss": 0.1488,
+      "step": 73800
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.567707915094087e-05,
+      "loss": 0.2498,
+      "step": 73810
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.5675740785352926e-05,
+      "loss": 0.1332,
+      "step": 73820
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.567440241976498e-05,
+      "loss": 0.1622,
+      "step": 73830
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.567306405417704e-05,
+      "loss": 0.1176,
+      "step": 73840
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.567172568858909e-05,
+      "loss": 0.3182,
+      "step": 73850
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.5670387323001154e-05,
+      "loss": 0.0579,
+      "step": 73860
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.566904895741321e-05,
+      "loss": 0.2567,
+      "step": 73870
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.5667710591825265e-05,
+      "loss": 0.1962,
+      "step": 73880
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.566637222623732e-05,
+      "loss": 0.1378,
+      "step": 73890
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.5665033860649376e-05,
+      "loss": 0.2705,
+      "step": 73900
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.566369549506143e-05,
+      "loss": 0.1833,
+      "step": 73910
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.5662357129473486e-05,
+      "loss": 0.2416,
+      "step": 73920
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.566101876388554e-05,
+      "loss": 0.1017,
+      "step": 73930
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.56596803982976e-05,
+      "loss": 0.2744,
+      "step": 73940
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.565834203270965e-05,
+      "loss": 0.0674,
+      "step": 73950
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.5657003667121714e-05,
+      "loss": 0.1043,
+      "step": 73960
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.565566530153377e-05,
+      "loss": 0.0945,
+      "step": 73970
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.5654326935945825e-05,
+      "loss": 0.1662,
+      "step": 73980
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.565298857035788e-05,
+      "loss": 0.4113,
+      "step": 73990
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.5651650204769936e-05,
+      "loss": 0.1466,
+      "step": 74000
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.565031183918199e-05,
+      "loss": 0.0336,
+      "step": 74010
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.5648973473594046e-05,
+      "loss": 0.2649,
+      "step": 74020
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.56476351080061e-05,
+      "loss": 0.2024,
+      "step": 74030
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.564629674241816e-05,
+      "loss": 0.2417,
+      "step": 74040
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.564495837683022e-05,
+      "loss": 0.125,
+      "step": 74050
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.5643620011242275e-05,
+      "loss": 0.0775,
+      "step": 74060
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.564228164565433e-05,
+      "loss": 0.2266,
+      "step": 74070
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.5640943280066385e-05,
+      "loss": 0.2519,
+      "step": 74080
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.563960491447844e-05,
+      "loss": 0.2581,
+      "step": 74090
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.5638266548890496e-05,
+      "loss": 0.0411,
+      "step": 74100
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.563692818330255e-05,
+      "loss": 0.1336,
+      "step": 74110
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.5635589817714607e-05,
+      "loss": 0.1598,
+      "step": 74120
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.563425145212666e-05,
+      "loss": 0.1823,
+      "step": 74130
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.563291308653872e-05,
+      "loss": 0.1914,
+      "step": 74140
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.563157472095078e-05,
+      "loss": 0.227,
+      "step": 74150
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.5630236355362835e-05,
+      "loss": 0.2615,
+      "step": 74160
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.562889798977489e-05,
+      "loss": 0.1314,
+      "step": 74170
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.5627559624186945e-05,
+      "loss": 0.137,
+      "step": 74180
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.5626221258599e-05,
+      "loss": 0.1995,
+      "step": 74190
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.5624882893011056e-05,
+      "loss": 0.1025,
+      "step": 74200
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.562354452742311e-05,
+      "loss": 0.1284,
+      "step": 74210
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.562220616183517e-05,
+      "loss": 0.0173,
+      "step": 74220
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.562086779624722e-05,
+      "loss": 0.2937,
+      "step": 74230
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.561952943065928e-05,
+      "loss": 0.1441,
+      "step": 74240
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.561819106507134e-05,
+      "loss": 0.1472,
+      "step": 74250
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.5616852699483395e-05,
+      "loss": 0.1755,
+      "step": 74260
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.561551433389545e-05,
+      "loss": 0.1772,
+      "step": 74270
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.5614175968307505e-05,
+      "loss": 0.218,
+      "step": 74280
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.561283760271956e-05,
+      "loss": 0.0599,
+      "step": 74290
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.5611499237131616e-05,
+      "loss": 0.2219,
+      "step": 74300
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.561016087154367e-05,
+      "loss": 0.2073,
+      "step": 74310
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.560882250595573e-05,
+      "loss": 0.2053,
+      "step": 74320
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.560748414036778e-05,
+      "loss": 0.2915,
+      "step": 74330
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.560614577477984e-05,
+      "loss": 0.1019,
+      "step": 74340
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.56048074091919e-05,
+      "loss": 0.2469,
+      "step": 74350
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.5603469043603955e-05,
+      "loss": 0.1988,
+      "step": 74360
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.560213067801601e-05,
+      "loss": 0.2457,
+      "step": 74370
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.5600792312428066e-05,
+      "loss": 0.0731,
+      "step": 74380
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.559945394684012e-05,
+      "loss": 0.3501,
+      "step": 74390
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.5598115581252176e-05,
+      "loss": 0.1985,
+      "step": 74400
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.559677721566423e-05,
+      "loss": 0.1599,
+      "step": 74410
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.559543885007629e-05,
+      "loss": 0.0976,
+      "step": 74420
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.559410048448834e-05,
+      "loss": 0.1578,
+      "step": 74430
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.55927621189004e-05,
+      "loss": 0.0475,
+      "step": 74440
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.559142375331246e-05,
+      "loss": 0.019,
+      "step": 74450
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.5590085387724515e-05,
+      "loss": 0.1295,
+      "step": 74460
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.558874702213657e-05,
+      "loss": 0.0834,
+      "step": 74470
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.5587408656548626e-05,
+      "loss": 0.208,
+      "step": 74480
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.558607029096068e-05,
+      "loss": 0.3912,
+      "step": 74490
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.5584731925372736e-05,
+      "loss": 0.141,
+      "step": 74500
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.558339355978479e-05,
+      "loss": 0.1864,
+      "step": 74510
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.558205519419685e-05,
+      "loss": 0.0517,
+      "step": 74520
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.55807168286089e-05,
+      "loss": 0.1711,
+      "step": 74530
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.557937846302096e-05,
+      "loss": 0.2527,
+      "step": 74540
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.557804009743302e-05,
+      "loss": 0.0293,
+      "step": 74550
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.5576701731845075e-05,
+      "loss": 0.2096,
+      "step": 74560
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.557536336625713e-05,
+      "loss": 0.2262,
+      "step": 74570
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.5574025000669186e-05,
+      "loss": 0.2072,
+      "step": 74580
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.557268663508124e-05,
+      "loss": 0.1093,
+      "step": 74590
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.55713482694933e-05,
+      "loss": 0.1043,
+      "step": 74600
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.557000990390535e-05,
+      "loss": 0.0729,
+      "step": 74610
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.556867153831741e-05,
+      "loss": 0.1958,
+      "step": 74620
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.556733317272946e-05,
+      "loss": 0.1492,
+      "step": 74630
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.556599480714152e-05,
+      "loss": 0.1121,
+      "step": 74640
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.556465644155358e-05,
+      "loss": 0.314,
+      "step": 74650
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.5563318075965635e-05,
+      "loss": 0.2359,
+      "step": 74660
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.556197971037769e-05,
+      "loss": 0.1544,
+      "step": 74670
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.5560641344789746e-05,
+      "loss": 0.2913,
+      "step": 74680
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.55593029792018e-05,
+      "loss": 0.2038,
+      "step": 74690
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.555796461361386e-05,
+      "loss": 0.1648,
+      "step": 74700
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.555662624802591e-05,
+      "loss": 0.0574,
+      "step": 74710
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.555528788243797e-05,
+      "loss": 0.09,
+      "step": 74720
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": {
+        "accuracy": 0.9436314363143632
+      },
+      "eval_f1": {
+        "f1": 0.9433683232067358
+      },
+      "eval_loss": 0.24802996218204498,
+      "eval_precision": {
+        "precision": 0.9452971145459653
+      },
+      "eval_recall": {
+        "recall": 0.9433746555197686
+      },
+      "eval_runtime": 431.6496,
+      "eval_samples_per_second": 38.469,
+      "eval_steps_per_second": 19.236,
+      "step": 74722
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.555394951685002e-05,
+      "loss": 0.1622,
+      "step": 74730
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.555261115126208e-05,
+      "loss": 0.1173,
+      "step": 74740
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.555127278567414e-05,
+      "loss": 0.2324,
+      "step": 74750
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.5549934420086196e-05,
+      "loss": 0.1687,
+      "step": 74760
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.554859605449825e-05,
+      "loss": 0.207,
+      "step": 74770
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.5547257688910306e-05,
+      "loss": 0.2133,
+      "step": 74780
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.554591932332236e-05,
+      "loss": 0.1438,
+      "step": 74790
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.554458095773442e-05,
+      "loss": 0.1442,
+      "step": 74800
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.554324259214647e-05,
+      "loss": 0.0517,
+      "step": 74810
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.554190422655853e-05,
+      "loss": 0.1272,
+      "step": 74820
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.554056586097058e-05,
+      "loss": 0.0815,
+      "step": 74830
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.553922749538264e-05,
+      "loss": 0.2313,
+      "step": 74840
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.55378891297947e-05,
+      "loss": 0.1228,
+      "step": 74850
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.5536550764206756e-05,
+      "loss": 0.2547,
+      "step": 74860
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.553521239861881e-05,
+      "loss": 0.1763,
+      "step": 74870
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.5533874033030866e-05,
+      "loss": 0.307,
+      "step": 74880
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.553253566744292e-05,
+      "loss": 0.1002,
+      "step": 74890
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.553119730185498e-05,
+      "loss": 0.2991,
+      "step": 74900
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.552985893626703e-05,
+      "loss": 0.2228,
+      "step": 74910
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.552852057067909e-05,
+      "loss": 0.1192,
+      "step": 74920
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.552718220509114e-05,
+      "loss": 0.1663,
+      "step": 74930
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.55258438395032e-05,
+      "loss": 0.1568,
+      "step": 74940
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.552450547391526e-05,
+      "loss": 0.1314,
+      "step": 74950
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.5523167108327316e-05,
+      "loss": 0.2175,
+      "step": 74960
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.552182874273937e-05,
+      "loss": 0.0896,
+      "step": 74970
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.5520490377151427e-05,
+      "loss": 0.1578,
+      "step": 74980
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.551915201156348e-05,
+      "loss": 0.2514,
+      "step": 74990
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.551781364597554e-05,
+      "loss": 0.0848,
+      "step": 75000
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.551647528038759e-05,
+      "loss": 0.1226,
+      "step": 75010
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.551513691479965e-05,
+      "loss": 0.1002,
+      "step": 75020
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.55137985492117e-05,
+      "loss": 0.0913,
+      "step": 75030
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.551246018362376e-05,
+      "loss": 0.074,
+      "step": 75040
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.551112181803582e-05,
+      "loss": 0.0968,
+      "step": 75050
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.5509783452447876e-05,
+      "loss": 0.2116,
+      "step": 75060
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.550844508685993e-05,
+      "loss": 0.0707,
+      "step": 75070
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.550710672127199e-05,
+      "loss": 0.2804,
+      "step": 75080
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.550576835568404e-05,
+      "loss": 0.1145,
+      "step": 75090
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.55044299900961e-05,
+      "loss": 0.1267,
+      "step": 75100
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.550309162450815e-05,
+      "loss": 0.1385,
+      "step": 75110
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.550175325892021e-05,
+      "loss": 0.0458,
+      "step": 75120
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.5500414893332263e-05,
+      "loss": 0.2094,
+      "step": 75130
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.549907652774432e-05,
+      "loss": 0.1856,
+      "step": 75140
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.549773816215638e-05,
+      "loss": 0.0454,
+      "step": 75150
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.5496399796568436e-05,
+      "loss": 0.2569,
+      "step": 75160
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.549506143098049e-05,
+      "loss": 0.1133,
+      "step": 75170
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.549372306539255e-05,
+      "loss": 0.0552,
+      "step": 75180
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.54923846998046e-05,
+      "loss": 0.0783,
+      "step": 75190
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.549104633421666e-05,
+      "loss": 0.1973,
+      "step": 75200
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.548970796862871e-05,
+      "loss": 0.1351,
+      "step": 75210
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.548836960304077e-05,
+      "loss": 0.1452,
+      "step": 75220
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.5487031237452824e-05,
+      "loss": 0.3062,
+      "step": 75230
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.548569287186488e-05,
+      "loss": 0.2562,
+      "step": 75240
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.548435450627694e-05,
+      "loss": 0.1525,
+      "step": 75250
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.5483016140688996e-05,
+      "loss": 0.2411,
+      "step": 75260
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.548167777510105e-05,
+      "loss": 0.0185,
+      "step": 75270
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.548033940951311e-05,
+      "loss": 0.1029,
+      "step": 75280
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.547900104392516e-05,
+      "loss": 0.2701,
+      "step": 75290
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.547766267833722e-05,
+      "loss": 0.103,
+      "step": 75300
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.547632431274927e-05,
+      "loss": 0.1329,
+      "step": 75310
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.547498594716133e-05,
+      "loss": 0.1588,
+      "step": 75320
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.5473647581573384e-05,
+      "loss": 0.121,
+      "step": 75330
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.5472309215985446e-05,
+      "loss": 0.4273,
+      "step": 75340
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.54709708503975e-05,
+      "loss": 0.1511,
+      "step": 75350
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.5469632484809556e-05,
+      "loss": 0.1718,
+      "step": 75360
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.546829411922161e-05,
+      "loss": 0.2011,
+      "step": 75370
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.546695575363367e-05,
+      "loss": 0.2176,
+      "step": 75380
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.546561738804572e-05,
+      "loss": 0.0911,
+      "step": 75390
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.546427902245778e-05,
+      "loss": 0.2535,
+      "step": 75400
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.546294065686983e-05,
+      "loss": 0.1276,
+      "step": 75410
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.546160229128189e-05,
+      "loss": 0.1492,
+      "step": 75420
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.5460263925693944e-05,
+      "loss": 0.1097,
+      "step": 75430
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.5458925560106006e-05,
+      "loss": 0.215,
+      "step": 75440
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545758719451806e-05,
+      "loss": 0.097,
+      "step": 75450
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545624882893012e-05,
+      "loss": 0.1527,
+      "step": 75460
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545491046334217e-05,
+      "loss": 0.0975,
+      "step": 75470
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545357209775423e-05,
+      "loss": 0.277,
+      "step": 75480
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545223373216628e-05,
+      "loss": 0.2236,
+      "step": 75490
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545089536657834e-05,
+      "loss": 0.2624,
+      "step": 75500
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.544955700099039e-05,
+      "loss": 0.2481,
+      "step": 75510
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.544821863540245e-05,
+      "loss": 0.0836,
+      "step": 75520
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.5446880269814504e-05,
+      "loss": 0.0769,
+      "step": 75530
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.5445541904226566e-05,
+      "loss": 0.2421,
+      "step": 75540
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.544420353863862e-05,
+      "loss": 0.2178,
+      "step": 75550
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.544286517305068e-05,
+      "loss": 0.2096,
+      "step": 75560
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.544152680746273e-05,
+      "loss": 0.1722,
+      "step": 75570
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.544018844187479e-05,
+      "loss": 0.1286,
+      "step": 75580
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.543885007628684e-05,
+      "loss": 0.2642,
+      "step": 75590
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.54375117106989e-05,
+      "loss": 0.1099,
+      "step": 75600
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.5436173345110953e-05,
+      "loss": 0.0811,
+      "step": 75610
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.543483497952301e-05,
+      "loss": 0.0525,
+      "step": 75620
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.5433496613935064e-05,
+      "loss": 0.1495,
+      "step": 75630
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.5432158248347126e-05,
+      "loss": 0.2537,
+      "step": 75640
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.543081988275918e-05,
+      "loss": 0.0945,
+      "step": 75650
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.542948151717124e-05,
+      "loss": 0.4523,
+      "step": 75660
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.542814315158329e-05,
+      "loss": 0.1553,
+      "step": 75670
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.542680478599535e-05,
+      "loss": 0.0365,
+      "step": 75680
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.54254664204074e-05,
+      "loss": 0.4207,
+      "step": 75690
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.542412805481946e-05,
+      "loss": 0.3954,
+      "step": 75700
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.5422789689231514e-05,
+      "loss": 0.0888,
+      "step": 75710
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.542145132364357e-05,
+      "loss": 0.2227,
+      "step": 75720
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.5420112958055624e-05,
+      "loss": 0.1566,
+      "step": 75730
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.5418774592467686e-05,
+      "loss": 0.1657,
+      "step": 75740
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.541743622687974e-05,
+      "loss": 0.2602,
+      "step": 75750
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.54160978612918e-05,
+      "loss": 0.2493,
+      "step": 75760
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.541475949570385e-05,
+      "loss": 0.2182,
+      "step": 75770
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.541342113011591e-05,
+      "loss": 0.3114,
+      "step": 75780
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.541208276452796e-05,
+      "loss": 0.0814,
+      "step": 75790
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.541074439894001e-05,
+      "loss": 0.1027,
+      "step": 75800
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.540940603335207e-05,
+      "loss": 0.3226,
+      "step": 75810
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.540806766776413e-05,
+      "loss": 0.1227,
+      "step": 75820
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.5406729302176184e-05,
+      "loss": 0.0648,
+      "step": 75830
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.540539093658824e-05,
+      "loss": 0.197,
+      "step": 75840
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.5404052571000295e-05,
+      "loss": 0.0872,
+      "step": 75850
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.540271420541235e-05,
+      "loss": 0.266,
+      "step": 75860
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.5401375839824406e-05,
+      "loss": 0.307,
+      "step": 75870
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.540003747423646e-05,
+      "loss": 0.1209,
+      "step": 75880
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.5398699108648516e-05,
+      "loss": 0.1596,
+      "step": 75890
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.539736074306057e-05,
+      "loss": 0.1273,
+      "step": 75900
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.539602237747263e-05,
+      "loss": 0.2087,
+      "step": 75910
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.539468401188469e-05,
+      "loss": 0.2031,
+      "step": 75920
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.5393345646296745e-05,
+      "loss": 0.1151,
+      "step": 75930
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.53920072807088e-05,
+      "loss": 0.1912,
+      "step": 75940
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.5390668915120855e-05,
+      "loss": 0.0987,
+      "step": 75950
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.538933054953291e-05,
+      "loss": 0.1435,
+      "step": 75960
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.5387992183944966e-05,
+      "loss": 0.1897,
+      "step": 75970
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.538665381835702e-05,
+      "loss": 0.1043,
+      "step": 75980
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.538531545276908e-05,
+      "loss": 0.2981,
+      "step": 75990
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.538397708718113e-05,
+      "loss": 0.2388,
+      "step": 76000
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.538263872159319e-05,
+      "loss": 0.3692,
+      "step": 76010
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.538130035600525e-05,
+      "loss": 0.1486,
+      "step": 76020
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.5379961990417305e-05,
+      "loss": 0.2352,
+      "step": 76030
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.537862362482936e-05,
+      "loss": 0.2215,
+      "step": 76040
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.5377285259241415e-05,
+      "loss": 0.2255,
+      "step": 76050
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.537594689365347e-05,
+      "loss": 0.2561,
+      "step": 76060
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.5374608528065526e-05,
+      "loss": 0.1693,
+      "step": 76070
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.537327016247758e-05,
+      "loss": 0.129,
+      "step": 76080
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.537193179688964e-05,
+      "loss": 0.1362,
+      "step": 76090
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.537059343130169e-05,
+      "loss": 0.0766,
+      "step": 76100
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.536925506571375e-05,
+      "loss": 0.1224,
+      "step": 76110
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.536791670012581e-05,
+      "loss": 0.1978,
+      "step": 76120
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.5366578334537865e-05,
+      "loss": 0.1397,
+      "step": 76130
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.536523996894992e-05,
+      "loss": 0.1939,
+      "step": 76140
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.5363901603361976e-05,
+      "loss": 0.2509,
+      "step": 76150
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.536256323777403e-05,
+      "loss": 0.1343,
+      "step": 76160
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.5361224872186086e-05,
+      "loss": 0.1308,
+      "step": 76170
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.535988650659814e-05,
+      "loss": 0.1471,
+      "step": 76180
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.53585481410102e-05,
+      "loss": 0.1342,
+      "step": 76190
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.535720977542225e-05,
+      "loss": 0.1949,
+      "step": 76200
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.535587140983431e-05,
+      "loss": 0.1332,
+      "step": 76210
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.535453304424637e-05,
+      "loss": 0.3212,
+      "step": 76220
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.5353194678658425e-05,
+      "loss": 0.0766,
+      "step": 76230
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.535185631307048e-05,
+      "loss": 0.1537,
+      "step": 76240
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.5350517947482536e-05,
+      "loss": 0.2361,
+      "step": 76250
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.534917958189459e-05,
+      "loss": 0.2258,
+      "step": 76260
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.5347841216306646e-05,
+      "loss": 0.1122,
+      "step": 76270
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.53465028507187e-05,
+      "loss": 0.2746,
+      "step": 76280
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.534516448513076e-05,
+      "loss": 0.1404,
+      "step": 76290
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.534382611954281e-05,
+      "loss": 0.2438,
+      "step": 76300
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.534248775395487e-05,
+      "loss": 0.2196,
+      "step": 76310
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.534114938836693e-05,
+      "loss": 0.0327,
+      "step": 76320
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.5339811022778985e-05,
+      "loss": 0.0656,
+      "step": 76330
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.533847265719104e-05,
+      "loss": 0.032,
+      "step": 76340
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.5337134291603096e-05,
+      "loss": 0.0276,
+      "step": 76350
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.533579592601515e-05,
+      "loss": 0.076,
+      "step": 76360
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.5334457560427207e-05,
+      "loss": 0.1307,
+      "step": 76370
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.533311919483926e-05,
+      "loss": 0.1575,
+      "step": 76380
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.533178082925132e-05,
+      "loss": 0.289,
+      "step": 76390
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.533044246366337e-05,
+      "loss": 0.1512,
+      "step": 76400
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.532910409807543e-05,
+      "loss": 0.2122,
+      "step": 76410
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.532776573248749e-05,
+      "loss": 0.1979,
+      "step": 76420
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.5326427366899545e-05,
+      "loss": 0.2141,
+      "step": 76430
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.53250890013116e-05,
+      "loss": 0.3101,
+      "step": 76440
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.5323750635723656e-05,
+      "loss": 0.1345,
+      "step": 76450
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.532241227013571e-05,
+      "loss": 0.132,
+      "step": 76460
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.532107390454777e-05,
+      "loss": 0.1819,
+      "step": 76470
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.531973553895982e-05,
+      "loss": 0.136,
+      "step": 76480
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.531839717337188e-05,
+      "loss": 0.0871,
+      "step": 76490
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.531705880778393e-05,
+      "loss": 0.152,
+      "step": 76500
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.531572044219599e-05,
+      "loss": 0.1744,
+      "step": 76510
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.531438207660805e-05,
+      "loss": 0.1668,
+      "step": 76520
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.5313043711020106e-05,
+      "loss": 0.0984,
+      "step": 76530
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.531170534543216e-05,
+      "loss": 0.0484,
+      "step": 76540
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.5310366979844216e-05,
+      "loss": 0.2115,
+      "step": 76550
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.530902861425627e-05,
+      "loss": 0.1721,
+      "step": 76560
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.530769024866833e-05,
+      "loss": 0.1003,
+      "step": 76570
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.530635188308038e-05,
+      "loss": 0.3268,
+      "step": 76580
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.530501351749244e-05,
+      "loss": 0.2341,
+      "step": 76590
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.530367515190449e-05,
+      "loss": 0.1457,
+      "step": 76600
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.530233678631655e-05,
+      "loss": 0.1353,
+      "step": 76610
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.530099842072861e-05,
+      "loss": 0.1138,
+      "step": 76620
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.5299660055140666e-05,
+      "loss": 0.0734,
+      "step": 76630
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.529832168955272e-05,
+      "loss": 0.0925,
+      "step": 76640
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.5296983323964776e-05,
+      "loss": 0.2622,
+      "step": 76650
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.529564495837683e-05,
+      "loss": 0.111,
+      "step": 76660
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.529430659278889e-05,
+      "loss": 0.2781,
+      "step": 76670
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.529296822720094e-05,
+      "loss": 0.1324,
+      "step": 76680
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.5291629861613e-05,
+      "loss": 0.2417,
+      "step": 76690
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.529029149602505e-05,
+      "loss": 0.1709,
+      "step": 76700
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.528895313043711e-05,
+      "loss": 0.0534,
+      "step": 76710
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.528761476484917e-05,
+      "loss": 0.0975,
+      "step": 76720
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.5286276399261226e-05,
+      "loss": 0.0577,
+      "step": 76730
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.528493803367328e-05,
+      "loss": 0.1205,
+      "step": 76740
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.5283599668085337e-05,
+      "loss": 0.3046,
+      "step": 76750
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.528226130249739e-05,
+      "loss": 0.3036,
+      "step": 76760
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.528092293690945e-05,
+      "loss": 0.2492,
+      "step": 76770
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.52795845713215e-05,
+      "loss": 0.2148,
+      "step": 76780
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.527824620573356e-05,
+      "loss": 0.2018,
+      "step": 76790
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.527690784014561e-05,
+      "loss": 0.1237,
+      "step": 76800
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.5275569474557675e-05,
+      "loss": 0.2647,
+      "step": 76810
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.527423110896973e-05,
+      "loss": 0.1516,
+      "step": 76820
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.5272892743381786e-05,
+      "loss": 0.3502,
+      "step": 76830
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.527155437779384e-05,
+      "loss": 0.1159,
+      "step": 76840
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.52702160122059e-05,
+      "loss": 0.2812,
+      "step": 76850
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.526887764661795e-05,
+      "loss": 0.1854,
+      "step": 76860
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.526753928103001e-05,
+      "loss": 0.1311,
+      "step": 76870
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.526620091544206e-05,
+      "loss": 0.085,
+      "step": 76880
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.526486254985412e-05,
+      "loss": 0.3039,
+      "step": 76890
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.526352418426617e-05,
+      "loss": 0.0802,
+      "step": 76900
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.5262185818678235e-05,
+      "loss": 0.1481,
+      "step": 76910
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.526084745309029e-05,
+      "loss": 0.1579,
+      "step": 76920
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.5259509087502346e-05,
+      "loss": 0.1257,
+      "step": 76930
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.52581707219144e-05,
+      "loss": 0.1024,
+      "step": 76940
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.525683235632646e-05,
+      "loss": 0.2152,
+      "step": 76950
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.525549399073851e-05,
+      "loss": 0.0063,
+      "step": 76960
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.525415562515057e-05,
+      "loss": 0.2347,
+      "step": 76970
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.525281725956262e-05,
+      "loss": 0.0425,
+      "step": 76980
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.525147889397468e-05,
+      "loss": 0.2864,
+      "step": 76990
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.5250140528386734e-05,
+      "loss": 0.1032,
+      "step": 77000
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.5248802162798796e-05,
+      "loss": 0.1187,
+      "step": 77010
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.524746379721085e-05,
+      "loss": 0.1446,
+      "step": 77020
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.5246125431622906e-05,
+      "loss": 0.2052,
+      "step": 77030
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.524478706603496e-05,
+      "loss": 0.2113,
+      "step": 77040
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.524344870044702e-05,
+      "loss": 0.1848,
+      "step": 77050
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.524211033485907e-05,
+      "loss": 0.3464,
+      "step": 77060
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.524077196927113e-05,
+      "loss": 0.0525,
+      "step": 77070
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.523943360368318e-05,
+      "loss": 0.101,
+      "step": 77080
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.523809523809524e-05,
+      "loss": 0.031,
+      "step": 77090
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.5236756872507294e-05,
+      "loss": 0.0844,
+      "step": 77100
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.5235418506919356e-05,
+      "loss": 0.1846,
+      "step": 77110
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.523408014133141e-05,
+      "loss": 0.0952,
+      "step": 77120
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.5232741775743466e-05,
+      "loss": 0.2437,
+      "step": 77130
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.523140341015552e-05,
+      "loss": 0.1246,
+      "step": 77140
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.523006504456758e-05,
+      "loss": 0.1286,
+      "step": 77150
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.522872667897963e-05,
+      "loss": 0.2218,
+      "step": 77160
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.522738831339169e-05,
+      "loss": 0.0588,
+      "step": 77170
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.522604994780374e-05,
+      "loss": 0.2372,
+      "step": 77180
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.52247115822158e-05,
+      "loss": 0.352,
+      "step": 77190
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.5223373216627854e-05,
+      "loss": 0.1469,
+      "step": 77200
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.5222034851039916e-05,
+      "loss": 0.1587,
+      "step": 77210
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.522069648545197e-05,
+      "loss": 0.1141,
+      "step": 77220
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.5219358119864027e-05,
+      "loss": 0.1418,
+      "step": 77230
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.521801975427608e-05,
+      "loss": 0.118,
+      "step": 77240
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.521668138868814e-05,
+      "loss": 0.2391,
+      "step": 77250
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.521534302310019e-05,
+      "loss": 0.2736,
+      "step": 77260
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.521400465751225e-05,
+      "loss": 0.1063,
+      "step": 77270
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.52126662919243e-05,
+      "loss": 0.4188,
+      "step": 77280
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.521132792633636e-05,
+      "loss": 0.1934,
+      "step": 77290
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.5209989560748414e-05,
+      "loss": 0.301,
+      "step": 77300
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.5208651195160476e-05,
+      "loss": 0.1953,
+      "step": 77310
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.520731282957253e-05,
+      "loss": 0.0535,
+      "step": 77320
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.520597446398459e-05,
+      "loss": 0.0939,
+      "step": 77330
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.520463609839664e-05,
+      "loss": 0.1057,
+      "step": 77340
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.52032977328087e-05,
+      "loss": 0.1959,
+      "step": 77350
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.520195936722075e-05,
+      "loss": 0.2501,
+      "step": 77360
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.520062100163281e-05,
+      "loss": 0.1466,
+      "step": 77370
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.5199282636044863e-05,
+      "loss": 0.1464,
+      "step": 77380
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.519794427045692e-05,
+      "loss": 0.0931,
+      "step": 77390
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.5196605904868974e-05,
+      "loss": 0.14,
+      "step": 77400
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.5195267539281036e-05,
+      "loss": 0.2312,
+      "step": 77410
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.519392917369309e-05,
+      "loss": 0.0686,
+      "step": 77420
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.519259080810515e-05,
+      "loss": 0.3221,
+      "step": 77430
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.51912524425172e-05,
+      "loss": 0.0134,
+      "step": 77440
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.518991407692926e-05,
+      "loss": 0.118,
+      "step": 77450
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.518857571134131e-05,
+      "loss": 0.236,
+      "step": 77460
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.518723734575337e-05,
+      "loss": 0.1449,
+      "step": 77470
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.5185898980165424e-05,
+      "loss": 0.0528,
+      "step": 77480
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.518456061457748e-05,
+      "loss": 0.1129,
+      "step": 77490
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.5183222248989534e-05,
+      "loss": 0.2563,
+      "step": 77500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.5181883883401596e-05,
+      "loss": 0.2779,
+      "step": 77510
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.518054551781365e-05,
+      "loss": 0.1851,
+      "step": 77520
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.517920715222571e-05,
+      "loss": 0.1515,
+      "step": 77530
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.517786878663776e-05,
+      "loss": 0.1103,
+      "step": 77540
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.517653042104982e-05,
+      "loss": 0.0361,
+      "step": 77550
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.517519205546187e-05,
+      "loss": 0.1688,
+      "step": 77560
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.517385368987393e-05,
+      "loss": 0.0736,
+      "step": 77570
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.5172515324285984e-05,
+      "loss": 0.0887,
+      "step": 77580
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.517117695869804e-05,
+      "loss": 0.126,
+      "step": 77590
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.5169838593110094e-05,
+      "loss": 0.3644,
+      "step": 77600
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.5168500227522157e-05,
+      "loss": 0.1354,
+      "step": 77610
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.516716186193421e-05,
+      "loss": 0.1036,
+      "step": 77620
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.516582349634627e-05,
+      "loss": 0.1996,
+      "step": 77630
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.516448513075832e-05,
+      "loss": 0.2186,
+      "step": 77640
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.516314676517038e-05,
+      "loss": 0.1006,
+      "step": 77650
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.516180839958243e-05,
+      "loss": 0.1933,
+      "step": 77660
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.516047003399449e-05,
+      "loss": 0.0922,
+      "step": 77670
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.5159131668406544e-05,
+      "loss": 0.0884,
+      "step": 77680
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.51577933028186e-05,
+      "loss": 0.201,
+      "step": 77690
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.5156454937230655e-05,
+      "loss": 0.1273,
+      "step": 77700
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.515511657164272e-05,
+      "loss": 0.1638,
+      "step": 77710
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.515377820605477e-05,
+      "loss": 0.1945,
+      "step": 77720
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.515243984046683e-05,
+      "loss": 0.1578,
+      "step": 77730
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.515110147487888e-05,
+      "loss": 0.3669,
+      "step": 77740
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.514976310929094e-05,
+      "loss": 0.0975,
+      "step": 77750
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.514842474370299e-05,
+      "loss": 0.2065,
+      "step": 77760
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.514708637811505e-05,
+      "loss": 0.1175,
+      "step": 77770
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.5145748012527104e-05,
+      "loss": 0.1295,
+      "step": 77780
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.514440964693916e-05,
+      "loss": 0.2666,
+      "step": 77790
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.5143071281351215e-05,
+      "loss": 0.1654,
+      "step": 77800
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.514173291576328e-05,
+      "loss": 0.1909,
+      "step": 77810
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.514039455017533e-05,
+      "loss": 0.1716,
+      "step": 77820
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.513905618458739e-05,
+      "loss": 0.2488,
+      "step": 77830
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.513771781899944e-05,
+      "loss": 0.2168,
+      "step": 77840
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.51363794534115e-05,
+      "loss": 0.1113,
+      "step": 77850
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.5135041087823554e-05,
+      "loss": 0.118,
+      "step": 77860
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.513370272223561e-05,
+      "loss": 0.0504,
+      "step": 77870
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.5132364356647664e-05,
+      "loss": 0.1181,
+      "step": 77880
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.513102599105972e-05,
+      "loss": 0.1754,
+      "step": 77890
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.5129687625471775e-05,
+      "loss": 0.1145,
+      "step": 77900
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.512834925988384e-05,
+      "loss": 0.2884,
+      "step": 77910
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.512701089429589e-05,
+      "loss": 0.2328,
+      "step": 77920
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.512567252870795e-05,
+      "loss": 0.3231,
+      "step": 77930
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.512433416312e-05,
+      "loss": 0.1214,
+      "step": 77940
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.512299579753206e-05,
+      "loss": 0.0923,
+      "step": 77950
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.5121657431944114e-05,
+      "loss": 0.0313,
+      "step": 77960
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.512031906635617e-05,
+      "loss": 0.1575,
+      "step": 77970
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.5118980700768224e-05,
+      "loss": 0.1505,
+      "step": 77980
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.511764233518028e-05,
+      "loss": 0.0976,
+      "step": 77990
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.5116303969592335e-05,
+      "loss": 0.0429,
+      "step": 78000
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.51149656040044e-05,
+      "loss": 0.1687,
+      "step": 78010
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.511362723841645e-05,
+      "loss": 0.0455,
+      "step": 78020
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.511228887282851e-05,
+      "loss": 0.1782,
+      "step": 78030
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.511095050724056e-05,
+      "loss": 0.0962,
+      "step": 78040
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.510961214165262e-05,
+      "loss": 0.0645,
+      "step": 78050
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.5108273776064674e-05,
+      "loss": 0.1275,
+      "step": 78060
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.510693541047673e-05,
+      "loss": 0.1846,
+      "step": 78070
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.5105597044888785e-05,
+      "loss": 0.1698,
+      "step": 78080
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.510425867930084e-05,
+      "loss": 0.2418,
+      "step": 78090
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.51029203137129e-05,
+      "loss": 0.2469,
+      "step": 78100
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.510158194812496e-05,
+      "loss": 0.1939,
+      "step": 78110
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.510024358253701e-05,
+      "loss": 0.1824,
+      "step": 78120
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.509890521694907e-05,
+      "loss": 0.1612,
+      "step": 78130
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.509756685136112e-05,
+      "loss": 0.1408,
+      "step": 78140
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.509622848577318e-05,
+      "loss": 0.1689,
+      "step": 78150
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.509489012018523e-05,
+      "loss": 0.1275,
+      "step": 78160
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.509355175459728e-05,
+      "loss": 0.167,
+      "step": 78170
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.509221338900934e-05,
+      "loss": 0.1169,
+      "step": 78180
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.50908750234214e-05,
+      "loss": 0.2017,
+      "step": 78190
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.5089536657833455e-05,
+      "loss": 0.2163,
+      "step": 78200
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.508819829224551e-05,
+      "loss": 0.0415,
+      "step": 78210
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.5086859926657566e-05,
+      "loss": 0.0198,
+      "step": 78220
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.508552156106962e-05,
+      "loss": 0.1197,
+      "step": 78230
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.508418319548168e-05,
+      "loss": 0.1429,
+      "step": 78240
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.508284482989373e-05,
+      "loss": 0.1328,
+      "step": 78250
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.508150646430579e-05,
+      "loss": 0.3347,
+      "step": 78260
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.508016809871784e-05,
+      "loss": 0.327,
+      "step": 78270
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.50788297331299e-05,
+      "loss": 0.0456,
+      "step": 78280
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.507749136754196e-05,
+      "loss": 0.1885,
+      "step": 78290
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.5076153001954015e-05,
+      "loss": 0.112,
+      "step": 78300
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.507481463636607e-05,
+      "loss": 0.1392,
+      "step": 78310
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.5073476270778126e-05,
+      "loss": 0.0943,
+      "step": 78320
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.507213790519018e-05,
+      "loss": 0.0934,
+      "step": 78330
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.507079953960224e-05,
+      "loss": 0.0813,
+      "step": 78340
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.506946117401429e-05,
+      "loss": 0.1701,
+      "step": 78350
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.506812280842635e-05,
+      "loss": 0.1481,
+      "step": 78360
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.50667844428384e-05,
+      "loss": 0.2738,
+      "step": 78370
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.5065446077250465e-05,
+      "loss": 0.2128,
+      "step": 78380
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.506410771166252e-05,
+      "loss": 0.0731,
+      "step": 78390
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.5062769346074576e-05,
+      "loss": 0.1488,
+      "step": 78400
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.506143098048663e-05,
+      "loss": 0.1989,
+      "step": 78410
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.5060092614898686e-05,
+      "loss": 0.2111,
+      "step": 78420
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.505875424931074e-05,
+      "loss": 0.2528,
+      "step": 78430
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.50574158837228e-05,
+      "loss": 0.1293,
+      "step": 78440
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.505607751813485e-05,
+      "loss": 0.279,
+      "step": 78450
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.505473915254691e-05,
+      "loss": 0.0682,
+      "step": 78460
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.505340078695896e-05,
+      "loss": 0.2166,
+      "step": 78470
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.5052062421371025e-05,
+      "loss": 0.2633,
+      "step": 78480
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.505072405578308e-05,
+      "loss": 0.1647,
+      "step": 78490
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.5049385690195136e-05,
+      "loss": 0.0541,
+      "step": 78500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.504804732460719e-05,
+      "loss": 0.2115,
+      "step": 78510
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.5046708959019246e-05,
+      "loss": 0.1087,
+      "step": 78520
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.50453705934313e-05,
+      "loss": 0.1916,
+      "step": 78530
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.504403222784336e-05,
+      "loss": 0.2252,
+      "step": 78540
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.504269386225541e-05,
+      "loss": 0.1643,
+      "step": 78550
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.504135549666747e-05,
+      "loss": 0.0921,
+      "step": 78560
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.504001713107952e-05,
+      "loss": 0.0996,
+      "step": 78570
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.5038678765491585e-05,
+      "loss": 0.2089,
+      "step": 78580
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.503734039990364e-05,
+      "loss": 0.1095,
+      "step": 78590
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.5036002034315696e-05,
+      "loss": 0.0162,
+      "step": 78600
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.503466366872775e-05,
+      "loss": 0.0497,
+      "step": 78610
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.503332530313981e-05,
+      "loss": 0.1048,
+      "step": 78620
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.503198693755186e-05,
+      "loss": 0.0897,
+      "step": 78630
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.503064857196392e-05,
+      "loss": 0.0697,
+      "step": 78640
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.502931020637597e-05,
+      "loss": 0.062,
+      "step": 78650
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.502797184078803e-05,
+      "loss": 0.1771,
+      "step": 78660
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.502663347520008e-05,
+      "loss": 0.1198,
+      "step": 78670
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.5025295109612145e-05,
+      "loss": 0.029,
+      "step": 78680
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.50239567440242e-05,
+      "loss": 0.1635,
+      "step": 78690
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.5022618378436256e-05,
+      "loss": 0.0915,
+      "step": 78700
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.502128001284831e-05,
+      "loss": 0.1435,
+      "step": 78710
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.501994164726037e-05,
+      "loss": 0.1587,
+      "step": 78720
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.501860328167242e-05,
+      "loss": 0.2237,
+      "step": 78730
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.501726491608448e-05,
+      "loss": 0.1105,
+      "step": 78740
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.501592655049653e-05,
+      "loss": 0.2237,
+      "step": 78750
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.501458818490859e-05,
+      "loss": 0.0174,
+      "step": 78760
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.5013249819320643e-05,
+      "loss": 0.0186,
+      "step": 78770
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.5011911453732706e-05,
+      "loss": 0.0829,
+      "step": 78780
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.501057308814476e-05,
+      "loss": 0.2998,
+      "step": 78790
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.5009234722556816e-05,
+      "loss": 0.1296,
+      "step": 78800
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.500789635696887e-05,
+      "loss": 0.1671,
+      "step": 78810
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.500655799138093e-05,
+      "loss": 0.1015,
+      "step": 78820
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.500521962579298e-05,
+      "loss": 0.2416,
+      "step": 78830
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.500388126020504e-05,
+      "loss": 0.1676,
+      "step": 78840
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.500254289461709e-05,
+      "loss": 0.0284,
+      "step": 78850
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.500120452902915e-05,
+      "loss": 0.0038,
+      "step": 78860
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.4999866163441204e-05,
+      "loss": 0.2672,
+      "step": 78870
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.4998527797853266e-05,
+      "loss": 0.2291,
+      "step": 78880
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.499718943226532e-05,
+      "loss": 0.1221,
+      "step": 78890
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.4995851066677376e-05,
+      "loss": 0.0948,
+      "step": 78900
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.499451270108943e-05,
+      "loss": 0.3149,
+      "step": 78910
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.499317433550149e-05,
+      "loss": 0.0678,
+      "step": 78920
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.499183596991354e-05,
+      "loss": 0.0987,
+      "step": 78930
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.49904976043256e-05,
+      "loss": 0.1982,
+      "step": 78940
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.498915923873765e-05,
+      "loss": 0.1954,
+      "step": 78950
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.498782087314971e-05,
+      "loss": 0.1571,
+      "step": 78960
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.4986482507561764e-05,
+      "loss": 0.2383,
+      "step": 78970
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.4985144141973826e-05,
+      "loss": 0.1413,
+      "step": 78980
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.498380577638588e-05,
+      "loss": 0.0336,
+      "step": 78990
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.4982467410797937e-05,
+      "loss": 0.1468,
+      "step": 79000
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.498112904520999e-05,
+      "loss": 0.1934,
+      "step": 79010
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.497979067962205e-05,
+      "loss": 0.2263,
+      "step": 79020
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.49784523140341e-05,
+      "loss": 0.0937,
+      "step": 79030
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.497711394844616e-05,
+      "loss": 0.0352,
+      "step": 79040
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.497577558285821e-05,
+      "loss": 0.1779,
+      "step": 79050
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.497443721727027e-05,
+      "loss": 0.1879,
+      "step": 79060
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.4973098851682324e-05,
+      "loss": 0.0962,
+      "step": 79070
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.4971760486094386e-05,
+      "loss": 0.1314,
+      "step": 79080
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.497042212050644e-05,
+      "loss": 0.2494,
+      "step": 79090
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.49690837549185e-05,
+      "loss": 0.2767,
+      "step": 79100
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.496774538933055e-05,
+      "loss": 0.2309,
+      "step": 79110
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.496640702374261e-05,
+      "loss": 0.2736,
+      "step": 79120
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.496506865815466e-05,
+      "loss": 0.1513,
+      "step": 79130
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.496373029256672e-05,
+      "loss": 0.26,
+      "step": 79140
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.4962391926978773e-05,
+      "loss": 0.1661,
+      "step": 79150
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.496105356139083e-05,
+      "loss": 0.0555,
+      "step": 79160
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.4959715195802884e-05,
+      "loss": 0.0248,
+      "step": 79170
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.4958376830214946e-05,
+      "loss": 0.0661,
+      "step": 79180
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.4957038464627e-05,
+      "loss": 0.3595,
+      "step": 79190
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.495570009903906e-05,
+      "loss": 0.2398,
+      "step": 79200
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.495436173345111e-05,
+      "loss": 0.2391,
+      "step": 79210
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.495302336786317e-05,
+      "loss": 0.0116,
+      "step": 79220
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.495168500227522e-05,
+      "loss": 0.0878,
+      "step": 79230
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.495034663668728e-05,
+      "loss": 0.0485,
+      "step": 79240
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.4949008271099334e-05,
+      "loss": 0.3616,
+      "step": 79250
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.494766990551139e-05,
+      "loss": 0.2126,
+      "step": 79260
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.4946331539923444e-05,
+      "loss": 0.1924,
+      "step": 79270
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.4944993174335506e-05,
+      "loss": 0.0526,
+      "step": 79280
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.494365480874756e-05,
+      "loss": 0.1963,
+      "step": 79290
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.494231644315962e-05,
+      "loss": 0.4718,
+      "step": 79300
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.494097807757167e-05,
+      "loss": 0.109,
+      "step": 79310
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.493963971198373e-05,
+      "loss": 0.1961,
+      "step": 79320
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.493830134639578e-05,
+      "loss": 0.3403,
+      "step": 79330
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.493696298080784e-05,
+      "loss": 0.096,
+      "step": 79340
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.4935624615219894e-05,
+      "loss": 0.2306,
+      "step": 79350
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.493428624963195e-05,
+      "loss": 0.0971,
+      "step": 79360
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.4932947884044004e-05,
+      "loss": 0.1585,
+      "step": 79370
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.4931609518456067e-05,
+      "loss": 0.0905,
+      "step": 79380
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.493027115286812e-05,
+      "loss": 0.1797,
+      "step": 79390
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.492893278728018e-05,
+      "loss": 0.1553,
+      "step": 79400
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.492759442169223e-05,
+      "loss": 0.2056,
+      "step": 79410
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.492625605610429e-05,
+      "loss": 0.1207,
+      "step": 79420
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.492491769051634e-05,
+      "loss": 0.1289,
+      "step": 79430
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.49235793249284e-05,
+      "loss": 0.1728,
+      "step": 79440
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.4922240959340454e-05,
+      "loss": 0.1315,
+      "step": 79450
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.492090259375251e-05,
+      "loss": 0.2242,
+      "step": 79460
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.4919564228164565e-05,
+      "loss": 0.2406,
+      "step": 79470
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.491822586257663e-05,
+      "loss": 0.2404,
+      "step": 79480
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.491688749698868e-05,
+      "loss": 0.1881,
+      "step": 79490
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.491554913140074e-05,
+      "loss": 0.0561,
+      "step": 79500
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.491421076581279e-05,
+      "loss": 0.0367,
+      "step": 79510
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.491287240022485e-05,
+      "loss": 0.0709,
+      "step": 79520
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.49115340346369e-05,
+      "loss": 0.2323,
+      "step": 79530
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.491019566904896e-05,
+      "loss": 0.2803,
+      "step": 79540
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.4908857303461014e-05,
+      "loss": 0.1801,
+      "step": 79550
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.490751893787307e-05,
+      "loss": 0.1255,
+      "step": 79560
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.490618057228513e-05,
+      "loss": 0.3588,
+      "step": 79570
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.490484220669719e-05,
+      "loss": 0.0124,
+      "step": 79580
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.490350384110924e-05,
+      "loss": 0.1977,
+      "step": 79590
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.49021654755213e-05,
+      "loss": 0.0952,
+      "step": 79600
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.490082710993335e-05,
+      "loss": 0.0976,
+      "step": 79610
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.489948874434541e-05,
+      "loss": 0.2524,
+      "step": 79620
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.4898150378757463e-05,
+      "loss": 0.1745,
+      "step": 79630
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.489681201316952e-05,
+      "loss": 0.195,
+      "step": 79640
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.4895473647581574e-05,
+      "loss": 0.1104,
+      "step": 79650
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.489413528199363e-05,
+      "loss": 0.1104,
+      "step": 79660
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.489279691640569e-05,
+      "loss": 0.1153,
+      "step": 79670
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.489145855081775e-05,
+      "loss": 0.113,
+      "step": 79680
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.48901201852298e-05,
+      "loss": 0.1954,
+      "step": 79690
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.488878181964186e-05,
+      "loss": 0.2773,
+      "step": 79700
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.488744345405391e-05,
+      "loss": 0.2909,
+      "step": 79710
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.488610508846597e-05,
+      "loss": 0.1449,
+      "step": 79720
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.4884766722878024e-05,
+      "loss": 0.091,
+      "step": 79730
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.488342835729008e-05,
+      "loss": 0.1473,
+      "step": 79740
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.4882089991702134e-05,
+      "loss": 0.0443,
+      "step": 79750
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.488075162611419e-05,
+      "loss": 0.183,
+      "step": 79760
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.487941326052625e-05,
+      "loss": 0.1011,
+      "step": 79770
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.487807489493831e-05,
+      "loss": 0.0957,
+      "step": 79780
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.487673652935036e-05,
+      "loss": 0.1535,
+      "step": 79790
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.487539816376242e-05,
+      "loss": 0.0882,
+      "step": 79800
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.487405979817447e-05,
+      "loss": 0.1397,
+      "step": 79810
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.487272143258653e-05,
+      "loss": 0.1138,
+      "step": 79820
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.4871383066998584e-05,
+      "loss": 0.1584,
+      "step": 79830
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.487004470141064e-05,
+      "loss": 0.0759,
+      "step": 79840
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.4868706335822694e-05,
+      "loss": 0.194,
+      "step": 79850
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.486736797023475e-05,
+      "loss": 0.1091,
+      "step": 79860
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.486602960464681e-05,
+      "loss": 0.1018,
+      "step": 79870
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.486469123905887e-05,
+      "loss": 0.1616,
+      "step": 79880
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.486335287347092e-05,
+      "loss": 0.122,
+      "step": 79890
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.486201450788298e-05,
+      "loss": 0.3423,
+      "step": 79900
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.486067614229503e-05,
+      "loss": 0.0429,
+      "step": 79910
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.485933777670709e-05,
+      "loss": 0.1058,
+      "step": 79920
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.4857999411119144e-05,
+      "loss": 0.1017,
+      "step": 79930
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.48566610455312e-05,
+      "loss": 0.2013,
+      "step": 79940
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.4855322679943255e-05,
+      "loss": 0.5023,
+      "step": 79950
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.485398431435531e-05,
+      "loss": 0.0237,
+      "step": 79960
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.485264594876737e-05,
+      "loss": 0.226,
+      "step": 79970
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.485130758317943e-05,
+      "loss": 0.0872,
+      "step": 79980
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.484996921759148e-05,
+      "loss": 0.0111,
+      "step": 79990
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.484863085200354e-05,
+      "loss": 0.0404,
+      "step": 80000
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.4847292486415593e-05,
+      "loss": 0.1276,
+      "step": 80010
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.484595412082765e-05,
+      "loss": 0.3996,
+      "step": 80020
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.4844615755239704e-05,
+      "loss": 0.2176,
+      "step": 80030
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.484327738965176e-05,
+      "loss": 0.0344,
+      "step": 80040
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.4841939024063815e-05,
+      "loss": 0.052,
+      "step": 80050
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.484060065847587e-05,
+      "loss": 0.1076,
+      "step": 80060
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.483926229288793e-05,
+      "loss": 0.1704,
+      "step": 80070
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.483792392729999e-05,
+      "loss": 0.3202,
+      "step": 80080
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.483658556171204e-05,
+      "loss": 0.3043,
+      "step": 80090
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.48352471961241e-05,
+      "loss": 0.2803,
+      "step": 80100
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.4833908830536154e-05,
+      "loss": 0.1473,
+      "step": 80110
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.483257046494821e-05,
+      "loss": 0.1776,
+      "step": 80120
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.4831232099360264e-05,
+      "loss": 0.1079,
+      "step": 80130
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.482989373377232e-05,
+      "loss": 0.4137,
+      "step": 80140
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.4828555368184375e-05,
+      "loss": 0.1088,
+      "step": 80150
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.482721700259643e-05,
+      "loss": 0.1024,
+      "step": 80160
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.482587863700849e-05,
+      "loss": 0.2363,
+      "step": 80170
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.482454027142055e-05,
+      "loss": 0.2161,
+      "step": 80180
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.48232019058326e-05,
+      "loss": 0.066,
+      "step": 80190
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.482186354024466e-05,
+      "loss": 0.1457,
+      "step": 80200
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.4820525174656714e-05,
+      "loss": 0.2017,
+      "step": 80210
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.481918680906877e-05,
+      "loss": 0.2849,
+      "step": 80220
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.4817848443480824e-05,
+      "loss": 0.4569,
+      "step": 80230
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.481651007789288e-05,
+      "loss": 0.1161,
+      "step": 80240
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.4815171712304935e-05,
+      "loss": 0.0713,
+      "step": 80250
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.481383334671699e-05,
+      "loss": 0.1128,
+      "step": 80260
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.481249498112905e-05,
+      "loss": 0.2751,
+      "step": 80270
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.481115661554111e-05,
+      "loss": 0.2714,
+      "step": 80280
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.480981824995316e-05,
+      "loss": 0.2622,
+      "step": 80290
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.480847988436522e-05,
+      "loss": 0.0966,
+      "step": 80300
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.4807141518777274e-05,
+      "loss": 0.1568,
+      "step": 80310
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.480580315318933e-05,
+      "loss": 0.1348,
+      "step": 80320
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.4804464787601385e-05,
+      "loss": 0.1046,
+      "step": 80330
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.480312642201344e-05,
+      "loss": 0.2769,
+      "step": 80340
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.4801788056425495e-05,
+      "loss": 0.0728,
+      "step": 80350
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.480044969083755e-05,
+      "loss": 0.0869,
+      "step": 80360
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.479911132524961e-05,
+      "loss": 0.0877,
+      "step": 80370
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.479777295966167e-05,
+      "loss": 0.1073,
+      "step": 80380
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.479643459407372e-05,
+      "loss": 0.1129,
+      "step": 80390
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.479509622848578e-05,
+      "loss": 0.3328,
+      "step": 80400
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.4793757862897834e-05,
+      "loss": 0.1122,
+      "step": 80410
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.479241949730989e-05,
+      "loss": 0.1318,
+      "step": 80420
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.4791081131721945e-05,
+      "loss": 0.1463,
+      "step": 80430
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.4789742766134e-05,
+      "loss": 0.2914,
+      "step": 80440
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.4788404400546055e-05,
+      "loss": 0.2748,
+      "step": 80450
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.478706603495811e-05,
+      "loss": 0.1118,
+      "step": 80460
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.478572766937017e-05,
+      "loss": 0.0731,
+      "step": 80470
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.478438930378223e-05,
+      "loss": 0.1367,
+      "step": 80480
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.4783050938194284e-05,
+      "loss": 0.264,
+      "step": 80490
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.478171257260634e-05,
+      "loss": 0.2722,
+      "step": 80500
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.4780374207018394e-05,
+      "loss": 0.1331,
+      "step": 80510
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.477903584143045e-05,
+      "loss": 0.0108,
+      "step": 80520
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.47776974758425e-05,
+      "loss": 0.2487,
+      "step": 80530
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.4776359110254553e-05,
+      "loss": 0.1644,
+      "step": 80540
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.4775020744666616e-05,
+      "loss": 0.0919,
+      "step": 80550
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.477368237907867e-05,
+      "loss": 0.1573,
+      "step": 80560
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.4772344013490726e-05,
+      "loss": 0.1711,
+      "step": 80570
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.477100564790278e-05,
+      "loss": 0.2368,
+      "step": 80580
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.476966728231484e-05,
+      "loss": 0.2217,
+      "step": 80590
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.476832891672689e-05,
+      "loss": 0.2854,
+      "step": 80600
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.476699055113895e-05,
+      "loss": 0.1211,
+      "step": 80610
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.4765652185551e-05,
+      "loss": 0.0537,
+      "step": 80620
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.476431381996306e-05,
+      "loss": 0.1442,
+      "step": 80630
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.4762975454375114e-05,
+      "loss": 0.0869,
+      "step": 80640
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.4761637088787176e-05,
+      "loss": 0.1006,
+      "step": 80650
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.476029872319923e-05,
+      "loss": 0.0361,
+      "step": 80660
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.4758960357611286e-05,
+      "loss": 0.1008,
+      "step": 80670
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.475762199202334e-05,
+      "loss": 0.1054,
+      "step": 80680
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.47562836264354e-05,
+      "loss": 0.2675,
+      "step": 80690
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.475494526084745e-05,
+      "loss": 0.3305,
+      "step": 80700
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.475360689525951e-05,
+      "loss": 0.2267,
+      "step": 80710
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.475226852967156e-05,
+      "loss": 0.1942,
+      "step": 80720
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.475093016408362e-05,
+      "loss": 0.2248,
+      "step": 80730
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.4749591798495674e-05,
+      "loss": 0.0943,
+      "step": 80740
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.4748253432907736e-05,
+      "loss": 0.1136,
+      "step": 80750
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.474691506731979e-05,
+      "loss": 0.0934,
+      "step": 80760
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.4745576701731847e-05,
+      "loss": 0.2176,
+      "step": 80770
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.47442383361439e-05,
+      "loss": 0.3002,
+      "step": 80780
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.474289997055596e-05,
+      "loss": 0.1082,
+      "step": 80790
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.474156160496801e-05,
+      "loss": 0.0953,
+      "step": 80800
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.474022323938007e-05,
+      "loss": 0.161,
+      "step": 80810
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.473888487379212e-05,
+      "loss": 0.1481,
+      "step": 80820
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.473754650820418e-05,
+      "loss": 0.1395,
+      "step": 80830
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.4736208142616234e-05,
+      "loss": 0.3138,
+      "step": 80840
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.4734869777028296e-05,
+      "loss": 0.3089,
+      "step": 80850
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.473353141144035e-05,
+      "loss": 0.0206,
+      "step": 80860
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.473219304585241e-05,
+      "loss": 0.1529,
+      "step": 80870
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.473085468026446e-05,
+      "loss": 0.0653,
+      "step": 80880
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.472951631467652e-05,
+      "loss": 0.1446,
+      "step": 80890
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.472817794908857e-05,
+      "loss": 0.2392,
+      "step": 80900
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.472683958350063e-05,
+      "loss": 0.1132,
+      "step": 80910
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.472550121791268e-05,
+      "loss": 0.148,
+      "step": 80920
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.472416285232474e-05,
+      "loss": 0.1937,
+      "step": 80930
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.4722824486736794e-05,
+      "loss": 0.0091,
+      "step": 80940
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.4721486121148856e-05,
+      "loss": 0.2314,
+      "step": 80950
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.472014775556091e-05,
+      "loss": 0.1714,
+      "step": 80960
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.471880938997297e-05,
+      "loss": 0.055,
+      "step": 80970
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.471747102438502e-05,
+      "loss": 0.1856,
+      "step": 80980
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.471613265879708e-05,
+      "loss": 0.1867,
+      "step": 80990
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.471479429320913e-05,
+      "loss": 0.1785,
+      "step": 81000
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.471345592762119e-05,
+      "loss": 0.0761,
+      "step": 81010
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.4712117562033244e-05,
+      "loss": 0.2285,
+      "step": 81020
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.47107791964453e-05,
+      "loss": 0.055,
+      "step": 81030
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.4709440830857354e-05,
+      "loss": 0.1111,
+      "step": 81040
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.4708102465269416e-05,
+      "loss": 0.1462,
+      "step": 81050
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.470676409968147e-05,
+      "loss": 0.1542,
+      "step": 81060
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.470542573409353e-05,
+      "loss": 0.0494,
+      "step": 81070
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.470408736850558e-05,
+      "loss": 0.1005,
+      "step": 81080
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.470274900291764e-05,
+      "loss": 0.1383,
+      "step": 81090
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.470141063732969e-05,
+      "loss": 0.4354,
+      "step": 81100
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.470007227174175e-05,
+      "loss": 0.0949,
+      "step": 81110
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.4698733906153804e-05,
+      "loss": 0.0166,
+      "step": 81120
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.469739554056586e-05,
+      "loss": 0.1247,
+      "step": 81130
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.469605717497792e-05,
+      "loss": 0.1531,
+      "step": 81140
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.4694718809389976e-05,
+      "loss": 0.1026,
+      "step": 81150
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.469338044380203e-05,
+      "loss": 0.1316,
+      "step": 81160
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.469204207821409e-05,
+      "loss": 0.124,
+      "step": 81170
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.469070371262614e-05,
+      "loss": 0.1032,
+      "step": 81180
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.46893653470382e-05,
+      "loss": 0.0808,
+      "step": 81190
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.468802698145025e-05,
+      "loss": 0.3321,
+      "step": 81200
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.468668861586231e-05,
+      "loss": 0.0884,
+      "step": 81210
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.4685350250274364e-05,
+      "loss": 0.1784,
+      "step": 81220
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.468401188468642e-05,
+      "loss": 0.0852,
+      "step": 81230
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.468267351909848e-05,
+      "loss": 0.0879,
+      "step": 81240
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.468133515351054e-05,
+      "loss": 0.1058,
+      "step": 81250
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.467999678792259e-05,
+      "loss": 0.0789,
+      "step": 81260
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.467865842233465e-05,
+      "loss": 0.1144,
+      "step": 81270
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.46773200567467e-05,
+      "loss": 0.2203,
+      "step": 81280
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.467598169115876e-05,
+      "loss": 0.2281,
+      "step": 81290
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.467464332557081e-05,
+      "loss": 0.2688,
+      "step": 81300
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.467330495998287e-05,
+      "loss": 0.1465,
+      "step": 81310
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.4671966594394924e-05,
+      "loss": 0.2361,
+      "step": 81320
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.467062822880698e-05,
+      "loss": 0.1269,
+      "step": 81330
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.466928986321904e-05,
+      "loss": 0.0929,
+      "step": 81340
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.46679514976311e-05,
+      "loss": 0.059,
+      "step": 81350
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.466661313204315e-05,
+      "loss": 0.0102,
+      "step": 81360
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.466527476645521e-05,
+      "loss": 0.0981,
+      "step": 81370
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.466393640086726e-05,
+      "loss": 0.1352,
+      "step": 81380
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.466259803527932e-05,
+      "loss": 0.0308,
+      "step": 81390
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.4661259669691373e-05,
+      "loss": 0.2146,
+      "step": 81400
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.465992130410343e-05,
+      "loss": 0.2053,
+      "step": 81410
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.4658582938515484e-05,
+      "loss": 0.2332,
+      "step": 81420
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.465724457292754e-05,
+      "loss": 0.1631,
+      "step": 81430
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.46559062073396e-05,
+      "loss": 0.0618,
+      "step": 81440
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.465456784175166e-05,
+      "loss": 0.2379,
+      "step": 81450
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.465322947616371e-05,
+      "loss": 0.2686,
+      "step": 81460
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.465189111057577e-05,
+      "loss": 0.3089,
+      "step": 81470
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.465055274498782e-05,
+      "loss": 0.2253,
+      "step": 81480
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.464921437939988e-05,
+      "loss": 0.1683,
+      "step": 81490
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.4647876013811934e-05,
+      "loss": 0.1262,
+      "step": 81500
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.464653764822399e-05,
+      "loss": 0.2736,
+      "step": 81510
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.4645199282636044e-05,
+      "loss": 0.2002,
+      "step": 81520
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.46438609170481e-05,
+      "loss": 0.0857,
+      "step": 81530
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.464252255146016e-05,
+      "loss": 0.2619,
+      "step": 81540
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.464118418587222e-05,
+      "loss": 0.2297,
+      "step": 81550
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.463984582028427e-05,
+      "loss": 0.0919,
+      "step": 81560
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.463850745469633e-05,
+      "loss": 0.0139,
+      "step": 81570
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.463716908910838e-05,
+      "loss": 0.1259,
+      "step": 81580
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.463583072352044e-05,
+      "loss": 0.1239,
+      "step": 81590
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.4634492357932494e-05,
+      "loss": 0.0954,
+      "step": 81600
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.463315399234455e-05,
+      "loss": 0.2486,
+      "step": 81610
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.4631815626756604e-05,
+      "loss": 0.2746,
+      "step": 81620
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.463047726116866e-05,
+      "loss": 0.3062,
+      "step": 81630
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.462913889558072e-05,
+      "loss": 0.1742,
+      "step": 81640
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.462780052999278e-05,
+      "loss": 0.3616,
+      "step": 81650
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.462646216440483e-05,
+      "loss": 0.1252,
+      "step": 81660
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.462512379881689e-05,
+      "loss": 0.1517,
+      "step": 81670
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.462378543322894e-05,
+      "loss": 0.2879,
+      "step": 81680
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.4622447067641e-05,
+      "loss": 0.3337,
+      "step": 81690
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.4621108702053054e-05,
+      "loss": 0.301,
+      "step": 81700
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.461977033646511e-05,
+      "loss": 0.1448,
+      "step": 81710
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.4618431970877165e-05,
+      "loss": 0.0241,
+      "step": 81720
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.461709360528922e-05,
+      "loss": 0.0978,
+      "step": 81730
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.461575523970128e-05,
+      "loss": 0.1522,
+      "step": 81740
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.461441687411334e-05,
+      "loss": 0.282,
+      "step": 81750
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.461307850852539e-05,
+      "loss": 0.2,
+      "step": 81760
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.461174014293745e-05,
+      "loss": 0.0647,
+      "step": 81770
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.4610401777349503e-05,
+      "loss": 0.2421,
+      "step": 81780
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.460906341176156e-05,
+      "loss": 0.1903,
+      "step": 81790
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.4607725046173614e-05,
+      "loss": 0.4563,
+      "step": 81800
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.460638668058567e-05,
+      "loss": 0.0743,
+      "step": 81810
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.4605048314997725e-05,
+      "loss": 0.1951,
+      "step": 81820
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.460370994940978e-05,
+      "loss": 0.0404,
+      "step": 81830
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.460237158382184e-05,
+      "loss": 0.1586,
+      "step": 81840
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.46010332182339e-05,
+      "loss": 0.1056,
+      "step": 81850
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.459969485264595e-05,
+      "loss": 0.0829,
+      "step": 81860
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.459835648705801e-05,
+      "loss": 0.2054,
+      "step": 81870
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.4597018121470064e-05,
+      "loss": 0.2645,
+      "step": 81880
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.459567975588212e-05,
+      "loss": 0.2778,
+      "step": 81890
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.4594341390294174e-05,
+      "loss": 0.1312,
+      "step": 81900
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.459300302470623e-05,
+      "loss": 0.1555,
+      "step": 81910
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.4591664659118285e-05,
+      "loss": 0.0891,
+      "step": 81920
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.459032629353034e-05,
+      "loss": 0.2141,
+      "step": 81930
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.45889879279424e-05,
+      "loss": 0.1685,
+      "step": 81940
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.458764956235446e-05,
+      "loss": 0.121,
+      "step": 81950
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.458631119676651e-05,
+      "loss": 0.2473,
+      "step": 81960
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.458497283117857e-05,
+      "loss": 0.1323,
+      "step": 81970
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.4583634465590624e-05,
+      "loss": 0.0963,
+      "step": 81980
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.458229610000268e-05,
+      "loss": 0.2483,
+      "step": 81990
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.4580957734414734e-05,
+      "loss": 0.0204,
+      "step": 82000
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.457961936882679e-05,
+      "loss": 0.1527,
+      "step": 82010
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.4578281003238845e-05,
+      "loss": 0.069,
+      "step": 82020
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.45769426376509e-05,
+      "loss": 0.2616,
+      "step": 82030
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.457560427206296e-05,
+      "loss": 0.1252,
+      "step": 82040
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.457426590647502e-05,
+      "loss": 0.1175,
+      "step": 82050
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.457292754088707e-05,
+      "loss": 0.0698,
+      "step": 82060
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.457158917529913e-05,
+      "loss": 0.1202,
+      "step": 82070
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.4570250809711184e-05,
+      "loss": 0.2893,
+      "step": 82080
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.456891244412324e-05,
+      "loss": 0.1929,
+      "step": 82090
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.4567574078535295e-05,
+      "loss": 0.2054,
+      "step": 82100
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.456623571294735e-05,
+      "loss": 0.0913,
+      "step": 82110
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.4564897347359405e-05,
+      "loss": 0.0994,
+      "step": 82120
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.456355898177146e-05,
+      "loss": 0.1567,
+      "step": 82130
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.456222061618352e-05,
+      "loss": 0.2795,
+      "step": 82140
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.456088225059558e-05,
+      "loss": 0.2348,
+      "step": 82150
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.455954388500763e-05,
+      "loss": 0.0475,
+      "step": 82160
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.455820551941969e-05,
+      "loss": 0.023,
+      "step": 82170
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.4556867153831744e-05,
+      "loss": 0.1121,
+      "step": 82180
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.45555287882438e-05,
+      "loss": 0.1037,
+      "step": 82190
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.4554190422655855e-05,
+      "loss": 0.0575,
+      "step": 82200
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.455285205706791e-05,
+      "loss": 0.1363,
+      "step": 82210
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.4551513691479965e-05,
+      "loss": 0.2205,
+      "step": 82220
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.455017532589202e-05,
+      "loss": 0.1401,
+      "step": 82230
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.454883696030408e-05,
+      "loss": 0.2354,
+      "step": 82240
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.454749859471614e-05,
+      "loss": 0.1093,
+      "step": 82250
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.4546160229128193e-05,
+      "loss": 0.2054,
+      "step": 82260
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.454482186354025e-05,
+      "loss": 0.0697,
