diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18661 @@
+{
+  "best_metric": 0.32348209619522095,
+  "best_model_checkpoint": "/leonardo_work/IscrC_AGENT/PROFES2025/results/full_weighted_loss/oss_large/checkpoint-10333",
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 30999,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998064453692055e-05,
+      "loss": 0.6702,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9996128907384112e-05,
+      "loss": 0.6534,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9994193361076166e-05,
+      "loss": 0.6504,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999225781476822e-05,
+      "loss": 0.6176,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9990322268460273e-05,
+      "loss": 0.5519,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998838672215233e-05,
+      "loss": 0.5601,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9986451175844384e-05,
+      "loss": 0.6152,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9984515629536437e-05,
+      "loss": 0.5033,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9982580083228495e-05,
+      "loss": 0.515,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9980644536920548e-05,
+      "loss": 0.6009,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9978708990612602e-05,
+      "loss": 0.6259,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9976773444304655e-05,
+      "loss": 0.5908,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997483789799671e-05,
+      "loss": 0.504,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9972902351688766e-05,
+      "loss": 0.6135,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997096680538082e-05,
+      "loss": 0.5737,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9969031259072877e-05,
+      "loss": 0.5727,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996709571276493e-05,
+      "loss": 0.5528,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9965160166456984e-05,
+      "loss": 0.5856,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9963224620149038e-05,
+      "loss": 0.5259,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996128907384109e-05,
+      "loss": 0.6569,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.995935352753315e-05,
+      "loss": 0.5854,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9957417981225202e-05,
+      "loss": 0.5532,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9955482434917256e-05,
+      "loss": 0.5784,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9953546888609313e-05,
+      "loss": 0.4125,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9951611342301366e-05,
+      "loss": 0.5534,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9949675795993423e-05,
+      "loss": 0.4797,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9947740249685474e-05,
+      "loss": 0.5552,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994580470337753e-05,
+      "loss": 0.4894,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943869157069584e-05,
+      "loss": 0.5284,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941933610761638e-05,
+      "loss": 0.5146,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9939998064453695e-05,
+      "loss": 0.5452,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.993806251814575e-05,
+      "loss": 0.4934,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9936126971837802e-05,
+      "loss": 0.5338,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.993419142552986e-05,
+      "loss": 0.5827,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9932255879221913e-05,
+      "loss": 0.5584,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9930320332913966e-05,
+      "loss": 0.4878,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992838478660602e-05,
+      "loss": 0.5613,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9926449240298077e-05,
+      "loss": 0.5864,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992451369399013e-05,
+      "loss": 0.5593,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9922578147682184e-05,
+      "loss": 0.5047,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992064260137424e-05,
+      "loss": 0.5795,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991870705506629e-05,
+      "loss": 0.4874,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991677150875835e-05,
+      "loss": 0.5935,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9914835962450402e-05,
+      "loss": 0.4714,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991290041614246e-05,
+      "loss": 0.4332,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9910964869834513e-05,
+      "loss": 0.5486,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909029323526567e-05,
+      "loss": 0.5106,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907093777218624e-05,
+      "loss": 0.4095,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905158230910677e-05,
+      "loss": 0.5018,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990322268460273e-05,
+      "loss": 0.4767,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9901287138294784e-05,
+      "loss": 0.5725,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899351591986838e-05,
+      "loss": 0.5087,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897416045678895e-05,
+      "loss": 0.5413,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989548049937095e-05,
+      "loss": 0.5887,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893544953063006e-05,
+      "loss": 0.5282,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989160940675506e-05,
+      "loss": 0.6246,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9889673860447113e-05,
+      "loss": 0.4761,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9887738314139167e-05,
+      "loss": 0.4526,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988580276783122e-05,
+      "loss": 0.5858,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9883867221523277e-05,
+      "loss": 0.4968,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988193167521533e-05,
+      "loss": 0.5345,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9879996128907385e-05,
+      "loss": 0.4835,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987806058259944e-05,
+      "loss": 0.5663,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9876125036291495e-05,
+      "loss": 0.4832,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987418948998355e-05,
+      "loss": 0.5491,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9872253943675603e-05,
+      "loss": 0.401,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987031839736766e-05,
+      "loss": 0.4879,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9868382851059713e-05,
+      "loss": 0.4662,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9866447304751767e-05,
+      "loss": 0.4306,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9864511758443824e-05,
+      "loss": 0.4819,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9862576212135878e-05,
+      "loss": 0.6248,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.986064066582793e-05,
+      "loss": 0.4766,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9858705119519985e-05,
+      "loss": 0.4401,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9856769573212042e-05,
+      "loss": 0.5127,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9854834026904095e-05,
+      "loss": 0.5708,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.985289848059615e-05,
+      "loss": 0.5533,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9850962934288206e-05,
+      "loss": 0.5077,
+      "step": 770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984902738798026e-05,
+      "loss": 0.5217,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847091841672313e-05,
+      "loss": 0.4703,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845156295364367e-05,
+      "loss": 0.4868,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984322074905642e-05,
+      "loss": 0.4672,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9841285202748478e-05,
+      "loss": 0.3784,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.983934965644053e-05,
+      "loss": 0.5398,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.983741411013259e-05,
+      "loss": 0.4794,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9835478563824642e-05,
+      "loss": 0.4084,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9833543017516696e-05,
+      "loss": 0.471,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.983160747120875e-05,
+      "loss": 0.5172,
+      "step": 870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829671924900803e-05,
+      "loss": 0.4677,
+      "step": 880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982773637859286e-05,
+      "loss": 0.3877,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825800832284914e-05,
+      "loss": 0.5423,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823865285976967e-05,
+      "loss": 0.4725,
+      "step": 910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821929739669024e-05,
+      "loss": 0.3293,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819994193361078e-05,
+      "loss": 0.4186,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818058647053135e-05,
+      "loss": 0.5641,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816123100745185e-05,
+      "loss": 0.4329,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814187554437242e-05,
+      "loss": 0.514,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812252008129296e-05,
+      "loss": 0.5057,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981031646182135e-05,
+      "loss": 0.4797,
+      "step": 980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9808380915513406e-05,
+      "loss": 0.4232,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.980644536920546e-05,
+      "loss": 0.5797,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9804509822897514e-05,
+      "loss": 0.5265,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.980257427658957e-05,
+      "loss": 0.4998,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9800638730281624e-05,
+      "loss": 0.4603,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9798703183973678e-05,
+      "loss": 0.5379,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.979676763766573e-05,
+      "loss": 0.4301,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.979483209135779e-05,
+      "loss": 0.3812,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9792896545049842e-05,
+      "loss": 0.4182,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9790960998741896e-05,
+      "loss": 0.5267,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9789025452433953e-05,
+      "loss": 0.4483,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9787089906126003e-05,
+      "loss": 0.4548,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.978515435981806e-05,
+      "loss": 0.5065,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9783218813510114e-05,
+      "loss": 0.4703,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.978128326720217e-05,
+      "loss": 0.4476,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9779347720894225e-05,
+      "loss": 0.4187,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9777412174586278e-05,
+      "loss": 0.4101,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9775476628278335e-05,
+      "loss": 0.4369,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.977354108197039e-05,
+      "loss": 0.4845,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9771605535662442e-05,
+      "loss": 0.463,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9769669989354496e-05,
+      "loss": 0.4338,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.976773444304655e-05,
+      "loss": 0.3844,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9765798896738607e-05,
+      "loss": 0.5985,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.976386335043066e-05,
+      "loss": 0.4553,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9761927804122717e-05,
+      "loss": 0.4168,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.975999225781477e-05,
+      "loss": 0.4401,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9758056711506825e-05,
+      "loss": 0.4112,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.975612116519888e-05,
+      "loss": 0.3185,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9754185618890932e-05,
+      "loss": 0.5604,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.975225007258299e-05,
+      "loss": 0.4632,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9750314526275043e-05,
+      "loss": 0.3998,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9748378979967096e-05,
+      "loss": 0.4544,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9746443433659153e-05,
+      "loss": 0.49,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9744507887351207e-05,
+      "loss": 0.4104,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.974257234104326e-05,
+      "loss": 0.48,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9740636794735314e-05,
+      "loss": 0.435,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.973870124842737e-05,
+      "loss": 0.4436,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9736765702119425e-05,
+      "loss": 0.4434,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.973483015581148e-05,
+      "loss": 0.4331,
+      "step": 1370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9732894609503535e-05,
+      "loss": 0.3765,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.973095906319559e-05,
+      "loss": 0.4924,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9729023516887643e-05,
+      "loss": 0.5318,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9727087970579696e-05,
+      "loss": 0.4324,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9725152424271753e-05,
+      "loss": 0.4705,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9723216877963807e-05,
+      "loss": 0.4485,
+      "step": 1430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.972128133165586e-05,
+      "loss": 0.4056,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9719345785347918e-05,
+      "loss": 0.335,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.971741023903997e-05,
+      "loss": 0.4291,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9715474692732025e-05,
+      "loss": 0.3785,
+      "step": 1470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.971353914642408e-05,
+      "loss": 0.5012,
+      "step": 1480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9711603600116132e-05,
+      "loss": 0.4799,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.970966805380819e-05,
+      "loss": 0.5421,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9707732507500243e-05,
+      "loss": 0.4406,
+      "step": 1510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.97057969611923e-05,
+      "loss": 0.3503,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9703861414884354e-05,
+      "loss": 0.4586,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9701925868576407e-05,
+      "loss": 0.341,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9699990322268464e-05,
+      "loss": 0.4023,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9698054775960514e-05,
+      "loss": 0.3642,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.969611922965257e-05,
+      "loss": 0.3413,
+      "step": 1570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9694183683344625e-05,
+      "loss": 0.4084,
+      "step": 1580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.969224813703668e-05,
+      "loss": 0.4107,
+      "step": 1590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9690312590728736e-05,
+      "loss": 0.3871,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.968837704442079e-05,
+      "loss": 0.4353,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9686441498112846e-05,
+      "loss": 0.396,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9684505951804897e-05,
+      "loss": 0.4557,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9682570405496954e-05,
+      "loss": 0.4336,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9680634859189007e-05,
+      "loss": 0.3693,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.967869931288106e-05,
+      "loss": 0.3659,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9676763766573118e-05,
+      "loss": 0.4539,
+      "step": 1670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.967482822026517e-05,
+      "loss": 0.6164,
+      "step": 1680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9672892673957225e-05,
+      "loss": 0.4749,
+      "step": 1690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9670957127649282e-05,
+      "loss": 0.4269,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9669021581341336e-05,
+      "loss": 0.5115,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.966708603503339e-05,
+      "loss": 0.3295,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9665150488725443e-05,
+      "loss": 0.3294,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.96632149424175e-05,
+      "loss": 0.3672,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9661279396109554e-05,
+      "loss": 0.4018,
+      "step": 1750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9659343849801608e-05,
+      "loss": 0.4524,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9657408303493665e-05,
+      "loss": 0.3589,
+      "step": 1770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9655472757185715e-05,
+      "loss": 0.4565,
+      "step": 1780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9653537210877772e-05,
+      "loss": 0.4833,
+      "step": 1790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9651601664569825e-05,
+      "loss": 0.3666,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9649666118261882e-05,
+      "loss": 0.4113,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9647730571953936e-05,
+      "loss": 0.3988,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.964579502564599e-05,
+      "loss": 0.3072,
+      "step": 1830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9643859479338047e-05,
+      "loss": 0.4404,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.96419239330301e-05,
+      "loss": 0.3852,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9639988386722154e-05,
+      "loss": 0.3689,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9638052840414208e-05,
+      "loss": 0.2922,
+      "step": 1870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.963611729410626e-05,
+      "loss": 0.4405,
+      "step": 1880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.963418174779832e-05,
+      "loss": 0.4831,
+      "step": 1890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9632246201490372e-05,
+      "loss": 0.3095,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.963031065518243e-05,
+      "loss": 0.4492,
+      "step": 1910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9628375108874483e-05,
+      "loss": 0.4074,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9626439562566536e-05,
+      "loss": 0.3934,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.962450401625859e-05,
+      "loss": 0.5242,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9622568469950644e-05,
+      "loss": 0.4109,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.96206329236427e-05,
+      "loss": 0.3953,
+      "step": 1960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9618697377334754e-05,
+      "loss": 0.4802,
+      "step": 1970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9616761831026808e-05,
+      "loss": 0.4173,
+      "step": 1980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9614826284718865e-05,
+      "loss": 0.3493,
+      "step": 1990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.961289073841092e-05,
+      "loss": 0.4284,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9610955192102972e-05,
+      "loss": 0.281,
+      "step": 2010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9609019645795026e-05,
+      "loss": 0.4064,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9607084099487083e-05,
+      "loss": 0.4693,
+      "step": 2030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9605148553179136e-05,
+      "loss": 0.3675,
+      "step": 2040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.960321300687119e-05,
+      "loss": 0.434,
+      "step": 2050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9601277460563247e-05,
+      "loss": 0.4622,
+      "step": 2060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.95993419142553e-05,
+      "loss": 0.3378,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9597406367947354e-05,
+      "loss": 0.3537,
+      "step": 2080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9595470821639408e-05,
+      "loss": 0.3734,
+      "step": 2090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9593535275331465e-05,
+      "loss": 0.5988,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.959159972902352e-05,
+      "loss": 0.3383,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9589664182715572e-05,
+      "loss": 0.3184,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.958772863640763e-05,
+      "loss": 0.4947,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9585793090099683e-05,
+      "loss": 0.4162,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9583857543791737e-05,
+      "loss": 0.5044,
+      "step": 2150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.958192199748379e-05,
+      "loss": 0.363,
+      "step": 2160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9579986451175844e-05,
+      "loss": 0.5765,
+      "step": 2170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.95780509048679e-05,
+      "loss": 0.3375,
+      "step": 2180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9576115358559955e-05,
+      "loss": 0.3109,
+      "step": 2190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.957417981225201e-05,
+      "loss": 0.4322,
+      "step": 2200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9572244265944065e-05,
+      "loss": 0.3918,
+      "step": 2210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.957030871963612e-05,
+      "loss": 0.5206,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9568373173328176e-05,
+      "loss": 0.395,
+      "step": 2230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9566437627020226e-05,
+      "loss": 0.2894,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9564502080712283e-05,
+      "loss": 0.4633,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9562566534404337e-05,
+      "loss": 0.4803,
+      "step": 2260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.956063098809639e-05,
+      "loss": 0.4833,
+      "step": 2270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9558695441788447e-05,
+      "loss": 0.3793,
+      "step": 2280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.95567598954805e-05,
+      "loss": 0.3934,
+      "step": 2290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9554824349172558e-05,
+      "loss": 0.4539,
+      "step": 2300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9552888802864608e-05,
+      "loss": 0.3877,
+      "step": 2310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9550953256556665e-05,
+      "loss": 0.3713,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.954901771024872e-05,
+      "loss": 0.4035,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9547082163940773e-05,
+      "loss": 0.3546,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.954514661763283e-05,
+      "loss": 0.436,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9543211071324883e-05,
+      "loss": 0.5304,
+      "step": 2360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9541275525016937e-05,
+      "loss": 0.331,
+      "step": 2370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9539339978708994e-05,
+      "loss": 0.3991,
+      "step": 2380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9537404432401048e-05,
+      "loss": 0.3438,
+      "step": 2390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.95354688860931e-05,
+      "loss": 0.2611,
+      "step": 2400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9533533339785155e-05,
+      "loss": 0.4093,
+      "step": 2410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9531597793477212e-05,
+      "loss": 0.4731,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9529662247169265e-05,
+      "loss": 0.4533,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.952772670086132e-05,
+      "loss": 0.417,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9525791154553376e-05,
+      "loss": 0.4411,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9523855608245426e-05,
+      "loss": 0.3764,
+      "step": 2460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9521920061937483e-05,
+      "loss": 0.3516,
+      "step": 2470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9519984515629537e-05,
+      "loss": 0.4926,
+      "step": 2480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9518048969321594e-05,
+      "loss": 0.3807,
+      "step": 2490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9516113423013648e-05,
+      "loss": 0.4714,
+      "step": 2500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95141778767057e-05,
+      "loss": 0.3999,
+      "step": 2510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.951224233039776e-05,
+      "loss": 0.4388,
+      "step": 2520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9510306784089812e-05,
+      "loss": 0.298,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508371237781866e-05,
+      "loss": 0.3143,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950643569147392e-05,
+      "loss": 0.444,
+      "step": 2550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9504500145165973e-05,
+      "loss": 0.4594,
+      "step": 2560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950256459885803e-05,
+      "loss": 0.3178,
+      "step": 2570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9500629052550084e-05,
+      "loss": 0.505,
+      "step": 2580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949869350624214e-05,
+      "loss": 0.3597,
+      "step": 2590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9496757959934194e-05,
+      "loss": 0.3297,
+      "step": 2600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9494822413626248e-05,
+      "loss": 0.3853,
+      "step": 2610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.94928868673183e-05,
+      "loss": 0.4641,
+      "step": 2620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9490951321010355e-05,
+      "loss": 0.4039,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9489015774702412e-05,
+      "loss": 0.269,
+      "step": 2640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9487080228394466e-05,
+      "loss": 0.4259,
+      "step": 2650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948514468208652e-05,
+      "loss": 0.497,
+      "step": 2660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9483209135778576e-05,
+      "loss": 0.4025,
+      "step": 2670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948127358947063e-05,
+      "loss": 0.3003,
+      "step": 2680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9479338043162684e-05,
+      "loss": 0.4762,
+      "step": 2690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9477402496854737e-05,
+      "loss": 0.4207,
+      "step": 2700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9475466950546794e-05,
+      "loss": 0.3517,
+      "step": 2710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9473531404238848e-05,
+      "loss": 0.4658,
+      "step": 2720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94715958579309e-05,
+      "loss": 0.3262,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.946966031162296e-05,
+      "loss": 0.3176,
+      "step": 2740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9467724765315012e-05,
+      "loss": 0.5095,
+      "step": 2750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9465789219007066e-05,
+      "loss": 0.3552,
+      "step": 2760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.946385367269912e-05,
+      "loss": 0.3196,
+      "step": 2770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9461918126391177e-05,
+      "loss": 0.3575,
+      "step": 2780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.945998258008323e-05,
+      "loss": 0.4509,
+      "step": 2790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9458047033775284e-05,
+      "loss": 0.2998,
+      "step": 2800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.945611148746734e-05,
+      "loss": 0.4856,
+      "step": 2810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9454175941159395e-05,
+      "loss": 0.454,
+      "step": 2820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9452240394851448e-05,
+      "loss": 0.319,
+      "step": 2830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9450304848543502e-05,
+      "loss": 0.4077,
+      "step": 2840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9448369302235555e-05,
+      "loss": 0.3323,
+      "step": 2850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9446433755927612e-05,
+      "loss": 0.3237,
+      "step": 2860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9444498209619666e-05,
+      "loss": 0.4422,
+      "step": 2870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9442562663311723e-05,
+      "loss": 0.432,
+      "step": 2880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9440627117003777e-05,
+      "loss": 0.3756,
+      "step": 2890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.943869157069583e-05,
+      "loss": 0.361,
+      "step": 2900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9436756024387887e-05,
+      "loss": 0.245,
+      "step": 2910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9434820478079938e-05,
+      "loss": 0.2705,
+      "step": 2920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9432884931771995e-05,
+      "loss": 0.3702,
+      "step": 2930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.943094938546405e-05,
+      "loss": 0.4298,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9429013839156102e-05,
+      "loss": 0.4333,
+      "step": 2950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.942707829284816e-05,
+      "loss": 0.3286,
+      "step": 2960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9425142746540213e-05,
+      "loss": 0.4895,
+      "step": 2970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.942320720023227e-05,
+      "loss": 0.3283,
+      "step": 2980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.942127165392432e-05,
+      "loss": 0.4404,
+      "step": 2990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9419336107616377e-05,
+      "loss": 0.3959,
+      "step": 3000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.941740056130843e-05,
+      "loss": 0.4659,
+      "step": 3010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9415465015000484e-05,
+      "loss": 0.3416,
+      "step": 3020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.941352946869254e-05,
+      "loss": 0.3768,
+      "step": 3030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9411593922384595e-05,
+      "loss": 0.3742,
+      "step": 3040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.940965837607665e-05,
+      "loss": 0.3142,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9407722829768706e-05,
+      "loss": 0.3651,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.940578728346076e-05,
+      "loss": 0.4814,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9403851737152813e-05,
+      "loss": 0.3414,
+      "step": 3080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9401916190844866e-05,
+      "loss": 0.3209,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9399980644536923e-05,
+      "loss": 0.3836,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9398045098228977e-05,
+      "loss": 0.3541,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.939610955192103e-05,
+      "loss": 0.3712,
+      "step": 3120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9394174005613088e-05,
+      "loss": 0.3334,
+      "step": 3130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9392238459305138e-05,
+      "loss": 0.3763,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9390302912997195e-05,
+      "loss": 0.3172,
+      "step": 3150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.938836736668925e-05,
+      "loss": 0.4085,
+      "step": 3160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9386431820381306e-05,
+      "loss": 0.3642,
+      "step": 3170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.938449627407336e-05,
+      "loss": 0.3908,
+      "step": 3180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9382560727765413e-05,
+      "loss": 0.4342,
+      "step": 3190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.938062518145747e-05,
+      "loss": 0.4296,
+      "step": 3200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9378689635149524e-05,
+      "loss": 0.3587,
+      "step": 3210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9376754088841577e-05,
+      "loss": 0.3221,
+      "step": 3220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.937481854253363e-05,
+      "loss": 0.3671,
+      "step": 3230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9372882996225685e-05,
+      "loss": 0.2918,
+      "step": 3240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.937094744991774e-05,
+      "loss": 0.4048,
+      "step": 3250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9369011903609795e-05,
+      "loss": 0.3566,
+      "step": 3260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9367076357301852e-05,
+      "loss": 0.4537,
+      "step": 3270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9365140810993906e-05,
+      "loss": 0.3226,
+      "step": 3280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.936320526468596e-05,
+      "loss": 0.4957,
+      "step": 3290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9361269718378013e-05,
+      "loss": 0.3721,
+      "step": 3300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9359334172070067e-05,
+      "loss": 0.3961,
+      "step": 3310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9357398625762124e-05,
+      "loss": 0.4128,
+      "step": 3320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9355463079454177e-05,
+      "loss": 0.268,
+      "step": 3330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.935352753314623e-05,
+      "loss": 0.3519,
+      "step": 3340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9351591986838288e-05,
+      "loss": 0.329,
+      "step": 3350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.934965644053034e-05,
+      "loss": 0.5037,
+      "step": 3360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9347720894222395e-05,
+      "loss": 0.3974,
+      "step": 3370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.934578534791445e-05,
+      "loss": 0.3695,
+      "step": 3380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9343849801606506e-05,
+      "loss": 0.4009,
+      "step": 3390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.934191425529856e-05,
+      "loss": 0.3532,
+      "step": 3400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9339978708990613e-05,
+      "loss": 0.3979,
+      "step": 3410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.933804316268267e-05,
+      "loss": 0.3243,
+      "step": 3420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9336107616374724e-05,
+      "loss": 0.315,
+      "step": 3430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9334172070066778e-05,
+      "loss": 0.4386,
+      "step": 3440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.933223652375883e-05,
+      "loss": 0.5655,
+      "step": 3450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9330300977450888e-05,
+      "loss": 0.4439,
+      "step": 3460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9328365431142942e-05,
+      "loss": 0.3392,
+      "step": 3470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9326429884834995e-05,
+      "loss": 0.4329,
+      "step": 3480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9324494338527053e-05,
+      "loss": 0.4039,
+      "step": 3490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9322558792219106e-05,
+      "loss": 0.5243,
+      "step": 3500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.932062324591116e-05,
+      "loss": 0.3983,
+      "step": 3510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9318687699603213e-05,
+      "loss": 0.3782,
+      "step": 3520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9316752153295267e-05,
+      "loss": 0.4054,
+      "step": 3530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9314816606987324e-05,
+      "loss": 0.4127,
+      "step": 3540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9312881060679378e-05,
+      "loss": 0.3722,
+      "step": 3550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9310945514371435e-05,
+      "loss": 0.551,
+      "step": 3560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.930900996806349e-05,
+      "loss": 0.2479,
+      "step": 3570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9307074421755542e-05,
+      "loss": 0.5031,
+      "step": 3580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.93051388754476e-05,
+      "loss": 0.3735,
+      "step": 3590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.930320332913965e-05,
+      "loss": 0.2578,
+      "step": 3600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9301267782831706e-05,
+      "loss": 0.3022,
+      "step": 3610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.929933223652376e-05,
+      "loss": 0.4937,
+      "step": 3620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9297396690215814e-05,
+      "loss": 0.4229,
+      "step": 3630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.929546114390787e-05,
+      "loss": 0.3849,
+      "step": 3640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9293525597599924e-05,
+      "loss": 0.3684,
+      "step": 3650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.929159005129198e-05,
+      "loss": 0.4077,
+      "step": 3660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.928965450498403e-05,
+      "loss": 0.4068,
+      "step": 3670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.928771895867609e-05,
+      "loss": 0.3478,
+      "step": 3680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9285783412368142e-05,
+      "loss": 0.34,
+      "step": 3690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9283847866060196e-05,
+      "loss": 0.3703,
+      "step": 3700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9281912319752253e-05,
+      "loss": 0.4306,
+      "step": 3710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9279976773444306e-05,
+      "loss": 0.2496,
+      "step": 3720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.927804122713636e-05,
+      "loss": 0.5016,
+      "step": 3730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9276105680828417e-05,
+      "loss": 0.3561,
+      "step": 3740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9274170134520467e-05,
+      "loss": 0.3425,
+      "step": 3750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9272234588212524e-05,
+      "loss": 0.4288,
+      "step": 3760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9270299041904578e-05,
+      "loss": 0.4374,
+      "step": 3770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9268363495596635e-05,
+      "loss": 0.3769,
+      "step": 3780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.926642794928869e-05,
+      "loss": 0.4381,
+      "step": 3790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9264492402980742e-05,
+      "loss": 0.2633,
+      "step": 3800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.92625568566728e-05,
+      "loss": 0.405,
+      "step": 3810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9260621310364853e-05,
+      "loss": 0.2709,
+      "step": 3820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9258685764056907e-05,
+      "loss": 0.5473,
+      "step": 3830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.925675021774896e-05,
+      "loss": 0.4159,
+      "step": 3840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9254814671441014e-05,
+      "loss": 0.3858,
+      "step": 3850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.925287912513307e-05,
+      "loss": 0.4223,
+      "step": 3860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9250943578825125e-05,
+      "loss": 0.3838,
+      "step": 3870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.924900803251718e-05,
+      "loss": 0.3273,
+      "step": 3880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9247072486209235e-05,
+      "loss": 0.3045,
+      "step": 3890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.924513693990129e-05,
+      "loss": 0.3945,
+      "step": 3900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9243201393593342e-05,
+      "loss": 0.3587,
+      "step": 3910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9241265847285396e-05,
+      "loss": 0.3258,
+      "step": 3920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9239330300977453e-05,
+      "loss": 0.4742,
+      "step": 3930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9237394754669507e-05,
+      "loss": 0.3207,
+      "step": 3940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9235459208361564e-05,
+      "loss": 0.3781,
+      "step": 3950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9233523662053617e-05,
+      "loss": 0.4528,
+      "step": 3960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.923158811574567e-05,
+      "loss": 0.3518,
+      "step": 3970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9229652569437725e-05,
+      "loss": 0.257,
+      "step": 3980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.922771702312978e-05,
+      "loss": 0.3939,
+      "step": 3990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9225781476821835e-05,
+      "loss": 0.4746,
+      "step": 4000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.922384593051389e-05,
+      "loss": 0.3215,
+      "step": 4010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9221910384205943e-05,
+      "loss": 0.3318,
+      "step": 4020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9219974837898e-05,
+      "loss": 0.4439,
+      "step": 4030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9218039291590053e-05,
+      "loss": 0.3374,
+      "step": 4040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9216103745282107e-05,
+      "loss": 0.3451,
+      "step": 4050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.921416819897416e-05,
+      "loss": 0.3094,
+      "step": 4060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9212232652666218e-05,
+      "loss": 0.2938,
+      "step": 4070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.921029710635827e-05,
+      "loss": 0.3519,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9208361560050325e-05,
+      "loss": 0.4223,
+      "step": 4090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9206426013742382e-05,
+      "loss": 0.3957,
+      "step": 4100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9204490467434436e-05,
+      "loss": 0.4627,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.920255492112649e-05,
+      "loss": 0.3205,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9200619374818543e-05,
+      "loss": 0.3666,
+      "step": 4130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9198683828510596e-05,
+      "loss": 0.3097,
+      "step": 4140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9196748282202653e-05,
+      "loss": 0.5067,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9194812735894707e-05,
+      "loss": 0.3477,
+      "step": 4160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9192877189586764e-05,
+      "loss": 0.4193,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9190941643278818e-05,
+      "loss": 0.2697,
+      "step": 4180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.918900609697087e-05,
+      "loss": 0.3975,
+      "step": 4190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9187070550662925e-05,
+      "loss": 0.3699,
+      "step": 4200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.918513500435498e-05,
+      "loss": 0.3219,
