diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,19821 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.97789367628756,
+  "eval_steps": 500,
+  "global_step": 33000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998518342914716e-05,
+      "loss": 1.1436,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9970366858294324e-05,
+      "loss": 1.0946,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9955550287441474e-05,
+      "loss": 1.0475,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994073371658863e-05,
+      "loss": 1.2085,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9925917145735795e-05,
+      "loss": 1.0201,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.991110057488295e-05,
+      "loss": 1.1401,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989628400403011e-05,
+      "loss": 1.2025,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.988146743317727e-05,
+      "loss": 1.0663,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9866650862324424e-05,
+      "loss": 1.0889,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.985183429147158e-05,
+      "loss": 1.1078,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9837017720618746e-05,
+      "loss": 1.0007,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.98222011497659e-05,
+      "loss": 1.1888,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.980738457891306e-05,
+      "loss": 1.1755,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.979256800806022e-05,
+      "loss": 1.1285,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9777751437207374e-05,
+      "loss": 1.0683,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.976293486635453e-05,
+      "loss": 1.0682,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.974811829550169e-05,
+      "loss": 1.0443,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973330172464885e-05,
+      "loss": 1.0094,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.971848515379601e-05,
+      "loss": 1.0373,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.970366858294316e-05,
+      "loss": 1.0986,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9688852012090325e-05,
+      "loss": 1.0402,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.967403544123748e-05,
+      "loss": 1.1408,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965921887038464e-05,
+      "loss": 1.1502,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96444022995318e-05,
+      "loss": 1.1159,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9629585728678954e-05,
+      "loss": 1.0239,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.961476915782611e-05,
+      "loss": 1.1654,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9599952586973275e-05,
+      "loss": 1.1863,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958513601612043e-05,
+      "loss": 1.0276,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.957031944526759e-05,
+      "loss": 1.1198,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9555502874414746e-05,
+      "loss": 1.0463,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9540686303561904e-05,
+      "loss": 0.966,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.952586973270906e-05,
+      "loss": 0.9627,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9511053161856225e-05,
+      "loss": 1.1312,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.949623659100338e-05,
+      "loss": 0.9223,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948142002015054e-05,
+      "loss": 1.1389,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.94666034492977e-05,
+      "loss": 0.8973,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9451786878444854e-05,
+      "loss": 1.0691,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.943697030759201e-05,
+      "loss": 1.099,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9422153736739175e-05,
+      "loss": 1.0559,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.940733716588633e-05,
+      "loss": 0.9345,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.939252059503349e-05,
+      "loss": 1.1638,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.937770402418064e-05,
+      "loss": 1.0137,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9362887453327804e-05,
+      "loss": 1.1298,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.934807088247496e-05,
+      "loss": 1.297,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.933325431162212e-05,
+      "loss": 1.1458,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931843774076928e-05,
+      "loss": 1.2104,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.930362116991643e-05,
+      "loss": 1.0494,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.928880459906359e-05,
+      "loss": 1.0527,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9273988028210754e-05,
+      "loss": 0.9439,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.925917145735791e-05,
+      "loss": 1.1118,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.924435488650507e-05,
+      "loss": 1.1182,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922953831565223e-05,
+      "loss": 1.0818,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921472174479938e-05,
+      "loss": 1.0487,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919990517394654e-05,
+      "loss": 1.1353,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9185088603093704e-05,
+      "loss": 0.9229,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917027203224086e-05,
+      "loss": 0.9945,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915545546138802e-05,
+      "loss": 1.0377,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9140638890535176e-05,
+      "loss": 0.9988,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.912582231968233e-05,
+      "loss": 1.0964,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911100574882949e-05,
+      "loss": 1.0607,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9096189177976655e-05,
+      "loss": 1.2273,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.908137260712381e-05,
+      "loss": 1.2177,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906655603627097e-05,
+      "loss": 0.9447,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9051739465418126e-05,
+      "loss": 1.0235,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9036922894565283e-05,
+      "loss": 0.8942,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.902210632371244e-05,
+      "loss": 1.1545,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9007289752859605e-05,
+      "loss": 0.9827,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.899247318200676e-05,
+      "loss": 1.004,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897765661115392e-05,
+      "loss": 1.1165,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.896284004030107e-05,
+      "loss": 0.9904,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8948023469448234e-05,
+      "loss": 1.0665,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.893320689859539e-05,
+      "loss": 0.9563,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891839032774255e-05,
+      "loss": 1.2268,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.890357375688971e-05,
+      "loss": 1.0352,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.888875718603686e-05,
+      "loss": 1.0372,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.887394061518402e-05,
+      "loss": 0.9848,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8859124044331184e-05,
+      "loss": 1.0746,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.884430747347834e-05,
+      "loss": 1.1633,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88294909026255e-05,
+      "loss": 1.1765,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8814674331772656e-05,
+      "loss": 0.9531,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879985776091981e-05,
+      "loss": 0.9954,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.878504119006697e-05,
+      "loss": 0.9633,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8770224619214134e-05,
+      "loss": 1.2382,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875540804836129e-05,
+      "loss": 1.005,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.874059147750845e-05,
+      "loss": 1.0923,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8725774906655606e-05,
+      "loss": 0.9299,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.871095833580276e-05,
+      "loss": 1.0834,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.869614176494992e-05,
+      "loss": 0.9882,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8681325194097084e-05,
+      "loss": 1.0145,
+      "step": 890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.866650862324424e-05,
+      "loss": 1.0234,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86516920523914e-05,
+      "loss": 1.1596,
+      "step": 910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8636875481538556e-05,
+      "loss": 1.1069,
+      "step": 920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.862205891068571e-05,
+      "loss": 1.012,
+      "step": 930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.860724233983287e-05,
+      "loss": 1.0147,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8592425768980034e-05,
+      "loss": 1.0027,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.857760919812719e-05,
+      "loss": 1.0529,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.856279262727434e-05,
+      "loss": 1.0556,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.85479760564215e-05,
+      "loss": 0.8887,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.853315948556866e-05,
+      "loss": 1.1216,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.851834291471582e-05,
+      "loss": 1.0372,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.850352634386298e-05,
+      "loss": 1.0452,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8488709773010135e-05,
+      "loss": 1.0786,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.847389320215729e-05,
+      "loss": 1.0466,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.845907663130445e-05,
+      "loss": 1.161,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8444260060451613e-05,
+      "loss": 1.0792,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.842944348959877e-05,
+      "loss": 1.1396,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.841462691874593e-05,
+      "loss": 1.0736,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8399810347893085e-05,
+      "loss": 1.104,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.838499377704024e-05,
+      "loss": 0.948,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.83701772061874e-05,
+      "loss": 0.965,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8355360635334564e-05,
+      "loss": 1.1474,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.834054406448172e-05,
+      "loss": 1.0499,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.832572749362888e-05,
+      "loss": 1.0691,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8310910922776035e-05,
+      "loss": 1.1725,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.829609435192319e-05,
+      "loss": 1.0933,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.828127778107035e-05,
+      "loss": 1.1244,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8266461210217514e-05,
+      "loss": 1.0848,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.825164463936467e-05,
+      "loss": 1.2447,
+      "step": 1180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.823682806851182e-05,
+      "loss": 1.0453,
+      "step": 1190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8222011497658985e-05,
+      "loss": 1.0077,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.820719492680614e-05,
+      "loss": 1.1749,
+      "step": 1210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.81923783559533e-05,
+      "loss": 1.021,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8177561785100464e-05,
+      "loss": 1.1465,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8162745214247614e-05,
+      "loss": 1.0847,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.814792864339477e-05,
+      "loss": 1.0417,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.813311207254193e-05,
+      "loss": 0.9796,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.811829550168909e-05,
+      "loss": 1.148,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.810347893083625e-05,
+      "loss": 1.0537,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.808866235998341e-05,
+      "loss": 1.0994,
+      "step": 1290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8073845789130565e-05,
+      "loss": 0.9678,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.805902921827772e-05,
+      "loss": 1.0571,
+      "step": 1310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.804421264742488e-05,
+      "loss": 1.0449,
+      "step": 1320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.802939607657204e-05,
+      "loss": 1.0071,
+      "step": 1330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.80145795057192e-05,
+      "loss": 0.8959,
+      "step": 1340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.799976293486636e-05,
+      "loss": 1.0767,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7984946364013515e-05,
+      "loss": 0.9891,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.797012979316067e-05,
+      "loss": 1.0674,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.795531322230783e-05,
+      "loss": 1.0322,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.794049665145499e-05,
+      "loss": 1.0522,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.792568008060215e-05,
+      "loss": 1.2132,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.79108635097493e-05,
+      "loss": 1.1291,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7896046938896465e-05,
+      "loss": 1.0099,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.788123036804362e-05,
+      "loss": 1.029,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.786641379719078e-05,
+      "loss": 0.9722,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7851597226337943e-05,
+      "loss": 1.023,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7836780655485094e-05,
+      "loss": 1.0579,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782196408463225e-05,
+      "loss": 1.0452,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7807147513779415e-05,
+      "loss": 1.2105,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.779233094292657e-05,
+      "loss": 1.0491,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.777751437207373e-05,
+      "loss": 0.9678,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7762697801220894e-05,
+      "loss": 0.9197,
+      "step": 1510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7747881230368044e-05,
+      "loss": 0.9753,
+      "step": 1520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.77330646595152e-05,
+      "loss": 1.0212,
+      "step": 1530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.771824808866236e-05,
+      "loss": 1.1102,
+      "step": 1540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.770343151780952e-05,
+      "loss": 1.0254,
+      "step": 1550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.768861494695668e-05,
+      "loss": 1.1006,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.767379837610384e-05,
+      "loss": 0.9692,
+      "step": 1570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7658981805250994e-05,
+      "loss": 1.076,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.764416523439815e-05,
+      "loss": 0.986,
+      "step": 1590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.762934866354531e-05,
+      "loss": 0.9922,
+      "step": 1600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.761453209269247e-05,
+      "loss": 1.1154,
+      "step": 1610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.759971552183963e-05,
+      "loss": 1.1606,
+      "step": 1620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.758489895098678e-05,
+      "loss": 1.0098,
+      "step": 1630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7570082380133944e-05,
+      "loss": 1.0313,
+      "step": 1640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.75552658092811e-05,
+      "loss": 1.0075,
+      "step": 1650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.754044923842826e-05,
+      "loss": 1.1056,
+      "step": 1660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.752563266757542e-05,
+      "loss": 1.0331,
+      "step": 1670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.751081609672258e-05,
+      "loss": 0.9307,
+      "step": 1680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.749599952586973e-05,
+      "loss": 0.9668,
+      "step": 1690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7481182955016895e-05,
+      "loss": 0.9923,
+      "step": 1700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.746636638416405e-05,
+      "loss": 0.9852,
+      "step": 1710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.745154981331121e-05,
+      "loss": 0.959,
+      "step": 1720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.743673324245837e-05,
+      "loss": 1.0467,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7421916671605523e-05,
+      "loss": 1.0341,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.740710010075268e-05,
+      "loss": 1.0286,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7392283529899845e-05,
+      "loss": 1.0069,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7377466959047e-05,
+      "loss": 0.9959,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.736265038819416e-05,
+      "loss": 1.032,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7347833817341316e-05,
+      "loss": 1.0334,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7333017246488474e-05,
+      "loss": 0.9271,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.731820067563563e-05,
+      "loss": 1.0898,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.730338410478279e-05,
+      "loss": 0.8218,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.728856753392995e-05,
+      "loss": 0.9664,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.727375096307711e-05,
+      "loss": 1.1085,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7258934392224267e-05,
+      "loss": 1.0249,
+      "step": 1850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7244117821371424e-05,
+      "loss": 0.9883,
+      "step": 1860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.722930125051858e-05,
+      "loss": 1.0239,
+      "step": 1870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.721448467966574e-05,
+      "loss": 0.9759,
+      "step": 1880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.71996681088129e-05,
+      "loss": 0.9996,
+      "step": 1890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.718485153796006e-05,
+      "loss": 0.8492,
+      "step": 1900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.717003496710721e-05,
+      "loss": 0.9642,
+      "step": 1910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7155218396254374e-05,
+      "loss": 0.9916,
+      "step": 1920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.714040182540153e-05,
+      "loss": 0.902,
+      "step": 1930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.712558525454869e-05,
+      "loss": 1.1098,
+      "step": 1940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.711076868369585e-05,
+      "loss": 1.0721,
+      "step": 1950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7095952112843e-05,
+      "loss": 1.0584,
+      "step": 1960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.708113554199016e-05,
+      "loss": 1.0544,
+      "step": 1970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7066318971137324e-05,
+      "loss": 1.1046,
+      "step": 1980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.705150240028448e-05,
+      "loss": 0.93,
+      "step": 1990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.703668582943164e-05,
+      "loss": 1.1538,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7021869258578796e-05,
+      "loss": 1.1012,
+      "step": 2010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.700705268772595e-05,
+      "loss": 1.056,
+      "step": 2020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.699223611687311e-05,
+      "loss": 1.0762,
+      "step": 2030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6977419546020274e-05,
+      "loss": 1.0884,
+      "step": 2040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.696260297516743e-05,
+      "loss": 1.1416,
+      "step": 2050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.694778640431459e-05,
+      "loss": 1.0391,
+      "step": 2060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6932969833461746e-05,
+      "loss": 1.1021,
+      "step": 2070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.69181532626089e-05,
+      "loss": 1.1112,
+      "step": 2080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.690333669175606e-05,
+      "loss": 1.1211,
+      "step": 2090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.688852012090322e-05,
+      "loss": 0.9748,
+      "step": 2100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.687370355005038e-05,
+      "loss": 0.945,
+      "step": 2110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.685888697919754e-05,
+      "loss": 1.1073,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.684407040834469e-05,
+      "loss": 1.0968,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6829253837491853e-05,
+      "loss": 1.0562,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.681443726663901e-05,
+      "loss": 1.0573,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.679962069578617e-05,
+      "loss": 1.0862,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.678480412493333e-05,
+      "loss": 1.0493,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.676998755408048e-05,
+      "loss": 1.019,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.675517098322764e-05,
+      "loss": 0.8864,
+      "step": 2190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6740354412374804e-05,
+      "loss": 0.9383,
+      "step": 2200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.672553784152196e-05,
+      "loss": 1.0499,
+      "step": 2210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.671072127066912e-05,
+      "loss": 0.9938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6695904699816275e-05,
+      "loss": 1.0305,
+      "step": 2230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.668108812896343e-05,
+      "loss": 0.9835,
+      "step": 2240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.666627155811059e-05,
+      "loss": 0.8994,
+      "step": 2250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6651454987257754e-05,
+      "loss": 1.0598,
+      "step": 2260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.663663841640491e-05,
+      "loss": 0.9745,
+      "step": 2270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.662182184555207e-05,
+      "loss": 0.9789,
+      "step": 2280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6607005274699225e-05,
+      "loss": 1.0458,
+      "step": 2290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.659218870384638e-05,
+      "loss": 0.9545,
+      "step": 2300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.657737213299354e-05,
+      "loss": 0.9021,
+      "step": 2310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6562555562140704e-05,
+      "loss": 0.971,
+      "step": 2320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.654773899128786e-05,
+      "loss": 1.0396,
+      "step": 2330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.653292242043502e-05,
+      "loss": 0.8743,
+      "step": 2340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6518105849582176e-05,
+      "loss": 0.967,
+      "step": 2350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.650328927872933e-05,
+      "loss": 1.1139,
+      "step": 2360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.648847270787649e-05,
+      "loss": 0.9882,
+      "step": 2370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.647365613702365e-05,
+      "loss": 1.0006,
+      "step": 2380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.645883956617081e-05,
+      "loss": 0.9794,
+      "step": 2390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.644402299531796e-05,
+      "loss": 0.9471,
+      "step": 2400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.642920642446512e-05,
+      "loss": 0.8593,
+      "step": 2410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.641438985361228e-05,
+      "loss": 1.0272,
+      "step": 2420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.639957328275944e-05,
+      "loss": 0.9176,
+      "step": 2430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.63847567119066e-05,
+      "loss": 0.9669,
+      "step": 2440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.636994014105376e-05,
+      "loss": 1.014,
+      "step": 2450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.635512357020091e-05,
+      "loss": 0.7286,
+      "step": 2460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.634030699934807e-05,
+      "loss": 1.114,
+      "step": 2470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.632549042849523e-05,
+      "loss": 0.9132,
+      "step": 2480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.631067385764239e-05,
+      "loss": 1.0984,
+      "step": 2490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.629585728678955e-05,
+      "loss": 1.028,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6281040715936705e-05,
+      "loss": 1.1322,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.626622414508386e-05,
+      "loss": 0.9703,
+      "step": 2520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.625140757423102e-05,
+      "loss": 1.0486,
+      "step": 2530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.623659100337818e-05,
+      "loss": 1.1308,
+      "step": 2540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.622177443252534e-05,
+      "loss": 0.9167,
+      "step": 2550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.62069578616725e-05,
+      "loss": 0.9322,
+      "step": 2560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6192141290819655e-05,
+      "loss": 1.1608,
+      "step": 2570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.617732471996681e-05,
+      "loss": 1.0079,
+      "step": 2580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.616250814911397e-05,
+      "loss": 0.9097,
+      "step": 2590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6147691578261134e-05,
+      "loss": 1.0079,
+      "step": 2600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.613287500740829e-05,
+      "loss": 1.0918,
+      "step": 2610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.611805843655545e-05,
+      "loss": 1.0862,
+      "step": 2620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6103241865702605e-05,
+      "loss": 1.0602,
+      "step": 2630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.608842529484976e-05,
+      "loss": 1.1089,
+      "step": 2640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.607360872399692e-05,
+      "loss": 0.9527,
+      "step": 2650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.605879215314408e-05,
+      "loss": 1.0643,
+      "step": 2660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.604397558229124e-05,
+      "loss": 0.9497,
+      "step": 2670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.602915901143839e-05,
+      "loss": 1.048,
+      "step": 2680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.601434244058555e-05,
+      "loss": 1.1039,
+      "step": 2690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.599952586973271e-05,
+      "loss": 0.9535,
+      "step": 2700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.598470929887987e-05,
+      "loss": 1.1689,
+      "step": 2710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.596989272802703e-05,
+      "loss": 0.9185,
+      "step": 2720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5955076157174184e-05,
+      "loss": 0.9434,
+      "step": 2730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.594025958632134e-05,
+      "loss": 1.085,
+      "step": 2740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.59254430154685e-05,
+      "loss": 0.9472,
+      "step": 2750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.591062644461566e-05,
+      "loss": 1.1535,
+      "step": 2760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.589580987376282e-05,
+      "loss": 0.9407,
+      "step": 2770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.588099330290998e-05,
+      "loss": 1.0422,
+      "step": 2780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5866176732057135e-05,
+      "loss": 0.9648,
+      "step": 2790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.585136016120429e-05,
+      "loss": 1.0928,
+      "step": 2800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.583654359035145e-05,
+      "loss": 0.9872,
+      "step": 2810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.582172701949861e-05,
+      "loss": 0.9209,
+      "step": 2820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.580691044864577e-05,
+      "loss": 1.114,
+      "step": 2830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.579209387779293e-05,
+      "loss": 1.1243,
+      "step": 2840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5777277306940085e-05,
+      "loss": 0.9924,
+      "step": 2850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.576246073608724e-05,
+      "loss": 0.9567,
+      "step": 2860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.57476441652344e-05,
+      "loss": 0.8971,
+      "step": 2870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.573282759438156e-05,
+      "loss": 0.9983,
+      "step": 2880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.571801102352872e-05,
+      "loss": 1.0459,
+      "step": 2890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.570319445267587e-05,
+      "loss": 1.0651,
+      "step": 2900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5688377881823035e-05,
+      "loss": 1.1088,
+      "step": 2910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.567356131097019e-05,
+      "loss": 1.1004,
+      "step": 2920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.565874474011735e-05,
+      "loss": 1.0982,
+      "step": 2930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5643928169264507e-05,
+      "loss": 1.1044,
+      "step": 2940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5629111598411664e-05,
+      "loss": 0.9959,
+      "step": 2950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.561429502755882e-05,
+      "loss": 0.9248,
+      "step": 2960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.559947845670598e-05,
+      "loss": 1.0083,
+      "step": 2970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.558466188585314e-05,
+      "loss": 1.0696,
+      "step": 2980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.55698453150003e-05,
+      "loss": 1.0809,
+      "step": 2990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.555502874414746e-05,
+      "loss": 0.9692,
+      "step": 3000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5540212173294614e-05,
+      "loss": 0.8689,
+      "step": 3010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.552539560244177e-05,
+      "loss": 1.0184,
+      "step": 3020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.551057903158893e-05,
+      "loss": 0.9662,
+      "step": 3030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.549576246073609e-05,
+      "loss": 0.9472,
+      "step": 3040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.548094588988325e-05,
+      "loss": 1.0923,
+      "step": 3050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.546612931903041e-05,
+      "loss": 0.9392,
+      "step": 3060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5451312748177564e-05,
+      "loss": 1.0266,
+      "step": 3070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.543649617732472e-05,
+      "loss": 1.0079,
+      "step": 3080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.542167960647188e-05,
+      "loss": 1.0063,
+      "step": 3090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.540686303561904e-05,
+      "loss": 0.9931,
+      "step": 3100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.53920464647662e-05,
+      "loss": 0.8938,
+      "step": 3110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.537722989391335e-05,
+      "loss": 1.0389,
+      "step": 3120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5362413323060514e-05,
+      "loss": 1.0111,
+      "step": 3130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.534759675220767e-05,
+      "loss": 0.9162,
+      "step": 3140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.533278018135483e-05,
+      "loss": 1.0123,
+      "step": 3150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.531796361050199e-05,
+      "loss": 1.007,
+      "step": 3160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.530314703964914e-05,
+      "loss": 1.0215,
+      "step": 3170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.52883304687963e-05,
+      "loss": 0.9833,
+      "step": 3180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5273513897943464e-05,
+      "loss": 1.0657,
+      "step": 3190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.525869732709062e-05,
+      "loss": 1.0104,
+      "step": 3200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.524388075623778e-05,
+      "loss": 1.136,
+      "step": 3210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5229064185384936e-05,
+      "loss": 1.0972,
+      "step": 3220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5214247614532093e-05,
+      "loss": 0.9414,
+      "step": 3230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.519943104367925e-05,
+      "loss": 1.0983,
+      "step": 3240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.518461447282641e-05,
+      "loss": 0.9094,
+      "step": 3250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.516979790197357e-05,
+      "loss": 0.9771,
+      "step": 3260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.515498133112073e-05,
+      "loss": 1.0007,
+      "step": 3270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5140164760267886e-05,
+      "loss": 1.0244,
+      "step": 3280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5125348189415044e-05,
+      "loss": 1.0073,
+      "step": 3290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.51105316185622e-05,
+      "loss": 0.9975,
+      "step": 3300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.509571504770936e-05,
+      "loss": 1.1689,
+      "step": 3310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.508089847685652e-05,
+      "loss": 0.8902,
+      "step": 3320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.506608190600368e-05,
+      "loss": 0.9794,
+      "step": 3330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.505126533515083e-05,
+      "loss": 0.8807,
+      "step": 3340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5036448764297994e-05,
+      "loss": 0.9263,
+      "step": 3350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.502163219344515e-05,
+      "loss": 0.8875,
+      "step": 3360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.500681562259231e-05,
+      "loss": 0.9376,
+      "step": 3370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.499199905173947e-05,
+      "loss": 1.0126,
+      "step": 3380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.497718248088662e-05,
+      "loss": 1.0124,
+      "step": 3390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.496236591003378e-05,
+      "loss": 0.7539,
+      "step": 3400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4947549339180944e-05,
+      "loss": 0.8332,
+      "step": 3410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.49327327683281e-05,
+      "loss": 1.027,
+      "step": 3420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.491791619747526e-05,
+      "loss": 0.8988,
+      "step": 3430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.490309962662242e-05,
+      "loss": 1.1417,
+      "step": 3440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.488828305576957e-05,
+      "loss": 1.0029,
+      "step": 3450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.487346648491673e-05,
+      "loss": 1.0582,
+      "step": 3460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4858649914063894e-05,
+      "loss": 1.0007,
+      "step": 3470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.484383334321105e-05,
+      "loss": 0.8922,
+      "step": 3480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.482901677235821e-05,
+      "loss": 1.0126,
+      "step": 3490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4814200201505366e-05,
+      "loss": 0.8675,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.479938363065252e-05,
+      "loss": 0.9286,
+      "step": 3510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.478456705979968e-05,
+      "loss": 1.0148,
+      "step": 3520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.476975048894684e-05,
+      "loss": 1.0747,
+      "step": 3530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4754933918094e-05,
+      "loss": 1.0847,
+      "step": 3540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.474011734724116e-05,
+      "loss": 1.0646,
+      "step": 3550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.472530077638831e-05,
+      "loss": 1.0878,
+      "step": 3560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.471048420553547e-05,
+      "loss": 0.9359,
+      "step": 3570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.469566763468263e-05,
+      "loss": 1.1106,
+      "step": 3580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.468085106382979e-05,
+      "loss": 0.949,
+      "step": 3590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.466603449297695e-05,
+      "loss": 0.9367,
+      "step": 3600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.465121792212411e-05,
+      "loss": 0.9948,
+      "step": 3610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.463640135127126e-05,
+      "loss": 1.0947,
+      "step": 3620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.462158478041842e-05,
+      "loss": 0.9545,
+      "step": 3630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.460676820956558e-05,
+      "loss": 1.0944,
+      "step": 3640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.459195163871274e-05,
+      "loss": 1.0387,
+      "step": 3650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.45771350678599e-05,
+      "loss": 0.9061,
+      "step": 3660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.456231849700705e-05,
+      "loss": 1.0306,
+      "step": 3670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.454750192615421e-05,
+      "loss": 1.0257,
+      "step": 3680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4532685355301374e-05,
+      "loss": 0.9344,
+      "step": 3690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.451786878444853e-05,
+      "loss": 0.9438,
+      "step": 3700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.450305221359569e-05,
+      "loss": 1.0341,
+      "step": 3710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4488235642742845e-05,
+      "loss": 1.0352,
+      "step": 3720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.447341907189e-05,
+      "loss": 1.1207,
+      "step": 3730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.445860250103716e-05,
+      "loss": 1.0834,
+      "step": 3740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4443785930184324e-05,
+      "loss": 0.8907,
+      "step": 3750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.442896935933148e-05,
+      "loss": 1.1111,
+      "step": 3760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.441415278847864e-05,
+      "loss": 0.9999,
+      "step": 3770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4399336217625795e-05,
+      "loss": 1.0504,
+      "step": 3780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.438451964677295e-05,
+      "loss": 0.9121,
+      "step": 3790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.436970307592011e-05,
+      "loss": 0.9015,
+      "step": 3800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.435488650506727e-05,
+      "loss": 1.0593,
+      "step": 3810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.434006993421443e-05,
+      "loss": 1.0612,
+      "step": 3820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.432525336336159e-05,
+      "loss": 0.9286,
+      "step": 3830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.431043679250874e-05,
+      "loss": 1.0359,
+      "step": 3840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.42956202216559e-05,
+      "loss": 1.0434,
+      "step": 3850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.428080365080306e-05,
+      "loss": 0.9933,
+      "step": 3860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.426598707995022e-05,
+      "loss": 0.9287,
+      "step": 3870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.425117050909738e-05,
+      "loss": 1.0855,
+      "step": 3880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.423635393824453e-05,
+      "loss": 1.0433,
+      "step": 3890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.422153736739169e-05,
+      "loss": 1.072,
+      "step": 3900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.420672079653885e-05,
+      "loss": 0.9849,
+      "step": 3910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.419190422568601e-05,
+      "loss": 0.999,
+      "step": 3920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.417708765483317e-05,
+      "loss": 0.8054,
+      "step": 3930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4162271083980325e-05,
+      "loss": 0.9708,
+      "step": 3940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.414745451312748e-05,
+      "loss": 1.0625,
+      "step": 3950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.413263794227464e-05,
+      "loss": 0.9717,
+      "step": 3960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.41178213714218e-05,
+      "loss": 0.9269,
+      "step": 3970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.410300480056896e-05,
+      "loss": 1.0109,
+      "step": 3980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.408818822971612e-05,
+      "loss": 1.1448,
+      "step": 3990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4073371658863275e-05,
+      "loss": 1.0363,
+      "step": 4000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.405855508801043e-05,
+      "loss": 1.0063,
+      "step": 4010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.404373851715759e-05,
+      "loss": 0.8822,
+      "step": 4020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.402892194630475e-05,
+      "loss": 1.0641,
+      "step": 4030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.401410537545191e-05,
+      "loss": 1.0352,
+      "step": 4040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.399928880459907e-05,
+      "loss": 1.073,
+      "step": 4050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.398447223374622e-05,
+      "loss": 1.0877,
