diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5149 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 11.998420151840962,
+  "global_step": 8544,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.945525291828794e-05,
+      "loss": 0.781,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.891050583657588e-05,
+      "loss": 0.7431,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.836575875486382e-05,
+      "loss": 0.7041,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.782101167315176e-05,
+      "loss": 0.6794,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.72762645914397e-05,
+      "loss": 0.6752,
+      "step": 50
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011673151750972763,
+      "loss": 0.6534,
+      "step": 60
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013618677042801555,
+      "loss": 0.6638,
+      "step": 70
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001556420233463035,
+      "loss": 0.6417,
+      "step": 80
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017509727626459142,
+      "loss": 0.6459,
+      "step": 90
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001945525291828794,
+      "loss": 0.6533,
+      "step": 100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002140077821011673,
+      "loss": 0.6434,
+      "step": 110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00023346303501945527,
+      "loss": 0.6427,
+      "step": 120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002529182879377432,
+      "loss": 0.6471,
+      "step": 130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002723735408560311,
+      "loss": 0.6407,
+      "step": 140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002918287937743191,
+      "loss": 0.6415,
+      "step": 150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000311284046692607,
+      "loss": 0.6475,
+      "step": 160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00033073929961089494,
+      "loss": 0.6396,
+      "step": 170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035019455252918285,
+      "loss": 0.628,
+      "step": 180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00036964980544747087,
+      "loss": 0.6363,
+      "step": 190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003891050583657588,
+      "loss": 0.6429,
+      "step": 200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004085603112840467,
+      "loss": 0.6441,
+      "step": 210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004280155642023346,
+      "loss": 0.6359,
+      "step": 220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00044747081712062257,
+      "loss": 0.6331,
+      "step": 230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00046692607003891054,
+      "loss": 0.6234,
+      "step": 240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00048638132295719845,
+      "loss": 0.6366,
+      "step": 250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004999998383195374,
+      "loss": 0.6313,
+      "step": 260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004999969640060184,
+      "loss": 0.6241,
+      "step": 270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004999904968408763,
+      "loss": 0.6166,
+      "step": 280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0004999804369170547,
+      "loss": 0.6133,
+      "step": 290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0004999667843791306,
+      "loss": 0.6304,
+      "step": 300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0004999495394233122,
+      "loss": 0.63,
+      "step": 310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0004999287022974372,
+      "loss": 0.6229,
+      "step": 320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0004999042733009678,
+      "loss": 0.615,
+      "step": 330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0004998762527849873,
+      "loss": 0.6052,
+      "step": 340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000499844641152195,
+      "loss": 0.6154,
+      "step": 350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0004998094388568999,
+      "loss": 0.6037,
+      "step": 360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004997706464050146,
+      "loss": 0.6154,
+      "step": 370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000499728264354048,
+      "loss": 0.6193,
+      "step": 380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004996822933130973,
+      "loss": 0.6221,
+      "step": 390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000499632733942839,
+      "loss": 0.6158,
+      "step": 400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004995795869555194,
+      "loss": 0.6066,
+      "step": 410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0004995228531149451,
+      "loss": 0.6023,
+      "step": 420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0004994625332364707,
+      "loss": 0.5985,
+      "step": 430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0004993986281869884,
+      "loss": 0.616,
+      "step": 440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0004993311388849148,
+      "loss": 0.601,
+      "step": 450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0004992600663001779,
+      "loss": 0.5982,
+      "step": 460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000499185411454203,
+      "loss": 0.6051,
+      "step": 470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0004991071754198985,
+      "loss": 0.6036,
+      "step": 480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004990253593216397,
+      "loss": 0.6029,
+      "step": 490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004989399643352531,
+      "loss": 0.5954,
+      "step": 500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0004988509916880002,
+      "loss": 0.5919,
+      "step": 510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0004987584426585581,
+      "loss": 0.5957,
+      "step": 520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000498662318577003,
+      "loss": 0.5944,
+      "step": 530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0004985626208247897,
+      "loss": 0.5957,
+      "step": 540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0004984593508347327,
+      "loss": 0.5898,
+      "step": 550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0004983525100909849,
+      "loss": 0.5943,
+      "step": 560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0004982421001290169,
+      "loss": 0.5925,
+      "step": 570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0004981281225355941,
+      "loss": 0.5915,
+      "step": 580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000498010578948755,
+      "loss": 0.5933,
+      "step": 590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0004978894710577867,
+      "loss": 0.5968,
+      "step": 600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0004977648006032011,
+      "loss": 0.5787,
+      "step": 610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0004976365693767095,
+      "loss": 0.5945,
+      "step": 620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0004975047792211976,
+      "loss": 0.5665,
+      "step": 630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0004973694320306983,
+      "loss": 0.5902,
+      "step": 640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0004972305297503646,
+      "loss": 0.5785,
+      "step": 650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0004970880743764421,
+      "loss": 0.5824,
+      "step": 660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0004969420679562394,
+      "loss": 0.5908,
+      "step": 670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0004967925125880998,
+      "loss": 0.5892,
+      "step": 680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0004966394104213704,
+      "loss": 0.5801,
+      "step": 690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0004964827636563713,
+      "loss": 0.5667,
+      "step": 700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0004963225745443643,
+      "loss": 0.5809,
+      "step": 710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0004961588453875202,
+      "loss": 0.5396,
+      "step": 720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004959915785388856,
+      "loss": 0.5388,
+      "step": 730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004958207764023499,
+      "loss": 0.5258,
+      "step": 740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004956464414326096,
+      "loss": 0.5234,
+      "step": 750
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004954685761351341,
+      "loss": 0.5338,
+      "step": 760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004952871830661286,
+      "loss": 0.5322,
+      "step": 770
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004951022648324985,
+      "loss": 0.5362,
+      "step": 780
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004949138240918111,
+      "loss": 0.5272,
+      "step": 790
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004947218635522579,
+      "loss": 0.5305,
+      "step": 800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004945263859726151,
+      "loss": 0.5225,
+      "step": 810
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004943273941622049,
+      "loss": 0.537,
+      "step": 820
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0004941248909808538,
+      "loss": 0.5299,
+      "step": 830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0004939188793388531,
+      "loss": 0.5238,
+      "step": 840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0004937093621969156,
+      "loss": 0.5193,
+      "step": 850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0004934963425661339,
+      "loss": 0.5326,
+      "step": 860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000493279823507937,
+      "loss": 0.524,
+      "step": 870
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0004930598081340459,
+      "loss": 0.5292,
+      "step": 880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0004928362996064295,
+      "loss": 0.5296,
+      "step": 890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0004926093011372587,
+      "loss": 0.5306,
+      "step": 900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0004923788159888602,
+      "loss": 0.5306,
+      "step": 910
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0004921448474736701,
+      "loss": 0.5259,
+      "step": 920
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0004919073989541861,
+      "loss": 0.5293,
+      "step": 930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0004916664738429187,
+      "loss": 0.5209,
+      "step": 940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0004914220756023425,
+      "loss": 0.5269,
+      "step": 950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0004911742077448466,
+      "loss": 0.52,
+      "step": 960
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0004909228738326844,
+      "loss": 0.5165,
+      "step": 970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000490668077477921,
+      "loss": 0.5188,
+      "step": 980
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0004904098223423833,
+      "loss": 0.5197,
+      "step": 990
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0004901481121376058,
+      "loss": 0.5286,
+      "step": 1000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0004898829506247778,
+      "loss": 0.532,
+      "step": 1010
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0004896143416146895,
+      "loss": 0.5249,
+      "step": 1020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0004893422889676772,
+      "loss": 0.5194,
+      "step": 1030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0004890667965935674,
+      "loss": 0.5264,
+      "step": 1040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000488787868451621,
+      "loss": 0.517,
+      "step": 1050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0004885055085504767,
+      "loss": 0.5213,
+      "step": 1060
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0004882197209480924,
+      "loss": 0.5186,
+      "step": 1070
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00048793050975168794,
+      "loss": 0.5155,
+      "step": 1080
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0004876378791176854,
+      "loss": 0.5207,