+      "step": 4210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9183199458047036e-05,
+      "loss": 0.4122,
+      "step": 4220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.918126391173909e-05,
+      "loss": 0.2784,
+      "step": 4230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9179328365431143e-05,
+      "loss": 0.3583,
+      "step": 4240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.91773928191232e-05,
+      "loss": 0.3288,
+      "step": 4250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9175457272815254e-05,
+      "loss": 0.4302,
+      "step": 4260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.917352172650731e-05,
+      "loss": 0.2683,
+      "step": 4270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.917158618019936e-05,
+      "loss": 0.3051,
+      "step": 4280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9169650633891418e-05,
+      "loss": 0.4471,
+      "step": 4290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.916771508758347e-05,
+      "loss": 0.4197,
+      "step": 4300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9165779541275525e-05,
+      "loss": 0.317,
+      "step": 4310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9163843994967582e-05,
+      "loss": 0.4387,
+      "step": 4320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9161908448659636e-05,
+      "loss": 0.3962,
+      "step": 4330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9159972902351693e-05,
+      "loss": 0.4048,
+      "step": 4340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9158037356043743e-05,
+      "loss": 0.2978,
+      "step": 4350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.91561018097358e-05,
+      "loss": 0.282,
+      "step": 4360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9154166263427854e-05,
+      "loss": 0.4118,
+      "step": 4370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9152230717119907e-05,
+      "loss": 0.4749,
+      "step": 4380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9150295170811964e-05,
+      "loss": 0.3257,
+      "step": 4390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9148359624504018e-05,
+      "loss": 0.262,
+      "step": 4400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.914642407819607e-05,
+      "loss": 0.4256,
+      "step": 4410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.914448853188813e-05,
+      "loss": 0.3689,
+      "step": 4420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.914255298558018e-05,
+      "loss": 0.348,
+      "step": 4430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9140617439272236e-05,
+      "loss": 0.3387,
+      "step": 4440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.913868189296429e-05,
+      "loss": 0.3537,
+      "step": 4450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9136746346656347e-05,
+      "loss": 0.397,
+      "step": 4460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.91348108003484e-05,
+      "loss": 0.4115,
+      "step": 4470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9132875254040454e-05,
+      "loss": 0.2933,
+      "step": 4480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.913093970773251e-05,
+      "loss": 0.3539,
+      "step": 4490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9129004161424565e-05,
+      "loss": 0.292,
+      "step": 4500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9127068615116618e-05,
+      "loss": 0.4652,
+      "step": 4510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9125133068808672e-05,
+      "loss": 0.4474,
+      "step": 4520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9123197522500725e-05,
+      "loss": 0.4126,
+      "step": 4530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9121261976192782e-05,
+      "loss": 0.4463,
+      "step": 4540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9119326429884836e-05,
+      "loss": 0.3133,
+      "step": 4550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9117390883576893e-05,
+      "loss": 0.3686,
+      "step": 4560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9115455337268947e-05,
+      "loss": 0.363,
+      "step": 4570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9113519790961e-05,
+      "loss": 0.4454,
+      "step": 4580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9111584244653054e-05,
+      "loss": 0.3419,
+      "step": 4590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9109648698345108e-05,
+      "loss": 0.3417,
+      "step": 4600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9107713152037165e-05,
+      "loss": 0.4106,
+      "step": 4610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.910577760572922e-05,
+      "loss": 0.332,
+      "step": 4620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9103842059421272e-05,
+      "loss": 0.4436,
+      "step": 4630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.910190651311333e-05,
+      "loss": 0.3219,
+      "step": 4640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9099970966805383e-05,
+      "loss": 0.4151,
+      "step": 4650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9098035420497436e-05,
+      "loss": 0.3221,
+      "step": 4660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.909609987418949e-05,
+      "loss": 0.367,
+      "step": 4670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9094164327881547e-05,
+      "loss": 0.3652,
+      "step": 4680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.90922287815736e-05,
+      "loss": 0.4228,
+      "step": 4690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9090293235265654e-05,
+      "loss": 0.3582,
+      "step": 4700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.908835768895771e-05,
+      "loss": 0.4286,
+      "step": 4710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9086422142649765e-05,
+      "loss": 0.3196,
+      "step": 4720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.908448659634182e-05,
+      "loss": 0.3407,
+      "step": 4730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9082551050033872e-05,
+      "loss": 0.3364,
+      "step": 4740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.908061550372593e-05,
+      "loss": 0.2977,
+      "step": 4750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9078679957417983e-05,
+      "loss": 0.3058,
+      "step": 4760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9076744411110036e-05,
+      "loss": 0.3094,
+      "step": 4770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9074808864802093e-05,
+      "loss": 0.3791,
+      "step": 4780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9072873318494147e-05,
+      "loss": 0.3833,
+      "step": 4790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.90709377721862e-05,
+      "loss": 0.4535,
+      "step": 4800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9069002225878254e-05,
+      "loss": 0.323,
+      "step": 4810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9067066679570308e-05,
+      "loss": 0.4877,
+      "step": 4820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9065131133262365e-05,
+      "loss": 0.3678,
+      "step": 4830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.906319558695442e-05,
+      "loss": 0.3449,
+      "step": 4840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9061260040646476e-05,
+      "loss": 0.3734,
+      "step": 4850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.905932449433853e-05,
+      "loss": 0.3421,
+      "step": 4860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9057388948030583e-05,
+      "loss": 0.3678,
+      "step": 4870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9055453401722637e-05,
+      "loss": 0.359,
+      "step": 4880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.905351785541469e-05,
+      "loss": 0.4503,
+      "step": 4890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9051582309106747e-05,
+      "loss": 0.3667,
+      "step": 4900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90496467627988e-05,
+      "loss": 0.3053,
+      "step": 4910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9047711216490855e-05,
+      "loss": 0.3386,
+      "step": 4920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.904577567018291e-05,
+      "loss": 0.3816,
+      "step": 4930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9043840123874965e-05,
+      "loss": 0.351,
+      "step": 4940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9041904577567022e-05,
+      "loss": 0.3716,
+      "step": 4950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9039969031259072e-05,
+      "loss": 0.285,
+      "step": 4960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.903803348495113e-05,
+      "loss": 0.2784,
+      "step": 4970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9036097938643183e-05,
+      "loss": 0.4255,
+      "step": 4980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9034162392335237e-05,
+      "loss": 0.399,
+      "step": 4990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9032226846027294e-05,
+      "loss": 0.3096,
+      "step": 5000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9030291299719347e-05,
+      "loss": 0.3515,
+      "step": 5010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90283557534114e-05,
+      "loss": 0.3837,
+      "step": 5020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9026420207103458e-05,
+      "loss": 0.2423,
+      "step": 5030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9024484660795512e-05,
+      "loss": 0.3709,
+      "step": 5040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9022549114487565e-05,
+      "loss": 0.423,
+      "step": 5050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.902061356817962e-05,
+      "loss": 0.3123,
+      "step": 5060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9018678021871676e-05,
+      "loss": 0.3179,
+      "step": 5070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.901674247556373e-05,
+      "loss": 0.3199,
+      "step": 5080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9014806929255783e-05,
+      "loss": 0.2427,
+      "step": 5090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.901287138294784e-05,
+      "loss": 0.3378,
+      "step": 5100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.901093583663989e-05,
+      "loss": 0.434,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9009000290331948e-05,
+      "loss": 0.3113,
+      "step": 5120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9007064744024e-05,
+      "loss": 0.2909,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9005129197716058e-05,
+      "loss": 0.3994,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9003193651408112e-05,
+      "loss": 0.3406,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9001258105100166e-05,
+      "loss": 0.4876,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8999322558792223e-05,
+      "loss": 0.237,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8997387012484276e-05,
+      "loss": 0.3439,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.899545146617633e-05,
+      "loss": 0.3168,
+      "step": 5190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8993515919868383e-05,
+      "loss": 0.4082,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8991580373560437e-05,
+      "loss": 0.4286,
+      "step": 5210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8989644827252494e-05,
+      "loss": 0.4109,
+      "step": 5220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987709280944548e-05,
+      "loss": 0.3401,
+      "step": 5230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985773734636605e-05,
+      "loss": 0.3369,
+      "step": 5240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898383818832866e-05,
+      "loss": 0.4073,
+      "step": 5250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981902642020712e-05,
+      "loss": 0.4037,
+      "step": 5260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979967095712766e-05,
+      "loss": 0.4549,
+      "step": 5270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897803154940482e-05,
+      "loss": 0.3165,
+      "step": 5280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976096003096876e-05,
+      "loss": 0.3346,
+      "step": 5290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897416045678893e-05,
+      "loss": 0.4709,
+      "step": 5300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972224910480984e-05,
+      "loss": 0.3583,
+      "step": 5310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897028936417304e-05,
+      "loss": 0.3309,
+      "step": 5320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8968353817865094e-05,
+      "loss": 0.3607,
+      "step": 5330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8966418271557148e-05,
+      "loss": 0.2926,
+      "step": 5340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89644827252492e-05,
+      "loss": 0.2346,
+      "step": 5350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896254717894126e-05,
+      "loss": 0.56,
+      "step": 5360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960611632633312e-05,
+      "loss": 0.3649,
+      "step": 5370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8958676086325366e-05,
+      "loss": 0.2929,
+      "step": 5380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956740540017423e-05,
+      "loss": 0.3738,
+      "step": 5390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8954804993709476e-05,
+      "loss": 0.4704,
+      "step": 5400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895286944740153e-05,
+      "loss": 0.4452,
+      "step": 5410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950933901093584e-05,
+      "loss": 0.2478,
+      "step": 5420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.894899835478564e-05,
+      "loss": 0.2551,
+      "step": 5430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8947062808477694e-05,
+      "loss": 0.4916,
+      "step": 5440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8945127262169748e-05,
+      "loss": 0.2724,
+      "step": 5450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8943191715861805e-05,
+      "loss": 0.3297,
+      "step": 5460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.894125616955386e-05,
+      "loss": 0.3322,
+      "step": 5470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8939320623245912e-05,
+      "loss": 0.3209,
+      "step": 5480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8937385076937966e-05,
+      "loss": 0.3994,
+      "step": 5490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.893544953063002e-05,
+      "loss": 0.3907,
+      "step": 5500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8933513984322077e-05,
+      "loss": 0.2728,
+      "step": 5510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.893157843801413e-05,
+      "loss": 0.3397,
+      "step": 5520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8929642891706187e-05,
+      "loss": 0.3714,
+      "step": 5530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.892770734539824e-05,
+      "loss": 0.2728,
+      "step": 5540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8925771799090295e-05,
+      "loss": 0.3526,
+      "step": 5550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8923836252782348e-05,
+      "loss": 0.4678,
+      "step": 5560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8921900706474402e-05,
+      "loss": 0.2778,
+      "step": 5570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.891996516016646e-05,
+      "loss": 0.445,
+      "step": 5580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8918029613858512e-05,
+      "loss": 0.3914,
+      "step": 5590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8916094067550566e-05,
+      "loss": 0.3185,
+      "step": 5600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8914158521242623e-05,
+      "loss": 0.3925,
+      "step": 5610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8912222974934677e-05,
+      "loss": 0.3708,
+      "step": 5620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8910287428626734e-05,
+      "loss": 0.3551,
+      "step": 5630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8908351882318784e-05,
+      "loss": 0.2759,
+      "step": 5640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.890641633601084e-05,
+      "loss": 0.3111,
+      "step": 5650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8904480789702895e-05,
+      "loss": 0.4198,
+      "step": 5660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.890254524339495e-05,
+      "loss": 0.277,
+      "step": 5670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8900609697087005e-05,
+      "loss": 0.3026,
+      "step": 5680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.889867415077906e-05,
+      "loss": 0.2805,
+      "step": 5690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8896738604471113e-05,
+      "loss": 0.3731,
+      "step": 5700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.889480305816317e-05,
+      "loss": 0.4065,
+      "step": 5710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8892867511855223e-05,
+      "loss": 0.3705,
+      "step": 5720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8890931965547277e-05,
+      "loss": 0.3406,
+      "step": 5730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.888899641923933e-05,
+      "loss": 0.3375,
+      "step": 5740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8887060872931388e-05,
+      "loss": 0.4255,
+      "step": 5750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.888512532662344e-05,
+      "loss": 0.3831,
+      "step": 5760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8883189780315495e-05,
+      "loss": 0.2839,
+      "step": 5770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8881254234007552e-05,
+      "loss": 0.5226,
+      "step": 5780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8879318687699602e-05,
+      "loss": 0.3575,
+      "step": 5790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.887738314139166e-05,
+      "loss": 0.3221,
+      "step": 5800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8875447595083713e-05,
+      "loss": 0.3982,
+      "step": 5810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.887351204877577e-05,
+      "loss": 0.2864,
+      "step": 5820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8871576502467823e-05,
+      "loss": 0.3452,
+      "step": 5830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8869640956159877e-05,
+      "loss": 0.3075,
+      "step": 5840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8867705409851934e-05,
+      "loss": 0.2773,
+      "step": 5850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8865769863543988e-05,
+      "loss": 0.3656,
+      "step": 5860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.886383431723604e-05,
+      "loss": 0.4547,
+      "step": 5870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8861898770928095e-05,
+      "loss": 0.3325,
+      "step": 5880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.885996322462015e-05,
+      "loss": 0.3701,
+      "step": 5890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8858027678312206e-05,
+      "loss": 0.3877,
+      "step": 5900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.885609213200426e-05,
+      "loss": 0.3257,
+      "step": 5910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8854156585696316e-05,
+      "loss": 0.3113,
+      "step": 5920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.885222103938837e-05,
+      "loss": 0.3695,
+      "step": 5930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8850285493080424e-05,
+      "loss": 0.3431,
+      "step": 5940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8848349946772477e-05,
+      "loss": 0.3997,
+      "step": 5950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.884641440046453e-05,
+      "loss": 0.3135,
+      "step": 5960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8844478854156588e-05,
+      "loss": 0.3042,
+      "step": 5970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.884254330784864e-05,
+      "loss": 0.3763,
+      "step": 5980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8840607761540695e-05,
+      "loss": 0.3337,
+      "step": 5990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8838672215232752e-05,
+      "loss": 0.4308,
+      "step": 6000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8836736668924806e-05,
+      "loss": 0.3014,
+      "step": 6010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.883480112261686e-05,
+      "loss": 0.3657,
+      "step": 6020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8832865576308913e-05,
+      "loss": 0.3828,
+      "step": 6030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.883093003000097e-05,
+      "loss": 0.4757,
+      "step": 6040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8828994483693024e-05,
+      "loss": 0.422,
+      "step": 6050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8827058937385077e-05,
+      "loss": 0.5819,
+      "step": 6060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8825123391077134e-05,
+      "loss": 0.2975,
+      "step": 6070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8823187844769188e-05,
+      "loss": 0.3541,
+      "step": 6080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8821252298461242e-05,
+      "loss": 0.4085,
+      "step": 6090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8819316752153295e-05,
+      "loss": 0.265,
+      "step": 6100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8817381205845352e-05,
+      "loss": 0.4095,
+      "step": 6110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8815445659537406e-05,
+      "loss": 0.3463,
+      "step": 6120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.881351011322946e-05,
+      "loss": 0.3064,
+      "step": 6130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8811574566921517e-05,
+      "loss": 0.3216,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.880963902061357e-05,
+      "loss": 0.4792,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8807703474305624e-05,
+      "loss": 0.33,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8805767927997678e-05,
+      "loss": 0.3185,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.880383238168973e-05,
+      "loss": 0.2732,
+      "step": 6180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8801896835381788e-05,
+      "loss": 0.3093,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8799961289073842e-05,
+      "loss": 0.352,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.87980257427659e-05,
+      "loss": 0.5289,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8796090196457953e-05,
+      "loss": 0.316,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8794154650150006e-05,
+      "loss": 0.2334,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8792219103842063e-05,
+      "loss": 0.3564,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8790283557534113e-05,
+      "loss": 0.3761,
+      "step": 6250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.878834801122617e-05,
+      "loss": 0.2734,
+      "step": 6260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8786412464918224e-05,
+      "loss": 0.3142,
+      "step": 6270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8784476918610278e-05,
+      "loss": 0.3894,
+      "step": 6280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8782541372302335e-05,
+      "loss": 0.3341,
+      "step": 6290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.878060582599439e-05,
+      "loss": 0.4785,
+      "step": 6300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8778670279686445e-05,
+      "loss": 0.2977,
+      "step": 6310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8776734733378496e-05,
+      "loss": 0.3926,
+      "step": 6320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8774799187070553e-05,
+      "loss": 0.3539,
+      "step": 6330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8772863640762606e-05,
+      "loss": 0.4025,
+      "step": 6340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.877092809445466e-05,
+      "loss": 0.3544,
+      "step": 6350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8768992548146717e-05,
+      "loss": 0.2944,
+      "step": 6360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.876705700183877e-05,
+      "loss": 0.352,
+      "step": 6370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8765121455530824e-05,
+      "loss": 0.5195,
+      "step": 6380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.876318590922288e-05,
+      "loss": 0.249,
+      "step": 6390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8761250362914935e-05,
+      "loss": 0.2658,
+      "step": 6400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.875931481660699e-05,
+      "loss": 0.4948,
+      "step": 6410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8757379270299042e-05,
+      "loss": 0.3107,
+      "step": 6420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.87554437239911e-05,
+      "loss": 0.3666,
+      "step": 6430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8753508177683153e-05,
+      "loss": 0.4307,
+      "step": 6440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8751572631375206e-05,
+      "loss": 0.3087,
+      "step": 6450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8749637085067263e-05,
+      "loss": 0.2977,
+      "step": 6460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8747701538759314e-05,
+      "loss": 0.3428,
+      "step": 6470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.874576599245137e-05,
+      "loss": 0.46,
+      "step": 6480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8743830446143424e-05,
+      "loss": 0.4305,
+      "step": 6490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.874189489983548e-05,
+      "loss": 0.3785,
+      "step": 6500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8739959353527535e-05,
+      "loss": 0.3144,
+      "step": 6510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.873802380721959e-05,
+      "loss": 0.4196,
+      "step": 6520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8736088260911646e-05,
+      "loss": 0.4398,
+      "step": 6530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.87341527146037e-05,
+      "loss": 0.3285,
+      "step": 6540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8732217168295753e-05,
+      "loss": 0.3091,
+      "step": 6550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8730281621987807e-05,
+      "loss": 0.4289,
+      "step": 6560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.872834607567986e-05,
+      "loss": 0.2996,
+      "step": 6570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8726410529371917e-05,
+      "loss": 0.2093,
+      "step": 6580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.872447498306397e-05,
+      "loss": 0.4607,
+      "step": 6590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8722539436756028e-05,
+      "loss": 0.3703,
+      "step": 6600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.872060389044808e-05,
+      "loss": 0.3473,
+      "step": 6610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8718668344140135e-05,
+      "loss": 0.3047,
+      "step": 6620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.871673279783219e-05,
+      "loss": 0.401,
+      "step": 6630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8714797251524242e-05,
+      "loss": 0.4218,
+      "step": 6640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.87128617052163e-05,
+      "loss": 0.2586,
+      "step": 6650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8710926158908353e-05,
+      "loss": 0.2766,
+      "step": 6660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8708990612600407e-05,
+      "loss": 0.2839,
+      "step": 6670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8707055066292464e-05,
+      "loss": 0.3874,
+      "step": 6680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8705119519984517e-05,
+      "loss": 0.408,
+      "step": 6690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.870318397367657e-05,
+      "loss": 0.3257,
+      "step": 6700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8701248427368625e-05,
+      "loss": 0.2693,
+      "step": 6710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8699312881060682e-05,
+      "loss": 0.471,
+      "step": 6720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8697377334752735e-05,
+      "loss": 0.3936,
+      "step": 6730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.869544178844479e-05,
+      "loss": 0.3143,
+      "step": 6740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8693506242136846e-05,
+      "loss": 0.3663,
+      "step": 6750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.86915706958289e-05,
+      "loss": 0.2528,
+      "step": 6760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8689635149520953e-05,
+      "loss": 0.3174,
+      "step": 6770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8687699603213007e-05,
+      "loss": 0.3816,
+      "step": 6780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8685764056905064e-05,
+      "loss": 0.3816,
+      "step": 6790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8683828510597118e-05,
+      "loss": 0.3833,
+      "step": 6800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.868189296428917e-05,
+      "loss": 0.3293,
+      "step": 6810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8679957417981228e-05,
+      "loss": 0.3789,
+      "step": 6820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8678021871673282e-05,
+      "loss": 0.3262,
+      "step": 6830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8676086325365336e-05,
+      "loss": 0.3449,
+      "step": 6840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.867415077905739e-05,
+      "loss": 0.2956,
+      "step": 6850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8672215232749443e-05,
+      "loss": 0.2883,
+      "step": 6860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.86702796864415e-05,
+      "loss": 0.3236,
+      "step": 6870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8668344140133553e-05,
+      "loss": 0.5661,
+      "step": 6880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.866640859382561e-05,
+      "loss": 0.2544,
+      "step": 6890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8664473047517664e-05,
+      "loss": 0.2934,
+      "step": 6900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8662537501209718e-05,
+      "loss": 0.3362,
+      "step": 6910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8660601954901775e-05,
+      "loss": 0.264,
+      "step": 6920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8658666408593825e-05,
+      "loss": 0.2582,
+      "step": 6930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8656730862285882e-05,
+      "loss": 0.3938,
+      "step": 6940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8654795315977936e-05,
+      "loss": 0.3906,
+      "step": 6950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.865285976966999e-05,
+      "loss": 0.2991,
+      "step": 6960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8650924223362046e-05,
+      "loss": 0.3517,
+      "step": 6970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.86489886770541e-05,
+      "loss": 0.4076,
+      "step": 6980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8647053130746157e-05,
+      "loss": 0.2732,
+      "step": 6990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8645117584438207e-05,
+      "loss": 0.3043,
+      "step": 7000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8643182038130264e-05,
+      "loss": 0.3398,
+      "step": 7010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8641246491822318e-05,
+      "loss": 0.2635,
+      "step": 7020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.863931094551437e-05,
+      "loss": 0.5238,
+      "step": 7030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.863737539920643e-05,
+      "loss": 0.3611,
+      "step": 7040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8635439852898482e-05,
+      "loss": 0.4443,
+      "step": 7050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8633504306590536e-05,
+      "loss": 0.2671,
+      "step": 7060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8631568760282593e-05,
+      "loss": 0.4157,
+      "step": 7070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8629633213974647e-05,
+      "loss": 0.4455,
+      "step": 7080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.86276976676667e-05,
+      "loss": 0.3122,
+      "step": 7090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8625762121358754e-05,
+      "loss": 0.3113,
+      "step": 7100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.862382657505081e-05,
+      "loss": 0.4185,
+      "step": 7110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8621891028742864e-05,
+      "loss": 0.5069,
+      "step": 7120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8619955482434918e-05,
+      "loss": 0.288,
+      "step": 7130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8618019936126975e-05,
+      "loss": 0.2011,
+      "step": 7140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8616084389819025e-05,
+      "loss": 0.3601,
+      "step": 7150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8614148843511082e-05,
+      "loss": 0.4025,
+      "step": 7160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8612213297203136e-05,
+      "loss": 0.3757,
+      "step": 7170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8610277750895193e-05,
+      "loss": 0.4149,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8608342204587247e-05,
+      "loss": 0.3174,
+      "step": 7190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.86064066582793e-05,
+      "loss": 0.3494,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8604471111971357e-05,
+      "loss": 0.3389,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.860253556566341e-05,
+      "loss": 0.5023,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8600600019355465e-05,
+      "loss": 0.3073,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8598664473047518e-05,
+      "loss": 0.3632,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8596728926739572e-05,
+      "loss": 0.3306,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.859479338043163e-05,
+      "loss": 0.4477,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8592857834123683e-05,
+      "loss": 0.3131,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.859092228781574e-05,
+      "loss": 0.3356,
+      "step": 7280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8588986741507793e-05,
+      "loss": 0.332,
+      "step": 7290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8587051195199847e-05,
+      "loss": 0.406,
+      "step": 7300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.85851156488919e-05,
+      "loss": 0.2352,
+      "step": 7310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8583180102583954e-05,
+      "loss": 0.3321,
+      "step": 7320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.858124455627601e-05,
+      "loss": 0.4071,
+      "step": 7330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8579309009968065e-05,
+      "loss": 0.1675,
+      "step": 7340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.857737346366012e-05,
+      "loss": 0.5997,
+      "step": 7350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8575437917352175e-05,
+      "loss": 0.4883,
+      "step": 7360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.857350237104423e-05,
+      "loss": 0.4592,
+      "step": 7370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8571566824736283e-05,
+      "loss": 0.2885,
+      "step": 7380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8569631278428336e-05,
+      "loss": 0.3584,
+      "step": 7390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8567695732120393e-05,
+      "loss": 0.3204,
+      "step": 7400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8565760185812447e-05,
+      "loss": 0.4171,
+      "step": 7410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.85638246395045e-05,
+      "loss": 0.2512,
+      "step": 7420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8561889093196558e-05,
+      "loss": 0.4065,
+      "step": 7430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.855995354688861e-05,
+      "loss": 0.4318,
+      "step": 7440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8558018000580665e-05,
+      "loss": 0.2588,
+      "step": 7450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.855608245427272e-05,
+      "loss": 0.2969,
+      "step": 7460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8554146907964776e-05,
+      "loss": 0.3779,
+      "step": 7470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.855221136165683e-05,
+      "loss": 0.3335,
+      "step": 7480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8550275815348883e-05,
+      "loss": 0.3099,
+      "step": 7490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.854834026904094e-05,
+      "loss": 0.4144,
+      "step": 7500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8546404722732993e-05,
+      "loss": 0.2843,
+      "step": 7510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8544469176425047e-05,
+      "loss": 0.3132,
+      "step": 7520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.85425336301171e-05,
+      "loss": 0.1861,
+      "step": 7530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8540598083809154e-05,
+      "loss": 0.4092,
+      "step": 7540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.853866253750121e-05,
+      "loss": 0.3256,
+      "step": 7550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8536726991193265e-05,
+      "loss": 0.347,
+      "step": 7560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8534791444885322e-05,
+      "loss": 0.48,
+      "step": 7570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8532855898577376e-05,
+      "loss": 0.2864,
+      "step": 7580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.853092035226943e-05,
+      "loss": 0.3129,
+      "step": 7590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8528984805961486e-05,
+      "loss": 0.3525,
+      "step": 7600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8527049259653537e-05,
+      "loss": 0.2666,
+      "step": 7610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8525113713345594e-05,
+      "loss": 0.4155,
+      "step": 7620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8523178167037647e-05,
+      "loss": 0.3725,
+      "step": 7630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.85212426207297e-05,
+      "loss": 0.3956,
+      "step": 7640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8519307074421758e-05,
+      "loss": 0.3877,
+      "step": 7650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.851737152811381e-05,
+      "loss": 0.238,
+      "step": 7660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.851543598180587e-05,
+      "loss": 0.3421,
+      "step": 7670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.851350043549792e-05,
+      "loss": 0.3279,
+      "step": 7680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8511564889189976e-05,
+      "loss": 0.3341,
+      "step": 7690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.850962934288203e-05,
+      "loss": 0.3034,
+      "step": 7700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8507693796574083e-05,
+      "loss": 0.3053,
+      "step": 7710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.850575825026614e-05,
+      "loss": 0.2807,
+      "step": 7720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8503822703958194e-05,
+      "loss": 0.3688,
+      "step": 7730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8501887157650247e-05,
+      "loss": 0.3633,
+      "step": 7740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8499951611342304e-05,
+      "loss": 0.3636,
+      "step": 7750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8498016065034358e-05,
+      "loss": 0.3237,
+      "step": 7760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8496080518726412e-05,
+      "loss": 0.2752,
+      "step": 7770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8494144972418465e-05,
+      "loss": 0.3541,
+      "step": 7780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8492209426110522e-05,
+      "loss": 0.2767,
+      "step": 7790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8490273879802576e-05,
+      "loss": 0.4879,
+      "step": 7800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.848833833349463e-05,
+      "loss": 0.3322,
+      "step": 7810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8486402787186687e-05,
+      "loss": 0.4919,
+      "step": 7820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8484467240878737e-05,
+      "loss": 0.3178,
+      "step": 7830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8482531694570794e-05,
+      "loss": 0.2615,
+      "step": 7840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8480596148262848e-05,
+      "loss": 0.3312,
+      "step": 7850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8478660601954905e-05,
+      "loss": 0.381,
+      "step": 7860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8476725055646958e-05,
+      "loss": 0.5812,
+      "step": 7870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8474789509339012e-05,
+      "loss": 0.2434,
+      "step": 7880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.847285396303107e-05,
+      "loss": 0.3828,
+      "step": 7890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8470918416723123e-05,
+      "loss": 0.2616,
+      "step": 7900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8468982870415176e-05,
+      "loss": 0.3708,
+      "step": 7910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.846704732410723e-05,
+      "loss": 0.2601,
+      "step": 7920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8465111777799283e-05,
+      "loss": 0.3226,
+      "step": 7930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.846317623149134e-05,
+      "loss": 0.2885,
+      "step": 7940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8461240685183394e-05,
+      "loss": 0.4451,
+      "step": 7950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.845930513887545e-05,
+      "loss": 0.3158,
+      "step": 7960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8457369592567505e-05,
+      "loss": 0.4061,
+      "step": 7970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.845543404625956e-05,
+      "loss": 0.2168,
+      "step": 7980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8453498499951612e-05,
+      "loss": 0.3073,
+      "step": 7990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8451562953643666e-05,
+      "loss": 0.3378,
+      "step": 8000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8449627407335723e-05,
+      "loss": 0.3039,
+      "step": 8010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8447691861027776e-05,
+      "loss": 0.3899,
+      "step": 8020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.844575631471983e-05,
+      "loss": 0.2474,
+      "step": 8030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8443820768411887e-05,
+      "loss": 0.33,
+      "step": 8040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.844188522210394e-05,
+      "loss": 0.3826,
+      "step": 8050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8439949675795994e-05,
+      "loss": 0.303,
+      "step": 8060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8438014129488048e-05,
+      "loss": 0.3189,
+      "step": 8070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8436078583180105e-05,
+      "loss": 0.2922,
+      "step": 8080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.843414303687216e-05,
+      "loss": 0.2596,
+      "step": 8090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8432207490564212e-05,
+      "loss": 0.5159,
+      "step": 8100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.843027194425627e-05,
+      "loss": 0.3181,
+      "step": 8110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8428336397948323e-05,
+      "loss": 0.4271,
+      "step": 8120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8426400851640376e-05,
+      "loss": 0.3453,
+      "step": 8130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.842446530533243e-05,