+      "step": 82270
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.4543483497952304e-05,
+      "loss": 0.1023,
+      "step": 82280
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.454214513236436e-05,
+      "loss": 0.1544,
+      "step": 82290
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.4540806766776415e-05,
+      "loss": 0.2874,
+      "step": 82300
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.453946840118847e-05,
+      "loss": 0.216,
+      "step": 82310
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.4538130035600526e-05,
+      "loss": 0.1783,
+      "step": 82320
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.453679167001259e-05,
+      "loss": 0.1186,
+      "step": 82330
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.453545330442464e-05,
+      "loss": 0.2681,
+      "step": 82340
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.45341149388367e-05,
+      "loss": 0.0726,
+      "step": 82350
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.4532776573248754e-05,
+      "loss": 0.2057,
+      "step": 82360
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.453143820766081e-05,
+      "loss": 0.2057,
+      "step": 82370
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.4530099842072864e-05,
+      "loss": 0.1462,
+      "step": 82380
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.452876147648492e-05,
+      "loss": 0.1637,
+      "step": 82390
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.4527423110896975e-05,
+      "loss": 0.0465,
+      "step": 82400
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.452608474530903e-05,
+      "loss": 0.0558,
+      "step": 82410
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.4524746379721086e-05,
+      "loss": 0.2219,
+      "step": 82420
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.452340801413315e-05,
+      "loss": 0.1019,
+      "step": 82430
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.45220696485452e-05,
+      "loss": 0.1288,
+      "step": 82440
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.452073128295726e-05,
+      "loss": 0.0742,
+      "step": 82450
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.4519392917369314e-05,
+      "loss": 0.0833,
+      "step": 82460
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.451805455178137e-05,
+      "loss": 0.1683,
+      "step": 82470
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.4516716186193424e-05,
+      "loss": 0.1095,
+      "step": 82480
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.451537782060548e-05,
+      "loss": 0.2702,
+      "step": 82490
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.4514039455017535e-05,
+      "loss": 0.2477,
+      "step": 82500
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.451270108942959e-05,
+      "loss": 0.2745,
+      "step": 82510
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.4511362723841646e-05,
+      "loss": 0.1943,
+      "step": 82520
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.451002435825371e-05,
+      "loss": 0.0844,
+      "step": 82530
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.450868599266576e-05,
+      "loss": 0.0716,
+      "step": 82540
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.450734762707782e-05,
+      "loss": 0.1555,
+      "step": 82550
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.4506009261489874e-05,
+      "loss": 0.2894,
+      "step": 82560
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.450467089590193e-05,
+      "loss": 0.2121,
+      "step": 82570
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.4503332530313985e-05,
+      "loss": 0.0966,
+      "step": 82580
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.450199416472604e-05,
+      "loss": 0.2168,
+      "step": 82590
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.4500655799138095e-05,
+      "loss": 0.1947,
+      "step": 82600
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.449931743355015e-05,
+      "loss": 0.1386,
+      "step": 82610
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.4497979067962206e-05,
+      "loss": 0.2206,
+      "step": 82620
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.449664070237427e-05,
+      "loss": 0.1835,
+      "step": 82630
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.4495302336786323e-05,
+      "loss": 0.2231,
+      "step": 82640
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.449396397119838e-05,
+      "loss": 0.2753,
+      "step": 82650
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.4492625605610434e-05,
+      "loss": 0.1184,
+      "step": 82660
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.449128724002249e-05,
+      "loss": 0.212,
+      "step": 82670
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.4489948874434545e-05,
+      "loss": 0.2003,
+      "step": 82680
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.44886105088466e-05,
+      "loss": 0.2147,
+      "step": 82690
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.4487272143258655e-05,
+      "loss": 0.1517,
+      "step": 82700
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.448593377767071e-05,
+      "loss": 0.1459,
+      "step": 82710
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.4484595412082766e-05,
+      "loss": 0.1346,
+      "step": 82720
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.448325704649483e-05,
+      "loss": 0.232,
+      "step": 82730
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.4481918680906884e-05,
+      "loss": 0.1092,
+      "step": 82740
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.448058031531894e-05,
+      "loss": 0.1942,
+      "step": 82750
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.4479241949730994e-05,
+      "loss": 0.1651,
+      "step": 82760
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.447790358414305e-05,
+      "loss": 0.1184,
+      "step": 82770
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.4476565218555105e-05,
+      "loss": 0.1795,
+      "step": 82780
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.447522685296716e-05,
+      "loss": 0.081,
+      "step": 82790
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.4473888487379216e-05,
+      "loss": 0.1329,
+      "step": 82800
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.447255012179127e-05,
+      "loss": 0.1546,
+      "step": 82810
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.4471211756203326e-05,
+      "loss": 0.0357,
+      "step": 82820
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.446987339061539e-05,
+      "loss": 0.1574,
+      "step": 82830
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.4468535025027444e-05,
+      "loss": 0.216,
+      "step": 82840
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.44671966594395e-05,
+      "loss": 0.0987,
+      "step": 82850
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.4465858293851554e-05,
+      "loss": 0.1091,
+      "step": 82860
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.446451992826361e-05,
+      "loss": 0.1319,
+      "step": 82870
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.4463181562675665e-05,
+      "loss": 0.1229,
+      "step": 82880
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.4461843197087714e-05,
+      "loss": 0.179,
+      "step": 82890
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.446050483149977e-05,
+      "loss": 0.1399,
+      "step": 82900
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.445916646591183e-05,
+      "loss": 0.1641,
+      "step": 82910
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.4457828100323886e-05,
+      "loss": 0.1587,
+      "step": 82920
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.445648973473594e-05,
+      "loss": 0.2489,
+      "step": 82930
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.4455151369148e-05,
+      "loss": 0.1891,
+      "step": 82940
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.445381300356005e-05,
+      "loss": 0.051,
+      "step": 82950
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.445247463797211e-05,
+      "loss": 0.0846,
+      "step": 82960
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.445113627238416e-05,
+      "loss": 0.0899,
+      "step": 82970
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.444979790679622e-05,
+      "loss": 0.1539,
+      "step": 82980
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.4448459541208274e-05,
+      "loss": 0.0413,
+      "step": 82990
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.444712117562033e-05,
+      "loss": 0.2817,
+      "step": 83000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.444578281003239e-05,
+      "loss": 0.1095,
+      "step": 83010
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.2275,
+      "step": 83020
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": {
+        "accuracy": 0.946582354712436
+      },
+      "eval_f1": {
+        "f1": 0.9462472081330006
+      },
+      "eval_loss": 0.2473287284374237,
+      "eval_precision": {
+        "precision": 0.9479482237973264
+      },
+      "eval_recall": {
+        "recall": 0.9463251646491099
+      },
+      "eval_runtime": 429.3588,
+      "eval_samples_per_second": 38.674,
+      "eval_steps_per_second": 19.338,
+      "step": 83025
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.44431060788565e-05,
+      "loss": 0.0984,
+      "step": 83030
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.444176771326856e-05,
+      "loss": 0.2262,
+      "step": 83040
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.444042934768061e-05,
+      "loss": 0.0867,
+      "step": 83050
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.443909098209267e-05,
+      "loss": 0.0956,
+      "step": 83060
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.443775261650472e-05,
+      "loss": 0.1536,
+      "step": 83070
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.443641425091678e-05,
+      "loss": 0.0058,
+      "step": 83080
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.4435075885328834e-05,
+      "loss": 0.2837,
+      "step": 83090
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.443373751974089e-05,
+      "loss": 0.217,
+      "step": 83100
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.443239915415295e-05,
+      "loss": 0.0678,
+      "step": 83110
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.443106078856501e-05,
+      "loss": 0.0101,
+      "step": 83120
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.442972242297706e-05,
+      "loss": 0.3147,
+      "step": 83130
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.442838405738912e-05,
+      "loss": 0.1598,
+      "step": 83140
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.442704569180117e-05,
+      "loss": 0.1927,
+      "step": 83150
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.442570732621323e-05,
+      "loss": 0.1164,
+      "step": 83160
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.4424368960625283e-05,
+      "loss": 0.0781,
+      "step": 83170
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.442303059503734e-05,
+      "loss": 0.3308,
+      "step": 83180
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.4421692229449394e-05,
+      "loss": 0.0359,
+      "step": 83190
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.442035386386145e-05,
+      "loss": 0.2274,
+      "step": 83200
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.441901549827351e-05,
+      "loss": 0.1601,
+      "step": 83210
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.441767713268557e-05,
+      "loss": 0.2136,
+      "step": 83220
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.441633876709762e-05,
+      "loss": 0.3091,
+      "step": 83230
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.441500040150968e-05,
+      "loss": 0.2931,
+      "step": 83240
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.441366203592173e-05,
+      "loss": 0.1947,
+      "step": 83250
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.441232367033379e-05,
+      "loss": 0.234,
+      "step": 83260
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.4410985304745844e-05,
+      "loss": 0.0963,
+      "step": 83270
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.44096469391579e-05,
+      "loss": 0.112,
+      "step": 83280
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.4408308573569954e-05,
+      "loss": 0.1739,
+      "step": 83290
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.440697020798201e-05,
+      "loss": 0.2325,
+      "step": 83300
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.440563184239407e-05,
+      "loss": 0.0859,
+      "step": 83310
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.440429347680613e-05,
+      "loss": 0.0574,
+      "step": 83320
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.440295511121818e-05,
+      "loss": 0.2503,
+      "step": 83330
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.440161674563024e-05,
+      "loss": 0.276,
+      "step": 83340
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.440027838004229e-05,
+      "loss": 0.166,
+      "step": 83350
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.439894001445435e-05,
+      "loss": 0.1534,
+      "step": 83360
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.4397601648866404e-05,
+      "loss": 0.1399,
+      "step": 83370
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.439626328327846e-05,
+      "loss": 0.197,
+      "step": 83380
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.4394924917690514e-05,
+      "loss": 0.0415,
+      "step": 83390
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.439358655210257e-05,
+      "loss": 0.2322,
+      "step": 83400
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.439224818651463e-05,
+      "loss": 0.0831,
+      "step": 83410
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.439090982092669e-05,
+      "loss": 0.1249,
+      "step": 83420
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.438957145533874e-05,
+      "loss": 0.1365,
+      "step": 83430
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.43882330897508e-05,
+      "loss": 0.1433,
+      "step": 83440
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.438689472416285e-05,
+      "loss": 0.1405,
+      "step": 83450
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.438555635857491e-05,
+      "loss": 0.3996,
+      "step": 83460
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.4384217992986964e-05,
+      "loss": 0.017,
+      "step": 83470
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.438287962739902e-05,
+      "loss": 0.0854,
+      "step": 83480
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.4381541261811075e-05,
+      "loss": 0.2159,
+      "step": 83490
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.438020289622313e-05,
+      "loss": 0.1827,
+      "step": 83500
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.437886453063519e-05,
+      "loss": 0.1096,
+      "step": 83510
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.437752616504725e-05,
+      "loss": 0.0983,
+      "step": 83520
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.43761877994593e-05,
+      "loss": 0.2077,
+      "step": 83530
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.437484943387136e-05,
+      "loss": 0.0187,
+      "step": 83540
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.437351106828341e-05,
+      "loss": 0.2087,
+      "step": 83550
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.437217270269547e-05,
+      "loss": 0.2124,
+      "step": 83560
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.4370834337107524e-05,
+      "loss": 0.1616,
+      "step": 83570
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.436949597151958e-05,
+      "loss": 0.2344,
+      "step": 83580
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.4368157605931635e-05,
+      "loss": 0.1936,
+      "step": 83590
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.436681924034369e-05,
+      "loss": 0.2191,
+      "step": 83600
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.436548087475575e-05,
+      "loss": 0.2947,
+      "step": 83610
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.436414250916781e-05,
+      "loss": 0.1757,
+      "step": 83620
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.436280414357986e-05,
+      "loss": 0.1029,
+      "step": 83630
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.436146577799192e-05,
+      "loss": 0.0344,
+      "step": 83640
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.4360127412403974e-05,
+      "loss": 0.1998,
+      "step": 83650
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.435878904681603e-05,
+      "loss": 0.2024,
+      "step": 83660
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.4357450681228084e-05,
+      "loss": 0.1893,
+      "step": 83670
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.435611231564014e-05,
+      "loss": 0.147,
+      "step": 83680
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.4354773950052195e-05,
+      "loss": 0.102,
+      "step": 83690
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.435343558446425e-05,
+      "loss": 0.0803,
+      "step": 83700
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.435209721887631e-05,
+      "loss": 0.3082,
+      "step": 83710
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.435075885328837e-05,
+      "loss": 0.0232,
+      "step": 83720
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.434942048770042e-05,
+      "loss": 0.1172,
+      "step": 83730
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.434808212211248e-05,
+      "loss": 0.1326,
+      "step": 83740
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.4346743756524534e-05,
+      "loss": 0.1163,
+      "step": 83750
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.434540539093659e-05,
+      "loss": 0.1966,
+      "step": 83760
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.4344067025348644e-05,
+      "loss": 0.2018,
+      "step": 83770
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.43427286597607e-05,
+      "loss": 0.1756,
+      "step": 83780
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.4341390294172755e-05,
+      "loss": 0.1791,
+      "step": 83790
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.434005192858481e-05,
+      "loss": 0.0449,
+      "step": 83800
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.433871356299687e-05,
+      "loss": 0.1082,
+      "step": 83810
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.433737519740893e-05,
+      "loss": 0.1354,
+      "step": 83820
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.433603683182098e-05,
+      "loss": 0.1366,
+      "step": 83830
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.433469846623304e-05,
+      "loss": 0.1093,
+      "step": 83840
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.4333360100645094e-05,
+      "loss": 0.1618,
+      "step": 83850
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.433202173505715e-05,
+      "loss": 0.4249,
+      "step": 83860
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.4330683369469204e-05,
+      "loss": 0.1966,
+      "step": 83870
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.432934500388126e-05,
+      "loss": 0.0752,
+      "step": 83880
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.4328006638293315e-05,
+      "loss": 0.1054,
+      "step": 83890
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.432666827270538e-05,
+      "loss": 0.1804,
+      "step": 83900
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.432532990711743e-05,
+      "loss": 0.0734,
+      "step": 83910
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.432399154152949e-05,
+      "loss": 0.1244,
+      "step": 83920
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.432265317594154e-05,
+      "loss": 0.1157,
+      "step": 83930
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.43213148103536e-05,
+      "loss": 0.1355,
+      "step": 83940
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.4319976444765654e-05,
+      "loss": 0.0992,
+      "step": 83950
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.431863807917771e-05,
+      "loss": 0.06,
+      "step": 83960
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.4317299713589765e-05,
+      "loss": 0.066,
+      "step": 83970
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.431596134800182e-05,
+      "loss": 0.0608,
+      "step": 83980
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4314622982413875e-05,
+      "loss": 0.2448,
+      "step": 83990
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.431328461682594e-05,
+      "loss": 0.2083,
+      "step": 84000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.431194625123799e-05,
+      "loss": 0.1121,
+      "step": 84010
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.431060788565005e-05,
+      "loss": 0.2058,
+      "step": 84020
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4309269520062103e-05,
+      "loss": 0.3279,
+      "step": 84030
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.430793115447416e-05,
+      "loss": 0.0738,
+      "step": 84040
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4306592788886214e-05,
+      "loss": 0.021,
+      "step": 84050
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.430525442329827e-05,
+      "loss": 0.0099,
+      "step": 84060
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.4303916057710325e-05,
+      "loss": 0.1945,
+      "step": 84070
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.430257769212238e-05,
+      "loss": 0.1585,
+      "step": 84080
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.4301239326534435e-05,
+      "loss": 0.2786,
+      "step": 84090
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.42999009609465e-05,
+      "loss": 0.1041,
+      "step": 84100
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.429856259535855e-05,
+      "loss": 0.047,
+      "step": 84110
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.429722422977061e-05,
+      "loss": 0.1844,
+      "step": 84120
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.4295885864182664e-05,
+      "loss": 0.1527,
+      "step": 84130
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.429454749859472e-05,
+      "loss": 0.2107,
+      "step": 84140
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.4293209133006774e-05,
+      "loss": 0.2544,
+      "step": 84150
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.429187076741883e-05,
+      "loss": 0.025,
+      "step": 84160
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.4290532401830885e-05,
+      "loss": 0.0189,
+      "step": 84170
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.428919403624294e-05,
+      "loss": 0.1377,
+      "step": 84180
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.4287855670654996e-05,
+      "loss": 0.1648,
+      "step": 84190
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.428651730506706e-05,
+      "loss": 0.1221,
+      "step": 84200
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.428517893947911e-05,
+      "loss": 0.0963,
+      "step": 84210
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.428384057389117e-05,
+      "loss": 0.122,
+      "step": 84220
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.4282502208303224e-05,
+      "loss": 0.1318,
+      "step": 84230
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.428116384271528e-05,
+      "loss": 0.0553,
+      "step": 84240
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.4279825477127334e-05,
+      "loss": 0.0408,
+      "step": 84250
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.427848711153939e-05,
+      "loss": 0.1916,
+      "step": 84260
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.4277148745951445e-05,
+      "loss": 0.0242,
+      "step": 84270
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.42758103803635e-05,
+      "loss": 0.1387,
+      "step": 84280
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.4274472014775556e-05,
+      "loss": 0.0704,
+      "step": 84290
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.427313364918762e-05,
+      "loss": 0.2015,
+      "step": 84300
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.427179528359967e-05,
+      "loss": 0.0602,
+      "step": 84310
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.427045691801173e-05,
+      "loss": 0.1497,
+      "step": 84320
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.4269118552423784e-05,
+      "loss": 0.0477,
+      "step": 84330
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.426778018683584e-05,
+      "loss": 0.0843,
+      "step": 84340
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.4266441821247895e-05,
+      "loss": 0.0928,
+      "step": 84350
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.426510345565995e-05,
+      "loss": 0.257,
+      "step": 84360
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.4263765090072005e-05,
+      "loss": 0.1729,
+      "step": 84370
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.426242672448406e-05,
+      "loss": 0.061,
+      "step": 84380
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.4261088358896116e-05,
+      "loss": 0.2762,
+      "step": 84390
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.425974999330818e-05,
+      "loss": 0.0683,
+      "step": 84400
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.425841162772023e-05,
+      "loss": 0.1556,
+      "step": 84410
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.425707326213229e-05,
+      "loss": 0.2008,
+      "step": 84420
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.4255734896544344e-05,
+      "loss": 0.1297,
+      "step": 84430
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.42543965309564e-05,
+      "loss": 0.148,
+      "step": 84440
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.4253058165368455e-05,
+      "loss": 0.2291,
+      "step": 84450
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.425171979978051e-05,
+      "loss": 0.0648,
+      "step": 84460
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.4250381434192565e-05,
+      "loss": 0.1294,
+      "step": 84470
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.424904306860462e-05,
+      "loss": 0.1708,
+      "step": 84480
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.4247704703016676e-05,
+      "loss": 0.168,
+      "step": 84490
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.424636633742874e-05,
+      "loss": 0.1797,
+      "step": 84500
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.4245027971840794e-05,
+      "loss": 0.2022,
+      "step": 84510
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.424368960625285e-05,
+      "loss": 0.3326,
+      "step": 84520
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.4242351240664904e-05,
+      "loss": 0.0933,
+      "step": 84530
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.424101287507696e-05,
+      "loss": 0.1287,
+      "step": 84540
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.4239674509489015e-05,
+      "loss": 0.1712,
+      "step": 84550
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.423833614390107e-05,
+      "loss": 0.1825,
+      "step": 84560
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.4236997778313126e-05,
+      "loss": 0.038,
+      "step": 84570
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.423565941272518e-05,
+      "loss": 0.1181,
+      "step": 84580
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.4234321047137236e-05,
+      "loss": 0.253,
+      "step": 84590
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.42329826815493e-05,
+      "loss": 0.1109,
+      "step": 84600
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.4231644315961354e-05,
+      "loss": 0.1351,
+      "step": 84610
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.423030595037341e-05,
+      "loss": 0.3246,
+      "step": 84620
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.4228967584785464e-05,
+      "loss": 0.2239,
+      "step": 84630
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.422762921919752e-05,
+      "loss": 0.1058,
+      "step": 84640
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.4226290853609575e-05,
+      "loss": 0.2556,
+      "step": 84650
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.422495248802163e-05,
+      "loss": 0.1876,
+      "step": 84660
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.4223614122433686e-05,
+      "loss": 0.1663,
+      "step": 84670
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.422227575684574e-05,
+      "loss": 0.1271,
+      "step": 84680
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.4220937391257796e-05,
+      "loss": 0.0069,
+      "step": 84690
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.421959902566986e-05,
+      "loss": 0.188,
+      "step": 84700
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.4218260660081914e-05,
+      "loss": 0.1978,
+      "step": 84710
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.421692229449397e-05,
+      "loss": 0.1101,
+      "step": 84720
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.4215583928906025e-05,
+      "loss": 0.1357,
+      "step": 84730
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.421424556331808e-05,
+      "loss": 0.1767,
+      "step": 84740
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.4212907197730135e-05,
+      "loss": 0.141,
+      "step": 84750
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.421156883214219e-05,
+      "loss": 0.2467,
+      "step": 84760
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.4210230466554246e-05,
+      "loss": 0.1792,
+      "step": 84770
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.42088921009663e-05,
+      "loss": 0.3379,
+      "step": 84780
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.4207553735378357e-05,
+      "loss": 0.0383,
+      "step": 84790
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.420621536979042e-05,
+      "loss": 0.2969,
+      "step": 84800
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.4204877004202474e-05,
+      "loss": 0.0777,
+      "step": 84810
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.420353863861453e-05,
+      "loss": 0.0549,
+      "step": 84820
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.4202200273026585e-05,
+      "loss": 0.1626,
+      "step": 84830
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.420086190743864e-05,
+      "loss": 0.1431,
+      "step": 84840
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.4199523541850695e-05,
+      "loss": 0.0588,
+      "step": 84850
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.419818517626275e-05,
+      "loss": 0.1221,
+      "step": 84860
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.4196846810674806e-05,
+      "loss": 0.0769,
+      "step": 84870
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.419550844508686e-05,
+      "loss": 0.1153,
+      "step": 84880
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.419417007949892e-05,
+      "loss": 0.0827,
+      "step": 84890
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.419283171391098e-05,
+      "loss": 0.1125,
+      "step": 84900
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.4191493348323034e-05,
+      "loss": 0.1332,
+      "step": 84910
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.419015498273509e-05,
+      "loss": 0.361,
+      "step": 84920
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.4188816617147145e-05,
+      "loss": 0.2602,
+      "step": 84930
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.41874782515592e-05,
+      "loss": 0.067,
+      "step": 84940
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.4186139885971255e-05,
+      "loss": 0.141,
+      "step": 84950
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.418480152038331e-05,
+      "loss": 0.1565,
+      "step": 84960
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.4183463154795366e-05,
+      "loss": 0.1737,
+      "step": 84970
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.418212478920742e-05,
+      "loss": 0.4148,
+      "step": 84980
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.418078642361948e-05,
+      "loss": 0.1327,
+      "step": 84990
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.417944805803154e-05,
+      "loss": 0.195,
+      "step": 85000
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.4178109692443594e-05,
+      "loss": 0.0786,
+      "step": 85010
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.417677132685565e-05,
+      "loss": 0.1364,
+      "step": 85020
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.4175432961267705e-05,
+      "loss": 0.164,
+      "step": 85030
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.417409459567976e-05,
+      "loss": 0.1466,
+      "step": 85040
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.4172756230091816e-05,
+      "loss": 0.2166,
+      "step": 85050
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.417141786450387e-05,
+      "loss": 0.107,
+      "step": 85060
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.4170079498915926e-05,
+      "loss": 0.0289,
+      "step": 85070
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.416874113332798e-05,
+      "loss": 0.0998,
+      "step": 85080
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.416740276774004e-05,
+      "loss": 0.1427,
+      "step": 85090
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.41660644021521e-05,
+      "loss": 0.2068,
+      "step": 85100
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.4164726036564154e-05,
+      "loss": 0.2923,
+      "step": 85110
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.416338767097621e-05,
+      "loss": 0.0521,
+      "step": 85120
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.4162049305388265e-05,
+      "loss": 0.087,
+      "step": 85130
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.416071093980032e-05,
+      "loss": 0.3271,
+      "step": 85140
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4159372574212376e-05,
+      "loss": 0.0507,
+      "step": 85150
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.415803420862443e-05,
+      "loss": 0.1856,
+      "step": 85160
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4156695843036486e-05,
+      "loss": 0.1553,
+      "step": 85170
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.415535747744854e-05,
+      "loss": 0.4072,
+      "step": 85180
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4154019111860604e-05,
+      "loss": 0.0525,
+      "step": 85190
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.415268074627266e-05,
+      "loss": 0.1998,
+      "step": 85200
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4151342380684715e-05,
+      "loss": 0.1088,
+      "step": 85210
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.415000401509677e-05,
+      "loss": 0.0958,
+      "step": 85220
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.4148665649508825e-05,
+      "loss": 0.2155,
+      "step": 85230
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.414732728392088e-05,
+      "loss": 0.2807,
+      "step": 85240
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.4145988918332936e-05,
+      "loss": 0.1161,
+      "step": 85250
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.4144650552744985e-05,
+      "loss": 0.1917,
+      "step": 85260
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.414331218715704e-05,
+      "loss": 0.1881,
+      "step": 85270
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.41419738215691e-05,
+      "loss": 0.078,
+      "step": 85280
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.414063545598116e-05,
+      "loss": 0.0643,
+      "step": 85290
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.413929709039321e-05,
+      "loss": 0.1194,
+      "step": 85300
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.413795872480527e-05,
+      "loss": 0.2315,
+      "step": 85310
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.413662035921732e-05,
+      "loss": 0.1531,
+      "step": 85320
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.413528199362938e-05,
+      "loss": 0.1429,
+      "step": 85330
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.4133943628041434e-05,
+      "loss": 0.2556,
+      "step": 85340
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.413260526245349e-05,
+      "loss": 0.1354,
+      "step": 85350
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.4131266896865545e-05,
+      "loss": 0.328,
+      "step": 85360
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.412992853127761e-05,
+      "loss": 0.1256,
+      "step": 85370
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.412859016568966e-05,
+      "loss": 0.0243,
+      "step": 85380
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.412725180010172e-05,
+      "loss": 0.1113,
+      "step": 85390
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.412591343451377e-05,
+      "loss": 0.0496,
+      "step": 85400
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.412457506892583e-05,
+      "loss": 0.219,
+      "step": 85410
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.4123236703337883e-05,
+      "loss": 0.1003,
+      "step": 85420
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.412189833774994e-05,
+      "loss": 0.2046,
+      "step": 85430
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.4120559972161994e-05,
+      "loss": 0.1923,
+      "step": 85440
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.411922160657405e-05,
+      "loss": 0.2031,
+      "step": 85450
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.4117883240986105e-05,
+      "loss": 0.2024,
+      "step": 85460
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.411654487539817e-05,
+      "loss": 0.0862,
+      "step": 85470
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.411520650981022e-05,
+      "loss": 0.0932,
+      "step": 85480
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.411386814422228e-05,
+      "loss": 0.1564,
+      "step": 85490
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.411252977863433e-05,
+      "loss": 0.2454,
+      "step": 85500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.411119141304639e-05,
+      "loss": 0.0442,
+      "step": 85510
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.4109853047458444e-05,
+      "loss": 0.1514,
+      "step": 85520
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.41085146818705e-05,
+      "loss": 0.1498,
+      "step": 85530
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.4107176316282554e-05,
+      "loss": 0.112,
+      "step": 85540
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.410583795069461e-05,
+      "loss": 0.2365,
+      "step": 85550
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.4104499585106665e-05,
+      "loss": 0.2129,
+      "step": 85560
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.410316121951873e-05,
+      "loss": 0.0974,
+      "step": 85570
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.410182285393078e-05,
+      "loss": 0.1574,
+      "step": 85580
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.410048448834284e-05,
+      "loss": 0.1477,
+      "step": 85590
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.409914612275489e-05,
+      "loss": 0.1302,
+      "step": 85600
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.409780775716695e-05,
+      "loss": 0.2544,
+      "step": 85610
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.4096469391579004e-05,
+      "loss": 0.0292,
+      "step": 85620
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.409513102599106e-05,
+      "loss": 0.2511,
+      "step": 85630
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.4093792660403114e-05,
+      "loss": 0.1705,
+      "step": 85640
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.409245429481517e-05,
+      "loss": 0.0885,
+      "step": 85650
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.4091115929227225e-05,
+      "loss": 0.1748,
+      "step": 85660
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.408977756363929e-05,
+      "loss": 0.1903,
+      "step": 85670
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.408843919805134e-05,
+      "loss": 0.3508,
+      "step": 85680
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.40871008324634e-05,
+      "loss": 0.0795,
+      "step": 85690
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.408576246687545e-05,
+      "loss": 0.2262,
+      "step": 85700
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.408442410128751e-05,
+      "loss": 0.1397,
+      "step": 85710
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.4083085735699564e-05,
+      "loss": 0.1442,
+      "step": 85720
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.408174737011162e-05,
+      "loss": 0.134,
+      "step": 85730
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.4080409004523675e-05,
+      "loss": 0.137,
+      "step": 85740
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.407907063893573e-05,
+      "loss": 0.2302,
+      "step": 85750
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.4077732273347785e-05,
+      "loss": 0.1201,
+      "step": 85760
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.407639390775985e-05,
+      "loss": 0.1556,
+      "step": 85770
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.40750555421719e-05,
+      "loss": 0.0538,
+      "step": 85780
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.407371717658396e-05,
+      "loss": 0.1327,
+      "step": 85790
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.4072378810996013e-05,
+      "loss": 0.204,
+      "step": 85800
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.407104044540807e-05,
+      "loss": 0.2239,
+      "step": 85810
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.4069702079820124e-05,
+      "loss": 0.1501,
+      "step": 85820
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.406836371423218e-05,
+      "loss": 0.0504,
+      "step": 85830
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.4067025348644235e-05,
+      "loss": 0.11,
+      "step": 85840
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.406568698305629e-05,
+      "loss": 0.1777,
+      "step": 85850
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.4064348617468345e-05,
+      "loss": 0.1051,
+      "step": 85860
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.406301025188041e-05,
+      "loss": 0.1863,
+      "step": 85870
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.406167188629246e-05,
+      "loss": 0.1146,
+      "step": 85880
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.406033352070452e-05,
+      "loss": 0.1632,
+      "step": 85890
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.4058995155116574e-05,
+      "loss": 0.1851,
+      "step": 85900
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.405765678952863e-05,
+      "loss": 0.184,
+      "step": 85910
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.4056318423940684e-05,
+      "loss": 0.0719,
+      "step": 85920
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.405498005835274e-05,
+      "loss": 0.0621,
+      "step": 85930
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.4053641692764795e-05,
+      "loss": 0.1886,
+      "step": 85940
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.405230332717685e-05,
+      "loss": 0.1377,
+      "step": 85950
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.4050964961588906e-05,
+      "loss": 0.0611,
+      "step": 85960
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.404962659600097e-05,
+      "loss": 0.204,
+      "step": 85970
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.404828823041302e-05,
+      "loss": 0.2841,
+      "step": 85980
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.404694986482508e-05,
+      "loss": 0.1272,
+      "step": 85990
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.4045611499237134e-05,
+      "loss": 0.2068,
+      "step": 86000
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.404427313364919e-05,
+      "loss": 0.2017,
+      "step": 86010
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.4042934768061244e-05,
+      "loss": 0.1205,
+      "step": 86020
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.40415964024733e-05,
+      "loss": 0.314,
+      "step": 86030
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.4040258036885355e-05,
+      "loss": 0.1157,
+      "step": 86040
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.403891967129741e-05,
+      "loss": 0.0735,
+      "step": 86050
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.4037581305709466e-05,
+      "loss": 0.1903,
+      "step": 86060
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.403624294012153e-05,
+      "loss": 0.194,
+      "step": 86070
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.403490457453358e-05,
+      "loss": 0.1247,
+      "step": 86080
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.403356620894564e-05,
+      "loss": 0.1332,
+      "step": 86090
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.4032227843357694e-05,
+      "loss": 0.1967,
+      "step": 86100
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.403088947776975e-05,
+      "loss": 0.5201,
+      "step": 86110
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.4029551112181805e-05,
+      "loss": 0.0641,
+      "step": 86120
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.402821274659386e-05,
+      "loss": 0.2287,
+      "step": 86130
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.4026874381005915e-05,
+      "loss": 0.0552,
+      "step": 86140
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.402553601541797e-05,
+      "loss": 0.1219,
+      "step": 86150
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.4024197649830026e-05,
+      "loss": 0.054,
+      "step": 86160
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.402285928424209e-05,
+      "loss": 0.0689,
+      "step": 86170
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.402152091865414e-05,
+      "loss": 0.0541,
+      "step": 86180
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.40201825530662e-05,
+      "loss": 0.2115,
+      "step": 86190
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.4018844187478254e-05,
+      "loss": 0.1172,
+      "step": 86200
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.401750582189031e-05,
+      "loss": 0.2639,
+      "step": 86210
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.4016167456302365e-05,
+      "loss": 0.239,
+      "step": 86220
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.401482909071442e-05,
+      "loss": 0.1669,
+      "step": 86230
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.4013490725126475e-05,
+      "loss": 0.0489,
+      "step": 86240
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.401215235953853e-05,
+      "loss": 0.1149,
+      "step": 86250
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.4010813993950586e-05,
+      "loss": 0.3338,
+      "step": 86260
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.400947562836265e-05,
+      "loss": 0.0808,
+      "step": 86270
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.4008137262774703e-05,
+      "loss": 0.1089,
+      "step": 86280
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.400679889718676e-05,
+      "loss": 0.0794,
+      "step": 86290
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.4005460531598814e-05,
+      "loss": 0.0837,
+      "step": 86300
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.400412216601087e-05,
+      "loss": 0.1889,
+      "step": 86310
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.4002783800422925e-05,
+      "loss": 0.102,
+      "step": 86320
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.400144543483498e-05,
+      "loss": 0.0675,
+      "step": 86330
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.4000107069247036e-05,
+      "loss": 0.1939,
+      "step": 86340
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.399876870365909e-05,
+      "loss": 0.1921,
+      "step": 86350
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.3997430338071146e-05,
+      "loss": 0.1054,
+      "step": 86360
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.399609197248321e-05,
+      "loss": 0.2439,
+      "step": 86370
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.3994753606895264e-05,
+      "loss": 0.2031,
+      "step": 86380
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.399341524130732e-05,
+      "loss": 0.355,
+      "step": 86390
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.3992076875719374e-05,
+      "loss": 0.0794,
+      "step": 86400
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.399073851013143e-05,
+      "loss": 0.0371,
+      "step": 86410
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.3989400144543485e-05,
+      "loss": 0.1118,
+      "step": 86420
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.398806177895554e-05,
+      "loss": 0.1552,
+      "step": 86430
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.3986723413367596e-05,
+      "loss": 0.1007,
+      "step": 86440
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.398538504777965e-05,
+      "loss": 0.2183,
+      "step": 86450
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.3984046682191706e-05,
+      "loss": 0.1538,
+      "step": 86460
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.398270831660377e-05,
+      "loss": 0.2129,
+      "step": 86470
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.3981369951015824e-05,
+      "loss": 0.1106,
+      "step": 86480
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.398003158542788e-05,
+      "loss": 0.1633,
+      "step": 86490
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.3978693219839934e-05,
+      "loss": 0.0574,
+      "step": 86500
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.397735485425199e-05,
+      "loss": 0.0864,
+      "step": 86510
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.3976016488664045e-05,
+      "loss": 0.0036,
+      "step": 86520
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.39746781230761e-05,
+      "loss": 0.1853,
+      "step": 86530
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.3973339757488156e-05,
+      "loss": 0.1508,
+      "step": 86540
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.397200139190021e-05,
+      "loss": 0.132,
+      "step": 86550
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.3970663026312266e-05,
+      "loss": 0.1732,
+      "step": 86560
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.396932466072433e-05,
+      "loss": 0.1434,
+      "step": 86570
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.3967986295136384e-05,
+      "loss": 0.2597,
+      "step": 86580
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.396664792954844e-05,
+      "loss": 0.057,
+      "step": 86590
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.3965309563960495e-05,
+      "loss": 0.2824,
+      "step": 86600
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.396397119837255e-05,
+      "loss": 0.2363,
+      "step": 86610
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.3962632832784605e-05,
+      "loss": 0.1253,
+      "step": 86620
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.396129446719666e-05,
+      "loss": 0.4468,
+      "step": 86630
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.3959956101608716e-05,
+      "loss": 0.1976,
+      "step": 86640
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.395861773602077e-05,
+      "loss": 0.0758,
+      "step": 86650
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.3957279370432833e-05,
+      "loss": 0.0541,
+      "step": 86660
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.395594100484489e-05,
+      "loss": 0.1162,
+      "step": 86670
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.3954602639256944e-05,
+      "loss": 0.2441,
+      "step": 86680
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.3953264273669e-05,
+      "loss": 0.0541,
+      "step": 86690
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.3951925908081055e-05,
+      "loss": 0.2325,
+      "step": 86700
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.395058754249311e-05,
+      "loss": 0.1577,
+      "step": 86710
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.3949249176905165e-05,
+      "loss": 0.2779,
+      "step": 86720
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.394791081131722e-05,
+      "loss": 0.231,
+      "step": 86730
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.3946572445729276e-05,
+      "loss": 0.1388,
+      "step": 86740
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.394523408014133e-05,
+      "loss": 0.0359,
+      "step": 86750
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.3943895714553394e-05,
+      "loss": 0.1731,
+      "step": 86760
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.394255734896545e-05,
+      "loss": 0.1057,
+      "step": 86770
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.3941218983377504e-05,
+      "loss": 0.1762,
+      "step": 86780
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.393988061778956e-05,
+      "loss": 0.2464,
+      "step": 86790
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.3938542252201615e-05,
+      "loss": 0.2295,
+      "step": 86800
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.393720388661367e-05,
+      "loss": 0.0088,
+      "step": 86810
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.3935865521025726e-05,
+      "loss": 0.1403,
+      "step": 86820
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.393452715543778e-05,
+      "loss": 0.161,
+      "step": 86830
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.3933188789849836e-05,
+      "loss": 0.2097,
+      "step": 86840
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.393185042426189e-05,
+      "loss": 0.1584,
+      "step": 86850
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.3930512058673954e-05,
+      "loss": 0.1698,
+      "step": 86860
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.392917369308601e-05,
+      "loss": 0.2002,
+      "step": 86870
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.3927835327498064e-05,
+      "loss": 0.1229,
+      "step": 86880
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.392649696191012e-05,
+      "loss": 0.0344,
+      "step": 86890
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.3925158596322175e-05,
+      "loss": 0.1033,
+      "step": 86900
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.392382023073423e-05,
+      "loss": 0.2442,
+      "step": 86910
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.3922481865146286e-05,
+      "loss": 0.1885,
+      "step": 86920
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.392114349955834e-05,
+      "loss": 0.0406,
+      "step": 86930
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.3919805133970396e-05,
+      "loss": 0.0911,
+      "step": 86940
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.391846676838245e-05,
+      "loss": 0.1749,
+      "step": 86950
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.3917128402794514e-05,
+      "loss": 0.2025,
+      "step": 86960
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.391579003720657e-05,
+      "loss": 0.1874,
+      "step": 86970
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.3914451671618625e-05,
+      "loss": 0.0731,
+      "step": 86980
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.391311330603068e-05,
+      "loss": 0.1788,
+      "step": 86990
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.3911774940442735e-05,
+      "loss": 0.1797,
+      "step": 87000
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.391043657485479e-05,
+      "loss": 0.0875,
+      "step": 87010
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.3909098209266846e-05,
+      "loss": 0.1293,
+      "step": 87020
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.39077598436789e-05,
+      "loss": 0.1035,
+      "step": 87030
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.3906421478090957e-05,
+      "loss": 0.2059,
+      "step": 87040
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.390508311250301e-05,
+      "loss": 0.0852,
+      "step": 87050
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.3903744746915074e-05,
+      "loss": 0.1248,
+      "step": 87060
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.390240638132713e-05,
+      "loss": 0.1364,
+      "step": 87070
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.3901068015739185e-05,
+      "loss": 0.1744,
+      "step": 87080
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.389972965015124e-05,
+      "loss": 0.0611,
+      "step": 87090
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.3898391284563295e-05,
+      "loss": 0.11,
+      "step": 87100
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.389705291897535e-05,
+      "loss": 0.0621,
+      "step": 87110
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.3895714553387406e-05,
+      "loss": 0.1018,
+      "step": 87120
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.389437618779946e-05,
+      "loss": 0.1697,
+      "step": 87130
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.389303782221152e-05,
+      "loss": 0.2049,
+      "step": 87140
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.389169945662357e-05,
+      "loss": 0.1827,
+      "step": 87150
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.3890361091035634e-05,
+      "loss": 0.1681,
+      "step": 87160
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.388902272544769e-05,
+      "loss": 0.1127,
+      "step": 87170
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.3887684359859745e-05,
+      "loss": 0.1351,
+      "step": 87180
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.38863459942718e-05,
+      "loss": 0.1679,
+      "step": 87190
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.3885007628683856e-05,
+      "loss": 0.0895,
+      "step": 87200
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.388366926309591e-05,
+      "loss": 0.1232,
+      "step": 87210
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.3882330897507966e-05,
+      "loss": 0.0854,
+      "step": 87220
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.388099253192002e-05,
+      "loss": 0.2943,
+      "step": 87230
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.387965416633208e-05,
+      "loss": 0.0223,
+      "step": 87240
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.387831580074413e-05,
+      "loss": 0.3025,
+      "step": 87250
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.3876977435156194e-05,
+      "loss": 0.1761,
+      "step": 87260
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.387563906956825e-05,
+      "loss": 0.0583,
+      "step": 87270
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.3874300703980305e-05,
+      "loss": 0.173,
+      "step": 87280
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.387296233839236e-05,
+      "loss": 0.1028,
+      "step": 87290
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.3871623972804416e-05,
+      "loss": 0.1335,
+      "step": 87300
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.387028560721647e-05,
+      "loss": 0.2556,
+      "step": 87310
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.3868947241628526e-05,
+      "loss": 0.01,
+      "step": 87320
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.386760887604058e-05,
+      "loss": 0.2505,
+      "step": 87330
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.386627051045264e-05,
+      "loss": 0.2125,
+      "step": 87340
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.386493214486469e-05,
+      "loss": 0.1278,
+      "step": 87350
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.3863593779276754e-05,
+      "loss": 0.1263,
+      "step": 87360
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.386225541368881e-05,
+      "loss": 0.1765,
+      "step": 87370
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.3860917048100865e-05,
+      "loss": 0.0385,
+      "step": 87380
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.385957868251292e-05,
+      "loss": 0.1734,
+      "step": 87390
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.3858240316924976e-05,
+      "loss": 0.0609,
+      "step": 87400
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.385690195133703e-05,
+      "loss": 0.261,
+      "step": 87410
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.3855563585749087e-05,
+      "loss": 0.2723,
+      "step": 87420
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.385422522016114e-05,
+      "loss": 0.1276,
+      "step": 87430
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.38528868545732e-05,
+      "loss": 0.1615,
+      "step": 87440
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.385154848898525e-05,
+      "loss": 0.1022,
+      "step": 87450
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.3850210123397315e-05,
+      "loss": 0.1238,
+      "step": 87460
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.384887175780937e-05,
+      "loss": 0.0991,
+      "step": 87470
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.3847533392221425e-05,
+      "loss": 0.2142,
+      "step": 87480
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.384619502663348e-05,
+      "loss": 0.2103,
+      "step": 87490
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.3844856661045536e-05,
+      "loss": 0.2359,
+      "step": 87500
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.384351829545759e-05,
+      "loss": 0.1305,
+      "step": 87510
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.384217992986965e-05,
+      "loss": 0.177,
+      "step": 87520
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.38408415642817e-05,
+      "loss": 0.1594,
+      "step": 87530
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.383950319869376e-05,
+      "loss": 0.1066,
+      "step": 87540
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.383816483310581e-05,
+      "loss": 0.1496,
+      "step": 87550
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.3836826467517875e-05,
+      "loss": 0.0767,
+      "step": 87560
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.383548810192993e-05,
+      "loss": 0.1071,
+      "step": 87570
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.3834149736341985e-05,
+      "loss": 0.0854,
+      "step": 87580
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.383281137075404e-05,
+      "loss": 0.1516,
+      "step": 87590
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.3831473005166096e-05,
+      "loss": 0.264,
+      "step": 87600
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.383013463957815e-05,
+      "loss": 0.1672,
+      "step": 87610
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.38287962739902e-05,
+      "loss": 0.198,
+      "step": 87620
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.3827457908402255e-05,
+      "loss": 0.1561,
+      "step": 87630
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.382611954281432e-05,
+      "loss": 0.0582,
+      "step": 87640
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.382478117722637e-05,
+      "loss": 0.2885,
+      "step": 87650
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.382344281163843e-05,
+      "loss": 0.1084,
+      "step": 87660
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.3822104446050484e-05,
+      "loss": 0.1821,
+      "step": 87670
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.382076608046254e-05,
+      "loss": 0.127,
+      "step": 87680
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.3819427714874594e-05,
+      "loss": 0.0687,
+      "step": 87690
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.381808934928665e-05,
+      "loss": 0.0406,
+      "step": 87700
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.3816750983698705e-05,
+      "loss": 0.0914,
+      "step": 87710
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.381541261811076e-05,
+      "loss": 0.1327,
+      "step": 87720
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.3814074252522816e-05,
+      "loss": 0.1342,
+      "step": 87730
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.381273588693488e-05,
+      "loss": 0.0115,
+      "step": 87740
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.381139752134693e-05,
+      "loss": 0.1493,
+      "step": 87750
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.381005915575899e-05,
+      "loss": 0.0089,
+      "step": 87760
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.3808720790171044e-05,
+      "loss": 0.1885,
+      "step": 87770
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.38073824245831e-05,
+      "loss": 0.0222,
+      "step": 87780
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.3806044058995154e-05,
+      "loss": 0.2648,
+      "step": 87790
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.380470569340721e-05,
+      "loss": 0.1739,
+      "step": 87800
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.3803367327819265e-05,
+      "loss": 0.0851,
+      "step": 87810
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.380202896223132e-05,
+      "loss": 0.1143,
+      "step": 87820
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.3800690596643376e-05,
+      "loss": 0.0483,
+      "step": 87830
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.379935223105544e-05,
+      "loss": 0.228,
+      "step": 87840
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.379801386546749e-05,
+      "loss": 0.0974,
+      "step": 87850
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.379667549987955e-05,
+      "loss": 0.1635,
+      "step": 87860
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.3795337134291604e-05,
+      "loss": 0.3348,
+      "step": 87870
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.379399876870366e-05,
+      "loss": 0.1344,
+      "step": 87880
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.3792660403115714e-05,
+      "loss": 0.0754,
+      "step": 87890
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.379132203752777e-05,
+      "loss": 0.1854,
+      "step": 87900
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.3789983671939825e-05,
+      "loss": 0.053,
+      "step": 87910
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.378864530635188e-05,
+      "loss": 0.0993,
+      "step": 87920
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.3787306940763936e-05,
+      "loss": 0.0921,
+      "step": 87930
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.3785968575176e-05,
+      "loss": 0.0408,
+      "step": 87940
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.378463020958805e-05,
+      "loss": 0.1607,
+      "step": 87950
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.378329184400011e-05,
+      "loss": 0.1075,
+      "step": 87960
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.3781953478412164e-05,
+      "loss": 0.3213,
+      "step": 87970
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.378061511282422e-05,
+      "loss": 0.1189,
+      "step": 87980
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.3779276747236275e-05,
+      "loss": 0.0114,
+      "step": 87990
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.377793838164833e-05,
+      "loss": 0.0216,
+      "step": 88000
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.3776600016060385e-05,
+      "loss": 0.1224,
+      "step": 88010
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.377526165047244e-05,
+      "loss": 0.07,
+      "step": 88020
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.3773923284884496e-05,
+      "loss": 0.1877,
+      "step": 88030
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.377258491929656e-05,
+      "loss": 0.031,
+      "step": 88040
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.3771246553708613e-05,
+      "loss": 0.2941,
+      "step": 88050
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.376990818812067e-05,
+      "loss": 0.1266,
+      "step": 88060
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.3768569822532724e-05,
+      "loss": 0.097,
+      "step": 88070
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.376723145694478e-05,
+      "loss": 0.0363,
+      "step": 88080
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.3765893091356835e-05,
+      "loss": 0.327,
+      "step": 88090
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.376455472576889e-05,
+      "loss": 0.1836,
+      "step": 88100
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.3763216360180945e-05,
+      "loss": 0.1696,
+      "step": 88110
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.3761877994593e-05,
+      "loss": 0.1585,
+      "step": 88120
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.376053962900506e-05,
+      "loss": 0.0887,
+      "step": 88130
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.375920126341712e-05,
+      "loss": 0.1394,
+      "step": 88140
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.3757862897829174e-05,
+      "loss": 0.2406,
+      "step": 88150
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.375652453224123e-05,
+      "loss": 0.3413,
+      "step": 88160
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.3755186166653284e-05,
+      "loss": 0.1035,
+      "step": 88170
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.375384780106534e-05,
+      "loss": 0.1159,
+      "step": 88180
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.3752509435477395e-05,
+      "loss": 0.1124,
+      "step": 88190
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.375117106988945e-05,
+      "loss": 0.2078,
+      "step": 88200
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.3749832704301506e-05,
+      "loss": 0.1083,
+      "step": 88210
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.374849433871356e-05,
+      "loss": 0.0755,
+      "step": 88220
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.374715597312562e-05,
+      "loss": 0.2505,
+      "step": 88230
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.374581760753768e-05,
+      "loss": 0.0867,
+      "step": 88240
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.3744479241949734e-05,
+      "loss": 0.1838,
+      "step": 88250
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.374314087636179e-05,
+      "loss": 0.2402,
+      "step": 88260
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.3741802510773844e-05,
+      "loss": 0.116,
+      "step": 88270
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.37404641451859e-05,
+      "loss": 0.1646,
+      "step": 88280
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.3739125779597955e-05,
+      "loss": 0.2147,
+      "step": 88290
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.373778741401001e-05,
+      "loss": 0.1212,
+      "step": 88300
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.3736449048422066e-05,
+      "loss": 0.0627,
+      "step": 88310
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.373511068283412e-05,
+      "loss": 0.2792,
+      "step": 88320
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.373377231724618e-05,
+      "loss": 0.2405,
+      "step": 88330
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.373243395165824e-05,
+      "loss": 0.2281,
+      "step": 88340
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.3731095586070294e-05,
+      "loss": 0.0775,
+      "step": 88350
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.372975722048235e-05,
+      "loss": 0.1793,
+      "step": 88360
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.3728418854894405e-05,
+      "loss": 0.4503,
+      "step": 88370
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.372708048930646e-05,
+      "loss": 0.2751,
+      "step": 88380
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.3725742123718515e-05,
+      "loss": 0.387,
+      "step": 88390
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.372440375813057e-05,
+      "loss": 0.0198,
+      "step": 88400
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.3723065392542626e-05,
+      "loss": 0.1662,
+      "step": 88410
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.372172702695468e-05,
+      "loss": 0.2321,
+      "step": 88420
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.372038866136674e-05,
+      "loss": 0.1723,
+      "step": 88430
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.37190502957788e-05,
+      "loss": 0.0809,
+      "step": 88440
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.3717711930190854e-05,
+      "loss": 0.1781,
+      "step": 88450
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.371637356460291e-05,
+      "loss": 0.1915,
+      "step": 88460
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.3715035199014965e-05,
+      "loss": 0.1263,
+      "step": 88470
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.371369683342702e-05,
+      "loss": 0.1305,
+      "step": 88480
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.3712358467839075e-05,
+      "loss": 0.1327,
+      "step": 88490
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.371102010225113e-05,
+      "loss": 0.1882,
+      "step": 88500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.3709681736663186e-05,
+      "loss": 0.2545,
+      "step": 88510
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.370834337107524e-05,
+      "loss": 0.0995,
+      "step": 88520
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.3707005005487304e-05,
+      "loss": 0.1357,
+      "step": 88530
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.370566663989936e-05,
+      "loss": 0.1231,
+      "step": 88540
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.3704328274311414e-05,
+      "loss": 0.2127,
+      "step": 88550
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.370298990872347e-05,
+      "loss": 0.067,
+      "step": 88560
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.3701651543135525e-05,
+      "loss": 0.0757,
+      "step": 88570
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.370031317754758e-05,
+      "loss": 0.2411,
+      "step": 88580
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.3698974811959636e-05,
+      "loss": 0.3534,
+      "step": 88590
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.369763644637169e-05,
+      "loss": 0.118,
+      "step": 88600
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.3696298080783746e-05,
+      "loss": 0.1683,
+      "step": 88610
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.36949597151958e-05,
+      "loss": 0.1217,
+      "step": 88620
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.3693621349607864e-05,
+      "loss": 0.1539,
+      "step": 88630
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.369228298401992e-05,
+      "loss": 0.1031,
+      "step": 88640
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.3690944618431974e-05,
+      "loss": 0.1051,
+      "step": 88650
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.368960625284403e-05,
+      "loss": 0.1902,
+      "step": 88660
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.3688267887256085e-05,
+      "loss": 0.209,
+      "step": 88670
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.368692952166814e-05,
+      "loss": 0.0699,
+      "step": 88680
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.3685591156080196e-05,
+      "loss": 0.0139,
+      "step": 88690
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.368425279049225e-05,
+      "loss": 0.0296,
+      "step": 88700
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.3682914424904306e-05,
+      "loss": 0.127,
+      "step": 88710
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.368157605931636e-05,
+      "loss": 0.0766,
+      "step": 88720
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.3680237693728424e-05,
+      "loss": 0.1022,
+      "step": 88730
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.367889932814048e-05,
+      "loss": 0.0903,
+      "step": 88740
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.3677560962552535e-05,
+      "loss": 0.1216,
+      "step": 88750
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.367622259696459e-05,
+      "loss": 0.1579,
+      "step": 88760
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.3674884231376645e-05,
+      "loss": 0.1897,
+      "step": 88770
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.36735458657887e-05,
+      "loss": 0.1969,
+      "step": 88780
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.3672207500200756e-05,
+      "loss": 0.2665,
+      "step": 88790
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.367086913461281e-05,
+      "loss": 0.0937,
+      "step": 88800
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.3669530769024867e-05,
+      "loss": 0.1139,
+      "step": 88810
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.366819240343692e-05,
+      "loss": 0.1044,
+      "step": 88820
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.3666854037848984e-05,
+      "loss": 0.2449,
+      "step": 88830
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.366551567226104e-05,
+      "loss": 0.3136,
+      "step": 88840
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.3664177306673095e-05,
+      "loss": 0.1761,
+      "step": 88850
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.366283894108515e-05,
+      "loss": 0.1458,
+      "step": 88860
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.3661500575497205e-05,
+      "loss": 0.1298,
+      "step": 88870
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.366016220990926e-05,
+      "loss": 0.3009,
+      "step": 88880
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.3658823844321316e-05,
+      "loss": 0.1459,
+      "step": 88890
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.365748547873337e-05,
+      "loss": 0.0917,
+      "step": 88900
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.365614711314543e-05,
+      "loss": 0.1515,
+      "step": 88910
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.365480874755748e-05,
+      "loss": 0.0378,
+      "step": 88920
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.3653470381969544e-05,
+      "loss": 0.1987,
+      "step": 88930
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.36521320163816e-05,
+      "loss": 0.2314,
+      "step": 88940
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.3650793650793655e-05,
+      "loss": 0.2131,
+      "step": 88950
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.364945528520571e-05,
+      "loss": 0.1317,
+      "step": 88960
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.3648116919617765e-05,
+      "loss": 0.1305,
+      "step": 88970
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.364677855402982e-05,
+      "loss": 0.176,
+      "step": 88980
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.3645440188441876e-05,
+      "loss": 0.1202,
+      "step": 88990
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.364410182285393e-05,
+      "loss": 0.1228,
+      "step": 89000
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.364276345726599e-05,
+      "loss": 0.233,
+      "step": 89010
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.364142509167804e-05,
+      "loss": 0.1248,
+      "step": 89020
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.3640086726090104e-05,
+      "loss": 0.0903,
+      "step": 89030
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.363874836050216e-05,
+      "loss": 0.2579,
+      "step": 89040
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.3637409994914215e-05,
+      "loss": 0.0199,
+      "step": 89050
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.363607162932627e-05,
+      "loss": 0.0393,
+      "step": 89060
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.3634733263738326e-05,
+      "loss": 0.1971,
+      "step": 89070
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.363339489815038e-05,
+      "loss": 0.1203,
+      "step": 89080
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.3632056532562436e-05,
+      "loss": 0.1986,
+      "step": 89090
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.363071816697449e-05,
+      "loss": 0.0974,
+      "step": 89100
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.362937980138655e-05,
+      "loss": 0.1482,
+      "step": 89110
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.36280414357986e-05,
+      "loss": 0.3273,
+      "step": 89120
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.3626703070210664e-05,
+      "loss": 0.0958,
+      "step": 89130
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.362536470462272e-05,
+      "loss": 0.1157,
+      "step": 89140
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.3624026339034775e-05,
+      "loss": 0.1596,
+      "step": 89150
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.362268797344683e-05,
+      "loss": 0.1619,
+      "step": 89160
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.3621349607858886e-05,
+      "loss": 0.1679,
+      "step": 89170
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.362001124227094e-05,
+      "loss": 0.0487,
+      "step": 89180
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.3618672876682996e-05,
+      "loss": 0.1162,
+      "step": 89190
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.361733451109505e-05,
+      "loss": 0.0966,
+      "step": 89200
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.361599614550711e-05,
+      "loss": 0.1413,
+      "step": 89210
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.361465777991916e-05,
+      "loss": 0.0859,
+      "step": 89220
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.3613319414331225e-05,
+      "loss": 0.2383,
+      "step": 89230
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.361198104874328e-05,
+      "loss": 0.1342,
+      "step": 89240
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.3610642683155335e-05,
+      "loss": 0.2104,
+      "step": 89250
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.360930431756739e-05,
+      "loss": 0.0673,
+      "step": 89260
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.3607965951979446e-05,
+      "loss": 0.0733,
+      "step": 89270
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.36066275863915e-05,
+      "loss": 0.1058,
+      "step": 89280
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.360528922080356e-05,
+      "loss": 0.1767,
+      "step": 89290
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.360395085521561e-05,
+      "loss": 0.1895,
+      "step": 89300
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.360261248962767e-05,
+      "loss": 0.4226,
+      "step": 89310
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.360127412403972e-05,
+      "loss": 0.1172,
+      "step": 89320
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.3599935758451785e-05,
+      "loss": 0.1175,
+      "step": 89330
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.359859739286384e-05,
+      "loss": 0.1998,
+      "step": 89340
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.3597259027275895e-05,
+      "loss": 0.1974,
+      "step": 89350
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.359592066168795e-05,
+      "loss": 0.1677,
+      "step": 89360
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.3594582296100006e-05,
+      "loss": 0.0845,
+      "step": 89370
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.359324393051206e-05,
+      "loss": 0.099,
+      "step": 89380
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.359190556492412e-05,
+      "loss": 0.1454,
+      "step": 89390
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.359056719933617e-05,
+      "loss": 0.053,
+      "step": 89400
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.358922883374823e-05,
+      "loss": 0.1586,
+      "step": 89410
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.358789046816029e-05,
+      "loss": 0.0682,
+      "step": 89420
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.3586552102572345e-05,
+      "loss": 0.1282,
+      "step": 89430
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.35852137369844e-05,
+      "loss": 0.123,
+      "step": 89440
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.3583875371396456e-05,
+      "loss": 0.0739,
+      "step": 89450
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.358253700580851e-05,
+      "loss": 0.1971,
+      "step": 89460
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.3581198640220566e-05,
+      "loss": 0.1701,
+      "step": 89470
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.357986027463262e-05,
+      "loss": 0.3956,
+      "step": 89480
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.357852190904468e-05,
+      "loss": 0.1917,
+      "step": 89490
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.357718354345673e-05,
+      "loss": 0.0612,
+      "step": 89500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.357584517786879e-05,
+      "loss": 0.1282,
+      "step": 89510
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.357450681228085e-05,
+      "loss": 0.1975,
+      "step": 89520
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.3573168446692905e-05,
+      "loss": 0.1056,
+      "step": 89530
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.357183008110496e-05,
+      "loss": 0.1114,
+      "step": 89540
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.3570491715517016e-05,
+      "loss": 0.0102,
+      "step": 89550
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.356915334992907e-05,
+      "loss": 0.3291,
+      "step": 89560
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.3567814984341126e-05,
+      "loss": 0.1519,
+      "step": 89570
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.356647661875318e-05,
+      "loss": 0.1022,
+      "step": 89580
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.356513825316524e-05,
+      "loss": 0.1873,
+      "step": 89590
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.356379988757729e-05,
+      "loss": 0.1087,
+      "step": 89600
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.356246152198935e-05,
+      "loss": 0.016,
+      "step": 89610
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.356112315640141e-05,
+      "loss": 0.2463,
+      "step": 89620
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.3559784790813465e-05,
+      "loss": 0.1293,
+      "step": 89630
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.355844642522552e-05,
+      "loss": 0.3741,
+      "step": 89640
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.3557108059637576e-05,
+      "loss": 0.0639,
+      "step": 89650
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.355576969404963e-05,
+      "loss": 0.1331,
+      "step": 89660
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.3554431328461687e-05,
+      "loss": 0.0809,
+      "step": 89670
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.355309296287374e-05,
+      "loss": 0.0422,
+      "step": 89680
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.35517545972858e-05,
+      "loss": 0.161,
+      "step": 89690
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.355041623169785e-05,
+      "loss": 0.1414,
+      "step": 89700
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.354907786610991e-05,
+      "loss": 0.1317,
+      "step": 89710
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.354773950052197e-05,
+      "loss": 0.0539,
+      "step": 89720
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.3546401134934025e-05,
+      "loss": 0.2654,
+      "step": 89730
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.354506276934608e-05,
+      "loss": 0.0157,
+      "step": 89740
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.3543724403758136e-05,
+      "loss": 0.0301,
+      "step": 89750
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.354238603817019e-05,
+      "loss": 0.1854,
+      "step": 89760
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.354104767258225e-05,
+      "loss": 0.2373,
+      "step": 89770
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.35397093069943e-05,
+      "loss": 0.1359,
+      "step": 89780
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.353837094140636e-05,
+      "loss": 0.1162,
+      "step": 89790
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.353703257581841e-05,
+      "loss": 0.1577,
+      "step": 89800
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.353569421023047e-05,
+      "loss": 0.1757,
+      "step": 89810
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.353435584464253e-05,
+      "loss": 0.0462,
+      "step": 89820
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.3533017479054586e-05,
+      "loss": 0.1233,
+      "step": 89830
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.353167911346664e-05,
+      "loss": 0.2973,
+      "step": 89840
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.3530340747878696e-05,
+      "loss": 0.2002,
+      "step": 89850
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.352900238229075e-05,
+      "loss": 0.2811,
+      "step": 89860
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.352766401670281e-05,
+      "loss": 0.2314,
+      "step": 89870
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.352632565111486e-05,
+      "loss": 0.163,
+      "step": 89880
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.352498728552692e-05,
+      "loss": 0.1323,
+      "step": 89890
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.352364891993897e-05,
+      "loss": 0.1056,
+      "step": 89900
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.352231055435103e-05,
+      "loss": 0.1453,
+      "step": 89910
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.352097218876309e-05,
+      "loss": 0.0035,
+      "step": 89920
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.3519633823175146e-05,
+      "loss": 0.1211,
+      "step": 89930
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.35182954575872e-05,
+      "loss": 0.0213,
+      "step": 89940
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.3516957091999256e-05,
+      "loss": 0.1233,
+      "step": 89950
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.351561872641131e-05,
+      "loss": 0.1654,
+      "step": 89960
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.351428036082337e-05,
+      "loss": 0.1513,
+      "step": 89970
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.351294199523542e-05,
+      "loss": 0.1832,
+      "step": 89980
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.351160362964747e-05,
+      "loss": 0.2989,
+      "step": 89990
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.351026526405953e-05,
+      "loss": 0.1272,
+      "step": 90000
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.350892689847159e-05,
+      "loss": 0.3105,
+      "step": 90010
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.3507588532883644e-05,
+      "loss": 0.2016,
+      "step": 90020
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.35062501672957e-05,
+      "loss": 0.0752,
+      "step": 90030
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.3504911801707754e-05,
+      "loss": 0.0813,
+      "step": 90040
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.350357343611981e-05,
+      "loss": 0.3127,
+      "step": 90050
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.3502235070531865e-05,
+      "loss": 0.116,
+      "step": 90060
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.350089670494392e-05,
+      "loss": 0.1515,
+      "step": 90070
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.3499558339355976e-05,
+      "loss": 0.1612,
+      "step": 90080
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.349821997376803e-05,
+      "loss": 0.214,
+      "step": 90090
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.349688160818009e-05,
+      "loss": 0.1133,
+      "step": 90100
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.349554324259215e-05,
+      "loss": 0.0378,
+      "step": 90110
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.3494204877004204e-05,
+      "loss": 0.0741,
+      "step": 90120
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.349286651141626e-05,
+      "loss": 0.2636,
+      "step": 90130
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.3491528145828315e-05,
+      "loss": 0.2128,
+      "step": 90140
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.349018978024037e-05,
+      "loss": 0.062,
+      "step": 90150
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.3488851414652425e-05,
+      "loss": 0.2106,
+      "step": 90160
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.348751304906448e-05,
+      "loss": 0.247,
+      "step": 90170
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.3486174683476536e-05,
+      "loss": 0.335,
+      "step": 90180
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.348483631788859e-05,
+      "loss": 0.0063,
+      "step": 90190
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.348349795230065e-05,
+      "loss": 0.2837,
+      "step": 90200
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.348215958671271e-05,
+      "loss": 0.1635,
+      "step": 90210
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.3480821221124764e-05,
+      "loss": 0.198,
+      "step": 90220
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.347948285553682e-05,
+      "loss": 0.2282,
+      "step": 90230
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.3478144489948875e-05,
+      "loss": 0.2169,
+      "step": 90240
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.347680612436093e-05,
+      "loss": 0.0695,
+      "step": 90250
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.3475467758772985e-05,
+      "loss": 0.2357,
+      "step": 90260
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.347412939318504e-05,
+      "loss": 0.1958,
+      "step": 90270
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.3472791027597096e-05,
+      "loss": 0.2004,
+      "step": 90280
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.347145266200915e-05,
+      "loss": 0.111,
+      "step": 90290
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.3470114296421213e-05,
+      "loss": 0.1315,
+      "step": 90300
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.346877593083327e-05,
+      "loss": 0.0969,
+      "step": 90310
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.3467437565245324e-05,
+      "loss": 0.1549,
+      "step": 90320
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.346609919965738e-05,
+      "loss": 0.1215,
+      "step": 90330
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.3464760834069435e-05,
+      "loss": 0.0768,
+      "step": 90340
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.346342246848149e-05,
+      "loss": 0.1714,
+      "step": 90350
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.3462084102893546e-05,
+      "loss": 0.1134,
+      "step": 90360
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.34607457373056e-05,
+      "loss": 0.2231,
+      "step": 90370
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.3459407371717656e-05,
+      "loss": 0.1513,
+      "step": 90380
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.345806900612971e-05,
+      "loss": 0.115,
+      "step": 90390
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.3456730640541774e-05,
+      "loss": 0.1981,
+      "step": 90400
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.345539227495383e-05,
+      "loss": 0.1775,
+      "step": 90410
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.3454053909365884e-05,
+      "loss": 0.1794,
+      "step": 90420
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.345271554377794e-05,
+      "loss": 0.1644,
+      "step": 90430
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.3451377178189995e-05,
+      "loss": 0.1454,
+      "step": 90440
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.345003881260205e-05,
+      "loss": 0.019,
+      "step": 90450
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.3448700447014106e-05,
+      "loss": 0.1343,
+      "step": 90460
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.344736208142616e-05,
+      "loss": 0.3144,
+      "step": 90470
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.3446023715838216e-05,
+      "loss": 0.1182,
+      "step": 90480
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.344468535025027e-05,
+      "loss": 0.1733,
+      "step": 90490
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.3443346984662334e-05,
+      "loss": 0.1226,
+      "step": 90500
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.344200861907439e-05,
+      "loss": 0.1691,
+      "step": 90510
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.3440670253486444e-05,
+      "loss": 0.2173,
+      "step": 90520
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.34393318878985e-05,
+      "loss": 0.1357,
+      "step": 90530
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.3437993522310555e-05,
+      "loss": 0.1064,
+      "step": 90540
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.343665515672261e-05,
+      "loss": 0.3441,
+      "step": 90550
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.3435316791134666e-05,
+      "loss": 0.1463,
+      "step": 90560
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.343397842554672e-05,
+      "loss": 0.1283,
+      "step": 90570
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.3432640059958777e-05,
+      "loss": 0.0802,
+      "step": 90580
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.343130169437083e-05,
+      "loss": 0.1058,
+      "step": 90590
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.3429963328782894e-05,
+      "loss": 0.1287,
+      "step": 90600
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.342862496319495e-05,
+      "loss": 0.0208,
+      "step": 90610
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.3427286597607005e-05,
+      "loss": 0.2861,
+      "step": 90620
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.342594823201906e-05,
+      "loss": 0.1299,
+      "step": 90630
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.3424609866431115e-05,
+      "loss": 0.0922,
+      "step": 90640
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.342327150084317e-05,
+      "loss": 0.1023,
+      "step": 90650
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.3421933135255226e-05,
+      "loss": 0.0476,
+      "step": 90660
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.342059476966728e-05,
+      "loss": 0.1298,
+      "step": 90670
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.341925640407934e-05,
+      "loss": 0.0958,
+      "step": 90680
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.341791803849139e-05,
+      "loss": 0.0793,
+      "step": 90690
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.3416579672903454e-05,
+      "loss": 0.1001,
+      "step": 90700
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.341524130731551e-05,
+      "loss": 0.2145,
+      "step": 90710
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.3413902941727565e-05,
+      "loss": 0.0848,
+      "step": 90720
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.341256457613962e-05,
+      "loss": 0.1308,
+      "step": 90730
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.3411226210551675e-05,