+      "step": 1090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0004873418332516497,
+      "loss": 0.5195,
+      "step": 1100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0004870423764082279,
+      "loss": 0.5245,
+      "step": 1110
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0004867395128910883,
+      "loss": 0.5111,
+      "step": 1120
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00048643324705285857,
+      "loss": 0.5216,
+      "step": 1130
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00048612358329506295,
+      "loss": 0.5218,
+      "step": 1140
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0004858105260680592,
+      "loss": 0.5182,
+      "step": 1150
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00048549407987097463,
+      "loss": 0.5097,
+      "step": 1160
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0004851742492516415,
+      "loss": 0.5179,
+      "step": 1170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00048485103880653135,
+      "loss": 0.5152,
+      "step": 1180
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00048452445318068933,
+      "loss": 0.51,
+      "step": 1190
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00048419449706766716,
+      "loss": 0.5118,
+      "step": 1200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000483861175209456,
+      "loss": 0.5153,
+      "step": 1210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004835244923964178,
+      "loss": 0.5048,
+      "step": 1220
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004831844534672169,
+      "loss": 0.5075,
+      "step": 1230
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00048284106330875035,
+      "loss": 0.5206,
+      "step": 1240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004824943268560776,
+      "loss": 0.4996,
+      "step": 1250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004821442490923498,
+      "loss": 0.5169,
+      "step": 1260
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004817908350487378,
+      "loss": 0.5214,
+      "step": 1270
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004814340898043601,
+      "loss": 0.5163,
+      "step": 1280
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00048107401848621,
+      "loss": 0.5093,
+      "step": 1290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00048071062626908154,
+      "loss": 0.5026,
+      "step": 1300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004803439183754954,
+      "loss": 0.521,
+      "step": 1310
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004799739000756239,
+      "loss": 0.5125,
+      "step": 1320
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004796005766872149,
+      "loss": 0.5066,
+      "step": 1330
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004792239535755157,
+      "loss": 0.5127,
+      "step": 1340
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004788440361531961,
+      "loss": 0.5121,
+      "step": 1350
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00047846082988026975,
+      "loss": 0.5147,
+      "step": 1360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0004780743402640169,
+      "loss": 0.5085,
+      "step": 1370
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00047768457285890433,
+      "loss": 0.5025,
+      "step": 1380
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0004772915332665059,
+      "loss": 0.5115,
+      "step": 1390
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0004768952271354219,
+      "loss": 0.5014,
+      "step": 1400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00047649566016119807,
+      "loss": 0.5115,
+      "step": 1410
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0004760928380862435,
+      "loss": 0.5008,
+      "step": 1420
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0004756867666997482,
+      "loss": 0.4657,
+      "step": 1430
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0004752774518376002,
+      "loss": 0.431,
+      "step": 1440
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00047486489938230103,
+      "loss": 0.4359,
+      "step": 1450
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00047444911526288186,
+      "loss": 0.4312,
+      "step": 1460
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00047403010545481774,
+      "loss": 0.4324,
+      "step": 1470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00047360787597994214,
+      "loss": 0.425,
+      "step": 1480
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00047318243290636017,
+      "loss": 0.4344,
+      "step": 1490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00047275378234836135,
+      "loss": 0.4367,
+      "step": 1500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00047232193046633196,
+      "loss": 0.4275,
+      "step": 1510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00047188688346666633,
+      "loss": 0.4341,
+      "step": 1520
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0004714486476016777,
+      "loss": 0.4233,
+      "step": 1530
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0004710072291695082,
+      "loss": 0.4431,
+      "step": 1540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00047056263451403864,
+      "loss": 0.4318,
+      "step": 1550
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00047011487002479715,
+      "loss": 0.4382,
+      "step": 1560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0004696639421368674,
+      "loss": 0.4306,
+      "step": 1570
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00046920985733079593,
+      "loss": 0.4323,
+      "step": 1580
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00046875262213249945,
+      "loss": 0.4456,
+      "step": 1590
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0004682922431131706,
+      "loss": 0.436,
+      "step": 1600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00046782872688918377,
+      "loss": 0.4473,
+      "step": 1610
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000467362080122,
+      "loss": 0.4321,
+      "step": 1620
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00046689230951807095,
+      "loss": 0.4399,
+      "step": 1630
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0004664194218287431,
+      "loss": 0.4327,
+      "step": 1640
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00046594342385015996,
+      "loss": 0.4359,
+      "step": 1650
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00046546432242316533,
+      "loss": 0.4375,
+      "step": 1660
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000464982124433204,
+      "loss": 0.4333,
+      "step": 1670
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0004644968368102235,
+      "loss": 0.439,
+      "step": 1680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0004640084665285741,
+      "loss": 0.4354,
+      "step": 1690
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00046351702060690907,
+      "loss": 0.4376,
+      "step": 1700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00046302250610808323,
+      "loss": 0.4409,
+      "step": 1710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00046252493013905174,
+      "loss": 0.4371,
+      "step": 1720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00046202429985076787,
+      "loss": 0.4426,
+      "step": 1730
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00046152062243808045,
+      "loss": 0.4343,
+      "step": 1740
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00046101390513963027,
+      "loss": 0.4331,
+      "step": 1750
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00046050415523774603,
+      "loss": 0.438,
+      "step": 1760
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00045999138005833965,
+      "loss": 0.427,
+      "step": 1770
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0004594755869708012,
+      "loss": 0.4404,
+      "step": 1780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0004589567833878929,
+      "loss": 0.4211,
+      "step": 1790
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0004584349767656424,
+      "loss": 0.4402,
+      "step": 1800
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0004579101746032359,
+      "loss": 0.4345,
+      "step": 1810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0004573823844429101,
+      "loss": 0.4302,
+      "step": 1820
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0004568516138698442,
+      "loss": 0.4383,
+      "step": 1830
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00045631787051205044,
+      "loss": 0.4281,
+      "step": 1840
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00045578116204026466,
+      "loss": 0.4321,
+      "step": 1850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00045524149616783615,
+      "loss": 0.4369,
+      "step": 1860
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00045469888065061685,
+      "loss": 0.4364,
+      "step": 1870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00045415332328684944,
+      "loss": 0.4346,
+      "step": 1880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00045360483191705594,
+      "loss": 0.4445,
+      "step": 1890
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00045305341442392446,
+      "loss": 0.4247,
+      "step": 1900
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0004524990787321961,
+      "loss": 0.4231,
+      "step": 1910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00045194183280855126,
+      "loss": 0.4283,
+      "step": 1920
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00045138168466149486,
+      "loss": 0.4373,
+      "step": 1930
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00045081864234124136,
+      "loss": 0.4258,
+      "step": 1940
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0004502527139395991,
+      "loss": 0.4416,
+      "step": 1950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00044968390758985395,
+      "loss": 0.4451,
+      "step": 1960
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0004491122314666525,
+      "loss": 0.4353,
+      "step": 1970
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00044853769378588426,
+      "loss": 0.4222,
+      "step": 1980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00044796030280456437,
+      "loss": 0.4324,
+      "step": 1990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0004473800668207139,
+      "loss": 0.4371,
+      "step": 2000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0004467969941732413,
+      "loss": 0.4438,
+      "step": 2010
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0004462110932418225,
+      "loss": 0.4434,
+      "step": 2020
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0004456223724467801,
+      "loss": 0.4326,
+      "step": 2030
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0004450308402489626,
+      "loss": 0.4306,
+      "step": 2040
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0004444365051496231,
+      "loss": 0.4331,
+      "step": 2050
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00044383937569029665,
+      "loss": 0.4325,
+      "step": 2060
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0004432394604526775,
+      "loss": 0.4299,
+      "step": 2070
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00044263676805849614,
+      "loss": 0.4369,
+      "step": 2080
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00044203130716939524,
+      "loss": 0.4287,
+      "step": 2090
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00044142308648680484,
+      "loss": 0.4289,
+      "step": 2100
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0004408121147518179,
+      "loss": 0.4308,
+      "step": 2110
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0004401984007450642,
+      "loss": 0.4281,
+      "step": 2120
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00043958195328658425,
+      "loss": 0.4309,
+      "step": 2130
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00043896278123570264,
+      "loss": 0.4064,
+      "step": 2140
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0004383408934909008,
+      "loss": 0.3512,
+      "step": 2150
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004377162989896887,
+      "loss": 0.3488,
+      "step": 2160
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00043708900670847686,