+      "loss": 0.2857,
+      "step": 8140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8422529759024487e-05,
+      "loss": 0.2519,
+      "step": 8150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.842059421271654e-05,
+      "loss": 0.4198,
+      "step": 8160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8418658666408594e-05,
+      "loss": 0.3964,
+      "step": 8170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.841672312010065e-05,
+      "loss": 0.3443,
+      "step": 8180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8414787573792705e-05,
+      "loss": 0.2781,
+      "step": 8190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.841285202748476e-05,
+      "loss": 0.3547,
+      "step": 8200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8410916481176812e-05,
+      "loss": 0.2769,
+      "step": 8210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8408980934868866e-05,
+      "loss": 0.3398,
+      "step": 8220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8407045388560923e-05,
+      "loss": 0.3051,
+      "step": 8230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8405109842252977e-05,
+      "loss": 0.4598,
+      "step": 8240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8403174295945034e-05,
+      "loss": 0.3766,
+      "step": 8250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8401238749637087e-05,
+      "loss": 0.4321,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.839930320332914e-05,
+      "loss": 0.3444,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8397367657021198e-05,
+      "loss": 0.2653,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8395432110713248e-05,
+      "loss": 0.3143,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8393496564405305e-05,
+      "loss": 0.4002,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.839156101809736e-05,
+      "loss": 0.3208,
+      "step": 8310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8389625471789413e-05,
+      "loss": 0.3039,
+      "step": 8320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838768992548147e-05,
+      "loss": 0.386,
+      "step": 8330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8385754379173523e-05,
+      "loss": 0.3154,
+      "step": 8340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838381883286558e-05,
+      "loss": 0.303,
+      "step": 8350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838188328655763e-05,
+      "loss": 0.3458,
+      "step": 8360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8379947740249687e-05,
+      "loss": 0.2572,
+      "step": 8370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837801219394174e-05,
+      "loss": 0.3554,
+      "step": 8380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8376076647633795e-05,
+      "loss": 0.2646,
+      "step": 8390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8374141101325852e-05,
+      "loss": 0.5278,
+      "step": 8400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8372205555017905e-05,
+      "loss": 0.3005,
+      "step": 8410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837027000870996e-05,
+      "loss": 0.3468,
+      "step": 8420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8368334462402016e-05,
+      "loss": 0.3188,
+      "step": 8430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.836639891609407e-05,
+      "loss": 0.3798,
+      "step": 8440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8364463369786123e-05,
+      "loss": 0.3706,
+      "step": 8450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8362527823478177e-05,
+      "loss": 0.3007,
+      "step": 8460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8360592277170234e-05,
+      "loss": 0.3033,
+      "step": 8470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8358656730862288e-05,
+      "loss": 0.2932,
+      "step": 8480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.835672118455434e-05,
+      "loss": 0.407,
+      "step": 8490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8354785638246398e-05,
+      "loss": 0.3566,
+      "step": 8500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.835285009193845e-05,
+      "loss": 0.3719,
+      "step": 8510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8350914545630506e-05,
+      "loss": 0.4093,
+      "step": 8520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.834897899932256e-05,
+      "loss": 0.3197,
+      "step": 8530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8347043453014616e-05,
+      "loss": 0.3003,
+      "step": 8540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.834510790670667e-05,
+      "loss": 0.212,
+      "step": 8550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8343172360398723e-05,
+      "loss": 0.349,
+      "step": 8560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.834123681409078e-05,
+      "loss": 0.2925,
+      "step": 8570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8339301267782834e-05,
+      "loss": 0.2735,
+      "step": 8580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8337365721474888e-05,
+      "loss": 0.3169,
+      "step": 8590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.833543017516694e-05,
+      "loss": 0.3693,
+      "step": 8600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8333494628858995e-05,
+      "loss": 0.3285,
+      "step": 8610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8331559082551052e-05,
+      "loss": 0.3351,
+      "step": 8620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8329623536243106e-05,
+      "loss": 0.265,
+      "step": 8630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8327687989935163e-05,
+      "loss": 0.3356,
+      "step": 8640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8325752443627216e-05,
+      "loss": 0.3967,
+      "step": 8650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.832381689731927e-05,
+      "loss": 0.362,
+      "step": 8660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8321881351011324e-05,
+      "loss": 0.3239,
+      "step": 8670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8319945804703377e-05,
+      "loss": 0.3021,
+      "step": 8680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8318010258395434e-05,
+      "loss": 0.2433,
+      "step": 8690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8316074712087488e-05,
+      "loss": 0.3874,
+      "step": 8700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.831413916577954e-05,
+      "loss": 0.3488,
+      "step": 8710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.83122036194716e-05,
+      "loss": 0.309,
+      "step": 8720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8310268073163652e-05,
+      "loss": 0.2218,
+      "step": 8730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8308332526855706e-05,
+      "loss": 0.3475,
+      "step": 8740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.830639698054776e-05,
+      "loss": 0.2832,
+      "step": 8750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8304461434239817e-05,
+      "loss": 0.2744,
+      "step": 8760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.830252588793187e-05,
+      "loss": 0.3072,
+      "step": 8770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8300590341623924e-05,
+      "loss": 0.3038,
+      "step": 8780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.829865479531598e-05,
+      "loss": 0.2906,
+      "step": 8790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8296719249008034e-05,
+      "loss": 0.2102,
+      "step": 8800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8294783702700088e-05,
+      "loss": 0.3616,
+      "step": 8810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8292848156392142e-05,
+      "loss": 0.2323,
+      "step": 8820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.82909126100842e-05,
+      "loss": 0.3345,
+      "step": 8830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8288977063776252e-05,
+      "loss": 0.567,
+      "step": 8840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8287041517468306e-05,
+      "loss": 0.3922,
+      "step": 8850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8285105971160363e-05,
+      "loss": 0.4524,
+      "step": 8860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8283170424852417e-05,
+      "loss": 0.4013,
+      "step": 8870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.828123487854447e-05,
+      "loss": 0.3345,
+      "step": 8880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8279299332236524e-05,
+      "loss": 0.4011,
+      "step": 8890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8277363785928578e-05,
+      "loss": 0.3421,
+      "step": 8900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8275428239620635e-05,
+      "loss": 0.3368,
+      "step": 8910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8273492693312688e-05,
+      "loss": 0.1746,
+      "step": 8920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8271557147004745e-05,
+      "loss": 0.2804,
+      "step": 8930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.82696216006968e-05,
+      "loss": 0.3459,
+      "step": 8940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8267686054388853e-05,
+      "loss": 0.5029,
+      "step": 8950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.826575050808091e-05,
+      "loss": 0.3974,
+      "step": 8960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.826381496177296e-05,
+      "loss": 0.4006,
+      "step": 8970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8261879415465017e-05,
+      "loss": 0.2823,
+      "step": 8980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.825994386915707e-05,
+      "loss": 0.3753,
+      "step": 8990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8258008322849124e-05,
+      "loss": 0.4216,
+      "step": 9000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.825607277654118e-05,
+      "loss": 0.3845,
+      "step": 9010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8254137230233235e-05,
+      "loss": 0.3455,
+      "step": 9020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8252201683925292e-05,
+      "loss": 0.2668,
+      "step": 9030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8250266137617342e-05,
+      "loss": 0.3399,
+      "step": 9040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.82483305913094e-05,
+      "loss": 0.3577,
+      "step": 9050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8246395045001453e-05,
+      "loss": 0.2828,
+      "step": 9060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8244459498693506e-05,
+      "loss": 0.3839,
+      "step": 9070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8242523952385563e-05,
+      "loss": 0.2374,
+      "step": 9080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8240588406077617e-05,
+      "loss": 0.2963,
+      "step": 9090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.823865285976967e-05,
+      "loss": 0.3409,
+      "step": 9100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8236717313461728e-05,
+      "loss": 0.4284,
+      "step": 9110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.823478176715378e-05,
+      "loss": 0.3127,
+      "step": 9120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8232846220845835e-05,
+      "loss": 0.4288,
+      "step": 9130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.823091067453789e-05,
+      "loss": 0.2945,
+      "step": 9140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8228975128229946e-05,
+      "loss": 0.3732,
+      "step": 9150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8227039581922e-05,
+      "loss": 0.3105,
+      "step": 9160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8225104035614053e-05,
+      "loss": 0.252,
+      "step": 9170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.822316848930611e-05,
+      "loss": 0.2462,
+      "step": 9180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8221232942998164e-05,
+      "loss": 0.3349,
+      "step": 9190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8219297396690217e-05,
+      "loss": 0.411,
+      "step": 9200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.821736185038227e-05,
+      "loss": 0.3199,
+      "step": 9210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8215426304074328e-05,
+      "loss": 0.3333,
+      "step": 9220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.821349075776638e-05,
+      "loss": 0.4106,
+      "step": 9230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8211555211458435e-05,
+      "loss": 0.4452,
+      "step": 9240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8209619665150492e-05,
+      "loss": 0.2888,
+      "step": 9250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8207684118842546e-05,
+      "loss": 0.3203,
+      "step": 9260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.82057485725346e-05,
+      "loss": 0.2887,
+      "step": 9270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8203813026226653e-05,
+      "loss": 0.4203,
+      "step": 9280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8201877479918707e-05,
+      "loss": 0.3805,
+      "step": 9290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8199941933610764e-05,
+      "loss": 0.3047,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8198006387302817e-05,
+      "loss": 0.3427,
+      "step": 9310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8196070840994874e-05,
+      "loss": 0.4192,
+      "step": 9320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8194135294686928e-05,
+      "loss": 0.2797,
+      "step": 9330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.819219974837898e-05,
+      "loss": 0.3262,
+      "step": 9340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8190264202071035e-05,
+      "loss": 0.3345,
+      "step": 9350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.818832865576309e-05,
+      "loss": 0.2887,
+      "step": 9360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8186393109455146e-05,
+      "loss": 0.4654,
+      "step": 9370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.81844575631472e-05,
+      "loss": 0.2379,
+      "step": 9380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8182522016839253e-05,
+      "loss": 0.2199,
+      "step": 9390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.818058647053131e-05,
+      "loss": 0.4024,
+      "step": 9400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8178650924223364e-05,
+      "loss": 0.3641,
+      "step": 9410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8176715377915417e-05,
+      "loss": 0.3297,
+      "step": 9420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.817477983160747e-05,
+      "loss": 0.2559,
+      "step": 9430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8172844285299528e-05,
+      "loss": 0.2178,
+      "step": 9440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8170908738991582e-05,
+      "loss": 0.3624,
+      "step": 9450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8168973192683635e-05,
+      "loss": 0.4301,
+      "step": 9460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8167037646375692e-05,
+      "loss": 0.465,
+      "step": 9470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8165102100067746e-05,
+      "loss": 0.2276,
+      "step": 9480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.81631665537598e-05,
+      "loss": 0.4161,
+      "step": 9490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8161231007451853e-05,
+      "loss": 0.314,
+      "step": 9500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.815929546114391e-05,
+      "loss": 0.3701,
+      "step": 9510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8157359914835964e-05,
+      "loss": 0.2331,
+      "step": 9520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8155424368528018e-05,
+      "loss": 0.226,
+      "step": 9530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8153488822220075e-05,
+      "loss": 0.2902,
+      "step": 9540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8151553275912128e-05,
+      "loss": 0.2461,
+      "step": 9550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8149617729604182e-05,
+      "loss": 0.3302,
+      "step": 9560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8147682183296236e-05,
+      "loss": 0.3362,
+      "step": 9570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.814574663698829e-05,
+      "loss": 0.3117,
+      "step": 9580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8143811090680346e-05,
+      "loss": 0.2561,
+      "step": 9590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.81418755443724e-05,
+      "loss": 0.2557,
+      "step": 9600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8139939998064457e-05,
+      "loss": 0.4308,
+      "step": 9610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.813800445175651e-05,
+      "loss": 0.4339,
+      "step": 9620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8136068905448564e-05,
+      "loss": 0.3919,
+      "step": 9630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.813413335914062e-05,
+      "loss": 0.2994,
+      "step": 9640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.813219781283267e-05,
+      "loss": 0.2443,
+      "step": 9650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.813026226652473e-05,
+      "loss": 0.4114,
+      "step": 9660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8128326720216782e-05,
+      "loss": 0.4617,
+      "step": 9670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8126391173908836e-05,
+      "loss": 0.272,
+      "step": 9680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8124455627600893e-05,
+      "loss": 0.2754,
+      "step": 9690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8122520081292946e-05,
+      "loss": 0.292,
+      "step": 9700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8120584534985003e-05,
+      "loss": 0.3178,
+      "step": 9710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8118648988677057e-05,
+      "loss": 0.4456,
+      "step": 9720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.811671344236911e-05,
+      "loss": 0.312,
+      "step": 9730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8114777896061164e-05,
+      "loss": 0.2437,
+      "step": 9740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8112842349753218e-05,
+      "loss": 0.3638,
+      "step": 9750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8110906803445275e-05,
+      "loss": 0.3005,
+      "step": 9760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.810897125713733e-05,
+      "loss": 0.2665,
+      "step": 9770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8107035710829382e-05,
+      "loss": 0.252,
+      "step": 9780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.810510016452144e-05,
+      "loss": 0.3701,
+      "step": 9790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8103164618213493e-05,
+      "loss": 0.303,
+      "step": 9800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8101229071905547e-05,
+      "loss": 0.2783,
+      "step": 9810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.80992935255976e-05,
+      "loss": 0.5224,
+      "step": 9820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8097357979289657e-05,
+      "loss": 0.3384,
+      "step": 9830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.809542243298171e-05,
+      "loss": 0.2959,
+      "step": 9840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8093486886673764e-05,
+      "loss": 0.4738,
+      "step": 9850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.809155134036582e-05,
+      "loss": 0.3816,
+      "step": 9860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8089615794057875e-05,
+      "loss": 0.2889,
+      "step": 9870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.808768024774993e-05,
+      "loss": 0.252,
+      "step": 9880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8085744701441982e-05,
+      "loss": 0.4023,
+      "step": 9890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.808380915513404e-05,
+      "loss": 0.2877,
+      "step": 9900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8081873608826093e-05,
+      "loss": 0.4409,
+      "step": 9910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8079938062518147e-05,
+      "loss": 0.4125,
+      "step": 9920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8078002516210204e-05,
+      "loss": 0.3123,
+      "step": 9930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8076066969902257e-05,
+      "loss": 0.381,
+      "step": 9940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.807413142359431e-05,
+      "loss": 0.1897,
+      "step": 9950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8072195877286365e-05,
+      "loss": 0.4724,
+      "step": 9960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8070260330978418e-05,
+      "loss": 0.361,
+      "step": 9970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8068324784670475e-05,
+      "loss": 0.4062,
+      "step": 9980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.806638923836253e-05,
+      "loss": 0.3966,
+      "step": 9990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8064453692054586e-05,
+      "loss": 0.2606,
+      "step": 10000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.806251814574664e-05,
+      "loss": 0.2878,
+      "step": 10010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8060582599438693e-05,
+      "loss": 0.3085,
+      "step": 10020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8058647053130747e-05,
+      "loss": 0.5354,
+      "step": 10030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.80567115068228e-05,
+      "loss": 0.3531,
+      "step": 10040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8054775960514857e-05,
+      "loss": 0.2622,
+      "step": 10050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.805284041420691e-05,
+      "loss": 0.3848,
+      "step": 10060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8050904867898965e-05,
+      "loss": 0.3465,
+      "step": 10070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8048969321591022e-05,
+      "loss": 0.301,
+      "step": 10080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8047033775283075e-05,
+      "loss": 0.4145,
+      "step": 10090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.804509822897513e-05,
+      "loss": 0.4173,
+      "step": 10100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8043162682667183e-05,
+      "loss": 0.3967,
+      "step": 10110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.804122713635924e-05,
+      "loss": 0.3905,
+      "step": 10120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8039291590051293e-05,
+      "loss": 0.255,
+      "step": 10130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8037356043743347e-05,
+      "loss": 0.3156,
+      "step": 10140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8035420497435404e-05,
+      "loss": 0.4377,
+      "step": 10150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8033484951127458e-05,
+      "loss": 0.3761,
+      "step": 10160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.803154940481951e-05,
+      "loss": 0.3284,
+      "step": 10170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8029613858511565e-05,
+      "loss": 0.3078,
+      "step": 10180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8027678312203622e-05,
+      "loss": 0.2337,
+      "step": 10190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8025742765895676e-05,
+      "loss": 0.3601,
+      "step": 10200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.802380721958773e-05,
+      "loss": 0.2557,
+      "step": 10210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8021871673279786e-05,
+      "loss": 0.197,
+      "step": 10220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.801993612697184e-05,
+      "loss": 0.4232,
+      "step": 10230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8018000580663894e-05,
+      "loss": 0.2984,
+      "step": 10240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8016065034355947e-05,
+      "loss": 0.2304,
+      "step": 10250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8014129488048e-05,
+      "loss": 0.3563,
+      "step": 10260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8012193941740058e-05,
+      "loss": 0.3423,
+      "step": 10270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.801025839543211e-05,
+      "loss": 0.2777,
+      "step": 10280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.800832284912417e-05,
+      "loss": 0.4114,
+      "step": 10290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8006387302816222e-05,
+      "loss": 0.2532,
+      "step": 10300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8004451756508276e-05,
+      "loss": 0.5116,
+      "step": 10310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8002516210200333e-05,
+      "loss": 0.3188,
+      "step": 10320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8000580663892383e-05,
+      "loss": 0.4189,
+      "step": 10330
+    },
+    {
+      "epoch": 1.0,
+      "eval_FN": 633,
+      "eval_FP": 1997,
+      "eval_TN": 13762,
+      "eval_TP": 4273,
+      "eval_accuracy": 0.8727316719090249,
+      "eval_f1": 0.7646743020758769,
+      "eval_loss": 0.32348209619522095,
+      "eval_precision": 0.6814992025518342,
+      "eval_recall": 0.8709743171626579,
+      "eval_runtime": 142.0455,
+      "eval_samples_per_second": 145.482,
+      "eval_steps_per_second": 9.096,
+      "step": 10333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.799864511758444e-05,
+      "loss": 0.2483,
+      "step": 10340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7996709571276494e-05,
+      "loss": 0.2275,
+      "step": 10350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7994774024968547e-05,
+      "loss": 0.3128,
+      "step": 10360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7992838478660604e-05,
+      "loss": 0.323,
+      "step": 10370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7990902932352658e-05,
+      "loss": 0.21,
+      "step": 10380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7988967386044715e-05,
+      "loss": 0.2717,
+      "step": 10390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.798703183973677e-05,
+      "loss": 0.2561,
+      "step": 10400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7985096293428822e-05,
+      "loss": 0.3568,
+      "step": 10410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7983160747120876e-05,
+      "loss": 0.3371,
+      "step": 10420
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.798122520081293e-05,
+      "loss": 0.3075,
+      "step": 10430
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7979289654504987e-05,
+      "loss": 0.2404,
+      "step": 10440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.797735410819704e-05,
+      "loss": 0.2847,
+      "step": 10450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7975418561889094e-05,
+      "loss": 0.2981,
+      "step": 10460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.797348301558115e-05,
+      "loss": 0.2101,
+      "step": 10470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7971547469273204e-05,
+      "loss": 0.2906,
+      "step": 10480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7969611922965258e-05,
+      "loss": 0.3312,
+      "step": 10490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7967676376657312e-05,
+      "loss": 0.3293,
+      "step": 10500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.796574083034937e-05,
+      "loss": 0.4588,
+      "step": 10510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7963805284041422e-05,
+      "loss": 0.1739,
+      "step": 10520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7961869737733476e-05,
+      "loss": 0.2689,
+      "step": 10530
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7959934191425533e-05,
+      "loss": 0.3254,
+      "step": 10540
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7957998645117587e-05,
+      "loss": 0.4014,
+      "step": 10550
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.795606309880964e-05,
+      "loss": 0.2781,
+      "step": 10560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7954127552501694e-05,
+      "loss": 0.2659,
+      "step": 10570
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.795219200619375e-05,
+      "loss": 0.2742,
+      "step": 10580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7950256459885805e-05,
+      "loss": 0.3584,
+      "step": 10590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7948320913577858e-05,
+      "loss": 0.2346,
+      "step": 10600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7946385367269915e-05,
+      "loss": 0.4227,
+      "step": 10610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.794444982096197e-05,
+      "loss": 0.3183,
+      "step": 10620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7942514274654023e-05,
+      "loss": 0.3454,
+      "step": 10630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7940578728346076e-05,
+      "loss": 0.3053,
+      "step": 10640
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.793864318203813e-05,
+      "loss": 0.357,
+      "step": 10650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7936707635730187e-05,
+      "loss": 0.2759,
+      "step": 10660
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.793477208942224e-05,
+      "loss": 0.2706,
+      "step": 10670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7932836543114298e-05,
+      "loss": 0.2408,
+      "step": 10680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.793090099680635e-05,
+      "loss": 0.3548,
+      "step": 10690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7928965450498405e-05,
+      "loss": 0.2619,
+      "step": 10700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.792702990419046e-05,
+      "loss": 0.301,
+      "step": 10710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7925094357882512e-05,
+      "loss": 0.3667,
+      "step": 10720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.792315881157457e-05,
+      "loss": 0.3468,
+      "step": 10730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7921223265266623e-05,
+      "loss": 0.2212,
+      "step": 10740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7919287718958676e-05,
+      "loss": 0.2348,
+      "step": 10750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7917352172650733e-05,
+      "loss": 0.3599,
+      "step": 10760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7915416626342787e-05,
+      "loss": 0.3403,
+      "step": 10770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.791348108003484e-05,
+      "loss": 0.2977,
+      "step": 10780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7911545533726894e-05,
+      "loss": 0.2981,
+      "step": 10790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.790960998741895e-05,
+      "loss": 0.2956,
+      "step": 10800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7907674441111005e-05,
+      "loss": 0.2545,
+      "step": 10810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.790573889480306e-05,
+      "loss": 0.3093,
+      "step": 10820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7903803348495116e-05,
+      "loss": 0.3307,
+      "step": 10830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.790186780218717e-05,
+      "loss": 0.2052,
+      "step": 10840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7899932255879223e-05,
+      "loss": 0.3603,
+      "step": 10850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7897996709571277e-05,
+      "loss": 0.4176,
+      "step": 10860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7896061163263334e-05,
+      "loss": 0.3206,
+      "step": 10870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7894125616955387e-05,
+      "loss": 0.2235,
+      "step": 10880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.789219007064744e-05,
+      "loss": 0.2051,
+      "step": 10890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7890254524339498e-05,
+      "loss": 0.4007,
+      "step": 10900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.788831897803155e-05,
+      "loss": 0.4649,
+      "step": 10910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7886383431723605e-05,
+      "loss": 0.2752,
+      "step": 10920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7884447885415662e-05,
+      "loss": 0.2034,
+      "step": 10930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7882512339107712e-05,
+      "loss": 0.2743,
+      "step": 10940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.788057679279977e-05,
+      "loss": 0.3253,
+      "step": 10950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7878641246491823e-05,
+      "loss": 0.2611,
+      "step": 10960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.787670570018388e-05,
+      "loss": 0.3317,
+      "step": 10970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7874770153875934e-05,
+      "loss": 0.3276,
+      "step": 10980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7872834607567987e-05,
+      "loss": 0.2423,
+      "step": 10990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7870899061260044e-05,
+      "loss": 0.2252,
+      "step": 11000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7868963514952095e-05,
+      "loss": 0.1972,
+      "step": 11010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.786702796864415e-05,
+      "loss": 0.392,
+      "step": 11020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7865092422336205e-05,
+      "loss": 0.6294,
+      "step": 11030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.786315687602826e-05,
+      "loss": 0.2671,
+      "step": 11040
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7861221329720316e-05,
+      "loss": 0.3807,
+      "step": 11050
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.785928578341237e-05,
+      "loss": 0.375,
+      "step": 11060
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7857350237104427e-05,
+      "loss": 0.3174,
+      "step": 11070
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.785541469079648e-05,
+      "loss": 0.2594,
+      "step": 11080
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7853479144488534e-05,
+      "loss": 0.2906,
+      "step": 11090
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7851543598180587e-05,
+      "loss": 0.3741,
+      "step": 11100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.784960805187264e-05,
+      "loss": 0.3104,
+      "step": 11110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7847672505564698e-05,
+      "loss": 0.2734,
+      "step": 11120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7845736959256752e-05,
+      "loss": 0.2159,
+      "step": 11130
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7843801412948805e-05,
+      "loss": 0.2701,
+      "step": 11140
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7841865866640862e-05,
+      "loss": 0.3687,
+      "step": 11150
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7839930320332913e-05,
+      "loss": 0.1689,
+      "step": 11160
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.783799477402497e-05,
+      "loss": 0.4662,
+      "step": 11170
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7836059227717023e-05,
+      "loss": 0.1815,
+      "step": 11180
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.783412368140908e-05,
+      "loss": 0.2901,
+      "step": 11190
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7832188135101134e-05,
+      "loss": 0.2974,
+      "step": 11200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7830252588793188e-05,
+      "loss": 0.364,
+      "step": 11210
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7828317042485245e-05,
+      "loss": 0.2594,
+      "step": 11220
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.78263814961773e-05,
+      "loss": 0.2335,
+      "step": 11230
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7824445949869352e-05,
+      "loss": 0.2905,
+      "step": 11240
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7822510403561406e-05,
+      "loss": 0.3135,
+      "step": 11250
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7820574857253463e-05,
+      "loss": 0.2641,
+      "step": 11260
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7818639310945516e-05,
+      "loss": 0.2023,
+      "step": 11270
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.781670376463757e-05,
+      "loss": 0.4067,
+      "step": 11280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7814768218329627e-05,
+      "loss": 0.3494,
+      "step": 11290
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.781283267202168e-05,
+      "loss": 0.3323,
+      "step": 11300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7810897125713734e-05,
+      "loss": 0.4988,
+      "step": 11310
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7808961579405788e-05,
+      "loss": 0.2556,
+      "step": 11320
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.780702603309784e-05,
+      "loss": 0.2121,
+      "step": 11330
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.78050904867899e-05,
+      "loss": 0.3893,
+      "step": 11340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7803154940481952e-05,
+      "loss": 0.2478,
+      "step": 11350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.780121939417401e-05,
+      "loss": 0.3192,
+      "step": 11360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7799283847866063e-05,
+      "loss": 0.3694,
+      "step": 11370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7797348301558116e-05,
+      "loss": 0.3107,
+      "step": 11380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.779541275525017e-05,
+      "loss": 0.2937,
+      "step": 11390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7793477208942224e-05,
+      "loss": 0.2507,
+      "step": 11400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.779154166263428e-05,
+      "loss": 0.2119,
+      "step": 11410
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7789606116326334e-05,
+      "loss": 0.3228,
+      "step": 11420
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7787670570018388e-05,
+      "loss": 0.3126,
+      "step": 11430
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7785735023710445e-05,
+      "loss": 0.3323,
+      "step": 11440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.77837994774025e-05,
+      "loss": 0.2664,
+      "step": 11450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7781863931094556e-05,
+      "loss": 0.401,
+      "step": 11460
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7779928384786606e-05,
+      "loss": 0.249,
+      "step": 11470
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7777992838478663e-05,
+      "loss": 0.2923,
+      "step": 11480
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7776057292170717e-05,
+      "loss": 0.2577,
+      "step": 11490
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.777412174586277e-05,
+      "loss": 0.3159,
+      "step": 11500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7772186199554827e-05,
+      "loss": 0.2538,
+      "step": 11510
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.777025065324688e-05,
+      "loss": 0.3503,
+      "step": 11520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7768315106938934e-05,
+      "loss": 0.1787,
+      "step": 11530
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7766379560630988e-05,
+      "loss": 0.3458,
+      "step": 11540
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7764444014323042e-05,
+      "loss": 0.2278,
+      "step": 11550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.77625084680151e-05,
+      "loss": 0.3523,
+      "step": 11560
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7760572921707152e-05,
+      "loss": 0.3537,
+      "step": 11570
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.775863737539921e-05,
+      "loss": 0.4065,
+      "step": 11580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7756701829091263e-05,
+      "loss": 0.3426,
+      "step": 11590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7754766282783317e-05,
+      "loss": 0.29,
+      "step": 11600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7752830736475374e-05,
+      "loss": 0.3502,
+      "step": 11610
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7750895190167424e-05,
+      "loss": 0.3568,
+      "step": 11620
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.774895964385948e-05,
+      "loss": 0.2594,
+      "step": 11630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7747024097551535e-05,
+      "loss": 0.3507,
+      "step": 11640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.774508855124359e-05,
+      "loss": 0.3133,
+      "step": 11650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7743153004935645e-05,
+      "loss": 0.2974,
+      "step": 11660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.77412174586277e-05,
+      "loss": 0.2756,
+      "step": 11670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7739281912319756e-05,
+      "loss": 0.4259,
+      "step": 11680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7737346366011806e-05,
+      "loss": 0.2921,
+      "step": 11690
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7735410819703863e-05,
+      "loss": 0.3598,
+      "step": 11700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7733475273395917e-05,
+      "loss": 0.2676,
+      "step": 11710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.773153972708797e-05,
+      "loss": 0.4373,
+      "step": 11720
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7729604180780028e-05,
+      "loss": 0.3675,
+      "step": 11730
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.772766863447208e-05,
+      "loss": 0.2378,
+      "step": 11740
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7725733088164138e-05,
+      "loss": 0.2922,
+      "step": 11750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7723797541856192e-05,
+      "loss": 0.2991,
+      "step": 11760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7721861995548245e-05,
+      "loss": 0.3666,
+      "step": 11770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.77199264492403e-05,
+      "loss": 0.1641,
+      "step": 11780
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7717990902932353e-05,
+      "loss": 0.4562,
+      "step": 11790
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.771605535662441e-05,
+      "loss": 0.4683,
+      "step": 11800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7714119810316463e-05,
+      "loss": 0.2242,
+      "step": 11810
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7712184264008517e-05,
+      "loss": 0.3182,
+      "step": 11820
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7710248717700574e-05,
+      "loss": 0.3053,
+      "step": 11830
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7708313171392624e-05,
+      "loss": 0.3082,
+      "step": 11840
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.770637762508468e-05,
+      "loss": 0.2097,
+      "step": 11850
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7704442078776735e-05,
+      "loss": 0.1834,
+      "step": 11860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7702506532468792e-05,
+      "loss": 0.4084,
+      "step": 11870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7700570986160846e-05,
+      "loss": 0.3549,
+      "step": 11880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.76986354398529e-05,
+      "loss": 0.2612,
+      "step": 11890
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7696699893544956e-05,
+      "loss": 0.2147,
+      "step": 11900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.769476434723701e-05,
+      "loss": 0.2528,
+      "step": 11910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7692828800929064e-05,
+      "loss": 0.2622,
+      "step": 11920
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7690893254621117e-05,
+      "loss": 0.3125,
+      "step": 11930
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.768895770831317e-05,
+      "loss": 0.248,
+      "step": 11940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7687022162005228e-05,
+      "loss": 0.2384,
+      "step": 11950
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.768508661569728e-05,
+      "loss": 0.3284,