+      "step": 4060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.396965566289338e-05,
+      "loss": 0.9055,
+      "step": 4070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.395483909204054e-05,
+      "loss": 1.0689,
+      "step": 4080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.39400225211877e-05,
+      "loss": 1.0445,
+      "step": 4090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.392520595033486e-05,
+      "loss": 0.9072,
+      "step": 4100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.391038937948201e-05,
+      "loss": 0.9371,
+      "step": 4110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.389557280862917e-05,
+      "loss": 0.8655,
+      "step": 4120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.388075623777633e-05,
+      "loss": 1.1038,
+      "step": 4130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.386593966692349e-05,
+      "loss": 1.0431,
+      "step": 4140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.385112309607065e-05,
+      "loss": 0.9737,
+      "step": 4150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3836306525217804e-05,
+      "loss": 1.0146,
+      "step": 4160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.382148995436496e-05,
+      "loss": 0.9463,
+      "step": 4170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.380667338351212e-05,
+      "loss": 1.048,
+      "step": 4180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.379185681265928e-05,
+      "loss": 0.9281,
+      "step": 4190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.377704024180644e-05,
+      "loss": 1.0834,
+      "step": 4200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.37622236709536e-05,
+      "loss": 1.007,
+      "step": 4210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3747407100100754e-05,
+      "loss": 0.9686,
+      "step": 4220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.373259052924791e-05,
+      "loss": 1.0291,
+      "step": 4230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.371777395839507e-05,
+      "loss": 1.0465,
+      "step": 4240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.370295738754223e-05,
+      "loss": 0.953,
+      "step": 4250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.368814081668939e-05,
+      "loss": 1.0508,
+      "step": 4260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.367332424583655e-05,
+      "loss": 1.0006,
+      "step": 4270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3658507674983704e-05,
+      "loss": 1.1245,
+      "step": 4280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.364369110413086e-05,
+      "loss": 0.9801,
+      "step": 4290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.362887453327802e-05,
+      "loss": 0.9444,
+      "step": 4300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.361405796242518e-05,
+      "loss": 1.0968,
+      "step": 4310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.359924139157234e-05,
+      "loss": 0.9368,
+      "step": 4320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.358442482071949e-05,
+      "loss": 0.9025,
+      "step": 4330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.356960824986665e-05,
+      "loss": 0.8605,
+      "step": 4340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.355479167901381e-05,
+      "loss": 0.9366,
+      "step": 4350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.353997510816097e-05,
+      "loss": 0.8986,
+      "step": 4360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3525158537308126e-05,
+      "loss": 0.9287,
+      "step": 4370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3510341966455284e-05,
+      "loss": 1.0341,
+      "step": 4380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.349552539560244e-05,
+      "loss": 1.0212,
+      "step": 4390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.34807088247496e-05,
+      "loss": 0.8867,
+      "step": 4400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.346589225389676e-05,
+      "loss": 0.9691,
+      "step": 4410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.345107568304392e-05,
+      "loss": 0.9443,
+      "step": 4420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3436259112191076e-05,
+      "loss": 1.0488,
+      "step": 4430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3421442541338234e-05,
+      "loss": 1.1335,
+      "step": 4440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.340662597048539e-05,
+      "loss": 0.9009,
+      "step": 4450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.339180939963255e-05,
+      "loss": 1.0512,
+      "step": 4460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.337699282877971e-05,
+      "loss": 0.89,
+      "step": 4470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.336217625792687e-05,
+      "loss": 0.9175,
+      "step": 4480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.334735968707403e-05,
+      "loss": 1.0487,
+      "step": 4490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3332543116221184e-05,
+      "loss": 0.9249,
+      "step": 4500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.331772654536834e-05,
+      "loss": 0.9176,
+      "step": 4510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.33029099745155e-05,
+      "loss": 1.0861,
+      "step": 4520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.328809340366266e-05,
+      "loss": 1.0183,
+      "step": 4530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.327327683280982e-05,
+      "loss": 0.9382,
+      "step": 4540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.325846026195697e-05,
+      "loss": 1.011,
+      "step": 4550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3243643691104134e-05,
+      "loss": 1.1504,
+      "step": 4560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.322882712025129e-05,
+      "loss": 0.9311,
+      "step": 4570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.321401054939845e-05,
+      "loss": 1.0303,
+      "step": 4580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.319919397854561e-05,
+      "loss": 0.9909,
+      "step": 4590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.318437740769277e-05,
+      "loss": 0.9847,
+      "step": 4600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.316956083683992e-05,
+      "loss": 0.8849,
+      "step": 4610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.315474426598708e-05,
+      "loss": 1.0116,
+      "step": 4620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.313992769513424e-05,
+      "loss": 1.1569,
+      "step": 4630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.31251111242814e-05,
+      "loss": 0.8751,
+      "step": 4640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3110294553428556e-05,
+      "loss": 0.9307,
+      "step": 4650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.309547798257571e-05,
+      "loss": 1.0062,
+      "step": 4660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.308066141172287e-05,
+      "loss": 0.9799,
+      "step": 4670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.306584484087003e-05,
+      "loss": 0.9787,
+      "step": 4680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.305102827001719e-05,
+      "loss": 0.9239,
+      "step": 4690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.303621169916435e-05,
+      "loss": 0.9535,
+      "step": 4700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3021395128311506e-05,
+      "loss": 1.0515,
+      "step": 4710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.300657855745866e-05,
+      "loss": 1.037,
+      "step": 4720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.299176198660582e-05,
+      "loss": 1.1239,
+      "step": 4730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.297694541575298e-05,
+      "loss": 1.0813,
+      "step": 4740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.296212884490014e-05,
+      "loss": 1.2004,
+      "step": 4750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.29473122740473e-05,
+      "loss": 0.9914,
+      "step": 4760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2932495703194456e-05,
+      "loss": 0.9044,
+      "step": 4770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2917679132341613e-05,
+      "loss": 1.0464,
+      "step": 4780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.290286256148877e-05,
+      "loss": 0.9826,
+      "step": 4790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.288804599063593e-05,
+      "loss": 0.9495,
+      "step": 4800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.287322941978309e-05,
+      "loss": 1.0386,
+      "step": 4810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.285841284893025e-05,
+      "loss": 0.9099,
+      "step": 4820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.28435962780774e-05,
+      "loss": 0.9316,
+      "step": 4830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2828779707224564e-05,
+      "loss": 1.0725,
+      "step": 4840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.281396313637172e-05,
+      "loss": 0.8463,
+      "step": 4850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.279914656551888e-05,
+      "loss": 1.1508,
+      "step": 4860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.278432999466604e-05,
+      "loss": 0.9393,
+      "step": 4870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.276951342381319e-05,
+      "loss": 1.0294,
+      "step": 4880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.275469685296035e-05,
+      "loss": 1.1232,
+      "step": 4890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.273988028210751e-05,
+      "loss": 0.9461,
+      "step": 4900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.272506371125467e-05,
+      "loss": 0.8287,
+      "step": 4910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.271024714040183e-05,
+      "loss": 1.0617,
+      "step": 4920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2695430569548986e-05,
+      "loss": 0.9236,
+      "step": 4930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.268061399869614e-05,
+      "loss": 0.9911,
+      "step": 4940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.26657974278433e-05,
+      "loss": 1.1111,
+      "step": 4950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.265098085699046e-05,
+      "loss": 0.9179,
+      "step": 4960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.263616428613762e-05,
+      "loss": 0.8431,
+      "step": 4970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.262134771528478e-05,
+      "loss": 0.9731,
+      "step": 4980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2606531144431936e-05,
+      "loss": 0.9344,
+      "step": 4990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.259171457357909e-05,
+      "loss": 1.0364,
+      "step": 5000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.257689800272625e-05,
+      "loss": 0.9216,
+      "step": 5010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.256208143187341e-05,
+      "loss": 0.9568,
+      "step": 5020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.254726486102057e-05,
+      "loss": 0.9962,
+      "step": 5030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.253244829016773e-05,
+      "loss": 1.1549,
+      "step": 5040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.251763171931488e-05,
+      "loss": 0.9061,
+      "step": 5050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.250281514846204e-05,
+      "loss": 0.9633,
+      "step": 5060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.24879985776092e-05,
+      "loss": 0.9426,
+      "step": 5070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.247318200675636e-05,
+      "loss": 1.0316,
+      "step": 5080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.245836543590352e-05,
+      "loss": 0.879,
+      "step": 5090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.244354886505067e-05,
+      "loss": 0.8395,
+      "step": 5100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.242873229419783e-05,
+      "loss": 1.022,
+      "step": 5110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.241391572334499e-05,
+      "loss": 0.9806,
+      "step": 5120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.239909915249215e-05,
+      "loss": 0.9603,
+      "step": 5130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.238428258163931e-05,
+      "loss": 0.8212,
+      "step": 5140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2369466010786465e-05,
+      "loss": 0.9402,
+      "step": 5150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.235464943993362e-05,
+      "loss": 0.9632,
+      "step": 5160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.233983286908078e-05,
+      "loss": 0.857,
+      "step": 5170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.232501629822794e-05,
+      "loss": 1.0024,
+      "step": 5180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.23101997273751e-05,
+      "loss": 0.9256,
+      "step": 5190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.229538315652226e-05,
+      "loss": 0.9668,
+      "step": 5200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2280566585669415e-05,
+      "loss": 0.9422,
+      "step": 5210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.226575001481657e-05,
+      "loss": 0.9228,
+      "step": 5220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.225093344396373e-05,
+      "loss": 0.9124,
+      "step": 5230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.223611687311089e-05,
+      "loss": 1.0929,
+      "step": 5240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.222130030225805e-05,
+      "loss": 0.9791,
+      "step": 5250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.220648373140521e-05,
+      "loss": 1.0369,
+      "step": 5260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.219166716055236e-05,
+      "loss": 1.0614,
+      "step": 5270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.217685058969952e-05,
+      "loss": 1.0704,
+      "step": 5280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.216203401884668e-05,
+      "loss": 0.977,
+      "step": 5290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.214721744799384e-05,
+      "loss": 1.0282,
+      "step": 5300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2132400877141e-05,
+      "loss": 0.9825,
+      "step": 5310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.211758430628815e-05,
+      "loss": 1.0764,
+      "step": 5320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.210276773543531e-05,
+      "loss": 0.9701,
+      "step": 5330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.208795116458247e-05,
+      "loss": 1.0994,
+      "step": 5340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.207313459372963e-05,
+      "loss": 0.8733,
+      "step": 5350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.205831802287679e-05,
+      "loss": 1.0822,
+      "step": 5360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.204350145202395e-05,
+      "loss": 0.9434,
+      "step": 5370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.20286848811711e-05,
+      "loss": 0.9525,
+      "step": 5380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.201386831031826e-05,
+      "loss": 0.9808,
+      "step": 5390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.199905173946542e-05,
+      "loss": 0.9915,
+      "step": 5400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.198423516861258e-05,
+      "loss": 0.9708,
+      "step": 5410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.196941859775974e-05,
+      "loss": 1.0643,
+      "step": 5420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1954602026906895e-05,
+      "loss": 1.0476,
+      "step": 5430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.193978545605405e-05,
+      "loss": 1.0005,
+      "step": 5440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.192496888520121e-05,
+      "loss": 1.0663,
+      "step": 5450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1910152314348366e-05,
+      "loss": 1.0395,
+      "step": 5460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.189533574349553e-05,
+      "loss": 1.0204,
+      "step": 5470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.188051917264269e-05,
+      "loss": 0.9644,
+      "step": 5480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.186570260178984e-05,
+      "loss": 0.983,
+      "step": 5490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1850886030937e-05,
+      "loss": 1.0131,
+      "step": 5500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.183606946008416e-05,
+      "loss": 1.0005,
+      "step": 5510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1821252889231316e-05,
+      "loss": 0.9998,
+      "step": 5520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.180643631837848e-05,
+      "loss": 0.9927,
+      "step": 5530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.179161974752564e-05,
+      "loss": 0.8264,
+      "step": 5540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.177680317667279e-05,
+      "loss": 1.0266,
+      "step": 5550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.176198660581995e-05,
+      "loss": 0.8833,
+      "step": 5560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.174717003496711e-05,
+      "loss": 0.9666,
+      "step": 5570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.173235346411427e-05,
+      "loss": 1.0156,
+      "step": 5580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.171753689326143e-05,
+      "loss": 1.0261,
+      "step": 5590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.170272032240858e-05,
+      "loss": 1.0929,
+      "step": 5600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.168790375155574e-05,
+      "loss": 0.9007,
+      "step": 5610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.16730871807029e-05,
+      "loss": 0.9849,
+      "step": 5620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.165827060985006e-05,
+      "loss": 0.9266,
+      "step": 5630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.164345403899722e-05,
+      "loss": 0.8687,
+      "step": 5640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1628637468144374e-05,
+      "loss": 1.0976,
+      "step": 5650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.161382089729153e-05,
+      "loss": 1.001,
+      "step": 5660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.159900432643869e-05,
+      "loss": 0.922,
+      "step": 5670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.158418775558585e-05,
+      "loss": 1.0987,
+      "step": 5680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.156937118473301e-05,
+      "loss": 1.035,
+      "step": 5690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.155455461388017e-05,
+      "loss": 0.9958,
+      "step": 5700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1539738043027324e-05,
+      "loss": 0.8683,
+      "step": 5710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.152492147217448e-05,
+      "loss": 0.8124,
+      "step": 5720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.151010490132164e-05,
+      "loss": 1.0338,
+      "step": 5730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1495288330468796e-05,
+      "loss": 0.9689,
+      "step": 5740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.148047175961596e-05,
+      "loss": 1.0404,
+      "step": 5750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.146565518876312e-05,
+      "loss": 0.9264,
+      "step": 5760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.145083861791027e-05,
+      "loss": 1.0461,
+      "step": 5770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.143602204705743e-05,
+      "loss": 0.8862,
+      "step": 5780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.142120547620459e-05,
+      "loss": 0.912,
+      "step": 5790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1406388905351746e-05,
+      "loss": 0.8659,
+      "step": 5800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.139157233449891e-05,
+      "loss": 0.9026,
+      "step": 5810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.137675576364606e-05,
+      "loss": 1.0321,
+      "step": 5820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.136193919279322e-05,
+      "loss": 0.9552,
+      "step": 5830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.134712262194038e-05,
+      "loss": 1.0345,
+      "step": 5840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.133230605108754e-05,
+      "loss": 0.9723,
+      "step": 5850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1317489480234696e-05,
+      "loss": 0.9258,
+      "step": 5860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1302672909381853e-05,
+      "loss": 0.9752,
+      "step": 5870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.128785633852901e-05,
+      "loss": 1.0206,
+      "step": 5880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.127303976767617e-05,
+      "loss": 0.9742,
+      "step": 5890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.125822319682333e-05,
+      "loss": 0.9583,
+      "step": 5900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.124340662597049e-05,
+      "loss": 0.8648,
+      "step": 5910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1228590055117646e-05,
+      "loss": 1.0787,
+      "step": 5920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1213773484264804e-05,
+      "loss": 0.9813,
+      "step": 5930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.119895691341196e-05,
+      "loss": 0.944,
+      "step": 5940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.118414034255912e-05,
+      "loss": 1.0075,
+      "step": 5950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.116932377170628e-05,
+      "loss": 0.8044,
+      "step": 5960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.115450720085344e-05,
+      "loss": 0.9448,
+      "step": 5970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1139690630000597e-05,
+      "loss": 0.7978,
+      "step": 5980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1124874059147754e-05,
+      "loss": 0.8473,
+      "step": 5990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.111005748829491e-05,
+      "loss": 1.087,
+      "step": 6000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.109524091744207e-05,
+      "loss": 1.024,
+      "step": 6010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1080424346589226e-05,
+      "loss": 1.0458,
+      "step": 6020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.106560777573639e-05,
+      "loss": 1.0961,
+      "step": 6030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.105079120488354e-05,
+      "loss": 0.9593,
+      "step": 6040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.10359746340307e-05,
+      "loss": 1.1049,
+      "step": 6050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.102115806317786e-05,
+      "loss": 1.0515,
+      "step": 6060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.100634149232502e-05,
+      "loss": 0.9561,
+      "step": 6070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0991524921472176e-05,
+      "loss": 0.9915,
+      "step": 6080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.097670835061933e-05,
+      "loss": 0.942,
+      "step": 6090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.096189177976649e-05,
+      "loss": 0.9765,
+      "step": 6100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.094707520891365e-05,
+      "loss": 1.0278,
+      "step": 6110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.093225863806081e-05,
+      "loss": 0.9689,
+      "step": 6120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.091744206720797e-05,
+      "loss": 1.0525,
+      "step": 6130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0902625496355126e-05,
+      "loss": 0.923,
+      "step": 6140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.088780892550228e-05,
+      "loss": 0.9831,
+      "step": 6150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.087299235464944e-05,
+      "loss": 0.919,
+      "step": 6160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.08581757837966e-05,
+      "loss": 1.0675,
+      "step": 6170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.084335921294376e-05,
+      "loss": 1.0158,
+      "step": 6180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.082854264209092e-05,
+      "loss": 0.837,
+      "step": 6190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0813726071238076e-05,
+      "loss": 1.0987,
+      "step": 6200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.079890950038523e-05,
+      "loss": 1.0349,
+      "step": 6210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.078409292953239e-05,
+      "loss": 0.8741,
+      "step": 6220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.076927635867955e-05,
+      "loss": 0.7855,
+      "step": 6230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.075445978782671e-05,
+      "loss": 0.8909,
+      "step": 6240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.073964321697387e-05,
+      "loss": 1.0483,
+      "step": 6250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.072482664612102e-05,
+      "loss": 1.0883,
+      "step": 6260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0710010075268183e-05,
+      "loss": 0.9039,
+      "step": 6270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.069519350441534e-05,
+      "loss": 0.9711,
+      "step": 6280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.06803769335625e-05,
+      "loss": 1.0803,
+      "step": 6290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0665560362709655e-05,
+      "loss": 1.0707,
+      "step": 6300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.065074379185681e-05,
+      "loss": 0.9018,
+      "step": 6310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.063592722100397e-05,
+      "loss": 0.866,
+      "step": 6320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.062111065015113e-05,
+      "loss": 0.9473,
+      "step": 6330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.060629407929829e-05,
+      "loss": 0.9391,
+      "step": 6340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.059147750844545e-05,
+      "loss": 0.9851,
+      "step": 6350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0576660937592605e-05,
+      "loss": 0.9714,
+      "step": 6360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.056184436673976e-05,
+      "loss": 0.9525,
+      "step": 6370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.054702779588692e-05,
+      "loss": 0.9674,
+      "step": 6380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.053221122503408e-05,
+      "loss": 0.9294,
+      "step": 6390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.051739465418124e-05,
+      "loss": 0.9353,
+      "step": 6400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.05025780833284e-05,
+      "loss": 0.9975,
+      "step": 6410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0487761512475555e-05,
+      "loss": 1.066,
+      "step": 6420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.047294494162271e-05,
+      "loss": 0.8707,
+      "step": 6430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.045812837076987e-05,
+      "loss": 0.957,
+      "step": 6440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.044331179991703e-05,
+      "loss": 1.0121,
+      "step": 6450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.042849522906419e-05,
+      "loss": 1.0795,
+      "step": 6460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.041367865821135e-05,
+      "loss": 0.9982,
+      "step": 6470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.03988620873585e-05,
+      "loss": 0.934,
+      "step": 6480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.038404551650566e-05,
+      "loss": 0.9686,
+      "step": 6490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.036922894565282e-05,
+      "loss": 1.0467,
+      "step": 6500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.035441237479998e-05,
+      "loss": 0.9294,
+      "step": 6510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.033959580394714e-05,
+      "loss": 1.1935,
+      "step": 6520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.03247792330943e-05,
+      "loss": 0.9153,
+      "step": 6530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.030996266224145e-05,
+      "loss": 0.9231,
+      "step": 6540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.029514609138861e-05,
+      "loss": 0.9228,
+      "step": 6550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.028032952053577e-05,
+      "loss": 0.9463,
+      "step": 6560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.026551294968293e-05,
+      "loss": 0.9007,
+      "step": 6570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0250696378830085e-05,
+      "loss": 0.8671,
+      "step": 6580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.023587980797724e-05,
+      "loss": 0.9874,
+      "step": 6590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.02210632371244e-05,
+      "loss": 1.0277,
+      "step": 6600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0206246666271556e-05,
+      "loss": 1.1138,
+      "step": 6610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.019143009541872e-05,
+      "loss": 0.9372,
+      "step": 6620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.017661352456588e-05,
+      "loss": 1.0057,
+      "step": 6630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0161796953713035e-05,
+      "loss": 0.8845,
+      "step": 6640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.014698038286019e-05,
+      "loss": 0.8917,
+      "step": 6650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.013216381200735e-05,
+      "loss": 1.1626,
+      "step": 6660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0117347241154507e-05,
+      "loss": 0.8921,
+      "step": 6670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.010253067030167e-05,
+      "loss": 1.0248,
+      "step": 6680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.008771409944883e-05,
+      "loss": 1.1001,
+      "step": 6690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0072897528595985e-05,
+      "loss": 0.969,
+      "step": 6700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.005808095774314e-05,
+      "loss": 0.8646,
+      "step": 6710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.00432643868903e-05,
+      "loss": 1.0792,
+      "step": 6720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.002844781603746e-05,
+      "loss": 0.9803,
+      "step": 6730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.001363124518462e-05,
+      "loss": 0.9235,
+      "step": 6740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.999881467433178e-05,
+      "loss": 0.9415,
+      "step": 6750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.998399810347893e-05,
+      "loss": 0.7717,
+      "step": 6760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.996918153262609e-05,
+      "loss": 0.9583,
+      "step": 6770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.995436496177325e-05,
+      "loss": 1.0054,
+      "step": 6780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.993954839092041e-05,
+      "loss": 1.0008,
+      "step": 6790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.992473182006757e-05,
+      "loss": 0.9701,
+      "step": 6800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.990991524921472e-05,
+      "loss": 0.8991,
+      "step": 6810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.989509867836188e-05,
+      "loss": 0.8607,
+      "step": 6820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.988028210750904e-05,
+      "loss": 0.8719,
+      "step": 6830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.98654655366562e-05,
+      "loss": 0.9115,
+      "step": 6840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.985064896580336e-05,
+      "loss": 1.0135,
+      "step": 6850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9835832394950514e-05,
+      "loss": 0.9497,
+      "step": 6860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.982101582409767e-05,
+      "loss": 0.9448,
+      "step": 6870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.980619925324483e-05,
+      "loss": 0.9963,
+      "step": 6880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9791382682391986e-05,
+      "loss": 0.8949,
+      "step": 6890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.977656611153915e-05,
+      "loss": 0.9696,
+      "step": 6900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.976174954068631e-05,
+      "loss": 1.01,
+      "step": 6910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9746932969833465e-05,
+      "loss": 0.8498,
+      "step": 6920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.973211639898062e-05,
+      "loss": 1.0444,
+      "step": 6930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.971729982812778e-05,
+      "loss": 0.9505,
+      "step": 6940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9702483257274936e-05,
+      "loss": 1.0077,
+      "step": 6950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.96876666864221e-05,
+      "loss": 0.9244,
+      "step": 6960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.967285011556926e-05,
+      "loss": 0.9844,
+      "step": 6970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.965803354471641e-05,
+      "loss": 1.0796,
+      "step": 6980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.964321697386357e-05,
+      "loss": 1.012,
+      "step": 6990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.962840040301073e-05,
+      "loss": 0.9539,
+      "step": 7000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9613583832157886e-05,
+      "loss": 0.9021,
+      "step": 7010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.959876726130505e-05,
+      "loss": 0.945,
+      "step": 7020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.95839506904522e-05,
+      "loss": 0.8773,
+      "step": 7030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.956913411959936e-05,
+      "loss": 0.8579,
+      "step": 7040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.955431754874652e-05,
+      "loss": 1.0599,
+      "step": 7050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.953950097789368e-05,
+      "loss": 0.9359,
+      "step": 7060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9524684407040837e-05,
+      "loss": 0.9445,
+      "step": 7070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9509867836187994e-05,
+      "loss": 1.0743,
+      "step": 7080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.949505126533515e-05,
+      "loss": 1.0425,
+      "step": 7090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.948023469448231e-05,
+      "loss": 0.861,
+      "step": 7100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.946541812362947e-05,
+      "loss": 1.0149,
+      "step": 7110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.945060155277663e-05,
+      "loss": 0.8354,
+      "step": 7120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.943578498192379e-05,
+      "loss": 1.02,
+      "step": 7130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9420968411070944e-05,
+      "loss": 1.0401,
+      "step": 7140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.94061518402181e-05,
+      "loss": 0.9268,
+      "step": 7150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.939133526936526e-05,
+      "loss": 0.9768,
+      "step": 7160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9376518698512416e-05,
+      "loss": 0.8638,
+      "step": 7170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.936170212765958e-05,
+      "loss": 1.0255,
+      "step": 7180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.934688555680674e-05,
+      "loss": 0.9134,
+      "step": 7190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.933206898595389e-05,
+      "loss": 0.8389,
+      "step": 7200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.931725241510105e-05,
+      "loss": 0.8027,
+      "step": 7210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.930243584424821e-05,
+      "loss": 0.8722,
+      "step": 7220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9287619273395366e-05,
+      "loss": 1.066,
+      "step": 7230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.927280270254253e-05,
+      "loss": 0.8792,
+      "step": 7240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.925798613168968e-05,
+      "loss": 0.9674,
+      "step": 7250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.924316956083684e-05,
+      "loss": 1.0199,
+      "step": 7260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9228352989984e-05,
+      "loss": 0.9654,
+      "step": 7270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.921353641913116e-05,
+      "loss": 0.8794,
+      "step": 7280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9198719848278316e-05,
+      "loss": 0.9812,
+      "step": 7290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.918390327742547e-05,
+      "loss": 0.979,
+      "step": 7300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.916908670657263e-05,
+      "loss": 0.9721,
+      "step": 7310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.915427013571979e-05,
+      "loss": 0.8746,
+      "step": 7320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.913945356486695e-05,
+      "loss": 0.9368,
+      "step": 7330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.912463699401411e-05,
+      "loss": 0.918,
+      "step": 7340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9109820423161266e-05,
+      "loss": 0.9911,
+      "step": 7350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9095003852308423e-05,
+      "loss": 0.8009,
+      "step": 7360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.908018728145558e-05,
+      "loss": 0.9735,
+      "step": 7370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.906537071060274e-05,
+      "loss": 0.9262,
+      "step": 7380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.90505541397499e-05,
+      "loss": 0.9529,
+      "step": 7390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.903573756889706e-05,
+      "loss": 0.8379,
+      "step": 7400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9020920998044216e-05,
+      "loss": 0.981,
+      "step": 7410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.900610442719137e-05,
+      "loss": 1.0949,
+      "step": 7420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.899128785633853e-05,
+      "loss": 1.0619,
+      "step": 7430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.897647128548569e-05,
+      "loss": 0.9317,
+      "step": 7440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8961654714632845e-05,
+      "loss": 1.0149,
+      "step": 7450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.894683814378001e-05,
+      "loss": 0.9187,
+      "step": 7460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.893202157292716e-05,
+      "loss": 0.8749,
+      "step": 7470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.891720500207432e-05,
+      "loss": 0.9824,
+      "step": 7480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.890238843122148e-05,
+      "loss": 0.8977,
+      "step": 7490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.888757186036864e-05,
+      "loss": 0.9572,
+      "step": 7500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8872755289515795e-05,
+      "loss": 0.8585,
+      "step": 7510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.885793871866296e-05,
+      "loss": 0.919,
+      "step": 7520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.884312214781011e-05,
+      "loss": 1.094,
+      "step": 7530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.882830557695727e-05,
+      "loss": 0.8935,
+      "step": 7540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.881348900610443e-05,
+      "loss": 1.0104,
+      "step": 7550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.879867243525159e-05,
+      "loss": 0.9139,
+      "step": 7560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8783855864398746e-05,
+      "loss": 0.901,
+      "step": 7570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.87690392935459e-05,
+      "loss": 0.9488,
+      "step": 7580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.875422272269306e-05,
+      "loss": 1.0754,
+      "step": 7590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.873940615184022e-05,
+      "loss": 0.9621,
+      "step": 7600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.872458958098738e-05,
+      "loss": 0.849,
+      "step": 7610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.870977301013454e-05,
+      "loss": 1.0123,
+      "step": 7620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8694956439281696e-05,
+      "loss": 1.0595,
+      "step": 7630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.868013986842885e-05,
+      "loss": 1.0262,
+      "step": 7640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.866532329757601e-05,
+      "loss": 1.1201,
+      "step": 7650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.865050672672317e-05,
+      "loss": 0.8926,
+      "step": 7660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.863569015587033e-05,
+      "loss": 0.8442,
+      "step": 7670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.862087358501749e-05,
+      "loss": 0.961,
+      "step": 7680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8606057014164646e-05,
+      "loss": 0.9387,
+      "step": 7690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8591240443311796e-05,
+      "loss": 0.8551,
+      "step": 7700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.857642387245896e-05,
+      "loss": 0.8485,
+      "step": 7710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.856160730160612e-05,
+      "loss": 1.1136,
+      "step": 7720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8546790730753275e-05,
+      "loss": 0.9127,
+      "step": 7730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.853197415990044e-05,
+      "loss": 0.9789,
+      "step": 7740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.851715758904759e-05,
+      "loss": 1.0526,
+      "step": 7750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8502341018194747e-05,
+      "loss": 0.8497,
+      "step": 7760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.848752444734191e-05,
+      "loss": 0.839,
+      "step": 7770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.847270787648907e-05,
+      "loss": 0.9596,
+      "step": 7780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8457891305636225e-05,
+      "loss": 0.8706,
+      "step": 7790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.844307473478338e-05,
+      "loss": 1.0107,
+      "step": 7800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.842825816393054e-05,
+      "loss": 0.9414,
+      "step": 7810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.84134415930777e-05,
+      "loss": 0.8719,
+      "step": 7820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.839862502222486e-05,
+      "loss": 0.8495,
+      "step": 7830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.838380845137202e-05,
+      "loss": 0.9224,
+      "step": 7840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8368991880519175e-05,
+      "loss": 0.9091,