+      "loss": 0.3448,
+      "step": 2170
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004364590256624471,
+      "loss": 0.3393,
+      "step": 2180
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00043582636490542315,
+      "loss": 0.3464,
+      "step": 2190
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004351910335297402,
+      "loss": 0.3463,
+      "step": 2200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004345530406661148,
+      "loss": 0.3453,
+      "step": 2210
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004339123954835128,
+      "loss": 0.3493,
+      "step": 2220
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004332691071890186,
+      "loss": 0.3453,
+      "step": 2230
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00043262318502770185,
+      "loss": 0.3447,
+      "step": 2240
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004319746382824854,
+      "loss": 0.3538,
+      "step": 2250
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004313234762740111,
+      "loss": 0.3551,
+      "step": 2260
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004306697083605068,
+      "loss": 0.3552,
+      "step": 2270
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004300133439376509,
+      "loss": 0.3477,
+      "step": 2280
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004293543924384379,
+      "loss": 0.3463,
+      "step": 2290
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004286928633330426,
+      "loss": 0.3519,
+      "step": 2300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00042802876612868426,
+      "loss": 0.3473,
+      "step": 2310
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00042736211036948946,
+      "loss": 0.3541,
+      "step": 2320
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004266929056363557,
+      "loss": 0.3524,
+      "step": 2330
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00042602116154681277,
+      "loss": 0.3564,
+      "step": 2340
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004253468877548855,
+      "loss": 0.3532,
+      "step": 2350
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004246700939509542,
+      "loss": 0.3518,
+      "step": 2360
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004239907898616159,
+      "loss": 0.3529,
+      "step": 2370
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00042330898524954423,
+      "loss": 0.3479,
+      "step": 2380
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004226246899133496,
+      "loss": 0.363,
+      "step": 2390
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004219379136874378,
+      "loss": 0.3484,
+      "step": 2400
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.000421248666441869,
+      "loss": 0.3534,
+      "step": 2410
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004205569580822158,
+      "loss": 0.3573,
+      "step": 2420
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004198627985494211,
+      "loss": 0.3574,
+      "step": 2430
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004191661978196546,
+      "loss": 0.3635,
+      "step": 2440
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004184671659041703,
+      "loss": 0.3582,
+      "step": 2450
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.000417765712849162,
+      "loss": 0.3479,
+      "step": 2460
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00041706184873561904,
+      "loss": 0.3549,
+      "step": 2470
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004163555836791816,
+      "loss": 0.3558,
+      "step": 2480
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004156469278299951,
+      "loss": 0.3561,
+      "step": 2490
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00041493589137256464,
+      "loss": 0.3466,
+      "step": 2500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00041422248452560817,
+      "loss": 0.3485,
+      "step": 2510
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041350671754190996,
+      "loss": 0.3461,
+      "step": 2520
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041278860070817323,
+      "loss": 0.3473,
+      "step": 2530
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004120681443448724,
+      "loss": 0.3609,
+      "step": 2540
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00041134535880610427,
+      "loss": 0.3651,
+      "step": 2550
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00041062025447943974,
+      "loss": 0.3526,
+      "step": 2560
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004098928417857746,
+      "loss": 0.3599,
+      "step": 2570
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004091631311791791,
+      "loss": 0.3468,
+      "step": 2580
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004084311331467484,
+      "loss": 0.355,
+      "step": 2590
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004076968582084517,
+      "loss": 0.3546,
+      "step": 2600
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004069603169169805,
+      "loss": 0.3615,
+      "step": 2610
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004062215198575978,
+      "loss": 0.3507,
+      "step": 2620
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00040548047764798545,
+      "loss": 0.3517,
+      "step": 2630
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004047372009380914,
+      "loss": 0.3607,
+      "step": 2640
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00040399170040997733,
+      "loss": 0.3558,
+      "step": 2650
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00040324398677766437,
+      "loss": 0.3591,
+      "step": 2660
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004024940707869796,
+      "loss": 0.3632,
+      "step": 2670
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00040174196321540143,
+      "loss": 0.3649,
+      "step": 2680
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00040098767487190467,
+      "loss": 0.3527,
+      "step": 2690
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004002312165968055,
+      "loss": 0.3571,
+      "step": 2700
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00039947259926160524,
+      "loss": 0.3615,
+      "step": 2710
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00039871183376883435,
+      "loss": 0.3533,
+      "step": 2720
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0003979489310518958,
+      "loss": 0.3638,
+      "step": 2730
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00039718390207490784,
+      "loss": 0.3555,
+      "step": 2740
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0003964167578325463,
+      "loss": 0.361,
+      "step": 2750
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0003956475093498869,
+      "loss": 0.3465,
+      "step": 2760
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00039487616768224646,
+      "loss": 0.365,
+      "step": 2770
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0003941027439150243,
+      "loss": 0.3541,
+      "step": 2780
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0003933272491635428,
+      "loss": 0.3599,
+      "step": 2790
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0003925496945728875,
+      "loss": 0.3559,
+      "step": 2800
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00039177009131774714,
+      "loss": 0.3569,
+      "step": 2810
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0003909884506022533,
+      "loss": 0.3439,
+      "step": 2820
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00039020478365981863,
+      "loss": 0.3478,
+      "step": 2830
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0003894191017529762,
+      "loss": 0.3588,
+      "step": 2840
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00038863141617321716,
+      "loss": 0.3464,
+      "step": 2850
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003878417382408287,
+      "loss": 0.2648,
+      "step": 2860
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00038705007930473113,
+      "loss": 0.2697,
+      "step": 2870
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00038625645074231497,
+      "loss": 0.2768,
+      "step": 2880
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003854608639592775,
+      "loss": 0.2681,
+      "step": 2890
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00038466333038945856,
+      "loss": 0.2624,
+      "step": 2900
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00038386386149467645,
+      "loss": 0.2696,
+      "step": 2910
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003830624687645633,
+      "loss": 0.2695,
+      "step": 2920
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00038225916371639946,
+      "loss": 0.2703,
+      "step": 2930
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003814539578949486,
+      "loss": 0.2641,
+      "step": 2940
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038064686287229147,
+      "loss": 0.2749,
+      "step": 2950
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00037983789024765945,
+      "loss": 0.2751,
+      "step": 2960
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00037902705164726837,
+      "loss": 0.2687,
+      "step": 2970
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003782143587241507,
+      "loss": 0.271,
+      "step": 2980
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00037739982315798865,
+      "loss": 0.2732,
+      "step": 2990
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003765834566549462,
+      "loss": 0.2747,
+      "step": 3000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037576527094750057,
+      "loss": 0.274,
+      "step": 3010
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037494527779427403,
+      "loss": 0.2738,
+      "step": 3020
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003741234889798645,
+      "loss": 0.2793,
+      "step": 3030
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037329991631467665,
+      "loss": 0.2807,
+      "step": 3040
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003724745716347516,
+      "loss": 0.2752,
+      "step": 3050
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003716474668015975,
+      "loss": 0.2722,
+      "step": 3060
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003708186137020184,
+      "loss": 0.2818,
+      "step": 3070
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036998802424794363,
+      "loss": 0.279,
+      "step": 3080
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036915571037625703,
+      "loss": 0.2844,
+      "step": 3090
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036832168404862467,
+      "loss": 0.2844,
+      "step": 3100
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036748595725132353,
+      "loss": 0.2792,
+      "step": 3110
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003666485419950691,
+      "loss": 0.2775,
+      "step": 3120
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036580945031484247,
+      "loss": 0.275,
+      "step": 3130
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003649686942697178,
+      "loss": 0.2822,
+      "step": 3140
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003641262859426885,
+      "loss": 0.2743,
+      "step": 3150
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00036328223744049433,
+      "loss": 0.2825,
+      "step": 3160
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003624365608934467,
+      "loss": 0.2835,
+      "step": 3170
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00036158926845525444,
+      "loss": 0.2716,
+      "step": 3180
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00036074037230284965,
+      "loss": 0.2765,
+      "step": 3190
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035988988463621206,
+      "loss": 0.2806,
+      "step": 3200
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035903781767819424,
+      "loss": 0.2841,
+      "step": 3210
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00035818418367434537,
+      "loss": 0.2844,
+      "step": 3220
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00035732899489273565,
+      "loss": 0.2758,
+      "step": 3230
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00035647226362378007,
+      "loss": 0.28,
+      "step": 3240
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00035561400218006133,