+      "step": 11960
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.768315106938934e-05,
+      "loss": 0.1896,
+      "step": 11970
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7681215523081392e-05,
+      "loss": 0.5095,
+      "step": 11980
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7679279976773446e-05,
+      "loss": 0.2764,
+      "step": 11990
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.76773444304655e-05,
+      "loss": 0.3324,
+      "step": 12000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7675408884157553e-05,
+      "loss": 0.304,
+      "step": 12010
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.767347333784961e-05,
+      "loss": 0.296,
+      "step": 12020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7671537791541664e-05,
+      "loss": 0.3058,
+      "step": 12030
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.766960224523372e-05,
+      "loss": 0.3414,
+      "step": 12040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7667666698925774e-05,
+      "loss": 0.3343,
+      "step": 12050
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7665731152617828e-05,
+      "loss": 0.4341,
+      "step": 12060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.766379560630988e-05,
+      "loss": 0.2519,
+      "step": 12070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7661860060001935e-05,
+      "loss": 0.3931,
+      "step": 12080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7659924513693992e-05,
+      "loss": 0.3777,
+      "step": 12090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7657988967386046e-05,
+      "loss": 0.1947,
+      "step": 12100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.76560534210781e-05,
+      "loss": 0.291,
+      "step": 12110
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7654117874770157e-05,
+      "loss": 0.4085,
+      "step": 12120
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.765218232846221e-05,
+      "loss": 0.3697,
+      "step": 12130
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7650246782154267e-05,
+      "loss": 0.2979,
+      "step": 12140
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7648311235846317e-05,
+      "loss": 0.2956,
+      "step": 12150
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7646375689538375e-05,
+      "loss": 0.1438,
+      "step": 12160
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7644440143230428e-05,
+      "loss": 0.2426,
+      "step": 12170
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7642504596922482e-05,
+      "loss": 0.373,
+      "step": 12180
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.764056905061454e-05,
+      "loss": 0.2025,
+      "step": 12190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7638633504306592e-05,
+      "loss": 0.3099,
+      "step": 12200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7636697957998646e-05,
+      "loss": 0.3478,
+      "step": 12210
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.76347624116907e-05,
+      "loss": 0.2691,
+      "step": 12220
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7632826865382753e-05,
+      "loss": 0.3796,
+      "step": 12230
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.763089131907481e-05,
+      "loss": 0.364,
+      "step": 12240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7628955772766864e-05,
+      "loss": 0.2784,
+      "step": 12250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.762702022645892e-05,
+      "loss": 0.2691,
+      "step": 12260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7625084680150975e-05,
+      "loss": 0.2386,
+      "step": 12270
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7623149133843028e-05,
+      "loss": 0.3177,
+      "step": 12280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7621213587535085e-05,
+      "loss": 0.2799,
+      "step": 12290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7619278041227136e-05,
+      "loss": 0.2812,
+      "step": 12300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7617342494919193e-05,
+      "loss": 0.2085,
+      "step": 12310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7615406948611246e-05,
+      "loss": 0.3404,
+      "step": 12320
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.76134714023033e-05,
+      "loss": 0.2502,
+      "step": 12330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7611535855995357e-05,
+      "loss": 0.2787,
+      "step": 12340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.760960030968741e-05,
+      "loss": 0.2544,
+      "step": 12350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7607664763379468e-05,
+      "loss": 0.2554,
+      "step": 12360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7605729217071518e-05,
+      "loss": 0.2304,
+      "step": 12370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7603793670763575e-05,
+      "loss": 0.3398,
+      "step": 12380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.760185812445563e-05,
+      "loss": 0.4181,
+      "step": 12390
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7599922578147682e-05,
+      "loss": 0.3358,
+      "step": 12400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.759798703183974e-05,
+      "loss": 0.2375,
+      "step": 12410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7596051485531793e-05,
+      "loss": 0.2598,
+      "step": 12420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.759411593922385e-05,
+      "loss": 0.2919,
+      "step": 12430
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7592180392915903e-05,
+      "loss": 0.2072,
+      "step": 12440
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7590244846607957e-05,
+      "loss": 0.2624,
+      "step": 12450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.758830930030001e-05,
+      "loss": 0.2373,
+      "step": 12460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7586373753992064e-05,
+      "loss": 0.3185,
+      "step": 12470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.758443820768412e-05,
+      "loss": 0.5107,
+      "step": 12480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7582502661376175e-05,
+      "loss": 0.341,
+      "step": 12490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.758056711506823e-05,
+      "loss": 0.2974,
+      "step": 12500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7578631568760286e-05,
+      "loss": 0.2554,
+      "step": 12510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7576696022452336e-05,
+      "loss": 0.2948,
+      "step": 12520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7574760476144393e-05,
+      "loss": 0.3229,
+      "step": 12530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7572824929836447e-05,
+      "loss": 0.214,
+      "step": 12540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7570889383528504e-05,
+      "loss": 0.268,
+      "step": 12550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7568953837220557e-05,
+      "loss": 0.4982,
+      "step": 12560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.756701829091261e-05,
+      "loss": 0.3273,
+      "step": 12570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7565082744604668e-05,
+      "loss": 0.2519,
+      "step": 12580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.756314719829672e-05,
+      "loss": 0.2431,
+      "step": 12590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7561211651988775e-05,
+      "loss": 0.4066,
+      "step": 12600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.755927610568083e-05,
+      "loss": 0.3407,
+      "step": 12610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7557340559372882e-05,
+      "loss": 0.3044,
+      "step": 12620
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.755540501306494e-05,
+      "loss": 0.3155,
+      "step": 12630
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7553469466756993e-05,
+      "loss": 0.2075,
+      "step": 12640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.755153392044905e-05,
+      "loss": 0.3167,
+      "step": 12650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7549598374141104e-05,
+      "loss": 0.3173,
+      "step": 12660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7547662827833157e-05,
+      "loss": 0.2738,
+      "step": 12670
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.754572728152521e-05,
+      "loss": 0.3184,
+      "step": 12680
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7543791735217265e-05,
+      "loss": 0.2201,
+      "step": 12690
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.754185618890932e-05,
+      "loss": 0.3613,
+      "step": 12700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7539920642601375e-05,
+      "loss": 0.2429,
+      "step": 12710
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.753798509629343e-05,
+      "loss": 0.2894,
+      "step": 12720
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7536049549985486e-05,
+      "loss": 0.2337,
+      "step": 12730
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.753411400367754e-05,
+      "loss": 0.2556,
+      "step": 12740
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7532178457369593e-05,
+      "loss": 0.2638,
+      "step": 12750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7530242911061647e-05,
+      "loss": 0.3512,
+      "step": 12760
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7528307364753704e-05,
+      "loss": 0.2178,
+      "step": 12770
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7526371818445758e-05,
+      "loss": 0.4166,
+      "step": 12780
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.752443627213781e-05,
+      "loss": 0.242,
+      "step": 12790
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7522500725829868e-05,
+      "loss": 0.2775,
+      "step": 12800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7520565179521922e-05,
+      "loss": 0.2914,
+      "step": 12810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7518629633213975e-05,
+      "loss": 0.2436,
+      "step": 12820
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.751669408690603e-05,
+      "loss": 0.2706,
+      "step": 12830
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7514758540598086e-05,
+      "loss": 0.3251,
+      "step": 12840
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.751282299429014e-05,
+      "loss": 0.2894,
+      "step": 12850
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7510887447982193e-05,
+      "loss": 0.2491,
+      "step": 12860
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.750895190167425e-05,
+      "loss": 0.4091,
+      "step": 12870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7507016355366304e-05,
+      "loss": 0.2688,
+      "step": 12880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7505080809058358e-05,
+      "loss": 0.2115,
+      "step": 12890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.750314526275041e-05,
+      "loss": 0.2996,
+      "step": 12900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7501209716442465e-05,
+      "loss": 0.3004,
+      "step": 12910
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7499274170134522e-05,
+      "loss": 0.3488,
+      "step": 12920
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7497338623826576e-05,
+      "loss": 0.3649,
+      "step": 12930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7495403077518633e-05,
+      "loss": 0.3626,
+      "step": 12940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7493467531210686e-05,
+      "loss": 0.3553,
+      "step": 12950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.749153198490274e-05,
+      "loss": 0.3368,
+      "step": 12960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7489596438594797e-05,
+      "loss": 0.3551,
+      "step": 12970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7487660892286847e-05,
+      "loss": 0.4314,
+      "step": 12980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7485725345978904e-05,
+      "loss": 0.2911,
+      "step": 12990
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7483789799670958e-05,
+      "loss": 0.2659,
+      "step": 13000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.748185425336301e-05,
+      "loss": 0.2418,
+      "step": 13010
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.747991870705507e-05,
+      "loss": 0.3578,
+      "step": 13020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7477983160747122e-05,
+      "loss": 0.313,
+      "step": 13030
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.747604761443918e-05,
+      "loss": 0.2388,
+      "step": 13040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.747411206813123e-05,
+      "loss": 0.3395,
+      "step": 13050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7472176521823286e-05,
+      "loss": 0.396,
+      "step": 13060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.747024097551534e-05,
+      "loss": 0.2968,
+      "step": 13070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7468305429207394e-05,
+      "loss": 0.3117,
+      "step": 13080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746636988289945e-05,
+      "loss": 0.2308,
+      "step": 13090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7464434336591504e-05,
+      "loss": 0.3432,
+      "step": 13100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7462498790283558e-05,
+      "loss": 0.2703,
+      "step": 13110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7460563243975615e-05,
+      "loss": 0.2355,
+      "step": 13120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.745862769766767e-05,
+      "loss": 0.3138,
+      "step": 13130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7456692151359722e-05,
+      "loss": 0.3389,
+      "step": 13140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7454756605051776e-05,
+      "loss": 0.3225,
+      "step": 13150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7452821058743833e-05,
+      "loss": 0.2878,
+      "step": 13160
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7450885512435887e-05,
+      "loss": 0.3208,
+      "step": 13170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744894996612794e-05,
+      "loss": 0.3255,
+      "step": 13180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7447014419819997e-05,
+      "loss": 0.1907,
+      "step": 13190
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7445078873512047e-05,
+      "loss": 0.2946,
+      "step": 13200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7443143327204105e-05,
+      "loss": 0.2404,
+      "step": 13210
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7441207780896158e-05,
+      "loss": 0.3631,
+      "step": 13220
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7439272234588215e-05,
+      "loss": 0.307,
+      "step": 13230
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743733668828027e-05,
+      "loss": 0.2548,
+      "step": 13240
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7435401141972322e-05,
+      "loss": 0.4253,
+      "step": 13250
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743346559566438e-05,
+      "loss": 0.2876,
+      "step": 13260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7431530049356433e-05,
+      "loss": 0.3833,
+      "step": 13270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7429594503048487e-05,
+      "loss": 0.2182,
+      "step": 13280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.742765895674054e-05,
+      "loss": 0.2656,
+      "step": 13290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7425723410432594e-05,
+      "loss": 0.4348,
+      "step": 13300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.742378786412465e-05,
+      "loss": 0.3498,
+      "step": 13310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7421852317816705e-05,
+      "loss": 0.378,
+      "step": 13320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741991677150876e-05,
+      "loss": 0.3365,
+      "step": 13330
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7417981225200815e-05,
+      "loss": 0.35,
+      "step": 13340
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741604567889287e-05,
+      "loss": 0.2568,
+      "step": 13350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7414110132584923e-05,
+      "loss": 0.2551,
+      "step": 13360
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7412174586276976e-05,
+      "loss": 0.3466,
+      "step": 13370
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7410239039969033e-05,
+      "loss": 0.3855,
+      "step": 13380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7408303493661087e-05,
+      "loss": 0.2287,
+      "step": 13390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.740636794735314e-05,
+      "loss": 0.1829,
+      "step": 13400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7404432401045198e-05,
+      "loss": 0.2454,
+      "step": 13410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.740249685473725e-05,
+      "loss": 0.3818,
+      "step": 13420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7400561308429305e-05,
+      "loss": 0.4368,
+      "step": 13430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.739862576212136e-05,
+      "loss": 0.2245,
+      "step": 13440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7396690215813415e-05,
+      "loss": 0.3694,
+      "step": 13450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.739475466950547e-05,
+      "loss": 0.3407,
+      "step": 13460
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7392819123197523e-05,
+      "loss": 0.326,
+      "step": 13470
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.739088357688958e-05,
+      "loss": 0.341,
+      "step": 13480
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7388948030581633e-05,
+      "loss": 0.2129,
+      "step": 13490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7387012484273687e-05,
+      "loss": 0.1966,
+      "step": 13500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.738507693796574e-05,
+      "loss": 0.3588,
+      "step": 13510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7383141391657798e-05,
+      "loss": 0.2129,
+      "step": 13520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.738120584534985e-05,
+      "loss": 0.2396,
+      "step": 13530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7379270299041905e-05,
+      "loss": 0.2834,
+      "step": 13540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7377334752733962e-05,
+      "loss": 0.2101,
+      "step": 13550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7375399206426016e-05,
+      "loss": 0.2817,
+      "step": 13560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.737346366011807e-05,
+      "loss": 0.382,
+      "step": 13570
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7371528113810123e-05,
+      "loss": 0.2905,
+      "step": 13580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7369592567502177e-05,
+      "loss": 0.3307,
+      "step": 13590
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7367657021194234e-05,
+      "loss": 0.2484,
+      "step": 13600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7365721474886287e-05,
+      "loss": 0.2224,
+      "step": 13610
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7363785928578344e-05,
+      "loss": 0.2508,
+      "step": 13620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7361850382270398e-05,
+      "loss": 0.3228,
+      "step": 13630
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.735991483596245e-05,
+      "loss": 0.2452,
+      "step": 13640
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.735797928965451e-05,
+      "loss": 0.1718,
+      "step": 13650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.735604374334656e-05,
+      "loss": 0.1921,
+      "step": 13660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7354108197038616e-05,
+      "loss": 0.3054,
+      "step": 13670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.735217265073067e-05,
+      "loss": 0.3517,
+      "step": 13680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7350237104422723e-05,
+      "loss": 0.251,
+      "step": 13690
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.734830155811478e-05,
+      "loss": 0.2865,
+      "step": 13700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7346366011806834e-05,
+      "loss": 0.4003,
+      "step": 13710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.734443046549889e-05,
+      "loss": 0.1633,
+      "step": 13720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.734249491919094e-05,
+      "loss": 0.322,
+      "step": 13730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7340559372882998e-05,
+      "loss": 0.3617,
+      "step": 13740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.733862382657505e-05,
+      "loss": 0.3212,
+      "step": 13750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7336688280267105e-05,
+      "loss": 0.3475,
+      "step": 13760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7334752733959162e-05,
+      "loss": 0.3098,
+      "step": 13770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7332817187651216e-05,
+      "loss": 0.1709,
+      "step": 13780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.733088164134327e-05,
+      "loss": 0.2931,
+      "step": 13790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7328946095035327e-05,
+      "loss": 0.4214,
+      "step": 13800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.732701054872738e-05,
+      "loss": 0.3188,
+      "step": 13810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7325075002419434e-05,
+      "loss": 0.3444,
+      "step": 13820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7323139456111488e-05,
+      "loss": 0.2546,
+      "step": 13830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7321203909803545e-05,
+      "loss": 0.3692,
+      "step": 13840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7319268363495598e-05,
+      "loss": 0.2862,
+      "step": 13850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7317332817187652e-05,
+      "loss": 0.1609,
+      "step": 13860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.731539727087971e-05,
+      "loss": 0.4133,
+      "step": 13870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7313461724571762e-05,
+      "loss": 0.4843,
+      "step": 13880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7311526178263816e-05,
+      "loss": 0.2882,
+      "step": 13890
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.730959063195587e-05,
+      "loss": 0.2981,
+      "step": 13900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7307655085647927e-05,
+      "loss": 0.2529,
+      "step": 13910
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.730571953933998e-05,
+      "loss": 0.3123,
+      "step": 13920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7303783993032034e-05,
+      "loss": 0.303,
+      "step": 13930
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.730184844672409e-05,
+      "loss": 0.3345,
+      "step": 13940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7299912900416145e-05,
+      "loss": 0.2852,
+      "step": 13950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.72979773541082e-05,
+      "loss": 0.3422,
+      "step": 13960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7296041807800252e-05,
+      "loss": 0.2504,
+      "step": 13970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7294106261492306e-05,
+      "loss": 0.2574,
+      "step": 13980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7292170715184363e-05,
+      "loss": 0.2679,
+      "step": 13990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7290235168876416e-05,
+      "loss": 0.2278,
+      "step": 14000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7288299622568473e-05,
+      "loss": 0.2479,
+      "step": 14010
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7286364076260527e-05,
+      "loss": 0.3253,
+      "step": 14020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.728442852995258e-05,
+      "loss": 0.3506,
+      "step": 14030
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7282492983644634e-05,
+      "loss": 0.2396,
+      "step": 14040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7280557437336688e-05,
+      "loss": 0.2969,
+      "step": 14050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7278621891028745e-05,
+      "loss": 0.2032,
+      "step": 14060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.72766863447208e-05,
+      "loss": 0.2813,
+      "step": 14070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7274750798412852e-05,
+      "loss": 0.3826,
+      "step": 14080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.727281525210491e-05,
+      "loss": 0.1839,
+      "step": 14090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7270879705796963e-05,
+      "loss": 0.3368,
+      "step": 14100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7268944159489016e-05,
+      "loss": 0.3362,
+      "step": 14110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.726700861318107e-05,
+      "loss": 0.306,
+      "step": 14120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7265073066873127e-05,
+      "loss": 0.2922,
+      "step": 14130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.726313752056518e-05,
+      "loss": 0.2158,
+      "step": 14140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7261201974257234e-05,
+      "loss": 0.4264,
+      "step": 14150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.725926642794929e-05,
+      "loss": 0.357,
+      "step": 14160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7257330881641345e-05,
+      "loss": 0.3419,
+      "step": 14170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.72553953353334e-05,
+      "loss": 0.2832,
+      "step": 14180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7253459789025452e-05,
+      "loss": 0.1968,
+      "step": 14190
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.725152424271751e-05,
+      "loss": 0.3244,
+      "step": 14200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7249588696409563e-05,
+      "loss": 0.3712,
+      "step": 14210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7247653150101617e-05,
+      "loss": 0.2123,
+      "step": 14220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7245717603793674e-05,
+      "loss": 0.1996,
+      "step": 14230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7243782057485727e-05,
+      "loss": 0.2396,
+      "step": 14240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.724184651117778e-05,
+      "loss": 0.3362,
+      "step": 14250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7239910964869835e-05,
+      "loss": 0.2388,
+      "step": 14260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7237975418561888e-05,
+      "loss": 0.2132,
+      "step": 14270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7236039872253945e-05,
+      "loss": 0.3061,
+      "step": 14280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7234104325946e-05,
+      "loss": 0.4071,
+      "step": 14290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7232168779638056e-05,
+      "loss": 0.3262,
+      "step": 14300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.723023323333011e-05,
+      "loss": 0.361,
+      "step": 14310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7228297687022163e-05,
+      "loss": 0.253,
+      "step": 14320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.722636214071422e-05,
+      "loss": 0.3543,
+      "step": 14330
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.722442659440627e-05,
+      "loss": 0.2824,
+      "step": 14340
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7222491048098327e-05,
+      "loss": 0.2834,
+      "step": 14350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.722055550179038e-05,
+      "loss": 0.2591,
+      "step": 14360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7218619955482435e-05,
+      "loss": 0.2181,
+      "step": 14370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.721668440917449e-05,
+      "loss": 0.3592,
+      "step": 14380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7214748862866545e-05,
+      "loss": 0.216,
+      "step": 14390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7212813316558602e-05,
+      "loss": 0.2801,
+      "step": 14400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7210877770250656e-05,
+      "loss": 0.4208,
+      "step": 14410
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.720894222394271e-05,
+      "loss": 0.3146,
+      "step": 14420
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7207006677634763e-05,
+      "loss": 0.3148,
+      "step": 14430
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7205071131326817e-05,
+      "loss": 0.234,
+      "step": 14440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7203135585018874e-05,
+      "loss": 0.2951,
+      "step": 14450
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7201200038710928e-05,
+      "loss": 0.2603,
+      "step": 14460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.719926449240298e-05,
+      "loss": 0.2924,
+      "step": 14470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7197328946095038e-05,
+      "loss": 0.2818,
+      "step": 14480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7195393399787092e-05,
+      "loss": 0.338,
+      "step": 14490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7193457853479145e-05,
+      "loss": 0.2182,
+      "step": 14500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.71915223071712e-05,
+      "loss": 0.2248,
+      "step": 14510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7189586760863256e-05,
+      "loss": 0.2366,
+      "step": 14520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.718765121455531e-05,
+      "loss": 0.3848,
+      "step": 14530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7185715668247363e-05,
+      "loss": 0.3134,
+      "step": 14540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.718378012193942e-05,
+      "loss": 0.4722,
+      "step": 14550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7181844575631474e-05,
+      "loss": 0.2455,
+      "step": 14560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7179909029323528e-05,
+      "loss": 0.3423,
+      "step": 14570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.717797348301558e-05,
+      "loss": 0.305,
+      "step": 14580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.717603793670764e-05,
+      "loss": 0.2622,
+      "step": 14590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7174102390399692e-05,
+      "loss": 0.3206,
+      "step": 14600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7172166844091746e-05,
+      "loss": 0.3733,
+      "step": 14610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7170231297783803e-05,
+      "loss": 0.3235,
+      "step": 14620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7168295751475856e-05,
+      "loss": 0.2555,
+      "step": 14630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.716636020516791e-05,
+      "loss": 0.2367,
+      "step": 14640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7164424658859964e-05,
+      "loss": 0.3549,
+      "step": 14650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7162489112552017e-05,
+      "loss": 0.3309,
+      "step": 14660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7160553566244074e-05,
+      "loss": 0.2453,
+      "step": 14670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7158618019936128e-05,
+      "loss": 0.3493,
+      "step": 14680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7156682473628185e-05,
+      "loss": 0.2987,
+      "step": 14690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.715474692732024e-05,
+      "loss": 0.3634,
+      "step": 14700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7152811381012292e-05,
+      "loss": 0.3987,
+      "step": 14710
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7150875834704346e-05,
+      "loss": 0.2969,
+      "step": 14720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.71489402883964e-05,
+      "loss": 0.3346,
+      "step": 14730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7147004742088456e-05,
+      "loss": 0.3005,
+      "step": 14740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.714506919578051e-05,
+      "loss": 0.3874,
+      "step": 14750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7143133649472564e-05,
+      "loss": 0.191,
+      "step": 14760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.714119810316462e-05,
+      "loss": 0.4449,
+      "step": 14770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7139262556856674e-05,
+      "loss": 0.3259,
+      "step": 14780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7137327010548728e-05,
+      "loss": 0.1921,
+      "step": 14790
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.713539146424078e-05,
+      "loss": 0.3625,
+      "step": 14800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.713345591793284e-05,
+      "loss": 0.3073,
+      "step": 14810
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7131520371624892e-05,
+      "loss": 0.3603,
+      "step": 14820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7129584825316946e-05,
+      "loss": 0.1769,
+      "step": 14830
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7127649279009003e-05,
+      "loss": 0.2774,
+      "step": 14840
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7125713732701057e-05,
+      "loss": 0.3222,
+      "step": 14850
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.712377818639311e-05,
+      "loss": 0.3113,
+      "step": 14860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7121842640085164e-05,
+      "loss": 0.1536,
+      "step": 14870
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.711990709377722e-05,
+      "loss": 0.3249,
+      "step": 14880
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7117971547469275e-05,
+      "loss": 0.3014,
+      "step": 14890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7116036001161328e-05,
+      "loss": 0.3849,
+      "step": 14900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7114100454853385e-05,
+      "loss": 0.3169,
+      "step": 14910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.711216490854544e-05,
+      "loss": 0.445,
+      "step": 14920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7110229362237492e-05,
+      "loss": 0.1737,
+      "step": 14930
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7108293815929546e-05,
+      "loss": 0.4172,
+      "step": 14940
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.71063582696216e-05,
+      "loss": 0.3779,
+      "step": 14950
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7104422723313657e-05,
+      "loss": 0.4008,
+      "step": 14960
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.710248717700571e-05,
+      "loss": 0.339,
+      "step": 14970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7100551630697767e-05,
+      "loss": 0.32,
+      "step": 14980
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.709861608438982e-05,
+      "loss": 0.2163,
+      "step": 14990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7096680538081875e-05,
+      "loss": 0.3445,
+      "step": 15000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7094744991773932e-05,
+      "loss": 0.2182,
+      "step": 15010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7092809445465982e-05,
+      "loss": 0.2312,
+      "step": 15020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.709087389915804e-05,
+      "loss": 0.4108,
+      "step": 15030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7088938352850093e-05,
+      "loss": 0.401,
+      "step": 15040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7087002806542146e-05,
+      "loss": 0.1518,
+      "step": 15050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7085067260234203e-05,
+      "loss": 0.3935,
+      "step": 15060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7083131713926257e-05,
+      "loss": 0.4534,
+      "step": 15070
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7081196167618314e-05,
+      "loss": 0.2556,
+      "step": 15080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7079260621310368e-05,
+      "loss": 0.175,
+      "step": 15090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.707732507500242e-05,
+      "loss": 0.4635,
+      "step": 15100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7075389528694475e-05,
+      "loss": 0.181,
+      "step": 15110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.707345398238653e-05,
+      "loss": 0.3349,
+      "step": 15120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7071518436078586e-05,
+      "loss": 0.2822,
+      "step": 15130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.706958288977064e-05,
+      "loss": 0.2728,
+      "step": 15140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7067647343462693e-05,
+      "loss": 0.1593,
+      "step": 15150
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.706571179715475e-05,
+      "loss": 0.3836,
+      "step": 15160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7063776250846803e-05,
+      "loss": 0.5097,
+      "step": 15170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7061840704538857e-05,
+      "loss": 0.2625,
+      "step": 15180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.705990515823091e-05,
+      "loss": 0.3634,
+      "step": 15190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7057969611922968e-05,
+      "loss": 0.3355,
+      "step": 15200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.705603406561502e-05,
+      "loss": 0.3,
+      "step": 15210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7054098519307075e-05,
+      "loss": 0.4448,
+      "step": 15220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7052162972999132e-05,
+      "loss": 0.3678,
+      "step": 15230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7050227426691186e-05,
+      "loss": 0.3687,
+      "step": 15240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.704829188038324e-05,
+      "loss": 0.3271,
+      "step": 15250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7046356334075293e-05,
+      "loss": 0.2735,
+      "step": 15260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.704442078776735e-05,
+      "loss": 0.1678,
+      "step": 15270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7042485241459404e-05,
+      "loss": 0.2928,
+      "step": 15280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7040549695151457e-05,
+      "loss": 0.3669,
+      "step": 15290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7038614148843514e-05,
+      "loss": 0.2997,
+      "step": 15300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7036678602535568e-05,
+      "loss": 0.2852,
+      "step": 15310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.703474305622762e-05,
+      "loss": 0.3684,
+      "step": 15320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7032807509919675e-05,
+      "loss": 0.2905,
+      "step": 15330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.703087196361173e-05,
+      "loss": 0.2983,
+      "step": 15340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7028936417303786e-05,
+      "loss": 0.3232,
+      "step": 15350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.702700087099584e-05,
+      "loss": 0.339,
+      "step": 15360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7025065324687896e-05,
+      "loss": 0.3196,
+      "step": 15370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.702312977837995e-05,
+      "loss": 0.2618,
+      "step": 15380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7021194232072004e-05,
+      "loss": 0.2814,
+      "step": 15390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7019258685764057e-05,
+      "loss": 0.3145,
+      "step": 15400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.701732313945611e-05,
+      "loss": 0.3224,
+      "step": 15410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7015387593148168e-05,
+      "loss": 0.4222,
+      "step": 15420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.701345204684022e-05,
+      "loss": 0.3898,
+      "step": 15430
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7011516500532275e-05,
+      "loss": 0.2236,
+      "step": 15440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7009580954224332e-05,
+      "loss": 0.3493,
+      "step": 15450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7007645407916386e-05,
+      "loss": 0.3444,
+      "step": 15460
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.700570986160844e-05,
+      "loss": 0.3962,
+      "step": 15470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7003774315300493e-05,
+      "loss": 0.1949,
+      "step": 15480
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.700183876899255e-05,
+      "loss": 0.2249,
+      "step": 15490
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6999903222684604e-05,
+      "loss": 0.2345,
+      "step": 15500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6997967676376658e-05,
+      "loss": 0.2555,
+      "step": 15510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6996032130068715e-05,
+      "loss": 0.2762,
+      "step": 15520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6994096583760768e-05,
+      "loss": 0.2083,
+      "step": 15530
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6992161037452822e-05,
+      "loss": 0.3651,
+      "step": 15540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6990225491144875e-05,
+      "loss": 0.307,
+      "step": 15550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6988289944836932e-05,
+      "loss": 0.5213,
+      "step": 15560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6986354398528986e-05,
+      "loss": 0.3041,
+      "step": 15570
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.698441885222104e-05,
+      "loss": 0.1966,
+      "step": 15580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6982483305913097e-05,
+      "loss": 0.19,
+      "step": 15590
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.698054775960515e-05,
+      "loss": 0.2828,
+      "step": 15600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6978612213297204e-05,
+      "loss": 0.4134,
+      "step": 15610
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.697667666698926e-05,
+      "loss": 0.4587,
+      "step": 15620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.697474112068131e-05,
+      "loss": 0.2614,
+      "step": 15630
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.697280557437337e-05,
+      "loss": 0.227,
+      "step": 15640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6970870028065422e-05,
+      "loss": 0.1846,
+      "step": 15650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.696893448175748e-05,
+      "loss": 0.1985,
+      "step": 15660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6966998935449533e-05,
+      "loss": 0.3917,
+      "step": 15670
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6965063389141586e-05,
+      "loss": 0.3362,
+      "step": 15680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6963127842833643e-05,
+      "loss": 0.2642,
+      "step": 15690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6961192296525694e-05,
+      "loss": 0.2277,
+      "step": 15700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.695925675021775e-05,
+      "loss": 0.2511,
+      "step": 15710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6957321203909804e-05,
+      "loss": 0.2165,
+      "step": 15720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6955385657601858e-05,
+      "loss": 0.199,
+      "step": 15730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6953450111293915e-05,
+      "loss": 0.4836,
+      "step": 15740
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.695151456498597e-05,
+      "loss": 0.3889,
+      "step": 15750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6949579018678026e-05,
+      "loss": 0.3399,
+      "step": 15760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.694764347237008e-05,
+      "loss": 0.3039,
+      "step": 15770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6945707926062133e-05,
+      "loss": 0.2901,
+      "step": 15780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6943772379754186e-05,
+      "loss": 0.3675,
+      "step": 15790
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.694183683344624e-05,