+      "step": 7850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.835417530966633e-05,
+      "loss": 0.9843,
+      "step": 7860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.833935873881349e-05,
+      "loss": 1.1209,
+      "step": 7870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.832454216796065e-05,
+      "loss": 1.0426,
+      "step": 7880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.830972559710781e-05,
+      "loss": 0.8406,
+      "step": 7890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.829490902625497e-05,
+      "loss": 0.861,
+      "step": 7900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8280092455402125e-05,
+      "loss": 1.0723,
+      "step": 7910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.826527588454928e-05,
+      "loss": 0.9372,
+      "step": 7920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.825045931369644e-05,
+      "loss": 1.0065,
+      "step": 7930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.82356427428436e-05,
+      "loss": 0.7633,
+      "step": 7940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.822082617199076e-05,
+      "loss": 0.9963,
+      "step": 7950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.820600960113792e-05,
+      "loss": 0.8949,
+      "step": 7960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.819119303028507e-05,
+      "loss": 1.0342,
+      "step": 7970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8176376459432226e-05,
+      "loss": 1.0523,
+      "step": 7980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.816155988857939e-05,
+      "loss": 0.9841,
+      "step": 7990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.814674331772655e-05,
+      "loss": 1.0401,
+      "step": 8000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8131926746873704e-05,
+      "loss": 0.8284,
+      "step": 8010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.811711017602086e-05,
+      "loss": 0.9808,
+      "step": 8020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.810229360516802e-05,
+      "loss": 0.8408,
+      "step": 8030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8087477034315176e-05,
+      "loss": 0.9917,
+      "step": 8040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.807266046346234e-05,
+      "loss": 0.904,
+      "step": 8050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.80578438926095e-05,
+      "loss": 0.9815,
+      "step": 8060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8043027321756655e-05,
+      "loss": 0.9306,
+      "step": 8070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.802821075090381e-05,
+      "loss": 0.9941,
+      "step": 8080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.801339418005097e-05,
+      "loss": 0.8978,
+      "step": 8090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7998577609198126e-05,
+      "loss": 0.941,
+      "step": 8100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.798376103834529e-05,
+      "loss": 0.918,
+      "step": 8110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.796894446749245e-05,
+      "loss": 0.7802,
+      "step": 8120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7954127896639605e-05,
+      "loss": 0.9219,
+      "step": 8130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.793931132578676e-05,
+      "loss": 1.0816,
+      "step": 8140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.792449475493392e-05,
+      "loss": 1.0741,
+      "step": 8150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7909678184081077e-05,
+      "loss": 1.0095,
+      "step": 8160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.789486161322824e-05,
+      "loss": 0.962,
+      "step": 8170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.78800450423754e-05,
+      "loss": 0.8198,
+      "step": 8180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.786522847152255e-05,
+      "loss": 0.9874,
+      "step": 8190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.785041190066971e-05,
+      "loss": 0.997,
+      "step": 8200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.783559532981687e-05,
+      "loss": 1.0218,
+      "step": 8210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.782077875896403e-05,
+      "loss": 0.7353,
+      "step": 8220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.780596218811119e-05,
+      "loss": 0.9453,
+      "step": 8230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.779114561725834e-05,
+      "loss": 0.9067,
+      "step": 8240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.77763290464055e-05,
+      "loss": 0.9107,
+      "step": 8250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7761512475552656e-05,
+      "loss": 0.8818,
+      "step": 8260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.774669590469982e-05,
+      "loss": 1.065,
+      "step": 8270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.773187933384698e-05,
+      "loss": 0.9564,
+      "step": 8280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7717062762994134e-05,
+      "loss": 1.0434,
+      "step": 8290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.770224619214129e-05,
+      "loss": 1.0674,
+      "step": 8300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.768742962128845e-05,
+      "loss": 1.0128,
+      "step": 8310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7672613050435606e-05,
+      "loss": 0.9528,
+      "step": 8320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.765779647958277e-05,
+      "loss": 1.088,
+      "step": 8330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.764297990872993e-05,
+      "loss": 1.0491,
+      "step": 8340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7628163337877084e-05,
+      "loss": 1.0906,
+      "step": 8350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.761334676702424e-05,
+      "loss": 0.9301,
+      "step": 8360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.75985301961714e-05,
+      "loss": 0.8296,
+      "step": 8370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7583713625318556e-05,
+      "loss": 0.9707,
+      "step": 8380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.756889705446572e-05,
+      "loss": 0.9587,
+      "step": 8390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.755408048361288e-05,
+      "loss": 1.0983,
+      "step": 8400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.753926391276003e-05,
+      "loss": 0.7814,
+      "step": 8410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.752444734190719e-05,
+      "loss": 0.9767,
+      "step": 8420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.750963077105435e-05,
+      "loss": 0.9191,
+      "step": 8430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7494814200201506e-05,
+      "loss": 1.133,
+      "step": 8440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.747999762934867e-05,
+      "loss": 0.7876,
+      "step": 8450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.746518105849583e-05,
+      "loss": 1.0342,
+      "step": 8460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.745036448764298e-05,
+      "loss": 0.9359,
+      "step": 8470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.743554791679014e-05,
+      "loss": 1.0841,
+      "step": 8480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.74207313459373e-05,
+      "loss": 0.9303,
+      "step": 8490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7405914775084456e-05,
+      "loss": 0.947,
+      "step": 8500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.739109820423162e-05,
+      "loss": 0.9999,
+      "step": 8510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.737628163337877e-05,
+      "loss": 0.9578,
+      "step": 8520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.736146506252593e-05,
+      "loss": 0.9773,
+      "step": 8530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7346648491673085e-05,
+      "loss": 0.9933,
+      "step": 8540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.733183192082025e-05,
+      "loss": 0.8242,
+      "step": 8550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7317015349967406e-05,
+      "loss": 1.1116,
+      "step": 8560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7302198779114564e-05,
+      "loss": 1.1778,
+      "step": 8570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.728738220826172e-05,
+      "loss": 0.9228,
+      "step": 8580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.727256563740888e-05,
+      "loss": 1.0696,
+      "step": 8590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7257749066556035e-05,
+      "loss": 1.104,
+      "step": 8600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.72429324957032e-05,
+      "loss": 0.7433,
+      "step": 8610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.722811592485036e-05,
+      "loss": 0.9487,
+      "step": 8620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7213299353997514e-05,
+      "loss": 0.9236,
+      "step": 8630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.719848278314467e-05,
+      "loss": 0.9623,
+      "step": 8640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.718366621229183e-05,
+      "loss": 1.021,
+      "step": 8650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7168849641438986e-05,
+      "loss": 0.749,
+      "step": 8660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.715403307058615e-05,
+      "loss": 0.8871,
+      "step": 8670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.713921649973331e-05,
+      "loss": 0.8794,
+      "step": 8680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.712439992888046e-05,
+      "loss": 1.0036,
+      "step": 8690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.710958335802762e-05,
+      "loss": 0.9535,
+      "step": 8700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.709476678717478e-05,
+      "loss": 0.9561,
+      "step": 8710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7079950216321936e-05,
+      "loss": 1.0284,
+      "step": 8720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.70651336454691e-05,
+      "loss": 0.9994,
+      "step": 8730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.705031707461625e-05,
+      "loss": 0.8966,
+      "step": 8740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.703550050376341e-05,
+      "loss": 0.8629,
+      "step": 8750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.702068393291057e-05,
+      "loss": 0.9028,
+      "step": 8760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.700586736205773e-05,
+      "loss": 0.949,
+      "step": 8770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6991050791204886e-05,
+      "loss": 0.9702,
+      "step": 8780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.697623422035204e-05,
+      "loss": 0.8712,
+      "step": 8790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.69614176494992e-05,
+      "loss": 0.8101,
+      "step": 8800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.694660107864636e-05,
+      "loss": 0.9015,
+      "step": 8810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6931784507793515e-05,
+      "loss": 1.0166,
+      "step": 8820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.691696793694068e-05,
+      "loss": 1.0416,
+      "step": 8830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6902151366087836e-05,
+      "loss": 0.9694,
+      "step": 8840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.688733479523499e-05,
+      "loss": 0.9738,
+      "step": 8850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.687251822438215e-05,
+      "loss": 1.0464,
+      "step": 8860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.685770165352931e-05,
+      "loss": 0.9831,
+      "step": 8870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6842885082676465e-05,
+      "loss": 0.8784,
+      "step": 8880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.682806851182363e-05,
+      "loss": 0.9262,
+      "step": 8890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6813251940970786e-05,
+      "loss": 0.9408,
+      "step": 8900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.679843537011794e-05,
+      "loss": 1.0725,
+      "step": 8910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.67836187992651e-05,
+      "loss": 0.9675,
+      "step": 8920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.676880222841226e-05,
+      "loss": 0.9087,
+      "step": 8930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6753985657559415e-05,
+      "loss": 0.9308,
+      "step": 8940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.673916908670658e-05,
+      "loss": 1.0129,
+      "step": 8950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.672435251585373e-05,
+      "loss": 0.7679,
+      "step": 8960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.670953594500089e-05,
+      "loss": 0.865,
+      "step": 8970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.669471937414805e-05,
+      "loss": 0.9606,
+      "step": 8980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.667990280329521e-05,
+      "loss": 0.9638,
+      "step": 8990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6665086232442365e-05,
+      "loss": 0.9136,
+      "step": 9000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.665026966158952e-05,
+      "loss": 1.0182,
+      "step": 9010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.663545309073668e-05,
+      "loss": 0.7749,
+      "step": 9020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.662063651988384e-05,
+      "loss": 0.9811,
+      "step": 9030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6605819949031e-05,
+      "loss": 0.9869,
+      "step": 9040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.659100337817816e-05,
+      "loss": 1.0581,
+      "step": 9050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6576186807325316e-05,
+      "loss": 0.8655,
+      "step": 9060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.656137023647247e-05,
+      "loss": 0.9068,
+      "step": 9070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.654655366561963e-05,
+      "loss": 0.8552,
+      "step": 9080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.653173709476679e-05,
+      "loss": 0.948,
+      "step": 9090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6516920523913944e-05,
+      "loss": 0.8968,
+      "step": 9100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.650210395306111e-05,
+      "loss": 0.9175,
+      "step": 9110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6487287382208266e-05,
+      "loss": 1.0115,
+      "step": 9120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6472470811355416e-05,
+      "loss": 0.8408,
+      "step": 9130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.645765424050258e-05,
+      "loss": 0.9666,
+      "step": 9140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.644283766964974e-05,
+      "loss": 0.861,
+      "step": 9150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6428021098796895e-05,
+      "loss": 0.8259,
+      "step": 9160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.641320452794406e-05,
+      "loss": 0.8378,
+      "step": 9170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.639838795709121e-05,
+      "loss": 0.9386,
+      "step": 9180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6383571386238366e-05,
+      "loss": 0.8758,
+      "step": 9190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.636875481538553e-05,
+      "loss": 0.9089,
+      "step": 9200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.635393824453269e-05,
+      "loss": 0.8981,
+      "step": 9210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6339121673679845e-05,
+      "loss": 1.0558,
+      "step": 9220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6324305102827e-05,
+      "loss": 0.9055,
+      "step": 9230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.630948853197416e-05,
+      "loss": 0.9934,
+      "step": 9240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6294671961121317e-05,
+      "loss": 0.9359,
+      "step": 9250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.627985539026848e-05,
+      "loss": 0.8586,
+      "step": 9260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.626503881941564e-05,
+      "loss": 0.9114,
+      "step": 9270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6250222248562795e-05,
+      "loss": 0.8345,
+      "step": 9280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.623540567770995e-05,
+      "loss": 0.9753,
+      "step": 9290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.622058910685711e-05,
+      "loss": 0.8734,
+      "step": 9300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.620577253600427e-05,
+      "loss": 0.9912,
+      "step": 9310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.619095596515143e-05,
+      "loss": 0.9848,
+      "step": 9320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.617613939429859e-05,
+      "loss": 0.9692,
+      "step": 9330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6161322823445745e-05,
+      "loss": 0.9421,
+      "step": 9340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.61465062525929e-05,
+      "loss": 0.9079,
+      "step": 9350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.613168968174006e-05,
+      "loss": 1.0107,
+      "step": 9360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.611687311088722e-05,
+      "loss": 0.902,
+      "step": 9370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6102056540034374e-05,
+      "loss": 0.9594,
+      "step": 9380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.608723996918154e-05,
+      "loss": 0.7266,
+      "step": 9390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.607242339832869e-05,
+      "loss": 1.0683,
+      "step": 9400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6057606827475846e-05,
+      "loss": 0.9873,
+      "step": 9410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.604279025662301e-05,
+      "loss": 0.9732,
+      "step": 9420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.602797368577017e-05,
+      "loss": 0.8969,
+      "step": 9430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6013157114917324e-05,
+      "loss": 0.9062,
+      "step": 9440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.599834054406449e-05,
+      "loss": 1.0581,
+      "step": 9450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.598352397321164e-05,
+      "loss": 0.882,
+      "step": 9460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5968707402358796e-05,
+      "loss": 0.755,
+      "step": 9470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.595389083150596e-05,
+      "loss": 0.8925,
+      "step": 9480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.593907426065312e-05,
+      "loss": 0.9063,
+      "step": 9490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5924257689800274e-05,
+      "loss": 1.0206,
+      "step": 9500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.590944111894743e-05,
+      "loss": 0.8922,
+      "step": 9510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.589462454809459e-05,
+      "loss": 0.9306,
+      "step": 9520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5879807977241746e-05,
+      "loss": 1.1392,
+      "step": 9530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.586499140638891e-05,
+      "loss": 0.9836,
+      "step": 9540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.585017483553607e-05,
+      "loss": 0.8159,
+      "step": 9550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5835358264683225e-05,
+      "loss": 0.7914,
+      "step": 9560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.582054169383038e-05,
+      "loss": 0.8551,
+      "step": 9570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.580572512297754e-05,
+      "loss": 0.8242,
+      "step": 9580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5790908552124696e-05,
+      "loss": 1.0202,
+      "step": 9590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.577609198127186e-05,
+      "loss": 0.9322,
+      "step": 9600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.576127541041902e-05,
+      "loss": 0.9435,
+      "step": 9610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5746458839566175e-05,
+      "loss": 0.7511,
+      "step": 9620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.573164226871333e-05,
+      "loss": 0.9171,
+      "step": 9630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.571682569786049e-05,
+      "loss": 0.8806,
+      "step": 9640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5702009127007646e-05,
+      "loss": 0.935,
+      "step": 9650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5687192556154804e-05,
+      "loss": 1.114,
+      "step": 9660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.567237598530197e-05,
+      "loss": 0.8805,
+      "step": 9670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.565755941444912e-05,
+      "loss": 1.0953,
+      "step": 9680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5642742843596275e-05,
+      "loss": 0.9546,
+      "step": 9690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.562792627274344e-05,
+      "loss": 1.0753,
+      "step": 9700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.56131097018906e-05,
+      "loss": 0.9717,
+      "step": 9710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5598293131037754e-05,
+      "loss": 0.8966,
+      "step": 9720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.558347656018491e-05,
+      "loss": 1.0347,
+      "step": 9730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.556865998933207e-05,
+      "loss": 0.9045,
+      "step": 9740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5553843418479226e-05,
+      "loss": 0.984,
+      "step": 9750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.553902684762639e-05,
+      "loss": 0.8778,
+      "step": 9760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.552421027677355e-05,
+      "loss": 0.9551,
+      "step": 9770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5509393705920704e-05,
+      "loss": 0.9584,
+      "step": 9780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.549457713506786e-05,
+      "loss": 0.948,
+      "step": 9790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.547976056421502e-05,
+      "loss": 1.0751,
+      "step": 9800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5464943993362176e-05,
+      "loss": 0.9411,
+      "step": 9810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.545012742250934e-05,
+      "loss": 0.8559,
+      "step": 9820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.54353108516565e-05,
+      "loss": 1.0811,
+      "step": 9830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5420494280803654e-05,
+      "loss": 0.7798,
+      "step": 9840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.540567770995081e-05,
+      "loss": 0.9944,
+      "step": 9850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.539086113909797e-05,
+      "loss": 0.8336,
+      "step": 9860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5376044568245126e-05,
+      "loss": 1.0403,
+      "step": 9870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.536122799739229e-05,
+      "loss": 0.8135,
+      "step": 9880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.534641142653945e-05,
+      "loss": 0.8156,
+      "step": 9890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.53315948556866e-05,
+      "loss": 0.95,
+      "step": 9900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.531677828483376e-05,
+      "loss": 0.8798,
+      "step": 9910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.530196171398092e-05,
+      "loss": 0.9948,
+      "step": 9920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5287145143128076e-05,
+      "loss": 0.9372,
+      "step": 9930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.527232857227523e-05,
+      "loss": 0.9383,
+      "step": 9940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.525751200142239e-05,
+      "loss": 1.0386,
+      "step": 9950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.524269543056955e-05,
+      "loss": 0.9028,
+      "step": 9960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5227878859716705e-05,
+      "loss": 1.0153,
+      "step": 9970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.521306228886387e-05,
+      "loss": 0.8812,
+      "step": 9980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5198245718011026e-05,
+      "loss": 1.0119,
+      "step": 9990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5183429147158183e-05,
+      "loss": 0.9994,
+      "step": 10000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.516861257630534e-05,
+      "loss": 0.869,
+      "step": 10010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.51537960054525e-05,
+      "loss": 0.8866,
+      "step": 10020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5138979434599655e-05,
+      "loss": 0.8339,
+      "step": 10030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.512416286374682e-05,
+      "loss": 0.7662,
+      "step": 10040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5109346292893976e-05,
+      "loss": 0.808,
+      "step": 10050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5094529722041134e-05,
+      "loss": 0.8775,
+      "step": 10060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.507971315118829e-05,
+      "loss": 0.8669,
+      "step": 10070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.506489658033545e-05,
+      "loss": 0.8483,
+      "step": 10080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5050080009482605e-05,
+      "loss": 1.1318,
+      "step": 10090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.503526343862977e-05,
+      "loss": 0.9721,
+      "step": 10100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5020446867776927e-05,
+      "loss": 0.9693,
+      "step": 10110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.500563029692408e-05,
+      "loss": 0.9194,
+      "step": 10120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.499081372607124e-05,
+      "loss": 1.0312,
+      "step": 10130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.49759971552184e-05,
+      "loss": 0.9082,
+      "step": 10140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4961180584365556e-05,
+      "loss": 0.969,
+      "step": 10150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.494636401351272e-05,
+      "loss": 0.9248,
+      "step": 10160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.493154744265987e-05,
+      "loss": 1.0493,
+      "step": 10170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.491673087180703e-05,
+      "loss": 0.8598,
+      "step": 10180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.490191430095419e-05,
+      "loss": 1.0603,
+      "step": 10190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.488709773010135e-05,
+      "loss": 1.0821,
+      "step": 10200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4872281159248506e-05,
+      "loss": 0.9239,
+      "step": 10210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.485746458839566e-05,
+      "loss": 0.9464,
+      "step": 10220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.484264801754282e-05,
+      "loss": 0.9686,
+      "step": 10230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.482783144668998e-05,
+      "loss": 0.9362,
+      "step": 10240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4813014875837135e-05,
+      "loss": 0.9344,
+      "step": 10250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.47981983049843e-05,
+      "loss": 1.0437,
+      "step": 10260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4783381734131456e-05,
+      "loss": 0.9584,
+      "step": 10270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.476856516327861e-05,
+      "loss": 0.9738,
+      "step": 10280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.475374859242577e-05,
+      "loss": 0.8704,
+      "step": 10290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.473893202157293e-05,
+      "loss": 0.8944,
+      "step": 10300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4724115450720085e-05,
+      "loss": 0.794,
+      "step": 10310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.470929887986725e-05,
+      "loss": 1.0281,
+      "step": 10320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4694482309014406e-05,
+      "loss": 1.0935,
+      "step": 10330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4679665738161556e-05,
+      "loss": 1.0274,
+      "step": 10340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.466484916730872e-05,
+      "loss": 1.0199,
+      "step": 10350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.465003259645588e-05,
+      "loss": 1.0275,
+      "step": 10360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4635216025603035e-05,
+      "loss": 0.9856,
+      "step": 10370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46203994547502e-05,
+      "loss": 1.0552,
+      "step": 10380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.460558288389735e-05,
+      "loss": 1.0258,
+      "step": 10390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.459076631304451e-05,
+      "loss": 0.7794,
+      "step": 10400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.457594974219167e-05,
+      "loss": 0.8032,
+      "step": 10410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.456113317133883e-05,
+      "loss": 0.9867,
+      "step": 10420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4546316600485985e-05,
+      "loss": 0.939,
+      "step": 10430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.453150002963315e-05,
+      "loss": 0.9754,
+      "step": 10440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.45166834587803e-05,
+      "loss": 0.9524,
+      "step": 10450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.450186688792746e-05,
+      "loss": 0.8299,
+      "step": 10460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.448705031707462e-05,
+      "loss": 0.8595,
+      "step": 10470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.447223374622178e-05,
+      "loss": 0.7695,
+      "step": 10480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4457417175368935e-05,
+      "loss": 0.774,
+      "step": 10490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.444260060451609e-05,
+      "loss": 0.8111,
+      "step": 10500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.442778403366325e-05,
+      "loss": 0.92,
+      "step": 10510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.441296746281041e-05,
+      "loss": 0.8905,
+      "step": 10520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4398150891957564e-05,
+      "loss": 0.7788,
+      "step": 10530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.438333432110473e-05,
+      "loss": 0.7675,
+      "step": 10540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4368517750251885e-05,
+      "loss": 0.9089,
+      "step": 10550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4353701179399036e-05,
+      "loss": 0.984,
+      "step": 10560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.43388846085462e-05,
+      "loss": 0.9988,
+      "step": 10570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.432406803769336e-05,
+      "loss": 0.9656,
+      "step": 10580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4309251466840514e-05,
+      "loss": 1.0129,
+      "step": 10590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.429443489598768e-05,
+      "loss": 0.9738,
+      "step": 10600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4279618325134836e-05,
+      "loss": 0.9136,
+      "step": 10610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4264801754281986e-05,
+      "loss": 0.9943,
+      "step": 10620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.424998518342915e-05,
+      "loss": 0.8487,
+      "step": 10630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.423516861257631e-05,
+      "loss": 0.8095,
+      "step": 10640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4220352041723465e-05,
+      "loss": 1.0404,
+      "step": 10650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.420553547087063e-05,
+      "loss": 0.886,
+      "step": 10660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.419071890001778e-05,
+      "loss": 1.0033,
+      "step": 10670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4175902329164936e-05,
+      "loss": 0.8785,
+      "step": 10680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.41610857583121e-05,
+      "loss": 0.9679,
+      "step": 10690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.414626918745926e-05,
+      "loss": 0.9523,
+      "step": 10700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4131452616606415e-05,
+      "loss": 0.8169,
+      "step": 10710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.411663604575357e-05,
+      "loss": 1.0053,
+      "step": 10720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.410181947490073e-05,
+      "loss": 0.8736,
+      "step": 10730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4087002904047886e-05,
+      "loss": 0.9577,
+      "step": 10740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.407218633319505e-05,
+      "loss": 0.8155,
+      "step": 10750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.405736976234221e-05,
+      "loss": 0.8046,
+      "step": 10760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4042553191489365e-05,
+      "loss": 0.8739,
+      "step": 10770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.402773662063652e-05,
+      "loss": 0.8769,
+      "step": 10780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.401292004978368e-05,
+      "loss": 0.9993,
+      "step": 10790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.399810347893084e-05,
+      "loss": 0.9002,
+      "step": 10800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3983286908077994e-05,
+      "loss": 1.0381,
+      "step": 10810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.396847033722516e-05,
+      "loss": 0.8421,
+      "step": 10820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3953653766372315e-05,
+      "loss": 0.8218,
+      "step": 10830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3938837195519466e-05,
+      "loss": 0.9304,
+      "step": 10840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.392402062466663e-05,
+      "loss": 0.8721,
+      "step": 10850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.390920405381379e-05,
+      "loss": 0.9466,
+      "step": 10860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3894387482960944e-05,
+      "loss": 0.8224,
+      "step": 10870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.387957091210811e-05,
+      "loss": 0.8728,
+      "step": 10880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.386475434125526e-05,
+      "loss": 1.0159,
+      "step": 10890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3849937770402416e-05,
+      "loss": 0.9289,
+      "step": 10900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.383512119954958e-05,
+      "loss": 0.8679,
+      "step": 10910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.382030462869674e-05,
+      "loss": 1.0493,
+      "step": 10920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3805488057843894e-05,
+      "loss": 0.8017,
+      "step": 10930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.379067148699105e-05,
+      "loss": 0.7417,
+      "step": 10940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.377585491613821e-05,
+      "loss": 0.8948,
+      "step": 10950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3761038345285366e-05,
+      "loss": 0.9638,
+      "step": 10960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.374622177443253e-05,
+      "loss": 0.9754,
+      "step": 10970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.373140520357969e-05,
+      "loss": 0.8849,
+      "step": 10980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3716588632726844e-05,
+      "loss": 0.9619,
+      "step": 10990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3701772061874e-05,
+      "loss": 1.0277,
+      "step": 11000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.368695549102116e-05,
+      "loss": 0.9002,
+      "step": 11010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3672138920168316e-05,
+      "loss": 1.0167,
+      "step": 11020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.365732234931548e-05,
+      "loss": 1.1125,
+      "step": 11030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.364250577846264e-05,
+      "loss": 0.9814,
+      "step": 11040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3627689207609795e-05,
+      "loss": 0.8165,
+      "step": 11050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3612872636756945e-05,
+      "loss": 1.0137,
+      "step": 11060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.359805606590411e-05,
+      "loss": 0.9903,
+      "step": 11070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3583239495051266e-05,
+      "loss": 0.9677,
+      "step": 11080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3568422924198423e-05,
+      "loss": 0.9325,
+      "step": 11090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.355360635334559e-05,
+      "loss": 0.9825,
+      "step": 11100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.353878978249274e-05,
+      "loss": 0.8533,
+      "step": 11110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3523973211639895e-05,
+      "loss": 1.0349,
+      "step": 11120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.350915664078706e-05,
+      "loss": 0.8865,
+      "step": 11130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3494340069934216e-05,
+      "loss": 1.0031,
+      "step": 11140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3479523499081374e-05,
+      "loss": 0.8539,
+      "step": 11150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.346470692822853e-05,
+      "loss": 0.8381,
+      "step": 11160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.344989035737569e-05,
+      "loss": 0.9522,
+      "step": 11170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3435073786522845e-05,
+      "loss": 0.9666,
+      "step": 11180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.342025721567001e-05,
+      "loss": 0.819,
+      "step": 11190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3405440644817167e-05,
+      "loss": 0.9633,
+      "step": 11200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3390624073964324e-05,
+      "loss": 1.0502,
+      "step": 11210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.337580750311148e-05,
+      "loss": 0.909,
+      "step": 11220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.336099093225864e-05,
+      "loss": 1.0321,
+      "step": 11230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3346174361405796e-05,
+      "loss": 1.063,
+      "step": 11240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.333135779055296e-05,
+      "loss": 0.9974,
+      "step": 11250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.331654121970012e-05,
+      "loss": 0.9965,
+      "step": 11260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3301724648847274e-05,
+      "loss": 0.7699,
+      "step": 11270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.328690807799443e-05,
+      "loss": 0.9671,
+      "step": 11280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.327209150714159e-05,
+      "loss": 0.9856,
+      "step": 11290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3257274936288746e-05,
+      "loss": 0.8764,
+      "step": 11300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.324245836543591e-05,
+      "loss": 0.9155,
+      "step": 11310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.322764179458307e-05,
+      "loss": 1.0399,
+      "step": 11320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.321282522373022e-05,
+      "loss": 0.8548,
+      "step": 11330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3198008652877375e-05,
+      "loss": 1.0151,
+      "step": 11340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.318319208202454e-05,
+      "loss": 0.7865,
+      "step": 11350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3168375511171696e-05,
+      "loss": 0.8713,
+      "step": 11360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.315355894031885e-05,
+      "loss": 1.0597,
+      "step": 11370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.313874236946602e-05,
+      "loss": 0.8394,
+      "step": 11380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.312392579861317e-05,
+      "loss": 0.9528,
+      "step": 11390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3109109227760325e-05,
+      "loss": 0.8585,
+      "step": 11400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.309429265690749e-05,
+      "loss": 0.8442,
+      "step": 11410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3079476086054646e-05,
+      "loss": 1.0275,
+      "step": 11420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.30646595152018e-05,
+      "loss": 1.0349,
+      "step": 11430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.304984294434896e-05,
+      "loss": 0.7873,
+      "step": 11440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.303502637349612e-05,
+      "loss": 0.9402,
+      "step": 11450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3020209802643275e-05,
+      "loss": 0.8423,
+      "step": 11460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.300539323179044e-05,
+      "loss": 0.846,
+      "step": 11470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2990576660937596e-05,
+      "loss": 0.9767,
+      "step": 11480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2975760090084753e-05,
+      "loss": 0.8893,
+      "step": 11490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.296094351923191e-05,
+      "loss": 1.0934,
+      "step": 11500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.294612694837907e-05,
+      "loss": 0.8449,
+      "step": 11510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2931310377526225e-05,
+      "loss": 0.9332,
+      "step": 11520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.291649380667339e-05,
+      "loss": 0.9488,
+      "step": 11530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2901677235820546e-05,
+      "loss": 0.7982,
+      "step": 11540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2886860664967704e-05,
+      "loss": 0.9872,
+      "step": 11550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.287204409411486e-05,
+      "loss": 0.8299,
+      "step": 11560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.285722752326202e-05,
+      "loss": 0.8706,
+      "step": 11570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2842410952409175e-05,
+      "loss": 1.0442,
+      "step": 11580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.282759438155634e-05,
+      "loss": 0.9559,
+      "step": 11590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2812777810703497e-05,
+      "loss": 0.8583,
+      "step": 11600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.279796123985065e-05,
+      "loss": 0.9606,
+      "step": 11610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2783144668997804e-05,
+      "loss": 0.9162,
+      "step": 11620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.276832809814497e-05,