+      "loss": 0.276,
+      "step": 3250
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003547542228961535,
+      "loss": 0.2836,
+      "step": 3260
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00035389293812844403,
+      "loss": 0.2802,
+      "step": 3270
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003530301602549568,
+      "loss": 0.2758,
+      "step": 3280
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.000352165901675174,
+      "loss": 0.2826,
+      "step": 3290
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003513001748098578,
+      "loss": 0.2766,
+      "step": 3300
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00035043299210087196,
+      "loss": 0.2911,
+      "step": 3310
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00034956436601100304,
+      "loss": 0.2834,
+      "step": 3320
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003486943090237813,
+      "loss": 0.2825,
+      "step": 3330
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00034782283364330116,
+      "loss": 0.281,
+      "step": 3340
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00034694995239404166,
+      "loss": 0.2726,
+      "step": 3350
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003460756778206865,
+      "loss": 0.2792,
+      "step": 3360
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00034520002248794335,
+      "loss": 0.2776,
+      "step": 3370
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.000344322998980364,
+      "loss": 0.2776,
+      "step": 3380
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003434446199021627,
+      "loss": 0.2809,
+      "step": 3390
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00034256489787703553,
+      "loss": 0.2823,
+      "step": 3400
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.000341683845547979,
+      "loss": 0.2797,
+      "step": 3410
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003408014755771078,
+      "loss": 0.2839,
+      "step": 3420
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00033991780064547377,
+      "loss": 0.2781,
+      "step": 3430
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003390328334528825,
+      "loss": 0.2813,
+      "step": 3440
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003381465867177118,
+      "loss": 0.279,
+      "step": 3450
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003372590731767284,
+      "loss": 0.2807,
+      "step": 3460
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00033637030558490515,
+      "loss": 0.2809,
+      "step": 3470
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003354802967152375,
+      "loss": 0.2759,
+      "step": 3480
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003345890593585599,
+      "loss": 0.2796,
+      "step": 3490
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00033369660632336237,
+      "loss": 0.2868,
+      "step": 3500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00033280295043560586,
+      "loss": 0.2814,
+      "step": 3510
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003319081045385385,
+      "loss": 0.2865,
+      "step": 3520
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003310120814925106,
+      "loss": 0.2819,
+      "step": 3530
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003301148941747899,
+      "loss": 0.2881,
+      "step": 3540
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00032921655547937665,
+      "loss": 0.279,
+      "step": 3550
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003283170783168181,
+      "loss": 0.2797,
+      "step": 3560
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00032741647561402334,
+      "loss": 0.2122,
+      "step": 3570
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.000326514760314077,
+      "loss": 0.2101,
+      "step": 3580
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0003256119453760535,
+      "loss": 0.2012,
+      "step": 3590
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.000324708043774831,
+      "loss": 0.2007,
+      "step": 3600
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00032380306850090444,
+      "loss": 0.205,
+      "step": 3610
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00032289703256019954,
+      "loss": 0.2007,
+      "step": 3620
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00032198994897388494,
+      "loss": 0.2006,
+      "step": 3630
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.000321081830778186,
+      "loss": 0.2027,
+      "step": 3640
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00032017269102419714,
+      "loss": 0.1977,
+      "step": 3650
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0003192625427776937,
+      "loss": 0.2067,
+      "step": 3660
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00031835139911894506,
+      "loss": 0.2031,
+      "step": 3670
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0003174392731425262,
+      "loss": 0.2023,
+      "step": 3680
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00031652617795712936,
+      "loss": 0.2035,
+      "step": 3690
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0003156121266853761,
+      "loss": 0.2056,
+      "step": 3700
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0003146971324636282,
+      "loss": 0.1988,
+      "step": 3710
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0003137812084417994,
+      "loss": 0.2117,
+      "step": 3720
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0003128643677831659,
+      "loss": 0.2051,
+      "step": 3730
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00031194662366417755,
+      "loss": 0.2006,
+      "step": 3740
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0003110279892742682,
+      "loss": 0.2065,
+      "step": 3750
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0003101084778156666,
+      "loss": 0.217,
+      "step": 3760
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00030918810250320607,
+      "loss": 0.1976,
+      "step": 3770
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0003082668765641349,
+      "loss": 0.2073,
+      "step": 3780
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0003073448132379264,
+      "loss": 0.2103,
+      "step": 3790
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0003064219257760884,
+      "loss": 0.2049,
+      "step": 3800
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0003054982274419728,
+      "loss": 0.2078,
+      "step": 3810
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00030457373151058506,
+      "loss": 0.2003,
+      "step": 3820
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00030364845126839326,
+      "loss": 0.2097,
+      "step": 3830
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0003027224000131375,
+      "loss": 0.2081,
+      "step": 3840
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00030179559105363837,
+      "loss": 0.2054,
+      "step": 3850
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.000300868037709606,
+      "loss": 0.2045,
+      "step": 3860
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00029993975331144826,
+      "loss": 0.2115,
+      "step": 3870
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002990107512000796,
+      "loss": 0.2076,
+      "step": 3880
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002980810447267294,
+      "loss": 0.2091,
+      "step": 3890
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00029715064725274927,
+      "loss": 0.214,
+      "step": 3900
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.000296219572149422,
+      "loss": 0.2082,
+      "step": 3910
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002952878327977688,
+      "loss": 0.2112,
+      "step": 3920
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002943554425883574,
+      "loss": 0.2111,
+      "step": 3930
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00029342241492110924,
+      "loss": 0.2136,
+      "step": 3940
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00029248876320510673,
+      "loss": 0.206,
+      "step": 3950
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002915545008584016,
+      "loss": 0.2024,
+      "step": 3960
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002906196413078205,
+      "loss": 0.2045,
+      "step": 3970
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00028968419798877334,
+      "loss": 0.2149,
+      "step": 3980
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00028874818434505964,
+      "loss": 0.2129,
+      "step": 3990
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00028781161382867513,
+      "loss": 0.2125,
+      "step": 4000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00028687449989961905,
+      "loss": 0.2102,
+      "step": 4010
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00028593685602570007,
+      "loss": 0.2076,
+      "step": 4020
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00028499869568234307,
+      "loss": 0.2046,
+      "step": 4030
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00028406003235239554,
+      "loss": 0.2127,
+      "step": 4040
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002831208795259334,
+      "loss": 0.2085,
+      "step": 4050
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00028218125070006786,
+      "loss": 0.2108,
+      "step": 4060
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002812411593787504,
+      "loss": 0.208,
+      "step": 4070
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002803006190725798,
+      "loss": 0.2128,
+      "step": 4080
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00027935964329860733,
+      "loss": 0.2158,
+      "step": 4090
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0002784182455801423,
+      "loss": 0.2123,
+      "step": 4100
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00027747643944655865,
+      "loss": 0.219,
+      "step": 4110
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002765342384330992,
+      "loss": 0.2175,
+      "step": 4120
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00027559165608068246,
+      "loss": 0.216,
+      "step": 4130
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00027464870593570676,
+      "loss": 0.2078,
+      "step": 4140
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002737054015498567,
+      "loss": 0.2096,
+      "step": 4150
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002727617564799077,
+      "loss": 0.209,
+      "step": 4160
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00027181778428753123,
+      "loss": 0.2123,
+      "step": 4170
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0002708734985391002,
+      "loss": 0.2185,
+      "step": 4180
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002699289128054938,
+      "loss": 0.2113,
+      "step": 4190
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002689840406619023,
+      "loss": 0.2098,
+      "step": 4200
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002680388956876324,
+      "loss": 0.2094,
+      "step": 4210
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00026709349146591157,
+      "loss": 0.2172,
+      "step": 4220
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00026614784158369325,
+      "loss": 0.2173,
+      "step": 4230
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002652019596314613,
+      "loss": 0.2144,
+      "step": 4240
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0002642558592030348,
+      "loss": 0.2267,
+      "step": 4250
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00026330955389537275,
+      "loss": 0.2127,
+      "step": 4260
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002623630573083786,
+      "loss": 0.2233,
+      "step": 4270
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00026141638304470467,
+      "loss": 0.1622,
+      "step": 4280
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0002604695447095568,
+      "loss": 0.1438,
+      "step": 4290
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.000259522555910499,
+      "loss": 0.1432,
+      "step": 4300
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0002585754302572574,
+      "loss": 0.1467,
+      "step": 4310
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.000257628181361525,
+      "loss": 0.1456,
+      "step": 4320