+      "loss": 0.4347,
+      "step": 15800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6939901287138297e-05,
+      "loss": 0.2878,
+      "step": 15810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.693796574083035e-05,
+      "loss": 0.3332,
+      "step": 15820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6936030194522404e-05,
+      "loss": 0.3304,
+      "step": 15830
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.693409464821446e-05,
+      "loss": 0.309,
+      "step": 15840
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6932159101906515e-05,
+      "loss": 0.192,
+      "step": 15850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.693022355559857e-05,
+      "loss": 0.3615,
+      "step": 15860
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6928288009290622e-05,
+      "loss": 0.3228,
+      "step": 15870
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.692635246298268e-05,
+      "loss": 0.4439,
+      "step": 15880
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6924416916674733e-05,
+      "loss": 0.2694,
+      "step": 15890
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6922481370366787e-05,
+      "loss": 0.2832,
+      "step": 15900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6920545824058844e-05,
+      "loss": 0.3126,
+      "step": 15910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6918610277750897e-05,
+      "loss": 0.4237,
+      "step": 15920
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.691667473144295e-05,
+      "loss": 0.3158,
+      "step": 15930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6914739185135005e-05,
+      "loss": 0.3419,
+      "step": 15940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.691280363882706e-05,
+      "loss": 0.3258,
+      "step": 15950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6910868092519115e-05,
+      "loss": 0.3537,
+      "step": 15960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.690893254621117e-05,
+      "loss": 0.336,
+      "step": 15970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6906996999903226e-05,
+      "loss": 0.2936,
+      "step": 15980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.690506145359528e-05,
+      "loss": 0.4628,
+      "step": 15990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6903125907287333e-05,
+      "loss": 0.161,
+      "step": 16000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6901190360979387e-05,
+      "loss": 0.1847,
+      "step": 16010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.689925481467144e-05,
+      "loss": 0.3364,
+      "step": 16020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6897319268363497e-05,
+      "loss": 0.1948,
+      "step": 16030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.689538372205555e-05,
+      "loss": 0.3878,
+      "step": 16040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6893448175747608e-05,
+      "loss": 0.1859,
+      "step": 16050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6891512629439662e-05,
+      "loss": 0.2988,
+      "step": 16060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6889577083131715e-05,
+      "loss": 0.1777,
+      "step": 16070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.688764153682377e-05,
+      "loss": 0.3561,
+      "step": 16080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6885705990515823e-05,
+      "loss": 0.2426,
+      "step": 16090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.688377044420788e-05,
+      "loss": 0.4224,
+      "step": 16100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6881834897899933e-05,
+      "loss": 0.3223,
+      "step": 16110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6879899351591987e-05,
+      "loss": 0.1873,
+      "step": 16120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6877963805284044e-05,
+      "loss": 0.1635,
+      "step": 16130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6876028258976098e-05,
+      "loss": 0.3436,
+      "step": 16140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6874092712668155e-05,
+      "loss": 0.2926,
+      "step": 16150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6872157166360205e-05,
+      "loss": 0.4817,
+      "step": 16160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6870221620052262e-05,
+      "loss": 0.3496,
+      "step": 16170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6868286073744315e-05,
+      "loss": 0.2869,
+      "step": 16180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.686635052743637e-05,
+      "loss": 0.2636,
+      "step": 16190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6864414981128426e-05,
+      "loss": 0.2686,
+      "step": 16200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.686247943482048e-05,
+      "loss": 0.2625,
+      "step": 16210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6860543888512533e-05,
+      "loss": 0.3205,
+      "step": 16220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6858608342204587e-05,
+      "loss": 0.3099,
+      "step": 16230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6856672795896644e-05,
+      "loss": 0.2928,
+      "step": 16240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6854737249588698e-05,
+      "loss": 0.361,
+      "step": 16250
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.685280170328075e-05,
+      "loss": 0.4134,
+      "step": 16260
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.685086615697281e-05,
+      "loss": 0.1888,
+      "step": 16270
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6848930610664862e-05,
+      "loss": 0.3368,
+      "step": 16280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6846995064356916e-05,
+      "loss": 0.2677,
+      "step": 16290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6845059518048973e-05,
+      "loss": 0.3155,
+      "step": 16300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6843123971741023e-05,
+      "loss": 0.2505,
+      "step": 16310
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.684118842543308e-05,
+      "loss": 0.383,
+      "step": 16320
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6839252879125134e-05,
+      "loss": 0.1897,
+      "step": 16330
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.683731733281719e-05,
+      "loss": 0.2652,
+      "step": 16340
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6835381786509244e-05,
+      "loss": 0.3664,
+      "step": 16350
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6833446240201298e-05,
+      "loss": 0.2596,
+      "step": 16360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6831510693893355e-05,
+      "loss": 0.3717,
+      "step": 16370
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6829575147585405e-05,
+      "loss": 0.2709,
+      "step": 16380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6827639601277462e-05,
+      "loss": 0.1962,
+      "step": 16390
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6825704054969516e-05,
+      "loss": 0.2585,
+      "step": 16400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.682376850866157e-05,
+      "loss": 0.2822,
+      "step": 16410
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6821832962353626e-05,
+      "loss": 0.225,
+      "step": 16420
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.681989741604568e-05,
+      "loss": 0.3595,
+      "step": 16430
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6817961869737737e-05,
+      "loss": 0.4229,
+      "step": 16440
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.681602632342979e-05,
+      "loss": 0.2663,
+      "step": 16450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6814090777121844e-05,
+      "loss": 0.1621,
+      "step": 16460
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6812155230813898e-05,
+      "loss": 0.2684,
+      "step": 16470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.681021968450595e-05,
+      "loss": 0.2743,
+      "step": 16480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.680828413819801e-05,
+      "loss": 0.3048,
+      "step": 16490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6806348591890062e-05,
+      "loss": 0.3265,
+      "step": 16500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6804413045582116e-05,
+      "loss": 0.1646,
+      "step": 16510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6802477499274173e-05,
+      "loss": 0.2915,
+      "step": 16520
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6800541952966227e-05,
+      "loss": 0.4261,
+      "step": 16530
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.679860640665828e-05,
+      "loss": 0.2924,
+      "step": 16540
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6796670860350334e-05,
+      "loss": 0.3038,
+      "step": 16550
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.679473531404239e-05,
+      "loss": 0.3289,
+      "step": 16560
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6792799767734445e-05,
+      "loss": 0.2898,
+      "step": 16570
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6790864221426498e-05,
+      "loss": 0.3192,
+      "step": 16580
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6788928675118555e-05,
+      "loss": 0.2594,
+      "step": 16590
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.678699312881061e-05,
+      "loss": 0.2906,
+      "step": 16600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6785057582502662e-05,
+      "loss": 0.2275,
+      "step": 16610
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6783122036194716e-05,
+      "loss": 0.2385,
+      "step": 16620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6781186489886773e-05,
+      "loss": 0.3149,
+      "step": 16630
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6779250943578827e-05,
+      "loss": 0.3113,
+      "step": 16640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.677731539727088e-05,
+      "loss": 0.2499,
+      "step": 16650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6775379850962937e-05,
+      "loss": 0.2521,
+      "step": 16660
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.677344430465499e-05,
+      "loss": 0.4164,
+      "step": 16670
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6771508758347045e-05,
+      "loss": 0.2658,
+      "step": 16680
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.67695732120391e-05,
+      "loss": 0.2284,
+      "step": 16690
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6767637665731152e-05,
+      "loss": 0.3776,
+      "step": 16700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.676570211942321e-05,
+      "loss": 0.1443,
+      "step": 16710
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6763766573115263e-05,
+      "loss": 0.4929,
+      "step": 16720
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.676183102680732e-05,
+      "loss": 0.1865,
+      "step": 16730
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6759895480499373e-05,
+      "loss": 0.3168,
+      "step": 16740
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6757959934191427e-05,
+      "loss": 0.307,
+      "step": 16750
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.675602438788348e-05,
+      "loss": 0.3092,
+      "step": 16760
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6754088841575534e-05,
+      "loss": 0.3512,
+      "step": 16770
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.675215329526759e-05,
+      "loss": 0.2605,
+      "step": 16780
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6750217748959645e-05,
+      "loss": 0.4026,
+      "step": 16790
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.67482822026517e-05,
+      "loss": 0.2115,
+      "step": 16800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6746346656343756e-05,
+      "loss": 0.2831,
+      "step": 16810
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.674441111003581e-05,
+      "loss": 0.2923,
+      "step": 16820
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6742475563727866e-05,
+      "loss": 0.2846,
+      "step": 16830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6740540017419916e-05,
+      "loss": 0.2642,
+      "step": 16840
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6738604471111973e-05,
+      "loss": 0.2147,
+      "step": 16850
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6736668924804027e-05,
+      "loss": 0.3616,
+      "step": 16860
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.673473337849608e-05,
+      "loss": 0.2808,
+      "step": 16870
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6732797832188138e-05,
+      "loss": 0.3273,
+      "step": 16880
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.673086228588019e-05,
+      "loss": 0.4113,
+      "step": 16890
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6728926739572245e-05,
+      "loss": 0.2302,
+      "step": 16900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.67269911932643e-05,
+      "loss": 0.2401,
+      "step": 16910
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6725055646956356e-05,
+      "loss": 0.3552,
+      "step": 16920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.672312010064841e-05,
+      "loss": 0.2627,
+      "step": 16930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6721184554340463e-05,
+      "loss": 0.2984,
+      "step": 16940
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.671924900803252e-05,
+      "loss": 0.3124,
+      "step": 16950
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6717313461724574e-05,
+      "loss": 0.2215,
+      "step": 16960
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6715377915416627e-05,
+      "loss": 0.2988,
+      "step": 16970
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6713442369108684e-05,
+      "loss": 0.3898,
+      "step": 16980
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6711506822800735e-05,
+      "loss": 0.212,
+      "step": 16990
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.670957127649279e-05,
+      "loss": 0.329,
+      "step": 17000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6707635730184845e-05,
+      "loss": 0.3684,
+      "step": 17010
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6705700183876902e-05,
+      "loss": 0.2544,
+      "step": 17020
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6703764637568956e-05,
+      "loss": 0.2989,
+      "step": 17030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.670182909126101e-05,
+      "loss": 0.2157,
+      "step": 17040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6699893544953067e-05,
+      "loss": 0.2282,
+      "step": 17050
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6697957998645117e-05,
+      "loss": 0.3462,
+      "step": 17060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6696022452337174e-05,
+      "loss": 0.241,
+      "step": 17070
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6694086906029227e-05,
+      "loss": 0.3652,
+      "step": 17080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.669215135972128e-05,
+      "loss": 0.2195,
+      "step": 17090
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6690215813413338e-05,
+      "loss": 0.3775,
+      "step": 17100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6688280267105392e-05,
+      "loss": 0.2928,
+      "step": 17110
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.668634472079745e-05,
+      "loss": 0.3723,
+      "step": 17120
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6684409174489502e-05,
+      "loss": 0.2314,
+      "step": 17130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6682473628181556e-05,
+      "loss": 0.2739,
+      "step": 17140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.668053808187361e-05,
+      "loss": 0.3677,
+      "step": 17150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6678602535565663e-05,
+      "loss": 0.2582,
+      "step": 17160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.667666698925772e-05,
+      "loss": 0.2447,
+      "step": 17170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6674731442949774e-05,
+      "loss": 0.2591,
+      "step": 17180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6672795896641828e-05,
+      "loss": 0.3266,
+      "step": 17190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6670860350333885e-05,
+      "loss": 0.2254,
+      "step": 17200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6668924804025938e-05,
+      "loss": 0.2504,
+      "step": 17210
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6666989257717992e-05,
+      "loss": 0.3569,
+      "step": 17220
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6665053711410045e-05,
+      "loss": 0.4493,
+      "step": 17230
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6663118165102103e-05,
+      "loss": 0.3604,
+      "step": 17240
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6661182618794156e-05,
+      "loss": 0.3084,
+      "step": 17250
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.665924707248621e-05,
+      "loss": 0.3298,
+      "step": 17260
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6657311526178267e-05,
+      "loss": 0.1678,
+      "step": 17270
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.665537597987032e-05,
+      "loss": 0.2274,
+      "step": 17280
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6653440433562374e-05,
+      "loss": 0.4518,
+      "step": 17290
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6651504887254428e-05,
+      "loss": 0.3267,
+      "step": 17300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6649569340946485e-05,
+      "loss": 0.3082,
+      "step": 17310
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.664763379463854e-05,
+      "loss": 0.3486,
+      "step": 17320
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6645698248330592e-05,
+      "loss": 0.3719,
+      "step": 17330
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.664376270202265e-05,
+      "loss": 0.2115,
+      "step": 17340
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6641827155714703e-05,
+      "loss": 0.1718,
+      "step": 17350
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6639891609406756e-05,
+      "loss": 0.1969,
+      "step": 17360
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.663795606309881e-05,
+      "loss": 0.2087,
+      "step": 17370
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6636020516790864e-05,
+      "loss": 0.551,
+      "step": 17380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.663408497048292e-05,
+      "loss": 0.2694,
+      "step": 17390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6632149424174974e-05,
+      "loss": 0.323,
+      "step": 17400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.663021387786703e-05,
+      "loss": 0.3427,
+      "step": 17410
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6628278331559085e-05,
+      "loss": 0.2691,
+      "step": 17420
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.662634278525114e-05,
+      "loss": 0.2281,
+      "step": 17430
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6624407238943192e-05,
+      "loss": 0.2337,
+      "step": 17440
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6622471692635246e-05,
+      "loss": 0.3465,
+      "step": 17450
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6620536146327303e-05,
+      "loss": 0.2241,
+      "step": 17460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6618600600019356e-05,
+      "loss": 0.4621,
+      "step": 17470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.661666505371141e-05,
+      "loss": 0.1698,
+      "step": 17480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6614729507403467e-05,
+      "loss": 0.1191,
+      "step": 17490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.661279396109552e-05,
+      "loss": 0.2842,
+      "step": 17500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6610858414787578e-05,
+      "loss": 0.3554,
+      "step": 17510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6608922868479628e-05,
+      "loss": 0.5019,
+      "step": 17520
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6606987322171685e-05,
+      "loss": 0.3847,
+      "step": 17530
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.660505177586374e-05,
+      "loss": 0.2181,
+      "step": 17540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6603116229555792e-05,
+      "loss": 0.336,
+      "step": 17550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.660118068324785e-05,
+      "loss": 0.3242,
+      "step": 17560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6599245136939903e-05,
+      "loss": 0.408,
+      "step": 17570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6597309590631957e-05,
+      "loss": 0.3222,
+      "step": 17580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.659537404432401e-05,
+      "loss": 0.1738,
+      "step": 17590
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6593438498016067e-05,
+      "loss": 0.2368,
+      "step": 17600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.659150295170812e-05,
+      "loss": 0.3298,
+      "step": 17610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6589567405400175e-05,
+      "loss": 0.4114,
+      "step": 17620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.658763185909223e-05,
+      "loss": 0.3732,
+      "step": 17630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6585696312784285e-05,
+      "loss": 0.2467,
+      "step": 17640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.658376076647634e-05,
+      "loss": 0.2376,
+      "step": 17650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6581825220168396e-05,
+      "loss": 0.1994,
+      "step": 17660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6579889673860446e-05,
+      "loss": 0.2954,
+      "step": 17670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6577954127552503e-05,
+      "loss": 0.282,
+      "step": 17680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6576018581244557e-05,
+      "loss": 0.4242,
+      "step": 17690
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6574083034936614e-05,
+      "loss": 0.2746,
+      "step": 17700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6572147488628667e-05,
+      "loss": 0.2311,
+      "step": 17710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.657021194232072e-05,
+      "loss": 0.239,
+      "step": 17720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6568276396012778e-05,
+      "loss": 0.3641,
+      "step": 17730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.656634084970483e-05,
+      "loss": 0.3492,
+      "step": 17740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6564405303396885e-05,
+      "loss": 0.3581,
+      "step": 17750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.656246975708894e-05,
+      "loss": 0.3805,
+      "step": 17760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6560534210780993e-05,
+      "loss": 0.2485,
+      "step": 17770
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.655859866447305e-05,
+      "loss": 0.3356,
+      "step": 17780
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6556663118165103e-05,
+      "loss": 0.3991,
+      "step": 17790
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.655472757185716e-05,
+      "loss": 0.2662,
+      "step": 17800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6552792025549214e-05,
+      "loss": 0.35,
+      "step": 17810
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6550856479241268e-05,
+      "loss": 0.2963,
+      "step": 17820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.654892093293332e-05,
+      "loss": 0.2075,
+      "step": 17830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6546985386625375e-05,
+      "loss": 0.2336,
+      "step": 17840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6545049840317432e-05,
+      "loss": 0.2172,
+      "step": 17850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6543114294009486e-05,
+      "loss": 0.3327,
+      "step": 17860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.654117874770154e-05,
+      "loss": 0.2162,
+      "step": 17870
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6539243201393596e-05,
+      "loss": 0.3116,
+      "step": 17880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.653730765508565e-05,
+      "loss": 0.4425,
+      "step": 17890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6535372108777703e-05,
+      "loss": 0.2908,
+      "step": 17900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6533436562469757e-05,
+      "loss": 0.3909,
+      "step": 17910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6531501016161814e-05,
+      "loss": 0.3242,
+      "step": 17920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6529565469853868e-05,
+      "loss": 0.3519,
+      "step": 17930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.652762992354592e-05,
+      "loss": 0.2546,
+      "step": 17940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.652569437723798e-05,
+      "loss": 0.2197,
+      "step": 17950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6523758830930032e-05,
+      "loss": 0.2583,
+      "step": 17960
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6521823284622086e-05,
+      "loss": 0.231,
+      "step": 17970
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.651988773831414e-05,
+      "loss": 0.4129,
+      "step": 17980
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6517952192006196e-05,
+      "loss": 0.3499,
+      "step": 17990
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.651601664569825e-05,
+      "loss": 0.2336,
+      "step": 18000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6514081099390304e-05,
+      "loss": 0.3743,
+      "step": 18010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.651214555308236e-05,
+      "loss": 0.1599,
+      "step": 18020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6510210006774414e-05,
+      "loss": 0.2085,
+      "step": 18030
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6508274460466468e-05,
+      "loss": 0.3065,
+      "step": 18040
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.650633891415852e-05,
+      "loss": 0.2503,
+      "step": 18050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6504403367850575e-05,
+      "loss": 0.3537,
+      "step": 18060
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6502467821542632e-05,
+      "loss": 0.3136,
+      "step": 18070
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6500532275234686e-05,
+      "loss": 0.3658,
+      "step": 18080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6498596728926743e-05,
+      "loss": 0.2757,
+      "step": 18090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6496661182618796e-05,
+      "loss": 0.3217,
+      "step": 18100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.649472563631085e-05,
+      "loss": 0.3167,
+      "step": 18110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6492790090002904e-05,
+      "loss": 0.2666,
+      "step": 18120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6490854543694957e-05,
+      "loss": 0.2854,
+      "step": 18130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6488918997387014e-05,
+      "loss": 0.2103,
+      "step": 18140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6486983451079068e-05,
+      "loss": 0.2305,
+      "step": 18150
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6485047904771122e-05,
+      "loss": 0.2785,
+      "step": 18160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.648311235846318e-05,
+      "loss": 0.2928,
+      "step": 18170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6481176812155232e-05,
+      "loss": 0.2166,
+      "step": 18180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.647924126584729e-05,
+      "loss": 0.2716,
+      "step": 18190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.647730571953934e-05,
+      "loss": 0.2099,
+      "step": 18200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6475370173231397e-05,
+      "loss": 0.3413,
+      "step": 18210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.647343462692345e-05,
+      "loss": 0.293,
+      "step": 18220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6471499080615504e-05,
+      "loss": 0.2806,
+      "step": 18230
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.646956353430756e-05,
+      "loss": 0.3088,
+      "step": 18240
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6467627987999615e-05,
+      "loss": 0.3929,
+      "step": 18250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6465692441691668e-05,
+      "loss": 0.1898,
+      "step": 18260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6463756895383722e-05,
+      "loss": 0.346,
+      "step": 18270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.646182134907578e-05,
+      "loss": 0.3438,
+      "step": 18280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6459885802767833e-05,
+      "loss": 0.3033,
+      "step": 18290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6457950256459886e-05,
+      "loss": 0.2333,
+      "step": 18300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6456014710151943e-05,
+      "loss": 0.2983,
+      "step": 18310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6454079163843997e-05,
+      "loss": 0.2193,
+      "step": 18320
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.645214361753605e-05,
+      "loss": 0.3257,
+      "step": 18330
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6450208071228107e-05,
+      "loss": 0.1836,
+      "step": 18340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6448272524920158e-05,
+      "loss": 0.4733,
+      "step": 18350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6446336978612215e-05,
+      "loss": 0.314,
+      "step": 18360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.644440143230427e-05,
+      "loss": 0.2677,
+      "step": 18370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6442465885996325e-05,
+      "loss": 0.335,
+      "step": 18380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.644053033968838e-05,
+      "loss": 0.2912,
+      "step": 18390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6438594793380433e-05,
+      "loss": 0.347,
+      "step": 18400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.643665924707249e-05,
+      "loss": 0.2567,
+      "step": 18410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.643472370076454e-05,
+      "loss": 0.284,
+      "step": 18420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6432788154456597e-05,
+      "loss": 0.3342,
+      "step": 18430
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.643085260814865e-05,
+      "loss": 0.2778,
+      "step": 18440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6428917061840704e-05,
+      "loss": 0.277,
+      "step": 18450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.642698151553276e-05,
+      "loss": 0.2798,
+      "step": 18460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6425045969224815e-05,
+      "loss": 0.1895,
+      "step": 18470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6423110422916872e-05,
+      "loss": 0.3073,
+      "step": 18480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6421174876608926e-05,
+      "loss": 0.3007,
+      "step": 18490
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.641923933030098e-05,
+      "loss": 0.2726,
+      "step": 18500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6417303783993033e-05,
+      "loss": 0.336,
+      "step": 18510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6415368237685086e-05,
+      "loss": 0.3537,
+      "step": 18520
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6413432691377143e-05,
+      "loss": 0.306,
+      "step": 18530
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6411497145069197e-05,
+      "loss": 0.3821,
+      "step": 18540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.640956159876125e-05,
+      "loss": 0.4145,
+      "step": 18550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6407626052453308e-05,
+      "loss": 0.2732,
+      "step": 18560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.640569050614536e-05,
+      "loss": 0.151,
+      "step": 18570
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6403754959837415e-05,
+      "loss": 0.4677,
+      "step": 18580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.640181941352947e-05,
+      "loss": 0.3165,
+      "step": 18590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6399883867221526e-05,
+      "loss": 0.321,
+      "step": 18600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.639794832091358e-05,
+      "loss": 0.2441,
+      "step": 18610
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6396012774605633e-05,
+      "loss": 0.3293,
+      "step": 18620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.639407722829769e-05,
+      "loss": 0.214,
+      "step": 18630
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6392141681989744e-05,
+      "loss": 0.2849,
+      "step": 18640
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6390206135681797e-05,
+      "loss": 0.2968,
+      "step": 18650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.638827058937385e-05,
+      "loss": 0.348,
+      "step": 18660
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6386335043065908e-05,
+      "loss": 0.3481,
+      "step": 18670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.638439949675796e-05,
+      "loss": 0.2479,
+      "step": 18680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6382463950450015e-05,
+      "loss": 0.3407,
+      "step": 18690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6380528404142072e-05,
+      "loss": 0.2994,
+      "step": 18700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6378592857834126e-05,
+      "loss": 0.2279,
+      "step": 18710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.637665731152618e-05,
+      "loss": 0.3436,
+      "step": 18720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6374721765218233e-05,
+      "loss": 0.2652,
+      "step": 18730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6372786218910287e-05,
+      "loss": 0.2188,
+      "step": 18740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6370850672602344e-05,
+      "loss": 0.1717,
+      "step": 18750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6368915126294397e-05,
+      "loss": 0.2485,
+      "step": 18760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6366979579986454e-05,
+      "loss": 0.2834,
+      "step": 18770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6365044033678508e-05,
+      "loss": 0.4029,
+      "step": 18780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6363108487370562e-05,
+      "loss": 0.3681,
+      "step": 18790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6361172941062615e-05,
+      "loss": 0.2701,
+      "step": 18800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.635923739475467e-05,
+      "loss": 0.2573,
+      "step": 18810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6357301848446726e-05,
+      "loss": 0.3545,
+      "step": 18820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.635536630213878e-05,
+      "loss": 0.4568,
+      "step": 18830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6353430755830833e-05,
+      "loss": 0.2919,
+      "step": 18840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.635149520952289e-05,
+      "loss": 0.3349,
+      "step": 18850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6349559663214944e-05,
+      "loss": 0.274,
+      "step": 18860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6347624116907e-05,
+      "loss": 0.2792,
+      "step": 18870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.634568857059905e-05,
+      "loss": 0.4413,
+      "step": 18880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6343753024291108e-05,
+      "loss": 0.3486,
+      "step": 18890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6341817477983162e-05,
+      "loss": 0.1878,
+      "step": 18900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6339881931675216e-05,
+      "loss": 0.2555,
+      "step": 18910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6337946385367273e-05,
+      "loss": 0.4034,
+      "step": 18920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6336010839059326e-05,
+      "loss": 0.2576,
+      "step": 18930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.633407529275138e-05,
+      "loss": 0.3104,
+      "step": 18940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6332139746443433e-05,
+      "loss": 0.355,
+      "step": 18950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.633020420013549e-05,
+      "loss": 0.3288,
+      "step": 18960
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6328268653827544e-05,
+      "loss": 0.2096,
+      "step": 18970
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6326333107519598e-05,
+      "loss": 0.3618,
+      "step": 18980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6324397561211655e-05,
+      "loss": 0.1729,
+      "step": 18990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.632246201490371e-05,
+      "loss": 0.2401,
+      "step": 19000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6320526468595762e-05,
+      "loss": 0.1676,
+      "step": 19010
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.631859092228782e-05,
+      "loss": 0.2947,
+      "step": 19020
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.631665537597987e-05,
+      "loss": 0.3383,
+      "step": 19030
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6314719829671926e-05,
+      "loss": 0.3709,
+      "step": 19040
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.631278428336398e-05,
+      "loss": 0.2779,
+      "step": 19050
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6310848737056037e-05,
+      "loss": 0.3207,
+      "step": 19060
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.630891319074809e-05,
+      "loss": 0.157,
+      "step": 19070
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6306977644440144e-05,
+      "loss": 0.447,
+      "step": 19080
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.63050420981322e-05,
+      "loss": 0.281,
+      "step": 19090
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6303106551824255e-05,
+      "loss": 0.3323,
+      "step": 19100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.630117100551631e-05,
+      "loss": 0.2207,
+      "step": 19110
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6299235459208362e-05,
+      "loss": 0.2224,
+      "step": 19120
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6297299912900416e-05,
+      "loss": 0.4151,
+      "step": 19130
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6295364366592473e-05,
+      "loss": 0.2612,
+      "step": 19140
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6293428820284526e-05,
+      "loss": 0.2603,
+      "step": 19150
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6291493273976584e-05,
+      "loss": 0.1827,
+      "step": 19160
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6289557727668637e-05,
+      "loss": 0.329,
+      "step": 19170
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.628762218136069e-05,
+      "loss": 0.3778,
+      "step": 19180
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6285686635052744e-05,
+      "loss": 0.3084,
+      "step": 19190
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6283751088744798e-05,
+      "loss": 0.3448,
+      "step": 19200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6281815542436855e-05,
+      "loss": 0.396,
+      "step": 19210
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.627987999612891e-05,
+      "loss": 0.1541,
+      "step": 19220
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6277944449820962e-05,
+      "loss": 0.4176,
+      "step": 19230
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.627600890351302e-05,
+      "loss": 0.2442,
+      "step": 19240
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6274073357205073e-05,
+      "loss": 0.4129,
+      "step": 19250
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6272137810897127e-05,
+      "loss": 0.2672,
+      "step": 19260
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.627020226458918e-05,
+      "loss": 0.2013,
+      "step": 19270
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6268266718281237e-05,
+      "loss": 0.2055,
+      "step": 19280
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.626633117197329e-05,
+      "loss": 0.3049,
+      "step": 19290
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6264395625665345e-05,
+      "loss": 0.2785,
+      "step": 19300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.62624600793574e-05,
+      "loss": 0.307,
+      "step": 19310
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6260524533049455e-05,
+      "loss": 0.3223,
+      "step": 19320
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.625858898674151e-05,
+      "loss": 0.4327,
+      "step": 19330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6256653440433563e-05,
+      "loss": 0.3987,
+      "step": 19340
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.625471789412562e-05,
+      "loss": 0.3321,
+      "step": 19350
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6252782347817673e-05,
+      "loss": 0.3003,
+      "step": 19360
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6250846801509727e-05,
+      "loss": 0.2172,
+      "step": 19370
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6248911255201784e-05,
+      "loss": 0.3768,
+      "step": 19380
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6246975708893837e-05,
+      "loss": 0.2885,
+      "step": 19390
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.624504016258589e-05,
+      "loss": 0.3331,
+      "step": 19400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6243104616277945e-05,
+      "loss": 0.2489,
+      "step": 19410
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.624116906997e-05,
+      "loss": 0.2379,
+      "step": 19420
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6239233523662055e-05,
+      "loss": 0.3037,
+      "step": 19430
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.623729797735411e-05,
+      "loss": 0.3502,
+      "step": 19440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6235362431046166e-05,
+      "loss": 0.3181,
+      "step": 19450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.623342688473822e-05,
+      "loss": 0.2894,
+      "step": 19460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6231491338430273e-05,
+      "loss": 0.3452,
+      "step": 19470
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6229555792122327e-05,
+      "loss": 0.2964,
+      "step": 19480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.622762024581438e-05,
+      "loss": 0.333,
+      "step": 19490
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6225684699506438e-05,
+      "loss": 0.3493,
+      "step": 19500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.622374915319849e-05,
+      "loss": 0.4679,
+      "step": 19510
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6221813606890545e-05,
+      "loss": 0.3884,
+      "step": 19520
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6219878060582602e-05,
+      "loss": 0.4092,
+      "step": 19530
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6217942514274656e-05,
+      "loss": 0.2879,
+      "step": 19540
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6216006967966713e-05,
+      "loss": 0.2012,
+      "step": 19550
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6214071421658763e-05,
+      "loss": 0.244,
+      "step": 19560
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.621213587535082e-05,
+      "loss": 0.3397,
+      "step": 19570
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6210200329042873e-05,
+      "loss": 0.2582,
+      "step": 19580
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6208264782734927e-05,
+      "loss": 0.2718,
+      "step": 19590
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6206329236426984e-05,
+      "loss": 0.2351,
+      "step": 19600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6204393690119038e-05,
+      "loss": 0.2996,
+      "step": 19610
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.620245814381109e-05,
+      "loss": 0.2787,