+      "loss": 1.0091,
+      "step": 11630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2753511527292125e-05,
+      "loss": 0.9522,
+      "step": 11640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.273869495643928e-05,
+      "loss": 0.9861,
+      "step": 11650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.272387838558644e-05,
+      "loss": 0.9244,
+      "step": 11660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.27090618147336e-05,
+      "loss": 0.9683,
+      "step": 11670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2694245243880754e-05,
+      "loss": 0.8495,
+      "step": 11680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.267942867302792e-05,
+      "loss": 0.876,
+      "step": 11690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2664612102175076e-05,
+      "loss": 0.8485,
+      "step": 11700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.264979553132223e-05,
+      "loss": 0.9864,
+      "step": 11710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.263497896046939e-05,
+      "loss": 0.9987,
+      "step": 11720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.262016238961655e-05,
+      "loss": 0.8928,
+      "step": 11730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2605345818763705e-05,
+      "loss": 0.8925,
+      "step": 11740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.259052924791087e-05,
+      "loss": 0.9792,
+      "step": 11750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2575712677058026e-05,
+      "loss": 0.953,
+      "step": 11760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.256089610620518e-05,
+      "loss": 0.8147,
+      "step": 11770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.254607953535234e-05,
+      "loss": 0.8235,
+      "step": 11780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.25312629644995e-05,
+      "loss": 0.8287,
+      "step": 11790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2516446393646655e-05,
+      "loss": 0.993,
+      "step": 11800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.250162982279382e-05,
+      "loss": 0.9838,
+      "step": 11810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2486813251940976e-05,
+      "loss": 1.0423,
+      "step": 11820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2471996681088126e-05,
+      "loss": 0.9387,
+      "step": 11830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.245718011023529e-05,
+      "loss": 0.893,
+      "step": 11840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.244236353938245e-05,
+      "loss": 0.9076,
+      "step": 11850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2427546968529605e-05,
+      "loss": 0.8877,
+      "step": 11860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.241273039767677e-05,
+      "loss": 0.8991,
+      "step": 11870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.239791382682392e-05,
+      "loss": 0.9059,
+      "step": 11880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2383097255971077e-05,
+      "loss": 0.9591,
+      "step": 11890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2368280685118234e-05,
+      "loss": 1.0124,
+      "step": 11900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.23534641142654e-05,
+      "loss": 0.9362,
+      "step": 11910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2338647543412555e-05,
+      "loss": 0.87,
+      "step": 11920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.232383097255971e-05,
+      "loss": 0.925,
+      "step": 11930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.230901440170687e-05,
+      "loss": 0.9574,
+      "step": 11940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.229419783085403e-05,
+      "loss": 0.936,
+      "step": 11950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2279381260001184e-05,
+      "loss": 0.7628,
+      "step": 11960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.226456468914835e-05,
+      "loss": 0.9247,
+      "step": 11970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2249748118295505e-05,
+      "loss": 1.0546,
+      "step": 11980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.223493154744266e-05,
+      "loss": 0.7548,
+      "step": 11990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.222011497658982e-05,
+      "loss": 0.8808,
+      "step": 12000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.220529840573698e-05,
+      "loss": 0.875,
+      "step": 12010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2190481834884134e-05,
+      "loss": 0.9583,
+      "step": 12020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.21756652640313e-05,
+      "loss": 0.9905,
+      "step": 12030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2160848693178455e-05,
+      "loss": 0.9819,
+      "step": 12040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2146032122325606e-05,
+      "loss": 0.8693,
+      "step": 12050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.213121555147277e-05,
+      "loss": 0.851,
+      "step": 12060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.211639898061993e-05,
+      "loss": 0.9138,
+      "step": 12070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2101582409767084e-05,
+      "loss": 0.95,
+      "step": 12080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.208676583891425e-05,
+      "loss": 1.0171,
+      "step": 12090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.20719492680614e-05,
+      "loss": 0.7627,
+      "step": 12100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2057132697208556e-05,
+      "loss": 1.0109,
+      "step": 12110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.204231612635572e-05,
+      "loss": 0.9426,
+      "step": 12120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.202749955550288e-05,
+      "loss": 0.9074,
+      "step": 12130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2012682984650035e-05,
+      "loss": 0.8692,
+      "step": 12140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.199786641379719e-05,
+      "loss": 0.8587,
+      "step": 12150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.198304984294435e-05,
+      "loss": 0.8806,
+      "step": 12160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1968233272091506e-05,
+      "loss": 0.8254,
+      "step": 12170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1953416701238663e-05,
+      "loss": 0.9484,
+      "step": 12180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.193860013038583e-05,
+      "loss": 1.0171,
+      "step": 12190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1923783559532985e-05,
+      "loss": 0.8648,
+      "step": 12200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.190896698868014e-05,
+      "loss": 0.9771,
+      "step": 12210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.18941504178273e-05,
+      "loss": 0.9566,
+      "step": 12220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1879333846974456e-05,
+      "loss": 1.0407,
+      "step": 12230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1864517276121614e-05,
+      "loss": 0.8478,
+      "step": 12240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.184970070526878e-05,
+      "loss": 0.9864,
+      "step": 12250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1834884134415935e-05,
+      "loss": 0.9417,
+      "step": 12260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1820067563563085e-05,
+      "loss": 0.8376,
+      "step": 12270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.180525099271025e-05,
+      "loss": 0.8979,
+      "step": 12280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1790434421857407e-05,
+      "loss": 0.9631,
+      "step": 12290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1775617851004564e-05,
+      "loss": 0.9826,
+      "step": 12300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.176080128015173e-05,
+      "loss": 0.8654,
+      "step": 12310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.174598470929888e-05,
+      "loss": 0.8159,
+      "step": 12320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1731168138446035e-05,
+      "loss": 1.0192,
+      "step": 12330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.17163515675932e-05,
+      "loss": 1.0343,
+      "step": 12340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.170153499674036e-05,
+      "loss": 0.8398,
+      "step": 12350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1686718425887514e-05,
+      "loss": 0.8501,
+      "step": 12360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.167190185503468e-05,
+      "loss": 0.8306,
+      "step": 12370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.165708528418183e-05,
+      "loss": 0.8984,
+      "step": 12380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1642268713328986e-05,
+      "loss": 0.8919,
+      "step": 12390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.162745214247615e-05,
+      "loss": 0.8088,
+      "step": 12400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.161263557162331e-05,
+      "loss": 0.9202,
+      "step": 12410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1597819000770464e-05,
+      "loss": 0.9711,
+      "step": 12420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.158300242991762e-05,
+      "loss": 0.8329,
+      "step": 12430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.156818585906478e-05,
+      "loss": 1.0967,
+      "step": 12440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1553369288211936e-05,
+      "loss": 0.9618,
+      "step": 12450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.153855271735909e-05,
+      "loss": 0.9283,
+      "step": 12460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.152373614650626e-05,
+      "loss": 1.1064,
+      "step": 12470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1508919575653414e-05,
+      "loss": 1.0143,
+      "step": 12480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1494103004800565e-05,
+      "loss": 0.8655,
+      "step": 12490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.147928643394773e-05,
+      "loss": 0.965,
+      "step": 12500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1464469863094886e-05,
+      "loss": 0.8912,
+      "step": 12510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.144965329224204e-05,
+      "loss": 0.8465,
+      "step": 12520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.143483672138921e-05,
+      "loss": 0.9123,
+      "step": 12530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1420020150536364e-05,
+      "loss": 0.8008,
+      "step": 12540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1405203579683515e-05,
+      "loss": 0.9382,
+      "step": 12550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.139038700883068e-05,
+      "loss": 0.8803,
+      "step": 12560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1375570437977836e-05,
+      "loss": 1.0119,
+      "step": 12570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1360753867124993e-05,
+      "loss": 0.9621,
+      "step": 12580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.134593729627216e-05,
+      "loss": 0.8976,
+      "step": 12590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.133112072541931e-05,
+      "loss": 0.9723,
+      "step": 12600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1316304154566465e-05,
+      "loss": 0.9753,
+      "step": 12610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.130148758371363e-05,
+      "loss": 0.8936,
+      "step": 12620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1286671012860786e-05,
+      "loss": 1.0737,
+      "step": 12630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1271854442007944e-05,
+      "loss": 0.8781,
+      "step": 12640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.12570378711551e-05,
+      "loss": 0.8997,
+      "step": 12650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.124222130030226e-05,
+      "loss": 0.9192,
+      "step": 12660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1227404729449415e-05,
+      "loss": 0.8524,
+      "step": 12670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.121258815859658e-05,
+      "loss": 1.0335,
+      "step": 12680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1197771587743737e-05,
+      "loss": 0.9639,
+      "step": 12690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1182955016890894e-05,
+      "loss": 0.818,
+      "step": 12700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.116813844603805e-05,
+      "loss": 0.9447,
+      "step": 12710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.115332187518521e-05,
+      "loss": 0.9139,
+      "step": 12720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1138505304332365e-05,
+      "loss": 1.004,
+      "step": 12730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.112368873347952e-05,
+      "loss": 0.8301,
+      "step": 12740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.110887216262669e-05,
+      "loss": 1.0328,
+      "step": 12750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1094055591773844e-05,
+      "loss": 0.9522,
+      "step": 12760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1079239020920994e-05,
+      "loss": 0.8784,
+      "step": 12770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.106442245006816e-05,
+      "loss": 0.9544,
+      "step": 12780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1049605879215316e-05,
+      "loss": 0.9545,
+      "step": 12790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.103478930836247e-05,
+      "loss": 0.8227,
+      "step": 12800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.101997273750964e-05,
+      "loss": 1.0182,
+      "step": 12810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.100515616665679e-05,
+      "loss": 0.9353,
+      "step": 12820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0990339595803945e-05,
+      "loss": 1.0409,
+      "step": 12830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.097552302495111e-05,
+      "loss": 0.9231,
+      "step": 12840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0960706454098266e-05,
+      "loss": 1.0812,
+      "step": 12850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.094588988324542e-05,
+      "loss": 0.8666,
+      "step": 12860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.093107331239258e-05,
+      "loss": 0.8488,
+      "step": 12870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.091625674153974e-05,
+      "loss": 0.8775,
+      "step": 12880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0901440170686895e-05,
+      "loss": 0.8833,
+      "step": 12890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.088662359983406e-05,
+      "loss": 0.8949,
+      "step": 12900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0871807028981216e-05,
+      "loss": 0.9946,
+      "step": 12910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.085699045812837e-05,
+      "loss": 0.9376,
+      "step": 12920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.084217388727553e-05,
+      "loss": 0.8217,
+      "step": 12930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.082735731642269e-05,
+      "loss": 0.8726,
+      "step": 12940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0812540745569845e-05,
+      "loss": 0.912,
+      "step": 12950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.079772417471701e-05,
+      "loss": 0.985,
+      "step": 12960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0782907603864166e-05,
+      "loss": 0.8384,
+      "step": 12970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.076809103301132e-05,
+      "loss": 0.8627,
+      "step": 12980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0753274462158474e-05,
+      "loss": 0.8514,
+      "step": 12990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.073845789130564e-05,
+      "loss": 0.8775,
+      "step": 13000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0723641320452795e-05,
+      "loss": 0.8567,
+      "step": 13010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.070882474959995e-05,
+      "loss": 0.7999,
+      "step": 13020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0694008178747116e-05,
+      "loss": 0.8365,
+      "step": 13030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.067919160789427e-05,
+      "loss": 0.83,
+      "step": 13040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0664375037041424e-05,
+      "loss": 0.9431,
+      "step": 13050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.064955846618859e-05,
+      "loss": 0.838,
+      "step": 13060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0634741895335745e-05,
+      "loss": 1.0418,
+      "step": 13070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.06199253244829e-05,
+      "loss": 0.7926,
+      "step": 13080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.060510875363006e-05,
+      "loss": 0.8476,
+      "step": 13090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.059029218277722e-05,
+      "loss": 0.9605,
+      "step": 13100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0575475611924374e-05,
+      "loss": 0.8607,
+      "step": 13110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.056065904107154e-05,
+      "loss": 1.0471,
+      "step": 13120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0545842470218695e-05,
+      "loss": 0.8444,
+      "step": 13130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.053102589936585e-05,
+      "loss": 0.9976,
+      "step": 13140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0516209328513013e-05,
+      "loss": 0.9886,
+      "step": 13150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.050139275766017e-05,
+      "loss": 0.8636,
+      "step": 13160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0486576186807324e-05,
+      "loss": 0.8408,
+      "step": 13170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.047175961595449e-05,
+      "loss": 0.8352,
+      "step": 13180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0456943045101642e-05,
+      "loss": 1.0622,
+      "step": 13190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.04421264742488e-05,
+      "loss": 0.9629,
+      "step": 13200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0427309903395963e-05,
+      "loss": 0.9512,
+      "step": 13210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0412493332543117e-05,
+      "loss": 0.8721,
+      "step": 13220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0397676761690274e-05,
+      "loss": 0.9748,
+      "step": 13230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0382860190837435e-05,
+      "loss": 1.0791,
+      "step": 13240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0368043619984592e-05,
+      "loss": 0.8108,
+      "step": 13250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.035322704913175e-05,
+      "loss": 0.9094,
+      "step": 13260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0338410478278907e-05,
+      "loss": 1.0127,
+      "step": 13270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0323593907426067e-05,
+      "loss": 0.8709,
+      "step": 13280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0308777336573225e-05,
+      "loss": 0.9389,
+      "step": 13290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0293960765720382e-05,
+      "loss": 0.9801,
+      "step": 13300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0279144194867543e-05,
+      "loss": 1.082,
+      "step": 13310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.02643276240147e-05,
+      "loss": 0.8681,
+      "step": 13320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0249511053161857e-05,
+      "loss": 0.8893,
+      "step": 13330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0234694482309018e-05,
+      "loss": 0.9865,
+      "step": 13340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0219877911456175e-05,
+      "loss": 0.8258,
+      "step": 13350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.020506134060333e-05,
+      "loss": 0.9767,
+      "step": 13360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0190244769750493e-05,
+      "loss": 0.8932,
+      "step": 13370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.017542819889765e-05,
+      "loss": 0.8748,
+      "step": 13380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0160611628044804e-05,
+      "loss": 0.9297,
+      "step": 13390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0145795057191968e-05,
+      "loss": 0.7237,
+      "step": 13400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.013097848633912e-05,
+      "loss": 1.0156,
+      "step": 13410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.011616191548628e-05,
+      "loss": 0.8948,
+      "step": 13420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0101345344633443e-05,
+      "loss": 0.9977,
+      "step": 13430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0086528773780597e-05,
+      "loss": 0.9071,
+      "step": 13440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0071712202927754e-05,
+      "loss": 0.9769,
+      "step": 13450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0056895632074915e-05,
+      "loss": 0.9851,
+      "step": 13460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0042079061222072e-05,
+      "loss": 0.9729,
+      "step": 13470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.002726249036923e-05,
+      "loss": 0.8864,
+      "step": 13480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.001244591951639e-05,
+      "loss": 1.0173,
+      "step": 13490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9997629348663547e-05,
+      "loss": 0.8762,
+      "step": 13500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9982812777810704e-05,
+      "loss": 1.0338,
+      "step": 13510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9967996206957865e-05,
+      "loss": 0.869,
+      "step": 13520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9953179636105022e-05,
+      "loss": 0.978,
+      "step": 13530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.993836306525218e-05,
+      "loss": 0.8195,
+      "step": 13540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9923546494399336e-05,
+      "loss": 1.0383,
+      "step": 13550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9908729923546497e-05,
+      "loss": 0.8594,
+      "step": 13560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9893913352693654e-05,
+      "loss": 0.9188,
+      "step": 13570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9879096781840808e-05,
+      "loss": 0.9374,
+      "step": 13580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9864280210987972e-05,
+      "loss": 0.7712,
+      "step": 13590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.984946364013513e-05,
+      "loss": 1.0269,
+      "step": 13600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9834647069282283e-05,
+      "loss": 0.9598,
+      "step": 13610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9819830498429447e-05,
+      "loss": 0.908,
+      "step": 13620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.98050139275766e-05,
+      "loss": 0.8749,
+      "step": 13630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9790197356723758e-05,
+      "loss": 0.9237,
+      "step": 13640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9775380785870922e-05,
+      "loss": 0.9706,
+      "step": 13650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9760564215018076e-05,
+      "loss": 0.9432,
+      "step": 13660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9745747644165233e-05,
+      "loss": 0.994,
+      "step": 13670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9730931073312397e-05,
+      "loss": 0.981,
+      "step": 13680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.971611450245955e-05,
+      "loss": 0.8332,
+      "step": 13690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.970129793160671e-05,
+      "loss": 0.8669,
+      "step": 13700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968648136075387e-05,
+      "loss": 0.8138,
+      "step": 13710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9671664789901026e-05,
+      "loss": 0.9631,
+      "step": 13720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9656848219048184e-05,
+      "loss": 0.8868,
+      "step": 13730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9642031648195344e-05,
+      "loss": 0.8991,
+      "step": 13740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.96272150773425e-05,
+      "loss": 0.8355,
+      "step": 13750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.961239850648966e-05,
+      "loss": 0.8919,
+      "step": 13760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.959758193563682e-05,
+      "loss": 0.8202,
+      "step": 13770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9582765364783976e-05,
+      "loss": 1.0218,
+      "step": 13780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9567948793931134e-05,
+      "loss": 0.8098,
+      "step": 13790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9553132223078294e-05,
+      "loss": 0.9244,
+      "step": 13800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.953831565222545e-05,
+      "loss": 0.9319,
+      "step": 13810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.952349908137261e-05,
+      "loss": 0.8676,
+      "step": 13820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9508682510519763e-05,
+      "loss": 1.0167,
+      "step": 13830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9493865939666927e-05,
+      "loss": 0.8974,
+      "step": 13840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9479049368814084e-05,
+      "loss": 0.9078,
+      "step": 13850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9464232797961238e-05,
+      "loss": 0.8083,
+      "step": 13860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9449416227108402e-05,
+      "loss": 0.83,
+      "step": 13870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9434599656255556e-05,
+      "loss": 0.9206,
+      "step": 13880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9419783085402713e-05,
+      "loss": 0.945,
+      "step": 13890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9404966514549877e-05,
+      "loss": 0.9216,
+      "step": 13900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.939014994369703e-05,
+      "loss": 0.7943,
+      "step": 13910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9375333372844188e-05,
+      "loss": 0.9959,
+      "step": 13920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.936051680199135e-05,
+      "loss": 0.971,
+      "step": 13930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9345700231138506e-05,
+      "loss": 0.9838,
+      "step": 13940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9330883660285663e-05,
+      "loss": 0.8578,
+      "step": 13950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9316067089432824e-05,
+      "loss": 0.9311,
+      "step": 13960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.930125051857998e-05,
+      "loss": 0.8394,
+      "step": 13970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9286433947727138e-05,
+      "loss": 0.88,
+      "step": 13980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.92716173768743e-05,
+      "loss": 0.8746,
+      "step": 13990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9256800806021456e-05,
+      "loss": 0.8464,
+      "step": 14000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9241984235168613e-05,
+      "loss": 0.8344,
+      "step": 14010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9227167664315774e-05,
+      "loss": 0.8508,
+      "step": 14020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.921235109346293e-05,
+      "loss": 0.9861,
+      "step": 14030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9197534522610088e-05,
+      "loss": 0.9741,
+      "step": 14040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.918271795175725e-05,
+      "loss": 0.8541,
+      "step": 14050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9167901380904406e-05,
+      "loss": 1.1122,
+      "step": 14060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9153084810051563e-05,
+      "loss": 0.8881,
+      "step": 14070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9138268239198724e-05,
+      "loss": 0.8445,
+      "step": 14080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.912345166834588e-05,
+      "loss": 0.7931,
+      "step": 14090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9108635097493035e-05,
+      "loss": 0.8219,
+      "step": 14100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9093818526640192e-05,
+      "loss": 0.9491,
+      "step": 14110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9079001955787356e-05,
+      "loss": 1.0196,
+      "step": 14120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.906418538493451e-05,
+      "loss": 1.0039,
+      "step": 14130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9049368814081667e-05,
+      "loss": 0.9614,
+      "step": 14140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.903455224322883e-05,
+      "loss": 1.0026,
+      "step": 14150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9019735672375985e-05,
+      "loss": 0.9555,
+      "step": 14160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9004919101523142e-05,
+      "loss": 0.9415,
+      "step": 14170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8990102530670303e-05,
+      "loss": 0.6998,
+      "step": 14180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.897528595981746e-05,
+      "loss": 0.9666,
+      "step": 14190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8960469388964618e-05,
+      "loss": 0.884,
+      "step": 14200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8945652818111778e-05,
+      "loss": 0.8459,
+      "step": 14210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8930836247258935e-05,
+      "loss": 0.8909,
+      "step": 14220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8916019676406093e-05,
+      "loss": 0.9377,
+      "step": 14230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8901203105553253e-05,
+      "loss": 0.9647,
+      "step": 14240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.888638653470041e-05,
+      "loss": 0.9309,
+      "step": 14250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8871569963847568e-05,
+      "loss": 1.0228,
+      "step": 14260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8856753392994728e-05,
+      "loss": 0.8874,
+      "step": 14270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8841936822141886e-05,
+      "loss": 0.845,
+      "step": 14280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8827120251289043e-05,
+      "loss": 0.8833,
+      "step": 14290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8812303680436203e-05,
+      "loss": 1.0485,
+      "step": 14300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.879748710958336e-05,
+      "loss": 0.859,
+      "step": 14310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8782670538730518e-05,
+      "loss": 0.8155,
+      "step": 14320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.876785396787768e-05,
+      "loss": 0.9114,
+      "step": 14330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8753037397024836e-05,
+      "loss": 0.8955,
+      "step": 14340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.873822082617199e-05,
+      "loss": 0.8879,
+      "step": 14350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8723404255319154e-05,
+      "loss": 0.9131,
+      "step": 14360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.870858768446631e-05,
+      "loss": 0.925,
+      "step": 14370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8693771113613465e-05,
+      "loss": 0.8615,
+      "step": 14380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8678954542760622e-05,
+      "loss": 0.9005,
+      "step": 14390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8664137971907782e-05,
+      "loss": 1.0335,
+      "step": 14400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.864932140105494e-05,
+      "loss": 0.8189,
+      "step": 14410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8634504830202097e-05,
+      "loss": 0.8096,
+      "step": 14420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8619688259349258e-05,
+      "loss": 0.8743,
+      "step": 14430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8604871688496415e-05,
+      "loss": 0.7303,
+      "step": 14440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8590055117643572e-05,
+      "loss": 0.9973,
+      "step": 14450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8575238546790733e-05,
+      "loss": 0.7832,
+      "step": 14460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.856042197593789e-05,
+      "loss": 0.9041,
+      "step": 14470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8545605405085047e-05,
+      "loss": 0.8215,
+      "step": 14480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8530788834232208e-05,
+      "loss": 0.9953,
+      "step": 14490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8515972263379365e-05,
+      "loss": 0.9047,
+      "step": 14500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8501155692526522e-05,
+      "loss": 0.9655,
+      "step": 14510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8486339121673683e-05,
+      "loss": 0.8112,
+      "step": 14520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.847152255082084e-05,
+      "loss": 0.9085,
+      "step": 14530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8456705979967997e-05,
+      "loss": 0.8962,
+      "step": 14540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8441889409115158e-05,
+      "loss": 0.8624,
+      "step": 14550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8427072838262315e-05,
+      "loss": 0.8384,
+      "step": 14560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.841225626740947e-05,
+      "loss": 0.8974,
+      "step": 14570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8397439696556633e-05,
+      "loss": 0.9387,
+      "step": 14580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.838262312570379e-05,
+      "loss": 0.9768,
+      "step": 14590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8367806554850944e-05,
+      "loss": 0.9538,
+      "step": 14600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8352989983998108e-05,
+      "loss": 0.8618,
+      "step": 14610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8338173413145265e-05,
+      "loss": 0.9872,
+      "step": 14620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.832335684229242e-05,
+      "loss": 0.9197,
+      "step": 14630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8308540271439583e-05,
+      "loss": 0.9187,
+      "step": 14640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8293723700586737e-05,
+      "loss": 0.9396,
+      "step": 14650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8278907129733894e-05,
+      "loss": 0.8117,
+      "step": 14660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.826409055888105e-05,
+      "loss": 0.834,
+      "step": 14670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8249273988028212e-05,
+      "loss": 0.8044,
+      "step": 14680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.823445741717537e-05,
+      "loss": 0.9714,
+      "step": 14690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8219640846322527e-05,
+      "loss": 0.7972,
+      "step": 14700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8204824275469687e-05,
+      "loss": 0.935,
+      "step": 14710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8190007704616844e-05,
+      "loss": 1.164,
+      "step": 14720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8175191133764e-05,
+      "loss": 0.8783,
+      "step": 14730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8160374562911162e-05,
+      "loss": 0.8075,
+      "step": 14740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.814555799205832e-05,
+      "loss": 0.8762,
+      "step": 14750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8130741421205477e-05,
+      "loss": 0.9275,
+      "step": 14760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8115924850352637e-05,
+      "loss": 0.8824,
+      "step": 14770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8101108279499795e-05,
+      "loss": 1.0268,
+      "step": 14780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8086291708646952e-05,
+      "loss": 1.143,
+      "step": 14790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8071475137794112e-05,
+      "loss": 0.8914,
+      "step": 14800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.805665856694127e-05,
+      "loss": 0.8631,
+      "step": 14810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8041841996088424e-05,
+      "loss": 0.8337,
+      "step": 14820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8027025425235588e-05,
+      "loss": 0.9228,
+      "step": 14830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8012208854382745e-05,
+      "loss": 0.8923,
+      "step": 14840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.79973922835299e-05,
+      "loss": 0.8587,
+      "step": 14850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7982575712677063e-05,
+      "loss": 0.9243,
+      "step": 14860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7967759141824216e-05,
+      "loss": 0.9459,
+      "step": 14870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7952942570971374e-05,
+      "loss": 0.9452,
+      "step": 14880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7938126000118538e-05,
+      "loss": 0.9123,
+      "step": 14890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.792330942926569e-05,
+      "loss": 0.8826,
+      "step": 14900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.790849285841285e-05,
+      "loss": 0.8847,
+      "step": 14910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.789367628756001e-05,
+      "loss": 1.0471,
+      "step": 14920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7878859716707167e-05,
+      "loss": 0.8778,
+      "step": 14930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7864043145854324e-05,
+      "loss": 0.9198,
+      "step": 14940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.784922657500148e-05,
+      "loss": 0.8459,
+      "step": 14950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7834410004148642e-05,
+      "loss": 0.9726,
+      "step": 14960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.78195934332958e-05,
+      "loss": 0.8508,
+      "step": 14970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7804776862442956e-05,
+      "loss": 0.9304,
+      "step": 14980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7789960291590117e-05,
+      "loss": 0.8876,
+      "step": 14990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7775143720737274e-05,
+      "loss": 0.9692,
+      "step": 15000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.776032714988443e-05,
+      "loss": 0.8446,
+      "step": 15010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7745510579031592e-05,
+      "loss": 0.8281,
+      "step": 15020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.773069400817875e-05,
+      "loss": 0.9253,
+      "step": 15030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7715877437325903e-05,
+      "loss": 0.9492,
+      "step": 15040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7701060866473067e-05,
+      "loss": 0.8651,
+      "step": 15050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7686244295620224e-05,
+      "loss": 0.863,
+      "step": 15060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7671427724767378e-05,
+      "loss": 0.8056,
+      "step": 15070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7656611153914542e-05,
+      "loss": 0.8961,
+      "step": 15080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7641794583061696e-05,
+      "loss": 0.9401,
+      "step": 15090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7626978012208853e-05,
+      "loss": 0.8435,
+      "step": 15100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7612161441356017e-05,
+      "loss": 0.9163,
+      "step": 15110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.759734487050317e-05,
+      "loss": 0.9933,
+      "step": 15120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7582528299650328e-05,
+      "loss": 0.7188,
+      "step": 15130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7567711728797492e-05,
+      "loss": 0.8124,
+      "step": 15140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7552895157944646e-05,
+      "loss": 0.9366,
+      "step": 15150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7538078587091803e-05,
+      "loss": 0.9292,
+      "step": 15160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7523262016238964e-05,
+      "loss": 0.8255,
+      "step": 15170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.750844544538612e-05,
+      "loss": 0.8795,
+      "step": 15180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.749362887453328e-05,
+      "loss": 0.8525,
+      "step": 15190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.747881230368044e-05,
+      "loss": 0.8968,
+      "step": 15200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7463995732827596e-05,
+      "loss": 0.8813,
+      "step": 15210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7449179161974753e-05,
+      "loss": 0.9808,
+      "step": 15220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.743436259112191e-05,
+      "loss": 0.9589,
+      "step": 15230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.741954602026907e-05,
+      "loss": 0.8174,
+      "step": 15240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.740472944941623e-05,
+      "loss": 0.9346,
+      "step": 15250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7389912878563382e-05,
+      "loss": 1.0015,
+      "step": 15260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7375096307710546e-05,
+      "loss": 0.8213,
+      "step": 15270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7360279736857704e-05,
+      "loss": 0.9302,
+      "step": 15280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7345463166004857e-05,
+      "loss": 0.8862,
+      "step": 15290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.733064659515202e-05,
+      "loss": 0.9485,
+      "step": 15300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.731583002429918e-05,
+      "loss": 0.8125,
+      "step": 15310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7301013453446333e-05,
+      "loss": 0.9412,
+      "step": 15320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7286196882593497e-05,
+      "loss": 0.9581,
+      "step": 15330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.727138031174065e-05,
+      "loss": 0.8735,
+      "step": 15340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7256563740887808e-05,
+      "loss": 0.9793,
+      "step": 15350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.724174717003497e-05,
+      "loss": 1.0512,
+      "step": 15360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7226930599182126e-05,
+      "loss": 0.907,
+      "step": 15370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7212114028329283e-05,
+      "loss": 0.9575,