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00025668082283676613,
+      "loss": 0.1444,
+      "step": 4330
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0002557333682980206,
+      "loss": 0.1409,
+      "step": 4340
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0002547858313617081,
+      "loss": 0.147,
+      "step": 4350
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00025383822564543234,
+      "loss": 0.147,
+      "step": 4360
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0002528905647677859,
+      "loss": 0.144,
+      "step": 4370
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00025194286234815356,
+      "loss": 0.152,
+      "step": 4380
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0002509951320065176,
+      "loss": 0.1465,
+      "step": 4390
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00025004738736326135,
+      "loss": 0.1459,
+      "step": 4400
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0002490996420389738,
+      "loss": 0.1481,
+      "step": 4410
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0002481519096542537,
+      "loss": 0.1475,
+      "step": 4420
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0002472042038295137,
+      "loss": 0.1481,
+      "step": 4430
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0002462565381847849,
+      "loss": 0.148,
+      "step": 4440
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00024530892633952084,
+      "loss": 0.1493,
+      "step": 4450
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.000244361381912402,
+      "loss": 0.1506,
+      "step": 4460
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00024341391852113984,
+      "loss": 0.1547,
+      "step": 4470
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0002424665497822812,
+      "loss": 0.1551,
+      "step": 4480
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0002415192893110128,
+      "loss": 0.152,
+      "step": 4490
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0002405721507209651,
+      "loss": 0.1429,
+      "step": 4500
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00023962514762401721,
+      "loss": 0.1484,
+      "step": 4510
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0002386782936301008,
+      "loss": 0.1504,
+      "step": 4520
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0002377316023470048,
+      "loss": 0.1505,
+      "step": 4530
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00023678508738017958,
+      "loss": 0.1553,
+      "step": 4540
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00023583876233254183,
+      "loss": 0.1504,
+      "step": 4550
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00023489264080427854,
+      "loss": 0.1478,
+      "step": 4560
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00023394673639265192,
+      "loss": 0.1548,
+      "step": 4570
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00023300106269180394,
+      "loss": 0.1418,
+      "step": 4580
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00023205563329256062,
+      "loss": 0.1541,
+      "step": 4590
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0002311104617822374,
+      "loss": 0.1409,
+      "step": 4600
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00023016556174444298,
+      "loss": 0.1489,
+      "step": 4610
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0002292209467588851,
+      "loss": 0.1523,
+      "step": 4620
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00022827663040117424,
+      "loss": 0.148,
+      "step": 4630
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00022733262624262976,
+      "loss": 0.1549,
+      "step": 4640
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00022638894785008358,
+      "loss": 0.1471,
+      "step": 4650
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00022544560878568645,
+      "loss": 0.1567,
+      "step": 4660
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00022450262260671202,
+      "loss": 0.1537,
+      "step": 4670
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00022356000286536266,
+      "loss": 0.1494,
+      "step": 4680
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00022261776310857434,
+      "loss": 0.1515,
+      "step": 4690
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00022167591687782186,
+      "loss": 0.1477,
+      "step": 4700
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00022073447770892487,
+      "loss": 0.152,
+      "step": 4710
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00021979345913185255,
+      "loss": 0.154,
+      "step": 4720
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00021885287467052973,
+      "loss": 0.1535,
+      "step": 4730
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00021791273784264203,
+      "loss": 0.1562,
+      "step": 4740
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00021697306215944222,
+      "loss": 0.1509,
+      "step": 4750
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0002160338611255554,
+      "loss": 0.1563,
+      "step": 4760
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00021509514823878537,
+      "loss": 0.1498,
+      "step": 4770
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0002141569369899204,
+      "loss": 0.1488,
+      "step": 4780
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00021321924086253953,
+      "loss": 0.1551,
+      "step": 4790
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0002122820733328186,
+      "loss": 0.1516,
+      "step": 4800
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00021134544786933663,
+      "loss": 0.1554,
+      "step": 4810
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00021040937793288242,
+      "loss": 0.1584,
+      "step": 4820
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00020947387697626085,
+      "loss": 0.1508,
+      "step": 4830
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00020853895844409983,
+      "loss": 0.1552,
+      "step": 4840
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0002076046357726566,
+      "loss": 0.1546,
+      "step": 4850
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00020667092238962537,
+      "loss": 0.1496,
+      "step": 4860
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00020573783171394343,
+      "loss": 0.1543,
+      "step": 4870
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00020480537715559927,
+      "loss": 0.1625,
+      "step": 4880
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00020387357211543882,
+      "loss": 0.1531,
+      "step": 4890
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00020294242998497404,
+      "loss": 0.1489,
+      "step": 4900
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00020201196414618936,
+      "loss": 0.1515,
+      "step": 4910
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0002010821879713498,
+      "loss": 0.1587,
+      "step": 4920
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00020015311482280925,
+      "loss": 0.1543,
+      "step": 4930
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00019922475805281758,
+      "loss": 0.1518,
+      "step": 4940
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00019829713100332942,
+      "loss": 0.1562,
+      "step": 4950
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00019737024700581195,
+      "loss": 0.1499,
+      "step": 4960
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0001964441193810538,
+      "loss": 0.1566,
+      "step": 4970
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00019551876143897298,
+      "loss": 0.1571,
+      "step": 4980
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00019459418647842632,
+      "loss": 0.1216,
+      "step": 4990
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0001936704077870176,
+      "loss": 0.1024,
+      "step": 5000
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0001927474386409072,
+      "loss": 0.1058,
+      "step": 5010
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00019182529230462106,
+      "loss": 0.1005,
+      "step": 5020
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0001909039820308598,
+      "loss": 0.1019,
+      "step": 5030
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00018998352106030882,
+      "loss": 0.099,
+      "step": 5040
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.0001890639226214474,
+      "loss": 0.0999,
+      "step": 5050
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00018814519993035913,
+      "loss": 0.1047,
+      "step": 5060
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0001872273661905415,
+      "loss": 0.1031,
+      "step": 5070
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00018631043459271657,
+      "loss": 0.0976,
+      "step": 5080
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00018539441831464092,
+      "loss": 0.0992,
+      "step": 5090
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00018447933052091698,
+      "loss": 0.1016,
+      "step": 5100
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0001835651843628028,
+      "loss": 0.1006,
+      "step": 5110
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00018265199297802425,
+      "loss": 0.1055,
+      "step": 5120
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00018173976949058512,
+      "loss": 0.1026,
+      "step": 5130
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00018082852701057915,
+      "loss": 0.1008,
+      "step": 5140
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0001799182786340016,
+      "loss": 0.1,
+      "step": 5150
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00017900903744256063,
+      "loss": 0.105,
+      "step": 5160
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0001781008165034898,
+      "loss": 0.0969,
+      "step": 5170
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00017719362886935985,
+      "loss": 0.104,
+      "step": 5180
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00017628748757789143,
+      "loss": 0.1071,
+      "step": 5190
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00017538240565176743,
+      "loss": 0.1048,
+      "step": 5200
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00017447839609844622,
+      "loss": 0.1007,
+      "step": 5210
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00017357547190997416,
+      "loss": 0.1044,
+      "step": 5220
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.00017267364606279954,
+      "loss": 0.0989,
+      "step": 5230
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00017177293151758553,
+      "loss": 0.1078,
+      "step": 5240
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0001708733412190241,
+      "loss": 0.107,
+      "step": 5250
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0001699748880956503,
+      "loss": 0.1089,
+      "step": 5260
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0001690775850596558,
+      "loss": 0.1037,
+      "step": 5270
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.000168181445006704,
+      "loss": 0.1077,
+      "step": 5280
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00016728648081574423,
+      "loss": 0.1037,
+      "step": 5290
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00016639270534882691,
+      "loss": 0.1036,
+      "step": 5300
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00016550013145091852,
+      "loss": 0.1055,
+      "step": 5310
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00016460877194971718,
+      "loss": 0.0999,
+      "step": 5320
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00016371863965546825,
+      "loss": 0.1087,
+      "step": 5330
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00016282974736077995,
+      "loss": 0.1061,
+      "step": 5340
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00016194210784044,
+      "loss": 0.101,
+      "step": 5350
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00016105573385123166,
+      "loss": 0.1049,
+      "step": 5360
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00016017063813175068,
+      "loss": 0.1053,
+      "step": 5370
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0001592868334022216,
+      "loss": 0.1021,
+      "step": 5380
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0001584043323643161,
+      "loss": 0.1102,
+      "step": 5390
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.00015752314770096893,