+      "step": 19620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6200522597503145e-05,
+      "loss": 0.2997,
+      "step": 19630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.61985870511952e-05,
+      "loss": 0.2313,
+      "step": 19640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6196651504887256e-05,
+      "loss": 0.3736,
+      "step": 19650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.619471595857931e-05,
+      "loss": 0.2657,
+      "step": 19660
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6192780412271366e-05,
+      "loss": 0.1712,
+      "step": 19670
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.619084486596342e-05,
+      "loss": 0.3004,
+      "step": 19680
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6188909319655474e-05,
+      "loss": 0.4106,
+      "step": 19690
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.618697377334753e-05,
+      "loss": 0.2653,
+      "step": 19700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.618503822703958e-05,
+      "loss": 0.2902,
+      "step": 19710
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6183102680731638e-05,
+      "loss": 0.5352,
+      "step": 19720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.618116713442369e-05,
+      "loss": 0.2977,
+      "step": 19730
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.617923158811575e-05,
+      "loss": 0.2813,
+      "step": 19740
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6177296041807802e-05,
+      "loss": 0.28,
+      "step": 19750
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6175360495499856e-05,
+      "loss": 0.3155,
+      "step": 19760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6173424949191913e-05,
+      "loss": 0.3484,
+      "step": 19770
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6171489402883967e-05,
+      "loss": 0.3432,
+      "step": 19780
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.616955385657602e-05,
+      "loss": 0.2154,
+      "step": 19790
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6167618310268074e-05,
+      "loss": 0.1846,
+      "step": 19800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6165682763960127e-05,
+      "loss": 0.2863,
+      "step": 19810
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6163747217652184e-05,
+      "loss": 0.2513,
+      "step": 19820
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6161811671344238e-05,
+      "loss": 0.2739,
+      "step": 19830
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6159876125036295e-05,
+      "loss": 0.1992,
+      "step": 19840
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.615794057872835e-05,
+      "loss": 0.4048,
+      "step": 19850
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6156005032420402e-05,
+      "loss": 0.3042,
+      "step": 19860
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6154069486112456e-05,
+      "loss": 0.3278,
+      "step": 19870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.615213393980451e-05,
+      "loss": 0.1881,
+      "step": 19880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6150198393496567e-05,
+      "loss": 0.3691,
+      "step": 19890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.614826284718862e-05,
+      "loss": 0.3856,
+      "step": 19900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6146327300880674e-05,
+      "loss": 0.2138,
+      "step": 19910
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.614439175457273e-05,
+      "loss": 0.1669,
+      "step": 19920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6142456208264785e-05,
+      "loss": 0.2613,
+      "step": 19930
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6140520661956838e-05,
+      "loss": 0.37,
+      "step": 19940
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6138585115648892e-05,
+      "loss": 0.4984,
+      "step": 19950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.613664956934095e-05,
+      "loss": 0.3017,
+      "step": 19960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6134714023033003e-05,
+      "loss": 0.3092,
+      "step": 19970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6132778476725056e-05,
+      "loss": 0.3016,
+      "step": 19980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6130842930417113e-05,
+      "loss": 0.4102,
+      "step": 19990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6128907384109167e-05,
+      "loss": 0.3715,
+      "step": 20000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.612697183780122e-05,
+      "loss": 0.2694,
+      "step": 20010
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6125036291493274e-05,
+      "loss": 0.3026,
+      "step": 20020
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6123100745185328e-05,
+      "loss": 0.2734,
+      "step": 20030
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6121165198877385e-05,
+      "loss": 0.2112,
+      "step": 20040
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.611922965256944e-05,
+      "loss": 0.4955,
+      "step": 20050
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6117294106261495e-05,
+      "loss": 0.304,
+      "step": 20060
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.611535855995355e-05,
+      "loss": 0.328,
+      "step": 20070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6113423013645603e-05,
+      "loss": 0.335,
+      "step": 20080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6111487467337656e-05,
+      "loss": 0.2388,
+      "step": 20090
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.610955192102971e-05,
+      "loss": 0.27,
+      "step": 20100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6107616374721767e-05,
+      "loss": 0.4306,
+      "step": 20110
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.610568082841382e-05,
+      "loss": 0.3143,
+      "step": 20120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6103745282105874e-05,
+      "loss": 0.2758,
+      "step": 20130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.610180973579793e-05,
+      "loss": 0.2612,
+      "step": 20140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6099874189489985e-05,
+      "loss": 0.2335,
+      "step": 20150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.609793864318204e-05,
+      "loss": 0.3537,
+      "step": 20160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6096003096874092e-05,
+      "loss": 0.2725,
+      "step": 20170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.609406755056615e-05,
+      "loss": 0.2461,
+      "step": 20180
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6092132004258203e-05,
+      "loss": 0.3714,
+      "step": 20190
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6090196457950256e-05,
+      "loss": 0.2012,
+      "step": 20200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6088260911642314e-05,
+      "loss": 0.281,
+      "step": 20210
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6086325365334367e-05,
+      "loss": 0.2717,
+      "step": 20220
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6084389819026424e-05,
+      "loss": 0.2419,
+      "step": 20230
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6082454272718474e-05,
+      "loss": 0.3028,
+      "step": 20240
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.608051872641053e-05,
+      "loss": 0.3571,
+      "step": 20250
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6078583180102585e-05,
+      "loss": 0.3057,
+      "step": 20260
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.607664763379464e-05,
+      "loss": 0.2074,
+      "step": 20270
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6074712087486696e-05,
+      "loss": 0.338,
+      "step": 20280
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.607277654117875e-05,
+      "loss": 0.2173,
+      "step": 20290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6070840994870803e-05,
+      "loss": 0.3407,
+      "step": 20300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.606890544856286e-05,
+      "loss": 0.2563,
+      "step": 20310
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.606696990225491e-05,
+      "loss": 0.2561,
+      "step": 20320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6065034355946967e-05,
+      "loss": 0.2578,
+      "step": 20330
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.606309880963902e-05,
+      "loss": 0.2365,
+      "step": 20340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6061163263331078e-05,
+      "loss": 0.2707,
+      "step": 20350
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.605922771702313e-05,
+      "loss": 0.3317,
+      "step": 20360
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6057292170715185e-05,
+      "loss": 0.4035,
+      "step": 20370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6055356624407242e-05,
+      "loss": 0.4422,
+      "step": 20380
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6053421078099293e-05,
+      "loss": 0.1566,
+      "step": 20390
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.605148553179135e-05,
+      "loss": 0.2934,
+      "step": 20400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6049549985483403e-05,
+      "loss": 0.2853,
+      "step": 20410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6047614439175457e-05,
+      "loss": 0.1882,
+      "step": 20420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6045678892867514e-05,
+      "loss": 0.3031,
+      "step": 20430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6043743346559567e-05,
+      "loss": 0.2034,
+      "step": 20440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6041807800251624e-05,
+      "loss": 0.3468,
+      "step": 20450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6039872253943678e-05,
+      "loss": 0.4122,
+      "step": 20460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6037936707635732e-05,
+      "loss": 0.2875,
+      "step": 20470
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6036001161327785e-05,
+      "loss": 0.2982,
+      "step": 20480
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.603406561501984e-05,
+      "loss": 0.3036,
+      "step": 20490
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6032130068711896e-05,
+      "loss": 0.2092,
+      "step": 20500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.603019452240395e-05,
+      "loss": 0.2688,
+      "step": 20510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6028258976096003e-05,
+      "loss": 0.3737,
+      "step": 20520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.602632342978806e-05,
+      "loss": 0.2973,
+      "step": 20530
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6024387883480114e-05,
+      "loss": 0.3122,
+      "step": 20540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6022452337172168e-05,
+      "loss": 0.2608,
+      "step": 20550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.602051679086422e-05,
+      "loss": 0.3414,
+      "step": 20560
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6018581244556278e-05,
+      "loss": 0.2154,
+      "step": 20570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6016645698248332e-05,
+      "loss": 0.33,
+      "step": 20580
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6014710151940386e-05,
+      "loss": 0.3083,
+      "step": 20590
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6012774605632443e-05,
+      "loss": 0.3268,
+      "step": 20600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6010839059324496e-05,
+      "loss": 0.3215,
+      "step": 20610
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.600890351301655e-05,
+      "loss": 0.3121,
+      "step": 20620
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6006967966708603e-05,
+      "loss": 0.4392,
+      "step": 20630
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.600503242040066e-05,
+      "loss": 0.2084,
+      "step": 20640
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6003096874092714e-05,
+      "loss": 0.2343,
+      "step": 20650
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6001161327784768e-05,
+      "loss": 0.2889,
+      "step": 20660
+    },
+    {
+      "epoch": 2.0,
+      "eval_FN": 919,
+      "eval_FP": 1130,
+      "eval_TN": 14629,
+      "eval_TP": 3987,
+      "eval_accuracy": 0.9008468424872974,
+      "eval_f1": 0.7955701885662976,
+      "eval_loss": 0.34464871883392334,
+      "eval_precision": 0.7791674809458667,
+      "eval_recall": 0.8126783530370975,
+      "eval_runtime": 142.084,
+      "eval_samples_per_second": 145.442,
+      "eval_steps_per_second": 9.093,
+      "step": 20666
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5999225781476825e-05,
+      "loss": 0.3432,
+      "step": 20670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.599729023516888e-05,
+      "loss": 0.2342,
+      "step": 20680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5995354688860932e-05,
+      "loss": 0.3926,
+      "step": 20690
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5993419142552986e-05,
+      "loss": 0.3598,
+      "step": 20700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.599148359624504e-05,
+      "loss": 0.2192,
+      "step": 20710
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5989548049937096e-05,
+      "loss": 0.2865,
+      "step": 20720
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.598761250362915e-05,
+      "loss": 0.2612,
+      "step": 20730
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5985676957321207e-05,
+      "loss": 0.3038,
+      "step": 20740
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.598374141101326e-05,
+      "loss": 0.2026,
+      "step": 20750
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5981805864705314e-05,
+      "loss": 0.3326,
+      "step": 20760
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5979870318397368e-05,
+      "loss": 0.2022,
+      "step": 20770
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.597793477208942e-05,
+      "loss": 0.2445,
+      "step": 20780
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.597599922578148e-05,
+      "loss": 0.2519,
+      "step": 20790
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5974063679473532e-05,
+      "loss": 0.2906,
+      "step": 20800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5972128133165586e-05,
+      "loss": 0.2045,
+      "step": 20810
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5970192586857643e-05,
+      "loss": 0.346,
+      "step": 20820
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5968257040549697e-05,
+      "loss": 0.3019,
+      "step": 20830
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.596632149424175e-05,
+      "loss": 0.2392,
+      "step": 20840
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5964385947933804e-05,
+      "loss": 0.2449,
+      "step": 20850
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.596245040162586e-05,
+      "loss": 0.2073,
+      "step": 20860
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5960514855317914e-05,
+      "loss": 0.3339,
+      "step": 20870
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5958579309009968e-05,
+      "loss": 0.4166,
+      "step": 20880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5956643762702025e-05,
+      "loss": 0.2809,
+      "step": 20890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.595470821639408e-05,
+      "loss": 0.1807,
+      "step": 20900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5952772670086132e-05,
+      "loss": 0.2121,
+      "step": 20910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5950837123778186e-05,
+      "loss": 0.1771,
+      "step": 20920
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5948901577470243e-05,
+      "loss": 0.2304,
+      "step": 20930
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5946966031162297e-05,
+      "loss": 0.2765,
+      "step": 20940
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.594503048485435e-05,
+      "loss": 0.249,
+      "step": 20950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5943094938546407e-05,
+      "loss": 0.2479,
+      "step": 20960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.594115939223846e-05,
+      "loss": 0.231,
+      "step": 20970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5939223845930515e-05,
+      "loss": 0.3658,
+      "step": 20980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.593728829962257e-05,
+      "loss": 0.2716,
+      "step": 20990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5935352753314622e-05,
+      "loss": 0.2607,
+      "step": 21000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.593341720700668e-05,
+      "loss": 0.3657,
+      "step": 21010
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5931481660698733e-05,
+      "loss": 0.2209,
+      "step": 21020
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.592954611439079e-05,
+      "loss": 0.3332,
+      "step": 21030
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5927610568082843e-05,
+      "loss": 0.3315,
+      "step": 21040
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5925675021774897e-05,
+      "loss": 0.3444,
+      "step": 21050
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5923739475466954e-05,
+      "loss": 0.2668,
+      "step": 21060
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5921803929159004e-05,
+      "loss": 0.2203,
+      "step": 21070
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.591986838285106e-05,
+      "loss": 0.2921,
+      "step": 21080
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5917932836543115e-05,
+      "loss": 0.3581,
+      "step": 21090
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.591599729023517e-05,
+      "loss": 0.2976,
+      "step": 21100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5914061743927225e-05,
+      "loss": 0.2713,
+      "step": 21110
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.591212619761928e-05,
+      "loss": 0.2514,
+      "step": 21120
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5910190651311336e-05,
+      "loss": 0.1008,
+      "step": 21130
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.590825510500339e-05,
+      "loss": 0.4273,
+      "step": 21140
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5906319558695443e-05,
+      "loss": 0.3706,
+      "step": 21150
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5904384012387497e-05,
+      "loss": 0.2325,
+      "step": 21160
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.590244846607955e-05,
+      "loss": 0.3872,
+      "step": 21170
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5900512919771608e-05,
+      "loss": 0.2218,
+      "step": 21180
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.589857737346366e-05,
+      "loss": 0.2195,
+      "step": 21190
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5896641827155715e-05,
+      "loss": 0.2542,
+      "step": 21200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5894706280847772e-05,
+      "loss": 0.2097,
+      "step": 21210
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5892770734539826e-05,
+      "loss": 0.31,
+      "step": 21220
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.589083518823188e-05,
+      "loss": 0.2332,
+      "step": 21230
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5888899641923933e-05,
+      "loss": 0.2174,
+      "step": 21240
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.588696409561599e-05,
+      "loss": 0.2457,
+      "step": 21250
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5885028549308044e-05,
+      "loss": 0.2047,
+      "step": 21260
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5883093003000097e-05,
+      "loss": 0.2354,
+      "step": 21270
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5881157456692154e-05,
+      "loss": 0.4077,
+      "step": 21280
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5879221910384208e-05,
+      "loss": 0.4172,
+      "step": 21290
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.587728636407626e-05,
+      "loss": 0.3534,
+      "step": 21300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5875350817768315e-05,
+      "loss": 0.2571,
+      "step": 21310
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5873415271460372e-05,
+      "loss": 0.2064,
+      "step": 21320
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5871479725152426e-05,
+      "loss": 0.1528,
+      "step": 21330
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.586954417884448e-05,
+      "loss": 0.2901,
+      "step": 21340
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5867608632536536e-05,
+      "loss": 0.4629,
+      "step": 21350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.586567308622859e-05,
+      "loss": 0.1366,
+      "step": 21360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5863737539920644e-05,
+      "loss": 0.3023,
+      "step": 21370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5861801993612697e-05,
+      "loss": 0.245,
+      "step": 21380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.585986644730475e-05,
+      "loss": 0.2692,
+      "step": 21390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5857930900996808e-05,
+      "loss": 0.1231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.585599535468886e-05,
+      "loss": 0.3895,
+      "step": 21410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.585405980838092e-05,
+      "loss": 0.1967,
+      "step": 21420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5852124262072972e-05,
+      "loss": 0.3142,
+      "step": 21430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5850188715765026e-05,
+      "loss": 0.2888,
+      "step": 21440
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.584825316945708e-05,
+      "loss": 0.4153,
+      "step": 21450
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5846317623149133e-05,
+      "loss": 0.2585,
+      "step": 21460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.584438207684119e-05,
+      "loss": 0.2564,
+      "step": 21470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5842446530533244e-05,
+      "loss": 0.251,
+      "step": 21480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5840510984225297e-05,
+      "loss": 0.1714,
+      "step": 21490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5838575437917354e-05,
+      "loss": 0.4766,
+      "step": 21500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5836639891609408e-05,
+      "loss": 0.2469,
+      "step": 21510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5834704345301465e-05,
+      "loss": 0.2161,
+      "step": 21520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5832768798993515e-05,
+      "loss": 0.2216,
+      "step": 21530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5830833252685572e-05,
+      "loss": 0.2184,
+      "step": 21540
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5828897706377626e-05,
+      "loss": 0.2276,
+      "step": 21550
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.582696216006968e-05,
+      "loss": 0.1708,
+      "step": 21560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5825026613761737e-05,
+      "loss": 0.3663,
+      "step": 21570
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.582309106745379e-05,
+      "loss": 0.4665,
+      "step": 21580
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5821155521145844e-05,
+      "loss": 0.1806,
+      "step": 21590
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5819219974837898e-05,
+      "loss": 0.2338,
+      "step": 21600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5817284428529955e-05,
+      "loss": 0.2788,
+      "step": 21610
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5815348882222008e-05,
+      "loss": 0.203,
+      "step": 21620
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5813413335914062e-05,
+      "loss": 0.2149,
+      "step": 21630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.581147778960612e-05,
+      "loss": 0.2371,
+      "step": 21640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5809542243298173e-05,
+      "loss": 0.5227,
+      "step": 21650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5807606696990226e-05,
+      "loss": 0.1922,
+      "step": 21660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5805671150682283e-05,
+      "loss": 0.256,
+      "step": 21670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5803735604374333e-05,
+      "loss": 0.4526,
+      "step": 21680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.580180005806639e-05,
+      "loss": 0.2586,
+      "step": 21690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5799864511758444e-05,
+      "loss": 0.1447,
+      "step": 21700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.57979289654505e-05,
+      "loss": 0.248,
+      "step": 21710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5795993419142555e-05,
+      "loss": 0.3199,
+      "step": 21720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.579405787283461e-05,
+      "loss": 0.3233,
+      "step": 21730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5792122326526665e-05,
+      "loss": 0.1917,
+      "step": 21740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5790186780218716e-05,
+      "loss": 0.3776,
+      "step": 21750
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5788251233910773e-05,
+      "loss": 0.2117,
+      "step": 21760
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5786315687602826e-05,
+      "loss": 0.1987,
+      "step": 21770
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.578438014129488e-05,
+      "loss": 0.3366,
+      "step": 21780
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5782444594986937e-05,
+      "loss": 0.2574,
+      "step": 21790
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.578050904867899e-05,
+      "loss": 0.3858,
+      "step": 21800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5778573502371048e-05,
+      "loss": 0.2742,
+      "step": 21810
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.57766379560631e-05,
+      "loss": 0.264,
+      "step": 21820
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5774702409755155e-05,
+      "loss": 0.3125,
+      "step": 21830
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.577276686344721e-05,
+      "loss": 0.321,
+      "step": 21840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5770831317139262e-05,
+      "loss": 0.3371,
+      "step": 21850
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.576889577083132e-05,
+      "loss": 0.1889,
+      "step": 21860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5766960224523373e-05,
+      "loss": 0.2423,
+      "step": 21870
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5765024678215427e-05,
+      "loss": 0.1854,
+      "step": 21880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5763089131907484e-05,
+      "loss": 0.2817,
+      "step": 21890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5761153585599537e-05,
+      "loss": 0.1499,
+      "step": 21900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.575921803929159e-05,
+      "loss": 0.2647,
+      "step": 21910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5757282492983644e-05,
+      "loss": 0.4387,
+      "step": 21920
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.57553469466757e-05,
+      "loss": 0.226,
+      "step": 21930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5753411400367755e-05,
+      "loss": 0.192,
+      "step": 21940
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.575147585405981e-05,
+      "loss": 0.2184,
+      "step": 21950
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5749540307751866e-05,
+      "loss": 0.3275,
+      "step": 21960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.574760476144392e-05,
+      "loss": 0.336,
+      "step": 21970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5745669215135973e-05,
+      "loss": 0.229,
+      "step": 21980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5743733668828027e-05,
+      "loss": 0.2626,
+      "step": 21990
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5741798122520084e-05,
+      "loss": 0.193,
+      "step": 22000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5739862576212137e-05,
+      "loss": 0.1983,
+      "step": 22010
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.573792702990419e-05,
+      "loss": 0.2912,
+      "step": 22020
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5735991483596248e-05,
+      "loss": 0.1933,
+      "step": 22030
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.57340559372883e-05,
+      "loss": 0.4074,
+      "step": 22040
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5732120390980355e-05,
+      "loss": 0.1966,
+      "step": 22050
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.573018484467241e-05,
+      "loss": 0.2,
+      "step": 22060
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5728249298364463e-05,
+      "loss": 0.2004,
+      "step": 22070
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.572631375205652e-05,
+      "loss": 0.2468,
+      "step": 22080
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5724378205748573e-05,
+      "loss": 0.1423,
+      "step": 22090
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.572244265944063e-05,
+      "loss": 0.2409,
+      "step": 22100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5720507113132684e-05,
+      "loss": 0.1016,
+      "step": 22110
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5718571566824737e-05,
+      "loss": 0.1909,
+      "step": 22120
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.571663602051679e-05,
+      "loss": 0.2789,
+      "step": 22130
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5714700474208845e-05,
+      "loss": 0.3647,
+      "step": 22140
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5712764927900902e-05,
+      "loss": 0.361,
+      "step": 22150
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5710829381592955e-05,
+      "loss": 0.3072,
+      "step": 22160
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.570889383528501e-05,
+      "loss": 0.3066,
+      "step": 22170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5706958288977066e-05,
+      "loss": 0.2989,
+      "step": 22180
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.570502274266912e-05,
+      "loss": 0.2243,
+      "step": 22190
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5703087196361177e-05,
+      "loss": 0.3565,
+      "step": 22200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5701151650053227e-05,
+      "loss": 0.3425,
+      "step": 22210
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5699216103745284e-05,
+      "loss": 0.2412,
+      "step": 22220
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5697280557437338e-05,
+      "loss": 0.1828,
+      "step": 22230
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.569534501112939e-05,
+      "loss": 0.3325,
+      "step": 22240
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5693409464821448e-05,
+      "loss": 0.2853,
+      "step": 22250
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5691473918513502e-05,
+      "loss": 0.16,
+      "step": 22260
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5689538372205556e-05,
+      "loss": 0.3891,
+      "step": 22270
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.568760282589761e-05,
+      "loss": 0.3791,
+      "step": 22280
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5685667279589666e-05,
+      "loss": 0.3009,
+      "step": 22290
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.568373173328172e-05,
+      "loss": 0.1741,
+      "step": 22300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5681796186973774e-05,
+      "loss": 0.1853,
+      "step": 22310
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.567986064066583e-05,
+      "loss": 0.2431,
+      "step": 22320
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5677925094357884e-05,
+      "loss": 0.2311,
+      "step": 22330
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5675989548049938e-05,
+      "loss": 0.2238,
+      "step": 22340
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5674054001741995e-05,
+      "loss": 0.3027,
+      "step": 22350
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5672118455434045e-05,
+      "loss": 0.2233,
+      "step": 22360
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5670182909126102e-05,
+      "loss": 0.2132,
+      "step": 22370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5668247362818156e-05,
+      "loss": 0.2421,
+      "step": 22380
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5666311816510213e-05,
+      "loss": 0.2628,
+      "step": 22390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5664376270202266e-05,
+      "loss": 0.3292,
+      "step": 22400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.566244072389432e-05,
+      "loss": 0.1986,
+      "step": 22410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5660505177586377e-05,
+      "loss": 0.3264,
+      "step": 22420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5658569631278427e-05,
+      "loss": 0.2737,
+      "step": 22430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5656634084970484e-05,
+      "loss": 0.167,
+      "step": 22440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5654698538662538e-05,
+      "loss": 0.3104,
+      "step": 22450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.565276299235459e-05,
+      "loss": 0.233,
+      "step": 22460
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.565082744604665e-05,
+      "loss": 0.2926,
+      "step": 22470
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5648891899738702e-05,
+      "loss": 0.1775,
+      "step": 22480
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.564695635343076e-05,
+      "loss": 0.3102,
+      "step": 22490
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5645020807122813e-05,
+      "loss": 0.2869,
+      "step": 22500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5643085260814867e-05,
+      "loss": 0.3027,
+      "step": 22510
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.564114971450692e-05,
+      "loss": 0.1698,
+      "step": 22520
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5639214168198974e-05,
+      "loss": 0.2865,
+      "step": 22530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.563727862189103e-05,
+      "loss": 0.4138,
+      "step": 22540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5635343075583084e-05,
+      "loss": 0.2366,
+      "step": 22550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5633407529275138e-05,
+      "loss": 0.1658,
+      "step": 22560
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5631471982967195e-05,
+      "loss": 0.3825,
+      "step": 22570
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.562953643665925e-05,
+      "loss": 0.251,
+      "step": 22580
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5627600890351302e-05,
+      "loss": 0.2205,
+      "step": 22590
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5625665344043356e-05,
+      "loss": 0.2218,
+      "step": 22600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5623729797735413e-05,
+      "loss": 0.3052,
+      "step": 22610
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5621794251427467e-05,
+      "loss": 0.3055,
+      "step": 22620
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.561985870511952e-05,
+      "loss": 0.2835,
+      "step": 22630
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5617923158811577e-05,
+      "loss": 0.3201,
+      "step": 22640
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.561598761250363e-05,
+      "loss": 0.2993,
+      "step": 22650
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5614052066195685e-05,
+      "loss": 0.2218,
+      "step": 22660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5612116519887738e-05,
+      "loss": 0.1821,
+      "step": 22670
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5610180973579795e-05,
+      "loss": 0.3121,
+      "step": 22680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.560824542727185e-05,
+      "loss": 0.2111,
+      "step": 22690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5606309880963903e-05,
+      "loss": 0.4287,
+      "step": 22700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.560437433465596e-05,
+      "loss": 0.2586,
+      "step": 22710
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5602438788348013e-05,
+      "loss": 0.3464,
+      "step": 22720
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5600503242040067e-05,
+      "loss": 0.2257,
+      "step": 22730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.559856769573212e-05,
+      "loss": 0.2017,
+      "step": 22740
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5596632149424174e-05,
+      "loss": 0.2797,
+      "step": 22750
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.559469660311623e-05,
+      "loss": 0.2968,
+      "step": 22760
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5592761056808285e-05,
+      "loss": 0.1737,
+      "step": 22770
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5590825510500342e-05,
+      "loss": 0.2217,
+      "step": 22780
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5588889964192395e-05,
+      "loss": 0.2424,
+      "step": 22790
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.558695441788445e-05,
+      "loss": 0.1722,
+      "step": 22800
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5585018871576503e-05,
+      "loss": 0.2643,
+      "step": 22810
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5583083325268556e-05,
+      "loss": 0.411,
+      "step": 22820
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5581147778960613e-05,
+      "loss": 0.2698,
+      "step": 22830
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5579212232652667e-05,
+      "loss": 0.2355,
+      "step": 22840
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.557727668634472e-05,
+      "loss": 0.2695,
+      "step": 22850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5575341140036778e-05,
+      "loss": 0.2977,
+      "step": 22860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.557340559372883e-05,
+      "loss": 0.245,
+      "step": 22870
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.557147004742089e-05,
+      "loss": 0.24,
+      "step": 22880
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.556953450111294e-05,
+      "loss": 0.2836,
+      "step": 22890
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5567598954804996e-05,
+      "loss": 0.2545,
+      "step": 22900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.556566340849705e-05,
+      "loss": 0.1615,
+      "step": 22910
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5563727862189103e-05,
+      "loss": 0.2826,
+      "step": 22920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.556179231588116e-05,
+      "loss": 0.2515,
+      "step": 22930
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5559856769573214e-05,
+      "loss": 0.4158,
+      "step": 22940
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5557921223265267e-05,
+      "loss": 0.4195,
+      "step": 22950
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.555598567695732e-05,
+      "loss": 0.3761,
+      "step": 22960
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5554050130649378e-05,
+      "loss": 0.1775,
+      "step": 22970
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.555211458434143e-05,
+      "loss": 0.2074,
+      "step": 22980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5550179038033485e-05,
+      "loss": 0.1675,
+      "step": 22990
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5548243491725542e-05,
+      "loss": 0.1886,
+      "step": 23000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5546307945417596e-05,
+      "loss": 0.2923,
+      "step": 23010
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.554437239910965e-05,
+      "loss": 0.3487,
+      "step": 23020
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5542436852801706e-05,
+      "loss": 0.2838,
+      "step": 23030
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5540501306493757e-05,
+      "loss": 0.2118,
+      "step": 23040
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5538565760185814e-05,
+      "loss": 0.2534,
+      "step": 23050
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5536630213877867e-05,
+      "loss": 0.1665,
+      "step": 23060
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5534694667569924e-05,
+      "loss": 0.216,
+      "step": 23070
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5532759121261978e-05,
+      "loss": 0.222,
+      "step": 23080
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.553082357495403e-05,
+      "loss": 0.3336,
+      "step": 23090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.552888802864609e-05,
+      "loss": 0.2988,
+      "step": 23100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.552695248233814e-05,
+      "loss": 0.2145,
+      "step": 23110
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5525016936030196e-05,
+      "loss": 0.3021,
+      "step": 23120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.552308138972225e-05,
+      "loss": 0.2743,
+      "step": 23130
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5521145843414303e-05,
+      "loss": 0.263,
+      "step": 23140
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.551921029710636e-05,
+      "loss": 0.1799,
+      "step": 23150
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5517274750798414e-05,
+      "loss": 0.2998,
+      "step": 23160
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.551533920449047e-05,
+      "loss": 0.3104,
+      "step": 23170
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5513403658182525e-05,
+      "loss": 0.3845,
+      "step": 23180
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5511468111874578e-05,
+      "loss": 0.2989,
+      "step": 23190
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5509532565566632e-05,
+      "loss": 0.1964,
+      "step": 23200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5507597019258685e-05,
+      "loss": 0.2604,
+      "step": 23210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5505661472950742e-05,
+      "loss": 0.2242,
+      "step": 23220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5503725926642796e-05,
+      "loss": 0.2149,
+      "step": 23230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.550179038033485e-05,
+      "loss": 0.1885,
+      "step": 23240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5499854834026907e-05,
+      "loss": 0.2021,
+      "step": 23250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.549791928771896e-05,
+      "loss": 0.1642,
+      "step": 23260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5495983741411014e-05,
+      "loss": 0.3228,
+      "step": 23270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5494048195103068e-05,
+      "loss": 0.2392,
+      "step": 23280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5492112648795125e-05,
+      "loss": 0.1153,
+      "step": 23290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5490177102487178e-05,
+      "loss": 0.2811,
+      "step": 23300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5488241556179232e-05,
+      "loss": 0.2784,
+      "step": 23310
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.548630600987129e-05,
+      "loss": 0.3352,
+      "step": 23320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5484370463563343e-05,
+      "loss": 0.3005,
+      "step": 23330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5482434917255396e-05,
+      "loss": 0.2148,
+      "step": 23340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.548049937094745e-05,
+      "loss": 0.2207,
+      "step": 23350
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5478563824639507e-05,
+      "loss": 0.2115,
+      "step": 23360
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.547662827833156e-05,
+      "loss": 0.2036,
+      "step": 23370
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5474692732023614e-05,
+      "loss": 0.365,
+      "step": 23380
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.547275718571567e-05,
+      "loss": 0.3763,
+      "step": 23390
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5470821639407725e-05,
+      "loss": 0.2203,
+      "step": 23400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.546888609309978e-05,