+      "step": 15380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7197297457476443e-05,
+      "loss": 0.8579,
+      "step": 15390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.71824808866236e-05,
+      "loss": 0.9854,
+      "step": 15400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7167664315770758e-05,
+      "loss": 0.7837,
+      "step": 15410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.715284774491792e-05,
+      "loss": 0.858,
+      "step": 15420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7138031174065076e-05,
+      "loss": 0.891,
+      "step": 15430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7123214603212233e-05,
+      "loss": 0.8212,
+      "step": 15440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7108398032359394e-05,
+      "loss": 0.8824,
+      "step": 15450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.709358146150655e-05,
+      "loss": 0.8595,
+      "step": 15460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7078764890653708e-05,
+      "loss": 0.8807,
+      "step": 15470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.706394831980087e-05,
+      "loss": 0.9255,
+      "step": 15480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7049131748948026e-05,
+      "loss": 0.8835,
+      "step": 15490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7034315178095183e-05,
+      "loss": 0.8357,
+      "step": 15500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7019498607242337e-05,
+      "loss": 0.8606,
+      "step": 15510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.70046820363895e-05,
+      "loss": 0.8943,
+      "step": 15520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6989865465536658e-05,
+      "loss": 0.9866,
+      "step": 15530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6975048894683812e-05,
+      "loss": 0.9325,
+      "step": 15540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6960232323830976e-05,
+      "loss": 0.851,
+      "step": 15550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.694541575297813e-05,
+      "loss": 0.9898,
+      "step": 15560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6930599182125287e-05,
+      "loss": 0.9487,
+      "step": 15570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.691578261127245e-05,
+      "loss": 0.8888,
+      "step": 15580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6900966040419605e-05,
+      "loss": 0.9397,
+      "step": 15590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6886149469566762e-05,
+      "loss": 0.8789,
+      "step": 15600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6871332898713926e-05,
+      "loss": 0.8987,
+      "step": 15610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.685651632786108e-05,
+      "loss": 1.0085,
+      "step": 15620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6841699757008237e-05,
+      "loss": 0.9579,
+      "step": 15630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6826883186155398e-05,
+      "loss": 0.8685,
+      "step": 15640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6812066615302555e-05,
+      "loss": 0.9063,
+      "step": 15650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6797250044449712e-05,
+      "loss": 1.0567,
+      "step": 15660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6782433473596873e-05,
+      "loss": 0.8241,
+      "step": 15670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.676761690274403e-05,
+      "loss": 0.8085,
+      "step": 15680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6752800331891187e-05,
+      "loss": 0.7809,
+      "step": 15690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6737983761038348e-05,
+      "loss": 0.8839,
+      "step": 15700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6723167190185505e-05,
+      "loss": 0.8962,
+      "step": 15710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6708350619332663e-05,
+      "loss": 0.9903,
+      "step": 15720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6693534048479823e-05,
+      "loss": 0.769,
+      "step": 15730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.667871747762698e-05,
+      "loss": 0.9137,
+      "step": 15740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6663900906774138e-05,
+      "loss": 0.8526,
+      "step": 15750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6649084335921298e-05,
+      "loss": 0.9874,
+      "step": 15760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6634267765068455e-05,
+      "loss": 0.9854,
+      "step": 15770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6619451194215613e-05,
+      "loss": 1.0488,
+      "step": 15780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6604634623362767e-05,
+      "loss": 0.8769,
+      "step": 15790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.658981805250993e-05,
+      "loss": 0.9002,
+      "step": 15800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6575001481657084e-05,
+      "loss": 0.7225,
+      "step": 15810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.656018491080424e-05,
+      "loss": 0.9178,
+      "step": 15820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6545368339951406e-05,
+      "loss": 0.8703,
+      "step": 15830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.653055176909856e-05,
+      "loss": 0.9041,
+      "step": 15840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6515735198245717e-05,
+      "loss": 0.9494,
+      "step": 15850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6500918627392877e-05,
+      "loss": 0.8609,
+      "step": 15860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6486102056540035e-05,
+      "loss": 1.0311,
+      "step": 15870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6471285485687192e-05,
+      "loss": 0.8774,
+      "step": 15880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6456468914834352e-05,
+      "loss": 0.8898,
+      "step": 15890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.644165234398151e-05,
+      "loss": 0.9353,
+      "step": 15900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6426835773128667e-05,
+      "loss": 0.8315,
+      "step": 15910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6412019202275828e-05,
+      "loss": 1.0566,
+      "step": 15920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6397202631422985e-05,
+      "loss": 0.9669,
+      "step": 15930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6382386060570142e-05,
+      "loss": 0.9258,
+      "step": 15940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6367569489717303e-05,
+      "loss": 0.7841,
+      "step": 15950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.635275291886446e-05,
+      "loss": 1.0183,
+      "step": 15960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6337936348011617e-05,
+      "loss": 0.9442,
+      "step": 15970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6323119777158778e-05,
+      "loss": 0.9813,
+      "step": 15980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6308303206305935e-05,
+      "loss": 0.9232,
+      "step": 15990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6293486635453092e-05,
+      "loss": 0.9389,
+      "step": 16000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6278670064600253e-05,
+      "loss": 0.8465,
+      "step": 16010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.626385349374741e-05,
+      "loss": 0.8455,
+      "step": 16020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6249036922894564e-05,
+      "loss": 0.9775,
+      "step": 16030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6234220352041728e-05,
+      "loss": 0.9338,
+      "step": 16040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6219403781188885e-05,
+      "loss": 0.7642,
+      "step": 16050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.620458721033604e-05,
+      "loss": 0.9005,
+      "step": 16060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6189770639483196e-05,
+      "loss": 0.8258,
+      "step": 16070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.617495406863036e-05,
+      "loss": 0.9772,
+      "step": 16080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6160137497777514e-05,
+      "loss": 0.9044,
+      "step": 16090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.614532092692467e-05,
+      "loss": 0.9335,
+      "step": 16100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6130504356071832e-05,
+      "loss": 0.8672,
+      "step": 16110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.611568778521899e-05,
+      "loss": 1.017,
+      "step": 16120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6100871214366146e-05,
+      "loss": 0.9566,
+      "step": 16130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6086054643513307e-05,
+      "loss": 0.7851,
+      "step": 16140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6071238072660464e-05,
+      "loss": 0.8653,
+      "step": 16150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.605642150180762e-05,
+      "loss": 0.9346,
+      "step": 16160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6041604930954782e-05,
+      "loss": 0.8218,
+      "step": 16170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.602678836010194e-05,
+      "loss": 0.8127,
+      "step": 16180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6011971789249096e-05,
+      "loss": 1.0239,
+      "step": 16190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5997155218396257e-05,
+      "loss": 0.9663,
+      "step": 16200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5982338647543414e-05,
+      "loss": 0.9385,
+      "step": 16210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.596752207669057e-05,
+      "loss": 0.8683,
+      "step": 16220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5952705505837732e-05,
+      "loss": 0.8967,
+      "step": 16230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.593788893498489e-05,
+      "loss": 0.6821,
+      "step": 16240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5923072364132047e-05,
+      "loss": 0.8757,
+      "step": 16250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5908255793279207e-05,
+      "loss": 0.8384,
+      "step": 16260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5893439222426365e-05,
+      "loss": 0.9881,
+      "step": 16270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.587862265157352e-05,
+      "loss": 1.0065,
+      "step": 16280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5863806080720682e-05,
+      "loss": 0.8451,
+      "step": 16290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.584898950986784e-05,
+      "loss": 0.8693,
+      "step": 16300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5834172939014993e-05,
+      "loss": 0.8972,
+      "step": 16310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5819356368162157e-05,
+      "loss": 0.9758,
+      "step": 16320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.580453979730931e-05,
+      "loss": 0.8606,
+      "step": 16330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.578972322645647e-05,
+      "loss": 0.9072,
+      "step": 16340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5774906655603626e-05,
+      "loss": 0.9198,
+      "step": 16350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5760090084750786e-05,
+      "loss": 0.847,
+      "step": 16360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5745273513897944e-05,
+      "loss": 0.9478,
+      "step": 16370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.57304569430451e-05,
+      "loss": 0.7333,
+      "step": 16380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.571564037219226e-05,
+      "loss": 1.0021,
+      "step": 16390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.570082380133942e-05,
+      "loss": 0.9394,
+      "step": 16400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5686007230486576e-05,
+      "loss": 0.8365,
+      "step": 16410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5671190659633737e-05,
+      "loss": 0.781,
+      "step": 16420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5656374088780894e-05,
+      "loss": 0.9276,
+      "step": 16430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.564155751792805e-05,
+      "loss": 0.9678,
+      "step": 16440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.562674094707521e-05,
+      "loss": 0.93,
+      "step": 16450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.561192437622237e-05,
+      "loss": 0.9031,
+      "step": 16460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5597107805369526e-05,
+      "loss": 1.0367,
+      "step": 16470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5582291234516687e-05,
+      "loss": 0.8493,
+      "step": 16480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5567474663663844e-05,
+      "loss": 0.7942,
+      "step": 16490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5552658092810998e-05,
+      "loss": 0.9409,
+      "step": 16500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5537841521958162e-05,
+      "loss": 0.7449,
+      "step": 16510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.552302495110532e-05,
+      "loss": 0.9053,
+      "step": 16520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5508208380252473e-05,
+      "loss": 0.8583,
+      "step": 16530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5493391809399637e-05,
+      "loss": 0.8682,
+      "step": 16540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.547857523854679e-05,
+      "loss": 0.9189,
+      "step": 16550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5463758667693948e-05,
+      "loss": 0.7976,
+      "step": 16560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5448942096841112e-05,
+      "loss": 0.8626,
+      "step": 16570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5434125525988266e-05,
+      "loss": 0.7991,
+      "step": 16580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5419308955135423e-05,
+      "loss": 0.9152,
+      "step": 16590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5404492384282587e-05,
+      "loss": 1.0176,
+      "step": 16600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.538967581342974e-05,
+      "loss": 0.8972,
+      "step": 16610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5374859242576898e-05,
+      "loss": 0.8773,
+      "step": 16620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5360042671724055e-05,
+      "loss": 0.9817,
+      "step": 16630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5345226100871216e-05,
+      "loss": 0.8889,
+      "step": 16640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5330409530018373e-05,
+      "loss": 0.8452,
+      "step": 16650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.531559295916553e-05,
+      "loss": 0.8898,
+      "step": 16660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.530077638831269e-05,
+      "loss": 0.8766,
+      "step": 16670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.528595981745985e-05,
+      "loss": 0.9172,
+      "step": 16680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5271143246607006e-05,
+      "loss": 0.8526,
+      "step": 16690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5256326675754166e-05,
+      "loss": 0.9177,
+      "step": 16700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5241510104901323e-05,
+      "loss": 0.8059,
+      "step": 16710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5226693534048477e-05,
+      "loss": 0.8278,
+      "step": 16720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.521187696319564e-05,
+      "loss": 0.9526,
+      "step": 16730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.51970603923428e-05,
+      "loss": 0.9538,
+      "step": 16740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5182243821489952e-05,
+      "loss": 0.9277,
+      "step": 16750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5167427250637116e-05,
+      "loss": 0.865,
+      "step": 16760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5152610679784274e-05,
+      "loss": 0.9414,
+      "step": 16770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5137794108931427e-05,
+      "loss": 0.7853,
+      "step": 16780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.512297753807859e-05,
+      "loss": 0.9213,
+      "step": 16790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5108160967225745e-05,
+      "loss": 0.9364,
+      "step": 16800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5093344396372903e-05,
+      "loss": 0.8423,
+      "step": 16810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5078527825520067e-05,
+      "loss": 0.9448,
+      "step": 16820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.506371125466722e-05,
+      "loss": 0.9168,
+      "step": 16830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5048894683814378e-05,
+      "loss": 0.9517,
+      "step": 16840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5034078112961538e-05,
+      "loss": 0.8643,
+      "step": 16850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5019261542108695e-05,
+      "loss": 0.916,
+      "step": 16860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5004444971255853e-05,
+      "loss": 1.0083,
+      "step": 16870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4989628400403013e-05,
+      "loss": 0.892,
+      "step": 16880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.497481182955017e-05,
+      "loss": 1.0555,
+      "step": 16890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4959995258697328e-05,
+      "loss": 0.8717,
+      "step": 16900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4945178687844485e-05,
+      "loss": 0.8993,
+      "step": 16910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4930362116991646e-05,
+      "loss": 0.8272,
+      "step": 16920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4915545546138803e-05,
+      "loss": 0.8452,
+      "step": 16930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.490072897528596e-05,
+      "loss": 0.9096,
+      "step": 16940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.488591240443312e-05,
+      "loss": 0.754,
+      "step": 16950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4871095833580278e-05,
+      "loss": 0.9685,
+      "step": 16960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4856279262727435e-05,
+      "loss": 0.8254,
+      "step": 16970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4841462691874592e-05,
+      "loss": 1.0172,
+      "step": 16980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4826646121021753e-05,
+      "loss": 0.8977,
+      "step": 16990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.481182955016891e-05,
+      "loss": 0.9804,
+      "step": 17000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4797012979316067e-05,
+      "loss": 0.9287,
+      "step": 17010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4782196408463225e-05,
+      "loss": 0.7927,
+      "step": 17020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4767379837610385e-05,
+      "loss": 0.9531,
+      "step": 17030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4752563266757543e-05,
+      "loss": 0.8374,
+      "step": 17040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.47377466959047e-05,
+      "loss": 0.8209,
+      "step": 17050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.472293012505186e-05,
+      "loss": 0.8736,
+      "step": 17060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4708113554199018e-05,
+      "loss": 0.8482,
+      "step": 17070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4693296983346175e-05,
+      "loss": 1.0184,
+      "step": 17080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4678480412493336e-05,
+      "loss": 0.9015,
+      "step": 17090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4663663841640493e-05,
+      "loss": 0.891,
+      "step": 17100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.464884727078765e-05,
+      "loss": 0.84,
+      "step": 17110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4634030699934807e-05,
+      "loss": 0.9163,
+      "step": 17120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4619214129081964e-05,
+      "loss": 0.8362,
+      "step": 17130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4604397558229125e-05,
+      "loss": 0.7383,
+      "step": 17140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4589580987376282e-05,
+      "loss": 0.8876,
+      "step": 17150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.457476441652344e-05,
+      "loss": 0.9511,
+      "step": 17160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.45599478456706e-05,
+      "loss": 0.9941,
+      "step": 17170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4545131274817757e-05,
+      "loss": 0.7721,
+      "step": 17180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4530314703964915e-05,
+      "loss": 0.7751,
+      "step": 17190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4515498133112075e-05,
+      "loss": 0.7806,
+      "step": 17200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4500681562259232e-05,
+      "loss": 0.926,
+      "step": 17210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.448586499140639e-05,
+      "loss": 0.9392,
+      "step": 17220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.447104842055355e-05,
+      "loss": 0.8768,
+      "step": 17230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4456231849700708e-05,
+      "loss": 0.7338,
+      "step": 17240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4441415278847865e-05,
+      "loss": 0.9202,
+      "step": 17250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4426598707995022e-05,
+      "loss": 0.9434,
+      "step": 17260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.441178213714218e-05,
+      "loss": 0.7249,
+      "step": 17270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.439696556628934e-05,
+      "loss": 0.8375,
+      "step": 17280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4382148995436497e-05,
+      "loss": 0.7855,
+      "step": 17290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4367332424583654e-05,
+      "loss": 0.9012,
+      "step": 17300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4352515853730815e-05,
+      "loss": 0.9847,
+      "step": 17310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4337699282877972e-05,
+      "loss": 0.8869,
+      "step": 17320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.432288271202513e-05,
+      "loss": 0.8767,
+      "step": 17330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.430806614117229e-05,
+      "loss": 0.815,
+      "step": 17340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4293249570319447e-05,
+      "loss": 0.8581,
+      "step": 17350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4278432999466604e-05,
+      "loss": 0.8126,
+      "step": 17360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4263616428613765e-05,
+      "loss": 0.8067,
+      "step": 17370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.424879985776092e-05,
+      "loss": 0.8889,
+      "step": 17380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.423398328690808e-05,
+      "loss": 0.8773,
+      "step": 17390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4219166716055237e-05,
+      "loss": 0.8586,
+      "step": 17400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4204350145202394e-05,
+      "loss": 0.8651,
+      "step": 17410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4189533574349555e-05,
+      "loss": 0.9208,
+      "step": 17420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4174717003496712e-05,
+      "loss": 1.0027,
+      "step": 17430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.415990043264387e-05,
+      "loss": 0.8523,
+      "step": 17440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.414508386179103e-05,
+      "loss": 0.9177,
+      "step": 17450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4130267290938187e-05,
+      "loss": 0.9654,
+      "step": 17460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4115450720085344e-05,
+      "loss": 0.9022,
+      "step": 17470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4100634149232505e-05,
+      "loss": 0.999,
+      "step": 17480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.408581757837966e-05,
+      "loss": 0.7946,
+      "step": 17490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.407100100752682e-05,
+      "loss": 0.806,
+      "step": 17500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.405618443667398e-05,
+      "loss": 0.9306,
+      "step": 17510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4041367865821134e-05,
+      "loss": 0.8505,
+      "step": 17520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4026551294968294e-05,
+      "loss": 0.7275,
+      "step": 17530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.401173472411545e-05,
+      "loss": 0.85,
+      "step": 17540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.399691815326261e-05,
+      "loss": 0.9039,
+      "step": 17550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.398210158240977e-05,
+      "loss": 0.908,
+      "step": 17560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3967285011556927e-05,
+      "loss": 0.774,
+      "step": 17570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3952468440704084e-05,
+      "loss": 0.9109,
+      "step": 17580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3937651869851245e-05,
+      "loss": 0.9756,
+      "step": 17590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.39228352989984e-05,
+      "loss": 0.8371,
+      "step": 17600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.390801872814556e-05,
+      "loss": 0.8419,
+      "step": 17610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.389320215729272e-05,
+      "loss": 0.9165,
+      "step": 17620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3878385586439873e-05,
+      "loss": 0.9916,
+      "step": 17630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3863569015587034e-05,
+      "loss": 1.0279,
+      "step": 17640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3848752444734195e-05,
+      "loss": 0.8075,
+      "step": 17650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.383393587388135e-05,
+      "loss": 0.7937,
+      "step": 17660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.381911930302851e-05,
+      "loss": 0.9099,
+      "step": 17670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3804302732175666e-05,
+      "loss": 0.8146,
+      "step": 17680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3789486161322824e-05,
+      "loss": 0.8404,
+      "step": 17690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3774669590469984e-05,
+      "loss": 0.8843,
+      "step": 17700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.375985301961714e-05,
+      "loss": 0.9091,
+      "step": 17710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.37450364487643e-05,
+      "loss": 0.7976,
+      "step": 17720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.373021987791146e-05,
+      "loss": 0.8443,
+      "step": 17730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3715403307058613e-05,
+      "loss": 0.8589,
+      "step": 17740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3700586736205774e-05,
+      "loss": 0.9244,
+      "step": 17750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3685770165352934e-05,
+      "loss": 0.6903,
+      "step": 17760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3670953594500088e-05,
+      "loss": 0.8427,
+      "step": 17770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.365613702364725e-05,
+      "loss": 1.0646,
+      "step": 17780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3641320452794406e-05,
+      "loss": 0.7852,
+      "step": 17790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3626503881941563e-05,
+      "loss": 0.7886,
+      "step": 17800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3611687311088724e-05,
+      "loss": 0.8728,
+      "step": 17810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.359687074023588e-05,
+      "loss": 0.8495,
+      "step": 17820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.358205416938304e-05,
+      "loss": 1.0353,
+      "step": 17830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.35672375985302e-05,
+      "loss": 0.7727,
+      "step": 17840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3552421027677353e-05,
+      "loss": 0.7972,
+      "step": 17850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3537604456824514e-05,
+      "loss": 0.8583,
+      "step": 17860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3522787885971674e-05,
+      "loss": 0.7446,
+      "step": 17870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3507971315118828e-05,
+      "loss": 0.7902,
+      "step": 17880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.349315474426599e-05,
+      "loss": 0.9152,
+      "step": 17890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3478338173413146e-05,
+      "loss": 0.7742,
+      "step": 17900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3463521602560303e-05,
+      "loss": 0.8286,
+      "step": 17910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3448705031707464e-05,
+      "loss": 0.9388,
+      "step": 17920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.343388846085462e-05,
+      "loss": 0.942,
+      "step": 17930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3419071890001778e-05,
+      "loss": 0.8437,
+      "step": 17940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.340425531914894e-05,
+      "loss": 0.8614,
+      "step": 17950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3389438748296093e-05,
+      "loss": 0.8642,
+      "step": 17960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3374622177443253e-05,
+      "loss": 0.7555,
+      "step": 17970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3359805606590414e-05,
+      "loss": 1.0767,
+      "step": 17980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3344989035737568e-05,
+      "loss": 0.8096,
+      "step": 17990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.333017246488473e-05,
+      "loss": 0.8931,
+      "step": 18000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3315355894031886e-05,
+      "loss": 0.8635,
+      "step": 18010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3300539323179043e-05,
+      "loss": 0.9228,
+      "step": 18020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3285722752326203e-05,
+      "loss": 0.9533,
+      "step": 18030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.327090618147336e-05,
+      "loss": 0.7538,
+      "step": 18040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3256089610620518e-05,
+      "loss": 0.8418,
+      "step": 18050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.324127303976768e-05,
+      "loss": 0.7138,
+      "step": 18060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3226456468914836e-05,
+      "loss": 0.8289,
+      "step": 18070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3211639898061993e-05,
+      "loss": 0.8393,
+      "step": 18080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3196823327209154e-05,
+      "loss": 0.8786,
+      "step": 18090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3182006756356307e-05,
+      "loss": 0.8824,
+      "step": 18100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3167190185503468e-05,
+      "loss": 0.8648,
+      "step": 18110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.315237361465063e-05,
+      "loss": 0.9172,
+      "step": 18120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3137557043797783e-05,
+      "loss": 0.7848,
+      "step": 18130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3122740472944943e-05,
+      "loss": 0.7298,
+      "step": 18140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.31079239020921e-05,
+      "loss": 0.867,
+      "step": 18150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3093107331239258e-05,
+      "loss": 0.7374,
+      "step": 18160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3078290760386418e-05,
+      "loss": 0.8786,
+      "step": 18170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3063474189533575e-05,
+      "loss": 1.009,
+      "step": 18180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3048657618680733e-05,
+      "loss": 0.8278,
+      "step": 18190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3033841047827893e-05,
+      "loss": 0.8863,
+      "step": 18200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.301902447697505e-05,
+      "loss": 1.0623,
+      "step": 18210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3004207906122208e-05,
+      "loss": 0.7678,
+      "step": 18220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.298939133526937e-05,
+      "loss": 0.9686,
+      "step": 18230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2974574764416522e-05,
+      "loss": 0.8615,
+      "step": 18240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2959758193563683e-05,
+      "loss": 0.8721,
+      "step": 18250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.294494162271084e-05,
+      "loss": 0.8698,
+      "step": 18260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2930125051857997e-05,
+      "loss": 0.8678,
+      "step": 18270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2915308481005158e-05,
+      "loss": 0.7442,
+      "step": 18280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2900491910152315e-05,
+      "loss": 0.9643,
+      "step": 18290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2885675339299472e-05,
+      "loss": 0.8196,
+      "step": 18300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2870858768446633e-05,
+      "loss": 0.8376,
+      "step": 18310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.285604219759379e-05,
+      "loss": 0.7705,
+      "step": 18320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2841225626740948e-05,
+      "loss": 0.88,
+      "step": 18330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2826409055888108e-05,
+      "loss": 0.7661,
+      "step": 18340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2811592485035265e-05,
+      "loss": 0.8063,
+      "step": 18350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2796775914182423e-05,
+      "loss": 0.8005,
+      "step": 18360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.278195934332958e-05,
+      "loss": 0.7737,
+      "step": 18370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2767142772476737e-05,
+      "loss": 0.8472,
+      "step": 18380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2752326201623898e-05,
+      "loss": 0.782,
+      "step": 18390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2737509630771055e-05,
+      "loss": 0.7141,
+      "step": 18400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2722693059918212e-05,
+      "loss": 0.9833,
+      "step": 18410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2707876489065373e-05,
+      "loss": 0.7897,
+      "step": 18420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.269305991821253e-05,
+      "loss": 0.8591,
+      "step": 18430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2678243347359687e-05,
+      "loss": 0.9508,
+      "step": 18440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2663426776506848e-05,
+      "loss": 0.8874,
+      "step": 18450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2648610205654005e-05,
+      "loss": 0.925,
+      "step": 18460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2633793634801162e-05,
+      "loss": 0.7757,
+      "step": 18470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.261897706394832e-05,
+      "loss": 0.7419,
+      "step": 18480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.260416049309548e-05,
+      "loss": 0.9423,
+      "step": 18490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2589343922242637e-05,
+      "loss": 0.8271,
+      "step": 18500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2574527351389795e-05,
+      "loss": 0.8333,
+      "step": 18510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2559710780536952e-05,
+      "loss": 0.9858,
+      "step": 18520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2544894209684112e-05,
+      "loss": 0.8814,
+      "step": 18530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.253007763883127e-05,
+      "loss": 1.0032,
+      "step": 18540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2515261067978427e-05,
+      "loss": 0.7925,
+      "step": 18550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2500444497125588e-05,
+      "loss": 0.7896,
+      "step": 18560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2485627926272745e-05,
+      "loss": 0.9071,
+      "step": 18570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2470811355419902e-05,
+      "loss": 0.999,
+      "step": 18580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.245599478456706e-05,
+      "loss": 0.8147,
+      "step": 18590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.244117821371422e-05,
+      "loss": 0.8457,
+      "step": 18600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2426361642861377e-05,
+      "loss": 0.9476,
+      "step": 18610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2411545072008534e-05,
+      "loss": 0.7915,
+      "step": 18620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2396728501155695e-05,
+      "loss": 0.8055,
+      "step": 18630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2381911930302852e-05,
+      "loss": 0.9684,
+      "step": 18640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.236709535945001e-05,
+      "loss": 0.8395,
+      "step": 18650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2352278788597167e-05,
+      "loss": 0.8448,
+      "step": 18660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2337462217744327e-05,
+      "loss": 0.8679,
+      "step": 18670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2322645646891485e-05,
+      "loss": 0.7937,
+      "step": 18680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2307829076038642e-05,
+      "loss": 0.9354,
+      "step": 18690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2293012505185802e-05,
+      "loss": 0.9462,
+      "step": 18700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.227819593433296e-05,
+      "loss": 0.7613,
+      "step": 18710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2263379363480117e-05,
+      "loss": 0.7807,
+      "step": 18720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2248562792627274e-05,
+      "loss": 0.8408,
+      "step": 18730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2233746221774435e-05,
+      "loss": 0.9313,
+      "step": 18740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2218929650921592e-05,
+      "loss": 0.8566,
+      "step": 18750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.220411308006875e-05,
+      "loss": 0.7942,
+      "step": 18760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.218929650921591e-05,
+      "loss": 0.9131,
+      "step": 18770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2174479938363067e-05,
+      "loss": 0.8862,
+      "step": 18780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2159663367510224e-05,
+      "loss": 0.8691,
+      "step": 18790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.214484679665738e-05,
+      "loss": 0.9793,
+      "step": 18800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2130030225804542e-05,
+      "loss": 0.9441,
+      "step": 18810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.21152136549517e-05,
+      "loss": 0.842,
+      "step": 18820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2100397084098857e-05,
+      "loss": 0.8368,
+      "step": 18830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2085580513246014e-05,
+      "loss": 0.9774,
+      "step": 18840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2070763942393174e-05,
+      "loss": 0.8356,
+      "step": 18850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.205594737154033e-05,
+      "loss": 0.7599,
+      "step": 18860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.204113080068749e-05,
+      "loss": 0.8256,
+      "step": 18870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.202631422983465e-05,
+      "loss": 0.9039,
+      "step": 18880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2011497658981807e-05,
+      "loss": 0.9825,
+      "step": 18890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1996681088128964e-05,
+      "loss": 0.8668,
+      "step": 18900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1981864517276125e-05,
+      "loss": 0.8308,
+      "step": 18910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1967047946423282e-05,
+      "loss": 0.9404,
+      "step": 18920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.195223137557044e-05,
+      "loss": 0.8151,
+      "step": 18930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1937414804717596e-05,
+      "loss": 0.9006,
+      "step": 18940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1922598233864754e-05,
+      "loss": 0.8403,
+      "step": 18950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1907781663011914e-05,
+      "loss": 0.8765,
+      "step": 18960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.189296509215907e-05,
+      "loss": 0.8613,
+      "step": 18970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.187814852130623e-05,
+      "loss": 0.7064,
+      "step": 18980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.186333195045339e-05,
+      "loss": 0.8803,
+      "step": 18990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1848515379600546e-05,
+      "loss": 0.8661,
+      "step": 19000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1833698808747704e-05,
+      "loss": 0.8495,
+      "step": 19010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1818882237894864e-05,
+      "loss": 0.8133,
+      "step": 19020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.180406566704202e-05,
+      "loss": 0.7198,
+      "step": 19030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.178924909618918e-05,
+      "loss": 0.8265,
+      "step": 19040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.177443252533634e-05,
+      "loss": 0.7314,
+      "step": 19050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1759615954483493e-05,
+      "loss": 0.8324,
+      "step": 19060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1744799383630654e-05,
+      "loss": 0.8576,
+      "step": 19070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.172998281277781e-05,
+      "loss": 0.7959,
+      "step": 19080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.171516624192497e-05,
+      "loss": 0.8669,
+      "step": 19090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.170034967107213e-05,
+      "loss": 0.9387,
+      "step": 19100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1685533100219286e-05,
+      "loss": 0.867,
+      "step": 19110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1670716529366443e-05,
+      "loss": 0.7616,
+      "step": 19120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1655899958513604e-05,
+      "loss": 0.9017,
+      "step": 19130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.164108338766076e-05,