+      "loss": 0.1045,
+      "step": 5400
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.00015664329207619714,
+      "loss": 0.1086,
+      "step": 5410
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00015576477813491706,
+      "loss": 0.1078,
+      "step": 5420
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.000154887618502763,
+      "loss": 0.1072,
+      "step": 5430
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.00015401182578590568,
+      "loss": 0.1063,
+      "step": 5440
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.0001531374125708709,
+      "loss": 0.1022,
+      "step": 5450
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.00015226439142435933,
+      "loss": 0.102,
+      "step": 5460
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.00015139277489306487,
+      "loss": 0.1087,
+      "step": 5470
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00015052257550349528,
+      "loss": 0.1047,
+      "step": 5480
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00014965380576179143,
+      "loss": 0.1047,
+      "step": 5490
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.00014878647815354823,
+      "loss": 0.1106,
+      "step": 5500
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.00014792060514363454,
+      "loss": 0.1093,
+      "step": 5510
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00014705619917601453,
+      "loss": 0.1049,
+      "step": 5520
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.00014619327267356857,
+      "loss": 0.1129,
+      "step": 5530
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00014533183803791487,
+      "loss": 0.0999,
+      "step": 5540
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00014447190764923108,
+      "loss": 0.1044,
+      "step": 5550
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.0001436134938660762,
+      "loss": 0.1055,
+      "step": 5560
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.0001427566090252139,
+      "loss": 0.1039,
+      "step": 5570
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00014190126544143378,
+      "loss": 0.1042,
+      "step": 5580
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00014104747540737551,
+      "loss": 0.1036,
+      "step": 5590
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0001401952511933518,
+      "loss": 0.108,
+      "step": 5600
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00013934460504717205,
+      "loss": 0.1061,
+      "step": 5610
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.0001384955491939661,
+      "loss": 0.1107,
+      "step": 5620
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.000137648095836009,
+      "loss": 0.1081,
+      "step": 5630
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.0001368022571525454,
+      "loss": 0.1095,
+      "step": 5640
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.0001359580452996145,
+      "loss": 0.0989,
+      "step": 5650
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.0001351154724098752,
+      "loss": 0.107,
+      "step": 5660
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.0001342745505924321,
+      "loss": 0.1087,
+      "step": 5670
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.00013343529193266137,
+      "loss": 0.1031,
+      "step": 5680
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0001325977084920366,
+      "loss": 0.1081,
+      "step": 5690
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.00013176181230795627,
+      "loss": 0.093,
+      "step": 5700
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.00013092761539356995,
+      "loss": 0.0773,
+      "step": 5710
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.0001300951297376064,
+      "loss": 0.0709,
+      "step": 5720
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.00012926436730420032,
+      "loss": 0.0717,
+      "step": 5730
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.00012843534003272177,
+      "loss": 0.0691,
+      "step": 5740
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.00012760805983760303,
+      "loss": 0.0675,
+      "step": 5750
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.00012678253860816857,
+      "loss": 0.0734,
+      "step": 5760
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.00012595878820846366,
+      "loss": 0.0706,
+      "step": 5770
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00012513682047708376,
+      "loss": 0.0704,
+      "step": 5780
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.00012431664722700492,
+      "loss": 0.0702,
+      "step": 5790
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.00012349828024541315,
+      "loss": 0.0694,
+      "step": 5800
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.00012268173129353625,
+      "loss": 0.0726,
+      "step": 5810
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.00012186701210647327,
+      "loss": 0.0732,
+      "step": 5820
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.00012105413439302718,
+      "loss": 0.0655,
+      "step": 5830
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 0.00012024310983553586,
+      "loss": 0.0664,
+      "step": 5840
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.00011943395008970456,
+      "loss": 0.0681,
+      "step": 5850
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.00011862666678443795,
+      "loss": 0.0746,
+      "step": 5860
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.00011782127152167355,
+      "loss": 0.0712,
+      "step": 5870
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.00011701777587621468,
+      "loss": 0.0744,
+      "step": 5880
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.000116216191395564,
+      "loss": 0.0714,
+      "step": 5890
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.00011541652959975781,
+      "loss": 0.0701,
+      "step": 5900
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.0001146188019812005,
+      "loss": 0.0722,
+      "step": 5910
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.00011382302000449921,
+      "loss": 0.0692,
+      "step": 5920
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.00011302919510629889,
+      "loss": 0.0678,
+      "step": 5930
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.00011223733869511852,
+      "loss": 0.0693,
+      "step": 5940
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.00011144746215118664,
+      "loss": 0.0704,
+      "step": 5950
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 0.00011065957682627811,
+      "loss": 0.0704,
+      "step": 5960
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.00010987369404355041,
+      "loss": 0.0709,
+      "step": 5970
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.00010908982509738199,
+      "loss": 0.0684,
+      "step": 5980
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.00010830798125320879,
+      "loss": 0.0651,
+      "step": 5990
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.0001075281737473627,
+      "loss": 0.0711,
+      "step": 6000
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 0.00010675041378691088,
+      "loss": 0.0727,
+      "step": 6010
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.00010597471254949337,
+      "loss": 0.0716,
+      "step": 6020
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.00010520108118316338,
+      "loss": 0.0712,
+      "step": 6030
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.0001044295308062268,
+      "loss": 0.0715,
+      "step": 6040
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.00010366007250708248,
+      "loss": 0.0698,
+      "step": 6050
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 0.00010289271734406253,
+      "loss": 0.0734,
+      "step": 6060
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.00010212747634527391,
+      "loss": 0.0686,
+      "step": 6070
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.00010136436050843967,
+      "loss": 0.069,
+      "step": 6080
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.00010060338080074094,
+      "loss": 0.0713,
+      "step": 6090
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 9.984454815865909e-05,
+      "loss": 0.0735,
+      "step": 6100
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 9.908787348781903e-05,
+      "loss": 0.0748,
+      "step": 6110
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 9.83333676628321e-05,
+      "loss": 0.0697,
+      "step": 6120
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 9.758104152713981e-05,
+      "loss": 0.0729,
+      "step": 6130
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 9.683090589285822e-05,
+      "loss": 0.0735,
+      "step": 6140
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.608297154062237e-05,
+      "loss": 0.0728,
+      "step": 6150
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 9.533724921943148e-05,
+      "loss": 0.0703,
+      "step": 6160
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 9.459374964649417e-05,
+      "loss": 0.068,
+      "step": 6170
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 9.385248350707487e-05,
+      "loss": 0.0695,
+      "step": 6180
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 9.311346145433991e-05,
+      "loss": 0.0641,
+      "step": 6190
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 9.237669410920474e-05,
+      "loss": 0.0706,
+      "step": 6200
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 9.164219206018076e-05,
+      "loss": 0.0728,
+      "step": 6210
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 9.090996586322379e-05,
+      "loss": 0.0756,
+      "step": 6220
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 9.018002604158196e-05,
+      "loss": 0.0689,
+      "step": 6230
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 8.945238308564438e-05,
+      "loss": 0.0677,
+      "step": 6240
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 8.872704745279104e-05,
+      "loss": 0.0722,
+      "step": 6250
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 8.800402956724148e-05,
+      "loss": 0.0703,
+      "step": 6260
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 8.728333981990596e-05,
+      "loss": 0.0757,
+      "step": 6270
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 8.656498856823558e-05,
+      "loss": 0.072,
+      "step": 6280
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 8.584898613607359e-05,
+      "loss": 0.0762,
+      "step": 6290
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 8.513534281350679e-05,
+      "loss": 0.0681,
+      "step": 6300
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 8.44240688567181e-05,
+      "loss": 0.0689,
+      "step": 6310
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 8.371517448783894e-05,
+      "loss": 0.0678,
+      "step": 6320
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 8.300866989480196e-05,
+      "loss": 0.0722,
+      "step": 6330
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 8.230456523119525e-05,
+      "loss": 0.0755,
+      "step": 6340
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 8.160287061611605e-05,
+      "loss": 0.0738,
+      "step": 6350
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 8.09035961340254e-05,
+      "loss": 0.0766,
+      "step": 6360
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 8.020675183460308e-05,
+      "loss": 0.0742,
+      "step": 6370
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.951234773260343e-05,
+      "loss": 0.073,
+      "step": 6380
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.88203938077113e-05,
+      "loss": 0.0748,
+      "step": 6390
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 7.813090000439863e-05,
+      "loss": 0.0714,
+      "step": 6400
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.744387623178134e-05,
+      "loss": 0.0735,
+      "step": 6410
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 7.675933236347729e-05,
+      "loss": 0.0494,
+      "step": 6420
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 7.607727823746427e-05,
+      "loss": 0.0488,
+      "step": 6430
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 7.539772365593822e-05,
+      "loss": 0.0487,
+      "step": 6440
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 7.472067838517296e-05,
+      "loss": 0.0501,
+      "step": 6450