+      "loss": 0.2369,
+      "step": 23410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5466950546791832e-05,
+      "loss": 0.2536,
+      "step": 23420
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5465015000483886e-05,
+      "loss": 0.3092,
+      "step": 23430
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5463079454175943e-05,
+      "loss": 0.2055,
+      "step": 23440
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5461143907867996e-05,
+      "loss": 0.208,
+      "step": 23450
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5459208361560053e-05,
+      "loss": 0.2115,
+      "step": 23460
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5457272815252107e-05,
+      "loss": 0.207,
+      "step": 23470
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.545533726894416e-05,
+      "loss": 0.2334,
+      "step": 23480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5453401722636214e-05,
+      "loss": 0.1667,
+      "step": 23490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5451466176328268e-05,
+      "loss": 0.326,
+      "step": 23500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5449530630020325e-05,
+      "loss": 0.3479,
+      "step": 23510
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.544759508371238e-05,
+      "loss": 0.398,
+      "step": 23520
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5445659537404432e-05,
+      "loss": 0.2493,
+      "step": 23530
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.544372399109649e-05,
+      "loss": 0.3138,
+      "step": 23540
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5441788444788543e-05,
+      "loss": 0.1349,
+      "step": 23550
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.54398528984806e-05,
+      "loss": 0.2156,
+      "step": 23560
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.543791735217265e-05,
+      "loss": 0.4263,
+      "step": 23570
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5435981805864707e-05,
+      "loss": 0.2646,
+      "step": 23580
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.543404625955676e-05,
+      "loss": 0.1706,
+      "step": 23590
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5432110713248814e-05,
+      "loss": 0.2546,
+      "step": 23600
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.543017516694087e-05,
+      "loss": 0.3396,
+      "step": 23610
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5428239620632925e-05,
+      "loss": 0.3997,
+      "step": 23620
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.542630407432498e-05,
+      "loss": 0.2301,
+      "step": 23630
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5424368528017032e-05,
+      "loss": 0.2737,
+      "step": 23640
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.542243298170909e-05,
+      "loss": 0.25,
+      "step": 23650
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5420497435401143e-05,
+      "loss": 0.4109,
+      "step": 23660
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5418561889093197e-05,
+      "loss": 0.2701,
+      "step": 23670
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5416626342785254e-05,
+      "loss": 0.2678,
+      "step": 23680
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5414690796477307e-05,
+      "loss": 0.2808,
+      "step": 23690
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.541275525016936e-05,
+      "loss": 0.3414,
+      "step": 23700
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5410819703861418e-05,
+      "loss": 0.2359,
+      "step": 23710
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5408884157553468e-05,
+      "loss": 0.3002,
+      "step": 23720
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5406948611245525e-05,
+      "loss": 0.2926,
+      "step": 23730
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.540501306493758e-05,
+      "loss": 0.3342,
+      "step": 23740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5403077518629636e-05,
+      "loss": 0.2507,
+      "step": 23750
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.540114197232169e-05,
+      "loss": 0.2502,
+      "step": 23760
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5399206426013743e-05,
+      "loss": 0.2024,
+      "step": 23770
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.53972708797058e-05,
+      "loss": 0.3257,
+      "step": 23780
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5395335333397854e-05,
+      "loss": 0.2146,
+      "step": 23790
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5393399787089908e-05,
+      "loss": 0.2141,
+      "step": 23800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.539146424078196e-05,
+      "loss": 0.3064,
+      "step": 23810
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5389528694474015e-05,
+      "loss": 0.2994,
+      "step": 23820
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5387593148166072e-05,
+      "loss": 0.2311,
+      "step": 23830
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5385657601858125e-05,
+      "loss": 0.28,
+      "step": 23840
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5383722055550182e-05,
+      "loss": 0.24,
+      "step": 23850
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5381786509242236e-05,
+      "loss": 0.2524,
+      "step": 23860
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.537985096293429e-05,
+      "loss": 0.3766,
+      "step": 23870
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5377915416626343e-05,
+      "loss": 0.3418,
+      "step": 23880
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5375979870318397e-05,
+      "loss": 0.2911,
+      "step": 23890
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5374044324010454e-05,
+      "loss": 0.1963,
+      "step": 23900
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5372108777702508e-05,
+      "loss": 0.2365,
+      "step": 23910
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.537017323139456e-05,
+      "loss": 0.3218,
+      "step": 23920
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.536823768508662e-05,
+      "loss": 0.3835,
+      "step": 23930
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5366302138778672e-05,
+      "loss": 0.1557,
+      "step": 23940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5364366592470726e-05,
+      "loss": 0.2646,
+      "step": 23950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.536243104616278e-05,
+      "loss": 0.2171,
+      "step": 23960
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5360495499854836e-05,
+      "loss": 0.2523,
+      "step": 23970
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.535855995354689e-05,
+      "loss": 0.2275,
+      "step": 23980
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5356624407238944e-05,
+      "loss": 0.1274,
+      "step": 23990
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5354688860931e-05,
+      "loss": 0.3385,
+      "step": 24000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5352753314623054e-05,
+      "loss": 0.3724,
+      "step": 24010
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5350817768315108e-05,
+      "loss": 0.1828,
+      "step": 24020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.534888222200716e-05,
+      "loss": 0.2572,
+      "step": 24030
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.534694667569922e-05,
+      "loss": 0.3118,
+      "step": 24040
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5345011129391272e-05,
+      "loss": 0.2422,
+      "step": 24050
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5343075583083326e-05,
+      "loss": 0.204,
+      "step": 24060
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5341140036775383e-05,
+      "loss": 0.219,
+      "step": 24070
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5339204490467436e-05,
+      "loss": 0.2625,
+      "step": 24080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.533726894415949e-05,
+      "loss": 0.1647,
+      "step": 24090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5335333397851544e-05,
+      "loss": 0.2272,
+      "step": 24100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5333397851543597e-05,
+      "loss": 0.1124,
+      "step": 24110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5331462305235654e-05,
+      "loss": 0.2735,
+      "step": 24120
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5329526758927708e-05,
+      "loss": 0.346,
+      "step": 24130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5327591212619765e-05,
+      "loss": 0.4084,
+      "step": 24140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.532565566631182e-05,
+      "loss": 0.2252,
+      "step": 24150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5323720120003872e-05,
+      "loss": 0.2596,
+      "step": 24160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5321784573695926e-05,
+      "loss": 0.213,
+      "step": 24170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.531984902738798e-05,
+      "loss": 0.2955,
+      "step": 24180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5317913481080037e-05,
+      "loss": 0.4828,
+      "step": 24190
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.531597793477209e-05,
+      "loss": 0.2915,
+      "step": 24200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5314042388464144e-05,
+      "loss": 0.2197,
+      "step": 24210
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.53121068421562e-05,
+      "loss": 0.1886,
+      "step": 24220
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5310171295848255e-05,
+      "loss": 0.2988,
+      "step": 24230
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.530823574954031e-05,
+      "loss": 0.3611,
+      "step": 24240
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5306300203232362e-05,
+      "loss": 0.2374,
+      "step": 24250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.530436465692442e-05,
+      "loss": 0.2828,
+      "step": 24260
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5302429110616472e-05,
+      "loss": 0.3351,
+      "step": 24270
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5300493564308526e-05,
+      "loss": 0.2127,
+      "step": 24280
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5298558018000583e-05,
+      "loss": 0.1585,
+      "step": 24290
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5296622471692637e-05,
+      "loss": 0.2887,
+      "step": 24300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.529468692538469e-05,
+      "loss": 0.357,
+      "step": 24310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5292751379076744e-05,
+      "loss": 0.3714,
+      "step": 24320
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.52908158327688e-05,
+      "loss": 0.3306,
+      "step": 24330
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5288880286460855e-05,
+      "loss": 0.2368,
+      "step": 24340
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5286944740152908e-05,
+      "loss": 0.2029,
+      "step": 24350
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5285009193844965e-05,
+      "loss": 0.2222,
+      "step": 24360
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.528307364753702e-05,
+      "loss": 0.272,
+      "step": 24370
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5281138101229073e-05,
+      "loss": 0.2652,
+      "step": 24380
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.527920255492113e-05,
+      "loss": 0.2019,
+      "step": 24390
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.527726700861318e-05,
+      "loss": 0.2738,
+      "step": 24400
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5275331462305237e-05,
+      "loss": 0.2367,
+      "step": 24410
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.527339591599729e-05,
+      "loss": 0.2871,
+      "step": 24420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5271460369689348e-05,
+      "loss": 0.2361,
+      "step": 24430
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.52695248233814e-05,
+      "loss": 0.2566,
+      "step": 24440
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5267589277073455e-05,
+      "loss": 0.1745,
+      "step": 24450
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5265653730765512e-05,
+      "loss": 0.2593,
+      "step": 24460
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5263718184457565e-05,
+      "loss": 0.1936,
+      "step": 24470
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.526178263814962e-05,
+      "loss": 0.1706,
+      "step": 24480
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5259847091841673e-05,
+      "loss": 0.6071,
+      "step": 24490
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5257911545533728e-05,
+      "loss": 0.2808,
+      "step": 24500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5255975999225783e-05,
+      "loss": 0.3278,
+      "step": 24510
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5254040452917837e-05,
+      "loss": 0.2336,
+      "step": 24520
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5252104906609892e-05,
+      "loss": 0.2399,
+      "step": 24530
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5250169360301948e-05,
+      "loss": 0.2623,
+      "step": 24540
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5248233813994e-05,
+      "loss": 0.267,
+      "step": 24550
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5246298267686055e-05,
+      "loss": 0.1675,
+      "step": 24560
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.524436272137811e-05,
+      "loss": 0.2988,
+      "step": 24570
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5242427175070164e-05,
+      "loss": 0.4235,
+      "step": 24580
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.524049162876222e-05,
+      "loss": 0.3269,
+      "step": 24590
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5238556082454275e-05,
+      "loss": 0.309,
+      "step": 24600
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.523662053614633e-05,
+      "loss": 0.3546,
+      "step": 24610
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5234684989838384e-05,
+      "loss": 0.2942,
+      "step": 24620
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5232749443530437e-05,
+      "loss": 0.2581,
+      "step": 24630
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5230813897222493e-05,
+      "loss": 0.2348,
+      "step": 24640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5228878350914546e-05,
+      "loss": 0.2448,
+      "step": 24650
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5226942804606601e-05,
+      "loss": 0.3796,
+      "step": 24660
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5225007258298657e-05,
+      "loss": 0.2222,
+      "step": 24670
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.522307171199071e-05,
+      "loss": 0.1959,
+      "step": 24680
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5221136165682766e-05,
+      "loss": 0.2233,
+      "step": 24690
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.521920061937482e-05,
+      "loss": 0.2237,
+      "step": 24700
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5217265073066873e-05,
+      "loss": 0.228,
+      "step": 24710
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5215329526758928e-05,
+      "loss": 0.3361,
+      "step": 24720
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5213393980450984e-05,
+      "loss": 0.1928,
+      "step": 24730
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5211458434143039e-05,
+      "loss": 0.2597,
+      "step": 24740
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5209522887835093e-05,
+      "loss": 0.3797,
+      "step": 24750
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5207587341527148e-05,
+      "loss": 0.4445,
+      "step": 24760
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5205651795219203e-05,
+      "loss": 0.1985,
+      "step": 24770
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5203716248911255e-05,
+      "loss": 0.227,
+      "step": 24780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.520178070260331e-05,
+      "loss": 0.2051,
+      "step": 24790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5199845156295366e-05,
+      "loss": 0.2848,
+      "step": 24800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.519790960998742e-05,
+      "loss": 0.268,
+      "step": 24810
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5195974063679475e-05,
+      "loss": 0.1352,
+      "step": 24820
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.519403851737153e-05,
+      "loss": 0.3649,
+      "step": 24830
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5192102971063586e-05,
+      "loss": 0.3621,
+      "step": 24840
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5190167424755638e-05,
+      "loss": 0.2592,
+      "step": 24850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5188231878447693e-05,
+      "loss": 0.1312,
+      "step": 24860
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5186296332139746e-05,
+      "loss": 0.1721,
+      "step": 24870
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5184360785831802e-05,
+      "loss": 0.1885,
+      "step": 24880
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5182425239523857e-05,
+      "loss": 0.3529,
+      "step": 24890
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5180489693215912e-05,
+      "loss": 0.2163,
+      "step": 24900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5178554146907966e-05,
+      "loss": 0.2092,
+      "step": 24910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5176618600600021e-05,
+      "loss": 0.4735,
+      "step": 24920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5174683054292075e-05,
+      "loss": 0.3433,
+      "step": 24930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5172747507984129e-05,
+      "loss": 0.2236,
+      "step": 24940
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5170811961676184e-05,
+      "loss": 0.2328,
+      "step": 24950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.516887641536824e-05,
+      "loss": 0.3398,
+      "step": 24960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5166940869060293e-05,
+      "loss": 0.2548,
+      "step": 24970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5165005322752348e-05,
+      "loss": 0.4283,
+      "step": 24980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5163069776444404e-05,
+      "loss": 0.3228,
+      "step": 24990
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5161134230136459e-05,
+      "loss": 0.1594,
+      "step": 25000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5159198683828511e-05,
+      "loss": 0.2234,
+      "step": 25010
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5157263137520566e-05,
+      "loss": 0.177,
+      "step": 25020
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5155327591212622e-05,
+      "loss": 0.3289,
+      "step": 25030
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5153392044904675e-05,
+      "loss": 0.1792,
+      "step": 25040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.515145649859673e-05,
+      "loss": 0.2679,
+      "step": 25050
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5149520952288786e-05,
+      "loss": 0.2904,
+      "step": 25060
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.514758540598084e-05,
+      "loss": 0.1335,
+      "step": 25070
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5145649859672893e-05,
+      "loss": 0.2145,
+      "step": 25080
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5143714313364948e-05,
+      "loss": 0.3014,
+      "step": 25090
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5141778767057002e-05,
+      "loss": 0.3526,
+      "step": 25100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5139843220749057e-05,
+      "loss": 0.2486,
+      "step": 25110
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5137907674441113e-05,
+      "loss": 0.2735,
+      "step": 25120
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5135972128133168e-05,
+      "loss": 0.1811,
+      "step": 25130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5134036581825222e-05,
+      "loss": 0.5123,
+      "step": 25140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5132101035517277e-05,
+      "loss": 0.2389,
+      "step": 25150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5130165489209329e-05,
+      "loss": 0.2648,
+      "step": 25160
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5128229942901384e-05,
+      "loss": 0.3129,
+      "step": 25170
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.512629439659344e-05,
+      "loss": 0.1738,
+      "step": 25180
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5124358850285495e-05,
+      "loss": 0.3011,
+      "step": 25190
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5122423303977549e-05,
+      "loss": 0.2271,
+      "step": 25200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5120487757669604e-05,
+      "loss": 0.184,
+      "step": 25210
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.511855221136166e-05,
+      "loss": 0.1571,
+      "step": 25220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5116616665053711e-05,
+      "loss": 0.2989,
+      "step": 25230
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5114681118745767e-05,
+      "loss": 0.266,
+      "step": 25240
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5112745572437822e-05,
+      "loss": 0.2801,
+      "step": 25250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5110810026129876e-05,
+      "loss": 0.3949,
+      "step": 25260
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5108874479821931e-05,
+      "loss": 0.1405,
+      "step": 25270
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5106938933513986e-05,
+      "loss": 0.1985,
+      "step": 25280
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5105003387206042e-05,
+      "loss": 0.2862,
+      "step": 25290
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5103067840898095e-05,
+      "loss": 0.3799,
+      "step": 25300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5101132294590149e-05,
+      "loss": 0.1691,
+      "step": 25310
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5099196748282204e-05,
+      "loss": 0.4998,
+      "step": 25320
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5097261201974258e-05,
+      "loss": 0.4441,
+      "step": 25330
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5095325655666313e-05,
+      "loss": 0.438,
+      "step": 25340
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5093390109358368e-05,
+      "loss": 0.2481,
+      "step": 25350
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5091454563050422e-05,
+      "loss": 0.293,
+      "step": 25360
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5089519016742477e-05,
+      "loss": 0.1942,
+      "step": 25370
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5087583470434531e-05,
+      "loss": 0.2414,
+      "step": 25380
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5085647924126585e-05,
+      "loss": 0.1907,
+      "step": 25390
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.508371237781864e-05,
+      "loss": 0.4879,
+      "step": 25400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5081776831510695e-05,
+      "loss": 0.3317,
+      "step": 25410
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.507984128520275e-05,
+      "loss": 0.3148,
+      "step": 25420
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5077905738894804e-05,
+      "loss": 0.2303,
+      "step": 25430
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.507597019258686e-05,
+      "loss": 0.2472,
+      "step": 25440
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5074034646278915e-05,
+      "loss": 0.2147,
+      "step": 25450
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5072099099970967e-05,
+      "loss": 0.2488,
+      "step": 25460
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5070163553663022e-05,
+      "loss": 0.2994,
+      "step": 25470
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5068228007355078e-05,
+      "loss": 0.2812,
+      "step": 25480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5066292461047131e-05,
+      "loss": 0.1946,
+      "step": 25490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5064356914739187e-05,
+      "loss": 0.2467,
+      "step": 25500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5062421368431242e-05,
+      "loss": 0.1528,
+      "step": 25510
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5060485822123297e-05,
+      "loss": 0.3006,
+      "step": 25520
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5058550275815349e-05,
+      "loss": 0.4677,
+      "step": 25530
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5056614729507404e-05,
+      "loss": 0.2474,
+      "step": 25540
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5054679183199458e-05,
+      "loss": 0.1248,
+      "step": 25550
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5052743636891513e-05,
+      "loss": 0.3798,
+      "step": 25560
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5050808090583569e-05,
+      "loss": 0.2847,
+      "step": 25570
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5048872544275624e-05,
+      "loss": 0.3379,
+      "step": 25580
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5046936997967678e-05,
+      "loss": 0.2291,
+      "step": 25590
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5045001451659733e-05,
+      "loss": 0.1923,
+      "step": 25600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5043065905351785e-05,
+      "loss": 0.2408,
+      "step": 25610
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.504113035904384e-05,
+      "loss": 0.1416,
+      "step": 25620
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5039194812735896e-05,
+      "loss": 0.2853,
+      "step": 25630
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5037259266427951e-05,
+      "loss": 0.2393,
+      "step": 25640
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5035323720120005e-05,
+      "loss": 0.267,
+      "step": 25650
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.503338817381206e-05,
+      "loss": 0.1804,
+      "step": 25660
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5031452627504115e-05,
+      "loss": 0.2628,
+      "step": 25670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.502951708119617e-05,
+      "loss": 0.3025,
+      "step": 25680
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5027581534888223e-05,
+      "loss": 0.3175,
+      "step": 25690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5025645988580278e-05,
+      "loss": 0.1822,
+      "step": 25700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5023710442272331e-05,
+      "loss": 0.2743,
+      "step": 25710
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5021774895964387e-05,
+      "loss": 0.309,
+      "step": 25720
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5019839349656442e-05,
+      "loss": 0.2074,
+      "step": 25730
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5017903803348497e-05,
+      "loss": 0.1872,
+      "step": 25740
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5015968257040551e-05,
+      "loss": 0.1124,
+      "step": 25750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5014032710732605e-05,
+      "loss": 0.2892,
+      "step": 25760
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.501209716442466e-05,
+      "loss": 0.4778,
+      "step": 25770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5010161618116714e-05,
+      "loss": 0.3698,
+      "step": 25780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5008226071808769e-05,
+      "loss": 0.2932,
+      "step": 25790
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5006290525500824e-05,
+      "loss": 0.1962,
+      "step": 25800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.500435497919288e-05,
+      "loss": 0.2199,
+      "step": 25810
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5002419432884933e-05,
+      "loss": 0.2596,
+      "step": 25820
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5000483886576989e-05,
+      "loss": 0.2447,
+      "step": 25830
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.499854834026904e-05,
+      "loss": 0.2165,
+      "step": 25840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4996612793961096e-05,
+      "loss": 0.2877,
+      "step": 25850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4994677247653151e-05,
+      "loss": 0.2111,
+      "step": 25860
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4992741701345207e-05,
+      "loss": 0.2909,
+      "step": 25870
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.499080615503726e-05,
+      "loss": 0.1299,
+      "step": 25880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4988870608729316e-05,
+      "loss": 0.2323,
+      "step": 25890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4986935062421371e-05,
+      "loss": 0.3201,
+      "step": 25900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4984999516113423e-05,
+      "loss": 0.3555,
+      "step": 25910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4983063969805478e-05,
+      "loss": 0.236,
+      "step": 25920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4981128423497533e-05,
+      "loss": 0.4017,
+      "step": 25930
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4979192877189587e-05,
+      "loss": 0.1974,
+      "step": 25940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4977257330881642e-05,
+      "loss": 0.1356,
+      "step": 25950
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4975321784573698e-05,
+      "loss": 0.3001,
+      "step": 25960
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4973386238265753e-05,
+      "loss": 0.4177,
+      "step": 25970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4971450691957807e-05,
+      "loss": 0.3657,
+      "step": 25980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.496951514564986e-05,
+      "loss": 0.292,
+      "step": 25990
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4967579599341914e-05,
+      "loss": 0.1794,
+      "step": 26000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.496564405303397e-05,
+      "loss": 0.2537,
+      "step": 26010
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4963708506726025e-05,
+      "loss": 0.1979,
+      "step": 26020
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.496177296041808e-05,
+      "loss": 0.3007,
+      "step": 26030
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4959837414110134e-05,
+      "loss": 0.2129,
+      "step": 26040
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4957901867802189e-05,
+      "loss": 0.2475,
+      "step": 26050
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4955966321494243e-05,
+      "loss": 0.3807,
+      "step": 26060
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4954030775186296e-05,
+      "loss": 0.221,
+      "step": 26070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4952095228878352e-05,
+      "loss": 0.2414,
+      "step": 26080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4950159682570407e-05,
+      "loss": 0.1752,
+      "step": 26090
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.494822413626246e-05,
+      "loss": 0.2884,
+      "step": 26100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4946288589954516e-05,
+      "loss": 0.2661,
+      "step": 26110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4944353043646571e-05,
+      "loss": 0.2591,
+      "step": 26120
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4942417497338627e-05,
+      "loss": 0.243,
+      "step": 26130
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4940481951030678e-05,
+      "loss": 0.2162,
+      "step": 26140
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4938546404722734e-05,
+      "loss": 0.3812,
+      "step": 26150
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4936610858414789e-05,
+      "loss": 0.287,
+      "step": 26160
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4934675312106843e-05,
+      "loss": 0.2226,
+      "step": 26170
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4932739765798898e-05,
+      "loss": 0.3723,
+      "step": 26180
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4930804219490953e-05,
+      "loss": 0.239,
+      "step": 26190
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4928868673183009e-05,
+      "loss": 0.2136,
+      "step": 26200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4926933126875062e-05,
+      "loss": 0.2376,
+      "step": 26210
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4924997580567116e-05,
+      "loss": 0.2996,
+      "step": 26220
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.492306203425917e-05,
+      "loss": 0.3334,
+      "step": 26230
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4921126487951225e-05,
+      "loss": 0.2724,
+      "step": 26240
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.491919094164328e-05,
+      "loss": 0.1923,
+      "step": 26250
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4917255395335336e-05,
+      "loss": 0.2667,
+      "step": 26260
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.491531984902739e-05,
+      "loss": 0.3354,
+      "step": 26270
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4913384302719445e-05,
+      "loss": 0.2806,
+      "step": 26280
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4911448756411497e-05,
+      "loss": 0.1809,
+      "step": 26290
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4909513210103552e-05,
+      "loss": 0.1758,
+      "step": 26300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4907577663795607e-05,
+      "loss": 0.3002,
+      "step": 26310
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4905642117487663e-05,
+      "loss": 0.134,
+      "step": 26320
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4903706571179716e-05,
+      "loss": 0.2614,
+      "step": 26330
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4901771024871772e-05,
+      "loss": 0.1511,
+      "step": 26340
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4899835478563827e-05,
+      "loss": 0.3437,
+      "step": 26350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4897899932255882e-05,
+      "loss": 0.2063,
+      "step": 26360
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4895964385947934e-05,
+      "loss": 0.2841,
+      "step": 26370
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.489402883963999e-05,
+      "loss": 0.2034,
+      "step": 26380
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4892093293332043e-05,
+      "loss": 0.1939,
+      "step": 26390
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4890157747024098e-05,
+      "loss": 0.2049,
+      "step": 26400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4888222200716154e-05,
+      "loss": 0.1779,
+      "step": 26410
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4886286654408209e-05,
+      "loss": 0.4042,
+      "step": 26420
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4884351108100263e-05,
+      "loss": 0.2894,
+      "step": 26430
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4882415561792316e-05,
+      "loss": 0.4123,
+      "step": 26440
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4880480015484372e-05,
+      "loss": 0.3456,
+      "step": 26450
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4878544469176425e-05,
+      "loss": 0.2407,
+      "step": 26460
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.487660892286848e-05,
+      "loss": 0.3756,
+      "step": 26470
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4874673376560536e-05,
+      "loss": 0.3314,
+      "step": 26480
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.487273783025259e-05,
+      "loss": 0.2056,
+      "step": 26490
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4870802283944645e-05,
+      "loss": 0.2872,
+      "step": 26500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.48688667376367e-05,
+      "loss": 0.2372,
+      "step": 26510
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4866931191328752e-05,
+      "loss": 0.2434,
+      "step": 26520
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4864995645020808e-05,
+      "loss": 0.2534,
+      "step": 26530
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4863060098712863e-05,
+      "loss": 0.354,
+      "step": 26540
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4861124552404918e-05,
+      "loss": 0.2381,
+      "step": 26550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4859189006096972e-05,
+      "loss": 0.2053,
+      "step": 26560
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4857253459789027e-05,
+      "loss": 0.3853,
+      "step": 26570
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4855317913481082e-05,
+      "loss": 0.193,
+      "step": 26580
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4853382367173134e-05,
+      "loss": 0.1801,
+      "step": 26590
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.485144682086519e-05,
+      "loss": 0.3922,
+      "step": 26600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4849511274557245e-05,
+      "loss": 0.4124,
+      "step": 26610
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4847575728249299e-05,
+      "loss": 0.2806,
+      "step": 26620
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4845640181941354e-05,
+      "loss": 0.2537,
+      "step": 26630
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.484370463563341e-05,
+      "loss": 0.2269,
+      "step": 26640
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4841769089325465e-05,
+      "loss": 0.2972,
+      "step": 26650
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4839833543017518e-05,
+      "loss": 0.2073,
+      "step": 26660
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4837897996709572e-05,
+      "loss": 0.2556,
+      "step": 26670
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4835962450401626e-05,
+      "loss": 0.1992,
+      "step": 26680
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4834026904093681e-05,
+      "loss": 0.2746,
+      "step": 26690
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4832091357785736e-05,
+      "loss": 0.2676,
+      "step": 26700
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4830155811477792e-05,
+      "loss": 0.2429,
+      "step": 26710
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4828220265169845e-05,
+      "loss": 0.3507,
+      "step": 26720
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.48262847188619e-05,
+      "loss": 0.3182,
+      "step": 26730
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4824349172553956e-05,
+      "loss": 0.2764,
+      "step": 26740
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4822413626246008e-05,
+      "loss": 0.1826,
+      "step": 26750
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4820478079938063e-05,
+      "loss": 0.3486,
+      "step": 26760
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4818542533630119e-05,
+      "loss": 0.278,
+      "step": 26770
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4816606987322172e-05,
+      "loss": 0.3849,
+      "step": 26780
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4814671441014227e-05,
+      "loss": 0.1967,
+      "step": 26790
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4812735894706283e-05,
+      "loss": 0.2413,
+      "step": 26800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4810800348398338e-05,
+      "loss": 0.2047,
+      "step": 26810
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.480886480209039e-05,
+      "loss": 0.2476,
+      "step": 26820
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4806929255782445e-05,
+      "loss": 0.3956,
+      "step": 26830
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.48049937094745e-05,
+      "loss": 0.2292,
+      "step": 26840
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4803058163166554e-05,
+      "loss": 0.3093,
+      "step": 26850
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.480112261685861e-05,
+      "loss": 0.3712,
+      "step": 26860
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4799187070550665e-05,
+      "loss": 0.084,
+      "step": 26870
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4797251524242719e-05,
+      "loss": 0.2526,
+      "step": 26880
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4795315977934774e-05,
+      "loss": 0.3308,
+      "step": 26890
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4793380431626828e-05,
+      "loss": 0.182,
+      "step": 26900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4791444885318881e-05,
+      "loss": 0.2719,
+      "step": 26910
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4789509339010937e-05,
+      "loss": 0.1014,
+      "step": 26920
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4787573792702992e-05,
+      "loss": 0.3962,
+      "step": 26930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4785638246395047e-05,
+      "loss": 0.1958,
+      "step": 26940
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4783702700087101e-05,
+      "loss": 0.3108,
+      "step": 26950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4781767153779156e-05,
+      "loss": 0.1816,
+      "step": 26960
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4779831607471208e-05,
+      "loss": 0.2616,
+      "step": 26970
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4777896061163263e-05,
+      "loss": 0.1886,
+      "step": 26980
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4775960514855319e-05,
+      "loss": 0.1593,
+      "step": 26990
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4774024968547374e-05,
+      "loss": 0.3204,
+      "step": 27000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4772089422239428e-05,
+      "loss": 0.1555,
+      "step": 27010
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4770153875931483e-05,
+      "loss": 0.2386,
+      "step": 27020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4768218329623538e-05,
+      "loss": 0.2418,
+      "step": 27030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4766282783315594e-05,
+      "loss": 0.2748,
+      "step": 27040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4764347237007646e-05,
+      "loss": 0.2214,
+      "step": 27050
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4762411690699701e-05,
+      "loss": 0.3179,
+      "step": 27060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4760476144391755e-05,
+      "loss": 0.2726,
+      "step": 27070
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.475854059808381e-05,
+      "loss": 0.2973,
+      "step": 27080
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4756605051775865e-05,
+      "loss": 0.1487,
+      "step": 27090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.475466950546792e-05,
+      "loss": 0.2203,
+      "step": 27100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4752733959159974e-05,
+      "loss": 0.2318,
+      "step": 27110
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4750798412852028e-05,
+      "loss": 0.2669,
+      "step": 27120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4748862866544083e-05,
+      "loss": 0.3329,
+      "step": 27130
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4746927320236137e-05,
+      "loss": 0.1712,
+      "step": 27140
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4744991773928192e-05,
+      "loss": 0.2331,
+      "step": 27150
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4743056227620248e-05,
+      "loss": 0.1854,
+      "step": 27160
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4741120681312301e-05,
+      "loss": 0.2261,
+      "step": 27170
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4739185135004357e-05,
+      "loss": 0.4182,
+      "step": 27180
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4737249588696412e-05,
+      "loss": 0.2584,
+      "step": 27190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4735314042388464e-05,
+      "loss": 0.2698,
+      "step": 27200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4733378496080519e-05,
+      "loss": 0.343,
+      "step": 27210
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4731442949772574e-05,
+      "loss": 0.1668,
+      "step": 27220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.472950740346463e-05,