+      "loss": 0.1349,
+      "step": 90740
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.340988784496373e-05,
+      "loss": 0.1896,
+      "step": 90750
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.3408549479375786e-05,
+      "loss": 0.3061,
+      "step": 90760
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.340721111378784e-05,
+      "loss": 0.3044,
+      "step": 90770
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.34058727481999e-05,
+      "loss": 0.1588,
+      "step": 90780
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.340453438261195e-05,
+      "loss": 0.0619,
+      "step": 90790
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.3403196017024014e-05,
+      "loss": 0.3704,
+      "step": 90800
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.340185765143607e-05,
+      "loss": 0.1465,
+      "step": 90810
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.3400519285848125e-05,
+      "loss": 0.2696,
+      "step": 90820
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.339918092026018e-05,
+      "loss": 0.0516,
+      "step": 90830
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.3397842554672236e-05,
+      "loss": 0.1436,
+      "step": 90840
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.339650418908429e-05,
+      "loss": 0.0213,
+      "step": 90850
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.3395165823496346e-05,
+      "loss": 0.161,
+      "step": 90860
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.33938274579084e-05,
+      "loss": 0.0668,
+      "step": 90870
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.339248909232046e-05,
+      "loss": 0.2264,
+      "step": 90880
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.339115072673252e-05,
+      "loss": 0.1756,
+      "step": 90890
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.3389812361144574e-05,
+      "loss": 0.2418,
+      "step": 90900
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.338847399555663e-05,
+      "loss": 0.0732,
+      "step": 90910
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.3387135629968685e-05,
+      "loss": 0.2162,
+      "step": 90920
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.338579726438074e-05,
+      "loss": 0.1476,
+      "step": 90930
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.3384458898792796e-05,
+      "loss": 0.237,
+      "step": 90940
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.338312053320485e-05,
+      "loss": 0.1306,
+      "step": 90950
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.3381782167616906e-05,
+      "loss": 0.0087,
+      "step": 90960
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.338044380202896e-05,
+      "loss": 0.2287,
+      "step": 90970
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.337910543644102e-05,
+      "loss": 0.0892,
+      "step": 90980
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.337776707085308e-05,
+      "loss": 0.2011,
+      "step": 90990
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.3376428705265135e-05,
+      "loss": 0.1468,
+      "step": 91000
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.337509033967719e-05,
+      "loss": 0.1272,
+      "step": 91010
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.3373751974089245e-05,
+      "loss": 0.2603,
+      "step": 91020
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.33724136085013e-05,
+      "loss": 0.1362,
+      "step": 91030
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.3371075242913356e-05,
+      "loss": 0.2381,
+      "step": 91040
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.336973687732541e-05,
+      "loss": 0.1956,
+      "step": 91050
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.3368398511737467e-05,
+      "loss": 0.0811,
+      "step": 91060
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.336706014614952e-05,
+      "loss": 0.1406,
+      "step": 91070
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.336572178056158e-05,
+      "loss": 0.1584,
+      "step": 91080
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.336438341497364e-05,
+      "loss": 0.0664,
+      "step": 91090
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.3363045049385695e-05,
+      "loss": 0.0407,
+      "step": 91100
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.336170668379775e-05,
+      "loss": 0.1803,
+      "step": 91110
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.3360368318209805e-05,
+      "loss": 0.1913,
+      "step": 91120
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.335902995262186e-05,
+      "loss": 0.3514,
+      "step": 91130
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.3357691587033916e-05,
+      "loss": 0.2752,
+      "step": 91140
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.335635322144597e-05,
+      "loss": 0.0097,
+      "step": 91150
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.335501485585803e-05,
+      "loss": 0.1859,
+      "step": 91160
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.335367649027008e-05,
+      "loss": 0.2992,
+      "step": 91170
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.335233812468214e-05,
+      "loss": 0.2964,
+      "step": 91180
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.33509997590942e-05,
+      "loss": 0.0603,
+      "step": 91190
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.3349661393506255e-05,
+      "loss": 0.1033,
+      "step": 91200
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.334832302791831e-05,
+      "loss": 0.2531,
+      "step": 91210
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.3346984662330366e-05,
+      "loss": 0.0674,
+      "step": 91220
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.334564629674242e-05,
+      "loss": 0.0861,
+      "step": 91230
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.3344307931154476e-05,
+      "loss": 0.2444,
+      "step": 91240
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.334296956556653e-05,
+      "loss": 0.0623,
+      "step": 91250
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.334163119997859e-05,
+      "loss": 0.1101,
+      "step": 91260
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.334029283439064e-05,
+      "loss": 0.1271,
+      "step": 91270
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.33389544688027e-05,
+      "loss": 0.0611,
+      "step": 91280
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.333761610321476e-05,
+      "loss": 0.1155,
+      "step": 91290
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.3336277737626815e-05,
+      "loss": 0.0796,
+      "step": 91300
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.333493937203887e-05,
+      "loss": 0.1501,
+      "step": 91310
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.3333601006450926e-05,
+      "loss": 0.0114,
+      "step": 91320
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": {
+        "accuracy": 0.9551942186088528
+      },
+      "eval_f1": {
+        "f1": 0.954959353992539
+      },
+      "eval_loss": 0.19532862305641174,
+      "eval_precision": {
+        "precision": 0.9555671952457011
+      },
+      "eval_recall": {
+        "recall": 0.9550120730050532
+      },
+      "eval_runtime": 429.5088,
+      "eval_samples_per_second": 38.66,
+      "eval_steps_per_second": 19.331,
+      "step": 91327
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.333226264086298e-05,
+      "loss": 0.1423,
+      "step": 91330
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.3330924275275036e-05,
+      "loss": 0.0063,
+      "step": 91340
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.332958590968709e-05,
+      "loss": 0.0362,
+      "step": 91350
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.332824754409915e-05,
+      "loss": 0.1192,
+      "step": 91360
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.33269091785112e-05,
+      "loss": 0.3259,
+      "step": 91370
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.332557081292326e-05,
+      "loss": 0.0698,
+      "step": 91380
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.332423244733532e-05,
+      "loss": 0.2779,
+      "step": 91390
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.3322894081747375e-05,
+      "loss": 0.1451,
+      "step": 91400
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.332155571615943e-05,
+      "loss": 0.2444,
+      "step": 91410
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.3320217350571486e-05,
+      "loss": 0.0296,
+      "step": 91420
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.331887898498354e-05,
+      "loss": 0.0808,
+      "step": 91430
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.3317540619395597e-05,
+      "loss": 0.1031,
+      "step": 91440
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.331620225380765e-05,
+      "loss": 0.1236,
+      "step": 91450
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.331486388821971e-05,
+      "loss": 0.2784,
+      "step": 91460
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.331352552263176e-05,
+      "loss": 0.0667,
+      "step": 91470
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.331218715704382e-05,
+      "loss": 0.0092,
+      "step": 91480
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.331084879145588e-05,
+      "loss": 0.1555,
+      "step": 91490
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.3309510425867935e-05,
+      "loss": 0.0995,
+      "step": 91500
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.330817206027999e-05,
+      "loss": 0.1253,
+      "step": 91510
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.3306833694692046e-05,
+      "loss": 0.0812,
+      "step": 91520
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.33054953291041e-05,
+      "loss": 0.0127,
+      "step": 91530
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.330415696351616e-05,
+      "loss": 0.0971,
+      "step": 91540
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.330281859792821e-05,
+      "loss": 0.1972,
+      "step": 91550
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.330148023234027e-05,
+      "loss": 0.0975,
+      "step": 91560
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.330014186675232e-05,
+      "loss": 0.2394,
+      "step": 91570
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.329880350116438e-05,
+      "loss": 0.2442,
+      "step": 91580
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.329746513557644e-05,
+      "loss": 0.2164,
+      "step": 91590
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.3296126769988495e-05,
+      "loss": 0.0735,
+      "step": 91600
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.329478840440055e-05,
+      "loss": 0.1909,
+      "step": 91610
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.3293450038812606e-05,
+      "loss": 0.131,
+      "step": 91620
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.329211167322466e-05,
+      "loss": 0.2757,
+      "step": 91630
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.329077330763672e-05,
+      "loss": 0.1827,
+      "step": 91640
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.328943494204877e-05,
+      "loss": 0.2003,
+      "step": 91650
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.328809657646083e-05,
+      "loss": 0.1213,
+      "step": 91660
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.328675821087288e-05,
+      "loss": 0.0779,
+      "step": 91670
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.328541984528494e-05,
+      "loss": 0.1038,
+      "step": 91680
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.3284081479697e-05,
+      "loss": 0.1453,
+      "step": 91690
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.3282743114109056e-05,
+      "loss": 0.0792,
+      "step": 91700
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.328140474852111e-05,
+      "loss": 0.054,
+      "step": 91710
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.3280066382933166e-05,
+      "loss": 0.2406,
+      "step": 91720
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.327872801734522e-05,
+      "loss": 0.1649,
+      "step": 91730
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.327738965175728e-05,
+      "loss": 0.0441,
+      "step": 91740
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.327605128616933e-05,
+      "loss": 0.137,
+      "step": 91750
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.327471292058139e-05,
+      "loss": 0.1852,
+      "step": 91760
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.327337455499344e-05,
+      "loss": 0.1268,
+      "step": 91770
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.32720361894055e-05,
+      "loss": 0.156,
+      "step": 91780
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.327069782381756e-05,
+      "loss": 0.086,
+      "step": 91790
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.3269359458229616e-05,
+      "loss": 0.3199,
+      "step": 91800
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.326802109264167e-05,
+      "loss": 0.0635,
+      "step": 91810
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.3266682727053726e-05,
+      "loss": 0.1129,
+      "step": 91820
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.326534436146578e-05,
+      "loss": 0.1325,
+      "step": 91830
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.326400599587784e-05,
+      "loss": 0.2299,
+      "step": 91840
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.326266763028989e-05,
+      "loss": 0.1114,
+      "step": 91850
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.326132926470195e-05,
+      "loss": 0.1814,
+      "step": 91860
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.3259990899114e-05,
+      "loss": 0.1602,
+      "step": 91870
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.325865253352606e-05,
+      "loss": 0.1361,
+      "step": 91880
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.325731416793812e-05,
+      "loss": 0.1869,
+      "step": 91890
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.3255975802350176e-05,
+      "loss": 0.2908,
+      "step": 91900
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.325463743676223e-05,
+      "loss": 0.2345,
+      "step": 91910
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.325329907117429e-05,
+      "loss": 0.144,
+      "step": 91920
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.325196070558634e-05,
+      "loss": 0.1791,
+      "step": 91930
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.32506223399984e-05,
+      "loss": 0.3312,
+      "step": 91940
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.324928397441045e-05,
+      "loss": 0.2083,
+      "step": 91950
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.324794560882251e-05,
+      "loss": 0.0579,
+      "step": 91960
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.324660724323456e-05,
+      "loss": 0.042,
+      "step": 91970
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.324526887764662e-05,
+      "loss": 0.0096,
+      "step": 91980
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.324393051205868e-05,
+      "loss": 0.0518,
+      "step": 91990
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.3242592146470736e-05,
+      "loss": 0.1919,
+      "step": 92000
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.324125378088279e-05,
+      "loss": 0.0571,
+      "step": 92010
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.323991541529485e-05,
+      "loss": 0.3159,
+      "step": 92020
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.32385770497069e-05,
+      "loss": 0.2312,
+      "step": 92030
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.323723868411896e-05,
+      "loss": 0.135,
+      "step": 92040
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.323590031853101e-05,
+      "loss": 0.0373,
+      "step": 92050
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.323456195294307e-05,
+      "loss": 0.0768,
+      "step": 92060
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.3233223587355123e-05,
+      "loss": 0.2078,
+      "step": 92070
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.323188522176718e-05,
+      "loss": 0.1192,
+      "step": 92080
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.323054685617924e-05,
+      "loss": 0.2478,
+      "step": 92090
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.3229208490591296e-05,
+      "loss": 0.0657,
+      "step": 92100
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.322787012500335e-05,
+      "loss": 0.2766,
+      "step": 92110
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.322653175941541e-05,
+      "loss": 0.2373,
+      "step": 92120
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.322519339382746e-05,
+      "loss": 0.0572,
+      "step": 92130
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.322385502823952e-05,
+      "loss": 0.105,
+      "step": 92140
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.322251666265157e-05,
+      "loss": 0.1909,
+      "step": 92150
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.322117829706363e-05,
+      "loss": 0.1935,
+      "step": 92160
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.3219839931475684e-05,
+      "loss": 0.0981,
+      "step": 92170
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.3218501565887746e-05,
+      "loss": 0.3706,
+      "step": 92180
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.32171632002998e-05,
+      "loss": 0.1542,
+      "step": 92190
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.3215824834711856e-05,
+      "loss": 0.2222,
+      "step": 92200
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.321448646912391e-05,
+      "loss": 0.1086,
+      "step": 92210
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.321314810353597e-05,
+      "loss": 0.0634,
+      "step": 92220
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.321180973794802e-05,
+      "loss": 0.1622,
+      "step": 92230
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.321047137236008e-05,
+      "loss": 0.1334,
+      "step": 92240
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.320913300677213e-05,
+      "loss": 0.0316,
+      "step": 92250
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.320779464118419e-05,
+      "loss": 0.1791,
+      "step": 92260
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.3206456275596244e-05,
+      "loss": 0.1207,
+      "step": 92270
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.3205117910008306e-05,
+      "loss": 0.2186,
+      "step": 92280
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.320377954442036e-05,
+      "loss": 0.2578,
+      "step": 92290
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.3202441178832417e-05,
+      "loss": 0.1915,
+      "step": 92300
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.320110281324447e-05,
+      "loss": 0.0879,
+      "step": 92310
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.319976444765653e-05,
+      "loss": 0.0532,
+      "step": 92320
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.319842608206858e-05,
+      "loss": 0.0048,
+      "step": 92330
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.319708771648064e-05,
+      "loss": 0.1171,
+      "step": 92340
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.3195749350892686e-05,
+      "loss": 0.1202,
+      "step": 92350
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.319441098530474e-05,
+      "loss": 0.071,
+      "step": 92360
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.3193072619716804e-05,
+      "loss": 0.1244,
+      "step": 92370
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.319173425412886e-05,
+      "loss": 0.2817,
+      "step": 92380
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.3190395888540915e-05,
+      "loss": 0.0995,
+      "step": 92390
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.318905752295297e-05,
+      "loss": 0.1084,
+      "step": 92400
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.3187719157365025e-05,
+      "loss": 0.037,
+      "step": 92410
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.318638079177708e-05,
+      "loss": 0.0664,
+      "step": 92420
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.3185042426189136e-05,
+      "loss": 0.1196,
+      "step": 92430
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.318370406060119e-05,
+      "loss": 0.1216,
+      "step": 92440
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.318236569501325e-05,
+      "loss": 0.1721,
+      "step": 92450
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.318102732942531e-05,
+      "loss": 0.0127,
+      "step": 92460
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.3179688963837364e-05,
+      "loss": 0.1287,
+      "step": 92470
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.317835059824942e-05,
+      "loss": 0.0634,
+      "step": 92480
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.3177012232661475e-05,
+      "loss": 0.0563,
+      "step": 92490
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.317567386707353e-05,
+      "loss": 0.2869,
+      "step": 92500
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.3174335501485585e-05,
+      "loss": 0.0981,
+      "step": 92510
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.317299713589764e-05,
+      "loss": 0.0465,
+      "step": 92520
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.3171658770309696e-05,
+      "loss": 0.0143,
+      "step": 92530
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.317032040472175e-05,
+      "loss": 0.0743,
+      "step": 92540
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.316898203913381e-05,
+      "loss": 0.1355,
+      "step": 92550
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.316764367354587e-05,
+      "loss": 0.085,
+      "step": 92560
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.3166305307957924e-05,
+      "loss": 0.2393,
+      "step": 92570
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.316496694236998e-05,
+      "loss": 0.1154,
+      "step": 92580
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.3163628576782035e-05,
+      "loss": 0.184,
+      "step": 92590
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.316229021119409e-05,
+      "loss": 0.3044,
+      "step": 92600
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.3160951845606146e-05,
+      "loss": 0.1126,
+      "step": 92610
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.31596134800182e-05,
+      "loss": 0.2641,
+      "step": 92620
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.3158275114430256e-05,
+      "loss": 0.2849,
+      "step": 92630
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.315693674884231e-05,
+      "loss": 0.1014,
+      "step": 92640
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.315559838325437e-05,
+      "loss": 0.0914,
+      "step": 92650
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.315426001766643e-05,
+      "loss": 0.2623,
+      "step": 92660
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.3152921652078484e-05,
+      "loss": 0.0564,
+      "step": 92670
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.315158328649054e-05,
+      "loss": 0.0955,
+      "step": 92680
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.3150244920902595e-05,
+      "loss": 0.0508,
+      "step": 92690
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.314890655531465e-05,
+      "loss": 0.1883,
+      "step": 92700
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.3147568189726706e-05,
+      "loss": 0.0299,
+      "step": 92710
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.314622982413876e-05,
+      "loss": 0.1412,
+      "step": 92720
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.3144891458550816e-05,
+      "loss": 0.159,
+      "step": 92730
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.314355309296287e-05,
+      "loss": 0.2267,
+      "step": 92740
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.314221472737493e-05,
+      "loss": 0.2705,
+      "step": 92750
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.314087636178699e-05,
+      "loss": 0.0314,
+      "step": 92760
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.3139537996199045e-05,
+      "loss": 0.0841,
+      "step": 92770
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.31381996306111e-05,
+      "loss": 0.0485,
+      "step": 92780
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.3136861265023155e-05,
+      "loss": 0.1243,
+      "step": 92790
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.313552289943521e-05,
+      "loss": 0.2219,
+      "step": 92800
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.3134184533847266e-05,
+      "loss": 0.2275,
+      "step": 92810
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.313284616825932e-05,
+      "loss": 0.0894,
+      "step": 92820
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.3131507802671377e-05,
+      "loss": 0.2015,
+      "step": 92830
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.313016943708343e-05,
+      "loss": 0.0364,
+      "step": 92840
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.312883107149549e-05,
+      "loss": 0.0196,
+      "step": 92850
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.312749270590755e-05,
+      "loss": 0.0366,
+      "step": 92860
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.3126154340319605e-05,
+      "loss": 0.0075,
+      "step": 92870
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.312481597473166e-05,
+      "loss": 0.1547,
+      "step": 92880
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.3123477609143715e-05,
+      "loss": 0.0689,
+      "step": 92890
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.312213924355577e-05,
+      "loss": 0.0415,
+      "step": 92900
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.3120800877967826e-05,
+      "loss": 0.1536,
+      "step": 92910
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.311946251237988e-05,
+      "loss": 0.0249,
+      "step": 92920
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.311812414679194e-05,
+      "loss": 0.2033,
+      "step": 92930
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.311678578120399e-05,
+      "loss": 0.2469,
+      "step": 92940
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.311544741561605e-05,
+      "loss": 0.1542,
+      "step": 92950
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.311410905002811e-05,
+      "loss": 0.0752,
+      "step": 92960
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.3112770684440165e-05,
+      "loss": 0.2869,
+      "step": 92970
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.311143231885222e-05,
+      "loss": 0.0849,
+      "step": 92980
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.3110093953264276e-05,
+      "loss": 0.0923,
+      "step": 92990
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.310875558767633e-05,
+      "loss": 0.0805,
+      "step": 93000
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.3107417222088386e-05,
+      "loss": 0.1949,
+      "step": 93010
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.310607885650044e-05,
+      "loss": 0.0381,
+      "step": 93020
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.31047404909125e-05,
+      "loss": 0.1546,
+      "step": 93030
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.310340212532455e-05,
+      "loss": 0.1183,
+      "step": 93040
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.310206375973661e-05,
+      "loss": 0.1082,
+      "step": 93050
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.310072539414867e-05,
+      "loss": 0.1254,
+      "step": 93060
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.3099387028560725e-05,
+      "loss": 0.1983,
+      "step": 93070
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.309804866297278e-05,
+      "loss": 0.1456,
+      "step": 93080
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.3096710297384836e-05,
+      "loss": 0.1332,
+      "step": 93090
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.309537193179689e-05,
+      "loss": 0.1463,
+      "step": 93100
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.3094033566208946e-05,
+      "loss": 0.0718,
+      "step": 93110
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.3092695200621e-05,
+      "loss": 0.0449,
+      "step": 93120
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.309135683503306e-05,
+      "loss": 0.1598,
+      "step": 93130
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.309001846944511e-05,
+      "loss": 0.1,
+      "step": 93140
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.308868010385717e-05,
+      "loss": 0.1837,
+      "step": 93150
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.308734173826923e-05,
+      "loss": 0.1198,
+      "step": 93160
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.3086003372681285e-05,
+      "loss": 0.0455,
+      "step": 93170
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.308466500709334e-05,
+      "loss": 0.083,
+      "step": 93180
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.3083326641505396e-05,
+      "loss": 0.135,
+      "step": 93190
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.308198827591745e-05,
+      "loss": 0.2508,
+      "step": 93200
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.3080649910329506e-05,
+      "loss": 0.0591,
+      "step": 93210
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.307931154474156e-05,
+      "loss": 0.1326,
+      "step": 93220
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.307797317915362e-05,
+      "loss": 0.1651,
+      "step": 93230
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.307663481356567e-05,
+      "loss": 0.1889,
+      "step": 93240
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.307529644797773e-05,
+      "loss": 0.2729,
+      "step": 93250
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.307395808238979e-05,
+      "loss": 0.1644,
+      "step": 93260
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.3072619716801845e-05,
+      "loss": 0.095,
+      "step": 93270
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.30712813512139e-05,
+      "loss": 0.1276,
+      "step": 93280
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.3069942985625956e-05,
+      "loss": 0.0516,
+      "step": 93290
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.306860462003801e-05,
+      "loss": 0.098,
+      "step": 93300
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.306726625445007e-05,
+      "loss": 0.3269,
+      "step": 93310
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.306592788886212e-05,
+      "loss": 0.2269,
+      "step": 93320
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.306458952327418e-05,
+      "loss": 0.3479,
+      "step": 93330
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.306325115768623e-05,
+      "loss": 0.2143,
+      "step": 93340
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.306191279209829e-05,
+      "loss": 0.2001,
+      "step": 93350
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.306057442651035e-05,
+      "loss": 0.0978,
+      "step": 93360
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.3059236060922405e-05,
+      "loss": 0.1056,
+      "step": 93370
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.305789769533446e-05,
+      "loss": 0.0678,
+      "step": 93380
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.3056559329746516e-05,
+      "loss": 0.1292,
+      "step": 93390
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.305522096415857e-05,
+      "loss": 0.1091,
+      "step": 93400
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.305388259857063e-05,
+      "loss": 0.1539,
+      "step": 93410
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.305254423298268e-05,
+      "loss": 0.1369,
+      "step": 93420
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.305120586739474e-05,
+      "loss": 0.1899,
+      "step": 93430
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.304986750180679e-05,
+      "loss": 0.1632,
+      "step": 93440
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.304852913621885e-05,
+      "loss": 0.1462,
+      "step": 93450
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.304719077063091e-05,
+      "loss": 0.3821,
+      "step": 93460
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.3045852405042966e-05,
+      "loss": 0.0992,
+      "step": 93470
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.304451403945502e-05,
+      "loss": 0.1924,
+      "step": 93480
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.3043175673867076e-05,
+      "loss": 0.0101,
+      "step": 93490
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.304183730827913e-05,
+      "loss": 0.1313,
+      "step": 93500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.304049894269119e-05,
+      "loss": 0.061,
+      "step": 93510
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.303916057710324e-05,
+      "loss": 0.0726,
+      "step": 93520
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.30378222115153e-05,
+      "loss": 0.0278,
+      "step": 93530
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.303648384592735e-05,
+      "loss": 0.0245,
+      "step": 93540
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.303514548033941e-05,
+      "loss": 0.0212,
+      "step": 93550
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.303380711475147e-05,
+      "loss": 0.3313,
+      "step": 93560
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.3032468749163526e-05,
+      "loss": 0.1333,
+      "step": 93570
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.303113038357558e-05,
+      "loss": 0.1124,
+      "step": 93580
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.3029792017987636e-05,
+      "loss": 0.0861,
+      "step": 93590
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.302845365239969e-05,
+      "loss": 0.0829,
+      "step": 93600
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.302711528681175e-05,
+      "loss": 0.3108,
+      "step": 93610
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.30257769212238e-05,
+      "loss": 0.0477,
+      "step": 93620
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.302443855563586e-05,
+      "loss": 0.11,
+      "step": 93630
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.302310019004791e-05,
+      "loss": 0.0957,
+      "step": 93640
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.302176182445997e-05,
+      "loss": 0.0306,
+      "step": 93650
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.302042345887203e-05,
+      "loss": 0.0054,
+      "step": 93660
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.3019085093284086e-05,
+      "loss": 0.2887,
+      "step": 93670
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.301774672769614e-05,
+      "loss": 0.4706,
+      "step": 93680
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.3016408362108197e-05,
+      "loss": 0.0832,
+      "step": 93690
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.301506999652025e-05,
+      "loss": 0.2895,
+      "step": 93700
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.301373163093231e-05,
+      "loss": 0.0155,
+      "step": 93710
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.301239326534436e-05,
+      "loss": 0.0171,
+      "step": 93720
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.301105489975642e-05,
+      "loss": 0.1292,
+      "step": 93730
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.300971653416847e-05,
+      "loss": 0.0798,
+      "step": 93740
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.3008378168580535e-05,
+      "loss": 0.0114,
+      "step": 93750
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.300703980299259e-05,
+      "loss": 0.1163,
+      "step": 93760
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.3005701437404646e-05,
+      "loss": 0.0899,
+      "step": 93770
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.30043630718167e-05,
+      "loss": 0.1517,
+      "step": 93780
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.300302470622876e-05,
+      "loss": 0.1051,
+      "step": 93790
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.300168634064081e-05,
+      "loss": 0.1027,
+      "step": 93800
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.300034797505287e-05,
+      "loss": 0.2987,
+      "step": 93810
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.299900960946492e-05,
+      "loss": 0.0685,
+      "step": 93820
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.299767124387698e-05,
+      "loss": 0.1937,
+      "step": 93830
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.2996332878289033e-05,
+      "loss": 0.2401,
+      "step": 93840
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.2994994512701096e-05,
+      "loss": 0.1135,
+      "step": 93850
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.299365614711315e-05,
+      "loss": 0.2939,
+      "step": 93860
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.2992317781525206e-05,
+      "loss": 0.0697,
+      "step": 93870
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.299097941593726e-05,
+      "loss": 0.1108,
+      "step": 93880
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.298964105034932e-05,
+      "loss": 0.2707,
+      "step": 93890
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.298830268476137e-05,
+      "loss": 0.0958,
+      "step": 93900
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.298696431917343e-05,
+      "loss": 0.2049,
+      "step": 93910
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.298562595358548e-05,
+      "loss": 0.0904,
+      "step": 93920
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.298428758799754e-05,
+      "loss": 0.1364,
+      "step": 93930
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.2982949222409594e-05,
+      "loss": 0.1733,
+      "step": 93940
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.2981610856821656e-05,
+      "loss": 0.1714,
+      "step": 93950
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.298027249123371e-05,
+      "loss": 0.2172,
+      "step": 93960
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.2978934125645766e-05,
+      "loss": 0.1311,
+      "step": 93970
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.297759576005782e-05,
+      "loss": 0.1004,
+      "step": 93980
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.297625739446988e-05,
+      "loss": 0.1902,
+      "step": 93990
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.297491902888193e-05,
+      "loss": 0.0303,
+      "step": 94000
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.297358066329399e-05,
+      "loss": 0.0653,
+      "step": 94010
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.297224229770604e-05,
+      "loss": 0.1392,
+      "step": 94020
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.29709039321181e-05,
+      "loss": 0.2015,
+      "step": 94030
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.2969565566530154e-05,
+      "loss": 0.1623,
+      "step": 94040
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.2968227200942216e-05,
+      "loss": 0.0568,
+      "step": 94050
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.296688883535427e-05,
+      "loss": 0.1134,
+      "step": 94060
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.2965550469766327e-05,
+      "loss": 0.0449,
+      "step": 94070
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.296421210417838e-05,
+      "loss": 0.3177,
+      "step": 94080
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.296287373859044e-05,
+      "loss": 0.1542,
+      "step": 94090
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.296153537300249e-05,
+      "loss": 0.1382,
+      "step": 94100
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.296019700741455e-05,
+      "loss": 0.0107,
+      "step": 94110
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.29588586418266e-05,
+      "loss": 0.1884,
+      "step": 94120
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.295752027623866e-05,
+      "loss": 0.0457,
+      "step": 94130
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.2956181910650714e-05,
+      "loss": 0.0216,
+      "step": 94140
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.2954843545062776e-05,
+      "loss": 0.0226,
+      "step": 94150
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.295350517947483e-05,
+      "loss": 0.1166,
+      "step": 94160
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.295216681388689e-05,
+      "loss": 0.0953,
+      "step": 94170
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.295082844829894e-05,
+      "loss": 0.125,
+      "step": 94180
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.2949490082711e-05,
+      "loss": 0.1951,
+      "step": 94190
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.294815171712305e-05,
+      "loss": 0.0281,
+      "step": 94200
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.294681335153511e-05,
+      "loss": 0.3484,
+      "step": 94210
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.294547498594716e-05,
+      "loss": 0.1113,
+      "step": 94220
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.294413662035922e-05,
+      "loss": 0.036,
+      "step": 94230
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.2942798254771274e-05,
+      "loss": 0.1316,
+      "step": 94240
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.2941459889183336e-05,
+      "loss": 0.2124,
+      "step": 94250
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.294012152359539e-05,
+      "loss": 0.129,
+      "step": 94260
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.293878315800745e-05,
+      "loss": 0.0959,
+      "step": 94270
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.29374447924195e-05,
+      "loss": 0.0784,
+      "step": 94280
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.293610642683156e-05,
+      "loss": 0.12,
+      "step": 94290
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.293476806124361e-05,
+      "loss": 0.0813,
+      "step": 94300
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.293342969565567e-05,
+      "loss": 0.1616,
+      "step": 94310
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.2932091330067724e-05,
+      "loss": 0.1966,
+      "step": 94320
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.293075296447978e-05,
+      "loss": 0.2808,
+      "step": 94330
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.2929414598891834e-05,
+      "loss": 0.1439,
+      "step": 94340
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.2928076233303896e-05,
+      "loss": 0.1545,
+      "step": 94350
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.292673786771595e-05,
+      "loss": 0.1714,
+      "step": 94360
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.292539950212801e-05,
+      "loss": 0.2786,
+      "step": 94370
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.292406113654006e-05,
+      "loss": 0.1612,
+      "step": 94380
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.292272277095212e-05,
+      "loss": 0.1816,
+      "step": 94390
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.292138440536417e-05,
+      "loss": 0.1727,
+      "step": 94400
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.292004603977623e-05,
+      "loss": 0.1509,
+      "step": 94410
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.2918707674188284e-05,
+      "loss": 0.0621,
+      "step": 94420
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.291736930860034e-05,
+      "loss": 0.2224,
+      "step": 94430
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.2916030943012394e-05,
+      "loss": 0.1533,
+      "step": 94440
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.2914692577424456e-05,
+      "loss": 0.1435,
+      "step": 94450
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.291335421183651e-05,
+      "loss": 0.1127,
+      "step": 94460
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.291201584624857e-05,
+      "loss": 0.1363,
+      "step": 94470
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.291067748066062e-05,
+      "loss": 0.0456,
+      "step": 94480
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.290933911507268e-05,
+      "loss": 0.0837,
+      "step": 94490
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.290800074948473e-05,
+      "loss": 0.1675,
+      "step": 94500
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.290666238389679e-05,
+      "loss": 0.1148,
+      "step": 94510
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.2905324018308844e-05,
+      "loss": 0.1136,
+      "step": 94520
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.29039856527209e-05,
+      "loss": 0.0469,
+      "step": 94530
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.2902647287132954e-05,
+      "loss": 0.2178,
+      "step": 94540
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.2901308921545017e-05,
+      "loss": 0.0904,
+      "step": 94550
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.289997055595707e-05,
+      "loss": 0.1204,
+      "step": 94560
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.289863219036913e-05,
+      "loss": 0.1362,
+      "step": 94570
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.289729382478118e-05,
+      "loss": 0.0576,
+      "step": 94580
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.289595545919324e-05,
+      "loss": 0.2002,
+      "step": 94590
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.289461709360529e-05,
+      "loss": 0.1332,
+      "step": 94600
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.289327872801735e-05,
+      "loss": 0.1726,
+      "step": 94610
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.2891940362429404e-05,
+      "loss": 0.0913,
+      "step": 94620
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.289060199684146e-05,
+      "loss": 0.2112,
+      "step": 94630
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.2889263631253515e-05,
+      "loss": 0.1157,
+      "step": 94640
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.288792526566558e-05,
+      "loss": 0.1634,
+      "step": 94650
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.288658690007763e-05,
+      "loss": 0.0931,
+      "step": 94660
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.288524853448969e-05,
+      "loss": 0.1026,
+      "step": 94670
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.288391016890174e-05,
+      "loss": 0.1302,
+      "step": 94680
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.28825718033138e-05,
+      "loss": 0.209,
+      "step": 94690
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.2881233437725853e-05,
+      "loss": 0.0896,
+      "step": 94700
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.287989507213791e-05,
+      "loss": 0.1098,
+      "step": 94710
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.287855670654996e-05,
+      "loss": 0.1695,
+      "step": 94720
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.287721834096202e-05,
+      "loss": 0.2887,
+      "step": 94730
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.2875879975374075e-05,
+      "loss": 0.1309,
+      "step": 94740
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.287454160978613e-05,
+      "loss": 0.269,
+      "step": 94750
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.2873203244198185e-05,
+      "loss": 0.092,
+      "step": 94760
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.287186487861024e-05,
+      "loss": 0.1669,
+      "step": 94770
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.2870526513022296e-05,
+      "loss": 0.1237,
+      "step": 94780
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.286918814743435e-05,
+      "loss": 0.1014,
+      "step": 94790
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.286784978184641e-05,
+      "loss": 0.0919,
+      "step": 94800
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.286651141625846e-05,
+      "loss": 0.2555,
+      "step": 94810
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.286517305067052e-05,
+      "loss": 0.2199,
+      "step": 94820
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.286383468508258e-05,
+      "loss": 0.2122,
+      "step": 94830
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.2862496319494635e-05,
+      "loss": 0.1252,
+      "step": 94840
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.286115795390669e-05,
+      "loss": 0.0335,
+      "step": 94850
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.2859819588318746e-05,
+      "loss": 0.2656,
+      "step": 94860
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.28584812227308e-05,
+      "loss": 0.0528,
+      "step": 94870
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.1623,
+      "step": 94880
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.285580449155491e-05,
+      "loss": 0.1245,
+      "step": 94890
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.285446612596697e-05,
+      "loss": 0.0208,
+      "step": 94900
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.285312776037902e-05,
+      "loss": 0.1751,
+      "step": 94910
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.285178939479108e-05,
+      "loss": 0.1655,
+      "step": 94920
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.285045102920314e-05,
+      "loss": 0.1261,
+      "step": 94930
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.2849112663615195e-05,
+      "loss": 0.0386,
+      "step": 94940
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.284777429802725e-05,
+      "loss": 0.2657,
+      "step": 94950
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.2846435932439306e-05,
+      "loss": 0.1218,
+      "step": 94960
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.284509756685136e-05,
+      "loss": 0.1834,
+      "step": 94970
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.2843759201263416e-05,
+      "loss": 0.0594,
+      "step": 94980
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.284242083567547e-05,
+      "loss": 0.0613,
+      "step": 94990
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.284108247008753e-05,
+      "loss": 0.1765,
+      "step": 95000
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.283974410449958e-05,
+      "loss": 0.1368,
+      "step": 95010
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.283840573891164e-05,
+      "loss": 0.0659,
+      "step": 95020
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.28370673733237e-05,
+      "loss": 0.0699,
+      "step": 95030
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.2835729007735755e-05,
+      "loss": 0.2533,
+      "step": 95040
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.283439064214781e-05,
+      "loss": 0.3461,
+      "step": 95050
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.2833052276559866e-05,
+      "loss": 0.0754,
+      "step": 95060
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.283171391097192e-05,
+      "loss": 0.1016,
+      "step": 95070
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.2830375545383977e-05,
+      "loss": 0.2209,
+      "step": 95080
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.282903717979603e-05,
+      "loss": 0.1637,
+      "step": 95090
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.282769881420809e-05,
+      "loss": 0.0749,
+      "step": 95100
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.282636044862014e-05,
+      "loss": 0.0317,
+      "step": 95110
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.28250220830322e-05,
+      "loss": 0.2005,
+      "step": 95120
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.282368371744426e-05,
+      "loss": 0.2895,
+      "step": 95130
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.2822345351856315e-05,
+      "loss": 0.3014,
+      "step": 95140
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.282100698626837e-05,
+      "loss": 0.1708,
+      "step": 95150
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.2819668620680426e-05,
+      "loss": 0.0343,
+      "step": 95160
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.281833025509248e-05,
+      "loss": 0.0628,
+      "step": 95170
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.281699188950454e-05,
+      "loss": 0.1736,
+      "step": 95180
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.281565352391659e-05,
+      "loss": 0.0805,
+      "step": 95190
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.281431515832865e-05,
+      "loss": 0.127,
+      "step": 95200
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.28129767927407e-05,
+      "loss": 0.1122,
+      "step": 95210
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.2811638427152765e-05,
+      "loss": 0.0746,
+      "step": 95220
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.281030006156482e-05,
+      "loss": 0.1252,
+      "step": 95230
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.2808961695976876e-05,
+      "loss": 0.1051,
+      "step": 95240
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.280762333038893e-05,
+      "loss": 0.1278,
+      "step": 95250
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.2806284964800986e-05,
+      "loss": 0.1592,
+      "step": 95260
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.280494659921304e-05,
+      "loss": 0.2691,
+      "step": 95270
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.28036082336251e-05,
+      "loss": 0.2102,
+      "step": 95280
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.280226986803715e-05,
+      "loss": 0.0801,
+      "step": 95290
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.280093150244921e-05,
+      "loss": 0.0157,
+      "step": 95300
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.279959313686126e-05,
+      "loss": 0.0785,
+      "step": 95310
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.2798254771273325e-05,
+      "loss": 0.1368,
+      "step": 95320
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.279691640568538e-05,
+      "loss": 0.175,
+      "step": 95330
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.2795578040097436e-05,
+      "loss": 0.0961,
+      "step": 95340
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.279423967450949e-05,
+      "loss": 0.1417,
+      "step": 95350
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.2792901308921546e-05,
+      "loss": 0.137,
+      "step": 95360
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.27915629433336e-05,
+      "loss": 0.1933,
+      "step": 95370
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.279022457774566e-05,
+      "loss": 0.1083,
+      "step": 95380
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.278888621215771e-05,
+      "loss": 0.1953,
+      "step": 95390
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.278754784656977e-05,
+      "loss": 0.0717,
+      "step": 95400
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.278620948098182e-05,
+      "loss": 0.2216,
+      "step": 95410
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.2784871115393885e-05,
+      "loss": 0.0694,
+      "step": 95420
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.278353274980594e-05,
+      "loss": 0.0628,
+      "step": 95430
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.2782194384217996e-05,
+      "loss": 0.2175,
+      "step": 95440
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.278085601863005e-05,
+      "loss": 0.1688,
+      "step": 95450
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.2779517653042107e-05,
+      "loss": 0.1434,
+      "step": 95460
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.277817928745416e-05,
+      "loss": 0.0809,
+      "step": 95470
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.277684092186622e-05,
+      "loss": 0.0427,
+      "step": 95480
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.277550255627827e-05,
+      "loss": 0.0725,
+      "step": 95490
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.277416419069033e-05,
+      "loss": 0.137,
+      "step": 95500
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.277282582510238e-05,
+      "loss": 0.0523,
+      "step": 95510
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.2771487459514445e-05,
+      "loss": 0.0634,
+      "step": 95520
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.27701490939265e-05,
+      "loss": 0.2184,
+      "step": 95530
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.2768810728338556e-05,
+      "loss": 0.0045,
+      "step": 95540
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.276747236275061e-05,
+      "loss": 0.4232,
+      "step": 95550
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.276613399716267e-05,
+      "loss": 0.0731,
+      "step": 95560
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.276479563157472e-05,
+      "loss": 0.3029,
+      "step": 95570
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.276345726598678e-05,
+      "loss": 0.2172,
+      "step": 95580
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.276211890039883e-05,
+      "loss": 0.0695,
+      "step": 95590
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.276078053481089e-05,
+      "loss": 0.2155,
+      "step": 95600
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.275944216922294e-05,
+      "loss": 0.1972,
+      "step": 95610
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.2758103803635005e-05,
+      "loss": 0.0933,
+      "step": 95620
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.275676543804706e-05,
+      "loss": 0.2141,
+      "step": 95630
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.2755427072459116e-05,
+      "loss": 0.1716,
+      "step": 95640
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.275408870687117e-05,
+      "loss": 0.1443,
+      "step": 95650
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.275275034128323e-05,
+      "loss": 0.1063,
+      "step": 95660
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.275141197569528e-05,
+      "loss": 0.1629,
+      "step": 95670
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.275007361010734e-05,
+      "loss": 0.0905,
+      "step": 95680
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.274873524451939e-05,
+      "loss": 0.1297,
+      "step": 95690
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.274739687893145e-05,
+      "loss": 0.1656,
+      "step": 95700
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.2746058513343504e-05,
+      "loss": 0.1155,
+      "step": 95710
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.2744720147755566e-05,
+      "loss": 0.14,
+      "step": 95720
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.274338178216762e-05,
+      "loss": 0.0861,
+      "step": 95730
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.2742043416579676e-05,
+      "loss": 0.0182,
+      "step": 95740
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.274070505099173e-05,
+      "loss": 0.2472,
+      "step": 95750
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.273936668540379e-05,
+      "loss": 0.0523,
+      "step": 95760
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.273802831981584e-05,
+      "loss": 0.0503,
+      "step": 95770
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.27366899542279e-05,
+      "loss": 0.1809,
+      "step": 95780
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.273535158863995e-05,
+      "loss": 0.1522,
+      "step": 95790
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.273401322305201e-05,
+      "loss": 0.0801,
+      "step": 95800
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.2732674857464064e-05,
+      "loss": 0.1643,
+      "step": 95810
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.2731336491876126e-05,
+      "loss": 0.0482,
+      "step": 95820
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.272999812628818e-05,
+      "loss": 0.103,
+      "step": 95830
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.2728659760700236e-05,
+      "loss": 0.0096,
+      "step": 95840
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.272732139511229e-05,
+      "loss": 0.2543,
+      "step": 95850
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.272598302952435e-05,
+      "loss": 0.3835,
+      "step": 95860
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.27246446639364e-05,
+      "loss": 0.1009,
+      "step": 95870
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.272330629834846e-05,
+      "loss": 0.2097,
+      "step": 95880
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.272196793276051e-05,
+      "loss": 0.1169,
+      "step": 95890
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.272062956717257e-05,
+      "loss": 0.1115,
+      "step": 95900
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.2719291201584624e-05,
+      "loss": 0.204,
+      "step": 95910
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.2717952835996686e-05,
+      "loss": 0.2057,
+      "step": 95920
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.271661447040874e-05,
+      "loss": 0.0841,
+      "step": 95930
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.27152761048208e-05,
+      "loss": 0.2331,
+      "step": 95940
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.271393773923285e-05,
+      "loss": 0.023,
+      "step": 95950
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.271259937364491e-05,
+      "loss": 0.1157,
+      "step": 95960
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.271126100805696e-05,
+      "loss": 0.0728,
+      "step": 95970
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.270992264246902e-05,
+      "loss": 0.1163,
+      "step": 95980
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.270858427688107e-05,
+      "loss": 0.2625,
+      "step": 95990
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.270724591129313e-05,
+      "loss": 0.214,
+      "step": 96000
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.2705907545705184e-05,
+      "loss": 0.0935,
+      "step": 96010
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.2704569180117246e-05,
+      "loss": 0.2376,
+      "step": 96020
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.27032308145293e-05,
+      "loss": 0.245,
+      "step": 96030
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.270189244894136e-05,
+      "loss": 0.1447,
+      "step": 96040
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.270055408335341e-05,
+      "loss": 0.0777,
+      "step": 96050
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.269921571776547e-05,
+      "loss": 0.1747,
+      "step": 96060
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.269787735217752e-05,
+      "loss": 0.1023,
+      "step": 96070
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.269653898658958e-05,
+      "loss": 0.1623,
+      "step": 96080
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.2695200621001633e-05,
+      "loss": 0.1345,
+      "step": 96090
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.269386225541369e-05,
+      "loss": 0.2071,
+      "step": 96100
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.2692523889825744e-05,
+      "loss": 0.1297,
+      "step": 96110
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.2691185524237806e-05,
+      "loss": 0.2247,
+      "step": 96120
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.268984715864986e-05,
+      "loss": 0.1109,
+      "step": 96130
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.268850879306192e-05,
+      "loss": 0.1498,
+      "step": 96140
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.268717042747397e-05,
+      "loss": 0.0535,
+      "step": 96150
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.268583206188603e-05,
+      "loss": 0.1257,
+      "step": 96160
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.268449369629808e-05,
+      "loss": 0.1579,
+      "step": 96170
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.268315533071014e-05,
+      "loss": 0.1851,
+      "step": 96180
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.2681816965122194e-05,
+      "loss": 0.14,
+      "step": 96190
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.268047859953425e-05,
+      "loss": 0.1439,
+      "step": 96200
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.2679140233946304e-05,
+      "loss": 0.215,
+      "step": 96210
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.2677801868358366e-05,
+      "loss": 0.2722,
+      "step": 96220
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.267646350277042e-05,
+      "loss": 0.1796,
+      "step": 96230
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.267512513718248e-05,
+      "loss": 0.3001,
+      "step": 96240
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.267378677159453e-05,
+      "loss": 0.0975,
+      "step": 96250
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.267244840600659e-05,
+      "loss": 0.0472,
+      "step": 96260
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.267111004041864e-05,
+      "loss": 0.2708,
+      "step": 96270
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.26697716748307e-05,
+      "loss": 0.277,
+      "step": 96280
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.2668433309242754e-05,
+      "loss": 0.0727,
+      "step": 96290
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.266709494365481e-05,
+      "loss": 0.3394,
+      "step": 96300
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.2665756578066864e-05,
+      "loss": 0.2708,
+      "step": 96310
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.2664418212478927e-05,
+      "loss": 0.1452,
+      "step": 96320
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.266307984689098e-05,
+      "loss": 0.1539,
+      "step": 96330
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.266174148130304e-05,
+      "loss": 0.105,
+      "step": 96340
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.266040311571509e-05,
+      "loss": 0.1031,
+      "step": 96350
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.265906475012715e-05,
+      "loss": 0.0971,
+      "step": 96360
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.26577263845392e-05,
+      "loss": 0.0626,
+      "step": 96370
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.265638801895126e-05,
+      "loss": 0.1334,
+      "step": 96380
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.2655049653363314e-05,
+      "loss": 0.0911,
+      "step": 96390
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.265371128777537e-05,
+      "loss": 0.0469,
+      "step": 96400
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.2652372922187425e-05,
+      "loss": 0.1743,
+      "step": 96410
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.265103455659949e-05,
+      "loss": 0.1074,
+      "step": 96420
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.264969619101154e-05,
+      "loss": 0.0841,
+      "step": 96430
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.26483578254236e-05,
+      "loss": 0.0922,
+      "step": 96440
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.264701945983565e-05,
+      "loss": 0.0496,
+      "step": 96450
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.264568109424771e-05,
+      "loss": 0.2424,
+      "step": 96460
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.2644342728659763e-05,
+      "loss": 0.0581,
+      "step": 96470
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.264300436307182e-05,
+      "loss": 0.0363,
+      "step": 96480
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.2641665997483874e-05,
+      "loss": 0.1471,
+      "step": 96490
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.264032763189593e-05,
+      "loss": 0.0779,
+      "step": 96500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.263898926630799e-05,
+      "loss": 0.1017,
+      "step": 96510
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.263765090072005e-05,
+      "loss": 0.2594,
+      "step": 96520
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.26363125351321e-05,
+      "loss": 0.3965,
+      "step": 96530
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.263497416954416e-05,
+      "loss": 0.1424,
+      "step": 96540
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.263363580395621e-05,
+      "loss": 0.1335,
+      "step": 96550
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.263229743836827e-05,
+      "loss": 0.1236,
+      "step": 96560
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.2630959072780324e-05,
+      "loss": 0.2714,
+      "step": 96570
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.262962070719238e-05,
+      "loss": 0.1486,
+      "step": 96580
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.2628282341604434e-05,
+      "loss": 0.1955,
+      "step": 96590
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.262694397601649e-05,
+      "loss": 0.1514,
+      "step": 96600
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.262560561042855e-05,
+      "loss": 0.092,
+      "step": 96610
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.262426724484061e-05,
+      "loss": 0.1042,
+      "step": 96620
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.262292887925266e-05,
+      "loss": 0.1704,
+      "step": 96630
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.262159051366472e-05,
+      "loss": 0.0525,
+      "step": 96640
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.262025214807677e-05,
+      "loss": 0.388,
+      "step": 96650
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.261891378248883e-05,
+      "loss": 0.1031,
+      "step": 96660
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.2617575416900884e-05,
+      "loss": 0.0441,
+      "step": 96670
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.261623705131294e-05,
+      "loss": 0.136,
+      "step": 96680
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.2614898685724994e-05,
+      "loss": 0.0787,
+      "step": 96690
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.261356032013705e-05,
+      "loss": 0.135,
+      "step": 96700
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.261222195454911e-05,
+      "loss": 0.1791,
+      "step": 96710
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.261088358896117e-05,
+      "loss": 0.0403,
+      "step": 96720
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.260954522337322e-05,
+      "loss": 0.0974,
+      "step": 96730
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.260820685778528e-05,
+      "loss": 0.1511,
+      "step": 96740
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.260686849219733e-05,
+      "loss": 0.0726,
+      "step": 96750
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.260553012660939e-05,
+      "loss": 0.0774,
+      "step": 96760
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.2604191761021444e-05,
+      "loss": 0.0056,
+      "step": 96770
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.26028533954335e-05,
+      "loss": 0.0063,
+      "step": 96780
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.2601515029845555e-05,
+      "loss": 0.0556,
+      "step": 96790
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.260017666425761e-05,
+      "loss": 0.1558,
+      "step": 96800
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.259883829866967e-05,
+      "loss": 0.2121,
+      "step": 96810
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.259749993308173e-05,
+      "loss": 0.0951,
+      "step": 96820
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.259616156749378e-05,
+      "loss": 0.0635,
+      "step": 96830
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.259482320190584e-05,
+      "loss": 0.1247,
+      "step": 96840
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.259348483631789e-05,
+      "loss": 0.0783,
+      "step": 96850
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.259214647072995e-05,
+      "loss": 0.2453,
+      "step": 96860
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.2590808105142004e-05,
+      "loss": 0.1359,
+      "step": 96870
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.258946973955406e-05,
+      "loss": 0.1144,
+      "step": 96880
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.2588131373966115e-05,
+      "loss": 0.1046,
+      "step": 96890
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.258679300837817e-05,
+      "loss": 0.024,
+      "step": 96900
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.258545464279023e-05,
+      "loss": 0.0569,
+      "step": 96910
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.258411627720229e-05,
+      "loss": 0.2252,
+      "step": 96920
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.258277791161434e-05,
+      "loss": 0.324,
+      "step": 96930
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.25814395460264e-05,
+      "loss": 0.0807,
+      "step": 96940
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.2580101180438453e-05,
+      "loss": 0.0774,
+      "step": 96950
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.257876281485051e-05,
+      "loss": 0.3507,
+      "step": 96960
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.2577424449262564e-05,
+      "loss": 0.3053,
+      "step": 96970
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.257608608367462e-05,
+      "loss": 0.2619,
+      "step": 96980
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.2574747718086675e-05,
+      "loss": 0.0569,
+      "step": 96990
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.257340935249873e-05,
+      "loss": 0.142,
+      "step": 97000
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.257207098691079e-05,
+      "loss": 0.1344,
+      "step": 97010
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.257073262132285e-05,
+      "loss": 0.1902,
+      "step": 97020
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.25693942557349e-05,
+      "loss": 0.2368,
+      "step": 97030
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.256805589014696e-05,
+      "loss": 0.0608,
+      "step": 97040
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.2566717524559014e-05,
+      "loss": 0.0037,
+      "step": 97050
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.256537915897107e-05,
+      "loss": 0.2736,
+      "step": 97060
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.2564040793383124e-05,
+      "loss": 0.2244,
+      "step": 97070
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.256270242779517e-05,
+      "loss": 0.1266,
+      "step": 97080
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.2561364062207235e-05,
+      "loss": 0.065,
+      "step": 97090
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.256002569661929e-05,
+      "loss": 0.0982,
+      "step": 97100
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.2558687331031346e-05,
+      "loss": 0.2722,
+      "step": 97110
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.25573489654434e-05,
+      "loss": 0.175,
+      "step": 97120
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.2556010599855456e-05,
+      "loss": 0.0502,
+      "step": 97130
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.255467223426751e-05,
+      "loss": 0.3727,
+      "step": 97140
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.255333386867957e-05,
+      "loss": 0.1311,
+      "step": 97150
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.255199550309162e-05,
+      "loss": 0.0785,
+      "step": 97160
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.255065713750368e-05,
+      "loss": 0.126,
+      "step": 97170
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.254931877191573e-05,
+      "loss": 0.113,
+      "step": 97180
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.2547980406327795e-05,
+      "loss": 0.1906,
+      "step": 97190
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.254664204073985e-05,
+      "loss": 0.1337,
+      "step": 97200
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.2545303675151906e-05,
+      "loss": 0.1012,
+      "step": 97210
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.254396530956396e-05,
+      "loss": 0.2247,
+      "step": 97220
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.2542626943976016e-05,
+      "loss": 0.0132,
+      "step": 97230
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.254128857838807e-05,
+      "loss": 0.1297,
+      "step": 97240
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.253995021280013e-05,
+      "loss": 0.2101,
+      "step": 97250
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.253861184721218e-05,
+      "loss": 0.0675,
+      "step": 97260
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.253727348162424e-05,
+      "loss": 0.124,
+      "step": 97270
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.253593511603629e-05,
+      "loss": 0.1601,
+      "step": 97280
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.2534596750448355e-05,
+      "loss": 0.0741,
+      "step": 97290
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.253325838486041e-05,
+      "loss": 0.1447,
+      "step": 97300
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.2531920019272466e-05,
+      "loss": 0.0789,
+      "step": 97310
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.253058165368452e-05,
+      "loss": 0.2205,
+      "step": 97320
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.252924328809658e-05,
+      "loss": 0.1924,
+      "step": 97330
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.252790492250863e-05,
+      "loss": 0.0815,
+      "step": 97340
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.252656655692069e-05,
+      "loss": 0.2473,
+      "step": 97350
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.252522819133274e-05,
+      "loss": 0.2482,
+      "step": 97360
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.25238898257448e-05,
+      "loss": 0.0756,
+      "step": 97370
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.252255146015685e-05,
+      "loss": 0.0115,
+      "step": 97380
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.2521213094568915e-05,
+      "loss": 0.1348,
+      "step": 97390
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.251987472898097e-05,
+      "loss": 0.1997,
+      "step": 97400
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.2518536363393026e-05,
+      "loss": 0.1295,
+      "step": 97410
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.251719799780508e-05,
+      "loss": 0.201,
+      "step": 97420
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.251585963221714e-05,
+      "loss": 0.1688,
+      "step": 97430
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.251452126662919e-05,
+      "loss": 0.1013,
+      "step": 97440
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.251318290104125e-05,
+      "loss": 0.0083,
+      "step": 97450
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.25118445354533e-05,
+      "loss": 0.1316,
+      "step": 97460
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.251050616986536e-05,
+      "loss": 0.0967,
+      "step": 97470
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.2509167804277413e-05,
+      "loss": 0.1288,
+      "step": 97480
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.2507829438689476e-05,
+      "loss": 0.1105,
+      "step": 97490
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.250649107310153e-05,
+      "loss": 0.1164,
+      "step": 97500
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.2505152707513586e-05,
+      "loss": 0.1432,
+      "step": 97510
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.250381434192564e-05,
+      "loss": 0.1765,
+      "step": 97520
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.25024759763377e-05,
+      "loss": 0.1849,
+      "step": 97530
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.250113761074975e-05,
+      "loss": 0.0792,
+      "step": 97540
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.249979924516181e-05,
+      "loss": 0.1163,
+      "step": 97550
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.249846087957386e-05,
+      "loss": 0.2642,
+      "step": 97560
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.249712251398592e-05,
+      "loss": 0.0941,
+      "step": 97570
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.2495784148397974e-05,
+      "loss": 0.1324,
+      "step": 97580
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.2494445782810036e-05,
+      "loss": 0.0867,
+      "step": 97590
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.249310741722209e-05,
+      "loss": 0.2082,
+      "step": 97600
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.2491769051634146e-05,
+      "loss": 0.2434,
+      "step": 97610
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.24904306860462e-05,
+      "loss": 0.1762,
+      "step": 97620
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.248909232045826e-05,
+      "loss": 0.2738,
+      "step": 97630
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.248775395487031e-05,
+      "loss": 0.0604,
+      "step": 97640
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.248641558928237e-05,
+      "loss": 0.1378,
+      "step": 97650
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.248507722369442e-05,
+      "loss": 0.0714,
+      "step": 97660
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.248373885810648e-05,
+      "loss": 0.0629,
+      "step": 97670
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.2482400492518534e-05,
+      "loss": 0.213,
+      "step": 97680
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.2481062126930596e-05,
+      "loss": 0.2271,
+      "step": 97690
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.247972376134265e-05,
+      "loss": 0.2424,
+      "step": 97700
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.2478385395754707e-05,
+      "loss": 0.109,
+      "step": 97710
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.247704703016676e-05,
+      "loss": 0.3158,
+      "step": 97720
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.247570866457882e-05,
+      "loss": 0.3341,
+      "step": 97730
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.247437029899087e-05,
+      "loss": 0.042,
+      "step": 97740
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.247303193340293e-05,
+      "loss": 0.0714,
+      "step": 97750
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.247169356781498e-05,
+      "loss": 0.1266,
+      "step": 97760
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.247035520222704e-05,
+      "loss": 0.1244,
+      "step": 97770
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.2469016836639094e-05,
+      "loss": 0.2669,
+      "step": 97780
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.2467678471051156e-05,
+      "loss": 0.1747,
+      "step": 97790
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.246634010546321e-05,
+      "loss": 0.0546,
+      "step": 97800
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.246500173987527e-05,
+      "loss": 0.0879,
+      "step": 97810
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.246366337428732e-05,
+      "loss": 0.0768,
+      "step": 97820
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.246232500869938e-05,
+      "loss": 0.1556,
+      "step": 97830
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.246098664311143e-05,
+      "loss": 0.2825,
+      "step": 97840
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.245964827752349e-05,
+      "loss": 0.1415,
+      "step": 97850
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.2458309911935543e-05,
+      "loss": 0.1726,
+      "step": 97860
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.24569715463476e-05,
+      "loss": 0.2729,
+      "step": 97870
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.2455633180759654e-05,
+      "loss": 0.0157,
+      "step": 97880
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.2454294815171716e-05,
+      "loss": 0.1676,
+      "step": 97890
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.245295644958377e-05,
+      "loss": 0.0814,
+      "step": 97900
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.245161808399583e-05,
+      "loss": 0.1264,
+      "step": 97910
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.245027971840788e-05,
+      "loss": 0.0837,
+      "step": 97920
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.244894135281994e-05,
+      "loss": 0.1854,
+      "step": 97930
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.244760298723199e-05,
+      "loss": 0.184,
+      "step": 97940
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.244626462164405e-05,
+      "loss": 0.2004,
+      "step": 97950
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.2444926256056104e-05,
+      "loss": 0.1609,
+      "step": 97960
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.244358789046816e-05,
+      "loss": 0.0714,
+      "step": 97970
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.244224952488022e-05,
+      "loss": 0.0479,
+      "step": 97980
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.2440911159292276e-05,
+      "loss": 0.1606,
+      "step": 97990
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.243957279370433e-05,
+      "loss": 0.1301,
+      "step": 98000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.243823442811639e-05,
+      "loss": 0.2678,
+      "step": 98010
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.243689606252844e-05,
+      "loss": 0.0931,
+      "step": 98020
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.24355576969405e-05,
+      "loss": 0.1404,
+      "step": 98030
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.243421933135255e-05,
+      "loss": 0.2966,
+      "step": 98040
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.243288096576461e-05,
+      "loss": 0.143,
+      "step": 98050
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.2431542600176664e-05,
+      "loss": 0.1187,
+      "step": 98060
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.243020423458872e-05,
+      "loss": 0.1524,
+      "step": 98070
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.242886586900078e-05,
+      "loss": 0.103,
+      "step": 98080
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.2427527503412837e-05,
+      "loss": 0.0852,
+      "step": 98090
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.242618913782489e-05,
+      "loss": 0.1022,
+      "step": 98100
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.242485077223695e-05,
+      "loss": 0.1743,
+      "step": 98110
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.2423512406649e-05,
+      "loss": 0.1704,
+      "step": 98120
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.242217404106106e-05,
+      "loss": 0.2159,
+      "step": 98130
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.242083567547311e-05,
+      "loss": 0.0231,
+      "step": 98140
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.241949730988517e-05,
+      "loss": 0.1815,
+      "step": 98150
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.2418158944297224e-05,
+      "loss": 0.2346,
+      "step": 98160
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.241682057870928e-05,
+      "loss": 0.1224,
+      "step": 98170
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.241548221312134e-05,
+      "loss": 0.0058,
+      "step": 98180
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.24141438475334e-05,
+      "loss": 0.166,
+      "step": 98190
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.241280548194545e-05,
+      "loss": 0.162,
+      "step": 98200
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.241146711635751e-05,
+      "loss": 0.145,
+      "step": 98210
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.241012875076956e-05,
+      "loss": 0.006,
+      "step": 98220
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.240879038518162e-05,
+      "loss": 0.1475,
+      "step": 98230
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.240745201959367e-05,
+      "loss": 0.1304,
+      "step": 98240
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.240611365400573e-05,
+      "loss": 0.2166,
+      "step": 98250
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.2404775288417784e-05,
+      "loss": 0.0434,
+      "step": 98260
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.240343692282984e-05,
+      "loss": 0.1796,
+      "step": 98270
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.24020985572419e-05,
+      "loss": 0.1684,
+      "step": 98280
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.240076019165396e-05,
+      "loss": 0.076,
+      "step": 98290
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.239942182606601e-05,
+      "loss": 0.16,
+      "step": 98300
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.239808346047807e-05,
+      "loss": 0.1361,
+      "step": 98310
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.239674509489012e-05,
+      "loss": 0.1403,
+      "step": 98320
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.239540672930218e-05,
+      "loss": 0.086,
+      "step": 98330
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.2394068363714234e-05,
+      "loss": 0.1937,
+      "step": 98340
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.239272999812629e-05,
+      "loss": 0.1329,
+      "step": 98350
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.2391391632538344e-05,
+      "loss": 0.0442,
+      "step": 98360
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.23900532669504e-05,
+      "loss": 0.1254,
+      "step": 98370
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.238871490136246e-05,
+      "loss": 0.2062,
+      "step": 98380
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.238737653577452e-05,
+      "loss": 0.1115,
+      "step": 98390
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.238603817018657e-05,
+      "loss": 0.1436,
+      "step": 98400
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.238469980459863e-05,
+      "loss": 0.2374,
+      "step": 98410
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.238336143901068e-05,
+      "loss": 0.1562,
+      "step": 98420
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.238202307342274e-05,
+      "loss": 0.1166,
+      "step": 98430
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.2380684707834794e-05,
+      "loss": 0.1749,
+      "step": 98440
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.237934634224685e-05,
+      "loss": 0.0646,
+      "step": 98450
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.2378007976658904e-05,
+      "loss": 0.0399,
+      "step": 98460
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.237666961107096e-05,
+      "loss": 0.0984,
+      "step": 98470
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.237533124548302e-05,
+      "loss": 0.1621,
+      "step": 98480
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.237399287989508e-05,
+      "loss": 0.0592,
+      "step": 98490
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.237265451430713e-05,
+      "loss": 0.2584,
+      "step": 98500
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.237131614871919e-05,
+      "loss": 0.2307,
+      "step": 98510
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.236997778313124e-05,
+      "loss": 0.0888,
+      "step": 98520
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.23686394175433e-05,
+      "loss": 0.0219,
+      "step": 98530
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.2367301051955354e-05,
+      "loss": 0.006,
+      "step": 98540
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.236596268636741e-05,
+      "loss": 0.1632,
+      "step": 98550
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.2364624320779464e-05,
+      "loss": 0.0046,
+      "step": 98560
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.236328595519152e-05,
+      "loss": 0.077,
+      "step": 98570
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.236194758960358e-05,
+      "loss": 0.121,
+      "step": 98580
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.236060922401564e-05,
+      "loss": 0.2479,
+      "step": 98590
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.235927085842769e-05,
+      "loss": 0.0425,
+      "step": 98600
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.235793249283975e-05,
+      "loss": 0.1702,
+      "step": 98610
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.23565941272518e-05,
+      "loss": 0.1372,
+      "step": 98620
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.235525576166386e-05,
+      "loss": 0.1514,
+      "step": 98630
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.2353917396075914e-05,
+      "loss": 0.1768,
+      "step": 98640
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.235257903048797e-05,
+      "loss": 0.2224,
+      "step": 98650
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.2351240664900025e-05,
+      "loss": 0.145,
+      "step": 98660
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.234990229931208e-05,
+      "loss": 0.153,
+      "step": 98670
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.234856393372414e-05,
+      "loss": 0.1914,
+      "step": 98680
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.23472255681362e-05,
+      "loss": 0.0591,
+      "step": 98690
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.234588720254825e-05,
+      "loss": 0.1794,
+      "step": 98700
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.234454883696031e-05,
+      "loss": 0.1775,
+      "step": 98710
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.2343210471372363e-05,
+      "loss": 0.2131,
+      "step": 98720
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.234187210578442e-05,
+      "loss": 0.1804,
+      "step": 98730
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.2340533740196474e-05,
+      "loss": 0.1866,
+      "step": 98740
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.233919537460853e-05,
+      "loss": 0.0924,
+      "step": 98750
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.2337857009020585e-05,
+      "loss": 0.0623,
+      "step": 98760
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.233651864343264e-05,
+      "loss": 0.0633,
+      "step": 98770
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.23351802778447e-05,
+      "loss": 0.0346,
+      "step": 98780
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.233384191225676e-05,
+      "loss": 0.1883,
+      "step": 98790
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.233250354666881e-05,
+      "loss": 0.1003,
+      "step": 98800
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.233116518108087e-05,
+      "loss": 0.2463,
+      "step": 98810
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.2329826815492924e-05,
+      "loss": 0.0534,
+      "step": 98820
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.232848844990498e-05,
+      "loss": 0.1212,
+      "step": 98830
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.2327150084317034e-05,
+      "loss": 0.0683,
+      "step": 98840
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.232581171872909e-05,
+      "loss": 0.0371,
+      "step": 98850
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.2324473353141145e-05,
+      "loss": 0.1351,
+      "step": 98860
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.23231349875532e-05,
+      "loss": 0.0993,
+      "step": 98870
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.232179662196526e-05,
+      "loss": 0.1154,
+      "step": 98880
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.232045825637732e-05,
+      "loss": 0.0435,
+      "step": 98890
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.231911989078937e-05,
+      "loss": 0.2642,
+      "step": 98900
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.231778152520143e-05,
+      "loss": 0.3405,
+      "step": 98910
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.2316443159613484e-05,
+      "loss": 0.0139,
+      "step": 98920
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.231510479402554e-05,
+      "loss": 0.1904,
+      "step": 98930
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.2313766428437594e-05,
+      "loss": 0.1224,
+      "step": 98940
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.231242806284965e-05,
+      "loss": 0.1727,
+      "step": 98950
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.2311089697261705e-05,
+      "loss": 0.1155,
+      "step": 98960
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.230975133167376e-05,
+      "loss": 0.074,
+      "step": 98970
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.230841296608582e-05,
+      "loss": 0.1636,
+      "step": 98980
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.230707460049788e-05,
+      "loss": 0.1969,
+      "step": 98990
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.230573623490993e-05,
+      "loss": 0.192,
+      "step": 99000
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.230439786932199e-05,
+      "loss": 0.2121,
+      "step": 99010
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.2303059503734044e-05,
+      "loss": 0.1165,
+      "step": 99020
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.23017211381461e-05,
+      "loss": 0.1635,
+      "step": 99030
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.2300382772558155e-05,
+      "loss": 0.1404,
+      "step": 99040
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.229904440697021e-05,
+      "loss": 0.0408,
+      "step": 99050
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.2297706041382265e-05,
+      "loss": 0.1464,
+      "step": 99060
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.229636767579432e-05,
+      "loss": 0.199,
+      "step": 99070
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.229502931020638e-05,
+      "loss": 0.0319,
+      "step": 99080
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.229369094461844e-05,
+      "loss": 0.0657,
+      "step": 99090
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.229235257903049e-05,
+      "loss": 0.0492,
+      "step": 99100
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.229101421344255e-05,
+      "loss": 0.2513,
+      "step": 99110
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.2289675847854604e-05,
+      "loss": 0.2614,
+      "step": 99120
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.228833748226666e-05,
+      "loss": 0.1844,
+      "step": 99130
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.2286999116678715e-05,
+      "loss": 0.1933,
+      "step": 99140
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.228566075109077e-05,
+      "loss": 0.1617,
+      "step": 99150
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.2284322385502825e-05,
+      "loss": 0.1273,
+      "step": 99160
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.228298401991488e-05,
+      "loss": 0.3322,
+      "step": 99170
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.228164565432694e-05,
+      "loss": 0.1319,
+      "step": 99180
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.2280307288739e-05,
+      "loss": 0.082,
+      "step": 99190
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.2278968923151054e-05,
+      "loss": 0.1066,
+      "step": 99200
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.227763055756311e-05,