+      "loss": 0.9809,
+      "step": 19140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.162626681680792e-05,
+      "loss": 0.911,
+      "step": 19150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.161145024595508e-05,
+      "loss": 0.9344,
+      "step": 19160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1596633675102236e-05,
+      "loss": 0.8076,
+      "step": 19170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1581817104249394e-05,
+      "loss": 0.8818,
+      "step": 19180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1567000533396554e-05,
+      "loss": 0.9721,
+      "step": 19190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1552183962543708e-05,
+      "loss": 0.8332,
+      "step": 19200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.153736739169087e-05,
+      "loss": 0.898,
+      "step": 19210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1522550820838026e-05,
+      "loss": 0.9453,
+      "step": 19220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1507734249985183e-05,
+      "loss": 0.7826,
+      "step": 19230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1492917679132344e-05,
+      "loss": 0.9795,
+      "step": 19240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.14781011082795e-05,
+      "loss": 0.9173,
+      "step": 19250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1463284537426658e-05,
+      "loss": 0.9476,
+      "step": 19260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.144846796657382e-05,
+      "loss": 0.964,
+      "step": 19270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1433651395720976e-05,
+      "loss": 0.9072,
+      "step": 19280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1418834824868133e-05,
+      "loss": 0.8462,
+      "step": 19290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1404018254015294e-05,
+      "loss": 0.874,
+      "step": 19300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1389201683162448e-05,
+      "loss": 1.0157,
+      "step": 19310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.137438511230961e-05,
+      "loss": 0.8508,
+      "step": 19320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.135956854145677e-05,
+      "loss": 0.9037,
+      "step": 19330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1344751970603923e-05,
+      "loss": 0.9253,
+      "step": 19340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1329935399751083e-05,
+      "loss": 0.7522,
+      "step": 19350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.131511882889824e-05,
+      "loss": 0.9502,
+      "step": 19360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1300302258045398e-05,
+      "loss": 0.7547,
+      "step": 19370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.128548568719256e-05,
+      "loss": 0.9124,
+      "step": 19380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1270669116339716e-05,
+      "loss": 0.8023,
+      "step": 19390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1255852545486873e-05,
+      "loss": 0.8538,
+      "step": 19400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1241035974634034e-05,
+      "loss": 0.7735,
+      "step": 19410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1226219403781187e-05,
+      "loss": 0.8246,
+      "step": 19420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1211402832928348e-05,
+      "loss": 0.755,
+      "step": 19430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.119658626207551e-05,
+      "loss": 0.7665,
+      "step": 19440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1181769691222663e-05,
+      "loss": 1.0283,
+      "step": 19450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1166953120369823e-05,
+      "loss": 0.8871,
+      "step": 19460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.115213654951698e-05,
+      "loss": 0.9079,
+      "step": 19470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1137319978664138e-05,
+      "loss": 0.7618,
+      "step": 19480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1122503407811298e-05,
+      "loss": 0.8599,
+      "step": 19490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1107686836958456e-05,
+      "loss": 0.842,
+      "step": 19500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1092870266105613e-05,
+      "loss": 0.8677,
+      "step": 19510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1078053695252773e-05,
+      "loss": 0.9027,
+      "step": 19520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1063237124399927e-05,
+      "loss": 0.8858,
+      "step": 19530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1048420553547088e-05,
+      "loss": 0.9126,
+      "step": 19540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.103360398269425e-05,
+      "loss": 0.9025,
+      "step": 19550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1018787411841402e-05,
+      "loss": 0.8532,
+      "step": 19560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1003970840988563e-05,
+      "loss": 0.8272,
+      "step": 19570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0989154270135724e-05,
+      "loss": 0.8753,
+      "step": 19580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0974337699282877e-05,
+      "loss": 0.9098,
+      "step": 19590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0959521128430038e-05,
+      "loss": 0.9761,
+      "step": 19600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0944704557577195e-05,
+      "loss": 0.8511,
+      "step": 19610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0929887986724352e-05,
+      "loss": 0.9308,
+      "step": 19620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0915071415871513e-05,
+      "loss": 0.9555,
+      "step": 19630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0900254845018667e-05,
+      "loss": 0.9046,
+      "step": 19640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0885438274165828e-05,
+      "loss": 0.7832,
+      "step": 19650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0870621703312988e-05,
+      "loss": 0.8923,
+      "step": 19660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0855805132460142e-05,
+      "loss": 0.8718,
+      "step": 19670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0840988561607303e-05,
+      "loss": 0.8448,
+      "step": 19680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0826171990754463e-05,
+      "loss": 1.1105,
+      "step": 19690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0811355419901617e-05,
+      "loss": 0.8333,
+      "step": 19700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0796538849048778e-05,
+      "loss": 0.8141,
+      "step": 19710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0781722278195935e-05,
+      "loss": 0.9571,
+      "step": 19720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0766905707343092e-05,
+      "loss": 0.7987,
+      "step": 19730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0752089136490253e-05,
+      "loss": 0.9294,
+      "step": 19740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.073727256563741e-05,
+      "loss": 0.9727,
+      "step": 19750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0722455994784567e-05,
+      "loss": 1.01,
+      "step": 19760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0707639423931728e-05,
+      "loss": 0.8694,
+      "step": 19770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0692822853078882e-05,
+      "loss": 0.8795,
+      "step": 19780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0678006282226042e-05,
+      "loss": 0.8247,
+      "step": 19790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0663189711373203e-05,
+      "loss": 0.9454,
+      "step": 19800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0648373140520357e-05,
+      "loss": 0.9298,
+      "step": 19810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0633556569667517e-05,
+      "loss": 1.0117,
+      "step": 19820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0618739998814675e-05,
+      "loss": 0.7869,
+      "step": 19830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0603923427961832e-05,
+      "loss": 0.9181,
+      "step": 19840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0589106857108993e-05,
+      "loss": 1.0305,
+      "step": 19850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.057429028625615e-05,
+      "loss": 0.814,
+      "step": 19860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0559473715403307e-05,
+      "loss": 0.7786,
+      "step": 19870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0544657144550468e-05,
+      "loss": 0.706,
+      "step": 19880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0529840573697625e-05,
+      "loss": 0.9476,
+      "step": 19890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0515024002844782e-05,
+      "loss": 0.7293,
+      "step": 19900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0500207431991943e-05,
+      "loss": 0.8174,
+      "step": 19910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0485390861139097e-05,
+      "loss": 0.9313,
+      "step": 19920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0470574290286257e-05,
+      "loss": 0.8605,
+      "step": 19930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0455757719433414e-05,
+      "loss": 0.9768,
+      "step": 19940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.044094114858057e-05,
+      "loss": 0.8925,
+      "step": 19950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0426124577727732e-05,
+      "loss": 0.8249,
+      "step": 19960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.041130800687489e-05,
+      "loss": 0.9222,
+      "step": 19970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0396491436022047e-05,
+      "loss": 0.8031,
+      "step": 19980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0381674865169207e-05,
+      "loss": 0.8355,
+      "step": 19990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0366858294316365e-05,
+      "loss": 0.7415,
+      "step": 20000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0352041723463522e-05,
+      "loss": 0.8848,
+      "step": 20010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0337225152610682e-05,
+      "loss": 0.8046,
+      "step": 20020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.032240858175784e-05,
+      "loss": 0.858,
+      "step": 20030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0307592010904997e-05,
+      "loss": 0.8371,
+      "step": 20040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0292775440052154e-05,
+      "loss": 0.9214,
+      "step": 20050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.027795886919931e-05,
+      "loss": 0.8854,
+      "step": 20060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0263142298346472e-05,
+      "loss": 0.8968,
+      "step": 20070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.024832572749363e-05,
+      "loss": 0.8748,
+      "step": 20080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0233509156640786e-05,
+      "loss": 1.0004,
+      "step": 20090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0218692585787947e-05,
+      "loss": 0.8403,
+      "step": 20100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0203876014935104e-05,
+      "loss": 0.8263,
+      "step": 20110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.018905944408226e-05,
+      "loss": 0.8777,
+      "step": 20120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0174242873229422e-05,
+      "loss": 0.9249,
+      "step": 20130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.015942630237658e-05,
+      "loss": 0.7976,
+      "step": 20140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0144609731523737e-05,
+      "loss": 0.8771,
+      "step": 20150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0129793160670897e-05,
+      "loss": 0.786,
+      "step": 20160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0114976589818054e-05,
+      "loss": 0.882,
+      "step": 20170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0100160018965212e-05,
+      "loss": 0.9749,
+      "step": 20180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.008534344811237e-05,
+      "loss": 0.9407,
+      "step": 20190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0070526877259526e-05,
+      "loss": 0.8734,
+      "step": 20200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0055710306406687e-05,
+      "loss": 0.824,
+      "step": 20210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0040893735553844e-05,
+      "loss": 0.8913,
+      "step": 20220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0026077164701e-05,
+      "loss": 0.9268,
+      "step": 20230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0011260593848162e-05,
+      "loss": 0.9078,
+      "step": 20240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999644402299532e-05,
+      "loss": 0.8387,
+      "step": 20250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9981627452142476e-05,
+      "loss": 0.851,
+      "step": 20260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9966810881289637e-05,
+      "loss": 0.8193,
+      "step": 20270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9951994310436794e-05,
+      "loss": 0.8916,
+      "step": 20280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.993717773958395e-05,
+      "loss": 0.8234,
+      "step": 20290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.992236116873111e-05,
+      "loss": 0.903,
+      "step": 20300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.990754459787827e-05,
+      "loss": 1.025,
+      "step": 20310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9892728027025427e-05,
+      "loss": 1.0031,
+      "step": 20320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9877911456172584e-05,
+      "loss": 0.9271,
+      "step": 20330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.986309488531974e-05,
+      "loss": 0.8388,
+      "step": 20340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.98482783144669e-05,
+      "loss": 0.8144,
+      "step": 20350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.983346174361406e-05,
+      "loss": 1.049,
+      "step": 20360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9818645172761216e-05,
+      "loss": 0.8028,
+      "step": 20370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9803828601908377e-05,
+      "loss": 0.9394,
+      "step": 20380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9789012031055534e-05,
+      "loss": 0.9326,
+      "step": 20390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.977419546020269e-05,
+      "loss": 0.9962,
+      "step": 20400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.975937888934985e-05,
+      "loss": 0.8867,
+      "step": 20410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.974456231849701e-05,
+      "loss": 0.7435,
+      "step": 20420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9729745747644166e-05,
+      "loss": 0.7816,
+      "step": 20430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9714929176791323e-05,
+      "loss": 0.9469,
+      "step": 20440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9700112605938484e-05,
+      "loss": 0.8169,
+      "step": 20450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.968529603508564e-05,
+      "loss": 0.8735,
+      "step": 20460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.96704794642328e-05,
+      "loss": 1.0336,
+      "step": 20470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9655662893379956e-05,
+      "loss": 0.8156,
+      "step": 20480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9640846322527116e-05,
+      "loss": 0.9808,
+      "step": 20490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9626029751674274e-05,
+      "loss": 0.79,
+      "step": 20500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.961121318082143e-05,
+      "loss": 0.8405,
+      "step": 20510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9596396609968588e-05,
+      "loss": 0.953,
+      "step": 20520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.958158003911575e-05,
+      "loss": 0.7861,
+      "step": 20530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9566763468262906e-05,
+      "loss": 0.7984,
+      "step": 20540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551946897410063e-05,
+      "loss": 0.9889,
+      "step": 20550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9537130326557224e-05,
+      "loss": 0.841,
+      "step": 20560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.952231375570438e-05,
+      "loss": 0.8406,
+      "step": 20570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9507497184851538e-05,
+      "loss": 0.9461,
+      "step": 20580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.94926806139987e-05,
+      "loss": 0.8538,
+      "step": 20590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9477864043145856e-05,
+      "loss": 0.854,
+      "step": 20600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9463047472293013e-05,
+      "loss": 0.8276,
+      "step": 20610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.944823090144017e-05,
+      "loss": 0.955,
+      "step": 20620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.943341433058733e-05,
+      "loss": 0.8104,
+      "step": 20630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.941859775973449e-05,
+      "loss": 0.9592,
+      "step": 20640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9403781188881646e-05,
+      "loss": 0.8689,
+      "step": 20650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9388964618028803e-05,
+      "loss": 0.9637,
+      "step": 20660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9374148047175964e-05,
+      "loss": 0.7728,
+      "step": 20670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.935933147632312e-05,
+      "loss": 0.7657,
+      "step": 20680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9344514905470278e-05,
+      "loss": 0.8607,
+      "step": 20690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.932969833461744e-05,
+      "loss": 0.8825,
+      "step": 20700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9314881763764596e-05,
+      "loss": 0.924,
+      "step": 20710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9300065192911753e-05,
+      "loss": 0.855,
+      "step": 20720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9285248622058914e-05,
+      "loss": 0.7599,
+      "step": 20730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.927043205120607e-05,
+      "loss": 0.8022,
+      "step": 20740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9255615480353228e-05,
+      "loss": 0.7713,
+      "step": 20750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9240798909500385e-05,
+      "loss": 0.9565,
+      "step": 20760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9225982338647543e-05,
+      "loss": 0.9293,
+      "step": 20770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9211165767794703e-05,
+      "loss": 0.8294,
+      "step": 20780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.919634919694186e-05,
+      "loss": 0.8606,
+      "step": 20790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9181532626089018e-05,
+      "loss": 0.8515,
+      "step": 20800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.916671605523618e-05,
+      "loss": 0.853,
+      "step": 20810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9151899484383336e-05,
+      "loss": 0.7633,
+      "step": 20820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9137082913530493e-05,
+      "loss": 0.8862,
+      "step": 20830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9122266342677653e-05,
+      "loss": 0.9932,
+      "step": 20840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.910744977182481e-05,
+      "loss": 0.8002,
+      "step": 20850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9092633200971968e-05,
+      "loss": 0.9051,
+      "step": 20860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.907781663011913e-05,
+      "loss": 0.9279,
+      "step": 20870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9063000059266282e-05,
+      "loss": 0.8137,
+      "step": 20880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9048183488413443e-05,
+      "loss": 0.8661,
+      "step": 20890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.90333669175606e-05,
+      "loss": 0.9539,
+      "step": 20900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9018550346707757e-05,
+      "loss": 0.8177,
+      "step": 20910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9003733775854918e-05,
+      "loss": 0.7602,
+      "step": 20920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8988917205002075e-05,
+      "loss": 0.7922,
+      "step": 20930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8974100634149233e-05,
+      "loss": 0.8677,
+      "step": 20940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8959284063296393e-05,
+      "loss": 0.7631,
+      "step": 20950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.894446749244355e-05,
+      "loss": 0.8979,
+      "step": 20960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8929650921590708e-05,
+      "loss": 0.8185,
+      "step": 20970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8914834350737868e-05,
+      "loss": 0.892,
+      "step": 20980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8900017779885022e-05,
+      "loss": 0.9548,
+      "step": 20990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8885201209032183e-05,
+      "loss": 0.8945,
+      "step": 21000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8870384638179343e-05,
+      "loss": 0.8548,
+      "step": 21010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8855568067326497e-05,
+      "loss": 0.9585,
+      "step": 21020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8840751496473658e-05,
+      "loss": 0.8668,
+      "step": 21030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8825934925620815e-05,
+      "loss": 0.9046,
+      "step": 21040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8811118354767972e-05,
+      "loss": 0.8567,
+      "step": 21050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8796301783915133e-05,
+      "loss": 0.791,
+      "step": 21060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.878148521306229e-05,
+      "loss": 0.8112,
+      "step": 21070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8766668642209447e-05,
+      "loss": 0.8138,
+      "step": 21080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8751852071356608e-05,
+      "loss": 0.9992,
+      "step": 21090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8737035500503762e-05,
+      "loss": 0.8576,
+      "step": 21100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8722218929650922e-05,
+      "loss": 0.7868,
+      "step": 21110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8707402358798083e-05,
+      "loss": 0.8043,
+      "step": 21120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8692585787945237e-05,
+      "loss": 0.7189,
+      "step": 21130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8677769217092397e-05,
+      "loss": 0.7823,
+      "step": 21140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8662952646239558e-05,
+      "loss": 0.7672,
+      "step": 21150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8648136075386712e-05,
+      "loss": 1.001,
+      "step": 21160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8633319504533873e-05,
+      "loss": 0.7417,
+      "step": 21170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.861850293368103e-05,
+      "loss": 0.8655,
+      "step": 21180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8603686362828187e-05,
+      "loss": 0.7832,
+      "step": 21190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8588869791975348e-05,
+      "loss": 0.924,
+      "step": 21200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8574053221122505e-05,
+      "loss": 0.9696,
+      "step": 21210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8559236650269662e-05,
+      "loss": 0.883,
+      "step": 21220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8544420079416823e-05,
+      "loss": 0.9714,
+      "step": 21230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8529603508563977e-05,
+      "loss": 0.9046,
+      "step": 21240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8514786937711137e-05,
+      "loss": 0.9156,
+      "step": 21250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8499970366858298e-05,
+      "loss": 0.9072,
+      "step": 21260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.848515379600545e-05,
+      "loss": 0.9005,
+      "step": 21270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8470337225152612e-05,
+      "loss": 0.8195,
+      "step": 21280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.845552065429977e-05,
+      "loss": 0.8322,
+      "step": 21290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8440704083446927e-05,
+      "loss": 0.9191,
+      "step": 21300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8425887512594087e-05,
+      "loss": 0.8699,
+      "step": 21310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8411070941741245e-05,
+      "loss": 0.9575,
+      "step": 21320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8396254370888402e-05,
+      "loss": 0.8498,
+      "step": 21330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8381437800035562e-05,
+      "loss": 0.8958,
+      "step": 21340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366621229182716e-05,
+      "loss": 0.864,
+      "step": 21350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8351804658329877e-05,
+      "loss": 0.9615,
+      "step": 21360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8336988087477038e-05,
+      "loss": 0.788,
+      "step": 21370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832217151662419e-05,
+      "loss": 0.8295,
+      "step": 21380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8307354945771352e-05,
+      "loss": 0.8751,
+      "step": 21390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.829253837491851e-05,
+      "loss": 0.8718,
+      "step": 21400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8277721804065666e-05,
+      "loss": 0.8962,
+      "step": 21410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8262905233212827e-05,
+      "loss": 0.8029,
+      "step": 21420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8248088662359984e-05,
+      "loss": 0.9122,
+      "step": 21430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.823327209150714e-05,
+      "loss": 0.7332,
+      "step": 21440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8218455520654302e-05,
+      "loss": 0.8847,
+      "step": 21450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8203638949801456e-05,
+      "loss": 0.8831,
+      "step": 21460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8188822378948617e-05,
+      "loss": 0.9425,
+      "step": 21470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8174005808095777e-05,
+      "loss": 0.7179,
+      "step": 21480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.815918923724293e-05,
+      "loss": 0.9403,
+      "step": 21490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8144372666390092e-05,
+      "loss": 0.8041,
+      "step": 21500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.812955609553725e-05,
+      "loss": 0.9635,
+      "step": 21510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8114739524684406e-05,
+      "loss": 0.8108,
+      "step": 21520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8099922953831567e-05,
+      "loss": 0.9086,
+      "step": 21530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8085106382978724e-05,
+      "loss": 0.7554,
+      "step": 21540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.807028981212588e-05,
+      "loss": 0.999,
+      "step": 21550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8055473241273042e-05,
+      "loss": 0.9111,
+      "step": 21560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.80406566704202e-05,
+      "loss": 1.0083,
+      "step": 21570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8025840099567356e-05,
+      "loss": 0.8531,
+      "step": 21580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8011023528714517e-05,
+      "loss": 0.8125,
+      "step": 21590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.799620695786167e-05,
+      "loss": 0.8432,
+      "step": 21600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.798139038700883e-05,
+      "loss": 0.9143,
+      "step": 21610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7966573816155992e-05,
+      "loss": 0.8225,
+      "step": 21620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7951757245303146e-05,
+      "loss": 0.8217,
+      "step": 21630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7936940674450307e-05,
+      "loss": 0.8318,
+      "step": 21640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7922124103597464e-05,
+      "loss": 0.8802,
+      "step": 21650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.790730753274462e-05,
+      "loss": 0.6746,
+      "step": 21660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.789249096189178e-05,
+      "loss": 0.8367,
+      "step": 21670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.787767439103894e-05,
+      "loss": 0.8144,
+      "step": 21680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7862857820186096e-05,
+      "loss": 0.8638,
+      "step": 21690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7848041249333257e-05,
+      "loss": 0.9311,
+      "step": 21700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7833224678480414e-05,
+      "loss": 0.8604,
+      "step": 21710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.781840810762757e-05,
+      "loss": 0.7928,
+      "step": 21720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7803591536774732e-05,
+      "loss": 0.7322,
+      "step": 21730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7788774965921886e-05,
+      "loss": 0.824,
+      "step": 21740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7773958395069046e-05,
+      "loss": 0.965,
+      "step": 21750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7759141824216203e-05,
+      "loss": 0.8722,
+      "step": 21760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.774432525336336e-05,
+      "loss": 0.8038,
+      "step": 21770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.772950868251052e-05,
+      "loss": 0.8402,
+      "step": 21780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.771469211165768e-05,
+      "loss": 0.9165,
+      "step": 21790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7699875540804836e-05,
+      "loss": 0.8432,
+      "step": 21800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7685058969951996e-05,
+      "loss": 0.8241,
+      "step": 21810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7670242399099154e-05,
+      "loss": 0.8838,
+      "step": 21820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.765542582824631e-05,
+      "loss": 0.8166,
+      "step": 21830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.764060925739347e-05,
+      "loss": 0.8159,
+      "step": 21840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.762579268654063e-05,
+      "loss": 0.8546,
+      "step": 21850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7610976115687786e-05,
+      "loss": 1.0096,
+      "step": 21860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7596159544834943e-05,
+      "loss": 0.821,
+      "step": 21870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.75813429739821e-05,
+      "loss": 0.9978,
+      "step": 21880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.756652640312926e-05,
+      "loss": 0.9032,
+      "step": 21890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7551709832276418e-05,
+      "loss": 0.6687,
+      "step": 21900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7536893261423576e-05,
+      "loss": 0.9232,
+      "step": 21910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7522076690570736e-05,
+      "loss": 0.8338,
+      "step": 21920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7507260119717893e-05,
+      "loss": 0.7658,
+      "step": 21930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.749244354886505e-05,
+      "loss": 0.7666,
+      "step": 21940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.747762697801221e-05,
+      "loss": 0.783,
+      "step": 21950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.746281040715937e-05,
+      "loss": 0.7576,
+      "step": 21960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7447993836306526e-05,
+      "loss": 0.8817,
+      "step": 21970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7433177265453683e-05,
+      "loss": 0.8559,
+      "step": 21980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7418360694600844e-05,
+      "loss": 0.9449,
+      "step": 21990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7403544123748e-05,
+      "loss": 0.9124,
+      "step": 22000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7388727552895158e-05,
+      "loss": 1.0082,
+      "step": 22010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7373910982042315e-05,
+      "loss": 0.7337,
+      "step": 22020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7359094411189476e-05,
+      "loss": 0.8386,
+      "step": 22030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7344277840336633e-05,
+      "loss": 0.825,
+      "step": 22040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.732946126948379e-05,
+      "loss": 0.7686,
+      "step": 22050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.731464469863095e-05,
+      "loss": 0.9501,
+      "step": 22060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7299828127778108e-05,
+      "loss": 0.8041,
+      "step": 22070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7285011556925265e-05,
+      "loss": 0.8971,
+      "step": 22080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7270194986072426e-05,
+      "loss": 0.8153,
+      "step": 22090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7255378415219583e-05,
+      "loss": 0.8948,
+      "step": 22100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.724056184436674e-05,
+      "loss": 0.9888,
+      "step": 22110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7225745273513898e-05,
+      "loss": 0.7854,
+      "step": 22120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.721092870266106e-05,
+      "loss": 0.8631,
+      "step": 22130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7196112131808216e-05,
+      "loss": 0.9226,
+      "step": 22140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7181295560955373e-05,
+      "loss": 0.876,
+      "step": 22150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.716647899010253e-05,
+      "loss": 0.8988,
+      "step": 22160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.715166241924969e-05,
+      "loss": 0.8474,
+      "step": 22170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7136845848396848e-05,
+      "loss": 0.9297,
+      "step": 22180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7122029277544005e-05,
+      "loss": 0.8784,
+      "step": 22190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7107212706691166e-05,
+      "loss": 0.7333,
+      "step": 22200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7092396135838323e-05,
+      "loss": 0.889,
+      "step": 22210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.707757956498548e-05,
+      "loss": 0.8144,
+      "step": 22220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7062762994132637e-05,
+      "loss": 0.9886,
+      "step": 22230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7047946423279798e-05,
+      "loss": 0.9524,
+      "step": 22240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7033129852426955e-05,
+      "loss": 0.9731,
+      "step": 22250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7018313281574113e-05,
+      "loss": 0.7887,
+      "step": 22260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7003496710721273e-05,
+      "loss": 0.81,
+      "step": 22270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.698868013986843e-05,
+      "loss": 0.9525,
+      "step": 22280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6973863569015588e-05,
+      "loss": 0.772,
+      "step": 22290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6959046998162745e-05,
+      "loss": 0.7318,
+      "step": 22300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6944230427309905e-05,
+      "loss": 0.9601,
+      "step": 22310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6929413856457063e-05,
+      "loss": 0.8658,
+      "step": 22320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.691459728560422e-05,
+      "loss": 0.9724,
+      "step": 22330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6899780714751377e-05,
+      "loss": 0.9571,
+      "step": 22340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6884964143898538e-05,
+      "loss": 0.8063,
+      "step": 22350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6870147573045695e-05,
+      "loss": 0.82,
+      "step": 22360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6855331002192852e-05,
+      "loss": 0.921,
+      "step": 22370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6840514431340013e-05,
+      "loss": 0.8703,
+      "step": 22380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.682569786048717e-05,
+      "loss": 1.0329,
+      "step": 22390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6810881289634327e-05,
+      "loss": 0.9818,
+      "step": 22400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6796064718781488e-05,
+      "loss": 0.9763,
+      "step": 22410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6781248147928645e-05,
+      "loss": 0.9212,
+      "step": 22420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6766431577075802e-05,
+      "loss": 0.7776,
+      "step": 22430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.675161500622296e-05,
+      "loss": 0.8514,
+      "step": 22440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6736798435370117e-05,
+      "loss": 1.021,
+      "step": 22450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6721981864517278e-05,
+      "loss": 0.9384,
+      "step": 22460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6707165293664435e-05,
+      "loss": 0.7759,
+      "step": 22470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6692348722811592e-05,
+      "loss": 0.8767,
+      "step": 22480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6677532151958753e-05,
+      "loss": 0.7918,
+      "step": 22490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.666271558110591e-05,
+      "loss": 0.8851,
+      "step": 22500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6647899010253067e-05,
+      "loss": 0.9851,
+      "step": 22510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6633082439400228e-05,
+      "loss": 0.9906,
+      "step": 22520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6618265868547385e-05,
+      "loss": 0.9234,
+      "step": 22530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6603449297694542e-05,
+      "loss": 0.8392,
+      "step": 22540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6588632726841703e-05,
+      "loss": 0.8439,
+      "step": 22550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6573816155988857e-05,
+      "loss": 0.8402,
+      "step": 22560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6558999585136017e-05,
+      "loss": 0.7782,
+      "step": 22570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6544183014283174e-05,
+      "loss": 0.9362,
+      "step": 22580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6529366443430332e-05,
+      "loss": 0.8373,
+      "step": 22590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6514549872577492e-05,
+      "loss": 0.8784,
+      "step": 22600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.649973330172465e-05,
+      "loss": 0.678,
+      "step": 22610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6484916730871807e-05,
+      "loss": 0.7156,
+      "step": 22620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6470100160018967e-05,
+      "loss": 0.7588,
+      "step": 22630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6455283589166125e-05,
+      "loss": 0.9054,
+      "step": 22640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6440467018313282e-05,
+      "loss": 0.7793,
+      "step": 22650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6425650447460443e-05,
+      "loss": 0.7883,
+      "step": 22660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.64108338766076e-05,
+      "loss": 0.8559,
+      "step": 22670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6396017305754757e-05,
+      "loss": 0.7894,
+      "step": 22680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6381200734901918e-05,
+      "loss": 0.7528,
+      "step": 22690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.636638416404907e-05,
+      "loss": 0.8207,
+      "step": 22700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6351567593196232e-05,
+      "loss": 1.0498,
+      "step": 22710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.633675102234339e-05,
+      "loss": 0.7424,
+      "step": 22720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6321934451490547e-05,
+      "loss": 1.0651,
+      "step": 22730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6307117880637707e-05,
+      "loss": 0.8204,
+      "step": 22740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6292301309784864e-05,
+      "loss": 0.8968,
+      "step": 22750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.627748473893202e-05,
+      "loss": 0.8417,
+      "step": 22760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6262668168079182e-05,
+      "loss": 0.7999,
+      "step": 22770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.624785159722634e-05,
+      "loss": 0.8191,
+      "step": 22780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6233035026373497e-05,
+      "loss": 0.8413,
+      "step": 22790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6218218455520657e-05,
+      "loss": 0.9028,
+      "step": 22800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.620340188466781e-05,
+      "loss": 0.9951,
+      "step": 22810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6188585313814972e-05,
+      "loss": 0.8284,
+      "step": 22820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6173768742962132e-05,
+      "loss": 0.8921,
+      "step": 22830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6158952172109286e-05,
+      "loss": 0.7294,
+      "step": 22840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6144135601256447e-05,
+      "loss": 0.892,
+      "step": 22850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6129319030403604e-05,
+      "loss": 0.7497,
+      "step": 22860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.611450245955076e-05,
+      "loss": 0.8052,
+      "step": 22870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6099685888697922e-05,
+      "loss": 0.7653,
+      "step": 22880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.608486931784508e-05,
+      "loss": 1.0207,