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 7.404615215537947e-05,
+      "loss": 0.0503,
+      "step": 6460
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 7.337415466056613e-05,
+      "loss": 0.0524,
+      "step": 6470
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 7.270469555839921e-05,
+      "loss": 0.0494,
+      "step": 6480
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 7.203778447006473e-05,
+      "loss": 0.0462,
+      "step": 6490
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 7.137343098012918e-05,
+      "loss": 0.0444,
+      "step": 6500
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 7.071164463640264e-05,
+      "loss": 0.0536,
+      "step": 6510
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 7.00524349498012e-05,
+      "loss": 0.0522,
+      "step": 6520
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 6.939581139421033e-05,
+      "loss": 0.0502,
+      "step": 6530
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 6.874178340634854e-05,
+      "loss": 0.0519,
+      "step": 6540
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 6.809036038563207e-05,
+      "loss": 0.0484,
+      "step": 6550
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 6.744155169403976e-05,
+      "loss": 0.05,
+      "step": 6560
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 6.679536665597813e-05,
+      "loss": 0.0521,
+      "step": 6570
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 6.615181455814783e-05,
+      "loss": 0.0476,
+      "step": 6580
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 6.551090464941006e-05,
+      "loss": 0.0503,
+      "step": 6590
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 6.487264614065353e-05,
+      "loss": 0.0491,
+      "step": 6600
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 6.423704820466197e-05,
+      "loss": 0.0522,
+      "step": 6610
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 6.360411997598268e-05,
+      "loss": 0.0505,
+      "step": 6620
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 6.297387055079495e-05,
+      "loss": 0.0503,
+      "step": 6630
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 6.234630898677959e-05,
+      "loss": 0.05,
+      "step": 6640
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 6.172144430298832e-05,
+      "loss": 0.0492,
+      "step": 6650
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 6.109928547971455e-05,
+      "loss": 0.0461,
+      "step": 6660
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 6.047984145836438e-05,
+      "loss": 0.0453,
+      "step": 6670
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.986312114132758e-05,
+      "loss": 0.0515,
+      "step": 6680
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.924913339185023e-05,
+      "loss": 0.0484,
+      "step": 6690
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.863788703390699e-05,
+      "loss": 0.052,
+      "step": 6700
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.802939085207451e-05,
+      "loss": 0.0485,
+      "step": 6710
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.742365359140472e-05,
+      "loss": 0.0469,
+      "step": 6720
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.6820683957300086e-05,
+      "loss": 0.0491,
+      "step": 6730
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.622049061538731e-05,
+      "loss": 0.0544,
+      "step": 6740
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.56230821913937e-05,
+      "loss": 0.0492,
+      "step": 6750
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.5028467271022896e-05,
+      "loss": 0.0485,
+      "step": 6760
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.443665439983128e-05,
+      "loss": 0.045,
+      "step": 6770
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.384765208310552e-05,
+      "loss": 0.049,
+      "step": 6780
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.326146878574015e-05,
+      "loss": 0.0513,
+      "step": 6790
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.267811293211597e-05,
+      "loss": 0.0488,
+      "step": 6800
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.209759290597879e-05,
+      "loss": 0.0517,
+      "step": 6810
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.151991705031922e-05,
+      "loss": 0.0514,
+      "step": 6820
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.094509366725269e-05,
+      "loss": 0.051,
+      "step": 6830
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.0373131017900034e-05,
+      "loss": 0.0499,
+      "step": 6840
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.980403732226868e-05,
+      "loss": 0.0534,
+      "step": 6850
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.923782075913488e-05,
+      "loss": 0.0492,
+      "step": 6860
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.8674489465925895e-05,
+      "loss": 0.0491,
+      "step": 6870
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.81140515386029e-05,
+      "loss": 0.0521,
+      "step": 6880
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.7556515031545064e-05,
+      "loss": 0.0497,
+      "step": 6890
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.700188795743346e-05,
+      "loss": 0.0487,
+      "step": 6900
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.6450178287136076e-05,
+      "loss": 0.051,
+      "step": 6910
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.5901393949593e-05,
+      "loss": 0.0507,
+      "step": 6920
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.53555428317029e-05,
+      "loss": 0.0473,
+      "step": 6930
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.4812632778209324e-05,
+      "loss": 0.0459,
+      "step": 6940
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.4272671591588154e-05,
+      "loss": 0.0491,
+      "step": 6950
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.3735667031935214e-05,
+      "loss": 0.0466,
+      "step": 6960
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.3201626816855075e-05,
+      "loss": 0.0498,
+      "step": 6970
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.2670558621350054e-05,
+      "loss": 0.0508,
+      "step": 6980
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.214247007770955e-05,
+      "loss": 0.0511,
+      "step": 6990
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.1617368775401124e-05,
+      "loss": 0.047,
+      "step": 7000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.1095262260960445e-05,
+      "loss": 0.0513,
+      "step": 7010
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.0576158037883696e-05,
+      "loss": 0.0473,
+      "step": 7020
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.0060063566519255e-05,
+      "loss": 0.0496,
+      "step": 7030
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.9546986263960695e-05,
+      "loss": 0.0493,
+      "step": 7040
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.9036933503939944e-05,
+      "loss": 0.0497,
+      "step": 7050
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.852991261672159e-05,
+      "loss": 0.0504,
+      "step": 7060
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.8025930888997473e-05,
+      "loss": 0.0523,
+      "step": 7070
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.752499556378169e-05,
+      "loss": 0.0501,
+      "step": 7080
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.7027113840306935e-05,
+      "loss": 0.0504,
+      "step": 7090
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.653229287392071e-05,
+      "loss": 0.0491,
+      "step": 7100
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.604053977598271e-05,
+      "loss": 0.0528,
+      "step": 7110
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.555186161376231e-05,
+      "loss": 0.05,
+      "step": 7120
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.5066265410337336e-05,
+      "loss": 0.039,
+      "step": 7130
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.458375814449296e-05,
+      "loss": 0.0416,
+      "step": 7140
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.4104346750621574e-05,
+      "loss": 0.0387,
+      "step": 7150
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.362803811862264e-05,
+      "loss": 0.0388,
+      "step": 7160
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.3154839093804484e-05,
+      "loss": 0.0377,
+      "step": 7170
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.2684756476785124e-05,
+      "loss": 0.0381,
+      "step": 7180
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.2217797023394915e-05,
+      "loss": 0.0368,
+      "step": 7190
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.1753967444579565e-05,
+      "loss": 0.0424,
+      "step": 7200
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.129327440630339e-05,
+      "loss": 0.0362,
+      "step": 7210
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.083572452945379e-05,
+      "loss": 0.039,
+      "step": 7220
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.038132438974575e-05,
+      "loss": 0.0384,
+      "step": 7230
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.993008051762791e-05,
+      "loss": 0.0344,
+      "step": 7240
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.9481999398187936e-05,
+      "loss": 0.0384,
+      "step": 7250
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.9037087471060014e-05,
+      "loss": 0.0368,
+      "step": 7260
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.8595351130331987e-05,
+      "loss": 0.0408,
+      "step": 7270
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.8156796724453453e-05,
+      "loss": 0.0372,
+      "step": 7280
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.7721430556144537e-05,
+      "loss": 0.0393,
+      "step": 7290
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.7289258882305407e-05,
+      "loss": 0.0367,
+      "step": 7300
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.686028791392639e-05,
+      "loss": 0.0382,
+      "step": 7310
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.6434523815998397e-05,
+      "loss": 0.0347,
+      "step": 7320
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.601197270742467e-05,
+      "loss": 0.0359,
+      "step": 7330
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.5592640660932727e-05,
+      "loss": 0.0364,
+      "step": 7340
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.5176533702987116e-05,
+      "loss": 0.0362,
+      "step": 7350
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4763657813702574e-05,
+      "loss": 0.0395,
+      "step": 7360
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.4354018926758426e-05,
+      "loss": 0.0367,
+      "step": 7370
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.3947622929313183e-05,
+      "loss": 0.0372,
+      "step": 7380
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.3544475661919863e-05,
+      "loss": 0.0365,
+      "step": 7390
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.3144582918441958e-05,
+      "loss": 0.0365,
+      "step": 7400
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.2747950445970527e-05,
+      "loss": 0.0394,
+      "step": 7410
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.2354583944741296e-05,
+      "loss": 0.0399,
+      "step": 7420
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.1964489068052734e-05,
+      "loss": 0.0355,
+      "step": 7430
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.1577671422185e-05,
+      "loss": 0.0409,
+      "step": 7440
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.1194136566319272e-05,
+      "loss": 0.0385,
+      "step": 7450
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.0813890012457788e-05,
+      "loss": 0.0398,
+      "step": 7460
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.0436937225344593e-05,
+      "loss": 0.0406,
+      "step": 7470
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.0063283622387407e-05,
+      "loss": 0.038,
+      "step": 7480
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.9692934573579123e-05,
+      "loss": 0.0396,
+      "step": 7490
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.932589540142113e-05,
+      "loss": 0.0377,
+      "step": 7500
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.896217138084666e-05,
+      "loss": 0.0389,
+      "step": 7510
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.8601767739144875e-05,