+      "loss": 0.0869,
+      "step": 99210
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.2276292191975164e-05,
+      "loss": 0.1052,
+      "step": 99220
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.227495382638722e-05,
+      "loss": 0.0879,
+      "step": 99230
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.2273615460799275e-05,
+      "loss": 0.0962,
+      "step": 99240
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.227227709521133e-05,
+      "loss": 0.1482,
+      "step": 99250
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.2270938729623386e-05,
+      "loss": 0.2974,
+      "step": 99260
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.226960036403545e-05,
+      "loss": 0.1083,
+      "step": 99270
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.22682619984475e-05,
+      "loss": 0.1931,
+      "step": 99280
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.226692363285956e-05,
+      "loss": 0.1602,
+      "step": 99290
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.2265585267271614e-05,
+      "loss": 0.0388,
+      "step": 99300
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.226424690168367e-05,
+      "loss": 0.2714,
+      "step": 99310
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.2262908536095724e-05,
+      "loss": 0.1271,
+      "step": 99320
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.226157017050778e-05,
+      "loss": 0.0414,
+      "step": 99330
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.2260231804919835e-05,
+      "loss": 0.0118,
+      "step": 99340
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.225889343933189e-05,
+      "loss": 0.0607,
+      "step": 99350
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.2257555073743946e-05,
+      "loss": 0.103,
+      "step": 99360
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.225621670815601e-05,
+      "loss": 0.1451,
+      "step": 99370
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.225487834256806e-05,
+      "loss": 0.1454,
+      "step": 99380
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.225353997698012e-05,
+      "loss": 0.1244,
+      "step": 99390
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.2252201611392174e-05,
+      "loss": 0.1825,
+      "step": 99400
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.225086324580423e-05,
+      "loss": 0.1213,
+      "step": 99410
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.2249524880216285e-05,
+      "loss": 0.2038,
+      "step": 99420
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.224818651462834e-05,
+      "loss": 0.0716,
+      "step": 99430
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.2246848149040395e-05,
+      "loss": 0.0657,
+      "step": 99440
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.224550978345245e-05,
+      "loss": 0.1191,
+      "step": 99450
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.2244171417864506e-05,
+      "loss": 0.2688,
+      "step": 99460
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.224283305227656e-05,
+      "loss": 0.2172,
+      "step": 99470
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.2241494686688617e-05,
+      "loss": 0.1796,
+      "step": 99480
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.224015632110067e-05,
+      "loss": 0.1855,
+      "step": 99490
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.223881795551273e-05,
+      "loss": 0.2244,
+      "step": 99500
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.223747958992478e-05,
+      "loss": 0.1678,
+      "step": 99510
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.223614122433684e-05,
+      "loss": 0.0865,
+      "step": 99520
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.223480285874889e-05,
+      "loss": 0.062,
+      "step": 99530
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.223346449316095e-05,
+      "loss": 0.1857,
+      "step": 99540
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.223212612757301e-05,
+      "loss": 0.1332,
+      "step": 99550
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.2230787761985066e-05,
+      "loss": 0.1287,
+      "step": 99560
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.222944939639712e-05,
+      "loss": 0.201,
+      "step": 99570
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.222811103080918e-05,
+      "loss": 0.0887,
+      "step": 99580
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.222677266522123e-05,
+      "loss": 0.0155,
+      "step": 99590
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.222543429963329e-05,
+      "loss": 0.109,
+      "step": 99600
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.222409593404534e-05,
+      "loss": 0.0165,
+      "step": 99610
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.22227575684574e-05,
+      "loss": 0.22,
+      "step": 99620
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.222141920286945e-05,
+      "loss": 0.0778,
+      "step": 99630
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": {
+        "accuracy": 0.948509485094851
+      },
+      "eval_f1": {
+        "f1": 0.9485863094568601
+      },
+      "eval_loss": 0.2245834469795227,
+      "eval_precision": {
+        "precision": 0.9496017185087666
+      },
+      "eval_recall": {
+        "recall": 0.9484435235390778
+      },
+      "eval_runtime": 429.3969,
+      "eval_samples_per_second": 38.671,
+      "eval_steps_per_second": 19.336,
+      "step": 99630
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.222008083728151e-05,
+      "loss": 0.1085,
+      "step": 99640
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.221874247169357e-05,
+      "loss": 0.0829,
+      "step": 99650
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.2217404106105626e-05,
+      "loss": 0.12,
+      "step": 99660
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.221606574051768e-05,
+      "loss": 0.3783,
+      "step": 99670
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.221472737492974e-05,
+      "loss": 0.069,
+      "step": 99680
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.221338900934179e-05,
+      "loss": 0.1298,
+      "step": 99690
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.221205064375385e-05,
+      "loss": 0.0523,
+      "step": 99700
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.22107122781659e-05,
+      "loss": 0.1258,
+      "step": 99710
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.220937391257796e-05,
+      "loss": 0.0873,
+      "step": 99720
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.2208035546990014e-05,
+      "loss": 0.1037,
+      "step": 99730
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.220669718140207e-05,
+      "loss": 0.1478,
+      "step": 99740
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.220535881581413e-05,
+      "loss": 0.1512,
+      "step": 99750
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.2204020450226186e-05,
+      "loss": 0.1234,
+      "step": 99760
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.220268208463824e-05,
+      "loss": 0.2452,
+      "step": 99770
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.22013437190503e-05,
+      "loss": 0.3885,
+      "step": 99780
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.220000535346235e-05,
+      "loss": 0.1629,
+      "step": 99790
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.219866698787441e-05,
+      "loss": 0.2557,
+      "step": 99800
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.219732862228646e-05,
+      "loss": 0.0688,
+      "step": 99810
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.219599025669852e-05,
+      "loss": 0.1447,
+      "step": 99820
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.2194651891110574e-05,
+      "loss": 0.1318,
+      "step": 99830
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.219331352552263e-05,
+      "loss": 0.0446,
+      "step": 99840
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.219197515993469e-05,
+      "loss": 0.0965,
+      "step": 99850
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.2190636794346746e-05,
+      "loss": 0.1278,
+      "step": 99860
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.21892984287588e-05,
+      "loss": 0.1673,
+      "step": 99870
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.218796006317086e-05,
+      "loss": 0.1773,
+      "step": 99880
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.218662169758291e-05,
+      "loss": 0.1042,
+      "step": 99890
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.218528333199497e-05,
+      "loss": 0.188,
+      "step": 99900
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.218394496640702e-05,
+      "loss": 0.1252,
+      "step": 99910
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.218260660081908e-05,
+      "loss": 0.1009,
+      "step": 99920
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.2181268235231134e-05,
+      "loss": 0.1499,
+      "step": 99930
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.217992986964319e-05,
+      "loss": 0.0412,
+      "step": 99940
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.217859150405525e-05,
+      "loss": 0.0503,
+      "step": 99950
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.217725313846731e-05,
+      "loss": 0.016,
+      "step": 99960
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.217591477287936e-05,
+      "loss": 0.1072,
+      "step": 99970
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.217457640729142e-05,
+      "loss": 0.2242,
+      "step": 99980
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.217323804170347e-05,
+      "loss": 0.1589,
+      "step": 99990
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.217189967611553e-05,
+      "loss": 0.1026,
+      "step": 100000
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.217056131052758e-05,
+      "loss": 0.1048,
+      "step": 100010
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.216922294493964e-05,
+      "loss": 0.166,
+      "step": 100020
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.2167884579351694e-05,
+      "loss": 0.0513,
+      "step": 100030
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.216654621376375e-05,
+      "loss": 0.0268,
+      "step": 100040
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.216520784817581e-05,
+      "loss": 0.0969,
+      "step": 100050
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.216386948258787e-05,
+      "loss": 0.1754,
+      "step": 100060
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.216253111699992e-05,
+      "loss": 0.0914,
+      "step": 100070
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.216119275141198e-05,
+      "loss": 0.0902,
+      "step": 100080
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.215985438582403e-05,
+      "loss": 0.091,
+      "step": 100090
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.215851602023609e-05,
+      "loss": 0.1166,
+      "step": 100100
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.2157177654648143e-05,
+      "loss": 0.0084,
+      "step": 100110
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.21558392890602e-05,
+      "loss": 0.182,
+      "step": 100120
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.2154500923472254e-05,
+      "loss": 0.0545,
+      "step": 100130
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.215316255788431e-05,
+      "loss": 0.0522,
+      "step": 100140
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.215182419229637e-05,
+      "loss": 0.1097,
+      "step": 100150
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.215048582670843e-05,
+      "loss": 0.1614,
+      "step": 100160
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.214914746112048e-05,
+      "loss": 0.1995,
+      "step": 100170
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.214780909553254e-05,
+      "loss": 0.2002,
+      "step": 100180
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.214647072994459e-05,
+      "loss": 0.2511,
+      "step": 100190
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.214513236435665e-05,
+      "loss": 0.0343,
+      "step": 100200
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.2143793998768704e-05,
+      "loss": 0.1338,
+      "step": 100210
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.214245563318076e-05,
+      "loss": 0.1934,
+      "step": 100220
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.2141117267592814e-05,
+      "loss": 0.1455,
+      "step": 100230
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.213977890200487e-05,
+      "loss": 0.2111,
+      "step": 100240
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.213844053641693e-05,
+      "loss": 0.165,
+      "step": 100250
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.213710217082899e-05,
+      "loss": 0.0152,
+      "step": 100260
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.213576380524104e-05,
+      "loss": 0.1638,
+      "step": 100270
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.21344254396531e-05,
+      "loss": 0.0598,
+      "step": 100280
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.213308707406515e-05,
+      "loss": 0.0839,
+      "step": 100290
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.213174870847721e-05,
+      "loss": 0.2458,
+      "step": 100300
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.2130410342889264e-05,
+      "loss": 0.0653,
+      "step": 100310
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.212907197730132e-05,
+      "loss": 0.1445,
+      "step": 100320
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.2127733611713374e-05,
+      "loss": 0.0046,
+      "step": 100330
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.212639524612543e-05,
+      "loss": 0.1733,
+      "step": 100340
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.212505688053749e-05,
+      "loss": 0.0315,
+      "step": 100350
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.212371851494955e-05,
+      "loss": 0.2345,
+      "step": 100360
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.21223801493616e-05,
+      "loss": 0.0101,
+      "step": 100370
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.212104178377366e-05,
+      "loss": 0.1543,
+      "step": 100380
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.211970341818571e-05,
+      "loss": 0.0624,
+      "step": 100390
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.211836505259777e-05,
+      "loss": 0.4457,
+      "step": 100400
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.2117026687009824e-05,
+      "loss": 0.1262,
+      "step": 100410
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.211568832142188e-05,
+      "loss": 0.1329,
+      "step": 100420
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.2114349955833935e-05,
+      "loss": 0.0606,
+      "step": 100430
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.211301159024599e-05,
+      "loss": 0.1317,
+      "step": 100440
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.211167322465805e-05,
+      "loss": 0.2131,
+      "step": 100450
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.211033485907011e-05,
+      "loss": 0.2077,
+      "step": 100460
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.210899649348216e-05,
+      "loss": 0.2383,
+      "step": 100470
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.210765812789422e-05,
+      "loss": 0.2271,
+      "step": 100480
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.2106319762306273e-05,
+      "loss": 0.2604,
+      "step": 100490
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.210498139671833e-05,
+      "loss": 0.0723,
+      "step": 100500
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.2103643031130384e-05,
+      "loss": 0.1092,
+      "step": 100510
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.210230466554244e-05,
+      "loss": 0.1248,
+      "step": 100520
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.2100966299954495e-05,
+      "loss": 0.0918,
+      "step": 100530
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.209962793436655e-05,
+      "loss": 0.2155,
+      "step": 100540
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.209828956877861e-05,
+      "loss": 0.1774,
+      "step": 100550
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.209695120319067e-05,
+      "loss": 0.1801,
+      "step": 100560
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.209561283760272e-05,
+      "loss": 0.1533,
+      "step": 100570
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.209427447201478e-05,
+      "loss": 0.2306,
+      "step": 100580
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.2092936106426834e-05,
+      "loss": 0.0768,
+      "step": 100590
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.209159774083889e-05,
+      "loss": 0.0383,
+      "step": 100600
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.2090259375250944e-05,
+      "loss": 0.1069,
+      "step": 100610
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.2088921009663e-05,
+      "loss": 0.0031,
+      "step": 100620
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.2087582644075055e-05,
+      "loss": 0.1672,
+      "step": 100630
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.208624427848711e-05,
+      "loss": 0.0872,
+      "step": 100640
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.208490591289917e-05,
+      "loss": 0.116,
+      "step": 100650
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.208356754731123e-05,
+      "loss": 0.018,
+      "step": 100660
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.208222918172328e-05,
+      "loss": 0.1595,
+      "step": 100670
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.208089081613534e-05,
+      "loss": 0.2565,
+      "step": 100680
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.2079552450547394e-05,
+      "loss": 0.0709,
+      "step": 100690
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.207821408495945e-05,
+      "loss": 0.0919,
+      "step": 100700
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.2076875719371504e-05,
+      "loss": 0.3221,
+      "step": 100710
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.207553735378356e-05,
+      "loss": 0.0751,
+      "step": 100720
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.2074198988195615e-05,
+      "loss": 0.066,
+      "step": 100730
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.207286062260768e-05,
+      "loss": 0.2423,
+      "step": 100740
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.207152225701973e-05,
+      "loss": 0.0568,
+      "step": 100750
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.207018389143179e-05,
+      "loss": 0.128,
+      "step": 100760
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.206884552584384e-05,
+      "loss": 0.2979,
+      "step": 100770
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.20675071602559e-05,
+      "loss": 0.1927,
+      "step": 100780
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.2066168794667954e-05,
+      "loss": 0.0565,
+      "step": 100790
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.206483042908001e-05,
+      "loss": 0.0151,
+      "step": 100800
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.2063492063492065e-05,
+      "loss": 0.1791,
+      "step": 100810
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.206215369790412e-05,
+      "loss": 0.1457,
+      "step": 100820
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.2060815332316175e-05,
+      "loss": 0.0314,
+      "step": 100830
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.205947696672824e-05,
+      "loss": 0.0433,
+      "step": 100840
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.205813860114029e-05,
+      "loss": 0.11,
+      "step": 100850
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.205680023555235e-05,
+      "loss": 0.1796,
+      "step": 100860
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.20554618699644e-05,
+      "loss": 0.1899,
+      "step": 100870
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.205412350437646e-05,
+      "loss": 0.1427,
+      "step": 100880
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.2052785138788514e-05,
+      "loss": 0.0419,
+      "step": 100890
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.205144677320057e-05,
+      "loss": 0.1684,
+      "step": 100900
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.2050108407612625e-05,
+      "loss": 0.0687,
+      "step": 100910
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.204877004202468e-05,
+      "loss": 0.0546,
+      "step": 100920
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.2047431676436735e-05,
+      "loss": 0.1531,
+      "step": 100930
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.20460933108488e-05,
+      "loss": 0.0035,
+      "step": 100940
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.204475494526085e-05,
+      "loss": 0.0942,
+      "step": 100950
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.204341657967291e-05,
+      "loss": 0.1475,
+      "step": 100960
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.2042078214084963e-05,
+      "loss": 0.2152,
+      "step": 100970
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.204073984849702e-05,
+      "loss": 0.0108,
+      "step": 100980
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.2039401482909074e-05,
+      "loss": 0.2628,
+      "step": 100990
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.203806311732113e-05,
+      "loss": 0.0576,
+      "step": 101000
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.2036724751733185e-05,
+      "loss": 0.047,
+      "step": 101010
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.203538638614524e-05,
+      "loss": 0.2238,
+      "step": 101020
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.2034048020557296e-05,
+      "loss": 0.2105,
+      "step": 101030
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.203270965496936e-05,
+      "loss": 0.1093,
+      "step": 101040
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.203137128938141e-05,
+      "loss": 0.1864,
+      "step": 101050
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.203003292379347e-05,
+      "loss": 0.3178,
+      "step": 101060
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.2028694558205524e-05,
+      "loss": 0.1203,
+      "step": 101070
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.202735619261758e-05,
+      "loss": 0.2196,
+      "step": 101080
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.2026017827029634e-05,
+      "loss": 0.2479,
+      "step": 101090
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.202467946144169e-05,
+      "loss": 0.2376,
+      "step": 101100
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.2023341095853745e-05,
+      "loss": 0.024,
+      "step": 101110
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.20220027302658e-05,
+      "loss": 0.1517,
+      "step": 101120
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.2020664364677856e-05,
+      "loss": 0.0684,
+      "step": 101130
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.201932599908992e-05,
+      "loss": 0.0204,
+      "step": 101140
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.201798763350197e-05,
+      "loss": 0.2313,
+      "step": 101150
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.201664926791403e-05,
+      "loss": 0.1928,
+      "step": 101160
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.2015310902326084e-05,
+      "loss": 0.1331,
+      "step": 101170
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.201397253673814e-05,
+      "loss": 0.1157,
+      "step": 101180
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.2012634171150194e-05,
+      "loss": 0.0094,
+      "step": 101190
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.201129580556225e-05,
+      "loss": 0.0318,
+      "step": 101200
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.2009957439974305e-05,
+      "loss": 0.1094,
+      "step": 101210
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.200861907438636e-05,
+      "loss": 0.174,
+      "step": 101220
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.2007280708798416e-05,
+      "loss": 0.1827,
+      "step": 101230
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.200594234321048e-05,
+      "loss": 0.0829,
+      "step": 101240
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.200460397762253e-05,
+      "loss": 0.1136,
+      "step": 101250
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.200326561203459e-05,
+      "loss": 0.0625,
+      "step": 101260
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.2001927246446644e-05,
+      "loss": 0.1137,
+      "step": 101270
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.20005888808587e-05,
+      "loss": 0.0858,
+      "step": 101280
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.1999250515270755e-05,
+      "loss": 0.1094,
+      "step": 101290
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.199791214968281e-05,
+      "loss": 0.042,
+      "step": 101300
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.1996573784094865e-05,
+      "loss": 0.1374,
+      "step": 101310
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.199523541850692e-05,
+      "loss": 0.0533,
+      "step": 101320
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.1993897052918976e-05,
+      "loss": 0.0905,
+      "step": 101330
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.199255868733104e-05,
+      "loss": 0.1011,
+      "step": 101340
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.1991220321743093e-05,
+      "loss": 0.1237,
+      "step": 101350
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.198988195615515e-05,
+      "loss": 0.1353,
+      "step": 101360
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.1988543590567204e-05,
+      "loss": 0.0265,
+      "step": 101370
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.198720522497926e-05,
+      "loss": 0.053,
+      "step": 101380
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.1985866859391315e-05,
+      "loss": 0.1724,
+      "step": 101390
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.198452849380337e-05,
+      "loss": 0.0932,
+      "step": 101400
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.1983190128215425e-05,
+      "loss": 0.0372,
+      "step": 101410
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.198185176262748e-05,
+      "loss": 0.1156,
+      "step": 101420
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.1980513397039536e-05,
+      "loss": 0.1908,
+      "step": 101430
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.19791750314516e-05,
+      "loss": 0.115,
+      "step": 101440
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.1977836665863654e-05,
+      "loss": 0.1786,
+      "step": 101450
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.197649830027571e-05,
+      "loss": 0.2042,
+      "step": 101460
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.1975159934687764e-05,
+      "loss": 0.1417,
+      "step": 101470
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.197382156909982e-05,
+      "loss": 0.0754,
+      "step": 101480
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.1972483203511875e-05,
+      "loss": 0.2732,
+      "step": 101490
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.197114483792393e-05,
+      "loss": 0.0904,
+      "step": 101500
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.1969806472335986e-05,
+      "loss": 0.2509,
+      "step": 101510
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.196846810674804e-05,
+      "loss": 0.0955,
+      "step": 101520
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.1967129741160096e-05,
+      "loss": 0.0941,
+      "step": 101530
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.196579137557216e-05,
+      "loss": 0.0388,
+      "step": 101540
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.1964453009984214e-05,
+      "loss": 0.0611,
+      "step": 101550
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.196311464439627e-05,
+      "loss": 0.0993,
+      "step": 101560
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.1961776278808324e-05,
+      "loss": 0.0534,
+      "step": 101570
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.196043791322038e-05,
+      "loss": 0.0719,
+      "step": 101580
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.1959099547632435e-05,
+      "loss": 0.145,
+      "step": 101590
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.195776118204449e-05,
+      "loss": 0.1333,
+      "step": 101600
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.1956422816456546e-05,
+      "loss": 0.0064,
+      "step": 101610
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.19550844508686e-05,
+      "loss": 0.191,
+      "step": 101620
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.1953746085280656e-05,
+      "loss": 0.0787,
+      "step": 101630
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.195240771969272e-05,
+      "loss": 0.2005,
+      "step": 101640
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.1951069354104774e-05,
+      "loss": 0.1959,
+      "step": 101650
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.194973098851683e-05,
+      "loss": 0.229,
+      "step": 101660
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.1948392622928885e-05,
+      "loss": 0.0833,
+      "step": 101670
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.194705425734094e-05,
+      "loss": 0.1466,
+      "step": 101680
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.1945715891752995e-05,
+      "loss": 0.0935,
+      "step": 101690
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.194437752616505e-05,
+      "loss": 0.0724,
+      "step": 101700
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.1943039160577106e-05,
+      "loss": 0.0471,
+      "step": 101710
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.194170079498916e-05,
+      "loss": 0.2328,
+      "step": 101720
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.1940362429401217e-05,
+      "loss": 0.143,
+      "step": 101730
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.193902406381328e-05,
+      "loss": 0.0366,
+      "step": 101740
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.1937685698225334e-05,
+      "loss": 0.0449,
+      "step": 101750
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.193634733263739e-05,
+      "loss": 0.1772,
+      "step": 101760
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.1935008967049445e-05,
+      "loss": 0.0892,
+      "step": 101770
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.19336706014615e-05,
+      "loss": 0.1799,
+      "step": 101780
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.1932332235873555e-05,
+      "loss": 0.0941,
+      "step": 101790
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.193099387028561e-05,
+      "loss": 0.136,
+      "step": 101800
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.192965550469766e-05,
+      "loss": 0.0942,
+      "step": 101810
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.192831713910972e-05,
+      "loss": 0.2511,
+      "step": 101820
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.192697877352178e-05,
+      "loss": 0.1222,
+      "step": 101830
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.192564040793383e-05,
+      "loss": 0.2296,
+      "step": 101840
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.192430204234589e-05,
+      "loss": 0.1324,
+      "step": 101850
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.192296367675794e-05,
+      "loss": 0.0524,
+      "step": 101860
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.192162531117e-05,
+      "loss": 0.1388,
+      "step": 101870
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.1920286945582053e-05,
+      "loss": 0.012,
+      "step": 101880
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.191894857999411e-05,
+      "loss": 0.223,
+      "step": 101890
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.1917610214406164e-05,
+      "loss": 0.123,
+      "step": 101900
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.191627184881822e-05,
+      "loss": 0.1331,
+      "step": 101910
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.191493348323028e-05,
+      "loss": 0.0964,
+      "step": 101920
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.191359511764234e-05,
+      "loss": 0.1367,
+      "step": 101930
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.191225675205439e-05,
+      "loss": 0.0944,
+      "step": 101940
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.191091838646645e-05,
+      "loss": 0.1424,
+      "step": 101950
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.19095800208785e-05,
+      "loss": 0.0716,
+      "step": 101960
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.190824165529056e-05,
+      "loss": 0.2164,
+      "step": 101970
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.1906903289702614e-05,
+      "loss": 0.0249,
+      "step": 101980
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.190556492411467e-05,
+      "loss": 0.1591,
+      "step": 101990
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.1904226558526724e-05,
+      "loss": 0.1797,
+      "step": 102000
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.190288819293878e-05,
+      "loss": 0.021,
+      "step": 102010
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.190154982735084e-05,
+      "loss": 0.101,
+      "step": 102020
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.19002114617629e-05,
+      "loss": 0.1699,
+      "step": 102030
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.189887309617495e-05,
+      "loss": 0.1106,
+      "step": 102040
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.189753473058701e-05,
+      "loss": 0.07,
+      "step": 102050
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.189619636499906e-05,
+      "loss": 0.0202,
+      "step": 102060
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.189485799941112e-05,
+      "loss": 0.2541,
+      "step": 102070
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.1893519633823174e-05,
+      "loss": 0.1734,
+      "step": 102080
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.189218126823523e-05,
+      "loss": 0.1094,
+      "step": 102090
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.1890842902647284e-05,
+      "loss": 0.109,
+      "step": 102100
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.188950453705934e-05,
+      "loss": 0.2892,
+      "step": 102110
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.18881661714714e-05,
+      "loss": 0.0301,
+      "step": 102120
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.188682780588346e-05,
+      "loss": 0.1025,
+      "step": 102130
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.188548944029551e-05,
+      "loss": 0.2166,
+      "step": 102140
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.188415107470757e-05,
+      "loss": 0.1,
+      "step": 102150
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.188281270911962e-05,
+      "loss": 0.2004,
+      "step": 102160
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.188147434353168e-05,
+      "loss": 0.0895,
+      "step": 102170
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.1880135977943734e-05,
+      "loss": 0.2201,
+      "step": 102180
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.187879761235579e-05,
+      "loss": 0.1412,
+      "step": 102190
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.1877459246767845e-05,
+      "loss": 0.1752,
+      "step": 102200
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.18761208811799e-05,
+      "loss": 0.0662,
+      "step": 102210
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.187478251559196e-05,
+      "loss": 0.0635,
+      "step": 102220
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.187344415000402e-05,
+      "loss": 0.0461,
+      "step": 102230
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.187210578441607e-05,
+      "loss": 0.1395,
+      "step": 102240
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.187076741882813e-05,
+      "loss": 0.0908,
+      "step": 102250
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.186942905324018e-05,
+      "loss": 0.197,
+      "step": 102260
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.186809068765224e-05,
+      "loss": 0.0902,
+      "step": 102270
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.1866752322064294e-05,
+      "loss": 0.1846,
+      "step": 102280
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.186541395647635e-05,
+      "loss": 0.1673,
+      "step": 102290
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.1864075590888405e-05,
+      "loss": 0.0736,
+      "step": 102300
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.186273722530047e-05,
+      "loss": 0.0389,
+      "step": 102310
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.186139885971252e-05,
+      "loss": 0.1388,
+      "step": 102320
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.186006049412458e-05,
+      "loss": 0.106,
+      "step": 102330
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.185872212853663e-05,
+      "loss": 0.0873,
+      "step": 102340
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.185738376294869e-05,
+      "loss": 0.1015,
+      "step": 102350
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.1856045397360744e-05,
+      "loss": 0.0747,
+      "step": 102360
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.18547070317728e-05,
+      "loss": 0.0078,
+      "step": 102370
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.1853368666184854e-05,
+      "loss": 0.2557,
+      "step": 102380
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.185203030059691e-05,
+      "loss": 0.1282,
+      "step": 102390
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.1850691935008965e-05,
+      "loss": 0.0042,
+      "step": 102400
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.184935356942103e-05,
+      "loss": 0.014,
+      "step": 102410
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.184801520383308e-05,
+      "loss": 0.2295,
+      "step": 102420
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.184667683824514e-05,
+      "loss": 0.1999,
+      "step": 102430
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.184533847265719e-05,
+      "loss": 0.2354,
+      "step": 102440
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.184400010706925e-05,
+      "loss": 0.1016,
+      "step": 102450
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.1842661741481304e-05,
+      "loss": 0.133,
+      "step": 102460
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.184132337589336e-05,
+      "loss": 0.1404,
+      "step": 102470
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.1839985010305414e-05,
+      "loss": 0.2732,
+      "step": 102480
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.183864664471747e-05,
+      "loss": 0.0908,
+      "step": 102490
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.1837308279129525e-05,
+      "loss": 0.1958,
+      "step": 102500
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.183596991354159e-05,
+      "loss": 0.1813,
+      "step": 102510
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.183463154795364e-05,
+      "loss": 0.1901,
+      "step": 102520
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.18332931823657e-05,
+      "loss": 0.2062,
+      "step": 102530
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.183195481677775e-05,
+      "loss": 0.157,
+      "step": 102540
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.183061645118981e-05,
+      "loss": 0.0382,
+      "step": 102550
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.1829278085601864e-05,
+      "loss": 0.1486,
+      "step": 102560
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.182793972001392e-05,
+      "loss": 0.1597,
+      "step": 102570
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.1826601354425974e-05,
+      "loss": 0.0809,
+      "step": 102580
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.182526298883803e-05,
+      "loss": 0.1273,
+      "step": 102590
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.1823924623250085e-05,
+      "loss": 0.1787,
+      "step": 102600
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.182258625766215e-05,
+      "loss": 0.0487,
+      "step": 102610
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.18212478920742e-05,
+      "loss": 0.0615,
+      "step": 102620
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.181990952648626e-05,
+      "loss": 0.0804,
+      "step": 102630
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.181857116089831e-05,
+      "loss": 0.0675,
+      "step": 102640
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.181723279531037e-05,
+      "loss": 0.2274,
+      "step": 102650
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.1815894429722424e-05,
+      "loss": 0.2115,
+      "step": 102660
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.181455606413448e-05,
+      "loss": 0.1599,
+      "step": 102670
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.1813217698546535e-05,
+      "loss": 0.1357,
+      "step": 102680
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.181187933295859e-05,
+      "loss": 0.0098,
+      "step": 102690
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.1810540967370645e-05,
+      "loss": 0.2395,
+      "step": 102700
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.180920260178271e-05,
+      "loss": 0.0131,
+      "step": 102710
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.180786423619476e-05,
+      "loss": 0.0576,
+      "step": 102720
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.180652587060682e-05,
+      "loss": 0.2854,
+      "step": 102730
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.1805187505018873e-05,
+      "loss": 0.1991,
+      "step": 102740
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.180384913943093e-05,
+      "loss": 0.0735,
+      "step": 102750
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.1802510773842984e-05,
+      "loss": 0.107,
+      "step": 102760
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.180117240825504e-05,
+      "loss": 0.2834,
+      "step": 102770
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.1799834042667095e-05,
+      "loss": 0.1786,
+      "step": 102780
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.179849567707915e-05,
+      "loss": 0.1765,
+      "step": 102790
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.1797157311491205e-05,
+      "loss": 0.2455,
+      "step": 102800
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.179581894590327e-05,
+      "loss": 0.079,
+      "step": 102810
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.179448058031532e-05,
+      "loss": 0.1639,
+      "step": 102820
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.179314221472738e-05,
+      "loss": 0.0245,
+      "step": 102830
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.1791803849139434e-05,
+      "loss": 0.1383,
+      "step": 102840
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.179046548355149e-05,
+      "loss": 0.1135,
+      "step": 102850
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.1789127117963544e-05,
+      "loss": 0.1009,
+      "step": 102860
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.17877887523756e-05,
+      "loss": 0.234,
+      "step": 102870
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.1786450386787655e-05,
+      "loss": 0.1274,
+      "step": 102880
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.178511202119971e-05,
+      "loss": 0.1205,
+      "step": 102890
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.1783773655611766e-05,
+      "loss": 0.2237,
+      "step": 102900
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.178243529002383e-05,
+      "loss": 0.1602,
+      "step": 102910
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.178109692443588e-05,
+      "loss": 0.0229,
+      "step": 102920
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.177975855884794e-05,
+      "loss": 0.1318,
+      "step": 102930
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.1778420193259994e-05,
+      "loss": 0.1587,
+      "step": 102940
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.177708182767205e-05,
+      "loss": 0.2121,
+      "step": 102950
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.1775743462084104e-05,
+      "loss": 0.2438,
+      "step": 102960
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.177440509649616e-05,
+      "loss": 0.2702,
+      "step": 102970
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.1773066730908215e-05,
+      "loss": 0.0737,
+      "step": 102980
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.177172836532027e-05,
+      "loss": 0.2145,
+      "step": 102990
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.1770389999732326e-05,
+      "loss": 0.1136,
+      "step": 103000
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.176905163414439e-05,
+      "loss": 0.1118,
+      "step": 103010
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.176771326855644e-05,
+      "loss": 0.148,
+      "step": 103020
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.17663749029685e-05,
+      "loss": 0.2562,
+      "step": 103030
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.1765036537380554e-05,
+      "loss": 0.0737,
+      "step": 103040
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.176369817179261e-05,
+      "loss": 0.0763,
+      "step": 103050
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.1762359806204665e-05,
+      "loss": 0.2283,
+      "step": 103060
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.176102144061672e-05,
+      "loss": 0.1589,
+      "step": 103070
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.1759683075028775e-05,
+      "loss": 0.1522,
+      "step": 103080
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.175834470944083e-05,
+      "loss": 0.0819,
+      "step": 103090
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.1757006343852886e-05,
+      "loss": 0.0986,
+      "step": 103100
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.175566797826495e-05,
+      "loss": 0.0188,
+      "step": 103110
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.1754329612677003e-05,
+      "loss": 0.2401,
+      "step": 103120
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.175299124708906e-05,
+      "loss": 0.0977,
+      "step": 103130
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.1751652881501114e-05,
+      "loss": 0.2283,
+      "step": 103140
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.175031451591317e-05,
+      "loss": 0.0648,
+      "step": 103150
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.1748976150325225e-05,
+      "loss": 0.2095,
+      "step": 103160
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.174763778473728e-05,
+      "loss": 0.1125,
+      "step": 103170
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.1746299419149335e-05,
+      "loss": 0.0708,
+      "step": 103180
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.174496105356139e-05,
+      "loss": 0.0976,
+      "step": 103190
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.1743622687973446e-05,
+      "loss": 0.0584,
+      "step": 103200
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.174228432238551e-05,
+      "loss": 0.1825,
+      "step": 103210
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.1740945956797564e-05,
+      "loss": 0.1437,
+      "step": 103220
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.173960759120962e-05,
+      "loss": 0.1002,
+      "step": 103230
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.1738269225621674e-05,
+      "loss": 0.2086,
+      "step": 103240
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.173693086003373e-05,
+      "loss": 0.1071,
+      "step": 103250
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.1735592494445785e-05,
+      "loss": 0.0343,
+      "step": 103260
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.173425412885784e-05,
+      "loss": 0.111,
+      "step": 103270
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.1732915763269896e-05,
+      "loss": 0.2573,
+      "step": 103280
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.173157739768195e-05,
+      "loss": 0.1619,
+      "step": 103290
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.1730239032094006e-05,
+      "loss": 0.0312,
+      "step": 103300
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.172890066650607e-05,
+      "loss": 0.2342,
+      "step": 103310
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.1727562300918124e-05,
+      "loss": 0.3657,
+      "step": 103320
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.172622393533018e-05,
+      "loss": 0.1482,
+      "step": 103330
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.1724885569742234e-05,
+      "loss": 0.0505,
+      "step": 103340
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.172354720415429e-05,
+      "loss": 0.0543,
+      "step": 103350
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.1722208838566345e-05,
+      "loss": 0.1512,
+      "step": 103360
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.17208704729784e-05,
+      "loss": 0.0894,
+      "step": 103370
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.1719532107390456e-05,
+      "loss": 0.2121,
+      "step": 103380
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.171819374180251e-05,
+      "loss": 0.0267,
+      "step": 103390
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.1716855376214566e-05,
+      "loss": 0.1753,
+      "step": 103400
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.171551701062663e-05,
+      "loss": 0.0484,
+      "step": 103410
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.1714178645038684e-05,
+      "loss": 0.1494,
+      "step": 103420
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.171284027945074e-05,
+      "loss": 0.157,
+      "step": 103430
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.1711501913862795e-05,
+      "loss": 0.1061,
+      "step": 103440
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.171016354827485e-05,
+      "loss": 0.2747,
+      "step": 103450
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.1708825182686905e-05,
+      "loss": 0.1237,
+      "step": 103460
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.170748681709896e-05,
+      "loss": 0.0952,
+      "step": 103470
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.1706148451511016e-05,
+      "loss": 0.0924,
+      "step": 103480
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.170481008592307e-05,
+      "loss": 0.1061,
+      "step": 103490
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.170347172033513e-05,
+      "loss": 0.1207,
+      "step": 103500
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.170213335474719e-05,
+      "loss": 0.0351,
+      "step": 103510
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.1700794989159244e-05,
+      "loss": 0.0526,
+      "step": 103520
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.16994566235713e-05,
+      "loss": 0.3567,
+      "step": 103530
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.1698118257983355e-05,
+      "loss": 0.0615,
+      "step": 103540
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.169677989239541e-05,
+      "loss": 0.0947,
+      "step": 103550
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.1695441526807465e-05,
+      "loss": 0.1664,
+      "step": 103560
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.169410316121952e-05,
+      "loss": 0.0705,
+      "step": 103570
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.1692764795631576e-05,
+      "loss": 0.177,
+      "step": 103580
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.169142643004363e-05,
+      "loss": 0.1334,
+      "step": 103590
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.1690088064455693e-05,
+      "loss": 0.0771,
+      "step": 103600
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.168874969886775e-05,
+      "loss": 0.0282,
+      "step": 103610
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.1687411333279804e-05,
+      "loss": 0.1382,
+      "step": 103620
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.168607296769186e-05,
+      "loss": 0.0539,
+      "step": 103630
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.1684734602103915e-05,
+      "loss": 0.2999,
+      "step": 103640
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.168339623651597e-05,
+      "loss": 0.2712,
+      "step": 103650
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.1682057870928026e-05,
+      "loss": 0.2454,
+      "step": 103660
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.168071950534008e-05,
+      "loss": 0.2024,
+      "step": 103670
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.1679381139752136e-05,
+      "loss": 0.1626,
+      "step": 103680
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.167804277416419e-05,
+      "loss": 0.2333,
+      "step": 103690
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.1676704408576254e-05,
+      "loss": 0.188,
+      "step": 103700
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.167536604298831e-05,
+      "loss": 0.2472,
+      "step": 103710
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.1674027677400364e-05,
+      "loss": 0.0783,
+      "step": 103720
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.167268931181242e-05,
+      "loss": 0.1113,
+      "step": 103730
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.1671350946224475e-05,
+      "loss": 0.158,
+      "step": 103740
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.167001258063653e-05,
+      "loss": 0.0974,
+      "step": 103750
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.1668674215048586e-05,
+      "loss": 0.1098,
+      "step": 103760
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.166733584946064e-05,
+      "loss": 0.1919,
+      "step": 103770
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.1665997483872696e-05,
+      "loss": 0.1004,
+      "step": 103780
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.166465911828475e-05,
+      "loss": 0.1364,
+      "step": 103790
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.1663320752696814e-05,
+      "loss": 0.0517,
+      "step": 103800
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.166198238710887e-05,
+      "loss": 0.078,
+      "step": 103810
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.1660644021520924e-05,
+      "loss": 0.0114,
+      "step": 103820
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.165930565593298e-05,
+      "loss": 0.0694,
+      "step": 103830
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.1657967290345035e-05,
+      "loss": 0.0834,
+      "step": 103840
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.165662892475709e-05,
+      "loss": 0.114,
+      "step": 103850
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.1655290559169146e-05,
+      "loss": 0.1204,
+      "step": 103860
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.16539521935812e-05,
+      "loss": 0.1086,
+      "step": 103870
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.1652613827993256e-05,
+      "loss": 0.0893,
+      "step": 103880
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.165127546240531e-05,
+      "loss": 0.1685,
+      "step": 103890
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.1649937096817374e-05,
+      "loss": 0.0202,
+      "step": 103900
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.164859873122943e-05,
+      "loss": 0.1713,
+      "step": 103910
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.1647260365641485e-05,
+      "loss": 0.0291,
+      "step": 103920
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.164592200005354e-05,
+      "loss": 0.0755,
+      "step": 103930
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.1644583634465595e-05,
+      "loss": 0.1598,
+      "step": 103940
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.164324526887765e-05,
+      "loss": 0.1514,
+      "step": 103950
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.1641906903289706e-05,
+      "loss": 0.0536,
+      "step": 103960
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.164056853770176e-05,
+      "loss": 0.2446,
+      "step": 103970
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.163923017211382e-05,
+      "loss": 0.0422,
+      "step": 103980
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.163789180652587e-05,
+      "loss": 0.1284,
+      "step": 103990
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.1636553440937934e-05,
+      "loss": 0.0783,
+      "step": 104000
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.163521507534999e-05,
+      "loss": 0.0065,
+      "step": 104010
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.1633876709762045e-05,
+      "loss": 0.0833,
+      "step": 104020
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.16325383441741e-05,
+      "loss": 0.1346,
+      "step": 104030
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.1631199978586155e-05,
+      "loss": 0.1368,
+      "step": 104040
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.162986161299821e-05,
+      "loss": 0.0976,
+      "step": 104050
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.1628523247410266e-05,
+      "loss": 0.1712,
+      "step": 104060
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.162718488182232e-05,
+      "loss": 0.2655,
+      "step": 104070
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.162584651623438e-05,
+      "loss": 0.2474,
+      "step": 104080
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.162450815064643e-05,
+      "loss": 0.2587,
+      "step": 104090
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.1623169785058494e-05,
+      "loss": 0.145,
+      "step": 104100
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.162183141947055e-05,
+      "loss": 0.0414,
+      "step": 104110
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.1620493053882605e-05,
+      "loss": 0.2574,
+      "step": 104120
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.161915468829466e-05,
+      "loss": 0.2041,
+      "step": 104130
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.1617816322706716e-05,
+      "loss": 0.1274,
+      "step": 104140
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.161647795711877e-05,
+      "loss": 0.2299,
+      "step": 104150
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.1615139591530826e-05,
+      "loss": 0.1373,
+      "step": 104160
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.1613801225942875e-05,
+      "loss": 0.2498,
+      "step": 104170
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.161246286035494e-05,
+      "loss": 0.2247,
+      "step": 104180
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.161112449476699e-05,
+      "loss": 0.1272,
+      "step": 104190
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.160978612917905e-05,
+      "loss": 0.1555,
+      "step": 104200
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.16084477635911e-05,
+      "loss": 0.185,
+      "step": 104210
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.160710939800316e-05,
+      "loss": 0.0934,
+      "step": 104220
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.1605771032415214e-05,
+      "loss": 0.0613,
+      "step": 104230
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.160443266682727e-05,
+      "loss": 0.2109,
+      "step": 104240
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.1603094301239324e-05,
+      "loss": 0.0302,
+      "step": 104250
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.160175593565138e-05,
+      "loss": 0.0505,
+      "step": 104260
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.1600417570063435e-05,
+      "loss": 0.1609,
+      "step": 104270
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.15990792044755e-05,
+      "loss": 0.0126,
+      "step": 104280
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.159774083888755e-05,
+      "loss": 0.0356,
+      "step": 104290
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.159640247329961e-05,
+      "loss": 0.2707,
+      "step": 104300
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.159506410771166e-05,
+      "loss": 0.0914,
+      "step": 104310
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.159372574212372e-05,
+      "loss": 0.0712,
+      "step": 104320
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.1592387376535774e-05,
+      "loss": 0.0653,
+      "step": 104330
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.159104901094783e-05,
+      "loss": 0.1222,
+      "step": 104340
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.1589710645359884e-05,
+      "loss": 0.1358,
+      "step": 104350
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.158837227977194e-05,
+      "loss": 0.1491,
+      "step": 104360
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.1587033914183995e-05,
+      "loss": 0.1961,
+      "step": 104370
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.158569554859606e-05,
+      "loss": 0.11,
+      "step": 104380
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.158435718300811e-05,
+      "loss": 0.1892,
+      "step": 104390
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.158301881742017e-05,
+      "loss": 0.0858,
+      "step": 104400
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.158168045183222e-05,
+      "loss": 0.1183,
+      "step": 104410
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.158034208624428e-05,
+      "loss": 0.0875,
+      "step": 104420
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.1579003720656334e-05,
+      "loss": 0.1998,
+      "step": 104430
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.157766535506839e-05,
+      "loss": 0.1443,
+      "step": 104440
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.1576326989480445e-05,
+      "loss": 0.2843,
+      "step": 104450
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.15749886238925e-05,
+      "loss": 0.1659,
+      "step": 104460
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.1573650258304555e-05,
+      "loss": 0.1347,
+      "step": 104470
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.157231189271662e-05,
+      "loss": 0.1246,
+      "step": 104480
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.157097352712867e-05,
+      "loss": 0.036,
+      "step": 104490
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.156963516154073e-05,
+      "loss": 0.027,
+      "step": 104500
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.1568296795952783e-05,
+      "loss": 0.0718,
+      "step": 104510
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.156695843036484e-05,
+      "loss": 0.1755,
+      "step": 104520
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.1565620064776894e-05,
+      "loss": 0.1339,
+      "step": 104530
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.156428169918895e-05,
+      "loss": 0.0771,
+      "step": 104540
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.1562943333601005e-05,
+      "loss": 0.3128,
+      "step": 104550
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.156160496801306e-05,
+      "loss": 0.1853,
+      "step": 104560
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.1560266602425115e-05,
+      "loss": 0.0761,
+      "step": 104570
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.155892823683718e-05,
+      "loss": 0.0953,
+      "step": 104580
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.155758987124923e-05,
+      "loss": 0.2457,
+      "step": 104590
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.155625150566129e-05,
+      "loss": 0.1379,
+      "step": 104600
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.1554913140073344e-05,
+      "loss": 0.0241,
+      "step": 104610
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.15535747744854e-05,
+      "loss": 0.0846,
+      "step": 104620
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.1552236408897454e-05,
+      "loss": 0.1799,
+      "step": 104630
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.155089804330951e-05,
+      "loss": 0.1301,
+      "step": 104640
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.1549559677721565e-05,
+      "loss": 0.0675,
+      "step": 104650
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.154822131213362e-05,
+      "loss": 0.1363,
+      "step": 104660
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.1546882946545676e-05,
+      "loss": 0.0768,
+      "step": 104670
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.154554458095774e-05,
+      "loss": 0.1451,
+      "step": 104680
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.154420621536979e-05,
+      "loss": 0.137,
+      "step": 104690
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.154286784978185e-05,
+      "loss": 0.2528,
+      "step": 104700
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.1541529484193904e-05,
+      "loss": 0.1719,
+      "step": 104710
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.154019111860596e-05,
+      "loss": 0.278,
+      "step": 104720
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.1538852753018014e-05,
+      "loss": 0.0726,
+      "step": 104730
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.153751438743007e-05,
+      "loss": 0.0578,
+      "step": 104740
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.1536176021842125e-05,
+      "loss": 0.0364,
+      "step": 104750
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.153483765625418e-05,
+      "loss": 0.1716,
+      "step": 104760
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.1533499290666236e-05,
+      "loss": 0.1127,
+      "step": 104770
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.15321609250783e-05,
+      "loss": 0.0411,
+      "step": 104780
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.153082255949035e-05,
+      "loss": 0.0629,
+      "step": 104790
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.152948419390241e-05,
+      "loss": 0.0042,
+      "step": 104800
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.1528145828314464e-05,
+      "loss": 0.0693,
+      "step": 104810
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.152680746272652e-05,
+      "loss": 0.2076,
+      "step": 104820
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.1525469097138575e-05,
+      "loss": 0.2697,
+      "step": 104830
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.152413073155063e-05,
+      "loss": 0.1484,
+      "step": 104840
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.1522792365962685e-05,
+      "loss": 0.2347,
+      "step": 104850
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.152145400037474e-05,
+      "loss": 0.2174,
+      "step": 104860
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.1520115634786796e-05,
+      "loss": 0.1864,
+      "step": 104870
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.151877726919886e-05,
+      "loss": 0.0751,
+      "step": 104880
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.151743890361091e-05,
+      "loss": 0.1766,
+      "step": 104890
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.151610053802297e-05,
+      "loss": 0.2314,
+      "step": 104900
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.1514762172435024e-05,
+      "loss": 0.2807,
+      "step": 104910
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.151342380684708e-05,
+      "loss": 0.3232,
+      "step": 104920
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.1512085441259135e-05,
+      "loss": 0.0508,
+      "step": 104930
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.151074707567119e-05,
+      "loss": 0.1403,
+      "step": 104940
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.1509408710083245e-05,
+      "loss": 0.1177,
+      "step": 104950
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.15080703444953e-05,
+      "loss": 0.0086,
+      "step": 104960
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.1506731978907356e-05,
+      "loss": 0.1083,
+      "step": 104970
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.150539361331942e-05,
+      "loss": 0.1335,
+      "step": 104980
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.1504055247731474e-05,
+      "loss": 0.1616,
+      "step": 104990
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.150271688214353e-05,
+      "loss": 0.233,
+      "step": 105000
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.1501378516555584e-05,
+      "loss": 0.1207,
+      "step": 105010
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.150004015096764e-05,
+      "loss": 0.0324,
+      "step": 105020
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.1498701785379695e-05,
+      "loss": 0.1736,
+      "step": 105030
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.149736341979175e-05,
+      "loss": 0.1413,
+      "step": 105040
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.1496025054203806e-05,
+      "loss": 0.1375,
+      "step": 105050
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.149468668861586e-05,
+      "loss": 0.1268,
+      "step": 105060
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.149334832302792e-05,
+      "loss": 0.0887,
+      "step": 105070
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.149200995743998e-05,
+      "loss": 0.1005,
+      "step": 105080
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.1490671591852034e-05,
+      "loss": 0.1065,
+      "step": 105090
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.148933322626409e-05,
+      "loss": 0.0456,
+      "step": 105100
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.1487994860676144e-05,
+      "loss": 0.1075,
+      "step": 105110
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.14866564950882e-05,
+      "loss": 0.1758,
+      "step": 105120
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.1485318129500255e-05,
+      "loss": 0.2615,
+      "step": 105130
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.148397976391231e-05,
+      "loss": 0.1168,
+      "step": 105140
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.1482641398324366e-05,
+      "loss": 0.2072,
+      "step": 105150
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.148130303273642e-05,
+      "loss": 0.0364,
+      "step": 105160
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.147996466714848e-05,
+      "loss": 0.0554,
+      "step": 105170
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.147862630156054e-05,
+      "loss": 0.1204,
+      "step": 105180
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.1477287935972594e-05,
+      "loss": 0.1382,
+      "step": 105190
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.147594957038465e-05,
+      "loss": 0.1228,
+      "step": 105200
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.1474611204796704e-05,
+      "loss": 0.0973,
+      "step": 105210
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.147327283920876e-05,
+      "loss": 0.231,
+      "step": 105220
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.1471934473620815e-05,
+      "loss": 0.2768,
+      "step": 105230
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.147059610803287e-05,
+      "loss": 0.3124,
+      "step": 105240
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.1469257742444926e-05,
+      "loss": 0.2496,
+      "step": 105250
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.146791937685698e-05,
+      "loss": 0.1195,
+      "step": 105260
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.146658101126904e-05,
+      "loss": 0.1402,
+      "step": 105270
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.14652426456811e-05,
+      "loss": 0.2264,
+      "step": 105280
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.1463904280093154e-05,
+      "loss": 0.1297,
+      "step": 105290
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.146256591450521e-05,
+      "loss": 0.1926,
+      "step": 105300
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.1461227548917265e-05,
+      "loss": 0.0943,
+      "step": 105310
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.145988918332932e-05,
+      "loss": 0.1546,
+      "step": 105320
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.1458550817741375e-05,
+      "loss": 0.1254,
+      "step": 105330
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.145721245215343e-05,
+      "loss": 0.1595,
+      "step": 105340
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.1455874086565486e-05,
+      "loss": 0.123,
+      "step": 105350
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.145453572097754e-05,
+      "loss": 0.1791,
+      "step": 105360
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.1453197355389603e-05,
+      "loss": 0.0995,
+      "step": 105370
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.145185898980166e-05,
+      "loss": 0.1532,
+      "step": 105380
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.1450520624213714e-05,
+      "loss": 0.0807,
+      "step": 105390
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.144918225862577e-05,
+      "loss": 0.1421,
+      "step": 105400
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.1447843893037825e-05,
+      "loss": 0.1593,
+      "step": 105410
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.144650552744988e-05,
+      "loss": 0.0922,
+      "step": 105420
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.1445167161861935e-05,
+      "loss": 0.1384,
+      "step": 105430
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.144382879627399e-05,
+      "loss": 0.2768,
+      "step": 105440
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.1442490430686046e-05,
+      "loss": 0.0947,
+      "step": 105450
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.14411520650981e-05,
+      "loss": 0.0485,
+      "step": 105460
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.1439813699510164e-05,
+      "loss": 0.0735,
+      "step": 105470
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.143847533392222e-05,
+      "loss": 0.0462,
+      "step": 105480
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.1437136968334274e-05,
+      "loss": 0.1805,
+      "step": 105490
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.143579860274633e-05,
+      "loss": 0.3828,
+      "step": 105500
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.1434460237158385e-05,
+      "loss": 0.0948,
+      "step": 105510
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.143312187157044e-05,
+      "loss": 0.1331,
+      "step": 105520
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.1431783505982496e-05,
+      "loss": 0.0838,
+      "step": 105530
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.143044514039455e-05,
+      "loss": 0.1833,
+      "step": 105540
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.1429106774806606e-05,
+      "loss": 0.1991,
+      "step": 105550
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.142776840921866e-05,
+      "loss": 0.0454,
+      "step": 105560
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.1426430043630724e-05,
+      "loss": 0.1393,
+      "step": 105570
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.142509167804278e-05,
+      "loss": 0.1968,
+      "step": 105580
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.1423753312454834e-05,
+      "loss": 0.1894,
+      "step": 105590
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.142241494686689e-05,
+      "loss": 0.2881,
+      "step": 105600
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.1421076581278945e-05,
+      "loss": 0.0484,
+      "step": 105610
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.1419738215691e-05,
+      "loss": 0.0571,
+      "step": 105620
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.1418399850103056e-05,
+      "loss": 0.1769,
+      "step": 105630
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.141706148451511e-05,
+      "loss": 0.1079,
+      "step": 105640
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.1415723118927166e-05,
+      "loss": 0.0688,
+      "step": 105650
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.141438475333922e-05,
+      "loss": 0.0615,
+      "step": 105660
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.1413046387751284e-05,
+      "loss": 0.1051,
+      "step": 105670
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.141170802216334e-05,
+      "loss": 0.11,
+      "step": 105680
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.1410369656575395e-05,
+      "loss": 0.0067,
+      "step": 105690
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.140903129098745e-05,
+      "loss": 0.0852,
+      "step": 105700
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.1407692925399505e-05,
+      "loss": 0.1116,
+      "step": 105710
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.140635455981156e-05,
+      "loss": 0.1035,
+      "step": 105720
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.1405016194223616e-05,
+      "loss": 0.1173,
+      "step": 105730
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.140367782863567e-05,
+      "loss": 0.0805,
+      "step": 105740
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.1402339463047727e-05,
+      "loss": 0.125,
+      "step": 105750
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.140100109745978e-05,
+      "loss": 0.2467,
+      "step": 105760
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.1399662731871844e-05,
+      "loss": 0.2109,
+      "step": 105770
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.13983243662839e-05,
+      "loss": 0.0819,
+      "step": 105780
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.1396986000695955e-05,
+      "loss": 0.0606,
+      "step": 105790
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.139564763510801e-05,
+      "loss": 0.0509,
+      "step": 105800
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.1394309269520065e-05,
+      "loss": 0.062,
+      "step": 105810
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.139297090393212e-05,
+      "loss": 0.2755,
+      "step": 105820
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.1391632538344176e-05,
+      "loss": 0.2977,
+      "step": 105830
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.139029417275623e-05,
+      "loss": 0.0942,
+      "step": 105840
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.138895580716829e-05,
+      "loss": 0.03,
+      "step": 105850
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.138761744158034e-05,
+      "loss": 0.2378,
+      "step": 105860
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.1386279075992404e-05,
+      "loss": 0.0917,
+      "step": 105870
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.138494071040446e-05,
+      "loss": 0.0187,
+      "step": 105880
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.1383602344816515e-05,
+      "loss": 0.3536,
+      "step": 105890
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.138226397922857e-05,
+      "loss": 0.1804,
+      "step": 105900
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.1380925613640626e-05,
+      "loss": 0.1938,
+      "step": 105910
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.137958724805268e-05,
+      "loss": 0.1009,
+      "step": 105920
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.1378248882464736e-05,
+      "loss": 0.05,
+      "step": 105930
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.137691051687679e-05,
+      "loss": 0.167,
+      "step": 105940
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.137557215128885e-05,
+      "loss": 0.1427,
+      "step": 105950
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.13742337857009e-05,
+      "loss": 0.0892,
+      "step": 105960
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.1372895420112964e-05,
+      "loss": 0.2191,
+      "step": 105970
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.137155705452502e-05,
+      "loss": 0.1043,
+      "step": 105980
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.1370218688937075e-05,
+      "loss": 0.0661,
+      "step": 105990
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.136888032334913e-05,
+      "loss": 0.0335,
+      "step": 106000
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.1367541957761186e-05,
+      "loss": 0.0425,
+      "step": 106010
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.136620359217324e-05,
+      "loss": 0.1887,
+      "step": 106020
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.1364865226585296e-05,
+      "loss": 0.2655,
+      "step": 106030
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.136352686099735e-05,
+      "loss": 0.1396,
+      "step": 106040
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.136218849540941e-05,
+      "loss": 0.1053,
+      "step": 106050
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.136085012982146e-05,
+      "loss": 0.2711,
+      "step": 106060
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.1359511764233525e-05,
+      "loss": 0.1535,
+      "step": 106070
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.135817339864558e-05,
+      "loss": 0.1987,
+      "step": 106080
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.1356835033057635e-05,
+      "loss": 0.1258,
+      "step": 106090
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.135549666746969e-05,
+      "loss": 0.1104,
+      "step": 106100
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.1354158301881746e-05,
+      "loss": 0.0628,
+      "step": 106110
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.13528199362938e-05,
+      "loss": 0.0549,
+      "step": 106120
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.1351481570705857e-05,
+      "loss": 0.2229,
+      "step": 106130
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.135014320511791e-05,
+      "loss": 0.1625,
+      "step": 106140
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.134880483952997e-05,
+      "loss": 0.0595,
+      "step": 106150
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.134746647394202e-05,
+      "loss": 0.1849,
+      "step": 106160
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.1346128108354085e-05,
+      "loss": 0.2511,
+      "step": 106170
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.134478974276614e-05,
+      "loss": 0.2932,
+      "step": 106180
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.1343451377178195e-05,
+      "loss": 0.3472,
+      "step": 106190
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.134211301159025e-05,
+      "loss": 0.2085,
+      "step": 106200
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.1340774646002306e-05,
+      "loss": 0.1225,
+      "step": 106210
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.133943628041436e-05,
+      "loss": 0.1207,
+      "step": 106220
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.133809791482642e-05,
+      "loss": 0.1208,
+      "step": 106230
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.133675954923847e-05,
+      "loss": 0.1305,
+      "step": 106240
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.133542118365053e-05,
+      "loss": 0.1269,
+      "step": 106250
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.133408281806259e-05,
+      "loss": 0.1576,
+      "step": 106260
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.1332744452474645e-05,
+      "loss": 0.088,
+      "step": 106270
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.13314060868867e-05,
+      "loss": 0.0308,
+      "step": 106280
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.1330067721298755e-05,
+      "loss": 0.1093,
+      "step": 106290
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.132872935571081e-05,
+      "loss": 0.1035,
+      "step": 106300
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.1327390990122866e-05,
+      "loss": 0.3029,
+      "step": 106310
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.132605262453492e-05,
+      "loss": 0.2785,
+      "step": 106320
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.132471425894698e-05,
+      "loss": 0.1245,
+      "step": 106330
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.132337589335903e-05,
+      "loss": 0.0563,
+      "step": 106340
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.132203752777109e-05,
+      "loss": 0.2312,
+      "step": 106350
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.132069916218315e-05,
+      "loss": 0.0119,
+      "step": 106360
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.1319360796595205e-05,
+      "loss": 0.1578,
+      "step": 106370
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.131802243100726e-05,
+      "loss": 0.117,
+      "step": 106380
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.1316684065419316e-05,
+      "loss": 0.1733,
+      "step": 106390
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.131534569983137e-05,
+      "loss": 0.2392,
+      "step": 106400
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.1314007334243426e-05,
+      "loss": 0.1585,
+      "step": 106410
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.131266896865548e-05,
+      "loss": 0.0971,
+      "step": 106420
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.131133060306754e-05,
+      "loss": 0.0656,
+      "step": 106430
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.130999223747959e-05,
+      "loss": 0.205,
+      "step": 106440
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.130865387189165e-05,
+      "loss": 0.1207,
+      "step": 106450
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.130731550630371e-05,
+      "loss": 0.0991,
+      "step": 106460
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.1305977140715765e-05,
+      "loss": 0.2299,
+      "step": 106470
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.130463877512782e-05,
+      "loss": 0.1203,
+      "step": 106480
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.1303300409539876e-05,
+      "loss": 0.0719,
+      "step": 106490
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.130196204395193e-05,
+      "loss": 0.0774,
+      "step": 106500
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.1300623678363986e-05,
+      "loss": 0.18,
+      "step": 106510
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.129928531277604e-05,
+      "loss": 0.1032,
+      "step": 106520
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.12979469471881e-05,
+      "loss": 0.1603,
+      "step": 106530
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.129660858160015e-05,
+      "loss": 0.1234,
+      "step": 106540
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.129527021601221e-05,
+      "loss": 0.089,
+      "step": 106550
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.129393185042426e-05,
+      "loss": 0.0899,
+      "step": 106560
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.129259348483632e-05,
+      "loss": 0.0244,
+      "step": 106570
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.1291255119248374e-05,
+      "loss": 0.3862,
+      "step": 106580
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.128991675366043e-05,
+      "loss": 0.0507,
+      "step": 106590
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.1288578388072485e-05,
+      "loss": 0.1503,
+      "step": 106600
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.128724002248454e-05,
+      "loss": 0.0576,
+      "step": 106610
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.1285901656896595e-05,
+      "loss": 0.1594,
+      "step": 106620
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.128456329130865e-05,
+      "loss": 0.178,
+      "step": 106630
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.128322492572071e-05,
+      "loss": 0.0083,
+      "step": 106640
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.128188656013277e-05,
+      "loss": 0.061,
+      "step": 106650
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.128054819454482e-05,
+      "loss": 0.1343,
+      "step": 106660
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.127920982895688e-05,
+      "loss": 0.1646,
+      "step": 106670
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.1277871463368934e-05,
+      "loss": 0.0962,
+      "step": 106680
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.127653309778099e-05,
+      "loss": 0.1637,
+      "step": 106690
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.1275194732193045e-05,
+      "loss": 0.2044,
+      "step": 106700
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.12738563666051e-05,
+      "loss": 0.0382,
+      "step": 106710
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.1272518001017155e-05,
+      "loss": 0.0279,
+      "step": 106720
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.127117963542921e-05,
+      "loss": 0.1053,
+      "step": 106730
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.126984126984127e-05,
+      "loss": 0.0964,
+      "step": 106740
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.126850290425333e-05,
+      "loss": 0.0106,
+      "step": 106750
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.1267164538665383e-05,
+      "loss": 0.0747,
+      "step": 106760
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.126582617307744e-05,
+      "loss": 0.0782,
+      "step": 106770
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.1264487807489494e-05,
+      "loss": 0.3482,
+      "step": 106780
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.126314944190155e-05,
+      "loss": 0.141,
+      "step": 106790
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.1261811076313605e-05,
+      "loss": 0.0201,
+      "step": 106800
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.126047271072566e-05,
+      "loss": 0.0139,
+      "step": 106810
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.1259134345137715e-05,
+      "loss": 0.0074,
+      "step": 106820
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.125779597954977e-05,
+      "loss": 0.1336,
+      "step": 106830
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.125645761396183e-05,
+      "loss": 0.1325,
+      "step": 106840
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.125511924837389e-05,
+      "loss": 0.0769,
+      "step": 106850
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.1253780882785944e-05,
+      "loss": 0.109,
+      "step": 106860
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.1252442517198e-05,
+      "loss": 0.0658,
+      "step": 106870
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.1251104151610054e-05,
+      "loss": 0.0385,
+      "step": 106880
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.124976578602211e-05,
+      "loss": 0.1684,
+      "step": 106890
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.1248427420434165e-05,
+      "loss": 0.1163,
+      "step": 106900
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.124708905484622e-05,
+      "loss": 0.1718,
+      "step": 106910
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.1245750689258276e-05,
+      "loss": 0.247,
+      "step": 106920
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.124441232367033e-05,
+      "loss": 0.3002,
+      "step": 106930
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.124307395808239e-05,
+      "loss": 0.1779,
+      "step": 106940
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.124173559249445e-05,
+      "loss": 0.0088,
+      "step": 106950
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.1240397226906504e-05,
+      "loss": 0.1158,
+      "step": 106960
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.123905886131856e-05,
+      "loss": 0.1104,
+      "step": 106970
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.1237720495730614e-05,
+      "loss": 0.131,
+      "step": 106980
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.123638213014267e-05,
+      "loss": 0.0529,
+      "step": 106990
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.1235043764554725e-05,
+      "loss": 0.3061,
+      "step": 107000
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.123370539896678e-05,
+      "loss": 0.0677,
+      "step": 107010
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.1232367033378836e-05,
+      "loss": 0.074,
+      "step": 107020
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.123102866779089e-05,
+      "loss": 0.2118,
+      "step": 107030
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.122969030220295e-05,
+      "loss": 0.1235,
+      "step": 107040
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.122835193661501e-05,
+      "loss": 0.0351,
+      "step": 107050
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.1227013571027064e-05,
+      "loss": 0.1599,
+      "step": 107060
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.122567520543912e-05,
+      "loss": 0.2242,
+      "step": 107070
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.1224336839851175e-05,
+      "loss": 0.3126,
+      "step": 107080
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.122299847426323e-05,
+      "loss": 0.1184,
+      "step": 107090
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.1221660108675285e-05,
+      "loss": 0.0768,
+      "step": 107100
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.122032174308734e-05,
+      "loss": 0.0454,
+      "step": 107110
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.1218983377499396e-05,
+      "loss": 0.1556,
+      "step": 107120
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.121764501191145e-05,
+      "loss": 0.2643,
+      "step": 107130
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.1216306646323513e-05,
+      "loss": 0.0077,
+      "step": 107140
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.121496828073557e-05,
+      "loss": 0.1608,
+      "step": 107150
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.1213629915147624e-05,
+      "loss": 0.1446,
+      "step": 107160
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.121229154955968e-05,
+      "loss": 0.2064,
+      "step": 107170
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.1210953183971735e-05,
+      "loss": 0.0575,
+      "step": 107180
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.120961481838379e-05,
+      "loss": 0.1648,
+      "step": 107190
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.1208276452795845e-05,
+      "loss": 0.1494,
+      "step": 107200
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.12069380872079e-05,
+      "loss": 0.1698,
+      "step": 107210
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.1205599721619956e-05,
+      "loss": 0.0761,
+      "step": 107220
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.120426135603201e-05,
+      "loss": 0.1105,
+      "step": 107230
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.1202922990444074e-05,
+      "loss": 0.1383,
+      "step": 107240
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.120158462485613e-05,
+      "loss": 0.0913,
+      "step": 107250
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.1200246259268184e-05,
+      "loss": 0.0752,
+      "step": 107260
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.119890789368024e-05,
+      "loss": 0.0072,
+      "step": 107270
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.1197569528092295e-05,
+      "loss": 0.0567,
+      "step": 107280
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.119623116250435e-05,
+      "loss": 0.1333,
+      "step": 107290
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.1194892796916406e-05,
+      "loss": 0.0143,
+      "step": 107300
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.119355443132846e-05,
+      "loss": 0.1837,
+      "step": 107310
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.1192216065740516e-05,
+      "loss": 0.1491,
+      "step": 107320
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.119087770015257e-05,
+      "loss": 0.2837,
+      "step": 107330
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.1189539334564634e-05,
+      "loss": 0.2563,
+      "step": 107340
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.118820096897669e-05,
+      "loss": 0.3463,
+      "step": 107350
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.1186862603388744e-05,
+      "loss": 0.373,
+      "step": 107360
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.11855242378008e-05,
+      "loss": 0.1212,
+      "step": 107370
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.1184185872212855e-05,
+      "loss": 0.1582,
+      "step": 107380
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.118284750662491e-05,
+      "loss": 0.1917,
+      "step": 107390
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.1181509141036966e-05,
+      "loss": 0.0564,
+      "step": 107400
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.118017077544902e-05,
+      "loss": 0.1597,
+      "step": 107410
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.1178832409861076e-05,
+      "loss": 0.0424,
+      "step": 107420
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.117749404427313e-05,
+      "loss": 0.1037,
+      "step": 107430
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.1176155678685194e-05,
+      "loss": 0.0394,
+      "step": 107440
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.117481731309725e-05,
+      "loss": 0.1082,
+      "step": 107450
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.1173478947509305e-05,
+      "loss": 0.0387,
+      "step": 107460
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.117214058192136e-05,
+      "loss": 0.1564,
+      "step": 107470
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.1170802216333415e-05,
+      "loss": 0.0648,
+      "step": 107480
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.116946385074547e-05,
+      "loss": 0.0779,
+      "step": 107490
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.1168125485157526e-05,
+      "loss": 0.0717,
+      "step": 107500
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.116678711956958e-05,
+      "loss": 0.0372,
+      "step": 107510
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.1165448753981637e-05,
+      "loss": 0.0854,
+      "step": 107520
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.116411038839369e-05,
+      "loss": 0.2256,
+      "step": 107530
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.1162772022805754e-05,
+      "loss": 0.2178,
+      "step": 107540
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.116143365721781e-05,
+      "loss": 0.0507,
+      "step": 107550
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.1160095291629865e-05,
+      "loss": 0.2501,
+      "step": 107560
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.115875692604192e-05,
+      "loss": 0.1118,
+      "step": 107570
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.1157418560453975e-05,
+      "loss": 0.134,
+      "step": 107580
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.115608019486603e-05,
+      "loss": 0.073,
+      "step": 107590
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.1154741829278086e-05,
+      "loss": 0.0957,
+      "step": 107600
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.115340346369014e-05,