+      "step": 22890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6070052746992236e-05,
+      "loss": 0.7666,
+      "step": 22900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6055236176139397e-05,
+      "loss": 0.7809,
+      "step": 22910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.604041960528655e-05,
+      "loss": 0.8804,
+      "step": 22920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.602560303443371e-05,
+      "loss": 0.8108,
+      "step": 22930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6010786463580872e-05,
+      "loss": 0.8421,
+      "step": 22940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5995969892728026e-05,
+      "loss": 0.9035,
+      "step": 22950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5981153321875187e-05,
+      "loss": 0.7245,
+      "step": 22960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5966336751022344e-05,
+      "loss": 0.8261,
+      "step": 22970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.59515201801695e-05,
+      "loss": 0.9342,
+      "step": 22980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.593670360931666e-05,
+      "loss": 0.7752,
+      "step": 22990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.592188703846382e-05,
+      "loss": 0.8349,
+      "step": 23000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5907070467610976e-05,
+      "loss": 0.7854,
+      "step": 23010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5892253896758137e-05,
+      "loss": 0.866,
+      "step": 23020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.587743732590529e-05,
+      "loss": 0.8379,
+      "step": 23030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.586262075505245e-05,
+      "loss": 0.771,
+      "step": 23040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5847804184199612e-05,
+      "loss": 0.7034,
+      "step": 23050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5832987613346766e-05,
+      "loss": 0.8434,
+      "step": 23060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5818171042493926e-05,
+      "loss": 0.7547,
+      "step": 23070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5803354471641087e-05,
+      "loss": 0.8964,
+      "step": 23080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.578853790078824e-05,
+      "loss": 0.8574,
+      "step": 23090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.57737213299354e-05,
+      "loss": 0.8381,
+      "step": 23100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.575890475908256e-05,
+      "loss": 0.8343,
+      "step": 23110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5744088188229716e-05,
+      "loss": 0.8274,
+      "step": 23120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5729271617376876e-05,
+      "loss": 0.8053,
+      "step": 23130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.571445504652403e-05,
+      "loss": 0.8456,
+      "step": 23140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.569963847567119e-05,
+      "loss": 0.969,
+      "step": 23150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.568482190481835e-05,
+      "loss": 0.8612,
+      "step": 23160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5670005333965505e-05,
+      "loss": 0.8344,
+      "step": 23170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5655188763112666e-05,
+      "loss": 0.9009,
+      "step": 23180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5640372192259827e-05,
+      "loss": 0.9595,
+      "step": 23190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.562555562140698e-05,
+      "loss": 0.8761,
+      "step": 23200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.561073905055414e-05,
+      "loss": 0.8111,
+      "step": 23210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.55959224797013e-05,
+      "loss": 0.8634,
+      "step": 23220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5581105908848456e-05,
+      "loss": 0.835,
+      "step": 23230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5566289337995616e-05,
+      "loss": 0.9587,
+      "step": 23240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5551472767142773e-05,
+      "loss": 0.7974,
+      "step": 23250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.553665619628993e-05,
+      "loss": 0.7252,
+      "step": 23260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.552183962543709e-05,
+      "loss": 1.0557,
+      "step": 23270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5507023054584245e-05,
+      "loss": 1.0059,
+      "step": 23280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5492206483731406e-05,
+      "loss": 0.8877,
+      "step": 23290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5477389912878566e-05,
+      "loss": 0.8916,
+      "step": 23300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.546257334202572e-05,
+      "loss": 0.8149,
+      "step": 23310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.544775677117288e-05,
+      "loss": 0.8561,
+      "step": 23320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5432940200320038e-05,
+      "loss": 0.8927,
+      "step": 23330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5418123629467195e-05,
+      "loss": 0.7797,
+      "step": 23340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5403307058614356e-05,
+      "loss": 0.7539,
+      "step": 23350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5388490487761513e-05,
+      "loss": 0.9578,
+      "step": 23360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.537367391690867e-05,
+      "loss": 0.9525,
+      "step": 23370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.535885734605583e-05,
+      "loss": 0.8471,
+      "step": 23380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5344040775202988e-05,
+      "loss": 0.864,
+      "step": 23390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5329224204350145e-05,
+      "loss": 0.8827,
+      "step": 23400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5314407633497306e-05,
+      "loss": 0.8778,
+      "step": 23410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.529959106264446e-05,
+      "loss": 0.8729,
+      "step": 23420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.528477449179162e-05,
+      "loss": 0.954,
+      "step": 23430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5269957920938778e-05,
+      "loss": 0.7923,
+      "step": 23440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5255141350085935e-05,
+      "loss": 0.8829,
+      "step": 23450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5240324779233096e-05,
+      "loss": 0.865,
+      "step": 23460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5225508208380255e-05,
+      "loss": 0.8871,
+      "step": 23470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.521069163752741e-05,
+      "loss": 0.8585,
+      "step": 23480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5195875066674569e-05,
+      "loss": 0.9678,
+      "step": 23490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.518105849582173e-05,
+      "loss": 0.8421,
+      "step": 23500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5166241924968885e-05,
+      "loss": 0.7487,
+      "step": 23510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5151425354116044e-05,
+      "loss": 0.86,
+      "step": 23520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5136608783263203e-05,
+      "loss": 0.8443,
+      "step": 23530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.512179221241036e-05,
+      "loss": 0.8046,
+      "step": 23540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.510697564155752e-05,
+      "loss": 0.9261,
+      "step": 23550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5092159070704676e-05,
+      "loss": 0.8575,
+      "step": 23560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5077342499851835e-05,
+      "loss": 0.8521,
+      "step": 23570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5062525928998994e-05,
+      "loss": 0.8765,
+      "step": 23580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.504770935814615e-05,
+      "loss": 0.9086,
+      "step": 23590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5032892787293309e-05,
+      "loss": 0.8513,
+      "step": 23600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.501807621644047e-05,
+      "loss": 0.9946,
+      "step": 23610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5003259645587625e-05,
+      "loss": 0.9462,
+      "step": 23620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4988443074734784e-05,
+      "loss": 0.8317,
+      "step": 23630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4973626503881943e-05,
+      "loss": 0.9471,
+      "step": 23640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.49588099330291e-05,
+      "loss": 0.8202,
+      "step": 23650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4943993362176259e-05,
+      "loss": 0.8416,
+      "step": 23660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4929176791323418e-05,
+      "loss": 0.8742,
+      "step": 23670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4914360220470575e-05,
+      "loss": 0.787,
+      "step": 23680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4899543649617734e-05,
+      "loss": 0.8054,
+      "step": 23690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.488472707876489e-05,
+      "loss": 0.8448,
+      "step": 23700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4869910507912048e-05,
+      "loss": 0.7907,
+      "step": 23710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4855093937059209e-05,
+      "loss": 0.7657,
+      "step": 23720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4840277366206365e-05,
+      "loss": 0.7114,
+      "step": 23730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4825460795353524e-05,
+      "loss": 0.9068,
+      "step": 23740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4810644224500682e-05,
+      "loss": 0.7937,
+      "step": 23750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.479582765364784e-05,
+      "loss": 0.7901,
+      "step": 23760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4781011082794999e-05,
+      "loss": 0.7857,
+      "step": 23770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4766194511942158e-05,
+      "loss": 0.8562,
+      "step": 23780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4751377941089315e-05,
+      "loss": 0.7484,
+      "step": 23790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4736561370236474e-05,
+      "loss": 0.8844,
+      "step": 23800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4721744799383633e-05,
+      "loss": 0.9456,
+      "step": 23810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.470692822853079e-05,
+      "loss": 0.8661,
+      "step": 23820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4692111657677949e-05,
+      "loss": 0.8893,
+      "step": 23830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4677295086825104e-05,
+      "loss": 0.8261,
+      "step": 23840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4662478515972263e-05,
+      "loss": 0.9312,
+      "step": 23850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4647661945119422e-05,
+      "loss": 0.764,
+      "step": 23860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.463284537426658e-05,
+      "loss": 0.7764,
+      "step": 23870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4618028803413738e-05,
+      "loss": 0.809,
+      "step": 23880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4603212232560897e-05,
+      "loss": 0.8024,
+      "step": 23890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4588395661708055e-05,
+      "loss": 1.019,
+      "step": 23900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4573579090855213e-05,
+      "loss": 0.8105,
+      "step": 23910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4558762520002372e-05,
+      "loss": 0.8173,
+      "step": 23920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.454394594914953e-05,
+      "loss": 0.8577,
+      "step": 23930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4529129378296689e-05,
+      "loss": 0.86,
+      "step": 23940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4514312807443847e-05,
+      "loss": 0.8168,
+      "step": 23950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4499496236591003e-05,
+      "loss": 0.7918,
+      "step": 23960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4484679665738164e-05,
+      "loss": 0.8372,
+      "step": 23970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.446986309488532e-05,
+      "loss": 0.8965,
+      "step": 23980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4455046524032478e-05,
+      "loss": 0.7732,
+      "step": 23990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4440229953179637e-05,
+      "loss": 0.9507,
+      "step": 24000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4425413382326794e-05,
+      "loss": 0.8227,
+      "step": 24010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4410596811473953e-05,
+      "loss": 0.7898,
+      "step": 24020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4395780240621112e-05,
+      "loss": 0.7134,
+      "step": 24030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.438096366976827e-05,
+      "loss": 0.7505,
+      "step": 24040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4366147098915428e-05,
+      "loss": 0.8048,
+      "step": 24050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4351330528062587e-05,
+      "loss": 0.7904,
+      "step": 24060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4336513957209743e-05,
+      "loss": 0.837,
+      "step": 24070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4321697386356903e-05,
+      "loss": 0.9069,
+      "step": 24080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4306880815504062e-05,
+      "loss": 0.9193,
+      "step": 24090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4292064244651218e-05,
+      "loss": 0.7904,
+      "step": 24100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4277247673798377e-05,
+      "loss": 0.811,
+      "step": 24110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4262431102945534e-05,
+      "loss": 0.6747,
+      "step": 24120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4247614532092693e-05,
+      "loss": 0.8461,
+      "step": 24130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4232797961239852e-05,
+      "loss": 0.9374,
+      "step": 24140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4217981390387009e-05,
+      "loss": 0.8721,
+      "step": 24150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4203164819534168e-05,
+      "loss": 0.8494,
+      "step": 24160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4188348248681327e-05,
+      "loss": 0.867,
+      "step": 24170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4173531677828482e-05,
+      "loss": 0.7724,
+      "step": 24180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4158715106975643e-05,
+      "loss": 0.8602,
+      "step": 24190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4143898536122802e-05,
+      "loss": 0.8801,
+      "step": 24200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4129081965269958e-05,
+      "loss": 0.9309,
+      "step": 24210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4114265394417116e-05,
+      "loss": 0.7233,
+      "step": 24220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4099448823564277e-05,
+      "loss": 0.8141,
+      "step": 24230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4084632252711433e-05,
+      "loss": 0.9368,
+      "step": 24240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4069815681858592e-05,
+      "loss": 1.0427,
+      "step": 24250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4054999111005749e-05,
+      "loss": 0.7932,
+      "step": 24260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4040182540152908e-05,
+      "loss": 0.7997,
+      "step": 24270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4025365969300067e-05,
+      "loss": 0.7913,
+      "step": 24280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4010549398447224e-05,
+      "loss": 0.6766,
+      "step": 24290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3995732827594383e-05,
+      "loss": 0.9067,
+      "step": 24300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3980916256741542e-05,
+      "loss": 0.8415,
+      "step": 24310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3966099685888697e-05,
+      "loss": 0.7203,
+      "step": 24320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3951283115035856e-05,
+      "loss": 0.7764,
+      "step": 24330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3936466544183017e-05,
+      "loss": 0.8258,
+      "step": 24340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3921649973330172e-05,
+      "loss": 0.8167,
+      "step": 24350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3906833402477331e-05,
+      "loss": 0.8011,
+      "step": 24360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.389201683162449e-05,
+      "loss": 0.8363,
+      "step": 24370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3877200260771647e-05,
+      "loss": 0.7716,
+      "step": 24380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3862383689918806e-05,
+      "loss": 0.8876,
+      "step": 24390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3847567119065964e-05,
+      "loss": 0.7226,
+      "step": 24400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3832750548213123e-05,
+      "loss": 0.7654,
+      "step": 24410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3817933977360281e-05,
+      "loss": 0.8115,
+      "step": 24420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3803117406507437e-05,
+      "loss": 0.8248,
+      "step": 24430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3788300835654596e-05,
+      "loss": 0.8159,
+      "step": 24440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3773484264801757e-05,
+      "loss": 0.7409,
+      "step": 24450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3758667693948912e-05,
+      "loss": 0.8088,
+      "step": 24460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3743851123096071e-05,
+      "loss": 0.7538,
+      "step": 24470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.372903455224323e-05,
+      "loss": 0.9526,
+      "step": 24480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3714217981390387e-05,
+      "loss": 0.9346,
+      "step": 24490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3699401410537546e-05,
+      "loss": 0.955,
+      "step": 24500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3684584839684705e-05,
+      "loss": 0.8281,
+      "step": 24510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3669768268831862e-05,
+      "loss": 0.7606,
+      "step": 24520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3654951697979021e-05,
+      "loss": 0.8452,
+      "step": 24530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3640135127126177e-05,
+      "loss": 0.7429,
+      "step": 24540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3625318556273337e-05,
+      "loss": 0.8333,
+      "step": 24550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3610501985420496e-05,
+      "loss": 0.8538,
+      "step": 24560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3595685414567652e-05,
+      "loss": 0.8278,
+      "step": 24570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.358086884371481e-05,
+      "loss": 0.8386,
+      "step": 24580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.356605227286197e-05,
+      "loss": 0.9109,
+      "step": 24590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3551235702009127e-05,
+      "loss": 0.66,
+      "step": 24600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3536419131156286e-05,
+      "loss": 0.9268,
+      "step": 24610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3521602560303445e-05,
+      "loss": 0.7134,
+      "step": 24620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3506785989450602e-05,
+      "loss": 0.6895,
+      "step": 24630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3491969418597761e-05,
+      "loss": 0.8673,
+      "step": 24640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.347715284774492e-05,
+      "loss": 0.7157,
+      "step": 24650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3462336276892077e-05,
+      "loss": 0.9058,
+      "step": 24660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3447519706039236e-05,
+      "loss": 1.0182,
+      "step": 24670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3432703135186391e-05,
+      "loss": 0.8496,
+      "step": 24680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.341788656433355e-05,
+      "loss": 0.7644,
+      "step": 24690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3403069993480711e-05,
+      "loss": 0.9253,
+      "step": 24700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3388253422627867e-05,
+      "loss": 0.873,
+      "step": 24710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3373436851775026e-05,
+      "loss": 0.7271,
+      "step": 24720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3358620280922184e-05,
+      "loss": 0.8746,
+      "step": 24730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3343803710069342e-05,
+      "loss": 0.83,
+      "step": 24740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.33289871392165e-05,
+      "loss": 0.7188,
+      "step": 24750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.331417056836366e-05,
+      "loss": 0.6843,
+      "step": 24760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3299353997510817e-05,
+      "loss": 0.8361,
+      "step": 24770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3284537426657976e-05,
+      "loss": 0.9013,
+      "step": 24780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3269720855805135e-05,
+      "loss": 0.9251,
+      "step": 24790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.325490428495229e-05,
+      "loss": 0.7374,
+      "step": 24800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.324008771409945e-05,
+      "loss": 0.9033,
+      "step": 24810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3225271143246606e-05,
+      "loss": 0.7938,
+      "step": 24820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3210454572393765e-05,
+      "loss": 0.8459,
+      "step": 24830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3195638001540924e-05,
+      "loss": 0.7765,
+      "step": 24840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3180821430688081e-05,
+      "loss": 0.7869,
+      "step": 24850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.316600485983524e-05,
+      "loss": 0.8801,
+      "step": 24860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.31511882889824e-05,
+      "loss": 0.9698,
+      "step": 24870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3136371718129556e-05,
+      "loss": 0.8312,
+      "step": 24880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3121555147276715e-05,
+      "loss": 0.9598,
+      "step": 24890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3106738576423874e-05,
+      "loss": 0.8063,
+      "step": 24900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.309192200557103e-05,
+      "loss": 0.964,
+      "step": 24910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.307710543471819e-05,
+      "loss": 0.7865,
+      "step": 24920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.306228886386535e-05,
+      "loss": 0.7829,
+      "step": 24930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3047472293012505e-05,
+      "loss": 0.8451,
+      "step": 24940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3032655722159664e-05,
+      "loss": 0.7222,
+      "step": 24950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3017839151306821e-05,
+      "loss": 0.7559,
+      "step": 24960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.300302258045398e-05,
+      "loss": 0.9221,
+      "step": 24970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2988206009601139e-05,
+      "loss": 0.809,
+      "step": 24980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2973389438748296e-05,
+      "loss": 0.7493,
+      "step": 24990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2958572867895455e-05,
+      "loss": 0.9133,
+      "step": 25000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2943756297042614e-05,
+      "loss": 0.796,
+      "step": 25010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2928939726189771e-05,
+      "loss": 0.7717,
+      "step": 25020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.291412315533693e-05,
+      "loss": 0.954,
+      "step": 25030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2899306584484089e-05,
+      "loss": 0.803,
+      "step": 25040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2884490013631245e-05,
+      "loss": 0.7036,
+      "step": 25050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2869673442778404e-05,
+      "loss": 0.8272,
+      "step": 25060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2854856871925564e-05,
+      "loss": 0.8529,
+      "step": 25070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.284004030107272e-05,
+      "loss": 0.9032,
+      "step": 25080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2825223730219879e-05,
+      "loss": 0.8864,
+      "step": 25090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2810407159367036e-05,
+      "loss": 0.7249,
+      "step": 25100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2795590588514195e-05,
+      "loss": 0.9738,
+      "step": 25110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2780774017661354e-05,
+      "loss": 0.8423,
+      "step": 25120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2765957446808511e-05,
+      "loss": 0.951,
+      "step": 25130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.275114087595567e-05,
+      "loss": 0.8822,
+      "step": 25140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2736324305102829e-05,
+      "loss": 0.9904,
+      "step": 25150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2721507734249984e-05,
+      "loss": 0.9531,
+      "step": 25160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2706691163397143e-05,
+      "loss": 0.7682,
+      "step": 25170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2691874592544304e-05,
+      "loss": 0.8224,
+      "step": 25180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.267705802169146e-05,
+      "loss": 0.8131,
+      "step": 25190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2662241450838618e-05,
+      "loss": 0.877,
+      "step": 25200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2647424879985777e-05,
+      "loss": 0.8206,
+      "step": 25210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2632608309132935e-05,
+      "loss": 0.8753,
+      "step": 25220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2617791738280093e-05,
+      "loss": 0.8928,
+      "step": 25230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.260297516742725e-05,
+      "loss": 0.8714,
+      "step": 25240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.258815859657441e-05,
+      "loss": 0.7754,
+      "step": 25250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2573342025721569e-05,
+      "loss": 1.0292,
+      "step": 25260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2558525454868724e-05,
+      "loss": 0.8371,
+      "step": 25270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2543708884015885e-05,
+      "loss": 0.9046,
+      "step": 25280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2528892313163044e-05,
+      "loss": 0.729,
+      "step": 25290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.25140757423102e-05,
+      "loss": 0.7027,
+      "step": 25300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2499259171457358e-05,
+      "loss": 0.7705,
+      "step": 25310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2484442600604517e-05,
+      "loss": 0.8154,
+      "step": 25320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2469626029751674e-05,
+      "loss": 0.9587,
+      "step": 25330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2454809458898833e-05,
+      "loss": 0.8896,
+      "step": 25340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.243999288804599e-05,
+      "loss": 0.8733,
+      "step": 25350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.242517631719315e-05,
+      "loss": 1.0025,
+      "step": 25360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2410359746340308e-05,
+      "loss": 0.7811,
+      "step": 25370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2395543175487466e-05,
+      "loss": 0.8052,
+      "step": 25380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2380726604634624e-05,
+      "loss": 0.8242,
+      "step": 25390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2365910033781782e-05,
+      "loss": 0.7086,
+      "step": 25400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.235109346292894e-05,
+      "loss": 0.8182,
+      "step": 25410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2336276892076098e-05,
+      "loss": 0.8977,
+      "step": 25420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2321460321223257e-05,
+      "loss": 0.8453,
+      "step": 25430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2306643750370416e-05,
+      "loss": 0.9202,
+      "step": 25440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2291827179517573e-05,
+      "loss": 0.8283,
+      "step": 25450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2277010608664732e-05,
+      "loss": 0.8236,
+      "step": 25460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2262194037811889e-05,
+      "loss": 0.7732,
+      "step": 25470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2247377466959048e-05,
+      "loss": 0.8318,
+      "step": 25480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2232560896106205e-05,
+      "loss": 0.6824,
+      "step": 25490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2217744325253364e-05,
+      "loss": 0.7716,
+      "step": 25500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2202927754400523e-05,
+      "loss": 0.8433,
+      "step": 25510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.218811118354768e-05,
+      "loss": 0.7985,
+      "step": 25520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2173294612694838e-05,
+      "loss": 0.7839,
+      "step": 25530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2158478041841996e-05,
+      "loss": 0.7955,
+      "step": 25540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2143661470989155e-05,
+      "loss": 0.9498,
+      "step": 25550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2128844900136313e-05,
+      "loss": 0.7866,
+      "step": 25560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2114028329283472e-05,
+      "loss": 0.9004,
+      "step": 25570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.209921175843063e-05,
+      "loss": 0.8698,
+      "step": 25580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2084395187577788e-05,
+      "loss": 0.8557,
+      "step": 25590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2069578616724945e-05,
+      "loss": 0.8167,
+      "step": 25600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2054762045872104e-05,
+      "loss": 0.9107,
+      "step": 25610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2039945475019263e-05,
+      "loss": 0.7753,
+      "step": 25620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.202512890416642e-05,
+      "loss": 0.8963,
+      "step": 25630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2010312333313577e-05,
+      "loss": 0.9922,
+      "step": 25640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1995495762460738e-05,
+      "loss": 0.8771,
+      "step": 25650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1980679191607895e-05,
+      "loss": 0.9021,
+      "step": 25660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1965862620755052e-05,
+      "loss": 0.8496,
+      "step": 25670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1951046049902211e-05,
+      "loss": 0.8117,
+      "step": 25680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.193622947904937e-05,
+      "loss": 0.8069,
+      "step": 25690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1921412908196527e-05,
+      "loss": 0.7271,
+      "step": 25700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1906596337343685e-05,
+      "loss": 0.7499,
+      "step": 25710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1891779766490845e-05,
+      "loss": 0.7856,
+      "step": 25720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1876963195638003e-05,
+      "loss": 0.7633,
+      "step": 25730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.186214662478516e-05,
+      "loss": 0.8777,
+      "step": 25740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1847330053932319e-05,
+      "loss": 0.8289,
+      "step": 25750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1832513483079478e-05,
+      "loss": 0.9075,
+      "step": 25760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1817696912226635e-05,
+      "loss": 0.8961,
+      "step": 25770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1802880341373792e-05,
+      "loss": 0.7867,
+      "step": 25780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1788063770520951e-05,
+      "loss": 0.7762,
+      "step": 25790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.177324719966811e-05,
+      "loss": 0.8488,
+      "step": 25800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1758430628815267e-05,
+      "loss": 0.8515,
+      "step": 25810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1743614057962424e-05,
+      "loss": 0.8932,
+      "step": 25820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1728797487109585e-05,
+      "loss": 0.965,
+      "step": 25830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1713980916256742e-05,
+      "loss": 0.8563,
+      "step": 25840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.16991643454039e-05,
+      "loss": 0.8934,
+      "step": 25850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1684347774551058e-05,
+      "loss": 0.7615,
+      "step": 25860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1669531203698217e-05,
+      "loss": 0.8583,
+      "step": 25870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1654714632845375e-05,
+      "loss": 0.7768,
+      "step": 25880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1639898061992532e-05,
+      "loss": 0.8065,
+      "step": 25890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.162508149113969e-05,
+      "loss": 0.7745,
+      "step": 25900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.161026492028685e-05,
+      "loss": 1.0746,
+      "step": 25910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1595448349434007e-05,
+      "loss": 0.8585,
+      "step": 25920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1580631778581166e-05,
+      "loss": 0.956,
+      "step": 25930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1565815207728325e-05,
+      "loss": 0.8679,
+      "step": 25940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1550998636875482e-05,
+      "loss": 0.7557,
+      "step": 25950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.153618206602264e-05,
+      "loss": 0.7569,
+      "step": 25960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1521365495169798e-05,
+      "loss": 0.7993,
+      "step": 25970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1506548924316957e-05,
+      "loss": 0.6295,
+      "step": 25980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1491732353464114e-05,
+      "loss": 0.7848,
+      "step": 25990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1476915782611273e-05,
+      "loss": 0.7587,
+      "step": 26000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1462099211758432e-05,
+      "loss": 0.7593,
+      "step": 26010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.144728264090559e-05,
+      "loss": 0.8481,
+      "step": 26020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1432466070052747e-05,
+      "loss": 0.8453,
+      "step": 26030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1417649499199906e-05,
+      "loss": 0.9776,
+      "step": 26040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1402832928347064e-05,
+      "loss": 0.8218,
+      "step": 26050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1388016357494222e-05,
+      "loss": 0.9226,
+      "step": 26060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.137319978664138e-05,
+      "loss": 0.8553,
+      "step": 26070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1358383215788538e-05,
+      "loss": 0.8774,
+      "step": 26080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1343566644935697e-05,
+      "loss": 0.8202,
+      "step": 26090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1328750074082854e-05,
+      "loss": 0.9213,
+      "step": 26100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1313933503230013e-05,
+      "loss": 0.8127,
+      "step": 26110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1299116932377172e-05,
+      "loss": 0.9204,
+      "step": 26120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1284300361524329e-05,
+      "loss": 0.8596,
+      "step": 26130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1269483790671488e-05,
+      "loss": 0.9033,
+      "step": 26140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1254667219818645e-05,
+      "loss": 0.9182,
+      "step": 26150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1239850648965804e-05,
+      "loss": 0.862,
+      "step": 26160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1225034078112961e-05,
+      "loss": 0.6718,
+      "step": 26170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.121021750726012e-05,
+      "loss": 0.7975,
+      "step": 26180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.119540093640728e-05,
+      "loss": 0.7768,
+      "step": 26190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1180584365554437e-05,
+      "loss": 0.8502,
+      "step": 26200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1165767794701595e-05,
+      "loss": 0.8791,
+      "step": 26210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1150951223848753e-05,
+      "loss": 0.7101,
+      "step": 26220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1136134652995912e-05,
+      "loss": 0.8722,
+      "step": 26230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1121318082143069e-05,
+      "loss": 0.9821,
+      "step": 26240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1106501511290228e-05,
+      "loss": 0.8882,
+      "step": 26250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1091684940437385e-05,
+      "loss": 0.8302,
+      "step": 26260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1076868369584544e-05,
+      "loss": 0.7746,
+      "step": 26270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1062051798731703e-05,
+      "loss": 0.8006,
+      "step": 26280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.104723522787886e-05,
+      "loss": 0.8414,
+      "step": 26290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1032418657026019e-05,
+      "loss": 0.9094,
+      "step": 26300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1017602086173176e-05,
+      "loss": 0.842,
+      "step": 26310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1002785515320335e-05,
+      "loss": 0.8943,
+      "step": 26320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0987968944467492e-05,
+      "loss": 0.7498,
+      "step": 26330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0973152373614651e-05,
+      "loss": 0.8495,
+      "step": 26340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.095833580276181e-05,
+      "loss": 0.8342,
+      "step": 26350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0943519231908967e-05,
+      "loss": 0.8103,
+      "step": 26360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0928702661056125e-05,
+      "loss": 0.7783,
+      "step": 26370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0913886090203284e-05,
+      "loss": 0.7801,
+      "step": 26380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0899069519350443e-05,
+      "loss": 0.8705,
+      "step": 26390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.08842529484976e-05,
+      "loss": 0.8933,
+      "step": 26400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0869436377644759e-05,
+      "loss": 0.6934,
+      "step": 26410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0854619806791918e-05,
+      "loss": 0.7823,
+      "step": 26420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0839803235939075e-05,
+      "loss": 0.819,
+      "step": 26430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0824986665086232e-05,
+      "loss": 0.7221,
+      "step": 26440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0810170094233391e-05,
+      "loss": 0.8498,
+      "step": 26450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.079535352338055e-05,
+      "loss": 0.7632,
+      "step": 26460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0780536952527707e-05,
+      "loss": 0.8229,
+      "step": 26470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0765720381674866e-05,
+      "loss": 0.7822,
+      "step": 26480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0750903810822025e-05,
+      "loss": 0.7216,
+      "step": 26490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0736087239969182e-05,
+      "loss": 0.855,
+      "step": 26500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.072127066911634e-05,
+      "loss": 0.9372,
+      "step": 26510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0706454098263498e-05,
+      "loss": 0.8273,
+      "step": 26520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0691637527410657e-05,
+      "loss": 0.8511,
+      "step": 26530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0676820956557815e-05,
+      "loss": 0.953,
+      "step": 26540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0662004385704972e-05,
+      "loss": 0.7749,
+      "step": 26550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0647187814852132e-05,
+      "loss": 0.9035,
+      "step": 26560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.063237124399929e-05,
+      "loss": 0.7956,
+      "step": 26570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0617554673146447e-05,
+      "loss": 0.8819,
+      "step": 26580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0602738102293606e-05,
+      "loss": 0.9241,
+      "step": 26590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0587921531440765e-05,
+      "loss": 0.8081,
+      "step": 26600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0573104960587922e-05,
+      "loss": 0.8689,
+      "step": 26610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.055828838973508e-05,
+      "loss": 0.8158,
+      "step": 26620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0543471818882238e-05,
+      "loss": 0.9474,
+      "step": 26630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0528655248029397e-05,
+      "loss": 0.8814,
+      "step": 26640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0513838677176554e-05,