+      "loss": 0.173,
+      "step": 27230
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4727571857156683e-05,
+      "loss": 0.185,
+      "step": 27240
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4725636310848739e-05,
+      "loss": 0.4081,
+      "step": 27250
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4723700764540794e-05,
+      "loss": 0.2937,
+      "step": 27260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4721765218232846e-05,
+      "loss": 0.3289,
+      "step": 27270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4719829671924901e-05,
+      "loss": 0.1939,
+      "step": 27280
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4717894125616957e-05,
+      "loss": 0.3251,
+      "step": 27290
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.471595857930901e-05,
+      "loss": 0.1675,
+      "step": 27300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4714023033001066e-05,
+      "loss": 0.2802,
+      "step": 27310
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4712087486693121e-05,
+      "loss": 0.2342,
+      "step": 27320
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4710151940385176e-05,
+      "loss": 0.2371,
+      "step": 27330
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.470821639407723e-05,
+      "loss": 0.3819,
+      "step": 27340
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4706280847769284e-05,
+      "loss": 0.2079,
+      "step": 27350
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4704345301461337e-05,
+      "loss": 0.2505,
+      "step": 27360
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4702409755153393e-05,
+      "loss": 0.2532,
+      "step": 27370
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4700474208845448e-05,
+      "loss": 0.1377,
+      "step": 27380
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4698538662537503e-05,
+      "loss": 0.2639,
+      "step": 27390
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4696603116229557e-05,
+      "loss": 0.3481,
+      "step": 27400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4694667569921612e-05,
+      "loss": 0.2866,
+      "step": 27410
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4692732023613668e-05,
+      "loss": 0.2265,
+      "step": 27420
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.469079647730572e-05,
+      "loss": 0.2888,
+      "step": 27430
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4688860930997775e-05,
+      "loss": 0.274,
+      "step": 27440
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.468692538468983e-05,
+      "loss": 0.4298,
+      "step": 27450
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4684989838381884e-05,
+      "loss": 0.2925,
+      "step": 27460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4683054292073939e-05,
+      "loss": 0.1869,
+      "step": 27470
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4681118745765994e-05,
+      "loss": 0.1973,
+      "step": 27480
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.467918319945805e-05,
+      "loss": 0.3509,
+      "step": 27490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4677247653150102e-05,
+      "loss": 0.2807,
+      "step": 27500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4675312106842157e-05,
+      "loss": 0.1575,
+      "step": 27510
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4673376560534212e-05,
+      "loss": 0.2494,
+      "step": 27520
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4671441014226266e-05,
+      "loss": 0.4497,
+      "step": 27530
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4669505467918321e-05,
+      "loss": 0.194,
+      "step": 27540
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4667569921610377e-05,
+      "loss": 0.2501,
+      "step": 27550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.466563437530243e-05,
+      "loss": 0.3324,
+      "step": 27560
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4663698828994486e-05,
+      "loss": 0.3145,
+      "step": 27570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.466176328268654e-05,
+      "loss": 0.1386,
+      "step": 27580
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4659827736378593e-05,
+      "loss": 0.3177,
+      "step": 27590
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4657892190070648e-05,
+      "loss": 0.3405,
+      "step": 27600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4655956643762704e-05,
+      "loss": 0.145,
+      "step": 27610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4654021097454759e-05,
+      "loss": 0.1704,
+      "step": 27620
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4652085551146812e-05,
+      "loss": 0.3639,
+      "step": 27630
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4650150004838868e-05,
+      "loss": 0.2982,
+      "step": 27640
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.464821445853092e-05,
+      "loss": 0.155,
+      "step": 27650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4646278912222975e-05,
+      "loss": 0.1992,
+      "step": 27660
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.464434336591503e-05,
+      "loss": 0.3348,
+      "step": 27670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4642407819607086e-05,
+      "loss": 0.355,
+      "step": 27680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.464047227329914e-05,
+      "loss": 0.1607,
+      "step": 27690
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4638536726991195e-05,
+      "loss": 0.3764,
+      "step": 27700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.463660118068325e-05,
+      "loss": 0.2203,
+      "step": 27710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4634665634375305e-05,
+      "loss": 0.3231,
+      "step": 27720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4632730088067357e-05,
+      "loss": 0.3265,
+      "step": 27730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4630794541759413e-05,
+      "loss": 0.2545,
+      "step": 27740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4628858995451466e-05,
+      "loss": 0.2414,
+      "step": 27750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4626923449143522e-05,
+      "loss": 0.1135,
+      "step": 27760
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4624987902835577e-05,
+      "loss": 0.3786,
+      "step": 27770
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4623052356527632e-05,
+      "loss": 0.1919,
+      "step": 27780
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4621116810219686e-05,
+      "loss": 0.274,
+      "step": 27790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.461918126391174e-05,
+      "loss": 0.2828,
+      "step": 27800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4617245717603795e-05,
+      "loss": 0.1811,
+      "step": 27810
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4615310171295849e-05,
+      "loss": 0.2046,
+      "step": 27820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4613374624987904e-05,
+      "loss": 0.3058,
+      "step": 27830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.461143907867996e-05,
+      "loss": 0.25,
+      "step": 27840
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4609503532372013e-05,
+      "loss": 0.2109,
+      "step": 27850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4607567986064068e-05,
+      "loss": 0.2716,
+      "step": 27860
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4605632439756123e-05,
+      "loss": 0.2419,
+      "step": 27870
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4603696893448175e-05,
+      "loss": 0.204,
+      "step": 27880
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.460176134714023e-05,
+      "loss": 0.2511,
+      "step": 27890
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4599825800832286e-05,
+      "loss": 0.2218,
+      "step": 27900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4597890254524341e-05,
+      "loss": 0.2769,
+      "step": 27910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4595954708216395e-05,
+      "loss": 0.2541,
+      "step": 27920
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.459401916190845e-05,
+      "loss": 0.2544,
+      "step": 27930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4592083615600506e-05,
+      "loss": 0.3541,
+      "step": 27940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.459014806929256e-05,
+      "loss": 0.2134,
+      "step": 27950
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4588212522984613e-05,
+      "loss": 0.2875,
+      "step": 27960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4586276976676668e-05,
+      "loss": 0.3241,
+      "step": 27970
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4584341430368722e-05,
+      "loss": 0.1871,
+      "step": 27980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4582405884060777e-05,
+      "loss": 0.3527,
+      "step": 27990
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4580470337752833e-05,
+      "loss": 0.4144,
+      "step": 28000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4578534791444888e-05,
+      "loss": 0.2121,
+      "step": 28010
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4576599245136942e-05,
+      "loss": 0.3028,
+      "step": 28020
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4574663698828995e-05,
+      "loss": 0.2307,
+      "step": 28030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4572728152521049e-05,
+      "loss": 0.3901,
+      "step": 28040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4570792606213104e-05,
+      "loss": 0.2299,
+      "step": 28050
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.456885705990516e-05,
+      "loss": 0.2552,
+      "step": 28060
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4566921513597215e-05,
+      "loss": 0.1482,
+      "step": 28070
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4564985967289268e-05,
+      "loss": 0.252,
+      "step": 28080
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4563050420981324e-05,
+      "loss": 0.2199,
+      "step": 28090
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4561114874673379e-05,
+      "loss": 0.1231,
+      "step": 28100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4559179328365431e-05,
+      "loss": 0.4297,
+      "step": 28110
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4557243782057486e-05,
+      "loss": 0.2647,
+      "step": 28120
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4555308235749542e-05,
+      "loss": 0.2182,
+      "step": 28130
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4553372689441595e-05,
+      "loss": 0.3366,
+      "step": 28140
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.455143714313365e-05,
+      "loss": 0.2902,
+      "step": 28150
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4549501596825706e-05,
+      "loss": 0.3628,
+      "step": 28160
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4547566050517761e-05,
+      "loss": 0.1948,
+      "step": 28170
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4545630504209813e-05,
+      "loss": 0.1214,
+      "step": 28180
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4543694957901869e-05,
+      "loss": 0.2589,
+      "step": 28190
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4541759411593924e-05,
+      "loss": 0.2155,
+      "step": 28200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4539823865285978e-05,
+      "loss": 0.2672,
+      "step": 28210
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4537888318978033e-05,
+      "loss": 0.2225,
+      "step": 28220
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4535952772670088e-05,
+      "loss": 0.2453,
+      "step": 28230
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4534017226362142e-05,
+      "loss": 0.5203,
+      "step": 28240
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4532081680054197e-05,
+      "loss": 0.2529,
+      "step": 28250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.453014613374625e-05,
+      "loss": 0.1855,
+      "step": 28260
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4528210587438304e-05,
+      "loss": 0.1696,
+      "step": 28270
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.452627504113036e-05,
+      "loss": 0.2074,
+      "step": 28280
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4524339494822415e-05,
+      "loss": 0.1351,
+      "step": 28290
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.452240394851447e-05,
+      "loss": 0.296,
+      "step": 28300
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4520468402206524e-05,
+      "loss": 0.3107,
+      "step": 28310
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.451853285589858e-05,
+      "loss": 0.2044,
+      "step": 28320
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4516597309590631e-05,
+      "loss": 0.327,
+      "step": 28330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4514661763282687e-05,
+      "loss": 0.2884,
+      "step": 28340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4512726216974742e-05,
+      "loss": 0.2972,
+      "step": 28350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4510790670666797e-05,
+      "loss": 0.1876,
+      "step": 28360
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4508855124358851e-05,
+      "loss": 0.2492,
+      "step": 28370
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4506919578050906e-05,
+      "loss": 0.3499,
+      "step": 28380
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4504984031742962e-05,
+      "loss": 0.2805,
+      "step": 28390
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4503048485435017e-05,
+      "loss": 0.2175,
+      "step": 28400
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4501112939127069e-05,
+      "loss": 0.2098,
+      "step": 28410
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4499177392819124e-05,
+      "loss": 0.2833,
+      "step": 28420
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4497241846511178e-05,
+      "loss": 0.3221,
+      "step": 28430
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4495306300203233e-05,
+      "loss": 0.2808,
+      "step": 28440
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4493370753895289e-05,
+      "loss": 0.219,
+      "step": 28450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4491435207587344e-05,
+      "loss": 0.2589,
+      "step": 28460
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4489499661279397e-05,
+      "loss": 0.3433,
+      "step": 28470
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4487564114971453e-05,
+      "loss": 0.2038,
+      "step": 28480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4485628568663506e-05,
+      "loss": 0.1694,
+      "step": 28490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.448369302235556e-05,
+      "loss": 0.2074,
+      "step": 28500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4481757476047615e-05,
+      "loss": 0.1701,
+      "step": 28510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.447982192973967e-05,
+      "loss": 0.3573,
+      "step": 28520
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4477886383431724e-05,
+      "loss": 0.1362,
+      "step": 28530
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.447595083712378e-05,
+      "loss": 0.2973,
+      "step": 28540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4474015290815835e-05,
+      "loss": 0.2874,
+      "step": 28550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4472079744507887e-05,
+      "loss": 0.184,
+      "step": 28560
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4470144198199942e-05,
+      "loss": 0.2943,
+      "step": 28570
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4468208651891998e-05,
+      "loss": 0.1801,
+      "step": 28580
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4466273105584053e-05,
+      "loss": 0.2355,
+      "step": 28590
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4464337559276107e-05,
+      "loss": 0.1263,
+      "step": 28600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4462402012968162e-05,
+      "loss": 0.3104,
+      "step": 28610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4460466466660217e-05,
+      "loss": 0.2239,
+      "step": 28620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4458530920352271e-05,
+      "loss": 0.2605,
+      "step": 28630
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4456595374044325e-05,
+      "loss": 0.3171,
+      "step": 28640
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.445465982773638e-05,
+      "loss": 0.3009,
+      "step": 28650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4452724281428434e-05,
+      "loss": 0.1776,
+      "step": 28660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4450788735120489e-05,
+      "loss": 0.3159,
+      "step": 28670
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4448853188812544e-05,
+      "loss": 0.2257,
+      "step": 28680
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.44469176425046e-05,
+      "loss": 0.2769,
+      "step": 28690
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4444982096196653e-05,
+      "loss": 0.2937,
+      "step": 28700
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4443046549888707e-05,
+      "loss": 0.298,
+      "step": 28710
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.444111100358076e-05,
+      "loss": 0.2963,
+      "step": 28720
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4439175457272816e-05,
+      "loss": 0.3083,
+      "step": 28730
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4437239910964871e-05,
+      "loss": 0.2638,
+      "step": 28740
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4435304364656926e-05,
+      "loss": 0.3045,
+      "step": 28750
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.443336881834898e-05,
+      "loss": 0.2259,
+      "step": 28760
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4431433272041035e-05,
+      "loss": 0.1571,
+      "step": 28770
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.442949772573309e-05,
+      "loss": 0.3628,
+      "step": 28780
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4427562179425143e-05,
+      "loss": 0.1869,
+      "step": 28790
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4425626633117198e-05,
+      "loss": 0.1748,
+      "step": 28800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4423691086809253e-05,
+      "loss": 0.3537,
+      "step": 28810
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4421755540501307e-05,
+      "loss": 0.4392,
+      "step": 28820
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4419819994193362e-05,
+      "loss": 0.3535,
+      "step": 28830
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4417884447885418e-05,
+      "loss": 0.2025,
+      "step": 28840
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4415948901577473e-05,
+      "loss": 0.2439,
+      "step": 28850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4414013355269525e-05,
+      "loss": 0.3039,
+      "step": 28860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.441207780896158e-05,
+      "loss": 0.2688,
+      "step": 28870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4410142262653636e-05,
+      "loss": 0.3481,
+      "step": 28880
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.440820671634569e-05,
+      "loss": 0.2812,
+      "step": 28890
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4406271170037744e-05,
+      "loss": 0.1184,
+      "step": 28900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.44043356237298e-05,
+      "loss": 0.2074,
+      "step": 28910
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4402400077421853e-05,
+      "loss": 0.1262,
+      "step": 28920
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4400464531113909e-05,
+      "loss": 0.2113,
+      "step": 28930
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4398528984805962e-05,
+      "loss": 0.1201,
+      "step": 28940
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4396593438498016e-05,
+      "loss": 0.2824,
+      "step": 28950
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4394657892190071e-05,
+      "loss": 0.321,
+      "step": 28960
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4392722345882127e-05,
+      "loss": 0.4536,
+      "step": 28970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4390786799574182e-05,
+      "loss": 0.2425,
+      "step": 28980
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4388851253266236e-05,
+      "loss": 0.1847,
+      "step": 28990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4386915706958291e-05,
+      "loss": 0.1989,
+      "step": 29000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4384980160650343e-05,
+      "loss": 0.2785,
+      "step": 29010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4383044614342398e-05,
+      "loss": 0.3678,
+      "step": 29020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4381109068034454e-05,
+      "loss": 0.0756,
+      "step": 29030
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4379173521726509e-05,
+      "loss": 0.3214,
+      "step": 29040
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4377237975418563e-05,
+      "loss": 0.2607,
+      "step": 29050
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4375302429110618e-05,
+      "loss": 0.3181,
+      "step": 29060
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4373366882802673e-05,
+      "loss": 0.2503,
+      "step": 29070
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4371431336494729e-05,
+      "loss": 0.2693,
+      "step": 29080
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.436949579018678e-05,
+      "loss": 0.3115,
+      "step": 29090
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4367560243878836e-05,
+      "loss": 0.2133,
+      "step": 29100
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.436562469757089e-05,
+      "loss": 0.1673,
+      "step": 29110
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4363689151262945e-05,
+      "loss": 0.3611,
+      "step": 29120
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4361753604955e-05,
+      "loss": 0.2956,
+      "step": 29130
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4359818058647055e-05,
+      "loss": 0.3251,
+      "step": 29140
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4357882512339109e-05,
+      "loss": 0.3151,
+      "step": 29150
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4355946966031164e-05,
+      "loss": 0.2919,
+      "step": 29160
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4354011419723218e-05,
+      "loss": 0.216,
+      "step": 29170
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4352075873415272e-05,
+      "loss": 0.3654,
+      "step": 29180
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4350140327107327e-05,
+      "loss": 0.2067,
+      "step": 29190
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4348204780799382e-05,
+      "loss": 0.2331,
+      "step": 29200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4346269234491436e-05,
+      "loss": 0.2959,
+      "step": 29210
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4344333688183491e-05,
+      "loss": 0.2683,
+      "step": 29220
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4342398141875547e-05,
+      "loss": 0.0799,
+      "step": 29230
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4340462595567599e-05,
+      "loss": 0.2815,
+      "step": 29240
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4338527049259654e-05,
+      "loss": 0.486,
+      "step": 29250
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.433659150295171e-05,
+      "loss": 0.2631,
+      "step": 29260
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4334655956643765e-05,
+      "loss": 0.2519,
+      "step": 29270
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4332720410335818e-05,
+      "loss": 0.1783,
+      "step": 29280
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4330784864027874e-05,
+      "loss": 0.3135,
+      "step": 29290
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4328849317719929e-05,
+      "loss": 0.2005,
+      "step": 29300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4326913771411983e-05,
+      "loss": 0.2708,
+      "step": 29310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4324978225104036e-05,
+      "loss": 0.1997,
+      "step": 29320
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4323042678796091e-05,
+      "loss": 0.3888,
+      "step": 29330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4321107132488145e-05,
+      "loss": 0.2877,
+      "step": 29340
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.43191715861802e-05,
+      "loss": 0.1878,
+      "step": 29350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4317236039872256e-05,
+      "loss": 0.2176,
+      "step": 29360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4315300493564311e-05,
+      "loss": 0.2757,
+      "step": 29370
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4313364947256365e-05,
+      "loss": 0.1714,
+      "step": 29380
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4311429400948418e-05,
+      "loss": 0.1498,
+      "step": 29390
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4309493854640472e-05,
+      "loss": 0.2338,
+      "step": 29400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4307558308332527e-05,
+      "loss": 0.2898,
+      "step": 29410
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4305622762024583e-05,
+      "loss": 0.2384,
+      "step": 29420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4303687215716638e-05,
+      "loss": 0.3367,
+      "step": 29430
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4301751669408692e-05,
+      "loss": 0.2036,
+      "step": 29440
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4299816123100747e-05,
+      "loss": 0.2871,
+      "step": 29450
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4297880576792802e-05,
+      "loss": 0.2601,
+      "step": 29460
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4295945030484854e-05,
+      "loss": 0.2395,
+      "step": 29470
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.429400948417691e-05,
+      "loss": 0.3363,
+      "step": 29480
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4292073937868965e-05,
+      "loss": 0.1838,
+      "step": 29490
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4290138391561019e-05,
+      "loss": 0.2444,
+      "step": 29500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4288202845253074e-05,
+      "loss": 0.3472,
+      "step": 29510
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.428626729894513e-05,
+      "loss": 0.259,
+      "step": 29520
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4284331752637185e-05,
+      "loss": 0.281,
+      "step": 29530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4282396206329236e-05,
+      "loss": 0.2478,
+      "step": 29540
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4280460660021292e-05,
+      "loss": 0.1996,
+      "step": 29550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4278525113713345e-05,
+      "loss": 0.2458,
+      "step": 29560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.42765895674054e-05,
+      "loss": 0.2276,
+      "step": 29570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4274654021097456e-05,
+      "loss": 0.2117,
+      "step": 29580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4272718474789511e-05,
+      "loss": 0.3127,
+      "step": 29590
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4270782928481565e-05,
+      "loss": 0.2574,
+      "step": 29600
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.426884738217362e-05,
+      "loss": 0.2272,
+      "step": 29610
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4266911835865674e-05,
+      "loss": 0.2762,
+      "step": 29620
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4264976289557728e-05,
+      "loss": 0.2168,
+      "step": 29630
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4263040743249783e-05,
+      "loss": 0.1974,
+      "step": 29640
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4261105196941838e-05,
+      "loss": 0.3304,
+      "step": 29650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4259169650633894e-05,
+      "loss": 0.2852,
+      "step": 29660
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4257234104325947e-05,
+      "loss": 0.4177,
+      "step": 29670
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4255298558018003e-05,
+      "loss": 0.2137,
+      "step": 29680
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4253363011710058e-05,
+      "loss": 0.2305,
+      "step": 29690
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.425142746540211e-05,
+      "loss": 0.3274,
+      "step": 29700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4249491919094165e-05,
+      "loss": 0.2512,
+      "step": 29710
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.424755637278622e-05,
+      "loss": 0.296,
+      "step": 29720
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4245620826478274e-05,
+      "loss": 0.3207,
+      "step": 29730
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.424368528017033e-05,
+      "loss": 0.2745,
+      "step": 29740
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4241749733862385e-05,
+      "loss": 0.2397,
+      "step": 29750
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.423981418755444e-05,
+      "loss": 0.2124,
+      "step": 29760
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4237878641246492e-05,
+      "loss": 0.2063,
+      "step": 29770
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4235943094938547e-05,
+      "loss": 0.2017,
+      "step": 29780
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4234007548630601e-05,
+      "loss": 0.3527,
+      "step": 29790
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4232072002322656e-05,
+      "loss": 0.2827,
+      "step": 29800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4230136456014712e-05,
+      "loss": 0.2066,
+      "step": 29810
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4228200909706767e-05,
+      "loss": 0.2824,
+      "step": 29820
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.422626536339882e-05,
+      "loss": 0.1978,
+      "step": 29830
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4224329817090876e-05,
+      "loss": 0.3146,
+      "step": 29840
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4222394270782928e-05,
+      "loss": 0.2401,
+      "step": 29850
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4220458724474983e-05,
+      "loss": 0.2976,
+      "step": 29860
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4218523178167039e-05,
+      "loss": 0.2992,
+      "step": 29870
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4216587631859094e-05,
+      "loss": 0.2115,
+      "step": 29880
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4214652085551148e-05,
+      "loss": 0.3979,
+      "step": 29890
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4212716539243203e-05,
+      "loss": 0.2222,
+      "step": 29900
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4210780992935258e-05,
+      "loss": 0.2731,
+      "step": 29910
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.420884544662731e-05,
+      "loss": 0.2675,
+      "step": 29920
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4206909900319366e-05,
+      "loss": 0.2629,
+      "step": 29930
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.420497435401142e-05,
+      "loss": 0.2326,
+      "step": 29940
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4203038807703474e-05,
+      "loss": 0.1312,
+      "step": 29950
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.420110326139553e-05,
+      "loss": 0.1405,
+      "step": 29960
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4199167715087585e-05,
+      "loss": 0.213,
+      "step": 29970
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.419723216877964e-05,
+      "loss": 0.2271,
+      "step": 29980
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4195296622471694e-05,
+      "loss": 0.3155,
+      "step": 29990
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4193361076163748e-05,
+      "loss": 0.3274,
+      "step": 30000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4191425529855803e-05,
+      "loss": 0.2308,
+      "step": 30010
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4189489983547857e-05,
+      "loss": 0.2568,
+      "step": 30020
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4187554437239912e-05,
+      "loss": 0.2902,
+      "step": 30030
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4185618890931967e-05,
+      "loss": 0.2355,
+      "step": 30040
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4183683344624023e-05,
+      "loss": 0.2098,
+      "step": 30050
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4181747798316076e-05,
+      "loss": 0.2254,
+      "step": 30060
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.417981225200813e-05,
+      "loss": 0.2631,
+      "step": 30070
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4177876705700184e-05,
+      "loss": 0.3234,
+      "step": 30080
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4175941159392239e-05,
+      "loss": 0.3809,
+      "step": 30090
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4174005613084294e-05,
+      "loss": 0.3574,
+      "step": 30100
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.417207006677635e-05,
+      "loss": 0.2859,
+      "step": 30110
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4170134520468403e-05,
+      "loss": 0.1965,
+      "step": 30120
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4168198974160459e-05,
+      "loss": 0.242,
+      "step": 30130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4166263427852514e-05,
+      "loss": 0.188,
+      "step": 30140
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4164327881544566e-05,
+      "loss": 0.3378,
+      "step": 30150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4162392335236621e-05,
+      "loss": 0.1605,
+      "step": 30160
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4160456788928676e-05,
+      "loss": 0.2144,
+      "step": 30170
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.415852124262073e-05,
+      "loss": 0.3108,
+      "step": 30180
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4156585696312785e-05,
+      "loss": 0.2123,
+      "step": 30190
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.415465015000484e-05,
+      "loss": 0.4109,
+      "step": 30200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4152714603696896e-05,
+      "loss": 0.2525,
+      "step": 30210
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4150779057388948e-05,
+      "loss": 0.2012,
+      "step": 30220
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4148843511081003e-05,
+      "loss": 0.2829,
+      "step": 30230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4146907964773057e-05,
+      "loss": 0.1983,
+      "step": 30240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4144972418465112e-05,
+      "loss": 0.332,
+      "step": 30250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4143036872157168e-05,
+      "loss": 0.2593,
+      "step": 30260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4141101325849223e-05,
+      "loss": 0.2094,
+      "step": 30270
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4139165779541277e-05,
+      "loss": 0.2982,
+      "step": 30280
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4137230233233332e-05,
+      "loss": 0.244,
+      "step": 30290
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4135294686925386e-05,
+      "loss": 0.3663,
+      "step": 30300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.413335914061744e-05,
+      "loss": 0.3346,
+      "step": 30310
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4131423594309495e-05,
+      "loss": 0.2247,
+      "step": 30320
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.412948804800155e-05,
+      "loss": 0.25,
+      "step": 30330
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4127552501693604e-05,
+      "loss": 0.3604,
+      "step": 30340
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4125616955385659e-05,
+      "loss": 0.301,
+      "step": 30350
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4123681409077714e-05,
+      "loss": 0.2704,
+      "step": 30360
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.412174586276977e-05,
+      "loss": 0.306,
+      "step": 30370
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4119810316461821e-05,
+      "loss": 0.3321,
+      "step": 30380
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4117874770153877e-05,
+      "loss": 0.1893,
+      "step": 30390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4115939223845932e-05,
+      "loss": 0.2796,
+      "step": 30400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4114003677537986e-05,
+      "loss": 0.2241,
+      "step": 30410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4112068131230041e-05,
+      "loss": 0.3775,
+      "step": 30420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4110132584922096e-05,
+      "loss": 0.1969,
+      "step": 30430
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4108197038614152e-05,
+      "loss": 0.434,
+      "step": 30440
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4106261492306204e-05,
+      "loss": 0.2823,
+      "step": 30450
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4104325945998259e-05,
+      "loss": 0.2502,
+      "step": 30460
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4102390399690313e-05,
+      "loss": 0.2992,
+      "step": 30470
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4100454853382368e-05,
+      "loss": 0.2202,
+      "step": 30480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4098519307074423e-05,
+      "loss": 0.2645,
+      "step": 30490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4096583760766479e-05,
+      "loss": 0.2621,
+      "step": 30500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4094648214458532e-05,
+      "loss": 0.2782,
+      "step": 30510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4092712668150588e-05,
+      "loss": 0.3818,
+      "step": 30520
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.409077712184264e-05,
+      "loss": 0.1424,
+      "step": 30530
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4088841575534695e-05,
+      "loss": 0.2392,
+      "step": 30540
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.408690602922675e-05,
+      "loss": 0.4666,
+      "step": 30550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4084970482918806e-05,
+      "loss": 0.2837,
+      "step": 30560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.408303493661086e-05,
+      "loss": 0.1709,
+      "step": 30570
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4081099390302915e-05,
+      "loss": 0.1784,
+      "step": 30580
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.407916384399497e-05,
+      "loss": 0.2981,
+      "step": 30590
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4077228297687022e-05,
+      "loss": 0.2701,
+      "step": 30600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4075292751379077e-05,
+      "loss": 0.2692,
+      "step": 30610
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4073357205071132e-05,
+      "loss": 0.3596,
+      "step": 30620
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4071421658763186e-05,
+      "loss": 0.3847,
+      "step": 30630
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4069486112455241e-05,
+      "loss": 0.0671,
+      "step": 30640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4067550566147297e-05,
+      "loss": 0.2907,
+      "step": 30650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4065615019839352e-05,
+      "loss": 0.2712,
+      "step": 30660
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4063679473531406e-05,
+      "loss": 0.3645,
+      "step": 30670
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.406174392722346e-05,
+      "loss": 0.2951,
+      "step": 30680
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4059808380915515e-05,
+      "loss": 0.2643,
+      "step": 30690
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4057872834607568e-05,
+      "loss": 0.3132,
+      "step": 30700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4055937288299624e-05,
+      "loss": 0.5021,
+      "step": 30710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4054001741991679e-05,
+      "loss": 0.2907,
+      "step": 30720
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4052066195683733e-05,
+      "loss": 0.2646,
+      "step": 30730
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4050130649375788e-05,
+      "loss": 0.3159,
+      "step": 30740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4048195103067842e-05,
+      "loss": 0.2856,
+      "step": 30750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4046259556759895e-05,
+      "loss": 0.2481,
+      "step": 30760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.404432401045195e-05,
+      "loss": 0.226,
+      "step": 30770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4042388464144006e-05,
+      "loss": 0.3285,
+      "step": 30780
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4040452917836061e-05,
+      "loss": 0.1426,
+      "step": 30790
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4038517371528115e-05,
+      "loss": 0.1899,
+      "step": 30800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.403658182522017e-05,
+      "loss": 0.1949,
+      "step": 30810
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4034646278912225e-05,
+      "loss": 0.19,
+      "step": 30820
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4032710732604277e-05,
+      "loss": 0.2083,
+      "step": 30830
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4030775186296333e-05,
+      "loss": 0.4168,
+      "step": 30840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4028839639988388e-05,
+      "loss": 0.2787,
+      "step": 30850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4026904093680442e-05,
+      "loss": 0.1361,
+      "step": 30860
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4024968547372497e-05,
+      "loss": 0.2287,
+      "step": 30870
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4023033001064552e-05,
+      "loss": 0.2991,
+      "step": 30880
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4021097454756608e-05,
+      "loss": 0.3338,
+      "step": 30890
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4019161908448661e-05,
+      "loss": 0.3493,
+      "step": 30900
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4017226362140715e-05,
+      "loss": 0.4598,
+      "step": 30910
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4015290815832769e-05,
+      "loss": 0.2868,
+      "step": 30920
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4013355269524824e-05,
+      "loss": 0.1644,
+      "step": 30930
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.401141972321688e-05,
+      "loss": 0.2678,
+      "step": 30940
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4009484176908935e-05,
+      "loss": 0.302,
+      "step": 30950
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4007548630600988e-05,
+      "loss": 0.2285,
+      "step": 30960
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4005613084293044e-05,
+      "loss": 0.2352,
+      "step": 30970
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4003677537985097e-05,
+      "loss": 0.2555,
+      "step": 30980
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.400174199167715e-05,
+      "loss": 0.3025,
+      "step": 30990
+    },
+    {
+      "epoch": 3.0,
+      "eval_FN": 926,
+      "eval_FP": 965,
+      "eval_TN": 14794,
+      "eval_TP": 3980,
+      "eval_accuracy": 0.9084926203726107,
+      "eval_f1": 0.8080397929144248,
+      "eval_loss": 0.38119348883628845,
+      "eval_precision": 0.8048533872598584,
+      "eval_recall": 0.811251528740318,
+      "eval_runtime": 142.04,
+      "eval_samples_per_second": 145.487,
+      "eval_steps_per_second": 9.096,
+      "step": 30999
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 103330,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 1.3049019101684736e+17,
+  "trial_name": null,
+  "trial_params": null
+}