+      "loss": 0.0415,
+      "step": 107610
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.11520650981022e-05,
+      "loss": 0.1939,
+      "step": 107620
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.115072673251425e-05,
+      "loss": 0.1454,
+      "step": 107630
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.1149388366926314e-05,
+      "loss": 0.2775,
+      "step": 107640
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.114805000133837e-05,
+      "loss": 0.2571,
+      "step": 107650
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.1146711635750425e-05,
+      "loss": 0.0464,
+      "step": 107660
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.114537327016248e-05,
+      "loss": 0.1945,
+      "step": 107670
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.1144034904574536e-05,
+      "loss": 0.0482,
+      "step": 107680
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.114269653898659e-05,
+      "loss": 0.0107,
+      "step": 107690
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.1141358173398646e-05,
+      "loss": 0.0913,
+      "step": 107700
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.11400198078107e-05,
+      "loss": 0.2284,
+      "step": 107710
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.113868144222276e-05,
+      "loss": 0.0226,
+      "step": 107720
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.113734307663481e-05,
+      "loss": 0.0852,
+      "step": 107730
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.1136004711046874e-05,
+      "loss": 0.0634,
+      "step": 107740
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.113466634545893e-05,
+      "loss": 0.0344,
+      "step": 107750
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.1133327979870985e-05,
+      "loss": 0.1024,
+      "step": 107760
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.113198961428304e-05,
+      "loss": 0.3203,
+      "step": 107770
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.1130651248695096e-05,
+      "loss": 0.1918,
+      "step": 107780
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.112931288310715e-05,
+      "loss": 0.2146,
+      "step": 107790
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.1127974517519206e-05,
+      "loss": 0.1833,
+      "step": 107800
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.112663615193126e-05,
+      "loss": 0.0988,
+      "step": 107810
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.112529778634332e-05,
+      "loss": 0.0409,
+      "step": 107820
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.112395942075538e-05,
+      "loss": 0.1563,
+      "step": 107830
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.1122621055167434e-05,
+      "loss": 0.0812,
+      "step": 107840
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.112128268957949e-05,
+      "loss": 0.033,
+      "step": 107850
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.1119944323991545e-05,
+      "loss": 0.321,
+      "step": 107860
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.11186059584036e-05,
+      "loss": 0.0541,
+      "step": 107870
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.1117267592815656e-05,
+      "loss": 0.2484,
+      "step": 107880
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.111592922722771e-05,
+      "loss": 0.0435,
+      "step": 107890
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.1114590861639766e-05,
+      "loss": 0.108,
+      "step": 107900
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.111325249605182e-05,
+      "loss": 0.2254,
+      "step": 107910
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.111191413046388e-05,
+      "loss": 0.0753,
+      "step": 107920
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.111057576487594e-05,
+      "loss": 0.1031,
+      "step": 107930
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": {
+        "accuracy": 0.9443541102077687
+      },
+      "eval_f1": {
+        "f1": 0.9443461050911817
+      },
+      "eval_loss": 0.24347323179244995,
+      "eval_precision": {
+        "precision": 0.9453218450441414
+      },
+      "eval_recall": {
+        "recall": 0.9442028500529185
+      },
+      "eval_runtime": 430.0055,
+      "eval_samples_per_second": 38.616,
+      "eval_steps_per_second": 19.309,
+      "step": 107932
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.1109237399287995e-05,
+      "loss": 0.3587,
+      "step": 107940
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.110789903370005e-05,
+      "loss": 0.1957,
+      "step": 107950
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.1106560668112105e-05,
+      "loss": 0.0862,
+      "step": 107960
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.110522230252416e-05,
+      "loss": 0.0722,
+      "step": 107970
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.1103883936936216e-05,
+      "loss": 0.0489,
+      "step": 107980
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.110254557134827e-05,
+      "loss": 0.1538,
+      "step": 107990
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.110120720576033e-05,
+      "loss": 0.072,
+      "step": 108000
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.109986884017238e-05,
+      "loss": 0.2535,
+      "step": 108010
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.109853047458444e-05,
+      "loss": 0.0043,
+      "step": 108020
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.10971921089965e-05,
+      "loss": 0.0081,
+      "step": 108030
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.1095853743408555e-05,
+      "loss": 0.1669,
+      "step": 108040
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.109451537782061e-05,
+      "loss": 0.1292,
+      "step": 108050
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.1093177012232665e-05,
+      "loss": 0.0782,
+      "step": 108060
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.109183864664472e-05,
+      "loss": 0.264,
+      "step": 108070
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.1090500281056776e-05,
+      "loss": 0.0305,
+      "step": 108080
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.108916191546883e-05,
+      "loss": 0.2785,
+      "step": 108090
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.108782354988089e-05,
+      "loss": 0.1048,
+      "step": 108100
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.108648518429294e-05,
+      "loss": 0.0523,
+      "step": 108110
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.1085146818705e-05,
+      "loss": 0.0582,
+      "step": 108120
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.108380845311706e-05,
+      "loss": 0.2441,
+      "step": 108130
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.1082470087529115e-05,
+      "loss": 0.1756,
+      "step": 108140
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.108113172194117e-05,
+      "loss": 0.3541,
+      "step": 108150
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.1079793356353226e-05,
+      "loss": 0.099,
+      "step": 108160
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.107845499076528e-05,
+      "loss": 0.2463,
+      "step": 108170
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.1077116625177336e-05,
+      "loss": 0.0781,
+      "step": 108180
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.107577825958939e-05,
+      "loss": 0.0828,
+      "step": 108190
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.107443989400145e-05,
+      "loss": 0.0063,
+      "step": 108200
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.10731015284135e-05,
+      "loss": 0.0175,
+      "step": 108210
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.107176316282556e-05,
+      "loss": 0.1063,
+      "step": 108220
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.107042479723762e-05,
+      "loss": 0.0934,
+      "step": 108230
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.1069086431649675e-05,
+      "loss": 0.0511,
+      "step": 108240
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.106774806606173e-05,
+      "loss": 0.0428,
+      "step": 108250
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.1066409700473786e-05,
+      "loss": 0.2017,
+      "step": 108260
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.106507133488584e-05,
+      "loss": 0.1093,
+      "step": 108270
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.1063732969297896e-05,
+      "loss": 0.199,
+      "step": 108280
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.106239460370995e-05,
+      "loss": 0.1379,
+      "step": 108290
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.106105623812201e-05,
+      "loss": 0.056,
+      "step": 108300
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.105971787253406e-05,
+      "loss": 0.1335,
+      "step": 108310
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.105837950694612e-05,
+      "loss": 0.2647,
+      "step": 108320
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.105704114135818e-05,
+      "loss": 0.1778,
+      "step": 108330
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.1055702775770235e-05,
+      "loss": 0.121,
+      "step": 108340
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.105436441018229e-05,
+      "loss": 0.0627,
+      "step": 108350
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.1053026044594346e-05,
+      "loss": 0.0307,
+      "step": 108360
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.10516876790064e-05,
+      "loss": 0.0426,
+      "step": 108370
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.1050349313418457e-05,
+      "loss": 0.1118,
+      "step": 108380
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.104901094783051e-05,
+      "loss": 0.1123,
+      "step": 108390
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.104767258224257e-05,
+      "loss": 0.0655,
+      "step": 108400
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.104633421665462e-05,
+      "loss": 0.1459,
+      "step": 108410
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.104499585106668e-05,
+      "loss": 0.2558,
+      "step": 108420
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.104365748547874e-05,
+      "loss": 0.0602,
+      "step": 108430
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.1042319119890795e-05,
+      "loss": 0.1836,
+      "step": 108440
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.104098075430285e-05,
+      "loss": 0.0793,
+      "step": 108450
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.1039642388714906e-05,
+      "loss": 0.2658,
+      "step": 108460
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.103830402312696e-05,
+      "loss": 0.1019,
+      "step": 108470
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.103696565753902e-05,
+      "loss": 0.1879,
+      "step": 108480
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.103562729195107e-05,
+      "loss": 0.1095,
+      "step": 108490
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.103428892636313e-05,
+      "loss": 0.2679,
+      "step": 108500
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.103295056077518e-05,
+      "loss": 0.1645,
+      "step": 108510
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.103161219518724e-05,
+      "loss": 0.0608,
+      "step": 108520
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.10302738295993e-05,
+      "loss": 0.1127,
+      "step": 108530
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.1028935464011356e-05,
+      "loss": 0.1953,
+      "step": 108540
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.102759709842341e-05,
+      "loss": 0.024,
+      "step": 108550
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.1026258732835466e-05,
+      "loss": 0.0464,
+      "step": 108560
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.102492036724752e-05,
+      "loss": 0.2025,
+      "step": 108570
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.102358200165958e-05,
+      "loss": 0.1083,
+      "step": 108580
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.102224363607163e-05,
+      "loss": 0.1924,
+      "step": 108590
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.102090527048369e-05,
+      "loss": 0.1184,
+      "step": 108600
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.101956690489574e-05,
+      "loss": 0.0914,
+      "step": 108610
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.10182285393078e-05,
+      "loss": 0.1468,
+      "step": 108620
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.101689017371986e-05,
+      "loss": 0.2693,
+      "step": 108630
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.1015551808131916e-05,
+      "loss": 0.0572,
+      "step": 108640
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.101421344254397e-05,
+      "loss": 0.1338,
+      "step": 108650
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.1012875076956026e-05,
+      "loss": 0.0895,
+      "step": 108660
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.101153671136808e-05,
+      "loss": 0.0469,
+      "step": 108670
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.101019834578014e-05,
+      "loss": 0.0148,
+      "step": 108680
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.100885998019219e-05,
+      "loss": 0.1327,
+      "step": 108690
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.100752161460425e-05,
+      "loss": 0.0683,
+      "step": 108700
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.10061832490163e-05,
+      "loss": 0.209,
+      "step": 108710
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.100484488342836e-05,
+      "loss": 0.1096,
+      "step": 108720
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.100350651784042e-05,
+      "loss": 0.0491,
+      "step": 108730
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.1002168152252476e-05,
+      "loss": 0.1185,
+      "step": 108740
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.100082978666453e-05,
+      "loss": 0.1227,
+      "step": 108750
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.0999491421076587e-05,
+      "loss": 0.0473,
+      "step": 108760
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.099815305548864e-05,
+      "loss": 0.0528,
+      "step": 108770
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.09968146899007e-05,
+      "loss": 0.1311,
+      "step": 108780
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.099547632431275e-05,
+      "loss": 0.1101,
+      "step": 108790
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.099413795872481e-05,
+      "loss": 0.2096,
+      "step": 108800
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.099279959313686e-05,
+      "loss": 0.1269,
+      "step": 108810
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.099146122754892e-05,
+      "loss": 0.2057,
+      "step": 108820
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.099012286196098e-05,
+      "loss": 0.2473,
+      "step": 108830
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.0988784496373036e-05,
+      "loss": 0.0604,
+      "step": 108840
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.098744613078509e-05,
+      "loss": 0.2575,
+      "step": 108850
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.098610776519715e-05,
+      "loss": 0.1298,
+      "step": 108860
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.09847693996092e-05,
+      "loss": 0.0696,
+      "step": 108870
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.098343103402126e-05,
+      "loss": 0.0802,
+      "step": 108880
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.098209266843331e-05,
+      "loss": 0.107,
+      "step": 108890
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.098075430284536e-05,
+      "loss": 0.2155,
+      "step": 108900
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.097941593725742e-05,
+      "loss": 0.2428,
+      "step": 108910
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.097807757166948e-05,
+      "loss": 0.3027,
+      "step": 108920
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.0976739206081534e-05,
+      "loss": 0.0917,
+      "step": 108930
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.097540084049359e-05,
+      "loss": 0.1903,
+      "step": 108940
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.0974062474905645e-05,
+      "loss": 0.0875,
+      "step": 108950
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.09727241093177e-05,
+      "loss": 0.0442,
+      "step": 108960
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.0971385743729755e-05,
+      "loss": 0.18,
+      "step": 108970
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.097004737814181e-05,
+      "loss": 0.124,
+      "step": 108980
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.0968709012553866e-05,
+      "loss": 0.1379,
+      "step": 108990
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.096737064696592e-05,
+      "loss": 0.0837,
+      "step": 109000
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.0966032281377984e-05,
+      "loss": 0.0447,
+      "step": 109010
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.096469391579004e-05,
+      "loss": 0.0347,
+      "step": 109020
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.0963355550202094e-05,
+      "loss": 0.0705,
+      "step": 109030
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.096201718461415e-05,
+      "loss": 0.1985,
+      "step": 109040
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.0960678819026205e-05,
+      "loss": 0.1158,
+      "step": 109050
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.095934045343826e-05,
+      "loss": 0.1127,
+      "step": 109060
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.0958002087850316e-05,
+      "loss": 0.0268,
+      "step": 109070
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.095666372226237e-05,
+      "loss": 0.2216,
+      "step": 109080
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.0955325356674426e-05,
+      "loss": 0.1097,
+      "step": 109090
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.095398699108648e-05,
+      "loss": 0.2477,
+      "step": 109100
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.0952648625498544e-05,
+      "loss": 0.1595,
+      "step": 109110
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.09513102599106e-05,
+      "loss": 0.0388,
+      "step": 109120
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.0949971894322654e-05,
+      "loss": 0.3489,
+      "step": 109130
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.094863352873471e-05,
+      "loss": 0.0963,
+      "step": 109140
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.0947295163146765e-05,
+      "loss": 0.2354,
+      "step": 109150
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.094595679755882e-05,
+      "loss": 0.1359,
+      "step": 109160
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.0944618431970876e-05,
+      "loss": 0.0796,
+      "step": 109170
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.094328006638293e-05,
+      "loss": 0.0747,
+      "step": 109180
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.0941941700794986e-05,
+      "loss": 0.158,
+      "step": 109190
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.094060333520704e-05,
+      "loss": 0.0697,
+      "step": 109200
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.0939264969619104e-05,
+      "loss": 0.0806,
+      "step": 109210
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.093792660403116e-05,
+      "loss": 0.2066,
+      "step": 109220
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.0936588238443214e-05,
+      "loss": 0.0554,
+      "step": 109230
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.093524987285527e-05,
+      "loss": 0.1383,
+      "step": 109240
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.0933911507267325e-05,
+      "loss": 0.1824,
+      "step": 109250
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.093257314167938e-05,
+      "loss": 0.2171,
+      "step": 109260
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.0931234776091436e-05,
+      "loss": 0.1031,
+      "step": 109270
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.092989641050349e-05,
+      "loss": 0.0753,
+      "step": 109280
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.0928558044915547e-05,
+      "loss": 0.1859,
+      "step": 109290
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.092721967932761e-05,
+      "loss": 0.1406,
+      "step": 109300
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.0925881313739664e-05,
+      "loss": 0.0424,
+      "step": 109310
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.092454294815172e-05,
+      "loss": 0.1993,
+      "step": 109320
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.0923204582563775e-05,
+      "loss": 0.0669,
+      "step": 109330
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.092186621697583e-05,
+      "loss": 0.0377,
+      "step": 109340
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.0920527851387885e-05,
+      "loss": 0.0849,
+      "step": 109350
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.091918948579994e-05,
+      "loss": 0.1963,
+      "step": 109360
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.0917851120211996e-05,
+      "loss": 0.0419,
+      "step": 109370
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.091651275462405e-05,
+      "loss": 0.1137,
+      "step": 109380
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.091517438903611e-05,
+      "loss": 0.2499,
+      "step": 109390
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.091383602344817e-05,
+      "loss": 0.1085,
+      "step": 109400
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.0912497657860224e-05,
+      "loss": 0.0972,
+      "step": 109410
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.091115929227228e-05,
+      "loss": 0.1536,
+      "step": 109420
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.0909820926684335e-05,
+      "loss": 0.1284,
+      "step": 109430
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.090848256109639e-05,
+      "loss": 0.082,
+      "step": 109440
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.0907144195508445e-05,
+      "loss": 0.1885,
+      "step": 109450
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.09058058299205e-05,
+      "loss": 0.0984,
+      "step": 109460
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.0904467464332556e-05,
+      "loss": 0.2073,
+      "step": 109470
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.090312909874461e-05,
+      "loss": 0.0631,
+      "step": 109480
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.090179073315667e-05,
+      "loss": 0.1308,
+      "step": 109490
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.090045236756873e-05,
+      "loss": 0.2594,
+      "step": 109500
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.0899114001980784e-05,
+      "loss": 0.2106,
+      "step": 109510
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.089777563639284e-05,
+      "loss": 0.0113,
+      "step": 109520
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.0896437270804895e-05,
+      "loss": 0.0701,
+      "step": 109530
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.089509890521695e-05,
+      "loss": 0.1839,
+      "step": 109540
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.0893760539629006e-05,
+      "loss": 0.2645,
+      "step": 109550
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.089242217404106e-05,
+      "loss": 0.0576,
+      "step": 109560
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.0891083808453116e-05,
+      "loss": 0.1078,
+      "step": 109570
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.088974544286517e-05,
+      "loss": 0.1426,
+      "step": 109580
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.088840707727723e-05,
+      "loss": 0.0605,
+      "step": 109590
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.088706871168929e-05,
+      "loss": 0.0795,
+      "step": 109600
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.0885730346101344e-05,
+      "loss": 0.1114,
+      "step": 109610
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.08843919805134e-05,
+      "loss": 0.1241,
+      "step": 109620
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.0883053614925455e-05,
+      "loss": 0.1194,
+      "step": 109630
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.088171524933751e-05,
+      "loss": 0.1156,
+      "step": 109640
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.0880376883749566e-05,
+      "loss": 0.1797,
+      "step": 109650
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.087903851816162e-05,
+      "loss": 0.0771,
+      "step": 109660
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.0877700152573676e-05,
+      "loss": 0.1263,
+      "step": 109670
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.087636178698573e-05,
+      "loss": 0.0218,
+      "step": 109680
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.087502342139779e-05,
+      "loss": 0.1191,
+      "step": 109690
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.087368505580985e-05,
+      "loss": 0.0748,
+      "step": 109700
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.0872346690221905e-05,
+      "loss": 0.0926,
+      "step": 109710
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.087100832463396e-05,
+      "loss": 0.0456,
+      "step": 109720
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.0869669959046015e-05,
+      "loss": 0.114,
+      "step": 109730
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.086833159345807e-05,
+      "loss": 0.1018,
+      "step": 109740
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.0866993227870126e-05,
+      "loss": 0.2542,
+      "step": 109750
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.086565486228218e-05,
+      "loss": 0.0455,
+      "step": 109760
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.0864316496694237e-05,
+      "loss": 0.1116,
+      "step": 109770
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.086297813110629e-05,
+      "loss": 0.1661,
+      "step": 109780
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.086163976551835e-05,
+      "loss": 0.0463,
+      "step": 109790
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.086030139993041e-05,
+      "loss": 0.0607,
+      "step": 109800
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.0858963034342465e-05,
+      "loss": 0.0984,
+      "step": 109810
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.085762466875452e-05,
+      "loss": 0.2106,
+      "step": 109820
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.0856286303166575e-05,
+      "loss": 0.0541,
+      "step": 109830
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.085494793757863e-05,
+      "loss": 0.056,
+      "step": 109840
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.0853609571990686e-05,
+      "loss": 0.0289,
+      "step": 109850
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.085227120640274e-05,
+      "loss": 0.1801,
+      "step": 109860
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.08509328408148e-05,
+      "loss": 0.2131,
+      "step": 109870
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.084959447522685e-05,
+      "loss": 0.035,
+      "step": 109880
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.084825610963891e-05,
+      "loss": 0.0397,
+      "step": 109890
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.084691774405097e-05,
+      "loss": 0.2334,
+      "step": 109900
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.0845579378463025e-05,
+      "loss": 0.0876,
+      "step": 109910
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.084424101287508e-05,
+      "loss": 0.1723,
+      "step": 109920
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.0842902647287136e-05,
+      "loss": 0.1612,
+      "step": 109930
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.084156428169919e-05,
+      "loss": 0.1019,
+      "step": 109940
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.0840225916111246e-05,
+      "loss": 0.0696,
+      "step": 109950
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.08388875505233e-05,
+      "loss": 0.2612,
+      "step": 109960
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.083754918493536e-05,
+      "loss": 0.1428,
+      "step": 109970
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.083621081934741e-05,
+      "loss": 0.3325,
+      "step": 109980
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.083487245375947e-05,
+      "loss": 0.068,
+      "step": 109990
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.083353408817153e-05,
+      "loss": 0.2493,
+      "step": 110000
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.0832195722583585e-05,
+      "loss": 0.2099,
+      "step": 110010
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.083085735699564e-05,
+      "loss": 0.2612,
+      "step": 110020
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.0829518991407696e-05,
+      "loss": 0.0711,
+      "step": 110030
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.082818062581975e-05,
+      "loss": 0.1042,
+      "step": 110040
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.0826842260231806e-05,
+      "loss": 0.0435,
+      "step": 110050
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.082550389464386e-05,
+      "loss": 0.0902,
+      "step": 110060
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.082416552905592e-05,
+      "loss": 0.147,
+      "step": 110070
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.082282716346797e-05,
+      "loss": 0.2337,
+      "step": 110080
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.082148879788003e-05,
+      "loss": 0.4818,
+      "step": 110090
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.082015043229209e-05,
+      "loss": 0.0665,
+      "step": 110100
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.0818812066704145e-05,
+      "loss": 0.1648,
+      "step": 110110
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.08174737011162e-05,
+      "loss": 0.0334,
+      "step": 110120
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.0816135335528256e-05,
+      "loss": 0.1065,
+      "step": 110130
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.081479696994031e-05,
+      "loss": 0.0976,
+      "step": 110140
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.0813458604352367e-05,
+      "loss": 0.0604,
+      "step": 110150
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.081212023876442e-05,
+      "loss": 0.162,
+      "step": 110160
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.081078187317648e-05,
+      "loss": 0.0666,
+      "step": 110170
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.080944350758853e-05,
+      "loss": 0.1205,
+      "step": 110180
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.080810514200059e-05,
+      "loss": 0.252,
+      "step": 110190
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.080676677641265e-05,
+      "loss": 0.1638,
+      "step": 110200
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.0805428410824705e-05,
+      "loss": 0.0602,
+      "step": 110210
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.080409004523676e-05,
+      "loss": 0.0053,
+      "step": 110220
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.0802751679648816e-05,
+      "loss": 0.0564,
+      "step": 110230
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.080141331406087e-05,
+      "loss": 0.3206,
+      "step": 110240
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.080007494847293e-05,
+      "loss": 0.0241,
+      "step": 110250
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.079873658288498e-05,
+      "loss": 0.0403,
+      "step": 110260
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.079739821729704e-05,
+      "loss": 0.0885,
+      "step": 110270
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.079605985170909e-05,
+      "loss": 0.0425,
+      "step": 110280
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.079472148612115e-05,
+      "loss": 0.2083,
+      "step": 110290
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.079338312053321e-05,
+      "loss": 0.2183,
+      "step": 110300
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.0792044754945265e-05,
+      "loss": 0.0393,
+      "step": 110310
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.079070638935732e-05,
+      "loss": 0.0342,
+      "step": 110320
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.0789368023769376e-05,
+      "loss": 0.0762,
+      "step": 110330
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.078802965818143e-05,
+      "loss": 0.1173,
+      "step": 110340
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.078669129259349e-05,
+      "loss": 0.1675,
+      "step": 110350
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.078535292700554e-05,
+      "loss": 0.1753,
+      "step": 110360
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.07840145614176e-05,
+      "loss": 0.0519,
+      "step": 110370
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.078267619582965e-05,
+      "loss": 0.1606,
+      "step": 110380
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.078133783024171e-05,
+      "loss": 0.0327,
+      "step": 110390
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.077999946465377e-05,
+      "loss": 0.1315,
+      "step": 110400
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.0778661099065826e-05,
+      "loss": 0.1345,
+      "step": 110410
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.077732273347788e-05,
+      "loss": 0.0026,
+      "step": 110420
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.0775984367889936e-05,
+      "loss": 0.0154,
+      "step": 110430
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.077464600230199e-05,
+      "loss": 0.003,
+      "step": 110440
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.077330763671405e-05,
+      "loss": 0.2007,
+      "step": 110450
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.07719692711261e-05,
+      "loss": 0.0098,
+      "step": 110460
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.077063090553816e-05,
+      "loss": 0.123,
+      "step": 110470
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.076929253995021e-05,
+      "loss": 0.1212,
+      "step": 110480
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.076795417436227e-05,
+      "loss": 0.3023,
+      "step": 110490
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.076661580877433e-05,
+      "loss": 0.1391,
+      "step": 110500
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.0765277443186386e-05,
+      "loss": 0.0793,
+      "step": 110510
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.076393907759844e-05,
+      "loss": 0.0031,
+      "step": 110520
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.0762600712010496e-05,
+      "loss": 0.1047,
+      "step": 110530
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.076126234642255e-05,
+      "loss": 0.1119,
+      "step": 110540
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.075992398083461e-05,
+      "loss": 0.1014,
+      "step": 110550
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.075858561524666e-05,
+      "loss": 0.1312,
+      "step": 110560
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.075724724965872e-05,
+      "loss": 0.3402,
+      "step": 110570
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.075590888407077e-05,
+      "loss": 0.1255,
+      "step": 110580
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.0754570518482835e-05,
+      "loss": 0.1007,
+      "step": 110590
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.075323215289489e-05,
+      "loss": 0.057,
+      "step": 110600
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.0751893787306946e-05,
+      "loss": 0.2482,
+      "step": 110610
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.0750555421719e-05,
+      "loss": 0.2991,
+      "step": 110620
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.074921705613106e-05,
+      "loss": 0.0597,
+      "step": 110630
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.074787869054311e-05,
+      "loss": 0.1582,
+      "step": 110640
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.074654032495517e-05,
+      "loss": 0.1648,
+      "step": 110650
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.074520195936722e-05,
+      "loss": 0.0334,
+      "step": 110660
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.074386359377928e-05,
+      "loss": 0.147,
+      "step": 110670
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.074252522819133e-05,
+      "loss": 0.193,
+      "step": 110680
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.0741186862603395e-05,
+      "loss": 0.2081,
+      "step": 110690
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.073984849701545e-05,
+      "loss": 0.1248,
+      "step": 110700
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.0738510131427506e-05,
+      "loss": 0.249,
+      "step": 110710
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.073717176583956e-05,
+      "loss": 0.2422,
+      "step": 110720
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.073583340025162e-05,
+      "loss": 0.1197,
+      "step": 110730
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.073449503466367e-05,
+      "loss": 0.006,
+      "step": 110740
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.073315666907573e-05,
+      "loss": 0.161,
+      "step": 110750
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.073181830348778e-05,
+      "loss": 0.2008,
+      "step": 110760
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.073047993789984e-05,
+      "loss": 0.1473,
+      "step": 110770
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.0729141572311893e-05,
+      "loss": 0.2236,
+      "step": 110780
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.0727803206723956e-05,
+      "loss": 0.1408,
+      "step": 110790
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.072646484113601e-05,
+      "loss": 0.0661,
+      "step": 110800
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.0725126475548066e-05,
+      "loss": 0.1126,
+      "step": 110810
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.072378810996012e-05,
+      "loss": 0.2327,
+      "step": 110820
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.072244974437218e-05,
+      "loss": 0.1634,
+      "step": 110830
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.072111137878423e-05,
+      "loss": 0.1629,
+      "step": 110840
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.071977301319629e-05,
+      "loss": 0.2442,
+      "step": 110850
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.071843464760834e-05,
+      "loss": 0.2355,
+      "step": 110860
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.07170962820204e-05,
+      "loss": 0.1717,
+      "step": 110870
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.0715757916432454e-05,
+      "loss": 0.1165,
+      "step": 110880
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.0714419550844516e-05,
+      "loss": 0.1021,
+      "step": 110890
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.071308118525657e-05,
+      "loss": 0.0873,
+      "step": 110900
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.0711742819668626e-05,
+      "loss": 0.077,
+      "step": 110910
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.071040445408068e-05,
+      "loss": 0.1126,
+      "step": 110920
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.070906608849274e-05,
+      "loss": 0.0229,
+      "step": 110930
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.070772772290479e-05,
+      "loss": 0.0409,
+      "step": 110940
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.070638935731685e-05,
+      "loss": 0.1611,
+      "step": 110950
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.07050509917289e-05,
+      "loss": 0.1737,
+      "step": 110960
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.070371262614096e-05,
+      "loss": 0.0944,
+      "step": 110970
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.0702374260553014e-05,
+      "loss": 0.0823,
+      "step": 110980
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.0701035894965076e-05,
+      "loss": 0.0779,
+      "step": 110990
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.069969752937713e-05,
+      "loss": 0.0869,
+      "step": 111000
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.0698359163789187e-05,
+      "loss": 0.2832,
+      "step": 111010
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.069702079820124e-05,
+      "loss": 0.1639,
+      "step": 111020
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.06956824326133e-05,
+      "loss": 0.0836,
+      "step": 111030
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.069434406702535e-05,
+      "loss": 0.1129,
+      "step": 111040
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.069300570143741e-05,
+      "loss": 0.1244,
+      "step": 111050
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.069166733584946e-05,
+      "loss": 0.3174,
+      "step": 111060
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.069032897026152e-05,
+      "loss": 0.1391,
+      "step": 111070
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.0688990604673574e-05,
+      "loss": 0.3125,
+      "step": 111080
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.0687652239085636e-05,
+      "loss": 0.0088,
+      "step": 111090
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.068631387349769e-05,
+      "loss": 0.0934,
+      "step": 111100
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.068497550790975e-05,
+      "loss": 0.1118,
+      "step": 111110
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.06836371423218e-05,
+      "loss": 0.0922,
+      "step": 111120
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.068229877673386e-05,
+      "loss": 0.0866,
+      "step": 111130
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.068096041114591e-05,
+      "loss": 0.2624,
+      "step": 111140
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.067962204555797e-05,
+      "loss": 0.1338,
+      "step": 111150
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.0678283679970023e-05,
+      "loss": 0.3822,
+      "step": 111160
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.067694531438208e-05,
+      "loss": 0.0873,
+      "step": 111170
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.0675606948794134e-05,
+      "loss": 0.1786,
+      "step": 111180
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.0674268583206196e-05,
+      "loss": 0.06,
+      "step": 111190
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.067293021761825e-05,
+      "loss": 0.1081,
+      "step": 111200
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.067159185203031e-05,
+      "loss": 0.0406,
+      "step": 111210
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.067025348644236e-05,
+      "loss": 0.0148,
+      "step": 111220
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.066891512085442e-05,
+      "loss": 0.0028,
+      "step": 111230
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.066757675526647e-05,
+      "loss": 0.1246,
+      "step": 111240
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.066623838967853e-05,
+      "loss": 0.054,
+      "step": 111250
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.0664900024090584e-05,
+      "loss": 0.2376,
+      "step": 111260
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.066356165850264e-05,
+      "loss": 0.1307,
+      "step": 111270
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.0662223292914694e-05,
+      "loss": 0.2381,
+      "step": 111280
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.066088492732675e-05,
+      "loss": 0.0394,
+      "step": 111290
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.0659546561738805e-05,
+      "loss": 0.1766,
+      "step": 111300
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.065820819615086e-05,
+      "loss": 0.0317,
+      "step": 111310
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.0656869830562916e-05,
+      "loss": 0.1133,
+      "step": 111320
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.065553146497497e-05,
+      "loss": 0.1592,
+      "step": 111330
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.0654193099387026e-05,
+      "loss": 0.0591,
+      "step": 111340
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.065285473379908e-05,
+      "loss": 0.381,
+      "step": 111350
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.065151636821114e-05,
+      "loss": 0.0744,
+      "step": 111360
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.06501780026232e-05,
+      "loss": 0.1324,
+      "step": 111370
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.0648839637035254e-05,
+      "loss": 0.1339,
+      "step": 111380
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.064750127144731e-05,
+      "loss": 0.1796,
+      "step": 111390
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.0646162905859365e-05,
+      "loss": 0.1774,
+      "step": 111400
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.064482454027142e-05,
+      "loss": 0.1356,
+      "step": 111410
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.0643486174683476e-05,
+      "loss": 0.1608,
+      "step": 111420
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.064214780909553e-05,
+      "loss": 0.0823,
+      "step": 111430
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.0640809443507586e-05,
+      "loss": 0.11,
+      "step": 111440
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.063947107791964e-05,
+      "loss": 0.1373,
+      "step": 111450
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.06381327123317e-05,
+      "loss": 0.0309,
+      "step": 111460
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.063679434674376e-05,
+      "loss": 0.0862,
+      "step": 111470
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.0635455981155815e-05,
+      "loss": 0.2842,
+      "step": 111480
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.063411761556787e-05,
+      "loss": 0.1624,
+      "step": 111490
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.0632779249979925e-05,
+      "loss": 0.0754,
+      "step": 111500
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.063144088439198e-05,
+      "loss": 0.0203,
+      "step": 111510
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.0630102518804036e-05,
+      "loss": 0.1874,
+      "step": 111520
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.062876415321609e-05,
+      "loss": 0.1318,
+      "step": 111530
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.0627425787628147e-05,
+      "loss": 0.105,
+      "step": 111540
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.06260874220402e-05,
+      "loss": 0.1557,
+      "step": 111550
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.062474905645226e-05,
+      "loss": 0.0341,
+      "step": 111560
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.062341069086432e-05,
+      "loss": 0.0831,
+      "step": 111570
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.0622072325276375e-05,
+      "loss": 0.0171,
+      "step": 111580
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.062073395968843e-05,
+      "loss": 0.1489,
+      "step": 111590
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.0619395594100485e-05,
+      "loss": 0.056,
+      "step": 111600
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.061805722851254e-05,
+      "loss": 0.1335,
+      "step": 111610
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.0616718862924596e-05,
+      "loss": 0.2523,
+      "step": 111620
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.061538049733665e-05,
+      "loss": 0.1415,
+      "step": 111630
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.061404213174871e-05,
+      "loss": 0.1714,
+      "step": 111640
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.061270376616076e-05,
+      "loss": 0.1661,
+      "step": 111650
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.061136540057282e-05,
+      "loss": 0.0062,
+      "step": 111660
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.061002703498488e-05,
+      "loss": 0.1073,
+      "step": 111670
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.0608688669396935e-05,
+      "loss": 0.0666,
+      "step": 111680
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.060735030380899e-05,
+      "loss": 0.22,
+      "step": 111690
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.0606011938221046e-05,
+      "loss": 0.0974,
+      "step": 111700
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.06046735726331e-05,
+      "loss": 0.1233,
+      "step": 111710
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.0603335207045156e-05,
+      "loss": 0.2096,
+      "step": 111720
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.060199684145721e-05,
+      "loss": 0.0126,
+      "step": 111730
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.060065847586927e-05,
+      "loss": 0.1046,
+      "step": 111740
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.059932011028132e-05,
+      "loss": 0.1767,
+      "step": 111750
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.059798174469338e-05,
+      "loss": 0.0734,
+      "step": 111760
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.059664337910544e-05,
+      "loss": 0.1157,
+      "step": 111770
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.0595305013517495e-05,
+      "loss": 0.1339,
+      "step": 111780
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.059396664792955e-05,
+      "loss": 0.0773,
+      "step": 111790
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.0592628282341606e-05,
+      "loss": 0.0139,
+      "step": 111800
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.059128991675366e-05,
+      "loss": 0.2843,
+      "step": 111810
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.0589951551165716e-05,
+      "loss": 0.1158,
+      "step": 111820
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.058861318557777e-05,
+      "loss": 0.1155,
+      "step": 111830
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.058727481998983e-05,
+      "loss": 0.0925,
+      "step": 111840
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.058593645440188e-05,
+      "loss": 0.056,
+      "step": 111850
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.058459808881394e-05,
+      "loss": 0.1522,
+      "step": 111860
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.0583259723226e-05,
+      "loss": 0.0469,
+      "step": 111870
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.0581921357638055e-05,
+      "loss": 0.0168,
+      "step": 111880
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.058058299205011e-05,
+      "loss": 0.0327,
+      "step": 111890
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.0579244626462166e-05,
+      "loss": 0.1231,
+      "step": 111900
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.057790626087422e-05,
+      "loss": 0.1542,
+      "step": 111910
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.0576567895286277e-05,
+      "loss": 0.0361,
+      "step": 111920
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.057522952969833e-05,
+      "loss": 0.0827,
+      "step": 111930
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.057389116411039e-05,
+      "loss": 0.2433,
+      "step": 111940
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.057255279852244e-05,
+      "loss": 0.1849,
+      "step": 111950
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.05712144329345e-05,
+      "loss": 0.3221,
+      "step": 111960
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.056987606734656e-05,
+      "loss": 0.1173,
+      "step": 111970
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.0568537701758615e-05,
+      "loss": 0.0734,
+      "step": 111980
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.056719933617067e-05,
+      "loss": 0.13,
+      "step": 111990
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.0565860970582726e-05,
+      "loss": 0.3774,
+      "step": 112000
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.056452260499478e-05,
+      "loss": 0.1019,
+      "step": 112010
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.056318423940684e-05,
+      "loss": 0.1973,
+      "step": 112020
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.056184587381889e-05,
+      "loss": 0.0982,
+      "step": 112030
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.056050750823095e-05,
+      "loss": 0.0354,
+      "step": 112040
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.0559169142643e-05,
+      "loss": 0.144,
+      "step": 112050
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.0557830777055065e-05,
+      "loss": 0.2106,
+      "step": 112060
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.055649241146712e-05,
+      "loss": 0.4723,
+      "step": 112070
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.0555154045879175e-05,
+      "loss": 0.1281,
+      "step": 112080
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.055381568029123e-05,
+      "loss": 0.2718,
+      "step": 112090
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.0552477314703286e-05,
+      "loss": 0.1231,
+      "step": 112100
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.055113894911534e-05,
+      "loss": 0.1817,
+      "step": 112110
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.05498005835274e-05,
+      "loss": 0.1075,
+      "step": 112120
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.054846221793945e-05,
+      "loss": 0.1731,
+      "step": 112130
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.054712385235151e-05,
+      "loss": 0.2211,
+      "step": 112140
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.054578548676356e-05,
+      "loss": 0.2291,
+      "step": 112150
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.0544447121175625e-05,
+      "loss": 0.2176,
+      "step": 112160
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.054310875558768e-05,
+      "loss": 0.0829,
+      "step": 112170
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.0541770389999736e-05,
+      "loss": 0.3169,
+      "step": 112180
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.054043202441179e-05,
+      "loss": 0.0504,
+      "step": 112190
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.0539093658823846e-05,
+      "loss": 0.1249,
+      "step": 112200
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.05377552932359e-05,
+      "loss": 0.0935,
+      "step": 112210
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.053641692764796e-05,
+      "loss": 0.1767,
+      "step": 112220
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.053507856206001e-05,
+      "loss": 0.0774,
+      "step": 112230
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.053374019647207e-05,
+      "loss": 0.1084,
+      "step": 112240
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.053240183088412e-05,
+      "loss": 0.0459,
+      "step": 112250
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.0531063465296185e-05,
+      "loss": 0.0399,
+      "step": 112260
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.052972509970824e-05,
+      "loss": 0.055,
+      "step": 112270
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.0528386734120296e-05,
+      "loss": 0.1085,
+      "step": 112280
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.052704836853235e-05,
+      "loss": 0.0969,
+      "step": 112290
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.0525710002944406e-05,
+      "loss": 0.1477,
+      "step": 112300
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.052437163735646e-05,
+      "loss": 0.0748,
+      "step": 112310
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.052303327176852e-05,
+      "loss": 0.0629,
+      "step": 112320
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.052169490618057e-05,
+      "loss": 0.1774,
+      "step": 112330
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.052035654059263e-05,
+      "loss": 0.0741,
+      "step": 112340
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.051901817500468e-05,
+      "loss": 0.1506,
+      "step": 112350
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.0517679809416745e-05,
+      "loss": 0.1073,
+      "step": 112360
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.05163414438288e-05,
+      "loss": 0.1661,
+      "step": 112370
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.0515003078240856e-05,
+      "loss": 0.0352,
+      "step": 112380
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.051366471265291e-05,
+      "loss": 0.0646,
+      "step": 112390
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.0512326347064967e-05,
+      "loss": 0.1553,
+      "step": 112400
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.051098798147702e-05,
+      "loss": 0.0552,
+      "step": 112410
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.050964961588908e-05,
+      "loss": 0.2318,
+      "step": 112420
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.050831125030113e-05,
+      "loss": 0.0632,
+      "step": 112430
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.050697288471319e-05,
+      "loss": 0.2928,
+      "step": 112440
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.050563451912524e-05,
+      "loss": 0.1599,
+      "step": 112450
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.0504296153537305e-05,
+      "loss": 0.2654,
+      "step": 112460
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.050295778794936e-05,
+      "loss": 0.1805,
+      "step": 112470
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.0501619422361416e-05,
+      "loss": 0.0659,
+      "step": 112480
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.050028105677347e-05,
+      "loss": 0.0801,
+      "step": 112490
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.049894269118553e-05,
+      "loss": 0.1397,
+      "step": 112500
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.049760432559758e-05,
+      "loss": 0.1073,
+      "step": 112510
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.049626596000964e-05,
+      "loss": 0.0921,
+      "step": 112520
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.049492759442169e-05,
+      "loss": 0.05,
+      "step": 112530
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.049358922883375e-05,
+      "loss": 0.1368,
+      "step": 112540
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.0492250863245803e-05,
+      "loss": 0.2621,
+      "step": 112550
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.0490912497657866e-05,
+      "loss": 0.0792,
+      "step": 112560
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.048957413206992e-05,
+      "loss": 0.1704,
+      "step": 112570
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.0488235766481976e-05,
+      "loss": 0.0426,
+      "step": 112580
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.048689740089403e-05,
+      "loss": 0.2087,
+      "step": 112590
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.048555903530609e-05,
+      "loss": 0.1622,
+      "step": 112600
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.048422066971814e-05,
+      "loss": 0.1055,
+      "step": 112610
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.04828823041302e-05,
+      "loss": 0.0925,
+      "step": 112620
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.048154393854225e-05,
+      "loss": 0.0701,
+      "step": 112630
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.048020557295431e-05,
+      "loss": 0.0117,
+      "step": 112640
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.0478867207366364e-05,
+      "loss": 0.003,
+      "step": 112650
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.0477528841778426e-05,
+      "loss": 0.0747,
+      "step": 112660
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.047619047619048e-05,
+      "loss": 0.1255,
+      "step": 112670
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.0474852110602536e-05,
+      "loss": 0.0247,
+      "step": 112680
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.047351374501459e-05,
+      "loss": 0.1295,
+      "step": 112690
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.047217537942665e-05,
+      "loss": 0.1724,
+      "step": 112700
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.04708370138387e-05,
+      "loss": 0.0584,
+      "step": 112710
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.046949864825076e-05,
+      "loss": 0.0062,
+      "step": 112720
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.046816028266281e-05,
+      "loss": 0.19,
+      "step": 112730
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.046682191707487e-05,
+      "loss": 0.1473,
+      "step": 112740
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.0465483551486924e-05,
+      "loss": 0.2101,
+      "step": 112750
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.0464145185898986e-05,
+      "loss": 0.1071,
+      "step": 112760
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.046280682031104e-05,
+      "loss": 0.1024,
+      "step": 112770
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.0461468454723097e-05,
+      "loss": 0.1146,
+      "step": 112780
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.046013008913515e-05,
+      "loss": 0.0061,
+      "step": 112790
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.045879172354721e-05,
+      "loss": 0.197,
+      "step": 112800
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.045745335795926e-05,
+      "loss": 0.071,
+      "step": 112810
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.045611499237132e-05,
+      "loss": 0.0965,
+      "step": 112820
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.045477662678337e-05,
+      "loss": 0.1375,
+      "step": 112830
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.045343826119543e-05,
+      "loss": 0.0367,
+      "step": 112840
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.0452099895607484e-05,
+      "loss": 0.1735,
+      "step": 112850
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.0450761530019546e-05,
+      "loss": 0.2159,
+      "step": 112860
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.04494231644316e-05,
+      "loss": 0.1863,
+      "step": 112870
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.044808479884366e-05,
+      "loss": 0.2515,
+      "step": 112880
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.044674643325571e-05,
+      "loss": 0.0077,
+      "step": 112890
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.044540806766777e-05,
+      "loss": 0.1164,
+      "step": 112900
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.044406970207982e-05,
+      "loss": 0.1658,
+      "step": 112910
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.044273133649188e-05,
+      "loss": 0.0252,
+      "step": 112920
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.044139297090393e-05,
+      "loss": 0.1372,
+      "step": 112930
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.044005460531599e-05,
+      "loss": 0.3894,
+      "step": 112940
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.0438716239728044e-05,
+      "loss": 0.0394,
+      "step": 112950
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.0437377874140106e-05,
+      "loss": 0.0853,
+      "step": 112960
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.043603950855216e-05,
+      "loss": 0.0952,
+      "step": 112970
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.043470114296422e-05,
+      "loss": 0.161,
+      "step": 112980
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.043336277737627e-05,
+      "loss": 0.1828,
+      "step": 112990
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.043202441178833e-05,
+      "loss": 0.0879,
+      "step": 113000
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.043068604620038e-05,
+      "loss": 0.2544,
+      "step": 113010
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.042934768061244e-05,
+      "loss": 0.1769,
+      "step": 113020
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.0428009315024494e-05,
+      "loss": 0.122,
+      "step": 113030
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.042667094943655e-05,
+      "loss": 0.0263,
+      "step": 113040
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.0425332583848604e-05,
+      "loss": 0.1215,
+      "step": 113050
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.0423994218260666e-05,
+      "loss": 0.0889,
+      "step": 113060
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.042265585267272e-05,
+      "loss": 0.1131,
+      "step": 113070
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.042131748708478e-05,
+      "loss": 0.0903,
+      "step": 113080
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.041997912149683e-05,
+      "loss": 0.0905,
+      "step": 113090
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.041864075590889e-05,
+      "loss": 0.1363,
+      "step": 113100
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.041730239032094e-05,
+      "loss": 0.1525,
+      "step": 113110
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.0415964024733e-05,
+      "loss": 0.0697,
+      "step": 113120
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.0414625659145054e-05,
+      "loss": 0.0114,
+      "step": 113130
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.041328729355711e-05,
+      "loss": 0.218,
+      "step": 113140
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.0411948927969164e-05,
+      "loss": 0.0606,
+      "step": 113150
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.0410610562381226e-05,
+      "loss": 0.0244,
+      "step": 113160
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.040927219679328e-05,
+      "loss": 0.1431,
+      "step": 113170
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.040793383120534e-05,
+      "loss": 0.0958,
+      "step": 113180
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.040659546561739e-05,
+      "loss": 0.0921,
+      "step": 113190
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.040525710002945e-05,
+      "loss": 0.0573,
+      "step": 113200
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.04039187344415e-05,
+      "loss": 0.3192,
+      "step": 113210
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.040258036885356e-05,
+      "loss": 0.2887,
+      "step": 113220
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.0401242003265614e-05,
+      "loss": 0.229,
+      "step": 113230
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.039990363767767e-05,
+      "loss": 0.048,
+      "step": 113240
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.0398565272089724e-05,
+      "loss": 0.1592,
+      "step": 113250
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.039722690650179e-05,
+      "loss": 0.0971,
+      "step": 113260
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.039588854091384e-05,
+      "loss": 0.153,
+      "step": 113270
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.03945501753259e-05,
+      "loss": 0.1067,
+      "step": 113280
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.039321180973795e-05,
+      "loss": 0.0358,
+      "step": 113290
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.039187344415001e-05,
+      "loss": 0.0123,
+      "step": 113300
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.039053507856206e-05,
+      "loss": 0.0067,
+      "step": 113310
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.038919671297412e-05,
+      "loss": 0.2057,
+      "step": 113320
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.0387858347386174e-05,
+      "loss": 0.1942,
+      "step": 113330
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.038651998179823e-05,
+      "loss": 0.1019,
+      "step": 113340
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.038518161621029e-05,
+      "loss": 0.1734,
+      "step": 113350
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.038384325062235e-05,
+      "loss": 0.0488,
+      "step": 113360
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.03825048850344e-05,
+      "loss": 0.1983,
+      "step": 113370
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.038116651944646e-05,
+      "loss": 0.0071,
+      "step": 113380
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.037982815385851e-05,
+      "loss": 0.0339,
+      "step": 113390
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.037848978827057e-05,
+      "loss": 0.1188,
+      "step": 113400
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.0377151422682623e-05,
+      "loss": 0.0618,
+      "step": 113410
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.037581305709468e-05,
+      "loss": 0.0355,
+      "step": 113420
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.0374474691506734e-05,
+      "loss": 0.0559,
+      "step": 113430
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.037313632591879e-05,
+      "loss": 0.0446,
+      "step": 113440
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.037179796033085e-05,
+      "loss": 0.0897,
+      "step": 113450
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.037045959474291e-05,
+      "loss": 0.2792,
+      "step": 113460
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.036912122915496e-05,
+      "loss": 0.2282,
+      "step": 113470
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.036778286356702e-05,
+      "loss": 0.1825,
+      "step": 113480
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.036644449797907e-05,
+      "loss": 0.0233,
+      "step": 113490
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.036510613239113e-05,
+      "loss": 0.0904,
+      "step": 113500
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.0363767766803184e-05,
+      "loss": 0.1631,
+      "step": 113510
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.036242940121524e-05,
+      "loss": 0.09,
+      "step": 113520
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.0361091035627294e-05,
+      "loss": 0.0842,
+      "step": 113530
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.035975267003935e-05,
+      "loss": 0.131,
+      "step": 113540
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.035841430445141e-05,
+      "loss": 0.0464,
+      "step": 113550
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.035707593886347e-05,
+      "loss": 0.2121,
+      "step": 113560
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.035573757327552e-05,
+      "loss": 0.1908,
+      "step": 113570
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.035439920768758e-05,
+      "loss": 0.1515,
+      "step": 113580
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.035306084209963e-05,
+      "loss": 0.1214,
+      "step": 113590
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.035172247651169e-05,
+      "loss": 0.0448,
+      "step": 113600
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0350384110923744e-05,
+      "loss": 0.0034,
+      "step": 113610
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.03490457453358e-05,
+      "loss": 0.1552,
+      "step": 113620
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0347707379747854e-05,
+      "loss": 0.1134,
+      "step": 113630
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.034636901415991e-05,
+      "loss": 0.1375,
+      "step": 113640
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0345030648571965e-05,
+      "loss": 0.0303,
+      "step": 113650
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.034369228298402e-05,
+      "loss": 0.142,
+      "step": 113660
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0342353917396076e-05,
+      "loss": 0.0964,
+      "step": 113670
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.034101555180813e-05,
+      "loss": 0.0015,
+      "step": 113680
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0339677186220186e-05,
+      "loss": 0.0952,
+      "step": 113690
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.033833882063224e-05,
+      "loss": 0.0685,
+      "step": 113700
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.03370004550443e-05,
+      "loss": 0.0904,
+      "step": 113710
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.033566208945635e-05,
+      "loss": 0.2375,
+      "step": 113720
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.0334323723868415e-05,
+      "loss": 0.0134,
+      "step": 113730
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.033298535828047e-05,
+      "loss": 0.0679,
+      "step": 113740
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.0331646992692525e-05,
+      "loss": 0.1344,
+      "step": 113750
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.033030862710458e-05,
+      "loss": 0.194,
+      "step": 113760
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.0328970261516636e-05,
+      "loss": 0.0652,
+      "step": 113770
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.032763189592869e-05,
+      "loss": 0.151,
+      "step": 113780
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.032629353034075e-05,
+      "loss": 0.1078,
+      "step": 113790
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.03249551647528e-05,
+      "loss": 0.1361,
+      "step": 113800
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.032361679916486e-05,
+      "loss": 0.2318,
+      "step": 113810
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.032227843357691e-05,
+      "loss": 0.2376,
+      "step": 113820
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.0320940067988975e-05,
+      "loss": 0.1941,
+      "step": 113830
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.031960170240103e-05,
+      "loss": 0.1074,
+      "step": 113840
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.0318263336813085e-05,
+      "loss": 0.0752,
+      "step": 113850
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.031692497122514e-05,
+      "loss": 0.0717,
+      "step": 113860
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.0315586605637196e-05,
+      "loss": 0.1871,
+      "step": 113870
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.031424824004925e-05,
+      "loss": 0.1058,
+      "step": 113880
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.031290987446131e-05,
+      "loss": 0.148,
+      "step": 113890
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.031157150887336e-05,
+      "loss": 0.0234,
+      "step": 113900
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.031023314328542e-05,
+      "loss": 0.085,
+      "step": 113910
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.030889477769747e-05,
+      "loss": 0.2401,
+      "step": 113920
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.0307556412109535e-05,
+      "loss": 0.1853,
+      "step": 113930
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.030621804652159e-05,
+      "loss": 0.0307,
+      "step": 113940
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.0304879680933646e-05,
+      "loss": 0.2147,
+      "step": 113950
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.03035413153457e-05,
+      "loss": 0.1015,
+      "step": 113960
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.0302202949757756e-05,
+      "loss": 0.1217,
+      "step": 113970
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.030086458416981e-05,
+      "loss": 0.2138,
+      "step": 113980
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.029952621858187e-05,
+      "loss": 0.0199,
+      "step": 113990
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.029818785299392e-05,
+      "loss": 0.1361,
+      "step": 114000
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.029684948740598e-05,
+      "loss": 0.0897,
+      "step": 114010
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.029551112181803e-05,
+      "loss": 0.0063,
+      "step": 114020
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.0294172756230095e-05,
+      "loss": 0.1153,
+      "step": 114030
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.029283439064215e-05,
+      "loss": 0.151,
+      "step": 114040
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.0291496025054206e-05,
+      "loss": 0.0426,
+      "step": 114050
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.029015765946626e-05,
+      "loss": 0.1662,
+      "step": 114060
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.0288819293878316e-05,
+      "loss": 0.0834,
+      "step": 114070
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.028748092829037e-05,
+      "loss": 0.2232,
+      "step": 114080
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.028614256270243e-05,
+      "loss": 0.233,
+      "step": 114090
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.028480419711448e-05,
+      "loss": 0.0596,
+      "step": 114100
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.028346583152654e-05,
+      "loss": 0.0563,
+      "step": 114110
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.028212746593859e-05,
+      "loss": 0.338,
+      "step": 114120
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0280789100350655e-05,
+      "loss": 0.0855,
+      "step": 114130
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.027945073476271e-05,
+      "loss": 0.0479,
+      "step": 114140
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0278112369174766e-05,
+      "loss": 0.0909,
+      "step": 114150
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.027677400358682e-05,
+      "loss": 0.0902,
+      "step": 114160
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0275435637998877e-05,
+      "loss": 0.2764,
+      "step": 114170
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.027409727241093e-05,
+      "loss": 0.1784,
+      "step": 114180
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.027275890682299e-05,
+      "loss": 0.0932,
+      "step": 114190
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.027142054123504e-05,
+      "loss": 0.1526,
+      "step": 114200
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.02700821756471e-05,
+      "loss": 0.1677,
+      "step": 114210
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.026874381005915e-05,
+      "loss": 0.0953,
+      "step": 114220
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.0267405444471215e-05,
+      "loss": 0.1002,
+      "step": 114230
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.026606707888327e-05,
+      "loss": 0.1064,
+      "step": 114240
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.0264728713295326e-05,
+      "loss": 0.1192,
+      "step": 114250
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.026339034770738e-05,
+      "loss": 0.1123,
+      "step": 114260
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.026205198211944e-05,
+      "loss": 0.0614,
+      "step": 114270
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.026071361653149e-05,
+      "loss": 0.263,
+      "step": 114280
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.025937525094355e-05,
+      "loss": 0.0153,
+      "step": 114290
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.02580368853556e-05,
+      "loss": 0.189,
+      "step": 114300
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.025669851976766e-05,
+      "loss": 0.145,
+      "step": 114310
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0255360154179713e-05,
+      "loss": 0.078,
+      "step": 114320
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0254021788591776e-05,
+      "loss": 0.0104,
+      "step": 114330
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.025268342300383e-05,
+      "loss": 0.0235,
+      "step": 114340
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0251345057415886e-05,
+      "loss": 0.1943,
+      "step": 114350
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.025000669182794e-05,
+      "loss": 0.2513,
+      "step": 114360
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.024866832624e-05,
+      "loss": 0.2692,
+      "step": 114370
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.024732996065205e-05,
+      "loss": 0.0053,
+      "step": 114380
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.024599159506411e-05,
+      "loss": 0.0111,
+      "step": 114390
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.024465322947616e-05,
+      "loss": 0.1988,
+      "step": 114400
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.024331486388822e-05,
+      "loss": 0.232,
+      "step": 114410
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0241976498300274e-05,
+      "loss": 0.0586,
+      "step": 114420
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0240638132712336e-05,
+      "loss": 0.0877,
+      "step": 114430
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.023929976712439e-05,
+      "loss": 0.2782,
+      "step": 114440
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0237961401536446e-05,
+      "loss": 0.1084,
+      "step": 114450
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.02366230359485e-05,
+      "loss": 0.0404,
+      "step": 114460
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.023528467036056e-05,
+      "loss": 0.0473,
+      "step": 114470
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.023394630477261e-05,
+      "loss": 0.2839,
+      "step": 114480
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.023260793918467e-05,
+      "loss": 0.0553,
+      "step": 114490
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.023126957359672e-05,
+      "loss": 0.0911,
+      "step": 114500
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.022993120800878e-05,
+      "loss": 0.2122,
+      "step": 114510
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0228592842420834e-05,
+      "loss": 0.3493,
+      "step": 114520
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0227254476832896e-05,
+      "loss": 0.0722,
+      "step": 114530
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.022591611124495e-05,
+      "loss": 0.2281,
+      "step": 114540
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.0224577745657006e-05,
+      "loss": 0.1256,
+      "step": 114550
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.022323938006906e-05,
+      "loss": 0.1071,
+      "step": 114560
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.022190101448112e-05,
+      "loss": 0.1033,
+      "step": 114570
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.022056264889317e-05,
+      "loss": 0.0297,
+      "step": 114580
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.021922428330523e-05,
+      "loss": 0.1388,
+      "step": 114590
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.021788591771728e-05,
+      "loss": 0.2275,
+      "step": 114600
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.021654755212934e-05,
+      "loss": 0.2083,
+      "step": 114610
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.0215209186541394e-05,
+      "loss": 0.1684,
+      "step": 114620
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.0213870820953456e-05,
+      "loss": 0.0581,
+      "step": 114630
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.021253245536551e-05,
+      "loss": 0.3035,
+      "step": 114640
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.021119408977757e-05,
+      "loss": 0.2166,
+      "step": 114650
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.020985572418962e-05,
+      "loss": 0.0701,
+      "step": 114660
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.020851735860168e-05,
+      "loss": 0.3274,
+      "step": 114670
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.020717899301373e-05,
+      "loss": 0.2974,
+      "step": 114680
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.020584062742579e-05,
+      "loss": 0.0605,
+      "step": 114690
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.020450226183784e-05,
+      "loss": 0.0737,
+      "step": 114700
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.02031638962499e-05,
+      "loss": 0.1357,
+      "step": 114710
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.0201825530661954e-05,
+      "loss": 0.07,
+      "step": 114720
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.0200487165074016e-05,
+      "loss": 0.1077,
+      "step": 114730
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.019914879948607e-05,
+      "loss": 0.1413,
+      "step": 114740
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.019781043389813e-05,
+      "loss": 0.1391,
+      "step": 114750
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.019647206831018e-05,
+      "loss": 0.1241,
+      "step": 114760
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.019513370272224e-05,
+      "loss": 0.1391,
+      "step": 114770
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.019379533713429e-05,
+      "loss": 0.2014,
+      "step": 114780
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.019245697154635e-05,
+      "loss": 0.1792,
+      "step": 114790
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.0191118605958403e-05,
+      "loss": 0.1445,
+      "step": 114800
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.018978024037046e-05,
+      "loss": 0.1439,
+      "step": 114810
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.018844187478252e-05,
+      "loss": 0.0705,
+      "step": 114820
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.0187103509194576e-05,
+      "loss": 0.1193,
+      "step": 114830
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.018576514360663e-05,
+      "loss": 0.0526,
+      "step": 114840
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.018442677801869e-05,
+      "loss": 0.1447,
+      "step": 114850
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.018308841243074e-05,
+      "loss": 0.0606,
+      "step": 114860
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.01817500468428e-05,
+      "loss": 0.0893,
+      "step": 114870
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.018041168125485e-05,
+      "loss": 0.1369,
+      "step": 114880
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.017907331566691e-05,
+      "loss": 0.0614,
+      "step": 114890
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.0177734950078964e-05,
+      "loss": 0.2837,
+      "step": 114900
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.017639658449102e-05,
+      "loss": 0.0848,
+      "step": 114910
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.017505821890308e-05,
+      "loss": 0.0671,
+      "step": 114920
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.0173719853315136e-05,
+      "loss": 0.1317,
+      "step": 114930
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.017238148772719e-05,
+      "loss": 0.0886,
+      "step": 114940
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.017104312213925e-05,
+      "loss": 0.0383,
+      "step": 114950
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.01697047565513e-05,
+      "loss": 0.0285,
+      "step": 114960
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.016836639096336e-05,
+      "loss": 0.0691,
+      "step": 114970
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.016702802537541e-05,
+      "loss": 0.1512,
+      "step": 114980
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.016568965978747e-05,
+      "loss": 0.229,
+      "step": 114990
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.0164351294199524e-05,
+      "loss": 0.0419,
+      "step": 115000
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.016301292861158e-05,
+      "loss": 0.0248,
+      "step": 115010
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.016167456302364e-05,
+      "loss": 0.1467,
+      "step": 115020
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.0160336197435697e-05,
+      "loss": 0.0458,
+      "step": 115030
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.015899783184775e-05,
+      "loss": 0.1478,
+      "step": 115040
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.015765946625981e-05,
+      "loss": 0.2374,
+      "step": 115050
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.015632110067186e-05,
+      "loss": 0.0273,
+      "step": 115060
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.015498273508392e-05,
+      "loss": 0.3074,
+      "step": 115070
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.015364436949597e-05,
+      "loss": 0.0913,
+      "step": 115080
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.015230600390803e-05,
+      "loss": 0.1258,
+      "step": 115090
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.0150967638320084e-05,
+      "loss": 0.0956,
+      "step": 115100
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.014962927273214e-05,
+      "loss": 0.0834,
+      "step": 115110
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.01482909071442e-05,
+      "loss": 0.2212,
+      "step": 115120
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.014695254155626e-05,
+      "loss": 0.1173,
+      "step": 115130
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.014561417596831e-05,
+      "loss": 0.0043,
+      "step": 115140
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.014427581038037e-05,
+      "loss": 0.3043,
+      "step": 115150
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.014293744479242e-05,
+      "loss": 0.2057,
+      "step": 115160
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.014159907920448e-05,
+      "loss": 0.0524,
+      "step": 115170
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.0140260713616533e-05,
+      "loss": 0.1399,
+      "step": 115180
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.013892234802859e-05,
+      "loss": 0.2297,
+      "step": 115190
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.0137583982440644e-05,
+      "loss": 0.0312,
+      "step": 115200
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.01362456168527e-05,
+      "loss": 0.071,
+      "step": 115210
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.013490725126476e-05,
+      "loss": 0.1232,
+      "step": 115220
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.013356888567682e-05,
+      "loss": 0.2746,
+      "step": 115230
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.013223052008887e-05,
+      "loss": 0.12,
+      "step": 115240
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.013089215450093e-05,
+      "loss": 0.0216,
+      "step": 115250
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.012955378891298e-05,
+      "loss": 0.3299,
+      "step": 115260
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.012821542332504e-05,
+      "loss": 0.2347,
+      "step": 115270
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.0126877057737094e-05,
+      "loss": 0.1109,
+      "step": 115280
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.012553869214915e-05,
+      "loss": 0.1102,
+      "step": 115290
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.0124200326561204e-05,
+      "loss": 0.1747,
+      "step": 115300
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.012286196097326e-05,
+      "loss": 0.1781,
+      "step": 115310
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.012152359538532e-05,
+      "loss": 0.0933,
+      "step": 115320
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.012018522979738e-05,
+      "loss": 0.1213,
+      "step": 115330
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.011884686420943e-05,
+      "loss": 0.1093,
+      "step": 115340
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.011750849862149e-05,
+      "loss": 0.156,
+      "step": 115350
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.011617013303354e-05,
+      "loss": 0.2067,
+      "step": 115360
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.01148317674456e-05,
+      "loss": 0.1546,
+      "step": 115370
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.0113493401857654e-05,
+      "loss": 0.109,
+      "step": 115380
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.011215503626971e-05,
+      "loss": 0.1586,
+      "step": 115390
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.0110816670681764e-05,
+      "loss": 0.1104,
+      "step": 115400
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.010947830509382e-05,
+      "loss": 0.1186,
+      "step": 115410
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.010813993950588e-05,
+      "loss": 0.1288,
+      "step": 115420
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.010680157391794e-05,
+      "loss": 0.1439,
+      "step": 115430
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.010546320832999e-05,
+      "loss": 0.1264,
+      "step": 115440
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.010412484274205e-05,
+      "loss": 0.1654,
+      "step": 115450
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.01027864771541e-05,
+      "loss": 0.0853,
+      "step": 115460
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.010144811156616e-05,
+      "loss": 0.0469,
+      "step": 115470
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.0100109745978214e-05,
+      "loss": 0.0328,
+      "step": 115480
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.009877138039027e-05,
+      "loss": 0.2789,
+      "step": 115490
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.0097433014802325e-05,
+      "loss": 0.0726,
+      "step": 115500
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.009609464921438e-05,
+      "loss": 0.1033,
+      "step": 115510
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.009475628362644e-05,
+      "loss": 0.1263,
+      "step": 115520
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.00934179180385e-05,
+      "loss": 0.134,
+      "step": 115530
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.009207955245055e-05,
+      "loss": 0.1989,
+      "step": 115540
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.009074118686261e-05,
+      "loss": 0.0693,
+      "step": 115550
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.008940282127466e-05,
+      "loss": 0.1014,
+      "step": 115560
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.008806445568672e-05,
+      "loss": 0.1116,
+      "step": 115570
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.0086726090098774e-05,
+      "loss": 0.2376,
+      "step": 115580
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.008538772451083e-05,
+      "loss": 0.0687,
+      "step": 115590
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.0084049358922885e-05,
+      "loss": 0.1027,
+      "step": 115600
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.008271099333494e-05,
+      "loss": 0.2884,
+      "step": 115610
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.0081372627747e-05,
+      "loss": 0.032,
+      "step": 115620
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.008003426215906e-05,
+      "loss": 0.0953,
+      "step": 115630
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.007869589657111e-05,
+      "loss": 0.0915,
+      "step": 115640
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.007735753098317e-05,
+      "loss": 0.2012,
+      "step": 115650
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.0076019165395224e-05,
+      "loss": 0.1277,
+      "step": 115660
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.007468079980728e-05,
+      "loss": 0.1396,
+      "step": 115670
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.0073342434219334e-05,
+      "loss": 0.1432,
+      "step": 115680
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.007200406863139e-05,
+      "loss": 0.0185,
+      "step": 115690
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.0070665703043445e-05,
+      "loss": 0.0694,
+      "step": 115700
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.00693273374555e-05,
+      "loss": 0.2262,
+      "step": 115710
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.006798897186756e-05,
+      "loss": 0.0822,
+      "step": 115720
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.006665060627962e-05,
+      "loss": 0.1162,
+      "step": 115730
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.006531224069167e-05,
+      "loss": 0.1192,
+      "step": 115740
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.006397387510373e-05,
+      "loss": 0.1717,
+      "step": 115750
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.0062635509515784e-05,
+      "loss": 0.0666,
+      "step": 115760
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.006129714392784e-05,
+      "loss": 0.1486,
+      "step": 115770
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0059958778339894e-05,
+      "loss": 0.0978,
+      "step": 115780
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.005862041275195e-05,
+      "loss": 0.1211,
+      "step": 115790
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0057282047164005e-05,
+      "loss": 0.1224,
+      "step": 115800
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.005594368157606e-05,
+      "loss": 0.2747,
+      "step": 115810
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.005460531598812e-05,
+      "loss": 0.105,
+      "step": 115820
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.005326695040018e-05,
+      "loss": 0.2502,
+      "step": 115830
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.005192858481223e-05,
+      "loss": 0.161,
+      "step": 115840
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.005059021922429e-05,
+      "loss": 0.2746,
+      "step": 115850
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0049251853636344e-05,
+      "loss": 0.1368,
+      "step": 115860
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.00479134880484e-05,
+      "loss": 0.0666,
+      "step": 115870
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.0046575122460454e-05,
+      "loss": 0.1559,
+      "step": 115880
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.004523675687251e-05,
+      "loss": 0.1419,
+      "step": 115890
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.0043898391284565e-05,
+      "loss": 0.2297,
+      "step": 115900
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.004256002569662e-05,
+      "loss": 0.203,
+      "step": 115910
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.004122166010868e-05,
+      "loss": 0.1164,
+      "step": 115920
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.003988329452074e-05,
+      "loss": 0.193,
+      "step": 115930
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.003854492893279e-05,
+      "loss": 0.0586,
+      "step": 115940
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.003720656334485e-05,
+      "loss": 0.0614,
+      "step": 115950
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0035868197756904e-05,
+      "loss": 0.1565,
+      "step": 115960
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.003452983216896e-05,
+      "loss": 0.0141,
+      "step": 115970
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0033191466581015e-05,
+      "loss": 0.3504,
+      "step": 115980
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.003185310099307e-05,
+      "loss": 0.2229,
+      "step": 115990
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0030514735405125e-05,
+      "loss": 0.2221,
+      "step": 116000
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.002917636981718e-05,