+      "loss": 0.0389,
+      "step": 7520
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.8244689655885965e-05,
+      "loss": 0.0365,
+      "step": 7530
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.7890942262846487e-05,
+      "loss": 0.0385,
+      "step": 7540
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.7540530643935874e-05,
+      "loss": 0.037,
+      "step": 7550
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.7193459835122965e-05,
+      "loss": 0.0376,
+      "step": 7560
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.6849734824364054e-05,
+      "loss": 0.0432,
+      "step": 7570
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.6509360551531015e-05,
+      "loss": 0.0385,
+      "step": 7580
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.617234190834027e-05,
+      "loss": 0.039,
+      "step": 7590
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.5838683738282538e-05,
+      "loss": 0.0376,
+      "step": 7600
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.5508390836553227e-05,
+      "loss": 0.037,
+      "step": 7610
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.5181467949983597e-05,
+      "loss": 0.0364,
+      "step": 7620
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4857919776972346e-05,
+      "loss": 0.0359,
+      "step": 7630
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4537750967418272e-05,
+      "loss": 0.039,
+      "step": 7640
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4220966122653334e-05,
+      "loss": 0.0354,
+      "step": 7650
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.390756979537669e-05,
+      "loss": 0.0391,
+      "step": 7660
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.3597566489588903e-05,
+      "loss": 0.0405,
+      "step": 7670
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.3290960660527735e-05,
+      "loss": 0.0367,
+      "step": 7680
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.2987756714603616e-05,
+      "loss": 0.04,
+      "step": 7690
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.2687959009336697e-05,
+      "loss": 0.0379,
+      "step": 7700
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.2391571853293843e-05,
+      "loss": 0.0335,
+      "step": 7710
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.2098599506027263e-05,
+      "loss": 0.0341,
+      "step": 7720
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.1809046178012677e-05,
+      "loss": 0.0377,
+      "step": 7730
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.1522916030589165e-05,
+      "loss": 0.036,
+      "step": 7740
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.1268329117016807e-05,
+      "loss": 0.0399,
+      "step": 7750
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.0988714300960739e-05,
+      "loss": 0.0373,
+      "step": 7760
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.071253445496187e-05,
+      "loss": 0.0388,
+      "step": 7770
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.0439793548161186e-05,
+      "loss": 0.0377,
+      "step": 7780
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.0170495500276739e-05,
+      "loss": 0.0369,
+      "step": 7790
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 9.904644181547211e-06,
+      "loss": 0.037,
+      "step": 7800
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 9.642243412676415e-06,
+      "loss": 0.0415,
+      "step": 7810
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 9.38329696477827e-06,
+      "loss": 0.0383,
+      "step": 7820
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 9.127808559322548e-06,
+      "loss": 0.0376,
+      "step": 7830
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 8.875781868081662e-06,
+      "loss": 0.034,
+      "step": 7840
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 8.62722051307757e-06,
+      "loss": 0.0326,
+      "step": 7850
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 8.382128066529937e-06,
+      "loss": 0.0327,
+      "step": 7860
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 8.140508050804741e-06,
+      "loss": 0.0374,
+      "step": 7870
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 7.90236393836355e-06,
+      "loss": 0.0327,
+      "step": 7880
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 7.667699151713864e-06,
+      "loss": 0.0307,
+      "step": 7890
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 7.436517063359571e-06,
+      "loss": 0.0312,
+      "step": 7900
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 7.208820995752874e-06,
+      "loss": 0.0329,
+      "step": 7910
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 6.984614221246188e-06,
+      "loss": 0.0298,
+      "step": 7920
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 6.763899962045433e-06,
+      "loss": 0.0333,
+      "step": 7930
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 6.546681390163317e-06,
+      "loss": 0.0324,
+      "step": 7940
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 6.3329616273742895e-06,
+      "loss": 0.0335,
+      "step": 7950
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 6.12274374516919e-06,
+      "loss": 0.0332,
+      "step": 7960
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 5.916030764711283e-06,
+      "loss": 0.0323,
+      "step": 7970
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 5.712825656793042e-06,
+      "loss": 0.0307,
+      "step": 7980
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 5.513131341793015e-06,
+      "loss": 0.0361,
+      "step": 7990
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 5.316950689634281e-06,
+      "loss": 0.0329,
+      "step": 8000
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 5.124286519742949e-06,
+      "loss": 0.0315,
+      "step": 8010
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.935141601007831e-06,
+      "loss": 0.0327,
+      "step": 8020
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.749518651740364e-06,
+      "loss": 0.0282,
+      "step": 8030
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.567420339635808e-06,
+      "loss": 0.0345,
+      "step": 8040
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.3888492817348576e-06,
+      "loss": 0.0344,
+      "step": 8050
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.213808044385925e-06,
+      "loss": 0.0342,
+      "step": 8060
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.042299143208306e-06,
+      "loss": 0.034,
+      "step": 8070
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.874325043056071e-06,
+      "loss": 0.0318,
+      "step": 8080
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.7098881579826492e-06,
+      "loss": 0.0359,
+      "step": 8090
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.5489908512060244e-06,
+      "loss": 0.0336,
+      "step": 8100
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.391635435074869e-06,
+      "loss": 0.0342,
+      "step": 8110
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.237824171035325e-06,
+      "loss": 0.0342,
+      "step": 8120
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.0875592695984443e-06,
+      "loss": 0.0334,
+      "step": 8130
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.9408428903083816e-06,
+      "loss": 0.0324,
+      "step": 8140
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.797677141711502e-06,
+      "loss": 0.0342,
+      "step": 8150
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.6580640813259616e-06,
+      "loss": 0.0363,
+      "step": 8160
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.5220057156121747e-06,
+      "loss": 0.0325,
+      "step": 8170
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.4025940675865833e-06,
+      "loss": 0.0365,
+      "step": 8180
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.273294966558287e-06,
+      "loss": 0.0343,
+      "step": 8190
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.1475560899421876e-06,
+      "loss": 0.0316,
+      "step": 8200
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.025379244804837e-06,
+      "loss": 0.032,
+      "step": 8210
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.9067661870205977e-06,
+      "loss": 0.0357,
+      "step": 8220
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.791718621246663e-06,
+      "loss": 0.0349,
+      "step": 8230
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.6802382008984374e-06,
+      "loss": 0.0338,
+      "step": 8240
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.5723265281258336e-06,
+      "loss": 0.0361,
+      "step": 8250
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.46798515379018e-06,
+      "loss": 0.0331,
+      "step": 8260
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.367215577441988e-06,
+      "loss": 0.033,
+      "step": 8270
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.2700192472993865e-06,
+      "loss": 0.0351,
+      "step": 8280
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.1763975602273325e-06,
+      "loss": 0.0294,
+      "step": 8290
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.0863518617174884e-06,
+      "loss": 0.0332,
+      "step": 8300
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 9.998834458688488e-07,
+      "loss": 0.0375,
+      "step": 8310
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 9.169935553693388e-07,
+      "loss": 0.033,
+      "step": 8320
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 8.376833814776886e-07,
+      "loss": 0.0317,
+      "step": 8330
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 7.61954064006476e-07,
+      "loss": 0.0307,
+      "step": 8340
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 6.898066913057777e-07,
+      "loss": 0.0354,
+      "step": 8350
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 6.212423002473489e-07,
+      "loss": 0.0341,
+      "step": 8360
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 5.562618762099125e-07,
+      "loss": 0.0298,
+      "step": 8370
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.948663530648934e-07,
+      "loss": 0.0303,
+      "step": 8380
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.3705661316303936e-07,
+      "loss": 0.0331,
+      "step": 8390
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.828334873217099e-07,
+      "loss": 0.0322,
+      "step": 8400
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.321977548129407e-07,
+      "loss": 0.0367,
+      "step": 8410
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.8515014335225853e-07,
+      "loss": 0.032,
+      "step": 8420
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.41691329088245e-07,
+      "loss": 0.0308,
+      "step": 8430
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.0182193659273873e-07,
+      "loss": 0.0369,
+      "step": 8440
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.6554253885198156e-07,
+      "loss": 0.0322,
+      "step": 8450
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.3285365725823616e-07,
+      "loss": 0.0329,
+      "step": 8460
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.037557616024587e-07,
+      "loss": 0.034,
+      "step": 8470
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 7.824927006735982e-08,
+      "loss": 0.0336,
+      "step": 8480
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 5.6334549221603856e-08,
+      "loss": 0.035,
+      "step": 8490
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.801191401436865e-08,
+      "loss": 0.034,
+      "step": 8500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.328162777084919e-08,
+      "loss": 0.032,
+      "step": 8510
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.2143902188593847e-08,
+      "loss": 0.0344,
+      "step": 8520
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.598897334284757e-09,
+      "loss": 0.0309,
+      "step": 8530
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 6.467216416006227e-10,
+      "loss": 0.0306,
+      "step": 8540
+    },
+    {
+      "epoch": 12.0,
+      "step": 8544,
+      "total_flos": 5.2716569292622004e+19,
+      "train_loss": 0.23854404472840135,
+      "train_runtime": 341050.2928,
+      "train_samples_per_second": 3.207,
+      "train_steps_per_second": 0.025
+    }
+  ],
+  "max_steps": 8544,
+  "num_train_epochs": 12,
+  "total_flos": 5.2716569292622004e+19,
+  "trial_name": null,
+  "trial_params": null
+}