+      "loss": 0.2062,
+      "step": 116010
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0027838004229236e-05,
+      "loss": 0.2119,
+      "step": 116020
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.002649963864129e-05,
+      "loss": 0.1094,
+      "step": 116030
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.002516127305335e-05,
+      "loss": 0.0197,
+      "step": 116040
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.00238229074654e-05,
+      "loss": 0.2103,
+      "step": 116050
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.002248454187746e-05,
+      "loss": 0.1065,
+      "step": 116060
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.002114617628951e-05,
+      "loss": 0.0631,
+      "step": 116070
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.001980781070157e-05,
+      "loss": 0.0312,
+      "step": 116080
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.001846944511362e-05,
+      "loss": 0.117,
+      "step": 116090
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.0017131079525685e-05,
+      "loss": 0.1863,
+      "step": 116100
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.001579271393774e-05,
+      "loss": 0.0293,
+      "step": 116110
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.0014454348349796e-05,
+      "loss": 0.1586,
+      "step": 116120
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.001311598276185e-05,
+      "loss": 0.1707,
+      "step": 116130
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.001177761717391e-05,
+      "loss": 0.019,
+      "step": 116140
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.001043925158596e-05,
+      "loss": 0.0821,
+      "step": 116150
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.000910088599802e-05,
+      "loss": 0.032,
+      "step": 116160
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.000776252041007e-05,
+      "loss": 0.0037,
+      "step": 116170
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.000642415482213e-05,
+      "loss": 0.2274,
+      "step": 116180
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.0005085789234184e-05,
+      "loss": 0.0098,
+      "step": 116190
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.0003747423646246e-05,
+      "loss": 0.1418,
+      "step": 116200
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.00024090580583e-05,
+      "loss": 0.1808,
+      "step": 116210
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.0001070692470356e-05,
+      "loss": 0.1181,
+      "step": 116220
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.999973232688241e-05,
+      "loss": 0.1419,
+      "step": 116230
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": {
+        "accuracy": 0.9580849141824752
+      },
+      "eval_f1": {
+        "f1": 0.9580811670883926
+      },
+      "eval_loss": 0.17507466673851013,
+      "eval_precision": {
+        "precision": 0.9586631550970829
+      },
+      "eval_recall": {
+        "recall": 0.9580178560027687
+      },
+      "eval_runtime": 428.5926,
+      "eval_samples_per_second": 38.743,
+      "eval_steps_per_second": 19.373,
+      "step": 116235
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.999839396129447e-05,
+      "loss": 0.0446,
+      "step": 116240
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.999705559570652e-05,
+      "loss": 0.0914,
+      "step": 116250
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.999571723011858e-05,
+      "loss": 0.1198,
+      "step": 116260
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.999437886453063e-05,
+      "loss": 0.2113,
+      "step": 116270
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.999304049894269e-05,
+      "loss": 0.1884,
+      "step": 116280
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.9991702133354744e-05,
+      "loss": 0.2065,
+      "step": 116290
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.9990363767766806e-05,
+      "loss": 0.035,
+      "step": 116300
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.998902540217886e-05,
+      "loss": 0.1573,
+      "step": 116310
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.9987687036590916e-05,
+      "loss": 0.0339,
+      "step": 116320
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.998634867100297e-05,
+      "loss": 0.0996,
+      "step": 116330
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.998501030541503e-05,
+      "loss": 0.1538,
+      "step": 116340
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.998367193982708e-05,
+      "loss": 0.0942,
+      "step": 116350
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.998233357423914e-05,
+      "loss": 0.1493,
+      "step": 116360
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.998099520865119e-05,
+      "loss": 0.0828,
+      "step": 116370
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.997965684306325e-05,
+      "loss": 0.0629,
+      "step": 116380
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.997831847747531e-05,
+      "loss": 0.143,
+      "step": 116390
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.9976980111887366e-05,
+      "loss": 0.0528,
+      "step": 116400
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.997564174629942e-05,
+      "loss": 0.0422,
+      "step": 116410
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.9974303380711477e-05,
+      "loss": 0.236,
+      "step": 116420
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.997296501512353e-05,
+      "loss": 0.1642,
+      "step": 116430
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.997162664953559e-05,
+      "loss": 0.0706,
+      "step": 116440
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.997028828394764e-05,
+      "loss": 0.0818,
+      "step": 116450
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.99689499183597e-05,
+      "loss": 0.1315,
+      "step": 116460
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996761155277175e-05,
+      "loss": 0.1193,
+      "step": 116470
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996627318718381e-05,
+      "loss": 0.0466,
+      "step": 116480
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996493482159587e-05,
+      "loss": 0.0549,
+      "step": 116490
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.9963596456007926e-05,
+      "loss": 0.0608,
+      "step": 116500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996225809041998e-05,
+      "loss": 0.1483,
+      "step": 116510
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996091972483204e-05,
+      "loss": 0.0677,
+      "step": 116520
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.995958135924409e-05,
+      "loss": 0.1749,
+      "step": 116530
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.995824299365615e-05,
+      "loss": 0.1059,
+      "step": 116540
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.99569046280682e-05,
+      "loss": 0.1082,
+      "step": 116550
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.995556626248026e-05,
+      "loss": 0.1139,
+      "step": 116560
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.9954227896892313e-05,
+      "loss": 0.0637,
+      "step": 116570
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.995288953130437e-05,
+      "loss": 0.0883,
+      "step": 116580
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.995155116571643e-05,
+      "loss": 0.0323,
+      "step": 116590
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.9950212800128486e-05,
+      "loss": 0.0962,
+      "step": 116600
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.994887443454054e-05,
+      "loss": 0.027,
+      "step": 116610
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.99475360689526e-05,
+      "loss": 0.0126,
+      "step": 116620
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.994619770336465e-05,
+      "loss": 0.1193,
+      "step": 116630
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.994485933777671e-05,
+      "loss": 0.1929,
+      "step": 116640
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.994352097218876e-05,
+      "loss": 0.0932,
+      "step": 116650
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.994218260660082e-05,
+      "loss": 0.2036,
+      "step": 116660
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9940844241012874e-05,
+      "loss": 0.0233,
+      "step": 116670
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.993950587542493e-05,
+      "loss": 0.0894,
+      "step": 116680
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.993816750983699e-05,
+      "loss": 0.085,
+      "step": 116690
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.9936829144249046e-05,
+      "loss": 0.0213,
+      "step": 116700
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.99354907786611e-05,
+      "loss": 0.2951,
+      "step": 116710
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.993415241307316e-05,
+      "loss": 0.1182,
+      "step": 116720
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.993281404748521e-05,
+      "loss": 0.1922,
+      "step": 116730
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.993147568189727e-05,
+      "loss": 0.091,
+      "step": 116740
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.993013731630932e-05,
+      "loss": 0.0669,
+      "step": 116750
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.992879895072138e-05,
+      "loss": 0.076,
+      "step": 116760
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.9927460585133434e-05,
+      "loss": 0.0564,
+      "step": 116770
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.992612221954549e-05,
+      "loss": 0.0051,
+      "step": 116780
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.992478385395755e-05,
+      "loss": 0.1411,
+      "step": 116790
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.9923445488369607e-05,
+      "loss": 0.186,
+      "step": 116800
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.992210712278166e-05,
+      "loss": 0.0472,
+      "step": 116810
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.992076875719372e-05,
+      "loss": 0.1606,
+      "step": 116820
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.991943039160577e-05,
+      "loss": 0.17,
+      "step": 116830
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.991809202601783e-05,
+      "loss": 0.0659,
+      "step": 116840
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.991675366042988e-05,
+      "loss": 0.1289,
+      "step": 116850
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.991541529484194e-05,
+      "loss": 0.0715,
+      "step": 116860
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.9914076929253994e-05,
+      "loss": 0.086,
+      "step": 116870
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.991273856366605e-05,
+      "loss": 0.1838,
+      "step": 116880
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.991140019807811e-05,
+      "loss": 0.1534,
+      "step": 116890
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.991006183249017e-05,
+      "loss": 0.0757,
+      "step": 116900
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.990872346690222e-05,
+      "loss": 0.0049,
+      "step": 116910
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.990738510131428e-05,
+      "loss": 0.1704,
+      "step": 116920
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.990604673572633e-05,
+      "loss": 0.0628,
+      "step": 116930
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.990470837013839e-05,
+      "loss": 0.1325,
+      "step": 116940
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.990337000455044e-05,
+      "loss": 0.1736,
+      "step": 116950
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.99020316389625e-05,
+      "loss": 0.1117,
+      "step": 116960
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.9900693273374554e-05,
+      "loss": 0.2926,
+      "step": 116970
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.989935490778661e-05,
+      "loss": 0.1297,
+      "step": 116980
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.989801654219867e-05,
+      "loss": 0.004,
+      "step": 116990
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.989667817661073e-05,
+      "loss": 0.0971,
+      "step": 117000
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.989533981102278e-05,
+      "loss": 0.1362,
+      "step": 117010
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.989400144543484e-05,
+      "loss": 0.1512,
+      "step": 117020
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.989266307984689e-05,
+      "loss": 0.0557,
+      "step": 117030
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.989132471425895e-05,
+      "loss": 0.0394,
+      "step": 117040
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.9889986348671004e-05,
+      "loss": 0.1907,
+      "step": 117050
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.988864798308306e-05,
+      "loss": 0.0622,
+      "step": 117060
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.9887309617495114e-05,
+      "loss": 0.162,
+      "step": 117070
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.988597125190717e-05,
+      "loss": 0.1622,
+      "step": 117080
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.988463288631923e-05,
+      "loss": 0.1855,
+      "step": 117090
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.988329452073129e-05,
+      "loss": 0.0751,
+      "step": 117100
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.988195615514334e-05,
+      "loss": 0.0673,
+      "step": 117110
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.98806177895554e-05,
+      "loss": 0.0848,
+      "step": 117120
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.987927942396745e-05,
+      "loss": 0.0508,
+      "step": 117130
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.987794105837951e-05,
+      "loss": 0.0699,
+      "step": 117140
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.9876602692791564e-05,
+      "loss": 0.0157,
+      "step": 117150
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.987526432720362e-05,
+      "loss": 0.0468,
+      "step": 117160
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.9873925961615674e-05,
+      "loss": 0.037,
+      "step": 117170
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.987258759602773e-05,
+      "loss": 0.1191,
+      "step": 117180
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.987124923043979e-05,
+      "loss": 0.0917,
+      "step": 117190
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.986991086485185e-05,
+      "loss": 0.0095,
+      "step": 117200
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.98685724992639e-05,
+      "loss": 0.1742,
+      "step": 117210
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.986723413367596e-05,
+      "loss": 0.0784,
+      "step": 117220
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.986589576808801e-05,
+      "loss": 0.0628,
+      "step": 117230
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.986455740250007e-05,
+      "loss": 0.0789,
+      "step": 117240
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.9863219036912124e-05,
+      "loss": 0.0781,
+      "step": 117250
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.986188067132418e-05,
+      "loss": 0.0402,
+      "step": 117260
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.9860542305736235e-05,
+      "loss": 0.2522,
+      "step": 117270
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.985920394014829e-05,
+      "loss": 0.1383,
+      "step": 117280
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.985786557456035e-05,
+      "loss": 0.4119,
+      "step": 117290
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.985652720897241e-05,
+      "loss": 0.0377,
+      "step": 117300
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.985518884338446e-05,
+      "loss": 0.0165,
+      "step": 117310
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.985385047779652e-05,
+      "loss": 0.079,
+      "step": 117320
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.985251211220857e-05,
+      "loss": 0.051,
+      "step": 117330
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.985117374662063e-05,
+      "loss": 0.3113,
+      "step": 117340
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.9849835381032684e-05,
+      "loss": 0.0574,
+      "step": 117350
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.984849701544474e-05,
+      "loss": 0.0938,
+      "step": 117360
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.9847158649856795e-05,
+      "loss": 0.0701,
+      "step": 117370
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.984582028426885e-05,
+      "loss": 0.0573,
+      "step": 117380
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.984448191868091e-05,
+      "loss": 0.0388,
+      "step": 117390
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.984314355309297e-05,
+      "loss": 0.1622,
+      "step": 117400
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.984180518750502e-05,
+      "loss": 0.1688,
+      "step": 117410
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.984046682191708e-05,
+      "loss": 0.2595,
+      "step": 117420
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.9839128456329133e-05,
+      "loss": 0.1018,
+      "step": 117430
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.983779009074119e-05,
+      "loss": 0.2303,
+      "step": 117440
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.9836451725153244e-05,
+      "loss": 0.0767,
+      "step": 117450
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.98351133595653e-05,
+      "loss": 0.0864,
+      "step": 117460
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.9833774993977355e-05,
+      "loss": 0.0491,
+      "step": 117470
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.983243662838941e-05,
+      "loss": 0.1197,
+      "step": 117480
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.983109826280147e-05,
+      "loss": 0.0543,
+      "step": 117490
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.982975989721353e-05,
+      "loss": 0.0967,
+      "step": 117500
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.982842153162558e-05,
+      "loss": 0.0812,
+      "step": 117510
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.982708316603764e-05,
+      "loss": 0.1517,
+      "step": 117520
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.9825744800449694e-05,
+      "loss": 0.1063,
+      "step": 117530
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.982440643486175e-05,
+      "loss": 0.2106,
+      "step": 117540
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.9823068069273804e-05,
+      "loss": 0.1239,
+      "step": 117550
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.982172970368586e-05,
+      "loss": 0.0758,
+      "step": 117560
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.9820391338097915e-05,
+      "loss": 0.0548,
+      "step": 117570
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.981905297250997e-05,
+      "loss": 0.0739,
+      "step": 117580
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.981771460692203e-05,
+      "loss": 0.2761,
+      "step": 117590
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.981637624133409e-05,
+      "loss": 0.0329,
+      "step": 117600
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.981503787574614e-05,
+      "loss": 0.1403,
+      "step": 117610
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.98136995101582e-05,
+      "loss": 0.1748,
+      "step": 117620
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9812361144570254e-05,
+      "loss": 0.232,
+      "step": 117630
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.981102277898231e-05,
+      "loss": 0.0202,
+      "step": 117640
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9809684413394364e-05,
+      "loss": 0.0376,
+      "step": 117650
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.980834604780642e-05,
+      "loss": 0.014,
+      "step": 117660
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9807007682218475e-05,
+      "loss": 0.0758,
+      "step": 117670
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.980566931663054e-05,
+      "loss": 0.2802,
+      "step": 117680
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.980433095104259e-05,
+      "loss": 0.0844,
+      "step": 117690
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.980299258545465e-05,
+      "loss": 0.1913,
+      "step": 117700
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.98016542198667e-05,
+      "loss": 0.1677,
+      "step": 117710
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.980031585427876e-05,
+      "loss": 0.1045,
+      "step": 117720
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9798977488690814e-05,
+      "loss": 0.2925,
+      "step": 117730
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.979763912310287e-05,
+      "loss": 0.1972,
+      "step": 117740
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9796300757514925e-05,
+      "loss": 0.2041,
+      "step": 117750
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.979496239192698e-05,
+      "loss": 0.2083,
+      "step": 117760
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9793624026339035e-05,
+      "loss": 0.005,
+      "step": 117770
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.97922856607511e-05,
+      "loss": 0.0175,
+      "step": 117780
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.979094729516315e-05,
+      "loss": 0.1214,
+      "step": 117790
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.978960892957521e-05,
+      "loss": 0.0909,
+      "step": 117800
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.9788270563987263e-05,
+      "loss": 0.0782,
+      "step": 117810
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.978693219839932e-05,
+      "loss": 0.085,
+      "step": 117820
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.9785593832811374e-05,
+      "loss": 0.1943,
+      "step": 117830
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.978425546722343e-05,
+      "loss": 0.2198,
+      "step": 117840
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.9782917101635485e-05,
+      "loss": 0.0963,
+      "step": 117850
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.978157873604754e-05,
+      "loss": 0.1503,
+      "step": 117860
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.9780240370459595e-05,
+      "loss": 0.1263,
+      "step": 117870
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.977890200487166e-05,
+      "loss": 0.1394,
+      "step": 117880
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.977756363928371e-05,
+      "loss": 0.0131,
+      "step": 117890
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.977622527369577e-05,
+      "loss": 0.0797,
+      "step": 117900
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.9774886908107824e-05,
+      "loss": 0.0135,
+      "step": 117910
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.977354854251988e-05,
+      "loss": 0.2573,
+      "step": 117920
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.9772210176931934e-05,
+      "loss": 0.0208,
+      "step": 117930
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.977087181134399e-05,
+      "loss": 0.1694,
+      "step": 117940
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9769533445756045e-05,
+      "loss": 0.0204,
+      "step": 117950
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.97681950801681e-05,
+      "loss": 0.1687,
+      "step": 117960
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9766856714580156e-05,
+      "loss": 0.0958,
+      "step": 117970
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.976551834899222e-05,
+      "loss": 0.1039,
+      "step": 117980
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.976417998340427e-05,
+      "loss": 0.0383,
+      "step": 117990
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.976284161781633e-05,
+      "loss": 0.1873,
+      "step": 118000
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9761503252228384e-05,
+      "loss": 0.1009,
+      "step": 118010
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.976016488664044e-05,
+      "loss": 0.0223,
+      "step": 118020
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9758826521052494e-05,
+      "loss": 0.0984,
+      "step": 118030
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.975748815546455e-05,
+      "loss": 0.0819,
+      "step": 118040
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9756149789876605e-05,
+      "loss": 0.2122,
+      "step": 118050
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.975481142428866e-05,
+      "loss": 0.1029,
+      "step": 118060
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9753473058700716e-05,
+      "loss": 0.1706,
+      "step": 118070
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.975213469311278e-05,
+      "loss": 0.2146,
+      "step": 118080
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.975079632752483e-05,
+      "loss": 0.0944,
+      "step": 118090
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.974945796193689e-05,
+      "loss": 0.0608,
+      "step": 118100
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9748119596348944e-05,
+      "loss": 0.1147,
+      "step": 118110
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9746781230761e-05,
+      "loss": 0.1119,
+      "step": 118120
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9745442865173055e-05,
+      "loss": 0.1693,
+      "step": 118130
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.974410449958511e-05,
+      "loss": 0.1851,
+      "step": 118140
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9742766133997165e-05,
+      "loss": 0.0621,
+      "step": 118150
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.974142776840922e-05,
+      "loss": 0.187,
+      "step": 118160
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9740089402821276e-05,
+      "loss": 0.2561,
+      "step": 118170
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.973875103723334e-05,
+      "loss": 0.3592,
+      "step": 118180
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.973741267164539e-05,
+      "loss": 0.0924,
+      "step": 118190
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.973607430605745e-05,
+      "loss": 0.1769,
+      "step": 118200
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.9734735940469504e-05,
+      "loss": 0.0838,
+      "step": 118210
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.973339757488156e-05,
+      "loss": 0.1524,
+      "step": 118220
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.9732059209293615e-05,
+      "loss": 0.2413,
+      "step": 118230
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.973072084370567e-05,
+      "loss": 0.3225,
+      "step": 118240
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.9729382478117725e-05,
+      "loss": 0.1025,
+      "step": 118250
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.972804411252978e-05,
+      "loss": 0.1174,
+      "step": 118260
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9726705746941836e-05,
+      "loss": 0.0895,
+      "step": 118270
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.97253673813539e-05,
+      "loss": 0.1755,
+      "step": 118280
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9724029015765953e-05,
+      "loss": 0.0919,
+      "step": 118290
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.972269065017801e-05,
+      "loss": 0.0917,
+      "step": 118300
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9721352284590064e-05,
+      "loss": 0.0144,
+      "step": 118310
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.972001391900212e-05,
+      "loss": 0.1542,
+      "step": 118320
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9718675553414175e-05,
+      "loss": 0.0374,
+      "step": 118330
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.971733718782623e-05,
+      "loss": 0.2287,
+      "step": 118340
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9715998822238286e-05,
+      "loss": 0.1104,
+      "step": 118350
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.971466045665034e-05,
+      "loss": 0.0887,
+      "step": 118360
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.9713322091062396e-05,
+      "loss": 0.1487,
+      "step": 118370
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.971198372547445e-05,
+      "loss": 0.0417,
+      "step": 118380
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.971064535988651e-05,
+      "loss": 0.0511,
+      "step": 118390
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.970930699429856e-05,
+      "loss": 0.1068,
+      "step": 118400
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.970796862871062e-05,
+      "loss": 0.0682,
+      "step": 118410
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.970663026312267e-05,
+      "loss": 0.2003,
+      "step": 118420
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.970529189753473e-05,
+      "loss": 0.0971,
+      "step": 118430
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.9703953531946784e-05,
+      "loss": 0.0156,
+      "step": 118440
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.970261516635884e-05,
+      "loss": 0.0541,
+      "step": 118450
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.97012768007709e-05,
+      "loss": 0.2408,
+      "step": 118460
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.9699938435182956e-05,
+      "loss": 0.1255,
+      "step": 118470
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.969860006959501e-05,
+      "loss": 0.0144,
+      "step": 118480
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.969726170400707e-05,
+      "loss": 0.116,
+      "step": 118490
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.969592333841912e-05,
+      "loss": 0.1608,
+      "step": 118500
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.969458497283118e-05,
+      "loss": 0.2404,
+      "step": 118510
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.969324660724323e-05,
+      "loss": 0.1352,
+      "step": 118520
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.969190824165529e-05,
+      "loss": 0.0048,
+      "step": 118530
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.9690569876067344e-05,
+      "loss": 0.193,
+      "step": 118540
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.96892315104794e-05,
+      "loss": 0.0551,
+      "step": 118550
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.968789314489146e-05,
+      "loss": 0.0352,
+      "step": 118560
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.9686554779303516e-05,
+      "loss": 0.0686,
+      "step": 118570
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.968521641371557e-05,
+      "loss": 0.2272,
+      "step": 118580
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.968387804812763e-05,
+      "loss": 0.1406,
+      "step": 118590
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.968253968253968e-05,
+      "loss": 0.17,
+      "step": 118600
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.968120131695174e-05,
+      "loss": 0.214,
+      "step": 118610
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.967986295136379e-05,
+      "loss": 0.026,
+      "step": 118620
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.967852458577585e-05,
+      "loss": 0.1454,
+      "step": 118630
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.9677186220187904e-05,
+      "loss": 0.0743,
+      "step": 118640
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.967584785459996e-05,
+      "loss": 0.0835,
+      "step": 118650
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.967450948901202e-05,
+      "loss": 0.0874,
+      "step": 118660
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.967317112342408e-05,
+      "loss": 0.0574,
+      "step": 118670
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.967183275783613e-05,
+      "loss": 0.1016,
+      "step": 118680
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.967049439224819e-05,
+      "loss": 0.2477,
+      "step": 118690
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.966915602666024e-05,
+      "loss": 0.2788,
+      "step": 118700
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.96678176610723e-05,
+      "loss": 0.2893,
+      "step": 118710
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.966647929548435e-05,
+      "loss": 0.0482,
+      "step": 118720
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.966514092989641e-05,
+      "loss": 0.1258,
+      "step": 118730
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.9663802564308464e-05,
+      "loss": 0.1044,
+      "step": 118740
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.966246419872052e-05,
+      "loss": 0.1541,
+      "step": 118750
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.966112583313258e-05,
+      "loss": 0.0877,
+      "step": 118760
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.965978746754464e-05,
+      "loss": 0.1321,
+      "step": 118770
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.965844910195669e-05,
+      "loss": 0.2447,
+      "step": 118780
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.965711073636875e-05,
+      "loss": 0.1532,
+      "step": 118790
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.96557723707808e-05,
+      "loss": 0.0677,
+      "step": 118800
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.965443400519286e-05,
+      "loss": 0.0128,
+      "step": 118810
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.9653095639604913e-05,
+      "loss": 0.0981,
+      "step": 118820
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.965175727401697e-05,
+      "loss": 0.0652,
+      "step": 118830
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.9650418908429024e-05,
+      "loss": 0.0383,
+      "step": 118840
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.964908054284108e-05,
+      "loss": 0.2413,
+      "step": 118850
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.964774217725314e-05,
+      "loss": 0.0748,
+      "step": 118860
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.96464038116652e-05,
+      "loss": 0.0518,
+      "step": 118870
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.964506544607725e-05,
+      "loss": 0.1855,
+      "step": 118880
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.964372708048931e-05,
+      "loss": 0.2833,
+      "step": 118890
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.964238871490136e-05,
+      "loss": 0.2403,
+      "step": 118900
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.964105034931342e-05,
+      "loss": 0.1777,
+      "step": 118910
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.9639711983725474e-05,
+      "loss": 0.1439,
+      "step": 118920
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.963837361813753e-05,
+      "loss": 0.2297,
+      "step": 118930
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.9637035252549584e-05,
+      "loss": 0.072,
+      "step": 118940
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.963569688696164e-05,
+      "loss": 0.1975,
+      "step": 118950
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.96343585213737e-05,
+      "loss": 0.1436,
+      "step": 118960
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.963302015578576e-05,
+      "loss": 0.116,
+      "step": 118970
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.963168179019781e-05,
+      "loss": 0.0342,
+      "step": 118980
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.963034342460987e-05,
+      "loss": 0.1313,
+      "step": 118990
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.962900505902192e-05,
+      "loss": 0.0352,
+      "step": 119000
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.962766669343398e-05,
+      "loss": 0.1638,
+      "step": 119010
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.9626328327846034e-05,
+      "loss": 0.1019,
+      "step": 119020
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.962498996225809e-05,
+      "loss": 0.142,
+      "step": 119030
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.9623651596670144e-05,
+      "loss": 0.1069,
+      "step": 119040
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.96223132310822e-05,
+      "loss": 0.0988,
+      "step": 119050
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.962097486549426e-05,
+      "loss": 0.1744,
+      "step": 119060
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.961963649990632e-05,
+      "loss": 0.123,
+      "step": 119070
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.961829813431837e-05,
+      "loss": 0.1571,
+      "step": 119080
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.961695976873043e-05,
+      "loss": 0.0321,
+      "step": 119090
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.961562140314248e-05,
+      "loss": 0.0175,
+      "step": 119100
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.961428303755454e-05,
+      "loss": 0.1012,
+      "step": 119110
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.9612944671966594e-05,
+      "loss": 0.2606,
+      "step": 119120
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.961160630637865e-05,
+      "loss": 0.1305,
+      "step": 119130
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.9610267940790705e-05,
+      "loss": 0.0471,
+      "step": 119140
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.960892957520277e-05,
+      "loss": 0.0814,
+      "step": 119150
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.960759120961482e-05,
+      "loss": 0.1085,
+      "step": 119160
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.960625284402688e-05,
+      "loss": 0.0106,
+      "step": 119170
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.960491447843893e-05,
+      "loss": 0.0572,
+      "step": 119180
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.960357611285099e-05,
+      "loss": 0.1128,
+      "step": 119190
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.9602237747263043e-05,
+      "loss": 0.0314,
+      "step": 119200
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.96008993816751e-05,
+      "loss": 0.2751,
+      "step": 119210
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.9599561016087154e-05,
+      "loss": 0.0787,
+      "step": 119220
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.959822265049921e-05,
+      "loss": 0.1332,
+      "step": 119230
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.9596884284911265e-05,
+      "loss": 0.2073,
+      "step": 119240
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.959554591932333e-05,
+      "loss": 0.1208,
+      "step": 119250
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.959420755373538e-05,
+      "loss": 0.0816,
+      "step": 119260
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.959286918814744e-05,
+      "loss": 0.1362,
+      "step": 119270
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.959153082255949e-05,
+      "loss": 0.033,
+      "step": 119280
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.959019245697155e-05,
+      "loss": 0.2091,
+      "step": 119290
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.9588854091383604e-05,
+      "loss": 0.0559,
+      "step": 119300
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.958751572579566e-05,
+      "loss": 0.0631,
+      "step": 119310
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.9586177360207714e-05,
+      "loss": 0.017,
+      "step": 119320
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.958483899461977e-05,
+      "loss": 0.0162,
+      "step": 119330
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.9583500629031825e-05,
+      "loss": 0.1251,
+      "step": 119340
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.958216226344389e-05,
+      "loss": 0.0434,
+      "step": 119350
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.958082389785594e-05,
+      "loss": 0.1857,
+      "step": 119360
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.9579485532268e-05,
+      "loss": 0.086,
+      "step": 119370
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.957814716668005e-05,
+      "loss": 0.181,
+      "step": 119380
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.957680880109211e-05,
+      "loss": 0.1911,
+      "step": 119390
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.9575470435504164e-05,
+      "loss": 0.0131,
+      "step": 119400
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.957413206991622e-05,
+      "loss": 0.0841,
+      "step": 119410
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.9572793704328274e-05,
+      "loss": 0.0285,
+      "step": 119420
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.957145533874033e-05,
+      "loss": 0.0262,
+      "step": 119430
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.9570116973152385e-05,
+      "loss": 0.0643,
+      "step": 119440
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.956877860756445e-05,
+      "loss": 0.1381,
+      "step": 119450
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.95674402419765e-05,
+      "loss": 0.0572,
+      "step": 119460
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.956610187638856e-05,
+      "loss": 0.2769,
+      "step": 119470
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.956476351080061e-05,
+      "loss": 0.0722,
+      "step": 119480
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.956342514521267e-05,
+      "loss": 0.0909,
+      "step": 119490
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.9562086779624724e-05,
+      "loss": 0.0202,
+      "step": 119500
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.956074841403678e-05,
+      "loss": 0.0441,
+      "step": 119510
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.9559410048448835e-05,
+      "loss": 0.0925,
+      "step": 119520
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.955807168286089e-05,
+      "loss": 0.0028,
+      "step": 119530
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.9556733317272945e-05,
+      "loss": 0.1839,
+      "step": 119540
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.955539495168501e-05,
+      "loss": 0.0116,
+      "step": 119550
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.955405658609706e-05,
+      "loss": 0.0185,
+      "step": 119560
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.955271822050912e-05,
+      "loss": 0.1187,
+      "step": 119570
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.955137985492117e-05,
+      "loss": 0.1856,
+      "step": 119580
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.955004148933323e-05,
+      "loss": 0.0642,
+      "step": 119590
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.9548703123745284e-05,
+      "loss": 0.1512,
+      "step": 119600
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.954736475815734e-05,
+      "loss": 0.1492,
+      "step": 119610
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.9546026392569395e-05,
+      "loss": 0.0613,
+      "step": 119620
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.954468802698145e-05,
+      "loss": 0.0692,
+      "step": 119630
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.9543349661393505e-05,
+      "loss": 0.2306,
+      "step": 119640
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.954201129580557e-05,
+      "loss": 0.1868,
+      "step": 119650
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.954067293021762e-05,
+      "loss": 0.0782,
+      "step": 119660
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.953933456462968e-05,
+      "loss": 0.0826,
+      "step": 119670
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.9537996199041734e-05,
+      "loss": 0.1274,
+      "step": 119680
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.953665783345379e-05,
+      "loss": 0.1122,
+      "step": 119690
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.9535319467865844e-05,
+      "loss": 0.1523,
+      "step": 119700
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.95339811022779e-05,
+      "loss": 0.1128,
+      "step": 119710
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.9532642736689955e-05,
+      "loss": 0.1678,
+      "step": 119720
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.953130437110201e-05,
+      "loss": 0.09,
+      "step": 119730
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.9529966005514066e-05,
+      "loss": 0.2271,
+      "step": 119740
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.952862763992613e-05,
+      "loss": 0.2286,
+      "step": 119750
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.952728927433818e-05,
+      "loss": 0.0797,
+      "step": 119760
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.952595090875024e-05,
+      "loss": 0.1367,
+      "step": 119770
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.9524612543162294e-05,
+      "loss": 0.2537,
+      "step": 119780
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.952327417757435e-05,
+      "loss": 0.1312,
+      "step": 119790
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.9521935811986404e-05,
+      "loss": 0.2387,
+      "step": 119800
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.952059744639846e-05,
+      "loss": 0.0351,
+      "step": 119810
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.9519259080810515e-05,
+      "loss": 0.2074,
+      "step": 119820
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.951792071522257e-05,
+      "loss": 0.2274,
+      "step": 119830
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.9516582349634626e-05,
+      "loss": 0.1445,
+      "step": 119840
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.951524398404669e-05,
+      "loss": 0.2342,
+      "step": 119850
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.951390561845874e-05,
+      "loss": 0.0371,
+      "step": 119860
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.95125672528708e-05,
+      "loss": 0.133,
+      "step": 119870
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.9511228887282854e-05,
+      "loss": 0.0918,
+      "step": 119880
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.950989052169491e-05,
+      "loss": 0.1193,
+      "step": 119890
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.9508552156106964e-05,
+      "loss": 0.1626,
+      "step": 119900
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.950721379051902e-05,
+      "loss": 0.3088,
+      "step": 119910
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.9505875424931075e-05,
+      "loss": 0.1017,
+      "step": 119920
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.950453705934313e-05,
+      "loss": 0.0116,
+      "step": 119930
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.9503198693755186e-05,
+      "loss": 0.0649,
+      "step": 119940
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.950186032816725e-05,
+      "loss": 0.0752,
+      "step": 119950
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.95005219625793e-05,
+      "loss": 0.0905,
+      "step": 119960
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.949918359699136e-05,
+      "loss": 0.0653,
+      "step": 119970
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.9497845231403414e-05,
+      "loss": 0.0201,
+      "step": 119980
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.949650686581547e-05,
+      "loss": 0.2141,
+      "step": 119990
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.9495168500227525e-05,
+      "loss": 0.0457,
+      "step": 120000
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.949383013463958e-05,
+      "loss": 0.0966,
+      "step": 120010
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.9492491769051635e-05,
+      "loss": 0.0964,
+      "step": 120020
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.949115340346369e-05,
+      "loss": 0.0768,
+      "step": 120030
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.9489815037875746e-05,
+      "loss": 0.1035,
+      "step": 120040
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.948847667228781e-05,
+      "loss": 0.0466,
+      "step": 120050
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.9487138306699863e-05,
+      "loss": 0.0548,
+      "step": 120060
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.948579994111192e-05,
+      "loss": 0.0125,
+      "step": 120070
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.9484461575523974e-05,
+      "loss": 0.0308,
+      "step": 120080
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.948312320993603e-05,
+      "loss": 0.2046,
+      "step": 120090
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.9481784844348085e-05,
+      "loss": 0.1256,
+      "step": 120100
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.948044647876014e-05,
+      "loss": 0.0783,
+      "step": 120110
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.9479108113172195e-05,
+      "loss": 0.0989,
+      "step": 120120
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.947776974758425e-05,
+      "loss": 0.0768,
+      "step": 120130
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.9476431381996306e-05,
+      "loss": 0.0214,
+      "step": 120140
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.947509301640837e-05,
+      "loss": 0.0989,
+      "step": 120150
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.9473754650820424e-05,
+      "loss": 0.109,
+      "step": 120160
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.947241628523248e-05,
+      "loss": 0.1802,
+      "step": 120170
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.9471077919644534e-05,
+      "loss": 0.1097,
+      "step": 120180
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.946973955405659e-05,
+      "loss": 0.0929,
+      "step": 120190
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.9468401188468645e-05,
+      "loss": 0.2003,
+      "step": 120200
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.94670628228807e-05,
+      "loss": 0.2824,
+      "step": 120210
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.9465724457292756e-05,
+      "loss": 0.0749,
+      "step": 120220
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.946438609170481e-05,
+      "loss": 0.2405,
+      "step": 120230
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.9463047726116866e-05,
+      "loss": 0.1677,
+      "step": 120240
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.946170936052893e-05,
+      "loss": 0.1069,
+      "step": 120250
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.9460370994940984e-05,
+      "loss": 0.0075,
+      "step": 120260
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.945903262935304e-05,
+      "loss": 0.1344,
+      "step": 120270
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.9457694263765094e-05,
+      "loss": 0.0761,
+      "step": 120280
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.945635589817715e-05,
+      "loss": 0.0728,
+      "step": 120290
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.9455017532589205e-05,
+      "loss": 0.0709,
+      "step": 120300
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.945367916700126e-05,
+      "loss": 0.1784,
+      "step": 120310
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.9452340801413316e-05,
+      "loss": 0.2397,
+      "step": 120320
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.945100243582537e-05,
+      "loss": 0.184,
+      "step": 120330
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.9449664070237426e-05,
+      "loss": 0.0757,
+      "step": 120340
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.944832570464949e-05,
+      "loss": 0.0029,
+      "step": 120350
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.9446987339061544e-05,
+      "loss": 0.0332,
+      "step": 120360
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.94456489734736e-05,
+      "loss": 0.1228,
+      "step": 120370
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.9444310607885655e-05,
+      "loss": 0.0865,
+      "step": 120380
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.944297224229771e-05,
+      "loss": 0.1214,
+      "step": 120390
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.9441633876709765e-05,
+      "loss": 0.1574,
+      "step": 120400
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.944029551112182e-05,
+      "loss": 0.0385,
+      "step": 120410
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.9438957145533876e-05,
+      "loss": 0.0701,
+      "step": 120420
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.943761877994593e-05,
+      "loss": 0.2753,
+      "step": 120430
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.943628041435799e-05,
+      "loss": 0.2366,
+      "step": 120440
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.943494204877005e-05,
+      "loss": 0.2599,
+      "step": 120450
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.9433603683182104e-05,
+      "loss": 0.0442,
+      "step": 120460
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.943226531759416e-05,
+      "loss": 0.0032,
+      "step": 120470
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.9430926952006215e-05,
+      "loss": 0.1636,
+      "step": 120480
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.942958858641827e-05,
+      "loss": 0.0585,
+      "step": 120490
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.9428250220830325e-05,
+      "loss": 0.1235,
+      "step": 120500
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.942691185524238e-05,
+      "loss": 0.0921,
+      "step": 120510
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.9425573489654436e-05,
+      "loss": 0.1978,
+      "step": 120520
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.942423512406649e-05,
+      "loss": 0.1342,
+      "step": 120530
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.9422896758478554e-05,
+      "loss": 0.0062,
+      "step": 120540
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.942155839289061e-05,
+      "loss": 0.2012,
+      "step": 120550
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.9420220027302664e-05,
+      "loss": 0.0433,
+      "step": 120560
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.941888166171472e-05,
+      "loss": 0.284,
+      "step": 120570
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.9417543296126775e-05,
+      "loss": 0.1306,
+      "step": 120580
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.941620493053883e-05,
+      "loss": 0.0345,
+      "step": 120590
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.9414866564950886e-05,
+      "loss": 0.1332,
+      "step": 120600
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.941352819936294e-05,
+      "loss": 0.1847,
+      "step": 120610
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.9412189833774996e-05,
+      "loss": 0.0251,
+      "step": 120620
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.941085146818705e-05,
+      "loss": 0.0595,
+      "step": 120630
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.9409513102599114e-05,
+      "loss": 0.2523,
+      "step": 120640
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.940817473701117e-05,
+      "loss": 0.1521,
+      "step": 120650
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.9406836371423224e-05,
+      "loss": 0.0852,
+      "step": 120660
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.940549800583528e-05,
+      "loss": 0.0674,
+      "step": 120670
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.9404159640247335e-05,
+      "loss": 0.165,
+      "step": 120680
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.940282127465939e-05,
+      "loss": 0.0587,
+      "step": 120690
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.9401482909071446e-05,
+      "loss": 0.1668,
+      "step": 120700
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.94001445434835e-05,
+      "loss": 0.0725,
+      "step": 120710
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.9398806177895556e-05,
+      "loss": 0.0424,
+      "step": 120720
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.939746781230761e-05,
+      "loss": 0.0605,
+      "step": 120730
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.939612944671967e-05,
+      "loss": 0.2026,
+      "step": 120740
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.939479108113172e-05,
+      "loss": 0.1579,
+      "step": 120750
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.939345271554378e-05,
+      "loss": 0.0611,
+      "step": 120760
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.939211434995583e-05,
+      "loss": 0.1484,
+      "step": 120770
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.939077598436789e-05,
+      "loss": 0.1316,
+      "step": 120780
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.9389437618779944e-05,
+      "loss": 0.1838,
+      "step": 120790
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.9388099253192e-05,
+      "loss": 0.1166,
+      "step": 120800
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.9386760887604054e-05,
+      "loss": 0.2454,
+      "step": 120810
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.9385422522016117e-05,
+      "loss": 0.167,
+      "step": 120820
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.938408415642817e-05,
+      "loss": 0.1246,
+      "step": 120830
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.938274579084023e-05,
+      "loss": 0.0988,
+      "step": 120840
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.938140742525228e-05,
+      "loss": 0.0709,
+      "step": 120850
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.938006905966434e-05,
+      "loss": 0.0489,
+      "step": 120860
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.937873069407639e-05,
+      "loss": 0.1198,
+      "step": 120870
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.937739232848845e-05,
+      "loss": 0.1738,
+      "step": 120880
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.9376053962900504e-05,
+      "loss": 0.15,
+      "step": 120890
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.937471559731256e-05,
+      "loss": 0.0977,
+      "step": 120900
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.9373377231724615e-05,
+      "loss": 0.2372,
+      "step": 120910
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.937203886613668e-05,
+      "loss": 0.0919,
+      "step": 120920
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.937070050054873e-05,
+      "loss": 0.0749,
+      "step": 120930
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.936936213496079e-05,
+      "loss": 0.4009,
+      "step": 120940
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.936802376937284e-05,
+      "loss": 0.0458,
+      "step": 120950
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.93666854037849e-05,
+      "loss": 0.1056,
+      "step": 120960
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.936534703819695e-05,
+      "loss": 0.0304,
+      "step": 120970
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.936400867260901e-05,
+      "loss": 0.1966,
+      "step": 120980
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.9362670307021064e-05,
+      "loss": 0.1038,
+      "step": 120990
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.936133194143312e-05,
+      "loss": 0.0973,
+      "step": 121000
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.9359993575845175e-05,
+      "loss": 0.0516,
+      "step": 121010
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.935865521025724e-05,
+      "loss": 0.0099,
+      "step": 121020
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.935731684466929e-05,
+      "loss": 0.025,
+      "step": 121030
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.935597847908135e-05,
+      "loss": 0.1377,
+      "step": 121040
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.93546401134934e-05,
+      "loss": 0.1529,
+      "step": 121050
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.935330174790546e-05,
+      "loss": 0.1811,
+      "step": 121060
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.9351963382317514e-05,
+      "loss": 0.1229,
+      "step": 121070
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.935062501672957e-05,
+      "loss": 0.1861,
+      "step": 121080
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.9349286651141624e-05,
+      "loss": 0.0842,
+      "step": 121090
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.934794828555368e-05,
+      "loss": 0.009,
+      "step": 121100
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.9346609919965735e-05,
+      "loss": 0.1324,
+      "step": 121110
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.93452715543778e-05,
+      "loss": 0.2261,
+      "step": 121120
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.934393318878985e-05,
+      "loss": 0.1429,
+      "step": 121130
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.934259482320191e-05,
+      "loss": 0.1913,
+      "step": 121140
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.934125645761396e-05,
+      "loss": 0.0996,
+      "step": 121150
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.933991809202602e-05,
+      "loss": 0.1311,
+      "step": 121160
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.9338579726438074e-05,
+      "loss": 0.0845,
+      "step": 121170
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.933724136085013e-05,
+      "loss": 0.151,
+      "step": 121180
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.9335902995262184e-05,
+      "loss": 0.0599,
+      "step": 121190
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.933456462967424e-05,
+      "loss": 0.2209,
+      "step": 121200
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.9333226264086295e-05,
+      "loss": 0.1045,
+      "step": 121210
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.933188789849836e-05,
+      "loss": 0.0912,
+      "step": 121220
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.933054953291041e-05,
+      "loss": 0.1285,
+      "step": 121230
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.932921116732247e-05,
+      "loss": 0.0406,
+      "step": 121240
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.932787280173452e-05,
+      "loss": 0.1069,
+      "step": 121250
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.932653443614658e-05,
+      "loss": 0.0614,
+      "step": 121260
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.9325196070558634e-05,
+      "loss": 0.1424,
+      "step": 121270
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.932385770497069e-05,
+      "loss": 0.1043,
+      "step": 121280
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.9322519339382745e-05,
+      "loss": 0.0355,
+      "step": 121290
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.93211809737948e-05,
+      "loss": 0.3208,
+      "step": 121300
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.9319842608206855e-05,
+      "loss": 0.2122,
+      "step": 121310
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.931850424261892e-05,
+      "loss": 0.0576,
+      "step": 121320
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.931716587703097e-05,
+      "loss": 0.164,
+      "step": 121330
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.931582751144303e-05,
+      "loss": 0.0749,
+      "step": 121340
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.931448914585508e-05,
+      "loss": 0.2018,
+      "step": 121350
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.931315078026714e-05,
+      "loss": 0.1176,
+      "step": 121360
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.9311812414679194e-05,
+      "loss": 0.0568,
+      "step": 121370
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.931047404909125e-05,
+      "loss": 0.0311,
+      "step": 121380
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.9309135683503305e-05,
+      "loss": 0.04,
+      "step": 121390
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.930779731791536e-05,
+      "loss": 0.2231,
+      "step": 121400
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.9306458952327415e-05,
+      "loss": 0.0954,
+      "step": 121410
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.930512058673948e-05,
+      "loss": 0.0356,
+      "step": 121420
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.930378222115153e-05,
+      "loss": 0.1065,
+      "step": 121430
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.930244385556359e-05,
+      "loss": 0.3039,
+      "step": 121440
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.9301105489975643e-05,
+      "loss": 0.0084,
+      "step": 121450
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.92997671243877e-05,
+      "loss": 0.3449,
+      "step": 121460
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.9298428758799754e-05,
+      "loss": 0.2391,
+      "step": 121470
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.929709039321181e-05,
+      "loss": 0.1078,
+      "step": 121480
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.9295752027623865e-05,
+      "loss": 0.0955,
+      "step": 121490
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.929441366203592e-05,
+      "loss": 0.1388,
+      "step": 121500
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.9293075296447975e-05,
+      "loss": 0.2232,
+      "step": 121510
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.929173693086004e-05,
+      "loss": 0.0977,
+      "step": 121520
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.929039856527209e-05,
+      "loss": 0.0758,
+      "step": 121530
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.928906019968415e-05,
+      "loss": 0.17,
+      "step": 121540
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.9287721834096204e-05,
+      "loss": 0.0301,
+      "step": 121550
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.928638346850826e-05,
+      "loss": 0.0371,
+      "step": 121560
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.9285045102920314e-05,
+      "loss": 0.3649,
+      "step": 121570
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.928370673733237e-05,
+      "loss": 0.1337,
+      "step": 121580
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.9282368371744425e-05,
+      "loss": 0.0327,
+      "step": 121590
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.928103000615648e-05,
+      "loss": 0.139,
+      "step": 121600
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.9279691640568536e-05,
+      "loss": 0.1427,
+      "step": 121610
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.92783532749806e-05,
+      "loss": 0.1617,
+      "step": 121620
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.927701490939265e-05,
+      "loss": 0.068,
+      "step": 121630
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.927567654380471e-05,
+      "loss": 0.2618,
+      "step": 121640
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.9274338178216764e-05,
+      "loss": 0.1462,
+      "step": 121650
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.927299981262882e-05,
+      "loss": 0.1247,
+      "step": 121660
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.9271661447040874e-05,
+      "loss": 0.1845,
+      "step": 121670
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.927032308145293e-05,
+      "loss": 0.1327,
+      "step": 121680
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.9268984715864985e-05,
+      "loss": 0.022,
+      "step": 121690
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.926764635027704e-05,
+      "loss": 0.1934,
+      "step": 121700
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.9266307984689096e-05,
+      "loss": 0.0702,
+      "step": 121710
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.926496961910116e-05,
+      "loss": 0.127,
+      "step": 121720
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.926363125351321e-05,
+      "loss": 0.0469,
+      "step": 121730
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.926229288792527e-05,
+      "loss": 0.0578,
+      "step": 121740
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.9260954522337324e-05,
+      "loss": 0.1657,
+      "step": 121750
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.925961615674938e-05,
+      "loss": 0.0468,
+      "step": 121760
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.9258277791161435e-05,
+      "loss": 0.1722,
+      "step": 121770
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.925693942557349e-05,
+      "loss": 0.1557,
+      "step": 121780
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.9255601059985545e-05,
+      "loss": 0.1238,
+      "step": 121790
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.92542626943976e-05,
+      "loss": 0.0865,
+      "step": 121800
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.9252924328809656e-05,
+      "loss": 0.0279,
+      "step": 121810
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.925158596322172e-05,
+      "loss": 0.1672,
+      "step": 121820
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.9250247597633773e-05,
+      "loss": 0.1258,
+      "step": 121830
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.924890923204583e-05,
+      "loss": 0.1142,
+      "step": 121840
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.9247570866457884e-05,
+      "loss": 0.2291,
+      "step": 121850
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.924623250086994e-05,
+      "loss": 0.0085,
+      "step": 121860
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.9244894135281995e-05,
+      "loss": 0.1417,
+      "step": 121870
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.924355576969405e-05,
+      "loss": 0.0902,
+      "step": 121880
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.9242217404106105e-05,
+      "loss": 0.033,
+      "step": 121890
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.924087903851816e-05,
+      "loss": 0.0485,
+      "step": 121900
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.923954067293022e-05,
+      "loss": 0.0691,
+      "step": 121910
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.923820230734228e-05,
+      "loss": 0.0676,
+      "step": 121920
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.9236863941754334e-05,
+      "loss": 0.0041,
+      "step": 121930
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.923552557616639e-05,
+      "loss": 0.1988,
+      "step": 121940
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.9234187210578444e-05,
+      "loss": 0.0734,
+      "step": 121950
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.92328488449905e-05,
+      "loss": 0.1772,
+      "step": 121960
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.9231510479402555e-05,
+      "loss": 0.025,
+      "step": 121970
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.923017211381461e-05,
+      "loss": 0.1372,
+      "step": 121980
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.9228833748226666e-05,
+      "loss": 0.0197,
+      "step": 121990
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.922749538263872e-05,
+      "loss": 0.2773,
+      "step": 122000
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.922615701705078e-05,
+      "loss": 0.0141,
+      "step": 122010
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.922481865146284e-05,
+      "loss": 0.0437,
+      "step": 122020
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.9223480285874894e-05,
+      "loss": 0.0424,
+      "step": 122030
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.922214192028695e-05,
+      "loss": 0.0327,
+      "step": 122040
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.9220803554699004e-05,
+      "loss": 0.1072,
+      "step": 122050
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.921946518911106e-05,
+      "loss": 0.181,
+      "step": 122060
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.9218126823523115e-05,
+      "loss": 0.0152,
+      "step": 122070
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.921678845793517e-05,
+      "loss": 0.3047,
+      "step": 122080
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.9215450092347226e-05,
+      "loss": 0.1972,
+      "step": 122090
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.921411172675928e-05,
+      "loss": 0.0565,
+      "step": 122100
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.921277336117134e-05,
+      "loss": 0.0625,
+      "step": 122110
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.92114349955834e-05,
+      "loss": 0.1997,
+      "step": 122120
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.9210096629995454e-05,
+      "loss": 0.0588,
+      "step": 122130
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.920875826440751e-05,
+      "loss": 0.1862,
+      "step": 122140
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.9207419898819565e-05,
+      "loss": 0.0791,
+      "step": 122150
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.920608153323162e-05,
+      "loss": 0.2742,
+      "step": 122160
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.9204743167643675e-05,
+      "loss": 0.0197,
+      "step": 122170
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.920340480205573e-05,
+      "loss": 0.0598,
+      "step": 122180
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.9202066436467786e-05,
+      "loss": 0.1948,
+      "step": 122190
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.920072807087984e-05,
+      "loss": 0.2258,
+      "step": 122200
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.91993897052919e-05,
+      "loss": 0.12,
+      "step": 122210
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.919805133970396e-05,
+      "loss": 0.1372,
+      "step": 122220
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.9196712974116014e-05,
+      "loss": 0.1176,
+      "step": 122230
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.919537460852807e-05,
+      "loss": 0.088,
+      "step": 122240
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.9194036242940125e-05,
+      "loss": 0.0995,
+      "step": 122250
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.919269787735218e-05,
+      "loss": 0.0697,
+      "step": 122260
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.9191359511764235e-05,
+      "loss": 0.0481,
+      "step": 122270
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.919002114617629e-05,
+      "loss": 0.0078,
+      "step": 122280
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.9188682780588346e-05,
+      "loss": 0.2098,
+      "step": 122290
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.91873444150004e-05,
+      "loss": 0.0743,
+      "step": 122300
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.9186006049412463e-05,
+      "loss": 0.1989,
+      "step": 122310
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.918466768382452e-05,
+      "loss": 0.0941,
+      "step": 122320
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.9183329318236574e-05,
+      "loss": 0.1844,
+      "step": 122330
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.918199095264863e-05,
+      "loss": 0.0552,
+      "step": 122340
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.9180652587060685e-05,
+      "loss": 0.0895,
+      "step": 122350
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.917931422147274e-05,
+      "loss": 0.0473,
+      "step": 122360
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.9177975855884796e-05,
+      "loss": 0.1947,
+      "step": 122370
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.917663749029685e-05,
+      "loss": 0.154,
+      "step": 122380
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.9175299124708906e-05,
+      "loss": 0.0182,
+      "step": 122390
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.917396075912096e-05,
+      "loss": 0.0799,
+      "step": 122400
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.9172622393533024e-05,
+      "loss": 0.0949,
+      "step": 122410
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.917128402794508e-05,
+      "loss": 0.0516,
+      "step": 122420
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.9169945662357134e-05,
+      "loss": 0.1653,
+      "step": 122430
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.916860729676919e-05,
+      "loss": 0.1378,
+      "step": 122440
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.9167268931181245e-05,
+      "loss": 0.168,
+      "step": 122450
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.91659305655933e-05,
+      "loss": 0.115,
+      "step": 122460
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.9164592200005356e-05,
+      "loss": 0.119,
+      "step": 122470
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.916325383441741e-05,
+      "loss": 0.1328,
+      "step": 122480
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.9161915468829466e-05,
+      "loss": 0.0351,
+      "step": 122490
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.916057710324152e-05,
+      "loss": 0.0036,
+      "step": 122500
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.9159238737653584e-05,
+      "loss": 0.1051,
+      "step": 122510
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.915790037206564e-05,
+      "loss": 0.0724,
+      "step": 122520
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.9156562006477694e-05,
+      "loss": 0.1224,
+      "step": 122530
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.915522364088975e-05,
+      "loss": 0.0034,
+      "step": 122540
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.9153885275301805e-05,
+      "loss": 0.099,
+      "step": 122550
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.915254690971386e-05,
+      "loss": 0.009,
+      "step": 122560
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.9151208544125916e-05,
+      "loss": 0.1201,
+      "step": 122570
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.914987017853797e-05,
+      "loss": 0.0773,
+      "step": 122580
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.9148531812950027e-05,
+      "loss": 0.1575,
+      "step": 122590
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.914719344736208e-05,
+      "loss": 0.0737,
+      "step": 122600
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.9145855081774144e-05,
+      "loss": 0.0901,
+      "step": 122610
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.91445167161862e-05,
+      "loss": 0.0292,
+      "step": 122620
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.9143178350598255e-05,
+      "loss": 0.0218,
+      "step": 122630
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.914183998501031e-05,
+      "loss": 0.138,
+      "step": 122640
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.9140501619422365e-05,
+      "loss": 0.0519,
+      "step": 122650
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.913916325383442e-05,
+      "loss": 0.2722,
+      "step": 122660
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.9137824888246476e-05,
+      "loss": 0.0418,
+      "step": 122670
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.913648652265853e-05,
+      "loss": 0.203,
+      "step": 122680
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.913514815707059e-05,
+      "loss": 0.1078,
+      "step": 122690
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.913380979148264e-05,
+      "loss": 0.014,
+      "step": 122700
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.9132471425894704e-05,
+      "loss": 0.0276,
+      "step": 122710
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.913113306030676e-05,
+      "loss": 0.225,
+      "step": 122720
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.9129794694718815e-05,
+      "loss": 0.1509,
+      "step": 122730
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.912845632913087e-05,
+      "loss": 0.1452,
+      "step": 122740
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.9127117963542925e-05,
+      "loss": 0.1268,
+      "step": 122750
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.912577959795498e-05,
+      "loss": 0.1153,
+      "step": 122760
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.9124441232367036e-05,
+      "loss": 0.1689,
+      "step": 122770
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.912310286677909e-05,
+      "loss": 0.2957,
+      "step": 122780
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.912176450119115e-05,
+      "loss": 0.1336,
+      "step": 122790
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.91204261356032e-05,
+      "loss": 0.1325,
+      "step": 122800
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.9119087770015264e-05,
+      "loss": 0.067,
+      "step": 122810
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.911774940442732e-05,
+      "loss": 0.1714,
+      "step": 122820
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.9116411038839375e-05,
+      "loss": 0.1281,
+      "step": 122830
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.911507267325143e-05,
+      "loss": 0.0354,
+      "step": 122840
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.9113734307663486e-05,
+      "loss": 0.2175,
+      "step": 122850
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.911239594207554e-05,
+      "loss": 0.2199,
+      "step": 122860
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.9111057576487596e-05,
+      "loss": 0.1073,
+      "step": 122870
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.910971921089965e-05,
+      "loss": 0.0826,
+      "step": 122880
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.910838084531171e-05,
+      "loss": 0.1056,
+      "step": 122890
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.910704247972376e-05,
+      "loss": 0.0263,
+      "step": 122900
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.9105704114135824e-05,
+      "loss": 0.2177,
+      "step": 122910
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.910436574854788e-05,
+      "loss": 0.0595,
+      "step": 122920
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.9103027382959935e-05,
+      "loss": 0.1386,
+      "step": 122930
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.910168901737199e-05,
+      "loss": 0.141,
+      "step": 122940
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.9100350651784046e-05,
+      "loss": 0.1633,
+      "step": 122950
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.90990122861961e-05,
+      "loss": 0.1169,
+      "step": 122960
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.9097673920608156e-05,
+      "loss": 0.2568,
+      "step": 122970
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.909633555502021e-05,
+      "loss": 0.0946,
+      "step": 122980
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.909499718943227e-05,
+      "loss": 0.0355,
+      "step": 122990
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.909365882384432e-05,
+      "loss": 0.2525,
+      "step": 123000
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.9092320458256385e-05,
+      "loss": 0.3362,
+      "step": 123010
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.909098209266844e-05,
+      "loss": 0.219,
+      "step": 123020
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.9089643727080495e-05,
+      "loss": 0.1888,
+      "step": 123030
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.908830536149255e-05,
+      "loss": 0.1736,
+      "step": 123040
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.9086966995904606e-05,
+      "loss": 0.2246,
+      "step": 123050
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.908562863031666e-05,
+      "loss": 0.1102,
+      "step": 123060
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.9084290264728717e-05,
+      "loss": 0.0085,
+      "step": 123070
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.908295189914077e-05,
+      "loss": 0.1066,
+      "step": 123080
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.908161353355283e-05,
+      "loss": 0.2294,
+      "step": 123090
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.908027516796488e-05,
+      "loss": 0.0314,
+      "step": 123100
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.907893680237694e-05,
+      "loss": 0.0677,
+      "step": 123110
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.907759843678899e-05,
+      "loss": 0.1914,
+      "step": 123120
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.907626007120105e-05,
+      "loss": 0.0497,
+      "step": 123130
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.9074921705613104e-05,
+      "loss": 0.1419,
+      "step": 123140
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.907358334002516e-05,
+      "loss": 0.121,
+      "step": 123150
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.9072244974437215e-05,
+      "loss": 0.031,
+      "step": 123160
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.907090660884927e-05,
+      "loss": 0.0441,
+      "step": 123170
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.9069568243261325e-05,
+      "loss": 0.101,
+      "step": 123180
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.906822987767339e-05,
+      "loss": 0.0933,
+      "step": 123190
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.906689151208544e-05,
+      "loss": 0.0728,
+      "step": 123200
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.90655531464975e-05,
+      "loss": 0.071,
+      "step": 123210
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.9064214780909553e-05,
+      "loss": 0.2445,
+      "step": 123220
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.906287641532161e-05,
+      "loss": 0.2051,
+      "step": 123230
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.9061538049733664e-05,
+      "loss": 0.1645,
+      "step": 123240
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.906019968414572e-05,
+      "loss": 0.088,
+      "step": 123250
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.9058861318557775e-05,
+      "loss": 0.0847,
+      "step": 123260
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.905752295296983e-05,
+      "loss": 0.1004,
+      "step": 123270
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.9056184587381885e-05,
+      "loss": 0.1123,
+      "step": 123280
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.905484622179395e-05,
+      "loss": 0.0025,
+      "step": 123290
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.9053507856206e-05,
+      "loss": 0.0842,
+      "step": 123300
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.905216949061806e-05,
+      "loss": 0.1512,
+      "step": 123310
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.9050831125030114e-05,
+      "loss": 0.0699,
+      "step": 123320
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.904949275944217e-05,
+      "loss": 0.0657,
+      "step": 123330
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.9048154393854224e-05,
+      "loss": 0.089,
+      "step": 123340
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.904681602826628e-05,
+      "loss": 0.1077,
+      "step": 123350
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.9045477662678335e-05,
+      "loss": 0.0923,
+      "step": 123360
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.904413929709039e-05,
+      "loss": 0.1047,
+      "step": 123370
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.904280093150245e-05,
+      "loss": 0.1583,
+      "step": 123380
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.904146256591451e-05,
+      "loss": 0.0662,
+      "step": 123390
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.904012420032656e-05,
+      "loss": 0.2609,
+      "step": 123400
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.903878583473862e-05,
+      "loss": 0.2695,
+      "step": 123410
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.9037447469150674e-05,
+      "loss": 0.0208,
+      "step": 123420
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.903610910356273e-05,
+      "loss": 0.0477,
+      "step": 123430
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.9034770737974784e-05,
+      "loss": 0.0053,
+      "step": 123440
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.903343237238684e-05,
+      "loss": 0.0869,
+      "step": 123450
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.9032094006798895e-05,
+      "loss": 0.0831,
+      "step": 123460
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.903075564121095e-05,
+      "loss": 0.061,
+      "step": 123470
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.902941727562301e-05,
+      "loss": 0.0049,
+      "step": 123480
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.902807891003507e-05,
+      "loss": 0.0776,
+      "step": 123490
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.902674054444712e-05,
+      "loss": 0.2019,
+      "step": 123500
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.902540217885918e-05,
+      "loss": 0.0461,
+      "step": 123510
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.9024063813271234e-05,
+      "loss": 0.1278,
+      "step": 123520
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.902272544768329e-05,
+      "loss": 0.0076,
+      "step": 123530
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.9021387082095345e-05,
+      "loss": 0.1286,
+      "step": 123540
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.90200487165074e-05,
+      "loss": 0.0123,
+      "step": 123550
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.9018710350919455e-05,
+      "loss": 0.0144,
+      "step": 123560
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.901737198533151e-05,
+      "loss": 0.1207,
+      "step": 123570
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.901603361974357e-05,
+      "loss": 0.2879,
+      "step": 123580
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.901469525415563e-05,
+      "loss": 0.2966,
+      "step": 123590
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.901335688856768e-05,
+      "loss": 0.2177,
+      "step": 123600
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.901201852297974e-05,
+      "loss": 0.1434,
+      "step": 123610
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.9010680157391794e-05,
+      "loss": 0.0374,
+      "step": 123620
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.900934179180385e-05,
+      "loss": 0.2469,
+      "step": 123630
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.9008003426215905e-05,
+      "loss": 0.1856,
+      "step": 123640
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.900666506062796e-05,
+      "loss": 0.0789,
+      "step": 123650
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.9005326695040015e-05,
+      "loss": 0.1145,
+      "step": 123660
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.900398832945207e-05,
+      "loss": 0.3756,
+      "step": 123670
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.900264996386413e-05,
+      "loss": 0.095,
+      "step": 123680
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.900131159827619e-05,
+      "loss": 0.152,
+      "step": 123690
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.8999973232688244e-05,
+      "loss": 0.0441,
+      "step": 123700
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.89986348671003e-05,
+      "loss": 0.1315,
+      "step": 123710
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.8997296501512354e-05,
+      "loss": 0.1144,
+      "step": 123720
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.899595813592441e-05,
+      "loss": 0.1044,
+      "step": 123730
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.8994619770336465e-05,
+      "loss": 0.0922,
+      "step": 123740
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.899328140474852e-05,
+      "loss": 0.2145,
+      "step": 123750
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.8991943039160576e-05,
+      "loss": 0.0692,
+      "step": 123760
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.899060467357263e-05,
+      "loss": 0.0892,
+      "step": 123770
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.898926630798469e-05,
+      "loss": 0.2382,
+      "step": 123780
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.898792794239675e-05,
+      "loss": 0.3371,
+      "step": 123790
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.8986589576808804e-05,
+      "loss": 0.1407,
+      "step": 123800
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.898525121122086e-05,
+      "loss": 0.1295,
+      "step": 123810
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.8983912845632914e-05,
+      "loss": 0.0517,
+      "step": 123820
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.898257448004497e-05,
+      "loss": 0.0063,
+      "step": 123830
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.8981236114457025e-05,
+      "loss": 0.1231,
+      "step": 123840
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.897989774886908e-05,
+      "loss": 0.1362,
+      "step": 123850
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.8978559383281136e-05,
+      "loss": 0.192,
+      "step": 123860
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.897722101769319e-05,
+      "loss": 0.1147,
+      "step": 123870
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.897588265210525e-05,
+      "loss": 0.2351,
+      "step": 123880
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.897454428651731e-05,
+      "loss": 0.211,
+      "step": 123890
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.8973205920929364e-05,
+      "loss": 0.1975,
+      "step": 123900
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.897186755534142e-05,
+      "loss": 0.1008,
+      "step": 123910
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.8970529189753474e-05,
+      "loss": 0.073,
+      "step": 123920
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.896919082416553e-05,
+      "loss": 0.1026,
+      "step": 123930
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.8967852458577585e-05,
+      "loss": 0.1707,
+      "step": 123940
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.896651409298964e-05,
+      "loss": 0.1012,
+      "step": 123950
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.8965175727401696e-05,
+      "loss": 0.0765,
+      "step": 123960
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.896383736181375e-05,
+      "loss": 0.0571,
+      "step": 123970
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.896249899622581e-05,
+      "loss": 0.12,
+      "step": 123980
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.896116063063787e-05,
+      "loss": 0.0761,
+      "step": 123990
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.8959822265049924e-05,
+      "loss": 0.1563,
+      "step": 124000
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.895848389946198e-05,
+      "loss": 0.0575,
+      "step": 124010
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.8957145533874035e-05,
+      "loss": 0.1345,
+      "step": 124020
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.895580716828609e-05,
+      "loss": 0.1532,
+      "step": 124030
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.8954468802698145e-05,
+      "loss": 0.2111,
+      "step": 124040
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.89531304371102e-05,
+      "loss": 0.0946,
+      "step": 124050
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.8951792071522256e-05,
+      "loss": 0.0863,
+      "step": 124060
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.895045370593431e-05,
+      "loss": 0.0501,
+      "step": 124070
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.8949115340346373e-05,
+      "loss": 0.2022,
+      "step": 124080
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.894777697475843e-05,
+      "loss": 0.097,
+      "step": 124090
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.8946438609170484e-05,
+      "loss": 0.0263,
+      "step": 124100
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.894510024358254e-05,
+      "loss": 0.1749,
+      "step": 124110
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.8943761877994595e-05,
+      "loss": 0.0047,
+      "step": 124120
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.894242351240665e-05,
+      "loss": 0.056,
+      "step": 124130
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.8941085146818705e-05,
+      "loss": 0.0706,
+      "step": 124140
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.893974678123076e-05,
+      "loss": 0.0231,
+      "step": 124150
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.8938408415642816e-05,
+      "loss": 0.0774,
+      "step": 124160
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.893707005005487e-05,
+      "loss": 0.0342,
+      "step": 124170
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.8935731684466934e-05,
+      "loss": 0.3501,
+      "step": 124180
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.893439331887899e-05,
+      "loss": 0.2001,
+      "step": 124190
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.8933054953291044e-05,
+      "loss": 0.0058,
+      "step": 124200
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.89317165877031e-05,
+      "loss": 0.0236,
+      "step": 124210
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.8930378222115155e-05,
+      "loss": 0.1592,
+      "step": 124220
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.892903985652721e-05,
+      "loss": 0.0897,
+      "step": 124230
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.8927701490939266e-05,
+      "loss": 0.1187,
+      "step": 124240
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.892636312535132e-05,
+      "loss": 0.1179,
+      "step": 124250
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.8925024759763376e-05,
+      "loss": 0.144,
+      "step": 124260
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.892368639417543e-05,
+      "loss": 0.0063,
+      "step": 124270
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.8922348028587494e-05,
+      "loss": 0.1334,
+      "step": 124280
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.892100966299955e-05,
+      "loss": 0.0705,
+      "step": 124290
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.8919671297411604e-05,
+      "loss": 0.0596,
+      "step": 124300
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.891833293182366e-05,
+      "loss": 0.135,
+      "step": 124310
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.8916994566235715e-05,
+      "loss": 0.1166,
+      "step": 124320
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.891565620064777e-05,
+      "loss": 0.1534,
+      "step": 124330
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.8914317835059826e-05,
+      "loss": 0.0921,
+      "step": 124340
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.891297946947188e-05,
+      "loss": 0.1649,
+      "step": 124350
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.8911641103883936e-05,
+      "loss": 0.0474,
+      "step": 124360
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.891030273829599e-05,
+      "loss": 0.1067,
+      "step": 124370
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.8908964372708054e-05,
+      "loss": 0.1446,
+      "step": 124380
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.890762600712011e-05,
+      "loss": 0.2143,
+      "step": 124390
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.8906287641532165e-05,
+      "loss": 0.1013,
+      "step": 124400
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.890494927594422e-05,
+      "loss": 0.1641,
+      "step": 124410
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.8903610910356275e-05,
+      "loss": 0.0091,
+      "step": 124420
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.890227254476833e-05,
+      "loss": 0.256,
+      "step": 124430
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.8900934179180386e-05,
+      "loss": 0.0866,
+      "step": 124440
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.889959581359244e-05,
+      "loss": 0.0833,
+      "step": 124450
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.88982574480045e-05,
+      "loss": 0.1186,
+      "step": 124460
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.889691908241655e-05,
+      "loss": 0.1822,
+      "step": 124470
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.8895580716828614e-05,
+      "loss": 0.1941,
+      "step": 124480
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.889424235124067e-05,
+      "loss": 0.1105,
+      "step": 124490
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.8892903985652725e-05,
+      "loss": 0.1242,
+      "step": 124500
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.889156562006478e-05,
+      "loss": 0.2182,
+      "step": 124510
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.8890227254476835e-05,
+      "loss": 0.0934,
+      "step": 124520
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.0993,
+      "step": 124530
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": {
+        "accuracy": 0.9542908762420957
+      },
+      "eval_f1": {
+        "f1": 0.9541061721417268
+      },
+      "eval_loss": 0.20993636548519135,
+      "eval_precision": {
+        "precision": 0.9541191566948424
+      },
+      "eval_recall": {
+        "recall": 0.9541611121516007
+      },
+      "eval_runtime": 430.887,
+      "eval_samples_per_second": 38.537,
+      "eval_steps_per_second": 19.27,
+      "step": 124537
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.8887550523300946e-05,
+      "loss": 0.3279,
+      "step": 124540
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.8886212157713e-05,
+      "loss": 0.0956,
+      "step": 124550
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.888487379212506e-05,
+      "loss": 0.1256,
+      "step": 124560
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.888353542653711e-05,
+      "loss": 0.0782,
+      "step": 124570
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.8882197060949174e-05,
+      "loss": 0.1396,
+      "step": 124580
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.888085869536123e-05,
+      "loss": 0.1063,
+      "step": 124590
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.8879520329773285e-05,
+      "loss": 0.1021,
+      "step": 124600
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.887818196418534e-05,
+      "loss": 0.0448,
+      "step": 124610
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.8876843598597396e-05,
+      "loss": 0.2384,
+      "step": 124620
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.887550523300945e-05,
+      "loss": 0.0504,
+      "step": 124630
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.8874166867421506e-05,
+      "loss": 0.0344,
+      "step": 124640
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.887282850183356e-05,
+      "loss": 0.1378,
+      "step": 124650
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.887149013624562e-05,
+      "loss": 0.1131,
+      "step": 124660
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.887015177065768e-05,
+      "loss": 0.064,
+      "step": 124670
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.8868813405069734e-05,
+      "loss": 0.097,
+      "step": 124680
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.886747503948179e-05,
+      "loss": 0.0458,
+      "step": 124690
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.8866136673893845e-05,
+      "loss": 0.0295,
+      "step": 124700
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.88647983083059e-05,
+      "loss": 0.2129,
+      "step": 124710
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.8863459942717956e-05,
+      "loss": 0.0896,
+      "step": 124720
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.886212157713001e-05,
+      "loss": 0.0717,
+      "step": 124730
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.8860783211542066e-05,
+      "loss": 0.0307,
+      "step": 124740
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.885944484595412e-05,
+      "loss": 0.0985,
+      "step": 124750
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.885810648036618e-05,
+      "loss": 0.219,
+      "step": 124760
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.885676811477824e-05,
+      "loss": 0.0388,
+      "step": 124770
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.8855429749190295e-05,
+      "loss": 0.092,
+      "step": 124780
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.885409138360235e-05,
+      "loss": 0.0875,
+      "step": 124790
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.8852753018014405e-05,
+      "loss": 0.1701,
+      "step": 124800
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.885141465242646e-05,
+      "loss": 0.1287,
+      "step": 124810
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.8850076286838516e-05,
+      "loss": 0.1068,
+      "step": 124820
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.884873792125057e-05,
+      "loss": 0.1658,
+      "step": 124830
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.8847399555662627e-05,
+      "loss": 0.0486,
+      "step": 124840
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.884606119007468e-05,
+      "loss": 0.2902,
+      "step": 124850
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.884472282448674e-05,
+      "loss": 0.1261,
+      "step": 124860
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.88433844588988e-05,
+      "loss": 0.1237,
+      "step": 124870
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.8842046093310855e-05,
+      "loss": 0.1681,
+      "step": 124880
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.884070772772291e-05,
+      "loss": 0.1581,
+      "step": 124890
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.8839369362134965e-05,
+      "loss": 0.1544,
+      "step": 124900
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.883803099654702e-05,
+      "loss": 0.0869,
+      "step": 124910
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.8836692630959076e-05,
+      "loss": 0.0564,
+      "step": 124920
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.883535426537113e-05,
+      "loss": 0.1004,
+      "step": 124930
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.883401589978319e-05,
+      "loss": 0.0824,
+      "step": 124940
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.883267753419524e-05,
+      "loss": 0.1455,
+      "step": 124950
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.88313391686073e-05,
+      "loss": 0.0545,
+      "step": 124960
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.883000080301936e-05,
+      "loss": 0.0991,
+      "step": 124970
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.8828662437431415e-05,
+      "loss": 0.1902,
+      "step": 124980
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.882732407184347e-05,
+      "loss": 0.0749,
+      "step": 124990
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.8825985706255526e-05,
+      "loss": 0.1195,
+      "step": 125000
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.882464734066758e-05,
+      "loss": 0.12,
+      "step": 125010
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.8823308975079636e-05,
+      "loss": 0.2415,
+      "step": 125020
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.882197060949169e-05,
+      "loss": 0.1907,
+      "step": 125030
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.882063224390375e-05,
+      "loss": 0.1457,
+      "step": 125040
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.88192938783158e-05,
+      "loss": 0.0454,
+      "step": 125050
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.881795551272786e-05,
+      "loss": 0.0817,
+      "step": 125060
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.881661714713992e-05,
+      "loss": 0.0624,
+      "step": 125070
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.8815278781551975e-05,
+      "loss": 0.13,
+      "step": 125080
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.881394041596403e-05,
+      "loss": 0.078,
+      "step": 125090
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.8812602050376086e-05,
+      "loss": 0.1066,
+      "step": 125100
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.881126368478814e-05,
+      "loss": 0.2156,
+      "step": 125110
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.8809925319200196e-05,
+      "loss": 0.2131,
+      "step": 125120
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.880858695361225e-05,
+      "loss": 0.163,
+      "step": 125130
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.880724858802431e-05,
+      "loss": 0.1287,
+      "step": 125140
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.880591022243636e-05,
+      "loss": 0.1647,
+      "step": 125150
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.880457185684842e-05,
+      "loss": 0.1076,
+      "step": 125160
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.880323349126048e-05,
+      "loss": 0.1357,
+      "step": 125170
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.8801895125672535e-05,
+      "loss": 0.1699,
+      "step": 125180
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.880055676008459e-05,
+      "loss": 0.1666,
+      "step": 125190
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.8799218394496646e-05,
+      "loss": 0.1877,
+      "step": 125200
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.87978800289087e-05,
+      "loss": 0.028,
+      "step": 125210
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.8796541663320756e-05,
+      "loss": 0.0293,
+      "step": 125220
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.879520329773281e-05,
+      "loss": 0.3008,
+      "step": 125230
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.879386493214487e-05,
+      "loss": 0.1965,
+      "step": 125240
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.879252656655692e-05,
+      "loss": 0.1669,
+      "step": 125250
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.879118820096898e-05,
+      "loss": 0.0155,
+      "step": 125260
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.878984983538104e-05,
+      "loss": 0.0825,
+      "step": 125270
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.8788511469793095e-05,
+      "loss": 0.2278,
+      "step": 125280
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.878717310420515e-05,
+      "loss": 0.1466,
+      "step": 125290
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.8785834738617206e-05,
+      "loss": 0.0756,
+      "step": 125300
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.878449637302926e-05,
+      "loss": 0.0415,
+      "step": 125310
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.878315800744132e-05,
+      "loss": 0.0263,
+      "step": 125320
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.878181964185337e-05,
+      "loss": 0.0396,
+      "step": 125330
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.878048127626543e-05,
+      "loss": 0.1504,
+      "step": 125340
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.877914291067748e-05,
+      "loss": 0.0722,
+      "step": 125350
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.877780454508954e-05,
+      "loss": 0.175,
+      "step": 125360
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.87764661795016e-05,
+      "loss": 0.0588,
+      "step": 125370
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.8775127813913655e-05,
+      "loss": 0.1721,
+      "step": 125380
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.877378944832571e-05,
+      "loss": 0.2677,
+      "step": 125390
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.8772451082737766e-05,
+      "loss": 0.0347,
+      "step": 125400
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.877111271714982e-05,
+      "loss": 0.2234,
+      "step": 125410
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.876977435156188e-05,
+      "loss": 0.1131,
+      "step": 125420
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.876843598597393e-05,
+      "loss": 0.0449,
+      "step": 125430
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.876709762038599e-05,
+      "loss": 0.1124,
+      "step": 125440
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.876575925479804e-05,
+      "loss": 0.1343,
+      "step": 125450
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.87644208892101e-05,
+      "loss": 0.1152,
+      "step": 125460
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.8763082523622153e-05,
+      "loss": 0.0515,
+      "step": 125470
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.876174415803421e-05,
+      "loss": 0.0471,
+      "step": 125480
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.8760405792446264e-05,
+      "loss": 0.0912,
+      "step": 125490
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.875906742685832e-05,
+      "loss": 0.0647,
+      "step": 125500
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.8757729061270375e-05,
+      "loss": 0.0592,
+      "step": 125510
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.875639069568243e-05,
+      "loss": 0.0864,
+      "step": 125520
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.8755052330094486e-05,
+      "loss": 0.2763,
+      "step": 125530
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.875371396450654e-05,
+      "loss": 0.0873,
+      "step": 125540
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.87523755989186e-05,
+      "loss": 0.1024,
+      "step": 125550
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.875103723333066e-05,
+      "loss": 0.0769,
+      "step": 125560
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.8749698867742714e-05,
+      "loss": 0.0942,
+      "step": 125570
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.874836050215477e-05,
+      "loss": 0.085,
+      "step": 125580
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.8747022136566824e-05,
+      "loss": 0.1959,
+      "step": 125590
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.874568377097888e-05,
+      "loss": 0.2483,
+      "step": 125600
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.8744345405390935e-05,
+      "loss": 0.0677,
+      "step": 125610
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.874300703980299e-05,
+      "loss": 0.1669,
+      "step": 125620
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.8741668674215046e-05,
+      "loss": 0.0572,
+      "step": 125630
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.87403303086271e-05,
+      "loss": 0.0627,
+      "step": 125640
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.873899194303916e-05,
+      "loss": 0.0986,
+      "step": 125650
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.873765357745122e-05,
+      "loss": 0.2362,
+      "step": 125660
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.8736315211863274e-05,
+      "loss": 0.0588,
+      "step": 125670
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.873497684627533e-05,
+      "loss": 0.2303,
+      "step": 125680
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.8733638480687384e-05,
+      "loss": 0.0812,
+      "step": 125690
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.873230011509944e-05,
+      "loss": 0.0636,
+      "step": 125700
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.8730961749511495e-05,
+      "loss": 0.1973,
+      "step": 125710
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.872962338392355e-05,
+      "loss": 0.1291,
+      "step": 125720
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.8728285018335606e-05,
+      "loss": 0.1904,
+      "step": 125730
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.872694665274766e-05,
+      "loss": 0.1424,
+      "step": 125740
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.872560828715972e-05,
+      "loss": 0.121,
+      "step": 125750
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.872426992157178e-05,
+      "loss": 0.0147,
+      "step": 125760
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.8722931555983834e-05,
+      "loss": 0.1329,
+      "step": 125770
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.872159319039589e-05,
+      "loss": 0.1386,
+      "step": 125780
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.8720254824807945e-05,
+      "loss": 0.0374,
+      "step": 125790
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.871891645922e-05,
+      "loss": 0.1943,
+      "step": 125800
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.8717578093632055e-05,
+      "loss": 0.0171,
+      "step": 125810
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.871623972804411e-05,
+      "loss": 0.1587,
+      "step": 125820
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.8714901362456166e-05,
+      "loss": 0.1201,
+      "step": 125830
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.871356299686822e-05,
+      "loss": 0.2053,
+      "step": 125840
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.8712224631280283e-05,
+      "loss": 0.0865,
+      "step": 125850
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.871088626569234e-05,
+      "loss": 0.3059,
+      "step": 125860
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.8709547900104394e-05,
+      "loss": 0.1774,
+      "step": 125870
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.870820953451645e-05,
+      "loss": 0.0119,
+      "step": 125880
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.8706871168928505e-05,
+      "loss": 0.1234,
+      "step": 125890
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.870553280334056e-05,
+      "loss": 0.1293,
+      "step": 125900
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.8704194437752615e-05,
+      "loss": 0.1707,
+      "step": 125910
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.870285607216467e-05,
+      "loss": 0.0414,
+      "step": 125920
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.8701517706576726e-05,
+      "loss": 0.0858,
+      "step": 125930
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.870017934098878e-05,
+      "loss": 0.1385,
+      "step": 125940
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.8698840975400844e-05,
+      "loss": 0.0207,
+      "step": 125950
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.86975026098129e-05,
+      "loss": 0.1285,
+      "step": 125960
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.8696164244224954e-05,
+      "loss": 0.2474,
+      "step": 125970
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.869482587863701e-05,
+      "loss": 0.1637,
+      "step": 125980
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.8693487513049065e-05,
+      "loss": 0.225,
+      "step": 125990
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.869214914746112e-05,
+      "loss": 0.1459,
+      "step": 126000
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.8690810781873176e-05,
+      "loss": 0.0819,
+      "step": 126010
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.868947241628523e-05,
+      "loss": 0.0952,
+      "step": 126020
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.8688134050697286e-05,
+      "loss": 0.0138,
+      "step": 126030
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.868679568510934e-05,
+      "loss": 0.1988,
+      "step": 126040
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.8685457319521404e-05,
+      "loss": 0.236,
+      "step": 126050
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.868411895393346e-05,
+      "loss": 0.0711,
+      "step": 126060
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.8682780588345514e-05,
+      "loss": 0.007,
+      "step": 126070
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.868144222275757e-05,
+      "loss": 0.0111,
+      "step": 126080
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.8680103857169625e-05,
+      "loss": 0.1303,
+      "step": 126090
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.867876549158168e-05,
+      "loss": 0.0593,
+      "step": 126100
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.8677427125993736e-05,
+      "loss": 0.2445,
+      "step": 126110
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.867608876040579e-05,
+      "loss": 0.0345,
+      "step": 126120
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.8674750394817846e-05,
+      "loss": 0.0788,
+      "step": 126130
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.86734120292299e-05,
+      "loss": 0.1164,
+      "step": 126140
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.8672073663641964e-05,
+      "loss": 0.0793,
+      "step": 126150
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.867073529805402e-05,
+      "loss": 0.182,
+      "step": 126160
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.8669396932466075e-05,
+      "loss": 0.0332,
+      "step": 126170
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.866805856687813e-05,
+      "loss": 0.2199,
+      "step": 126180
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.8666720201290185e-05,
+      "loss": 0.1044,
+      "step": 126190
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.866538183570224e-05,
+      "loss": 0.0546,
+      "step": 126200
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.8664043470114296e-05,
+      "loss": 0.1071,
+      "step": 126210
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.866270510452635e-05,
+      "loss": 0.1223,
+      "step": 126220
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.8661366738938407e-05,
+      "loss": 0.4066,
+      "step": 126230
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.866002837335047e-05,
+      "loss": 0.1669,
+      "step": 126240
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.8658690007762524e-05,
+      "loss": 0.1334,
+      "step": 126250
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.865735164217458e-05,
+      "loss": 0.1083,
+      "step": 126260
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.8656013276586635e-05,
+      "loss": 0.1496,
+      "step": 126270
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.865467491099869e-05,
+      "loss": 0.0987,
+      "step": 126280
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.8653336545410745e-05,
+      "loss": 0.096,
+      "step": 126290
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.86519981798228e-05,
+      "loss": 0.0406,
+      "step": 126300
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.8650659814234856e-05,
+      "loss": 0.1643,
+      "step": 126310
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.864932144864691e-05,
+      "loss": 0.0626,
+      "step": 126320
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.864798308305897e-05,
+      "loss": 0.1363,
+      "step": 126330
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.864664471747103e-05,
+      "loss": 0.1812,
+      "step": 126340
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.8645306351883084e-05,
+      "loss": 0.1763,
+      "step": 126350
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.864396798629514e-05,
+      "loss": 0.0854,
+      "step": 126360
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.8642629620707195e-05,
+      "loss": 0.1152,
+      "step": 126370
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.864129125511925e-05,
+      "loss": 0.1069,
+      "step": 126380
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.8639952889531306e-05,
+      "loss": 0.1524,
+      "step": 126390
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.863861452394336e-05,
+      "loss": 0.1303,
+      "step": 126400
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.8637276158355416e-05,
+      "loss": 0.0744,
+      "step": 126410
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.863593779276747e-05,
+      "loss": 0.171,
+      "step": 126420
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.863459942717953e-05,
+      "loss": 0.0939,
+      "step": 126430
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.863326106159159e-05,
+      "loss": 0.0747,
+      "step": 126440
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.8631922696003644e-05,
+      "loss": 0.0439,
+      "step": 126450
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.86305843304157e-05,
+      "loss": 0.1781,
+      "step": 126460
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.8629245964827755e-05,
+      "loss": 0.0624,
+      "step": 126470
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.862790759923981e-05,
+      "loss": 0.1532,
+      "step": 126480
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.8626569233651866e-05,
+      "loss": 0.0686,
+      "step": 126490
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.862523086806392e-05,
+      "loss": 0.0411,
+      "step": 126500
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.8623892502475976e-05,
+      "loss": 0.1697,
+      "step": 126510
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.862255413688803e-05,
+      "loss": 0.1237,
+      "step": 126520
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.862121577130009e-05,
+      "loss": 0.0626,
+      "step": 126530
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.861987740571215e-05,
+      "loss": 0.0058,
+      "step": 126540
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.8618539040124204e-05,
+      "loss": 0.1188,
+      "step": 126550
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.861720067453626e-05,
+      "loss": 0.1034,
+      "step": 126560
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.8615862308948315e-05,
+      "loss": 0.1015,
+      "step": 126570
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.861452394336037e-05,
+      "loss": 0.1413,
+      "step": 126580
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.8613185577772426e-05,
+      "loss": 0.1307,
+      "step": 126590
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.861184721218448e-05,
+      "loss": 0.1244,
+      "step": 126600
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.8610508846596537e-05,
+      "loss": 0.2915,
+      "step": 126610
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.860917048100859e-05,
+      "loss": 0.1253,
+      "step": 126620
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.860783211542065e-05,
+      "loss": 0.2639,
+      "step": 126630
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.860649374983271e-05,
+      "loss": 0.0168,
+      "step": 126640
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.8605155384244765e-05,
+      "loss": 0.0601,
+      "step": 126650
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.860381701865682e-05,
+      "loss": 0.0974,
+      "step": 126660
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.8602478653068875e-05,
+      "loss": 0.0515,
+      "step": 126670
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.860114028748093e-05,
+      "loss": 0.0806,
+      "step": 126680
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.8599801921892986e-05,
+      "loss": 0.1599,
+      "step": 126690
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.859846355630504e-05,
+      "loss": 0.1153,
+      "step": 126700
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.85971251907171e-05,
+      "loss": 0.1547,
+      "step": 126710
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.859578682512915e-05,
+      "loss": 0.074,
+      "step": 126720
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.859444845954121e-05,
+      "loss": 0.044,
+      "step": 126730
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.859311009395327e-05,
+      "loss": 0.0793,
+      "step": 126740
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.8591771728365325e-05,
+      "loss": 0.0841,
+      "step": 126750
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.859043336277738e-05,
+      "loss": 0.1768,
+      "step": 126760
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.8589094997189435e-05,
+      "loss": 0.1254,
+      "step": 126770
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.858775663160149e-05,
+      "loss": 0.1216,
+      "step": 126780
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.8586418266013546e-05,
+      "loss": 0.2635,
+      "step": 126790
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.85850799004256e-05,
+      "loss": 0.0941,
+      "step": 126800
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.858374153483766e-05,
+      "loss": 0.0293,
+      "step": 126810
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.858240316924971e-05,
+      "loss": 0.0485,
+      "step": 126820
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.858106480366177e-05,
+      "loss": 0.1863,
+      "step": 126830
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.857972643807383e-05,
+      "loss": 0.1025,
+      "step": 126840
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.8578388072485885e-05,
+      "loss": 0.1128,
+      "step": 126850
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.857704970689794e-05,
+      "loss": 0.0795,
+      "step": 126860
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.8575711341309996e-05,
+      "loss": 0.0035,
+      "step": 126870
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.857437297572205e-05,
+      "loss": 0.2014,
+      "step": 126880
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.8573034610134106e-05,
+      "loss": 0.1102,
+      "step": 126890
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.857169624454616e-05,
+      "loss": 0.0105,
+      "step": 126900
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.857035787895822e-05,
+      "loss": 0.1359,
+      "step": 126910
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.856901951337027e-05,
+      "loss": 0.28,
+      "step": 126920
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.856768114778233e-05,
+      "loss": 0.2259,
+      "step": 126930
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.856634278219439e-05,
+      "loss": 0.0063,
+      "step": 126940
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.8565004416606445e-05,
+      "loss": 0.0424,
+      "step": 126950
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.85636660510185e-05,
+      "loss": 0.0141,
+      "step": 126960
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.8562327685430556e-05,
+      "loss": 0.0394,
+      "step": 126970
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.856098931984261e-05,
+      "loss": 0.0647,
+      "step": 126980
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.8559650954254666e-05,
+      "loss": 0.0753,
+      "step": 126990
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.855831258866672e-05,
+      "loss": 0.0085,
+      "step": 127000
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.855697422307878e-05,
+      "loss": 0.1424,
+      "step": 127010
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.855563585749083e-05,
+      "loss": 0.0045,
+      "step": 127020
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.855429749190289e-05,
+      "loss": 0.1349,
+      "step": 127030
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.855295912631495e-05,
+      "loss": 0.0777,
+      "step": 127040
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.8551620760727005e-05,
+      "loss": 0.0783,
+      "step": 127050
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.855028239513906e-05,
+      "loss": 0.1707,
+      "step": 127060
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.8548944029551116e-05,
+      "loss": 0.2003,
+      "step": 127070
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.854760566396317e-05,
+      "loss": 0.0189,
+      "step": 127080
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.8546267298375227e-05,
+      "loss": 0.0661,
+      "step": 127090
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.854492893278728e-05,
+      "loss": 0.1017,
+      "step": 127100
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.854359056719934e-05,
+      "loss": 0.1262,
+      "step": 127110
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.854225220161139e-05,
+      "loss": 0.197,
+      "step": 127120
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.854091383602345e-05,
+      "loss": 0.1995,
+      "step": 127130
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.853957547043551e-05,
+      "loss": 0.126,
+      "step": 127140
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.8538237104847565e-05,
+      "loss": 0.1146,
+      "step": 127150
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.853689873925962e-05,
+      "loss": 0.0443,
+      "step": 127160
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.8535560373671676e-05,
+      "loss": 0.0807,
+      "step": 127170
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.853422200808373e-05,
+      "loss": 0.1087,
+      "step": 127180
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.853288364249579e-05,
+      "loss": 0.1188,
+      "step": 127190
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.853154527690784e-05,
+      "loss": 0.0592,
+      "step": 127200
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.85302069113199e-05,
+      "loss": 0.09,
+      "step": 127210
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.852886854573195e-05,
+      "loss": 0.0097,
+      "step": 127220
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.852753018014401e-05,
+      "loss": 0.1355,
+      "step": 127230
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.852619181455607e-05,
+      "loss": 0.1556,
+      "step": 127240
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.8524853448968126e-05,
+      "loss": 0.0575,
+      "step": 127250
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.852351508338018e-05,
+      "loss": 0.0785,
+      "step": 127260
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.8522176717792236e-05,
+      "loss": 0.1103,
+      "step": 127270
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.852083835220429e-05,
+      "loss": 0.0423,
+      "step": 127280
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.851949998661635e-05,
+      "loss": 0.0576,
+      "step": 127290
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.85181616210284e-05,
+      "loss": 0.097,
+      "step": 127300
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.851682325544046e-05,
+      "loss": 0.189,
+      "step": 127310
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.851548488985251e-05,
+      "loss": 0.1445,
+      "step": 127320
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.851414652426457e-05,
+      "loss": 0.0892,
+      "step": 127330
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.851280815867663e-05,
+      "loss": 0.1979,
+      "step": 127340
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.8511469793088686e-05,
+      "loss": 0.1005,
+      "step": 127350
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.851013142750074e-05,
+      "loss": 0.0286,
+      "step": 127360
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.8508793061912796e-05,
+      "loss": 0.0896,
+      "step": 127370
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.850745469632485e-05,
+      "loss": 0.1535,
+      "step": 127380
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.850611633073691e-05,
+      "loss": 0.1861,
+      "step": 127390
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.850477796514896e-05,
+      "loss": 0.1204,
+      "step": 127400
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.850343959956102e-05,
+      "loss": 0.1818,
+      "step": 127410
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.850210123397307e-05,
+      "loss": 0.0519,
+      "step": 127420
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.8500762868385135e-05,
+      "loss": 0.0078,
+      "step": 127430
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.849942450279719e-05,
+      "loss": 0.0032,
+      "step": 127440
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.8498086137209246e-05,
+      "loss": 0.0653,
+      "step": 127450
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.84967477716213e-05,
+      "loss": 0.0035,
+      "step": 127460
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.8495409406033357e-05,
+      "loss": 0.1147,
+      "step": 127470
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.849407104044541e-05,
+      "loss": 0.1329,
+      "step": 127480
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.849273267485747e-05,
+      "loss": 0.1302,
+      "step": 127490
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.849139430926952e-05,
+      "loss": 0.1735,
+      "step": 127500
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.849005594368158e-05,
+      "loss": 0.2085,
+      "step": 127510
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.848871757809363e-05,
+      "loss": 0.2044,
+      "step": 127520
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.8487379212505695e-05,
+      "loss": 0.0736,
+      "step": 127530
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.848604084691775e-05,
+      "loss": 0.0049,
+      "step": 127540
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.8484702481329806e-05,
+      "loss": 0.0053,
+      "step": 127550
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.848336411574186e-05,
+      "loss": 0.1775,
+      "step": 127560
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.848202575015392e-05,
+      "loss": 0.1,
+      "step": 127570
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.848068738456597e-05,
+      "loss": 0.2291,
+      "step": 127580
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.847934901897803e-05,
+      "loss": 0.2905,
+      "step": 127590
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.847801065339008e-05,
+      "loss": 0.1137,
+      "step": 127600
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.847667228780214e-05,
+      "loss": 0.0617,
+      "step": 127610
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.847533392221419e-05,
+      "loss": 0.0222,
+      "step": 127620
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.8473995556626255e-05,
+      "loss": 0.1471,
+      "step": 127630
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.847265719103831e-05,
+      "loss": 0.1089,
+      "step": 127640
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.8471318825450366e-05,
+      "loss": 0.1071,
+      "step": 127650
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.846998045986242e-05,
+      "loss": 0.19,
+      "step": 127660
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.846864209427448e-05,
+      "loss": 0.1232,
+      "step": 127670
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.846730372868653e-05,
+      "loss": 0.1076,
+      "step": 127680
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.846596536309859e-05,
+      "loss": 0.1456,
+      "step": 127690
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.846462699751064e-05,
+      "loss": 0.0022,
+      "step": 127700
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.84632886319227e-05,
+      "loss": 0.0793,
+      "step": 127710
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.8461950266334754e-05,
+      "loss": 0.0436,
+      "step": 127720
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.8460611900746816e-05,
+      "loss": 0.2055,
+      "step": 127730
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.845927353515887e-05,
+      "loss": 0.1228,
+      "step": 127740
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.8457935169570926e-05,
+      "loss": 0.06,
+      "step": 127750
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.845659680398298e-05,
+      "loss": 0.0459,
+      "step": 127760
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.845525843839504e-05,
+      "loss": 0.1539,
+      "step": 127770
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.845392007280709e-05,
+      "loss": 0.0382,
+      "step": 127780
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.845258170721915e-05,
+      "loss": 0.0646,
+      "step": 127790
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.84512433416312e-05,
+      "loss": 0.0342,
+      "step": 127800
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.844990497604326e-05,
+      "loss": 0.003,
+      "step": 127810
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.8448566610455314e-05,
+      "loss": 0.0581,
+      "step": 127820
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.844722824486737e-05,
+      "loss": 0.0538,
+      "step": 127830
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.8445889879279424e-05,
+      "loss": 0.1017,
+      "step": 127840
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.844455151369148e-05,
+      "loss": 0.0127,
+      "step": 127850
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.8443213148103535e-05,
+      "loss": 0.0536,
+      "step": 127860
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.844187478251559e-05,
+      "loss": 0.1515,
+      "step": 127870
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.8440536416927646e-05,
+      "loss": 0.0489,
+      "step": 127880
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.84391980513397e-05,
+      "loss": 0.1027,
+      "step": 127890
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.8437859685751756e-05,
+      "loss": 0.1066,
+      "step": 127900
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.843652132016382e-05,
+      "loss": 0.1375,
+      "step": 127910
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.8435182954575874e-05,
+      "loss": 0.1417,
+      "step": 127920
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.843384458898793e-05,
+      "loss": 0.0341,
+      "step": 127930
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.8432506223399985e-05,
+      "loss": 0.0461,
+      "step": 127940
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.843116785781204e-05,
+      "loss": 0.1037,
+      "step": 127950
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.8429829492224095e-05,
+      "loss": 0.1221,
+      "step": 127960
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.842849112663615e-05,
+      "loss": 0.0649,
+      "step": 127970
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.8427152761048206e-05,
+      "loss": 0.1021,
+      "step": 127980
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.842581439546026e-05,
+      "loss": 0.0374,
+      "step": 127990
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.8424476029872317e-05,
+      "loss": 0.1209,
+      "step": 128000
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.842313766428438e-05,
+      "loss": 0.2283,
+      "step": 128010
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.8421799298696434e-05,
+      "loss": 0.1435,
+      "step": 128020
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.842046093310849e-05,
+      "loss": 0.1726,
+      "step": 128030
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.8419122567520545e-05,
+      "loss": 0.0276,
+      "step": 128040
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.84177842019326e-05,
+      "loss": 0.151,
+      "step": 128050
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.8416445836344655e-05,
+      "loss": 0.0622,
+      "step": 128060
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.841510747075671e-05,
+      "loss": 0.0217,
+      "step": 128070
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.8413769105168766e-05,
+      "loss": 0.1248,
+      "step": 128080
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.841243073958082e-05,
+      "loss": 0.016,
+      "step": 128090
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.841109237399288e-05,
+      "loss": 0.2686,
+      "step": 128100
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.840975400840494e-05,
+      "loss": 0.1165,
+      "step": 128110
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.8408415642816994e-05,
+      "loss": 0.0149,
+      "step": 128120
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.840707727722905e-05,
+      "loss": 0.1693,
+      "step": 128130
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.8405738911641105e-05,
+      "loss": 0.2084,
+      "step": 128140
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.840440054605316e-05,
+      "loss": 0.0447,
+      "step": 128150
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.8403062180465215e-05,
+      "loss": 0.0914,
+      "step": 128160
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.840172381487727e-05,
+      "loss": 0.3813,
+      "step": 128170
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.8400385449289326e-05,
+      "loss": 0.082,
+      "step": 128180
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.839904708370138e-05,
+      "loss": 0.0452,
+      "step": 128190
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.839770871811344e-05,
+      "loss": 0.2109,
+      "step": 128200
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.83963703525255e-05,
+      "loss": 0.1667,
+      "step": 128210
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.8395031986937554e-05,
+      "loss": 0.0652,
+      "step": 128220
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.839369362134961e-05,
+      "loss": 0.1174,
+      "step": 128230
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.8392355255761665e-05,
+      "loss": 0.1906,
+      "step": 128240
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.839101689017372e-05,
+      "loss": 0.2795,
+      "step": 128250
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.8389678524585776e-05,
+      "loss": 0.2305,
+      "step": 128260
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.838834015899783e-05,
+      "loss": 0.1376,
+      "step": 128270
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.8387001793409886e-05,
+      "loss": 0.2835,
+      "step": 128280
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.838566342782194e-05,
+      "loss": 0.1511,
+      "step": 128290
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.8384325062234e-05,
+      "loss": 0.175,
+      "step": 128300
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.838298669664606e-05,
+      "loss": 0.092,
+      "step": 128310
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.8381648331058114e-05,
+      "loss": 0.1726,
+      "step": 128320
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.838030996547017e-05,
+      "loss": 0.078,
+      "step": 128330
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.8378971599882225e-05,
+      "loss": 0.0797,
+      "step": 128340
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.837763323429428e-05,
+      "loss": 0.1563,
+      "step": 128350
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.8376294868706336e-05,
+      "loss": 0.2176,
+      "step": 128360
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.837495650311839e-05,
+      "loss": 0.0808,
+      "step": 128370
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.8373618137530446e-05,
+      "loss": 0.0331,
+      "step": 128380
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.83722797719425e-05,
+      "loss": 0.1257,
+      "step": 128390
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.837094140635456e-05,
+      "loss": 0.1998,
+      "step": 128400
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.836960304076662e-05,
+      "loss": 0.0822,
+      "step": 128410
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.8368264675178675e-05,
+      "loss": 0.0192,
+      "step": 128420
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.836692630959073e-05,
+      "loss": 0.0606,
+      "step": 128430
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.8365587944002785e-05,
+      "loss": 0.0413,
+      "step": 128440
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.836424957841484e-05,
+      "loss": 0.0781,
+      "step": 128450
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.8362911212826896e-05,
+      "loss": 0.105,
+      "step": 128460
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.836157284723895e-05,
+      "loss": 0.0834,
+      "step": 128470
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.836023448165101e-05,
+      "loss": 0.0628,
+      "step": 128480
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.835889611606306e-05,
+      "loss": 0.231,
+      "step": 128490
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.835755775047512e-05,
+      "loss": 0.2066,
+      "step": 128500
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.835621938488718e-05,
+      "loss": 0.1036,
+      "step": 128510
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.8354881019299235e-05,
+      "loss": 0.2243,
+      "step": 128520
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.835354265371129e-05,
+      "loss": 0.017,
+      "step": 128530
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.8352204288123345e-05,
+      "loss": 0.11,
+      "step": 128540
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.83508659225354e-05,
+      "loss": 0.0703,
+      "step": 128550
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.8349527556947456e-05,
+      "loss": 0.0187,
+      "step": 128560
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.834818919135951e-05,
+      "loss": 0.0087,
+      "step": 128570
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.834685082577157e-05,
+      "loss": 0.1711,
+      "step": 128580
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.834551246018362e-05,
+      "loss": 0.0749,
+      "step": 128590
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.834417409459568e-05,
+      "loss": 0.0252,
+      "step": 128600
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.834283572900774e-05,
+      "loss": 0.2043,
+      "step": 128610
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.8341497363419795e-05,
+      "loss": 0.2747,
+      "step": 128620
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.834015899783185e-05,
+      "loss": 0.0501,
+      "step": 128630
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.8338820632243906e-05,
+      "loss": 0.0936,
+      "step": 128640
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.833748226665596e-05,
+      "loss": 0.0609,
+      "step": 128650
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.8336143901068016e-05,
+      "loss": 0.1705,
+      "step": 128660
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.833480553548007e-05,
+      "loss": 0.1015,
+      "step": 128670
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.833346716989213e-05,
+      "loss": 0.0148,
+      "step": 128680
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.833212880430418e-05,
+      "loss": 0.1261,
+      "step": 128690
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.833079043871624e-05,
+      "loss": 0.3082,
+      "step": 128700
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.83294520731283e-05,
+      "loss": 0.1041,
+      "step": 128710
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.8328113707540355e-05,
+      "loss": 0.1722,
+      "step": 128720
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.832677534195241e-05,
+      "loss": 0.2083,
+      "step": 128730
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.8325436976364466e-05,
+      "loss": 0.0117,
+      "step": 128740
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.832409861077652e-05,
+      "loss": 0.0337,
+      "step": 128750
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.8322760245188576e-05,
+      "loss": 0.1682,
+      "step": 128760
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.832142187960063e-05,
+      "loss": 0.0944,
+      "step": 128770
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.832008351401269e-05,
+      "loss": 0.1224,
+      "step": 128780
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.831874514842474e-05,
+      "loss": 0.1073,
+      "step": 128790
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.83174067828368e-05,
+      "loss": 0.0891,
+      "step": 128800
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.831606841724886e-05,
+      "loss": 0.0601,
+      "step": 128810
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.8314730051660915e-05,
+      "loss": 0.4048,
+      "step": 128820
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.831339168607297e-05,
+      "loss": 0.0313,
+      "step": 128830
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.8312053320485026e-05,
+      "loss": 0.1425,
+      "step": 128840
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.831071495489708e-05,
+      "loss": 0.1076,
+      "step": 128850
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.8309376589309137e-05,
+      "loss": 0.2147,
+      "step": 128860
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.830803822372119e-05,
+      "loss": 0.1249,
+      "step": 128870
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.830669985813325e-05,
+      "loss": 0.1496,
+      "step": 128880
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.83053614925453e-05,
+      "loss": 0.0452,
+      "step": 128890
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.830402312695736e-05,
+      "loss": 0.0068,
+      "step": 128900
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.830268476136942e-05,
+      "loss": 0.2792,
+      "step": 128910
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.8301346395781475e-05,
+      "loss": 0.2318,
+      "step": 128920
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.830000803019353e-05,
+      "loss": 0.0095,
+      "step": 128930
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.8298669664605586e-05,
+      "loss": 0.0459,
+      "step": 128940
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.829733129901764e-05,
+      "loss": 0.089,
+      "step": 128950
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.82959929334297e-05,
+      "loss": 0.0862,
+      "step": 128960
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.829465456784175e-05,
+      "loss": 0.0777,
+      "step": 128970
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.829331620225381e-05,
+      "loss": 0.123,
+      "step": 128980
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.829197783666586e-05,
+      "loss": 0.0103,
+      "step": 128990
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.8290639471077925e-05,
+      "loss": 0.1622,
+      "step": 129000
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.828930110548998e-05,
+      "loss": 0.0968,
+      "step": 129010
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.8287962739902036e-05,
+      "loss": 0.0102,
+      "step": 129020
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.828662437431409e-05,
+      "loss": 0.0711,
+      "step": 129030
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.8285286008726146e-05,
+      "loss": 0.1805,
+      "step": 129040
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.82839476431382e-05,
+      "loss": 0.0934,
+      "step": 129050
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.828260927755026e-05,
+      "loss": 0.1507,
+      "step": 129060
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.828127091196231e-05,
+      "loss": 0.1026,
+      "step": 129070
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.827993254637437e-05,
+      "loss": 0.0229,
+      "step": 129080
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.827859418078642e-05,
+      "loss": 0.1925,
+      "step": 129090
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.8277255815198485e-05,
+      "loss": 0.04,
+      "step": 129100
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.827591744961054e-05,
+      "loss": 0.0856,
+      "step": 129110
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.8274579084022596e-05,
+      "loss": 0.0918,
+      "step": 129120
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.827324071843465e-05,
+      "loss": 0.1084,
+      "step": 129130
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.8271902352846706e-05,
+      "loss": 0.0113,
+      "step": 129140
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.827056398725876e-05,
+      "loss": 0.0932,
+      "step": 129150
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.826922562167082e-05,
+      "loss": 0.1327,
+      "step": 129160
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.826788725608287e-05,
+      "loss": 0.1408,
+      "step": 129170
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.826654889049493e-05,
+      "loss": 0.0659,
+      "step": 129180
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.826521052490698e-05,
+      "loss": 0.119,
+      "step": 129190
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.8263872159319045e-05,
+      "loss": 0.2249,
+      "step": 129200
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.82625337937311e-05,
+      "loss": 0.1617,
+      "step": 129210
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.8261195428143156e-05,
+      "loss": 0.1964,
+      "step": 129220
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.825985706255521e-05,
+      "loss": 0.1375,
+      "step": 129230
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.8258518696967266e-05,
+      "loss": 0.1526,
+      "step": 129240
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.825718033137932e-05,
+      "loss": 0.1615,
+      "step": 129250
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.825584196579138e-05,
+      "loss": 0.0627,
+      "step": 129260
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.825450360020343e-05,
+      "loss": 0.1139,
+      "step": 129270
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.825316523461549e-05,
+      "loss": 0.0993,
+      "step": 129280
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.825182686902754e-05,
+      "loss": 0.1578,
+      "step": 129290
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.8250488503439605e-05,
+      "loss": 0.1394,
+      "step": 129300
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.824915013785166e-05,
+      "loss": 0.0512,
+      "step": 129310
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.8247811772263716e-05,
+      "loss": 0.0803,
+      "step": 129320
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.824647340667577e-05,
+      "loss": 0.1115,
+      "step": 129330
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.824513504108783e-05,
+      "loss": 0.0765,
+      "step": 129340
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.824379667549988e-05,
+      "loss": 0.0324,
+      "step": 129350
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.824245830991194e-05,
+      "loss": 0.0785,
+      "step": 129360
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.824111994432399e-05,
+      "loss": 0.0055,
+      "step": 129370
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.823978157873605e-05,
+      "loss": 0.2286,
+      "step": 129380
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.82384432131481e-05,
+      "loss": 0.1882,
+      "step": 129390
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.8237104847560165e-05,
+      "loss": 0.1239,
+      "step": 129400
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.823576648197222e-05,
+      "loss": 0.0254,
+      "step": 129410
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.8234428116384276e-05,
+      "loss": 0.017,
+      "step": 129420
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.823308975079633e-05,
+      "loss": 0.0956,
+      "step": 129430
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.823175138520839e-05,
+      "loss": 0.2115,
+      "step": 129440
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.823041301962044e-05,
+      "loss": 0.029,
+      "step": 129450
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.82290746540325e-05,
+      "loss": 0.1033,
+      "step": 129460
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.822773628844455e-05,
+      "loss": 0.1321,
+      "step": 129470
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.822639792285661e-05,
+      "loss": 0.059,
+      "step": 129480
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.8225059557268663e-05,
+      "loss": 0.1453,
+      "step": 129490
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.8223721191680726e-05,
+      "loss": 0.1122,
+      "step": 129500
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.822238282609278e-05,
+      "loss": 0.177,
+      "step": 129510
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.8221044460504836e-05,
+      "loss": 0.0196,
+      "step": 129520
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.821970609491689e-05,
+      "loss": 0.102,
+      "step": 129530
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.821836772932895e-05,
+      "loss": 0.0046,
+      "step": 129540
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.8217029363741e-05,
+      "loss": 0.0807,
+      "step": 129550
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.821569099815306e-05,
+      "loss": 0.1478,
+      "step": 129560
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.821435263256511e-05,
+      "loss": 0.0836,
+      "step": 129570
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.821301426697717e-05,
+      "loss": 0.2027,
+      "step": 129580
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.8211675901389224e-05,
+      "loss": 0.2084,
+      "step": 129590
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.8210337535801286e-05,
+      "loss": 0.1491,
+      "step": 129600
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.820899917021334e-05,
+      "loss": 0.1676,
+      "step": 129610
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.8207660804625396e-05,
+      "loss": 0.0614,
+      "step": 129620
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.820632243903745e-05,
+      "loss": 0.2157,
+      "step": 129630
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.820498407344951e-05,
+      "loss": 0.1003,
+      "step": 129640
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.820364570786156e-05,
+      "loss": 0.13,
+      "step": 129650
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.820230734227362e-05,
+      "loss": 0.0898,
+      "step": 129660
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.820096897668567e-05,
+      "loss": 0.1016,
+      "step": 129670
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.819963061109773e-05,
+      "loss": 0.1477,
+      "step": 129680
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.8198292245509784e-05,
+      "loss": 0.1061,
+      "step": 129690
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.8196953879921846e-05,
+      "loss": 0.1082,
+      "step": 129700
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.81956155143339e-05,
+      "loss": 0.0911,
+      "step": 129710
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.8194277148745957e-05,
+      "loss": 0.3847,
+      "step": 129720
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.819293878315801e-05,
+      "loss": 0.071,
+      "step": 129730
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.819160041757007e-05,
+      "loss": 0.1213,
+      "step": 129740
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.819026205198212e-05,
+      "loss": 0.0888,
+      "step": 129750
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.818892368639418e-05,
+      "loss": 0.0703,
+      "step": 129760
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.818758532080623e-05,
+      "loss": 0.0523,
+      "step": 129770
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.818624695521829e-05,
+      "loss": 0.1449,
+      "step": 129780
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.8184908589630344e-05,
+      "loss": 0.1153,
+      "step": 129790
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.8183570224042406e-05,
+      "loss": 0.2722,
+      "step": 129800
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.818223185845446e-05,
+      "loss": 0.1187,
+      "step": 129810
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.818089349286652e-05,
+      "loss": 0.2043,
+      "step": 129820
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.817955512727857e-05,
+      "loss": 0.1583,
+      "step": 129830
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.817821676169063e-05,
+      "loss": 0.054,
+      "step": 129840
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.817687839610268e-05,
+      "loss": 0.1103,
+      "step": 129850
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.817554003051474e-05,
+      "loss": 0.0384,
+      "step": 129860
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.8174201664926793e-05,
+      "loss": 0.0646,
+      "step": 129870
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.817286329933885e-05,
+      "loss": 0.1588,
+      "step": 129880
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.8171524933750904e-05,
+      "loss": 0.0979,
+      "step": 129890
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.8170186568162966e-05,
+      "loss": 0.0624,
+      "step": 129900
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.816884820257502e-05,
+      "loss": 0.0503,
+      "step": 129910
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.816750983698708e-05,
+      "loss": 0.0164,
+      "step": 129920
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.816617147139913e-05,
+      "loss": 0.098,
+      "step": 129930
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.816483310581119e-05,
+      "loss": 0.1619,
+      "step": 129940
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.816349474022324e-05,
+      "loss": 0.0391,
+      "step": 129950
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.81621563746353e-05,
+      "loss": 0.0532,
+      "step": 129960
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.8160818009047354e-05,
+      "loss": 0.1108,
+      "step": 129970
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.815947964345941e-05,
+      "loss": 0.0614,
+      "step": 129980
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.8158141277871464e-05,
+      "loss": 0.05,
+      "step": 129990
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.8156802912283526e-05,
+      "loss": 0.0052,
+      "step": 130000
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.815546454669558e-05,
+      "loss": 0.0541,
+      "step": 130010
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.815412618110764e-05,
+      "loss": 0.1864,
+      "step": 130020
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.815278781551969e-05,
+      "loss": 0.0598,
+      "step": 130030
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.815144944993175e-05,
+      "loss": 0.2748,
+      "step": 130040
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.81501110843438e-05,
+      "loss": 0.1455,
+      "step": 130050
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.814877271875586e-05,
+      "loss": 0.0676,
+      "step": 130060
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.8147434353167914e-05,
+      "loss": 0.0829,
+      "step": 130070
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.814609598757997e-05,
+      "loss": 0.1637,
+      "step": 130080
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.8144757621992024e-05,
+      "loss": 0.1368,
+      "step": 130090
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.8143419256404087e-05,
+      "loss": 0.1301,
+      "step": 130100
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.814208089081614e-05,
+      "loss": 0.1986,
+      "step": 130110
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.81407425252282e-05,
+      "loss": 0.1398,
+      "step": 130120
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.813940415964025e-05,
+      "loss": 0.0539,
+      "step": 130130
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.813806579405231e-05,
+      "loss": 0.1559,
+      "step": 130140
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.813672742846436e-05,
+      "loss": 0.035,
+      "step": 130150
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.813538906287642e-05,
+      "loss": 0.2437,
+      "step": 130160
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.8134050697288474e-05,
+      "loss": 0.3136,
+      "step": 130170
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.813271233170053e-05,
+      "loss": 0.1662,
+      "step": 130180
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.8131373966112585e-05,
+      "loss": 0.1367,
+      "step": 130190
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.813003560052464e-05,
+      "loss": 0.1423,
+      "step": 130200
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.8128697234936695e-05,
+      "loss": 0.1803,
+      "step": 130210
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.812735886934875e-05,
+      "loss": 0.1122,
+      "step": 130220
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.8126020503760806e-05,
+      "loss": 0.0889,
+      "step": 130230
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.812468213817286e-05,
+      "loss": 0.0764,
+      "step": 130240
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.8123343772584917e-05,
+      "loss": 0.3557,
+      "step": 130250
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.812200540699697e-05,
+      "loss": 0.1058,
+      "step": 130260
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.812066704140903e-05,
+      "loss": 0.0118,
+      "step": 130270
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.811932867582109e-05,
+      "loss": 0.1319,
+      "step": 130280
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.8117990310233145e-05,
+      "loss": 0.0839,
+      "step": 130290
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.81166519446452e-05,
+      "loss": 0.0085,
+      "step": 130300
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.8115313579057255e-05,
+      "loss": 0.1464,
+      "step": 130310
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.811397521346931e-05,
+      "loss": 0.0094,
+      "step": 130320
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.8112636847881366e-05,
+      "loss": 0.0905,
+      "step": 130330
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.811129848229342e-05,
+      "loss": 0.0527,
+      "step": 130340
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.810996011670548e-05,
+      "loss": 0.1008,
+      "step": 130350
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.810862175111753e-05,
+      "loss": 0.1201,
+      "step": 130360
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.810728338552959e-05,
+      "loss": 0.0795,
+      "step": 130370
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.810594501994165e-05,
+      "loss": 0.2972,
+      "step": 130380
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.8104606654353705e-05,
+      "loss": 0.134,
+      "step": 130390
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.810326828876576e-05,
+      "loss": 0.0344,
+      "step": 130400
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.8101929923177816e-05,
+      "loss": 0.2317,
+      "step": 130410
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.810059155758987e-05,
+      "loss": 0.1809,
+      "step": 130420
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.8099253192001926e-05,
+      "loss": 0.0986,
+      "step": 130430
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.809791482641398e-05,
+      "loss": 0.0493,
+      "step": 130440
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.809657646082604e-05,
+      "loss": 0.1878,
+      "step": 130450
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 0.1823,
+      "step": 130460
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.8093899729650154e-05,
+      "loss": 0.0946,
+      "step": 130470
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.809256136406221e-05,
+      "loss": 0.0086,
+      "step": 130480
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.8091222998474265e-05,
+      "loss": 0.0983,
+      "step": 130490
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.808988463288632e-05,
+      "loss": 0.0155,
+      "step": 130500
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.8088546267298376e-05,
+      "loss": 0.1317,
+      "step": 130510
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.808720790171043e-05,
+      "loss": 0.0915,
+      "step": 130520
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.8085869536122486e-05,
+      "loss": 0.0604,
+      "step": 130530
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.808453117053454e-05,
+      "loss": 0.1723,
+      "step": 130540
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.80831928049466e-05,
+      "loss": 0.0813,
+      "step": 130550
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.808185443935865e-05,
+      "loss": 0.0387,
+      "step": 130560
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.8080516073770714e-05,
+      "loss": 0.1816,
+      "step": 130570
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.807917770818277e-05,
+      "loss": 0.0452,
+      "step": 130580
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.8077839342594825e-05,
+      "loss": 0.0921,
+      "step": 130590
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.807650097700688e-05,
+      "loss": 0.2858,
+      "step": 130600
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.8075162611418936e-05,
+      "loss": 0.0511,
+      "step": 130610
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.807382424583099e-05,
+      "loss": 0.2479,
+      "step": 130620
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.8072485880243047e-05,
+      "loss": 0.1259,
+      "step": 130630
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.80711475146551e-05,
+      "loss": 0.0423,
+      "step": 130640
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.806980914906716e-05,
+      "loss": 0.1011,
+      "step": 130650
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.806847078347921e-05,
+      "loss": 0.1077,
+      "step": 130660
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.8067132417891275e-05,
+      "loss": 0.0992,
+      "step": 130670
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.806579405230333e-05,
+      "loss": 0.1341,
+      "step": 130680
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.8064455686715385e-05,
+      "loss": 0.1692,
+      "step": 130690
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.806311732112744e-05,
+      "loss": 0.0594,
+      "step": 130700
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.8061778955539496e-05,
+      "loss": 0.0615,
+      "step": 130710
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.806044058995155e-05,
+      "loss": 0.0156,
+      "step": 130720
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.805910222436361e-05,
+      "loss": 0.1438,
+      "step": 130730
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.805776385877566e-05,
+      "loss": 0.1342,
+      "step": 130740
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.805642549318772e-05,
+      "loss": 0.1673,
+      "step": 130750
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.805508712759977e-05,
+      "loss": 0.0584,
+      "step": 130760
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.8053748762011835e-05,
+      "loss": 0.055,
+      "step": 130770
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.805241039642389e-05,
+      "loss": 0.0879,
+      "step": 130780
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.8051072030835945e-05,
+      "loss": 0.11,
+      "step": 130790
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.8049733665248e-05,
+      "loss": 0.0729,
+      "step": 130800
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.8048395299660056e-05,
+      "loss": 0.0035,
+      "step": 130810
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.804705693407211e-05,
+      "loss": 0.3474,
+      "step": 130820
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.804571856848417e-05,
+      "loss": 0.2645,
+      "step": 130830
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.804438020289622e-05,
+      "loss": 0.1114,
+      "step": 130840
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.804304183730828e-05,
+      "loss": 0.1962,
+      "step": 130850
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.804170347172033e-05,
+      "loss": 0.0084,
+      "step": 130860
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.8040365106132395e-05,
+      "loss": 0.0383,
+      "step": 130870
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.803902674054445e-05,
+      "loss": 0.0048,
+      "step": 130880
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.8037688374956506e-05,
+      "loss": 0.1047,
+      "step": 130890
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.803635000936856e-05,
+      "loss": 0.1481,
+      "step": 130900
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.8035011643780616e-05,
+      "loss": 0.0719,
+      "step": 130910
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.803367327819267e-05,
+      "loss": 0.1601,
+      "step": 130920
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.803233491260473e-05,
+      "loss": 0.0885,
+      "step": 130930
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.803099654701678e-05,
+      "loss": 0.1919,
+      "step": 130940
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.802965818142884e-05,
+      "loss": 0.1188,
+      "step": 130950
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.802831981584089e-05,
+      "loss": 0.0732,
+      "step": 130960
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.8026981450252955e-05,
+      "loss": 0.1165,
+      "step": 130970
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.802564308466501e-05,
+      "loss": 0.3899,
+      "step": 130980
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.8024304719077066e-05,
+      "loss": 0.0801,
+      "step": 130990
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.802296635348912e-05,
+      "loss": 0.1416,
+      "step": 131000
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.8021627987901176e-05,
+      "loss": 0.0896,
+      "step": 131010
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.802028962231323e-05,
+      "loss": 0.0393,
+      "step": 131020
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.801895125672529e-05,
+      "loss": 0.0215,
+      "step": 131030
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.801761289113734e-05,
+      "loss": 0.0735,
+      "step": 131040
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.80162745255494e-05,
+      "loss": 0.1558,
+      "step": 131050
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.801493615996145e-05,
+      "loss": 0.1314,
+      "step": 131060
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.8013597794373515e-05,
+      "loss": 0.251,
+      "step": 131070
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.801225942878557e-05,
+      "loss": 0.0774,
+      "step": 131080
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.8010921063197626e-05,
+      "loss": 0.1288,
+      "step": 131090
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.800958269760968e-05,
+      "loss": 0.1598,
+      "step": 131100
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.8008244332021737e-05,
+      "loss": 0.1046,
+      "step": 131110
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.800690596643379e-05,
+      "loss": 0.3714,
+      "step": 131120
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.800556760084585e-05,
+      "loss": 0.1269,
+      "step": 131130
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.80042292352579e-05,
+      "loss": 0.03,
+      "step": 131140
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.800289086966996e-05,
+      "loss": 0.2717,
+      "step": 131150
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.800155250408201e-05,
+      "loss": 0.0775,
+      "step": 131160
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.8000214138494075e-05,
+      "loss": 0.0928,
+      "step": 131170
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.799887577290613e-05,
+      "loss": 0.2762,
+      "step": 131180
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.7997537407318186e-05,
+      "loss": 0.0688,
+      "step": 131190
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.799619904173024e-05,
+      "loss": 0.1517,
+      "step": 131200
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.79948606761423e-05,
+      "loss": 0.0068,
+      "step": 131210
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.799352231055435e-05,
+      "loss": 0.0858,
+      "step": 131220
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.799218394496641e-05,
+      "loss": 0.1411,
+      "step": 131230
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.799084557937846e-05,
+      "loss": 0.1488,
+      "step": 131240
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.798950721379052e-05,
+      "loss": 0.2153,
+      "step": 131250
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.7988168848202573e-05,
+      "loss": 0.0228,
+      "step": 131260
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.7986830482614636e-05,
+      "loss": 0.0852,
+      "step": 131270
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.798549211702669e-05,
+      "loss": 0.1254,
+      "step": 131280
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.7984153751438746e-05,
+      "loss": 0.1143,
+      "step": 131290
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.79828153858508e-05,
+      "loss": 0.1223,
+      "step": 131300
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.798147702026286e-05,
+      "loss": 0.0053,
+      "step": 131310
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.798013865467491e-05,
+      "loss": 0.0123,
+      "step": 131320
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.797880028908697e-05,
+      "loss": 0.1557,
+      "step": 131330
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.797746192349902e-05,
+      "loss": 0.1015,
+      "step": 131340
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.797612355791108e-05,
+      "loss": 0.1208,
+      "step": 131350
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.7974785192323134e-05,
+      "loss": 0.0888,
+      "step": 131360
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.7973446826735196e-05,
+      "loss": 0.0714,
+      "step": 131370
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.797210846114725e-05,
+      "loss": 0.2481,
+      "step": 131380
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.7970770095559306e-05,
+      "loss": 0.2343,
+      "step": 131390
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.796943172997136e-05,
+      "loss": 0.1448,
+      "step": 131400
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.796809336438342e-05,
+      "loss": 0.1207,
+      "step": 131410
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.796675499879547e-05,
+      "loss": 0.183,
+      "step": 131420
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.796541663320753e-05,
+      "loss": 0.1302,
+      "step": 131430
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.796407826761958e-05,
+      "loss": 0.1288,
+      "step": 131440
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.796273990203164e-05,
+      "loss": 0.1008,
+      "step": 131450
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.7961401536443694e-05,
+      "loss": 0.2165,
+      "step": 131460
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.7960063170855756e-05,
+      "loss": 0.0711,
+      "step": 131470
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.795872480526781e-05,
+      "loss": 0.0483,
+      "step": 131480
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.7957386439679867e-05,
+      "loss": 0.2014,
+      "step": 131490
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.795604807409192e-05,
+      "loss": 0.0974,
+      "step": 131500
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.795470970850398e-05,
+      "loss": 0.0104,
+      "step": 131510
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.795337134291603e-05,
+      "loss": 0.0055,
+      "step": 131520
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.795203297732809e-05,
+      "loss": 0.1487,
+      "step": 131530
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.795069461174014e-05,
+      "loss": 0.0816,
+      "step": 131540
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.79493562461522e-05,
+      "loss": 0.0812,
+      "step": 131550
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.7948017880564254e-05,
+      "loss": 0.301,
+      "step": 131560
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.7946679514976316e-05,
+      "loss": 0.0054,
+      "step": 131570
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.794534114938837e-05,
+      "loss": 0.082,
+      "step": 131580
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.794400278380043e-05,
+      "loss": 0.0621,
+      "step": 131590
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.794266441821248e-05,
+      "loss": 0.0974,
+      "step": 131600
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.794132605262454e-05,
+      "loss": 0.1043,
+      "step": 131610
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.793998768703659e-05,
+      "loss": 0.1233,
+      "step": 131620
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.793864932144865e-05,
+      "loss": 0.2325,
+      "step": 131630
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.79373109558607e-05,
+      "loss": 0.0717,
+      "step": 131640
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.793597259027276e-05,
+      "loss": 0.0864,
+      "step": 131650
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.7934634224684814e-05,
+      "loss": 0.3511,
+      "step": 131660
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.7933295859096876e-05,
+      "loss": 0.2229,
+      "step": 131670
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.793195749350893e-05,
+      "loss": 0.1044,
+      "step": 131680
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.793061912792099e-05,
+      "loss": 0.0669,
+      "step": 131690
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.792928076233304e-05,
+      "loss": 0.0483,
+      "step": 131700
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.79279423967451e-05,
+      "loss": 0.1855,
+      "step": 131710
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.792660403115715e-05,
+      "loss": 0.1603,
+      "step": 131720
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.792526566556921e-05,
+      "loss": 0.1628,
+      "step": 131730
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.7923927299981264e-05,
+      "loss": 0.185,
+      "step": 131740
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.792258893439332e-05,
+      "loss": 0.2368,
+      "step": 131750
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.792125056880538e-05,
+      "loss": 0.1292,
+      "step": 131760
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.7919912203217436e-05,
+      "loss": 0.0303,
+      "step": 131770
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.791857383762949e-05,
+      "loss": 0.0212,
+      "step": 131780
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.791723547204155e-05,
+      "loss": 0.2243,
+      "step": 131790
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.79158971064536e-05,
+      "loss": 0.108,
+      "step": 131800
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.791455874086566e-05,
+      "loss": 0.1931,
+      "step": 131810
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.791322037527771e-05,
+      "loss": 0.1827,
+      "step": 131820
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.791188200968977e-05,
+      "loss": 0.0777,
+      "step": 131830
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.7910543644101824e-05,
+      "loss": 0.0296,
+      "step": 131840
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.790920527851388e-05,
+      "loss": 0.0345,
+      "step": 131850
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.790786691292594e-05,
+      "loss": 0.1125,
+      "step": 131860
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.7906528547337996e-05,
+      "loss": 0.0411,
+      "step": 131870
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.790519018175005e-05,
+      "loss": 0.0463,
+      "step": 131880
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.790385181616211e-05,
+      "loss": 0.1312,
+      "step": 131890
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.790251345057416e-05,
+      "loss": 0.1164,
+      "step": 131900
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.790117508498622e-05,
+      "loss": 0.0606,
+      "step": 131910
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.789983671939827e-05,
+      "loss": 0.1445,
+      "step": 131920
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.789849835381033e-05,
+      "loss": 0.0286,
+      "step": 131930
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.7897159988222384e-05,
+      "loss": 0.143,
+      "step": 131940
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.789582162263444e-05,
+      "loss": 0.1145,
+      "step": 131950
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.78944832570465e-05,
+      "loss": 0.0439,
+      "step": 131960
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.789314489145856e-05,
+      "loss": 0.1134,
+      "step": 131970
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.789180652587061e-05,
+      "loss": 0.1215,
+      "step": 131980
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.789046816028267e-05,
+      "loss": 0.1395,
+      "step": 131990
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.788912979469472e-05,
+      "loss": 0.2594,
+      "step": 132000
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.788779142910678e-05,
+      "loss": 0.1018,
+      "step": 132010
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.788645306351883e-05,
+      "loss": 0.0028,
+      "step": 132020
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.788511469793089e-05,
+      "loss": 0.0207,
+      "step": 132030
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.7883776332342944e-05,
+      "loss": 0.1307,
+      "step": 132040
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.7882437966755e-05,
+      "loss": 0.0827,
+      "step": 132050
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.788109960116706e-05,
+      "loss": 0.0192,
+      "step": 132060
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.787976123557912e-05,
+      "loss": 0.0387,
+      "step": 132070
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.787842286999117e-05,
+      "loss": 0.0042,
+      "step": 132080
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.787708450440323e-05,
+      "loss": 0.136,
+      "step": 132090
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.787574613881528e-05,
+      "loss": 0.0714,
+      "step": 132100
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.787440777322734e-05,
+      "loss": 0.0358,
+      "step": 132110
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.7873069407639393e-05,
+      "loss": 0.3344,
+      "step": 132120
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.787173104205145e-05,
+      "loss": 0.0809,
+      "step": 132130
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.7870392676463504e-05,
+      "loss": 0.0739,
+      "step": 132140
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.786905431087556e-05,
+      "loss": 0.161,
+      "step": 132150
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.786771594528762e-05,
+      "loss": 0.0457,
+      "step": 132160
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.786637757969968e-05,
+      "loss": 0.1419,
+      "step": 132170
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.786503921411173e-05,
+      "loss": 0.2242,
+      "step": 132180
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.786370084852379e-05,
+      "loss": 0.1422,
+      "step": 132190
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.786236248293584e-05,
+      "loss": 0.0772,
+      "step": 132200
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.78610241173479e-05,
+      "loss": 0.0992,
+      "step": 132210
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.7859685751759954e-05,
+      "loss": 0.0712,
+      "step": 132220
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.785834738617201e-05,
+      "loss": 0.0954,
+      "step": 132230
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.7857009020584064e-05,
+      "loss": 0.1395,
+      "step": 132240
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.785567065499612e-05,
+      "loss": 0.0453,
+      "step": 132250
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.785433228940818e-05,
+      "loss": 0.1147,
+      "step": 132260
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.785299392382024e-05,
+      "loss": 0.0596,
+      "step": 132270
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.785165555823229e-05,
+      "loss": 0.1651,
+      "step": 132280
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.785031719264435e-05,
+      "loss": 0.122,
+      "step": 132290
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.78489788270564e-05,
+      "loss": 0.0272,
+      "step": 132300
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.784764046146846e-05,
+      "loss": 0.0523,
+      "step": 132310
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.7846302095880514e-05,
+      "loss": 0.127,
+      "step": 132320
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.784496373029257e-05,
+      "loss": 0.1875,
+      "step": 132330
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.7843625364704624e-05,
+      "loss": 0.083,
+      "step": 132340
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.784228699911668e-05,
+      "loss": 0.3104,
+      "step": 132350
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.784094863352874e-05,
+      "loss": 0.0646,
+      "step": 132360
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.78396102679408e-05,
+      "loss": 0.1241,
+      "step": 132370
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.783827190235285e-05,
+      "loss": 0.0529,
+      "step": 132380
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.783693353676491e-05,
+      "loss": 0.0628,
+      "step": 132390
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.783559517117696e-05,
+      "loss": 0.0904,
+      "step": 132400
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.783425680558902e-05,
+      "loss": 0.0633,
+      "step": 132410
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.7832918440001074e-05,
+      "loss": 0.0017,
+      "step": 132420
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.783158007441313e-05,
+      "loss": 0.1592,
+      "step": 132430
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.7830241708825185e-05,
+      "loss": 0.0666,
+      "step": 132440
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.782890334323724e-05,
+      "loss": 0.0734,
+      "step": 132450
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.78275649776493e-05,
+      "loss": 0.1509,
+      "step": 132460
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.782622661206136e-05,
+      "loss": 0.1284,
+      "step": 132470
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.782488824647341e-05,
+      "loss": 0.2162,
+      "step": 132480
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.782354988088547e-05,
+      "loss": 0.0455,
+      "step": 132490
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.7822211515297523e-05,
+      "loss": 0.0633,
+      "step": 132500
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.782087314970958e-05,
+      "loss": 0.083,
+      "step": 132510
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.7819534784121634e-05,
+      "loss": 0.0722,
+      "step": 132520
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.781819641853369e-05,
+      "loss": 0.0425,
+      "step": 132530
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.7816858052945745e-05,
+      "loss": 0.0592,
+      "step": 132540
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.78155196873578e-05,
+      "loss": 0.0637,
+      "step": 132550
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.7814181321769855e-05,
+      "loss": 0.0982,
+      "step": 132560
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.781284295618191e-05,
+      "loss": 0.0582,
+      "step": 132570
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.7811504590593966e-05,
+      "loss": 0.0572,
+      "step": 132580
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.781016622500602e-05,
+      "loss": 0.1616,
+      "step": 132590
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.780882785941808e-05,
+      "loss": 0.0586,
+      "step": 132600
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.780748949383013e-05,
+      "loss": 0.0644,
+      "step": 132610
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.780615112824219e-05,
+      "loss": 0.0931,
+      "step": 132620
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.780481276265424e-05,
+      "loss": 0.0914,
+      "step": 132630
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.7803474397066305e-05,
+      "loss": 0.1183,
+      "step": 132640
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.780213603147836e-05,
+      "loss": 0.1955,
+      "step": 132650
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.7800797665890416e-05,
+      "loss": 0.04,
+      "step": 132660
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.779945930030247e-05,
+      "loss": 0.125,
+      "step": 132670
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.7798120934714526e-05,
+      "loss": 0.1134,
+      "step": 132680
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.779678256912658e-05,
+      "loss": 0.1289,
+      "step": 132690
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.779544420353864e-05,
+      "loss": 0.0072,
+      "step": 132700
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.779410583795069e-05,
+      "loss": 0.1257,
+      "step": 132710
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.779276747236275e-05,
+      "loss": 0.0497,
+      "step": 132720
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.77914291067748e-05,
+      "loss": 0.0494,
+      "step": 132730
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.7790090741186865e-05,
+      "loss": 0.1528,
+      "step": 132740
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.778875237559892e-05,
+      "loss": 0.0445,
+      "step": 132750
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.7787414010010976e-05,
+      "loss": 0.0821,
+      "step": 132760
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.778607564442303e-05,
+      "loss": 0.1002,
+      "step": 132770
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.7784737278835086e-05,
+      "loss": 0.1336,
+      "step": 132780
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.778339891324714e-05,
+      "loss": 0.0203,
+      "step": 132790
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.77820605476592e-05,
+      "loss": 0.0278,
+      "step": 132800
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.778072218207125e-05,
+      "loss": 0.1152,
+      "step": 132810
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.777938381648331e-05,
+      "loss": 0.1197,
+      "step": 132820
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.777804545089536e-05,
+      "loss": 0.0711,
+      "step": 132830
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.7776707085307425e-05,
+      "loss": 0.0696,
+      "step": 132840
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": {
+        "accuracy": 0.955736224028907
+      },
+      "eval_f1": {
+        "f1": 0.9555782982813351
+      },
+      "eval_loss": 0.22395330667495728,
+      "eval_precision": {
+        "precision": 0.9563626555520048
+      },
+      "eval_recall": {
+        "recall": 0.9555607789866469
+      },
+      "eval_runtime": 431.0348,
+      "eval_samples_per_second": 38.524,
+      "eval_steps_per_second": 19.263,
+      "step": 132840
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.777536871971948e-05,
+      "loss": 0.1063,
+      "step": 132850
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.7774030354131536e-05,
+      "loss": 0.0358,
+      "step": 132860
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.777269198854359e-05,
+      "loss": 0.2528,
+      "step": 132870
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.7771353622955647e-05,
+      "loss": 0.0909,
+      "step": 132880
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.77700152573677e-05,
+      "loss": 0.0506,
+      "step": 132890
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.776867689177976e-05,
+      "loss": 0.1164,
+      "step": 132900
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.776733852619181e-05,
+      "loss": 0.0014,
+      "step": 132910
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.776600016060387e-05,
+      "loss": 0.1446,
+      "step": 132920
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.776466179501592e-05,
+      "loss": 0.1255,
+      "step": 132930
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.7763323429427985e-05,
+      "loss": 0.1726,
+      "step": 132940
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.776198506384004e-05,
+      "loss": 0.1186,
+      "step": 132950
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.7760646698252096e-05,
+      "loss": 0.1243,
+      "step": 132960
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.775930833266415e-05,
+      "loss": 0.0834,
+      "step": 132970
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.775796996707621e-05,
+      "loss": 0.1084,
+      "step": 132980
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.775663160148826e-05,
+      "loss": 0.0052,
+      "step": 132990
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.775529323590032e-05,
+      "loss": 0.1627,
+      "step": 133000
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.775395487031237e-05,
+      "loss": 0.0137,
+      "step": 133010
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.775261650472443e-05,
+      "loss": 0.0209,
+      "step": 133020
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.7751278139136483e-05,
+      "loss": 0.1434,
+      "step": 133030
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.7749939773548546e-05,
+      "loss": 0.2024,
+      "step": 133040
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.77486014079606e-05,
+      "loss": 0.1415,
+      "step": 133050
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.7747263042372656e-05,
+      "loss": 0.0417,
+      "step": 133060
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.774592467678471e-05,
+      "loss": 0.007,
+      "step": 133070
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.774458631119677e-05,
+      "loss": 0.1248,
+      "step": 133080
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.774324794560882e-05,
+      "loss": 0.1308,
+      "step": 133090
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.774190958002088e-05,
+      "loss": 0.0933,
+      "step": 133100
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.774057121443293e-05,
+      "loss": 0.1389,
+      "step": 133110
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.773923284884499e-05,
+      "loss": 0.1394,
+      "step": 133120
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.7737894483257044e-05,
+      "loss": 0.11,
+      "step": 133130
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.7736556117669106e-05,
+      "loss": 0.0441,
+      "step": 133140
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.773521775208116e-05,
+      "loss": 0.1366,
+      "step": 133150
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.7733879386493216e-05,
+      "loss": 0.2251,
+      "step": 133160
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.773254102090527e-05,
+      "loss": 0.1258,
+      "step": 133170
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.773120265531733e-05,
+      "loss": 0.0482,
+      "step": 133180
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.772986428972938e-05,
+      "loss": 0.1622,
+      "step": 133190
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.772852592414144e-05,
+      "loss": 0.0967,
+      "step": 133200
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.772718755855349e-05,
+      "loss": 0.1018,
+      "step": 133210
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.772584919296555e-05,
+      "loss": 0.2386,
+      "step": 133220
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.772451082737761e-05,
+      "loss": 0.0897,
+      "step": 133230
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.7723172461789666e-05,
+      "loss": 0.0189,
+      "step": 133240
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.772183409620172e-05,
+      "loss": 0.4471,
+      "step": 133250
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.7720495730613776e-05,
+      "loss": 0.2156,
+      "step": 133260
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.771915736502583e-05,
+      "loss": 0.0556,
+      "step": 133270
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.771781899943789e-05,
+      "loss": 0.0361,
+      "step": 133280
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.771648063384994e-05,
+      "loss": 0.0691,
+      "step": 133290
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.7715142268262e-05,
+      "loss": 0.0134,
+      "step": 133300
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.771380390267405e-05,
+      "loss": 0.0712,
+      "step": 133310
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.771246553708611e-05,
+      "loss": 0.0376,
+      "step": 133320
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.771112717149817e-05,
+      "loss": 0.0238,
+      "step": 133330
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.7709788805910226e-05,
+      "loss": 0.0533,
+      "step": 133340
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.770845044032228e-05,
+      "loss": 0.1095,
+      "step": 133350
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.770711207473434e-05,
+      "loss": 0.0419,
+      "step": 133360
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.770577370914639e-05,
+      "loss": 0.098,
+      "step": 133370
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.770443534355845e-05,
+      "loss": 0.0359,
+      "step": 133380
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.77030969779705e-05,
+      "loss": 0.0572,
+      "step": 133390
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.770175861238256e-05,
+      "loss": 0.1105,
+      "step": 133400
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.770042024679461e-05,
+      "loss": 0.165,
+      "step": 133410
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.769908188120667e-05,
+      "loss": 0.2032,
+      "step": 133420
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.769774351561873e-05,
+      "loss": 0.0325,
+      "step": 133430
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.7696405150030786e-05,
+      "loss": 0.0386,
+      "step": 133440
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.769506678444284e-05,
+      "loss": 0.0832,
+      "step": 133450
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.76937284188549e-05,
+      "loss": 0.0341,
+      "step": 133460
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.769239005326695e-05,
+      "loss": 0.1429,
+      "step": 133470
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.769105168767901e-05,
+      "loss": 0.1778,
+      "step": 133480
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.768971332209106e-05,
+      "loss": 0.0269,
+      "step": 133490
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.768837495650312e-05,
+      "loss": 0.0606,
+      "step": 133500
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.7687036590915173e-05,
+      "loss": 0.0339,
+      "step": 133510
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.768569822532723e-05,
+      "loss": 0.0764,
+      "step": 133520
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.768435985973929e-05,
+      "loss": 0.1373,
+      "step": 133530
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.7683021494151346e-05,
+      "loss": 0.1492,
+      "step": 133540
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.76816831285634e-05,
+      "loss": 0.2024,
+      "step": 133550
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.768034476297546e-05,
+      "loss": 0.1371,
+      "step": 133560
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.767900639738751e-05,
+      "loss": 0.1639,
+      "step": 133570
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.767766803179957e-05,
+      "loss": 0.1911,
+      "step": 133580
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.767632966621162e-05,
+      "loss": 0.1476,
+      "step": 133590
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.767499130062368e-05,
+      "loss": 0.3284,
+      "step": 133600
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.7673652935035734e-05,
+      "loss": 0.0076,
+      "step": 133610
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.767231456944779e-05,
+      "loss": 0.2155,
+      "step": 133620
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.767097620385985e-05,
+      "loss": 0.0494,
+      "step": 133630
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.7669637838271906e-05,
+      "loss": 0.0563,
+      "step": 133640
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.766829947268396e-05,
+      "loss": 0.147,
+      "step": 133650
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.766696110709602e-05,
+      "loss": 0.1381,
+      "step": 133660
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.766562274150807e-05,
+      "loss": 0.0696,
+      "step": 133670
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.766428437592013e-05,
+      "loss": 0.0196,
+      "step": 133680
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.766294601033218e-05,
+      "loss": 0.1057,
+      "step": 133690
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.766160764474424e-05,
+      "loss": 0.0998,
+      "step": 133700
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.7660269279156294e-05,
+      "loss": 0.0828,
+      "step": 133710
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.765893091356835e-05,
+      "loss": 0.0444,
+      "step": 133720
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.765759254798041e-05,
+      "loss": 0.1168,
+      "step": 133730
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.7656254182392467e-05,
+      "loss": 0.0217,
+      "step": 133740
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.765491581680452e-05,
+      "loss": 0.0037,
+      "step": 133750
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.765357745121658e-05,
+      "loss": 0.1536,
+      "step": 133760
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.765223908562863e-05,
+      "loss": 0.1865,
+      "step": 133770
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.765090072004069e-05,
+      "loss": 0.005,
+      "step": 133780
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.764956235445274e-05,
+      "loss": 0.0392,
+      "step": 133790
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.76482239888648e-05,
+      "loss": 0.047,
+      "step": 133800
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.7646885623276854e-05,
+      "loss": 0.1551,
+      "step": 133810
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.764554725768891e-05,
+      "loss": 0.1146,
+      "step": 133820
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.764420889210097e-05,
+      "loss": 0.1409,
+      "step": 133830
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.764287052651303e-05,
+      "loss": 0.3301,
+      "step": 133840
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.764153216092508e-05,
+      "loss": 0.0965,
+      "step": 133850
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.764019379533714e-05,
+      "loss": 0.2041,
+      "step": 133860
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.763885542974919e-05,
+      "loss": 0.0947,
+      "step": 133870
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.763751706416125e-05,
+      "loss": 0.0043,
+      "step": 133880
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.7636178698573303e-05,
+      "loss": 0.1216,
+      "step": 133890
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.763484033298536e-05,
+      "loss": 0.0896,
+      "step": 133900
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.7633501967397414e-05,
+      "loss": 0.1057,
+      "step": 133910
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.763216360180947e-05,
+      "loss": 0.1961,
+      "step": 133920
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.763082523622153e-05,
+      "loss": 0.1185,
+      "step": 133930
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.762948687063359e-05,
+      "loss": 0.0738,
+      "step": 133940
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.762814850504564e-05,
+      "loss": 0.0966,
+      "step": 133950
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.76268101394577e-05,
+      "loss": 0.1572,
+      "step": 133960
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.762547177386975e-05,
+      "loss": 0.2185,
+      "step": 133970
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.762413340828181e-05,
+      "loss": 0.0317,
+      "step": 133980
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.7622795042693864e-05,
+      "loss": 0.128,
+      "step": 133990
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.762145667710592e-05,
+      "loss": 0.0579,
+      "step": 134000
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.7620118311517974e-05,
+      "loss": 0.1737,
+      "step": 134010
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.761877994593003e-05,
+      "loss": 0.0054,
+      "step": 134020
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.761744158034209e-05,
+      "loss": 0.11,
+      "step": 134030
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.761610321475415e-05,
+      "loss": 0.1642,
+      "step": 134040
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.76147648491662e-05,
+      "loss": 0.2125,
+      "step": 134050
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.761342648357826e-05,
+      "loss": 0.1055,
+      "step": 134060
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.761208811799031e-05,
+      "loss": 0.0679,
+      "step": 134070
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.761074975240237e-05,
+      "loss": 0.1582,
+      "step": 134080
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.7609411386814424e-05,
+      "loss": 0.1487,
+      "step": 134090
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.760807302122648e-05,
+      "loss": 0.2205,
+      "step": 134100
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.7606734655638534e-05,
+      "loss": 0.078,
+      "step": 134110
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.760539629005059e-05,
+      "loss": 0.1139,
+      "step": 134120
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.760405792446265e-05,
+      "loss": 0.1291,
+      "step": 134130
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.760271955887471e-05,
+      "loss": 0.1614,
+      "step": 134140
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.760138119328676e-05,
+      "loss": 0.0481,
+      "step": 134150
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.760004282769882e-05,
+      "loss": 0.002,
+      "step": 134160
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.759870446211087e-05,
+      "loss": 0.082,
+      "step": 134170
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.759736609652293e-05,
+      "loss": 0.1766,
+      "step": 134180
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.7596027730934984e-05,
+      "loss": 0.0137,
+      "step": 134190
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.759468936534704e-05,
+      "loss": 0.2144,
+      "step": 134200
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.7593350999759095e-05,
+      "loss": 0.1711,
+      "step": 134210
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.759201263417115e-05,
+      "loss": 0.0108,
+      "step": 134220
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.759067426858321e-05,
+      "loss": 0.0888,
+      "step": 134230
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.758933590299527e-05,
+      "loss": 0.0103,
+      "step": 134240
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.758799753740732e-05,
+      "loss": 0.0352,
+      "step": 134250
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.758665917181938e-05,
+      "loss": 0.0909,
+      "step": 134260
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.758532080623143e-05,
+      "loss": 0.3444,
+      "step": 134270
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.758398244064349e-05,
+      "loss": 0.231,
+      "step": 134280
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.7582644075055544e-05,
+      "loss": 0.1263,
+      "step": 134290
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.75813057094676e-05,
+      "loss": 0.1956,
+      "step": 134300
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.7579967343879655e-05,
+      "loss": 0.3812,
+      "step": 134310
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.757862897829171e-05,
+      "loss": 0.0141,
+      "step": 134320
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.757729061270377e-05,
+      "loss": 0.0725,
+      "step": 134330
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.757595224711583e-05,
+      "loss": 0.0465,
+      "step": 134340
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.757461388152788e-05,
+      "loss": 0.0646,
+      "step": 134350
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.757327551593994e-05,
+      "loss": 0.1439,
+      "step": 134360
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.7571937150351994e-05,
+      "loss": 0.1623,
+      "step": 134370
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.757059878476405e-05,
+      "loss": 0.1312,
+      "step": 134380
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.7569260419176104e-05,
+      "loss": 0.0123,
+      "step": 134390
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.756792205358816e-05,
+      "loss": 0.1339,
+      "step": 134400
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.7566583688000215e-05,
+      "loss": 0.0043,
+      "step": 134410
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.756524532241227e-05,
+      "loss": 0.1685,
+      "step": 134420
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.756390695682433e-05,
+      "loss": 0.1378,
+      "step": 134430
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.756256859123639e-05,
+      "loss": 0.1085,
+      "step": 134440
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.756123022564844e-05,
+      "loss": 0.1106,
+      "step": 134450
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.75598918600605e-05,
+      "loss": 0.1103,
+      "step": 134460
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.7558553494472554e-05,
+      "loss": 0.0635,
+      "step": 134470
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.755721512888461e-05,
+      "loss": 0.082,
+      "step": 134480
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.7555876763296664e-05,
+      "loss": 0.1325,
+      "step": 134490
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.755453839770872e-05,
+      "loss": 0.1383,
+      "step": 134500
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.7553200032120775e-05,
+      "loss": 0.0738,
+      "step": 134510
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.755186166653284e-05,
+      "loss": 0.0733,
+      "step": 134520
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.755052330094489e-05,
+      "loss": 0.1737,
+      "step": 134530
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.754918493535695e-05,
+      "loss": 0.0738,
+      "step": 134540
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.7547846569769e-05,
+      "loss": 0.2657,
+      "step": 134550
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.754650820418106e-05,
+      "loss": 0.0561,
+      "step": 134560
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.7545169838593114e-05,
+      "loss": 0.0587,
+      "step": 134570
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.754383147300517e-05,
+      "loss": 0.2063,
+      "step": 134580
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.7542493107417224e-05,
+      "loss": 0.0465,
+      "step": 134590
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.754115474182928e-05,
+      "loss": 0.0717,
+      "step": 134600
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.7539816376241335e-05,
+      "loss": 0.0341,
+      "step": 134610
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.75384780106534e-05,
+      "loss": 0.1054,
+      "step": 134620
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.753713964506545e-05,
+      "loss": 0.0917,
+      "step": 134630
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.753580127947751e-05,
+      "loss": 0.1211,
+      "step": 134640
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.753446291388956e-05,
+      "loss": 0.2662,
+      "step": 134650
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.753312454830162e-05,
+      "loss": 0.0664,
+      "step": 134660
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.7531786182713674e-05,
+      "loss": 0.0651,
+      "step": 134670
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.753044781712573e-05,
+      "loss": 0.1455,
+      "step": 134680
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.7529109451537785e-05,
+      "loss": 0.084,
+      "step": 134690
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.752777108594984e-05,
+      "loss": 0.2586,
+      "step": 134700
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.7526432720361895e-05,
+      "loss": 0.1302,
+      "step": 134710
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.752509435477396e-05,
+      "loss": 0.1575,
+      "step": 134720
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.752375598918601e-05,
+      "loss": 0.1374,
+      "step": 134730
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.752241762359807e-05,
+      "loss": 0.2353,
+      "step": 134740
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.7521079258010123e-05,
+      "loss": 0.0467,
+      "step": 134750
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.751974089242218e-05,
+      "loss": 0.0709,
+      "step": 134760
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.7518402526834234e-05,
+      "loss": 0.0689,
+      "step": 134770
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.751706416124629e-05,
+      "loss": 0.0919,
+      "step": 134780
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.7515725795658345e-05,
+      "loss": 0.0126,
+      "step": 134790
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.75143874300704e-05,
+      "loss": 0.0357,
+      "step": 134800
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.7513049064482455e-05,
+      "loss": 0.1334,
+      "step": 134810
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.751171069889452e-05,
+      "loss": 0.0964,
+      "step": 134820
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.751037233330657e-05,
+      "loss": 0.0935,
+      "step": 134830
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.750903396771863e-05,
+      "loss": 0.1391,
+      "step": 134840
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.7507695602130684e-05,
+      "loss": 0.3993,
+      "step": 134850
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.750635723654274e-05,
+      "loss": 0.0043,
+      "step": 134860
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.7505018870954794e-05,
+      "loss": 0.16,
+      "step": 134870
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.750368050536685e-05,
+      "loss": 0.1183,
+      "step": 134880
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.7502342139778905e-05,
+      "loss": 0.0682,
+      "step": 134890
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.750100377419096e-05,
+      "loss": 0.1763,
+      "step": 134900
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.7499665408603016e-05,
+      "loss": 0.1615,
+      "step": 134910
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.749832704301507e-05,
+      "loss": 0.3485,
+      "step": 134920
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.7496988677427126e-05,
+      "loss": 0.2729,
+      "step": 134930
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.749565031183918e-05,
+      "loss": 0.194,
+      "step": 134940
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.749431194625124e-05,
+      "loss": 0.08,
+      "step": 134950
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.749297358066329e-05,
+      "loss": 0.1532,
+      "step": 134960
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.749163521507535e-05,
+      "loss": 0.5254,
+      "step": 134970
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.74902968494874e-05,
+      "loss": 0.137,
+      "step": 134980
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.748895848389946e-05,
+      "loss": 0.1535,
+      "step": 134990
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.748762011831152e-05,
+      "loss": 0.1785,
+      "step": 135000
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.7486281752723576e-05,
+      "loss": 0.0586,
+      "step": 135010
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.748494338713563e-05,
+      "loss": 0.1047,
+      "step": 135020
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.7483605021547686e-05,
+      "loss": 0.0659,
+      "step": 135030
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.748226665595974e-05,
+      "loss": 0.1267,
+      "step": 135040
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.74809282903718e-05,
+      "loss": 0.2281,
+      "step": 135050
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.747958992478385e-05,
+      "loss": 0.1314,
+      "step": 135060
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.747825155919591e-05,
+      "loss": 0.2842,
+      "step": 135070
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.747691319360796e-05,
+      "loss": 0.1368,
+      "step": 135080
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.747557482802002e-05,
+      "loss": 0.03,
+      "step": 135090
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.747423646243208e-05,
+      "loss": 0.0679,
+      "step": 135100
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.7472898096844136e-05,
+      "loss": 0.1973,
+      "step": 135110
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.747155973125619e-05,
+      "loss": 0.0103,
+      "step": 135120
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.747022136566825e-05,
+      "loss": 0.1555,
+      "step": 135130
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.74688830000803e-05,
+      "loss": 0.0677,
+      "step": 135140
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.746754463449236e-05,
+      "loss": 0.1132,
+      "step": 135150
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.746620626890441e-05,
+      "loss": 0.0552,
+      "step": 135160
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.746486790331647e-05,
+      "loss": 0.1376,
+      "step": 135170
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.746352953772852e-05,
+      "loss": 0.1147,
+      "step": 135180
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.746219117214058e-05,
+      "loss": 0.0479,
+      "step": 135190
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.746085280655264e-05,
+      "loss": 0.0367,
+      "step": 135200
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.7459514440964696e-05,
+      "loss": 0.0144,
+      "step": 135210
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.745817607537675e-05,
+      "loss": 0.1204,
+      "step": 135220
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.745683770978881e-05,
+      "loss": 0.143,
+      "step": 135230
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.745549934420086e-05,
+      "loss": 0.0702,
+      "step": 135240
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.745416097861292e-05,
+      "loss": 0.0168,
+      "step": 135250
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.745282261302497e-05,
+      "loss": 0.1493,
+      "step": 135260
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.745148424743703e-05,
+      "loss": 0.1188,
+      "step": 135270
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.7450145881849083e-05,
+      "loss": 0.0032,
+      "step": 135280
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.744880751626114e-05,
+      "loss": 0.1298,
+      "step": 135290
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.74474691506732e-05,
+      "loss": 0.0209,
+      "step": 135300
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.7446130785085256e-05,
+      "loss": 0.0569,
+      "step": 135310
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.744479241949731e-05,
+      "loss": 0.0256,
+      "step": 135320
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.744345405390937e-05,
+      "loss": 0.0036,
+      "step": 135330
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.744211568832142e-05,
+      "loss": 0.1187,
+      "step": 135340
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.744077732273348e-05,
+      "loss": 0.0047,
+      "step": 135350
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.743943895714553e-05,
+      "loss": 0.1296,
+      "step": 135360
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.743810059155759e-05,
+      "loss": 0.1305,
+      "step": 135370
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.7436762225969644e-05,
+      "loss": 0.102,
+      "step": 135380
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.74354238603817e-05,
+      "loss": 0.087,
+      "step": 135390
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.743408549479376e-05,
+      "loss": 0.0419,
+      "step": 135400
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.7432747129205816e-05,
+      "loss": 0.0554,
+      "step": 135410
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.743140876361787e-05,
+      "loss": 0.2191,
+      "step": 135420
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.743007039802993e-05,
+      "loss": 0.004,
+      "step": 135430
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.742873203244198e-05,
+      "loss": 0.2101,
+      "step": 135440
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.742739366685404e-05,
+      "loss": 0.1476,
+      "step": 135450
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.742605530126609e-05,
+      "loss": 0.0054,
+      "step": 135460
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.742471693567815e-05,
+      "loss": 0.2373,
+      "step": 135470
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.7423378570090204e-05,
+      "loss": 0.1242,
+      "step": 135480
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.742204020450226e-05,
+      "loss": 0.0839,
+      "step": 135490
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.742070183891432e-05,
+      "loss": 0.0545,
+      "step": 135500
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.7419363473326377e-05,
+      "loss": 0.0815,
+      "step": 135510
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.741802510773843e-05,
+      "loss": 0.0656,
+      "step": 135520
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.741668674215049e-05,
+      "loss": 0.0746,
+      "step": 135530
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.741534837656254e-05,
+      "loss": 0.14,
+      "step": 135540
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.74140100109746e-05,
+      "loss": 0.0549,
+      "step": 135550
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.741267164538665e-05,
+      "loss": 0.0526,
+      "step": 135560
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.741133327979871e-05,
+      "loss": 0.179,
+      "step": 135570
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.7409994914210764e-05,
+      "loss": 0.0478,
+      "step": 135580
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.740865654862282e-05,
+      "loss": 0.0195,
+      "step": 135590
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.740731818303488e-05,
+      "loss": 0.1922,
+      "step": 135600
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.740597981744694e-05,
+      "loss": 0.0194,
+      "step": 135610
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.740464145185899e-05,
+      "loss": 0.0428,
+      "step": 135620
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.740330308627105e-05,
+      "loss": 0.1435,
+      "step": 135630
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.74019647206831e-05,
+      "loss": 0.1389,
+      "step": 135640
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.740062635509516e-05,
+      "loss": 0.0454,
+      "step": 135650
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.7399287989507213e-05,
+      "loss": 0.2567,
+      "step": 135660
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.739794962391927e-05,
+      "loss": 0.0048,
+      "step": 135670
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.7396611258331324e-05,
+      "loss": 0.1138,
+      "step": 135680
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.739527289274338e-05,
+      "loss": 0.094,
+      "step": 135690
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.739393452715544e-05,
+      "loss": 0.1454,
+      "step": 135700
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.73925961615675e-05,
+      "loss": 0.1049,
+      "step": 135710
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.739125779597955e-05,
+      "loss": 0.1353,
+      "step": 135720
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.738991943039161e-05,
+      "loss": 0.0924,
+      "step": 135730
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.738858106480366e-05,
+      "loss": 0.1559,
+      "step": 135740
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.738724269921572e-05,
+      "loss": 0.1174,
+      "step": 135750
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.7385904333627774e-05,
+      "loss": 0.1476,
+      "step": 135760
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.738456596803983e-05,
+      "loss": 0.0901,
+      "step": 135770
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.7383227602451884e-05,
+      "loss": 0.0061,
+      "step": 135780
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.738188923686394e-05,
+      "loss": 0.0866,
+      "step": 135790
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.7380550871276e-05,
+      "loss": 0.055,
+      "step": 135800
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.737921250568806e-05,
+      "loss": 0.1672,
+      "step": 135810
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.737787414010011e-05,
+      "loss": 0.0765,
+      "step": 135820
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.737653577451217e-05,
+      "loss": 0.2719,
+      "step": 135830
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.737519740892422e-05,
+      "loss": 0.1822,
+      "step": 135840
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.737385904333628e-05,
+      "loss": 0.0843,
+      "step": 135850
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.7372520677748334e-05,
+      "loss": 0.0061,
+      "step": 135860
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.737118231216039e-05,
+      "loss": 0.0917,
+      "step": 135870
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.7369843946572444e-05,
+      "loss": 0.1647,
+      "step": 135880
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.73685055809845e-05,
+      "loss": 0.0704,
+      "step": 135890
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.736716721539656e-05,
+      "loss": 0.0903,
+      "step": 135900
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.736582884980862e-05,
+      "loss": 0.0326,
+      "step": 135910
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.736449048422067e-05,
+      "loss": 0.205,
+      "step": 135920
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.736315211863273e-05,
+      "loss": 0.3352,
+      "step": 135930
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.736181375304478e-05,
+      "loss": 0.0338,
+      "step": 135940
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.736047538745684e-05,
+      "loss": 0.2238,
+      "step": 135950
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.7359137021868894e-05,
+      "loss": 0.1637,
+      "step": 135960
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.735779865628095e-05,
+      "loss": 0.0666,
+      "step": 135970
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.7356460290693005e-05,
+      "loss": 0.2892,
+      "step": 135980
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.735512192510507e-05,
+      "loss": 0.0134,
+      "step": 135990
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.735378355951712e-05,
+      "loss": 0.0521,
+      "step": 136000
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.735244519392918e-05,
+      "loss": 0.1494,
+      "step": 136010
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.735110682834123e-05,
+      "loss": 0.2193,
+      "step": 136020
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.734976846275329e-05,
+      "loss": 0.13,
+      "step": 136030
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.734843009716534e-05,
+      "loss": 0.0925,
+      "step": 136040
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.73470917315774e-05,
+      "loss": 0.0244,
+      "step": 136050
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.7345753365989454e-05,
+      "loss": 0.0698,
+      "step": 136060
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.734441500040151e-05,
+      "loss": 0.0239,
+      "step": 136070
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.7343076634813565e-05,
+      "loss": 0.1151,
+      "step": 136080
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.734173826922563e-05,
+      "loss": 0.182,
+      "step": 136090
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.734039990363768e-05,
+      "loss": 0.3527,
+      "step": 136100
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.733906153804974e-05,
+      "loss": 0.2211,
+      "step": 136110
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.733772317246179e-05,
+      "loss": 0.1106,
+      "step": 136120
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.733638480687385e-05,
+      "loss": 0.0842,
+      "step": 136130
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.7335046441285903e-05,
+      "loss": 0.1212,
+      "step": 136140
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.733370807569796e-05,
+      "loss": 0.0535,
+      "step": 136150
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.7332369710110014e-05,
+      "loss": 0.0901,
+      "step": 136160
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.733103134452207e-05,
+      "loss": 0.0385,
+      "step": 136170
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.7329692978934125e-05,
+      "loss": 0.1037,
+      "step": 136180
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.732835461334619e-05,
+      "loss": 0.0808,
+      "step": 136190
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.732701624775824e-05,
+      "loss": 0.2105,
+      "step": 136200
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.73256778821703e-05,
+      "loss": 0.1107,
+      "step": 136210
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.732433951658235e-05,
+      "loss": 0.0885,
+      "step": 136220
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.732300115099441e-05,
+      "loss": 0.141,
+      "step": 136230
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.7321662785406464e-05,
+      "loss": 0.1869,
+      "step": 136240
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.732032441981852e-05,
+      "loss": 0.0521,
+      "step": 136250
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.7318986054230574e-05,
+      "loss": 0.0645,
+      "step": 136260
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.731764768864263e-05,
+      "loss": 0.3357,
+      "step": 136270
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.7316309323054685e-05,
+      "loss": 0.0807,
+      "step": 136280
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.731497095746675e-05,
+      "loss": 0.1141,
+      "step": 136290
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.73136325918788e-05,
+      "loss": 0.0874,
+      "step": 136300
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.731229422629086e-05,
+      "loss": 0.0679,
+      "step": 136310
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.731095586070291e-05,
+      "loss": 0.0568,
+      "step": 136320
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.730961749511497e-05,
+      "loss": 0.1483,
+      "step": 136330
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.7308279129527024e-05,
+      "loss": 0.1046,
+      "step": 136340
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.730694076393908e-05,
+      "loss": 0.0549,
+      "step": 136350
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.7305602398351134e-05,
+      "loss": 0.0492,
+      "step": 136360
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.730426403276319e-05,
+      "loss": 0.2599,
+      "step": 136370
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.7302925667175245e-05,
+      "loss": 0.1427,
+      "step": 136380
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.730158730158731e-05,
+      "loss": 0.0748,
+      "step": 136390
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.730024893599936e-05,
+      "loss": 0.2311,
+      "step": 136400
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.729891057041142e-05,
+      "loss": 0.0964,
+      "step": 136410
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.729757220482347e-05,
+      "loss": 0.1652,
+      "step": 136420
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.729623383923553e-05,
+      "loss": 0.1117,
+      "step": 136430
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.7294895473647584e-05,
+      "loss": 0.1406,
+      "step": 136440
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.729355710805964e-05,
+      "loss": 0.1229,
+      "step": 136450
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.7292218742471695e-05,
+      "loss": 0.1001,
+      "step": 136460
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.729088037688375e-05,
+      "loss": 0.1209,
+      "step": 136470
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.7289542011295805e-05,
+      "loss": 0.2069,
+      "step": 136480
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.728820364570787e-05,
+      "loss": 0.1366,
+      "step": 136490
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.728686528011992e-05,
+      "loss": 0.0751,
+      "step": 136500
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.728552691453198e-05,
+      "loss": 0.1007,
+      "step": 136510
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.7284188548944033e-05,
+      "loss": 0.1579,
+      "step": 136520
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.728285018335609e-05,
+      "loss": 0.2176,
+      "step": 136530
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.7281511817768144e-05,
+      "loss": 0.0805,
+      "step": 136540
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.72801734521802e-05,
+      "loss": 0.0149,
+      "step": 136550
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.7278835086592255e-05,
+      "loss": 0.2511,
+      "step": 136560
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.727749672100431e-05,
+      "loss": 0.0941,
+      "step": 136570
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.7276158355416365e-05,
+      "loss": 0.1398,
+      "step": 136580
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.727481998982843e-05,
+      "loss": 0.19,
+      "step": 136590
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.727348162424048e-05,
+      "loss": 0.09,
+      "step": 136600
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.727214325865254e-05,
+      "loss": 0.1397,
+      "step": 136610
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.7270804893064594e-05,
+      "loss": 0.1986,
+      "step": 136620
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.726946652747665e-05,
+      "loss": 0.2554,
+      "step": 136630
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.7268128161888704e-05,
+      "loss": 0.0524,
+      "step": 136640
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.726678979630076e-05,
+      "loss": 0.0663,
+      "step": 136650
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.7265451430712815e-05,
+      "loss": 0.0086,
+      "step": 136660
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.726411306512487e-05,
+      "loss": 0.155,
+      "step": 136670
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.7262774699536926e-05,
+      "loss": 0.0288,
+      "step": 136680
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.726143633394899e-05,
+      "loss": 0.0871,
+      "step": 136690
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.726009796836104e-05,
+      "loss": 0.0035,
+      "step": 136700
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.72587596027731e-05,
+      "loss": 0.2756,
+      "step": 136710
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.7257421237185154e-05,
+      "loss": 0.0796,
+      "step": 136720
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.725608287159721e-05,
+      "loss": 0.1297,
+      "step": 136730
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.7254744506009264e-05,
+      "loss": 0.0059,
+      "step": 136740
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.725340614042132e-05,
+      "loss": 0.0589,
+      "step": 136750
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.7252067774833375e-05,
+      "loss": 0.1144,
+      "step": 136760
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.725072940924543e-05,
+      "loss": 0.0057,
+      "step": 136770
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.7249391043657486e-05,
+      "loss": 0.1285,
+      "step": 136780
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.724805267806955e-05,
+      "loss": 0.0934,
+      "step": 136790
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.72467143124816e-05,
+      "loss": 0.0263,
+      "step": 136800
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.724537594689366e-05,
+      "loss": 0.0668,
+      "step": 136810
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.7244037581305714e-05,
+      "loss": 0.1137,
+      "step": 136820
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.724269921571777e-05,
+      "loss": 0.1238,
+      "step": 136830
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.7241360850129825e-05,
+      "loss": 0.0194,
+      "step": 136840
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.724002248454188e-05,
+      "loss": 0.0724,
+      "step": 136850
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.7238684118953935e-05,
+      "loss": 0.1335,
+      "step": 136860
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.723734575336599e-05,
+      "loss": 0.0488,
+      "step": 136870
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.7236007387778046e-05,
+      "loss": 0.1726,
+      "step": 136880
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.723466902219011e-05,
+      "loss": 0.0059,
+      "step": 136890
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.723333065660216e-05,
+      "loss": 0.138,
+      "step": 136900
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.723199229101422e-05,
+      "loss": 0.0908,
+      "step": 136910
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.7230653925426274e-05,
+      "loss": 0.0765,
+      "step": 136920
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.722931555983833e-05,
+      "loss": 0.1246,
+      "step": 136930
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.7227977194250385e-05,
+      "loss": 0.1269,
+      "step": 136940
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.722663882866244e-05,
+      "loss": 0.1288,
+      "step": 136950
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.7225300463074495e-05,
+      "loss": 0.0401,
+      "step": 136960
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.722396209748655e-05,
+      "loss": 0.0346,
+      "step": 136970
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.7222623731898606e-05,
+      "loss": 0.1303,
+      "step": 136980
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.722128536631067e-05,
+      "loss": 0.0234,
+      "step": 136990
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.7219947000722724e-05,
+      "loss": 0.0638,
+      "step": 137000
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.721860863513478e-05,
+      "loss": 0.0662,
+      "step": 137010
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.7217270269546834e-05,
+      "loss": 0.1912,
+      "step": 137020
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.721593190395889e-05,
+      "loss": 0.0752,
+      "step": 137030
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.7214593538370945e-05,
+      "loss": 0.0927,
+      "step": 137040
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.7213255172783e-05,
+      "loss": 0.0315,
+      "step": 137050
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.7211916807195056e-05,
+      "loss": 0.0228,
+      "step": 137060
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.721057844160711e-05,
+      "loss": 0.1217,
+      "step": 137070
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.7209240076019166e-05,
+      "loss": 0.224,
+      "step": 137080
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.720790171043123e-05,
+      "loss": 0.0158,
+      "step": 137090
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.7206563344843284e-05,
+      "loss": 0.0042,
+      "step": 137100
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.720522497925534e-05,
+      "loss": 0.2657,
+      "step": 137110
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.7203886613667394e-05,
+      "loss": 0.2109,
+      "step": 137120
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.720254824807945e-05,
+      "loss": 0.0294,
+      "step": 137130
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.7201209882491505e-05,
+      "loss": 0.0739,
+      "step": 137140
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.719987151690356e-05,
+      "loss": 0.1244,
+      "step": 137150
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.7198533151315616e-05,
+      "loss": 0.0112,
+      "step": 137160
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.719719478572767e-05,
+      "loss": 0.1468,
+      "step": 137170
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.7195856420139726e-05,
+      "loss": 0.0881,
+      "step": 137180
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.719451805455179e-05,
+      "loss": 0.0875,
+      "step": 137190
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.7193179688963844e-05,
+      "loss": 0.0932,
+      "step": 137200
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.71918413233759e-05,
+      "loss": 0.0229,
+      "step": 137210
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.7190502957787954e-05,
+      "loss": 0.1887,
+      "step": 137220
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.718916459220001e-05,
+      "loss": 0.1393,
+      "step": 137230
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.7187826226612065e-05,
+      "loss": 0.191,
+      "step": 137240
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.718648786102412e-05,
+      "loss": 0.2449,
+      "step": 137250
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.7185149495436176e-05,
+      "loss": 0.1819,
+      "step": 137260
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.718381112984823e-05,
+      "loss": 0.0726,
+      "step": 137270
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.7182472764260287e-05,
+      "loss": 0.0767,
+      "step": 137280
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.718113439867234e-05,
+      "loss": 0.1064,
+      "step": 137290
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.71797960330844e-05,
+      "loss": 0.1621,
+      "step": 137300
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.717845766749645e-05,
+      "loss": 0.0036,
+      "step": 137310
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.717711930190851e-05,
+      "loss": 0.0453,
+      "step": 137320
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.717578093632056e-05,
+      "loss": 0.052,
+      "step": 137330
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.717444257073262e-05,
+      "loss": 0.0451,
+      "step": 137340
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.7173104205144674e-05,
+      "loss": 0.0209,
+      "step": 137350
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.717176583955673e-05,
+      "loss": 0.1388,
+      "step": 137360
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.717042747396879e-05,
+      "loss": 0.2657,
+      "step": 137370
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.716908910838085e-05,
+      "loss": 0.1894,
+      "step": 137380
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.71677507427929e-05,
+      "loss": 0.0829,
+      "step": 137390
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.716641237720496e-05,
+      "loss": 0.0093,
+      "step": 137400
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.716507401161701e-05,
+      "loss": 0.0074,
+      "step": 137410
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.716373564602907e-05,
+      "loss": 0.1319,
+      "step": 137420
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.716239728044112e-05,
+      "loss": 0.0797,
+      "step": 137430
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.716105891485318e-05,
+      "loss": 0.0636,
+      "step": 137440
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.7159720549265234e-05,
+      "loss": 0.0257,
+      "step": 137450
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.715838218367729e-05,
+      "loss": 0.0723,
+      "step": 137460
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.715704381808935e-05,
+      "loss": 0.2042,
+      "step": 137470
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.715570545250141e-05,
+      "loss": 0.1929,
+      "step": 137480
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.715436708691346e-05,
+      "loss": 0.0663,
+      "step": 137490
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.715302872132552e-05,
+      "loss": 0.0197,
+      "step": 137500
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.715169035573757e-05,
+      "loss": 0.2083,
+      "step": 137510
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.715035199014963e-05,
+      "loss": 0.0738,
+      "step": 137520
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.7149013624561684e-05,
+      "loss": 0.2002,
+      "step": 137530
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.714767525897374e-05,
+      "loss": 0.0146,
+      "step": 137540
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.7146336893385794e-05,
+      "loss": 0.1114,
+      "step": 137550
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.7144998527797856e-05,
+      "loss": 0.2075,
+      "step": 137560
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.714366016220991e-05,
+      "loss": 0.2261,
+      "step": 137570
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.714232179662197e-05,
+      "loss": 0.1278,
+      "step": 137580
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.714098343103402e-05,
+      "loss": 0.0065,
+      "step": 137590
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.713964506544608e-05,
+      "loss": 0.1981,
+      "step": 137600
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.713830669985813e-05,
+      "loss": 0.0574,
+      "step": 137610
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.713696833427019e-05,
+      "loss": 0.1605,
+      "step": 137620
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.7135629968682244e-05,
+      "loss": 0.2189,
+      "step": 137630
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.71342916030943e-05,
+      "loss": 0.1223,
+      "step": 137640
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.7132953237506354e-05,
+      "loss": 0.1615,
+      "step": 137650
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.7131614871918416e-05,
+      "loss": 0.0237,
+      "step": 137660
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.713027650633047e-05,
+      "loss": 0.2505,
+      "step": 137670
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.712893814074253e-05,
+      "loss": 0.0595,
+      "step": 137680
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.712759977515458e-05,
+      "loss": 0.0245,
+      "step": 137690
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.712626140956664e-05,
+      "loss": 0.1066,
+      "step": 137700
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.712492304397869e-05,
+      "loss": 0.0638,
+      "step": 137710
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.712358467839075e-05,
+      "loss": 0.1731,
+      "step": 137720
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.7122246312802804e-05,
+      "loss": 0.1342,
+      "step": 137730
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.712090794721486e-05,
+      "loss": 0.1042,
+      "step": 137740
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.7119569581626914e-05,
+      "loss": 0.0521,
+      "step": 137750
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.7118231216038977e-05,
+      "loss": 0.0502,
+      "step": 137760
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.711689285045103e-05,
+      "loss": 0.153,
+      "step": 137770
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.711555448486309e-05,
+      "loss": 0.2959,
+      "step": 137780
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.711421611927514e-05,
+      "loss": 0.1977,
+      "step": 137790
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.71128777536872e-05,
+      "loss": 0.0653,
+      "step": 137800
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.711153938809925e-05,
+      "loss": 0.249,
+      "step": 137810
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.711020102251131e-05,
+      "loss": 0.072,
+      "step": 137820
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.7108862656923364e-05,
+      "loss": 0.021,
+      "step": 137830
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.710752429133542e-05,
+      "loss": 0.0346,
+      "step": 137840
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.7106185925747475e-05,
+      "loss": 0.2126,
+      "step": 137850
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.710484756015954e-05,
+      "loss": 0.1706,
+      "step": 137860
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.710350919457159e-05,
+      "loss": 0.0092,
+      "step": 137870
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.710217082898365e-05,
+      "loss": 0.0106,
+      "step": 137880
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.71008324633957e-05,
+      "loss": 0.1931,
+      "step": 137890
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.709949409780776e-05,
+      "loss": 0.1835,
+      "step": 137900
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.7098155732219813e-05,
+      "loss": 0.2182,
+      "step": 137910
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.709681736663187e-05,
+      "loss": 0.1418,
+      "step": 137920
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.7095479001043924e-05,
+      "loss": 0.0457,
+      "step": 137930
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.709414063545598e-05,
+      "loss": 0.0785,
+      "step": 137940
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.7092802269868035e-05,
+      "loss": 0.1646,
+      "step": 137950
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.70914639042801e-05,
+      "loss": 0.169,
+      "step": 137960
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.709012553869215e-05,
+      "loss": 0.0957,
+      "step": 137970
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.708878717310421e-05,
+      "loss": 0.0495,
+      "step": 137980
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.708744880751626e-05,
+      "loss": 0.0365,
+      "step": 137990
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.708611044192832e-05,
+      "loss": 0.0797,
+      "step": 138000
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.7084772076340374e-05,
+      "loss": 0.0431,
+      "step": 138010
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.708343371075243e-05,
+      "loss": 0.1418,
+      "step": 138020
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.7082095345164484e-05,
+      "loss": 0.0497,
+      "step": 138030
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.708075697957654e-05,
+      "loss": 0.1709,
+      "step": 138040
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.7079418613988595e-05,
+      "loss": 0.022,
+      "step": 138050
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.707808024840066e-05,
+      "loss": 0.0123,
+      "step": 138060
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.707674188281271e-05,
+      "loss": 0.1365,
+      "step": 138070
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.707540351722477e-05,
+      "loss": 0.0714,
+      "step": 138080
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.707406515163682e-05,
+      "loss": 0.0097,
+      "step": 138090
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.707272678604888e-05,
+      "loss": 0.0714,
+      "step": 138100
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.7071388420460934e-05,
+      "loss": 0.1072,
+      "step": 138110
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.707005005487299e-05,
+      "loss": 0.2686,
+      "step": 138120
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.7068711689285044e-05,
+      "loss": 0.0734,
+      "step": 138130
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.70673733236971e-05,
+      "loss": 0.113,
+      "step": 138140
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.7066034958109155e-05,
+      "loss": 0.1832,
+      "step": 138150
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.706469659252122e-05,
+      "loss": 0.1231,
+      "step": 138160
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.706335822693327e-05,
+      "loss": 0.1249,
+      "step": 138170
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.706201986134533e-05,
+      "loss": 0.2049,
+      "step": 138180
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.706068149575738e-05,
+      "loss": 0.0093,
+      "step": 138190
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.705934313016944e-05,
+      "loss": 0.0745,
+      "step": 138200
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.7058004764581494e-05,
+      "loss": 0.189,
+      "step": 138210
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.705666639899355e-05,
+      "loss": 0.2174,
+      "step": 138220
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.7055328033405605e-05,
+      "loss": 0.1751,
+      "step": 138230
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.705398966781766e-05,
+      "loss": 0.0277,
+      "step": 138240
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.7052651302229715e-05,
+      "loss": 0.0992,
+      "step": 138250
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.705131293664178e-05,
+      "loss": 0.0165,
+      "step": 138260
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.704997457105383e-05,
+      "loss": 0.2298,
+      "step": 138270
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.704863620546589e-05,
+      "loss": 0.187,
+      "step": 138280
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.704729783987794e-05,
+      "loss": 0.0911,
+      "step": 138290
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.704595947429e-05,
+      "loss": 0.1608,
+      "step": 138300
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.7044621108702054e-05,
+      "loss": 0.0576,
+      "step": 138310
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.704328274311411e-05,
+      "loss": 0.0476,
+      "step": 138320
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.7041944377526165e-05,
+      "loss": 0.1263,
+      "step": 138330
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.704060601193822e-05,
+      "loss": 0.0575,
+      "step": 138340
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.7039267646350275e-05,
+      "loss": 0.0061,
+      "step": 138350
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.703792928076234e-05,
+      "loss": 0.0467,
+      "step": 138360
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.703659091517439e-05,
+      "loss": 0.0029,
+      "step": 138370
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.703525254958645e-05,
+      "loss": 0.0493,
+      "step": 138380
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.7033914183998504e-05,
+      "loss": 0.2076,
+      "step": 138390
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.703257581841056e-05,
+      "loss": 0.0291,
+      "step": 138400
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.7031237452822614e-05,
+      "loss": 0.0687,
+      "step": 138410
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.702989908723467e-05,
+      "loss": 0.1237,
+      "step": 138420
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.7028560721646725e-05,
+      "loss": 0.1047,
+      "step": 138430
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.702722235605878e-05,
+      "loss": 0.252,
+      "step": 138440
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.7025883990470836e-05,
+      "loss": 0.1386,
+      "step": 138450
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.70245456248829e-05,
+      "loss": 0.1613,
+      "step": 138460
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.702320725929495e-05,
+      "loss": 0.2075,
+      "step": 138470
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.702186889370701e-05,
+      "loss": 0.0118,
+      "step": 138480
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.7020530528119064e-05,
+      "loss": 0.1016,
+      "step": 138490
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.701919216253112e-05,
+      "loss": 0.1357,
+      "step": 138500
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.7017853796943174e-05,
+      "loss": 0.031,
+      "step": 138510
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.701651543135523e-05,
+      "loss": 0.1502,
+      "step": 138520
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.7015177065767285e-05,
+      "loss": 0.1787,
+      "step": 138530
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.701383870017934e-05,
+      "loss": 0.0097,
+      "step": 138540
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.7012500334591396e-05,
+      "loss": 0.0038,
+      "step": 138550
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.701116196900346e-05,
+      "loss": 0.0739,
+      "step": 138560
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.700982360341551e-05,
+      "loss": 0.0751,
+      "step": 138570
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.700848523782757e-05,
+      "loss": 0.0529,
+      "step": 138580
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.7007146872239624e-05,
+      "loss": 0.0128,
+      "step": 138590
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.700580850665168e-05,
+      "loss": 0.05,
+      "step": 138600
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.7004470141063735e-05,
+      "loss": 0.0844,
+      "step": 138610
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.700313177547579e-05,
+      "loss": 0.1973,
+      "step": 138620
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.7001793409887845e-05,
+      "loss": 0.0305,
+      "step": 138630
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.70004550442999e-05,
+      "loss": 0.2556,
+      "step": 138640
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.6999116678711956e-05,
+      "loss": 0.1748,
+      "step": 138650
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.699777831312402e-05,
+      "loss": 0.1222,
+      "step": 138660
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.699643994753607e-05,
+      "loss": 0.0652,
+      "step": 138670
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.699510158194813e-05,
+      "loss": 0.0666,
+      "step": 138680
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.6993763216360184e-05,
+      "loss": 0.035,
+      "step": 138690
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.699242485077224e-05,
+      "loss": 0.1128,
+      "step": 138700
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.6991086485184295e-05,
+      "loss": 0.0526,
+      "step": 138710
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.698974811959635e-05,
+      "loss": 0.1375,
+      "step": 138720
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.6988409754008405e-05,
+      "loss": 0.2857,
+      "step": 138730
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.698707138842046e-05,
+      "loss": 0.0892,
+      "step": 138740
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.6985733022832516e-05,
+      "loss": 0.0113,
+      "step": 138750
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.698439465724458e-05,
+      "loss": 0.0962,
+      "step": 138760
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.6983056291656633e-05,
+      "loss": 0.07,
+      "step": 138770
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.698171792606869e-05,
+      "loss": 0.1858,
+      "step": 138780
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.6980379560480744e-05,
+      "loss": 0.1835,
+      "step": 138790
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.69790411948928e-05,
+      "loss": 0.0211,
+      "step": 138800
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.6977702829304855e-05,
+      "loss": 0.0167,
+      "step": 138810
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.697636446371691e-05,
+      "loss": 0.0537,
+      "step": 138820
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.6975026098128965e-05,
+      "loss": 0.1178,
+      "step": 138830
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.697368773254102e-05,
+      "loss": 0.082,
+      "step": 138840
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.697234936695308e-05,
+      "loss": 0.0131,
+      "step": 138850
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.697101100136514e-05,
+      "loss": 0.0539,
+      "step": 138860
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.6969672635777194e-05,
+      "loss": 0.0077,
+      "step": 138870
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.696833427018925e-05,
+      "loss": 0.0517,
+      "step": 138880
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.6966995904601304e-05,
+      "loss": 0.0072,
+      "step": 138890
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.696565753901336e-05,
+      "loss": 0.0747,
+      "step": 138900
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.6964319173425415e-05,
+      "loss": 0.2774,
+      "step": 138910
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.696298080783747e-05,
+      "loss": 0.0982,
+      "step": 138920
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.6961642442249526e-05,
+      "loss": 0.0784,
+      "step": 138930
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.696030407666158e-05,
+      "loss": 0.0842,
+      "step": 138940
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.695896571107364e-05,
+      "loss": 0.1063,
+      "step": 138950
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.69576273454857e-05,
+      "loss": 0.1791,
+      "step": 138960
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.6956288979897754e-05,
+      "loss": 0.0206,
+      "step": 138970
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.695495061430981e-05,
+      "loss": 0.0051,
+      "step": 138980
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.6953612248721864e-05,
+      "loss": 0.1955,
+      "step": 138990
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.695227388313392e-05,
+      "loss": 0.1572,
+      "step": 139000
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.6950935517545975e-05,
+      "loss": 0.1796,
+      "step": 139010
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.694959715195803e-05,
+      "loss": 0.1969,
+      "step": 139020
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.6948258786370086e-05,
+      "loss": 0.1569,
+      "step": 139030
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.694692042078214e-05,
+      "loss": 0.1456,
+      "step": 139040
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.69455820551942e-05,
+      "loss": 0.0661,
+      "step": 139050
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.694424368960626e-05,
+      "loss": 0.1206,
+      "step": 139060
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.6942905324018314e-05,
+      "loss": 0.0868,
+      "step": 139070
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.694156695843037e-05,
+      "loss": 0.1338,
+      "step": 139080
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.6940228592842425e-05,
+      "loss": 0.0067,
+      "step": 139090
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.693889022725448e-05,
+      "loss": 0.1099,
+      "step": 139100
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.6937551861666535e-05,
+      "loss": 0.057,
+      "step": 139110
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.693621349607859e-05,
+      "loss": 0.0976,
+      "step": 139120
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.6934875130490646e-05,
+      "loss": 0.0539,
+      "step": 139130
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.69335367649027e-05,
+      "loss": 0.0988,
+      "step": 139140
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.6932198399314763e-05,
+      "loss": 0.1052,
+      "step": 139150
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.693086003372682e-05,
+      "loss": 0.0043,
+      "step": 139160
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.6929521668138874e-05,
+      "loss": 0.02,
+      "step": 139170
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.692818330255093e-05,
+      "loss": 0.1022,
+      "step": 139180
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.6926844936962985e-05,
+      "loss": 0.1679,
+      "step": 139190
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.692550657137504e-05,
+      "loss": 0.0842,
+      "step": 139200
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.6924168205787095e-05,
+      "loss": 0.2266,
+      "step": 139210
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.692282984019915e-05,
+      "loss": 0.066,
+      "step": 139220
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.6921491474611206e-05,
+      "loss": 0.1237,
+      "step": 139230
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.692015310902326e-05,
+      "loss": 0.1769,
+      "step": 139240
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.6918814743435324e-05,
+      "loss": 0.0051,
+      "step": 139250
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.691747637784738e-05,
+      "loss": 0.1111,
+      "step": 139260
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.6916138012259434e-05,
+      "loss": 0.4332,
+      "step": 139270
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.691479964667149e-05,
+      "loss": 0.1323,
+      "step": 139280
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.6913461281083545e-05,
+      "loss": 0.0834,
+      "step": 139290
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.69121229154956e-05,
+      "loss": 0.2807,
+      "step": 139300
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.6910784549907656e-05,
+      "loss": 0.0124,
+      "step": 139310
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.690944618431971e-05,
+      "loss": 0.0482,
+      "step": 139320
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.6908107818731766e-05,
+      "loss": 0.0202,
+      "step": 139330
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.690676945314382e-05,
+      "loss": 0.0978,
+      "step": 139340
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.6905431087555884e-05,
+      "loss": 0.4246,
+      "step": 139350
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.690409272196794e-05,
+      "loss": 0.1115,
+      "step": 139360
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.6902754356379994e-05,
+      "loss": 0.0549,
+      "step": 139370
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.690141599079205e-05,
+      "loss": 0.0061,
+      "step": 139380
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.6900077625204105e-05,
+      "loss": 0.12,
+      "step": 139390
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.689873925961616e-05,
+      "loss": 0.0318,
+      "step": 139400
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.6897400894028216e-05,
+      "loss": 0.005,
+      "step": 139410
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.689606252844027e-05,
+      "loss": 0.0363,
+      "step": 139420
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.6894724162852326e-05,
+      "loss": 0.0821,
+      "step": 139430
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.689338579726438e-05,
+      "loss": 0.1667,
+      "step": 139440
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.6892047431676444e-05,
+      "loss": 0.1312,
+      "step": 139450
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.68907090660885e-05,
+      "loss": 0.2539,
+      "step": 139460
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.6889370700500555e-05,
+      "loss": 0.09,
+      "step": 139470
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.688803233491261e-05,
+      "loss": 0.0447,
+      "step": 139480
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.6886693969324665e-05,
+      "loss": 0.0776,
+      "step": 139490
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.688535560373672e-05,
+      "loss": 0.0337,
+      "step": 139500
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.6884017238148776e-05,
+      "loss": 0.1377,
+      "step": 139510
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.688267887256083e-05,
+      "loss": 0.0232,
+      "step": 139520
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.6881340506972887e-05,
+      "loss": 0.1006,
+      "step": 139530
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.688000214138494e-05,
+      "loss": 0.1741,
+      "step": 139540
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.6878663775797004e-05,
+      "loss": 0.0378,
+      "step": 139550
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.687732541020906e-05,
+      "loss": 0.1543,
+      "step": 139560
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.6875987044621115e-05,
+      "loss": 0.0197,
+      "step": 139570
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.687464867903317e-05,
+      "loss": 0.2698,
+      "step": 139580
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.6873310313445225e-05,
+      "loss": 0.1642,
+      "step": 139590
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.687197194785728e-05,
+      "loss": 0.0503,
+      "step": 139600
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.6870633582269336e-05,
+      "loss": 0.1622,
+      "step": 139610
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.686929521668139e-05,
+      "loss": 0.2044,
+      "step": 139620
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.686795685109345e-05,
+      "loss": 0.0101,
+      "step": 139630
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.68666184855055e-05,
+      "loss": 0.0526,
+      "step": 139640
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.686528011991756e-05,
+      "loss": 0.0203,
+      "step": 139650
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.686394175432961e-05,
+      "loss": 0.1558,
+      "step": 139660
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.686260338874167e-05,
+      "loss": 0.0154,
+      "step": 139670
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.6861265023153723e-05,
+      "loss": 0.0516,
+      "step": 139680
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.685992665756578e-05,
+      "loss": 0.1125,
+      "step": 139690
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.6858588291977834e-05,
+      "loss": 0.0823,
+      "step": 139700
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.685724992638989e-05,
+      "loss": 0.0939,
+      "step": 139710
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.6855911560801945e-05,
+      "loss": 0.1195,
+      "step": 139720
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.685457319521401e-05,
+      "loss": 0.0906,
+      "step": 139730
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.685323482962606e-05,
+      "loss": 0.1456,
+      "step": 139740
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.685189646403812e-05,
+      "loss": 0.0589,
+      "step": 139750
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.685055809845017e-05,
+      "loss": 0.1247,
+      "step": 139760
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.684921973286223e-05,
+      "loss": 0.0101,
+      "step": 139770
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.6847881367274284e-05,
+      "loss": 0.01,
+      "step": 139780
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.684654300168634e-05,
+      "loss": 0.1576,
+      "step": 139790
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.6845204636098394e-05,
+      "loss": 0.1099,
+      "step": 139800
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.684386627051045e-05,
+      "loss": 0.0091,
+      "step": 139810
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.6842527904922505e-05,
+      "loss": 0.126,
+      "step": 139820
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.684118953933457e-05,
+      "loss": 0.1656,
+      "step": 139830
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.683985117374662e-05,
+      "loss": 0.0298,
+      "step": 139840
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.683851280815868e-05,
+      "loss": 0.1505,
+      "step": 139850
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.683717444257073e-05,
+      "loss": 0.0596,
+      "step": 139860
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.683583607698279e-05,
+      "loss": 0.0057,
+      "step": 139870
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.6834497711394844e-05,
+      "loss": 0.0099,
+      "step": 139880
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.68331593458069e-05,
+      "loss": 0.0618,
+      "step": 139890
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.6831820980218954e-05,
+      "loss": 0.1588,
+      "step": 139900
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.683048261463101e-05,
+      "loss": 0.0296,
+      "step": 139910
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.6829144249043065e-05,
+      "loss": 0.0272,
+      "step": 139920
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.682780588345513e-05,
+      "loss": 0.1472,
+      "step": 139930
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.682646751786718e-05,
+      "loss": 0.0302,
+      "step": 139940
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.682512915227924e-05,
+      "loss": 0.1397,
+      "step": 139950
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.682379078669129e-05,
+      "loss": 0.1072,
+      "step": 139960
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.682245242110335e-05,
+      "loss": 0.3071,
+      "step": 139970
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.6821114055515404e-05,
+      "loss": 0.0556,
+      "step": 139980
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.681977568992746e-05,
+      "loss": 0.1033,
+      "step": 139990
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.6818437324339515e-05,
+      "loss": 0.2471,
+      "step": 140000
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.681709895875157e-05,
+      "loss": 0.1687,
+      "step": 140010
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.6815760593163625e-05,
+      "loss": 0.0483,
+      "step": 140020
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.681442222757569e-05,
+      "loss": 0.3119,
+      "step": 140030
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.681308386198774e-05,
+      "loss": 0.0868,
+      "step": 140040
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.68117454963998e-05,
+      "loss": 0.1218,
+      "step": 140050
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.681040713081185e-05,
+      "loss": 0.011,
+      "step": 140060
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.680906876522391e-05,
+      "loss": 0.0717,
+      "step": 140070
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.6807730399635964e-05,
+      "loss": 0.1017,
+      "step": 140080
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.680639203404802e-05,
+      "loss": 0.1317,
+      "step": 140090
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.6805053668460075e-05,
+      "loss": 0.0121,
+      "step": 140100
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.680371530287213e-05,
+      "loss": 0.1132,
+      "step": 140110
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.6802376937284185e-05,
+      "loss": 0.1364,
+      "step": 140120
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.680103857169625e-05,
+      "loss": 0.1203,
+      "step": 140130
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.67997002061083e-05,
+      "loss": 0.0068,
+      "step": 140140
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.679836184052036e-05,
+      "loss": 0.0151,
+      "step": 140150
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.6797023474932413e-05,
+      "loss": 0.0497,
+      "step": 140160
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.679568510934447e-05,
+      "loss": 0.1106,
+      "step": 140170
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.6794346743756524e-05,
+      "loss": 0.036,
+      "step": 140180
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.679300837816858e-05,
+      "loss": 0.0707,
+      "step": 140190
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.6791670012580635e-05,
+      "loss": 0.0943,
+      "step": 140200
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.679033164699269e-05,
+      "loss": 0.0107,
+      "step": 140210
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.6788993281404746e-05,
+      "loss": 0.2366,
+      "step": 140220
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.678765491581681e-05,
+      "loss": 0.0692,
+      "step": 140230
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.678631655022886e-05,
+      "loss": 0.1197,
+      "step": 140240
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.678497818464092e-05,
+      "loss": 0.2243,
+      "step": 140250
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.6783639819052974e-05,
+      "loss": 0.0543,
+      "step": 140260
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.678230145346503e-05,
+      "loss": 0.2197,
+      "step": 140270
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.6780963087877084e-05,
+      "loss": 0.0299,
+      "step": 140280
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.677962472228914e-05,
+      "loss": 0.1022,
+      "step": 140290
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.6778286356701195e-05,
+      "loss": 0.1667,
+      "step": 140300
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.677694799111325e-05,
+      "loss": 0.0323,
+      "step": 140310
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.677560962552531e-05,
+      "loss": 0.0528,
+      "step": 140320
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.677427125993737e-05,
+      "loss": 0.0849,
+      "step": 140330
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.677293289434942e-05,
+      "loss": 0.0079,
+      "step": 140340
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.677159452876148e-05,
+      "loss": 0.0796,
+      "step": 140350
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.6770256163173534e-05,
+      "loss": 0.1251,
+      "step": 140360
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.676891779758559e-05,
+      "loss": 0.0539,
+      "step": 140370
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.6767579431997644e-05,
+      "loss": 0.1906,
+      "step": 140380
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.67662410664097e-05,
+      "loss": 0.0625,
+      "step": 140390
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.6764902700821755e-05,
+      "loss": 0.186,
+      "step": 140400
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.676356433523381e-05,
+      "loss": 0.0481,
+      "step": 140410
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.676222596964587e-05,
+      "loss": 0.2284,
+      "step": 140420
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.676088760405793e-05,
+      "loss": 0.0468,
+      "step": 140430
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.675954923846998e-05,
+      "loss": 0.0466,
+      "step": 140440
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.675821087288204e-05,
+      "loss": 0.1208,
+      "step": 140450
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.6756872507294094e-05,
+      "loss": 0.0679,
+      "step": 140460
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.675553414170615e-05,
+      "loss": 0.093,
+      "step": 140470
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.6754195776118205e-05,
+      "loss": 0.1183,
+      "step": 140480
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.675285741053026e-05,
+      "loss": 0.0732,
+      "step": 140490
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.6751519044942315e-05,
+      "loss": 0.1998,
+      "step": 140500
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.675018067935437e-05,
+      "loss": 0.2573,
+      "step": 140510
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.674884231376643e-05,
+      "loss": 0.0035,
+      "step": 140520
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.674750394817849e-05,
+      "loss": 0.2522,
+      "step": 140530
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.6746165582590543e-05,
+      "loss": 0.0706,
+      "step": 140540
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.67448272170026e-05,
+      "loss": 0.2156,
+      "step": 140550
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.6743488851414654e-05,
+      "loss": 0.0283,
+      "step": 140560
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.674215048582671e-05,
+      "loss": 0.2459,
+      "step": 140570
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.6740812120238765e-05,
+      "loss": 0.1015,
+      "step": 140580
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.673947375465082e-05,
+      "loss": 0.0443,
+      "step": 140590
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.6738135389062875e-05,
+      "loss": 0.0181,
+      "step": 140600
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.673679702347493e-05,
+      "loss": 0.0248,
+      "step": 140610
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.673545865788699e-05,
+      "loss": 0.1452,
+      "step": 140620
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.673412029229905e-05,
+      "loss": 0.0264,
+      "step": 140630
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.6732781926711104e-05,
+      "loss": 0.0399,
+      "step": 140640
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.673144356112316e-05,
+      "loss": 0.1087,
+      "step": 140650
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.6730105195535214e-05,
+      "loss": 0.2115,
+      "step": 140660
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.672876682994727e-05,
+      "loss": 0.0078,
+      "step": 140670
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.6727428464359325e-05,
+      "loss": 0.2343,
+      "step": 140680
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.672609009877138e-05,
+      "loss": 0.2144,
+      "step": 140690
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.6724751733183436e-05,
+      "loss": 0.0388,
+      "step": 140700
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.672341336759549e-05,
+      "loss": 0.0394,
+      "step": 140710
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.672207500200755e-05,
+      "loss": 0.0201,
+      "step": 140720
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.672073663641961e-05,
+      "loss": 0.0889,
+      "step": 140730
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.6719398270831664e-05,
+      "loss": 0.0015,
+      "step": 140740
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.671805990524372e-05,
+      "loss": 0.0598,
+      "step": 140750
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.6716721539655774e-05,
+      "loss": 0.059,
+      "step": 140760
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.671538317406783e-05,
+      "loss": 0.0149,
+      "step": 140770
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.6714044808479885e-05,
+      "loss": 0.042,
+      "step": 140780
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.671270644289194e-05,
+      "loss": 0.1628,
+      "step": 140790
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.6711368077303996e-05,
+      "loss": 0.0874,
+      "step": 140800
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.671002971171605e-05,
+      "loss": 0.0051,
+      "step": 140810
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.670869134612811e-05,
+      "loss": 0.0441,
+      "step": 140820
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.670735298054017e-05,
+      "loss": 0.057,
+      "step": 140830
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.6706014614952224e-05,
+      "loss": 0.0273,
+      "step": 140840
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.670467624936428e-05,
+      "loss": 0.0032,
+      "step": 140850
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.6703337883776335e-05,
+      "loss": 0.3239,
+      "step": 140860
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.670199951818839e-05,
+      "loss": 0.1028,
+      "step": 140870
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.6700661152600445e-05,
+      "loss": 0.0614,
+      "step": 140880
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.66993227870125e-05,
+      "loss": 0.0052,
+      "step": 140890
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.6697984421424556e-05,
+      "loss": 0.1115,
+      "step": 140900
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.669664605583661e-05,
+      "loss": 0.0252,
+      "step": 140910
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.669530769024867e-05,
+      "loss": 0.0465,
+      "step": 140920
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.669396932466073e-05,
+      "loss": 0.0367,
+      "step": 140930
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.6692630959072784e-05,
+      "loss": 0.083,
+      "step": 140940
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.669129259348484e-05,
+      "loss": 0.0784,
+      "step": 140950
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.6689954227896895e-05,
+      "loss": 0.1154,
+      "step": 140960
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.668861586230895e-05,
+      "loss": 0.1701,
+      "step": 140970
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.6687277496721005e-05,
+      "loss": 0.1291,
+      "step": 140980
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.668593913113306e-05,
+      "loss": 0.1439,
+      "step": 140990
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.6684600765545116e-05,
+      "loss": 0.0569,
+      "step": 141000
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.668326239995717e-05,
+      "loss": 0.0791,
+      "step": 141010
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.6681924034369234e-05,
+      "loss": 0.16,
+      "step": 141020
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.668058566878129e-05,
+      "loss": 0.0925,
+      "step": 141030
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.6679247303193344e-05,
+      "loss": 0.0443,
+      "step": 141040
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.66779089376054e-05,
+      "loss": 0.3466,
+      "step": 141050
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.6676570572017455e-05,
+      "loss": 0.1013,
+      "step": 141060
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.667523220642951e-05,
+      "loss": 0.2278,
+      "step": 141070
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.6673893840841566e-05,
+      "loss": 0.0467,
+      "step": 141080
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.667255547525362e-05,
+      "loss": 0.1639,
+      "step": 141090
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.6671217109665676e-05,
+      "loss": 0.1839,
+      "step": 141100
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.666987874407773e-05,
+      "loss": 0.0221,
+      "step": 141110
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.6668540378489794e-05,
+      "loss": 0.1428,
+      "step": 141120
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.666720201290185e-05,
+      "loss": 0.1114,
+      "step": 141130
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.6665863647313904e-05,
+      "loss": 0.1697,
+      "step": 141140
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": {
+        "accuracy": 0.9579644685335742
+      },
+      "eval_f1": {
+        "f1": 0.9577653922157884
+      },
+      "eval_loss": 0.19044803082942963,
+      "eval_precision": {
+        "precision": 0.9581933285912818
+      },
+      "eval_recall": {
+        "recall": 0.9578259452834421
+      },
+      "eval_runtime": 426.2483,
+      "eval_samples_per_second": 38.956,
+      "eval_steps_per_second": 19.479,
+      "step": 141142
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 415100,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 50,
+  "save_steps": 500,
+  "total_flos": 8.750034043996312e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}