+      "loss": 0.8688,
+      "step": 26650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0499022106323712e-05,
+      "loss": 0.8209,
+      "step": 26660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0484205535470872e-05,
+      "loss": 0.7129,
+      "step": 26670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.046938896461803e-05,
+      "loss": 0.8338,
+      "step": 26680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0454572393765187e-05,
+      "loss": 0.876,
+      "step": 26690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0439755822912346e-05,
+      "loss": 0.8838,
+      "step": 26700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0424939252059504e-05,
+      "loss": 0.9687,
+      "step": 26710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0410122681206662e-05,
+      "loss": 0.9331,
+      "step": 26720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0395306110353819e-05,
+      "loss": 0.8596,
+      "step": 26730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.038048953950098e-05,
+      "loss": 0.9814,
+      "step": 26740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0365672968648137e-05,
+      "loss": 0.8254,
+      "step": 26750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0350856397795294e-05,
+      "loss": 0.8314,
+      "step": 26760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0336039826942453e-05,
+      "loss": 0.8254,
+      "step": 26770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0321223256089612e-05,
+      "loss": 0.8073,
+      "step": 26780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0306406685236769e-05,
+      "loss": 0.8626,
+      "step": 26790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0291590114383926e-05,
+      "loss": 0.8962,
+      "step": 26800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0276773543531085e-05,
+      "loss": 0.8781,
+      "step": 26810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0261956972678244e-05,
+      "loss": 0.8088,
+      "step": 26820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0247140401825401e-05,
+      "loss": 0.9101,
+      "step": 26830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.023232383097256e-05,
+      "loss": 0.8275,
+      "step": 26840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.021750726011972e-05,
+      "loss": 0.7513,
+      "step": 26850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0202690689266877e-05,
+      "loss": 0.9617,
+      "step": 26860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0187874118414034e-05,
+      "loss": 0.8926,
+      "step": 26870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0173057547561193e-05,
+      "loss": 0.8115,
+      "step": 26880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0158240976708352e-05,
+      "loss": 0.8416,
+      "step": 26890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0143424405855509e-05,
+      "loss": 0.9499,
+      "step": 26900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0128607835002668e-05,
+      "loss": 0.8245,
+      "step": 26910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0113791264149827e-05,
+      "loss": 0.8692,
+      "step": 26920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0098974693296984e-05,
+      "loss": 0.7294,
+      "step": 26930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0084158122444141e-05,
+      "loss": 0.911,
+      "step": 26940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.00693415515913e-05,
+      "loss": 0.8318,
+      "step": 26950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0054524980738459e-05,
+      "loss": 0.8141,
+      "step": 26960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0039708409885616e-05,
+      "loss": 0.7381,
+      "step": 26970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0024891839032775e-05,
+      "loss": 0.8144,
+      "step": 26980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0010075268179932e-05,
+      "loss": 0.825,
+      "step": 26990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.995258697327091e-06,
+      "loss": 0.7575,
+      "step": 27000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.980442126474249e-06,
+      "loss": 0.8501,
+      "step": 27010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.965625555621407e-06,
+      "loss": 0.8956,
+      "step": 27020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.950808984768566e-06,
+      "loss": 0.8825,
+      "step": 27030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.935992413915724e-06,
+      "loss": 0.9027,
+      "step": 27040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.921175843062883e-06,
+      "loss": 0.9106,
+      "step": 27050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.90635927221004e-06,
+      "loss": 0.8476,
+      "step": 27060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.891542701357199e-06,
+      "loss": 0.7358,
+      "step": 27070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.876726130504356e-06,
+      "loss": 0.856,
+      "step": 27080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.861909559651515e-06,
+      "loss": 0.7659,
+      "step": 27090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.847092988798672e-06,
+      "loss": 0.9216,
+      "step": 27100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.832276417945831e-06,
+      "loss": 0.729,
+      "step": 27110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.81745984709299e-06,
+      "loss": 0.8307,
+      "step": 27120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.802643276240147e-06,
+      "loss": 0.8572,
+      "step": 27130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.787826705387306e-06,
+      "loss": 0.8346,
+      "step": 27140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.773010134534463e-06,
+      "loss": 1.0176,
+      "step": 27150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.758193563681622e-06,
+      "loss": 0.8436,
+      "step": 27160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.74337699282878e-06,
+      "loss": 0.8001,
+      "step": 27170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.728560421975938e-06,
+      "loss": 0.7491,
+      "step": 27180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.713743851123097e-06,
+      "loss": 0.7306,
+      "step": 27190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.698927280270255e-06,
+      "loss": 0.8969,
+      "step": 27200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.684110709417414e-06,
+      "loss": 0.7704,
+      "step": 27210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.66929413856457e-06,
+      "loss": 0.8897,
+      "step": 27220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.65447756771173e-06,
+      "loss": 0.7944,
+      "step": 27230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.639660996858887e-06,
+      "loss": 0.7888,
+      "step": 27240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.624844426006046e-06,
+      "loss": 0.8273,
+      "step": 27250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.610027855153205e-06,
+      "loss": 0.8328,
+      "step": 27260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.595211284300362e-06,
+      "loss": 0.8889,
+      "step": 27270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.58039471344752e-06,
+      "loss": 0.7882,
+      "step": 27280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.565578142594678e-06,
+      "loss": 0.8158,
+      "step": 27290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.550761571741837e-06,
+      "loss": 0.7459,
+      "step": 27300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.535945000888994e-06,
+      "loss": 0.87,
+      "step": 27310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.521128430036153e-06,
+      "loss": 0.7732,
+      "step": 27320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.506311859183312e-06,
+      "loss": 0.8097,
+      "step": 27330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.49149528833047e-06,
+      "loss": 0.8793,
+      "step": 27340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.476678717477627e-06,
+      "loss": 0.9013,
+      "step": 27350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.461862146624786e-06,
+      "loss": 0.8694,
+      "step": 27360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.447045575771945e-06,
+      "loss": 0.8583,
+      "step": 27370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.432229004919102e-06,
+      "loss": 0.8503,
+      "step": 27380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.417412434066259e-06,
+      "loss": 0.8637,
+      "step": 27390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.40259586321342e-06,
+      "loss": 0.7903,
+      "step": 27400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.387779292360577e-06,
+      "loss": 0.8728,
+      "step": 27410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.372962721507734e-06,
+      "loss": 0.7439,
+      "step": 27420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.358146150654893e-06,
+      "loss": 0.8156,
+      "step": 27430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.343329579802052e-06,
+      "loss": 0.9474,
+      "step": 27440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.328513008949209e-06,
+      "loss": 0.9365,
+      "step": 27450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.313696438096366e-06,
+      "loss": 0.786,
+      "step": 27460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.298879867243527e-06,
+      "loss": 0.7976,
+      "step": 27470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.284063296390684e-06,
+      "loss": 0.9132,
+      "step": 27480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.269246725537841e-06,
+      "loss": 0.8022,
+      "step": 27490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.254430154685e-06,
+      "loss": 0.8411,
+      "step": 27500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.23961358383216e-06,
+      "loss": 0.8011,
+      "step": 27510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.224797012979317e-06,
+      "loss": 0.9507,
+      "step": 27520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.209980442126474e-06,
+      "loss": 0.7688,
+      "step": 27530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.195163871273633e-06,
+      "loss": 0.7427,
+      "step": 27540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.180347300420792e-06,
+      "loss": 0.7929,
+      "step": 27550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.165530729567949e-06,
+      "loss": 0.7609,
+      "step": 27560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.150714158715106e-06,
+      "loss": 0.8253,
+      "step": 27570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.135897587862267e-06,
+      "loss": 0.8494,
+      "step": 27580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.121081017009424e-06,
+      "loss": 0.7224,
+      "step": 27590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.106264446156581e-06,
+      "loss": 0.8241,
+      "step": 27600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.09144787530374e-06,
+      "loss": 0.8178,
+      "step": 27610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.076631304450899e-06,
+      "loss": 0.83,
+      "step": 27620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.061814733598056e-06,
+      "loss": 0.6989,
+      "step": 27630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.046998162745214e-06,
+      "loss": 1.1115,
+      "step": 27640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.032181591892374e-06,
+      "loss": 0.8682,
+      "step": 27650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.017365021039531e-06,
+      "loss": 0.8742,
+      "step": 27660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.002548450186689e-06,
+      "loss": 0.9065,
+      "step": 27670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.987731879333848e-06,
+      "loss": 0.7826,
+      "step": 27680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.972915308481006e-06,
+      "loss": 0.6103,
+      "step": 27690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.958098737628164e-06,
+      "loss": 0.795,
+      "step": 27700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.943282166775321e-06,
+      "loss": 0.9199,
+      "step": 27710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.92846559592248e-06,
+      "loss": 0.8204,
+      "step": 27720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.913649025069639e-06,
+      "loss": 0.7308,
+      "step": 27730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.898832454216796e-06,
+      "loss": 0.6076,
+      "step": 27740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.884015883363955e-06,
+      "loss": 0.9196,
+      "step": 27750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.869199312511114e-06,
+      "loss": 0.8956,
+      "step": 27760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.854382741658271e-06,
+      "loss": 0.8484,
+      "step": 27770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.839566170805428e-06,
+      "loss": 0.8989,
+      "step": 27780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.824749599952587e-06,
+      "loss": 0.8727,
+      "step": 27790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.809933029099746e-06,
+      "loss": 0.7134,
+      "step": 27800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.795116458246903e-06,
+      "loss": 0.8573,
+      "step": 27810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.780299887394062e-06,
+      "loss": 0.812,
+      "step": 27820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.76548331654122e-06,
+      "loss": 0.7814,
+      "step": 27830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.750666745688378e-06,
+      "loss": 0.7689,
+      "step": 27840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.735850174835536e-06,
+      "loss": 0.8401,
+      "step": 27850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.721033603982695e-06,
+      "loss": 0.7971,
+      "step": 27860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.706217033129854e-06,
+      "loss": 0.7765,
+      "step": 27870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.69140046227701e-06,
+      "loss": 0.9027,
+      "step": 27880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.67658389142417e-06,
+      "loss": 0.964,
+      "step": 27890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.661767320571327e-06,
+      "loss": 0.7856,
+      "step": 27900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.646950749718486e-06,
+      "loss": 1.0029,
+      "step": 27910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.632134178865643e-06,
+      "loss": 0.8419,
+      "step": 27920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.617317608012802e-06,
+      "loss": 0.8328,
+      "step": 27930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.602501037159961e-06,
+      "loss": 0.75,
+      "step": 27940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.587684466307118e-06,
+      "loss": 0.9438,
+      "step": 27950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.572867895454277e-06,
+      "loss": 0.8421,
+      "step": 27960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.558051324601434e-06,
+      "loss": 1.1142,
+      "step": 27970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.543234753748593e-06,
+      "loss": 1.0156,
+      "step": 27980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.52841818289575e-06,
+      "loss": 0.9018,
+      "step": 27990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.51360161204291e-06,
+      "loss": 0.7118,
+      "step": 28000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.498785041190067e-06,
+      "loss": 0.8623,
+      "step": 28010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.483968470337226e-06,
+      "loss": 0.773,
+      "step": 28020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.469151899484385e-06,
+      "loss": 0.8335,
+      "step": 28030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.454335328631542e-06,
+      "loss": 0.7075,
+      "step": 28040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.4395187577787e-06,
+      "loss": 0.9465,
+      "step": 28050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.424702186925858e-06,
+      "loss": 0.8742,
+      "step": 28060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.409885616073017e-06,
+      "loss": 0.7308,
+      "step": 28070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.395069045220174e-06,
+      "loss": 0.8197,
+      "step": 28080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.380252474367333e-06,
+      "loss": 0.752,
+      "step": 28090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.365435903514492e-06,
+      "loss": 0.8653,
+      "step": 28100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.35061933266165e-06,
+      "loss": 0.7343,
+      "step": 28110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.335802761808806e-06,
+      "loss": 0.7,
+      "step": 28120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.320986190955965e-06,
+      "loss": 0.7569,
+      "step": 28130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.306169620103124e-06,
+      "loss": 1.0034,
+      "step": 28140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.291353049250281e-06,
+      "loss": 0.8393,
+      "step": 28150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.27653647839744e-06,
+      "loss": 0.8237,
+      "step": 28160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.2617199075446e-06,
+      "loss": 0.787,
+      "step": 28170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.246903336691757e-06,
+      "loss": 0.8261,
+      "step": 28180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.232086765838914e-06,
+      "loss": 0.7429,
+      "step": 28190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.217270194986073e-06,
+      "loss": 0.7722,
+      "step": 28200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.202453624133232e-06,
+      "loss": 0.8481,
+      "step": 28210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.187637053280389e-06,
+      "loss": 0.8171,
+      "step": 28220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.172820482427548e-06,
+      "loss": 0.6574,
+      "step": 28230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.158003911574707e-06,
+      "loss": 0.8484,
+      "step": 28240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.143187340721864e-06,
+      "loss": 0.8342,
+      "step": 28250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.128370769869021e-06,
+      "loss": 0.9326,
+      "step": 28260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.11355419901618e-06,
+      "loss": 0.724,
+      "step": 28270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.098737628163339e-06,
+      "loss": 0.9796,
+      "step": 28280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.083921057310496e-06,
+      "loss": 0.7743,
+      "step": 28290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.069104486457654e-06,
+      "loss": 0.9398,
+      "step": 28300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.054287915604814e-06,
+      "loss": 0.7642,
+      "step": 28310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.039471344751971e-06,
+      "loss": 0.8557,
+      "step": 28320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.024654773899129e-06,
+      "loss": 0.787,
+      "step": 28330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.009838203046288e-06,
+      "loss": 0.8425,
+      "step": 28340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.995021632193446e-06,
+      "loss": 0.9097,
+      "step": 28350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.980205061340604e-06,
+      "loss": 0.8853,
+      "step": 28360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.965388490487761e-06,
+      "loss": 0.8632,
+      "step": 28370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.950571919634922e-06,
+      "loss": 0.6685,
+      "step": 28380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.935755348782079e-06,
+      "loss": 0.7471,
+      "step": 28390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.920938777929236e-06,
+      "loss": 0.9071,
+      "step": 28400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.906122207076393e-06,
+      "loss": 0.8453,
+      "step": 28410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.891305636223554e-06,
+      "loss": 1.0258,
+      "step": 28420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.876489065370711e-06,
+      "loss": 0.8297,
+      "step": 28430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.861672494517868e-06,
+      "loss": 0.6788,
+      "step": 28440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.846855923665027e-06,
+      "loss": 0.8631,
+      "step": 28450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.832039352812186e-06,
+      "loss": 0.8629,
+      "step": 28460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.817222781959343e-06,
+      "loss": 0.8019,
+      "step": 28470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.8024062111065e-06,
+      "loss": 0.8317,
+      "step": 28480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.787589640253661e-06,
+      "loss": 0.722,
+      "step": 28490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.772773069400818e-06,
+      "loss": 0.8622,
+      "step": 28500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.757956498547976e-06,
+      "loss": 0.8397,
+      "step": 28510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.743139927695135e-06,
+      "loss": 0.7164,
+      "step": 28520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.728323356842294e-06,
+      "loss": 0.853,
+      "step": 28530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.71350678598945e-06,
+      "loss": 0.7922,
+      "step": 28540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.698690215136608e-06,
+      "loss": 0.893,
+      "step": 28550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.683873644283767e-06,
+      "loss": 0.8251,
+      "step": 28560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.669057073430926e-06,
+      "loss": 0.9007,
+      "step": 28570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.654240502578083e-06,
+      "loss": 0.7733,
+      "step": 28580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.639423931725242e-06,
+      "loss": 0.822,
+      "step": 28590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.6246073608724e-06,
+      "loss": 0.7719,
+      "step": 28600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.609790790019558e-06,
+      "loss": 0.8517,
+      "step": 28610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5949742191667155e-06,
+      "loss": 0.855,
+      "step": 28620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.580157648313875e-06,
+      "loss": 0.9967,
+      "step": 28630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5653410774610325e-06,
+      "loss": 0.9014,
+      "step": 28640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5505245066081905e-06,
+      "loss": 0.8207,
+      "step": 28650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5357079357553495e-06,
+      "loss": 0.742,
+      "step": 28660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5208913649025075e-06,
+      "loss": 0.6991,
+      "step": 28670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.506074794049666e-06,
+      "loss": 0.8417,
+      "step": 28680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.491258223196823e-06,
+      "loss": 0.6742,
+      "step": 28690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.476441652343983e-06,
+      "loss": 0.811,
+      "step": 28700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.46162508149114e-06,
+      "loss": 0.8786,
+      "step": 28710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.446808510638298e-06,
+      "loss": 0.8787,
+      "step": 28720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.431991939785457e-06,
+      "loss": 0.918,
+      "step": 28730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.417175368932615e-06,
+      "loss": 0.7166,
+      "step": 28740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.402358798079773e-06,
+      "loss": 0.754,
+      "step": 28750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.38754222722693e-06,
+      "loss": 0.9353,
+      "step": 28760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.372725656374089e-06,
+      "loss": 0.7204,
+      "step": 28770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.357909085521247e-06,
+      "loss": 0.8485,
+      "step": 28780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.343092514668405e-06,
+      "loss": 0.7644,
+      "step": 28790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.328275943815564e-06,
+      "loss": 0.7375,
+      "step": 28800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.313459372962722e-06,
+      "loss": 0.9146,
+      "step": 28810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.29864280210988e-06,
+      "loss": 0.8822,
+      "step": 28820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.283826231257038e-06,
+      "loss": 0.8327,
+      "step": 28830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.269009660404197e-06,
+      "loss": 0.6424,
+      "step": 28840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.254193089551355e-06,
+      "loss": 0.7733,
+      "step": 28850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.239376518698513e-06,
+      "loss": 0.8612,
+      "step": 28860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.224559947845672e-06,
+      "loss": 0.7965,
+      "step": 28870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.20974337699283e-06,
+      "loss": 0.8674,
+      "step": 28880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.194926806139987e-06,
+      "loss": 0.8873,
+      "step": 28890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.180110235287145e-06,
+      "loss": 0.8673,
+      "step": 28900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.165293664434304e-06,
+      "loss": 0.8005,
+      "step": 28910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.150477093581462e-06,
+      "loss": 0.7605,
+      "step": 28920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.135660522728619e-06,
+      "loss": 0.7572,
+      "step": 28930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.120843951875779e-06,
+      "loss": 0.8536,
+      "step": 28940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.106027381022936e-06,
+      "loss": 0.8039,
+      "step": 28950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.091210810170094e-06,
+      "loss": 0.6013,
+      "step": 28960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.0763942393172525e-06,
+      "loss": 0.8377,
+      "step": 28970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.061577668464411e-06,
+      "loss": 0.7033,
+      "step": 28980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.0467610976115695e-06,
+      "loss": 0.692,
+      "step": 28990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.031944526758727e-06,
+      "loss": 0.8825,
+      "step": 29000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.0171279559058865e-06,
+      "loss": 0.8131,
+      "step": 29010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.002311385053044e-06,
+      "loss": 0.7374,
+      "step": 29020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.987494814200202e-06,
+      "loss": 0.637,
+      "step": 29030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.97267824334736e-06,
+      "loss": 0.8886,
+      "step": 29040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.957861672494519e-06,
+      "loss": 0.9611,
+      "step": 29050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.943045101641676e-06,
+      "loss": 0.8604,
+      "step": 29060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.928228530788834e-06,
+      "loss": 0.8934,
+      "step": 29070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.913411959935993e-06,
+      "loss": 0.768,
+      "step": 29080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.898595389083151e-06,
+      "loss": 0.8102,
+      "step": 29090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.883778818230309e-06,
+      "loss": 0.8218,
+      "step": 29100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8689622473774664e-06,
+      "loss": 0.8032,
+      "step": 29110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.854145676524626e-06,
+      "loss": 0.8869,
+      "step": 29120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8393291056717834e-06,
+      "loss": 0.8738,
+      "step": 29130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8245125348189415e-06,
+      "loss": 0.6768,
+      "step": 29140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8096959639661004e-06,
+      "loss": 0.9605,
+      "step": 29150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.7948793931132585e-06,
+      "loss": 0.7273,
+      "step": 29160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.780062822260417e-06,
+      "loss": 0.8115,
+      "step": 29170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.765246251407574e-06,
+      "loss": 0.7178,
+      "step": 29180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.750429680554734e-06,
+      "loss": 0.8241,
+      "step": 29190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.735613109701891e-06,
+      "loss": 0.6951,
+      "step": 29200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.720796538849049e-06,
+      "loss": 0.9257,
+      "step": 29210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.705979967996208e-06,
+      "loss": 0.8817,
+      "step": 29220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.691163397143366e-06,
+      "loss": 0.8851,
+      "step": 29230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.676346826290523e-06,
+      "loss": 0.7645,
+      "step": 29240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.661530255437681e-06,
+      "loss": 0.8227,
+      "step": 29250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.64671368458484e-06,
+      "loss": 0.8884,
+      "step": 29260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.631897113731998e-06,
+      "loss": 0.9422,
+      "step": 29270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.617080542879156e-06,
+      "loss": 0.7692,
+      "step": 29280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.602263972026315e-06,
+      "loss": 0.7218,
+      "step": 29290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.587447401173473e-06,
+      "loss": 0.7579,
+      "step": 29300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5726308303206306e-06,
+      "loss": 0.8306,
+      "step": 29310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.557814259467789e-06,
+      "loss": 0.7826,
+      "step": 29320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5429976886149476e-06,
+      "loss": 0.8808,
+      "step": 29330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.528181117762106e-06,
+      "loss": 0.7251,
+      "step": 29340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.513364546909263e-06,
+      "loss": 0.879,
+      "step": 29350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.498547976056423e-06,
+      "loss": 0.7672,
+      "step": 29360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.48373140520358e-06,
+      "loss": 0.8023,
+      "step": 29370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.468914834350738e-06,
+      "loss": 0.7502,
+      "step": 29380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.454098263497896e-06,
+      "loss": 0.7522,
+      "step": 29390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.439281692645055e-06,
+      "loss": 0.8307,
+      "step": 29400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.424465121792213e-06,
+      "loss": 0.8703,
+      "step": 29410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.40964855093937e-06,
+      "loss": 0.8241,
+      "step": 29420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.39483198008653e-06,
+      "loss": 0.8221,
+      "step": 29430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.380015409233687e-06,
+      "loss": 0.8893,
+      "step": 29440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.365198838380845e-06,
+      "loss": 0.8828,
+      "step": 29450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.3503822675280034e-06,
+      "loss": 0.8587,
+      "step": 29460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.335565696675162e-06,
+      "loss": 0.7406,
+      "step": 29470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.3207491258223204e-06,
+      "loss": 0.7206,
+      "step": 29480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.305932554969478e-06,
+      "loss": 0.8015,
+      "step": 29490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.291115984116637e-06,
+      "loss": 0.7798,
+      "step": 29500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.276299413263795e-06,
+      "loss": 0.8116,
+      "step": 29510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.261482842410953e-06,
+      "loss": 0.8226,
+      "step": 29520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.246666271558111e-06,
+      "loss": 0.8893,
+      "step": 29530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.231849700705269e-06,
+      "loss": 0.7837,
+      "step": 29540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.217033129852427e-06,
+      "loss": 0.6093,
+      "step": 29550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.202216558999585e-06,
+      "loss": 0.789,
+      "step": 29560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.187399988146743e-06,
+      "loss": 0.8502,
+      "step": 29570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.172583417293902e-06,
+      "loss": 0.8546,
+      "step": 29580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.15776684644106e-06,
+      "loss": 0.9091,
+      "step": 29590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.142950275588218e-06,
+      "loss": 0.7525,
+      "step": 29600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.128133704735376e-06,
+      "loss": 0.8636,
+      "step": 29610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.113317133882534e-06,
+      "loss": 0.7831,
+      "step": 29620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0985005630296925e-06,
+      "loss": 0.833,
+      "step": 29630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0836839921768506e-06,
+      "loss": 0.8562,
+      "step": 29640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0688674213240095e-06,
+      "loss": 0.7481,
+      "step": 29650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.054050850471167e-06,
+      "loss": 0.8869,
+      "step": 29660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.039234279618326e-06,
+      "loss": 0.6972,
+      "step": 29670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.024417708765484e-06,
+      "loss": 0.8546,
+      "step": 29680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.009601137912642e-06,
+      "loss": 0.7603,
+      "step": 29690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9947845670598e-06,
+      "loss": 0.8197,
+      "step": 29700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.979967996206958e-06,
+      "loss": 0.73,
+      "step": 29710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.965151425354117e-06,
+      "loss": 0.7717,
+      "step": 29720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.950334854501274e-06,
+      "loss": 0.6476,
+      "step": 29730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.935518283648433e-06,
+      "loss": 0.8989,
+      "step": 29740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.92070171279559e-06,
+      "loss": 0.9831,
+      "step": 29750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.905885141942749e-06,
+      "loss": 0.9013,
+      "step": 29760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.891068571089907e-06,
+      "loss": 0.7805,
+      "step": 29770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.876252000237065e-06,
+      "loss": 0.7722,
+      "step": 29780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8614354293842235e-06,
+      "loss": 0.8622,
+      "step": 29790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8466188585313815e-06,
+      "loss": 0.7649,
+      "step": 29800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8318022876785405e-06,
+      "loss": 0.867,
+      "step": 29810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.816985716825698e-06,
+      "loss": 0.8248,
+      "step": 29820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.802169145972857e-06,
+      "loss": 0.8189,
+      "step": 29830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.787352575120014e-06,
+      "loss": 0.7743,
+      "step": 29840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.772536004267173e-06,
+      "loss": 0.8254,
+      "step": 29850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.757719433414331e-06,
+      "loss": 0.8056,
+      "step": 29860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.742902862561489e-06,
+      "loss": 0.7944,
+      "step": 29870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.728086291708647e-06,
+      "loss": 0.9064,
+      "step": 29880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.713269720855805e-06,
+      "loss": 0.8732,
+      "step": 29890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.698453150002964e-06,
+      "loss": 0.804,
+      "step": 29900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.683636579150121e-06,
+      "loss": 0.9746,
+      "step": 29910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.66882000829728e-06,
+      "loss": 0.8479,
+      "step": 29920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.654003437444438e-06,
+      "loss": 0.8559,
+      "step": 29930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.639186866591596e-06,
+      "loss": 0.8752,
+      "step": 29940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.624370295738754e-06,
+      "loss": 0.7864,
+      "step": 29950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.6095537248859125e-06,
+      "loss": 0.8694,
+      "step": 29960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.594737154033071e-06,
+      "loss": 0.8,
+      "step": 29970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.579920583180229e-06,
+      "loss": 0.8002,
+      "step": 29980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.565104012327388e-06,
+      "loss": 0.895,
+      "step": 29990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.550287441474546e-06,
+      "loss": 0.7581,
+      "step": 30000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.535470870621704e-06,
+      "loss": 0.9158,
+      "step": 30010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.520654299768862e-06,
+      "loss": 0.9058,
+      "step": 30020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.50583772891602e-06,
+      "loss": 0.8668,
+      "step": 30030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.491021158063178e-06,
+      "loss": 0.7331,
+      "step": 30040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.476204587210336e-06,
+      "loss": 0.8737,
+      "step": 30050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.461388016357494e-06,
+      "loss": 0.7617,
+      "step": 30060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.446571445504653e-06,
+      "loss": 0.8632,
+      "step": 30070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.43175487465181e-06,
+      "loss": 0.8363,
+      "step": 30080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.416938303798969e-06,
+      "loss": 0.8232,
+      "step": 30090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.402121732946127e-06,
+      "loss": 0.8688,
+      "step": 30100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.387305162093285e-06,
+      "loss": 0.7766,
+      "step": 30110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3724885912404435e-06,
+      "loss": 0.8947,
+      "step": 30120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3576720203876015e-06,
+      "loss": 0.6847,
+      "step": 30130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3428554495347605e-06,
+      "loss": 0.7965,
+      "step": 30140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.328038878681918e-06,
+      "loss": 0.8281,
+      "step": 30150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.313222307829077e-06,
+      "loss": 1.0558,
+      "step": 30160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.298405736976234e-06,
+      "loss": 0.8634,
+      "step": 30170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.283589166123393e-06,
+      "loss": 0.7995,
+      "step": 30180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.268772595270551e-06,
+      "loss": 0.7261,
+      "step": 30190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.253956024417709e-06,
+      "loss": 0.8498,
+      "step": 30200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.239139453564868e-06,
+      "loss": 0.8425,
+      "step": 30210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.224322882712025e-06,
+      "loss": 0.926,
+      "step": 30220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.209506311859184e-06,
+      "loss": 0.726,
+      "step": 30230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.194689741006341e-06,
+      "loss": 0.8751,
+      "step": 30240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1798731701535e-06,
+      "loss": 0.8552,
+      "step": 30250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.165056599300657e-06,
+      "loss": 0.7732,
+      "step": 30260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.150240028447816e-06,
+      "loss": 0.8612,
+      "step": 30270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1354234575949744e-06,
+      "loss": 0.685,
+      "step": 30280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1206068867421325e-06,
+      "loss": 0.799,
+      "step": 30290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.105790315889291e-06,
+      "loss": 0.74,
+      "step": 30300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.090973745036449e-06,
+      "loss": 0.8859,
+      "step": 30310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.076157174183608e-06,
+      "loss": 0.7469,
+      "step": 30320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.061340603330765e-06,
+      "loss": 0.7738,
+      "step": 30330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.046524032477924e-06,
+      "loss": 0.8269,
+      "step": 30340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.031707461625082e-06,
+      "loss": 0.9851,
+      "step": 30350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.01689089077224e-06,
+      "loss": 0.7928,
+      "step": 30360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.002074319919398e-06,
+      "loss": 0.7639,
+      "step": 30370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.987257749066556e-06,
+      "loss": 0.9153,
+      "step": 30380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.972441178213714e-06,
+      "loss": 0.9317,
+      "step": 30390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.957624607360872e-06,
+      "loss": 0.8111,
+      "step": 30400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.942808036508031e-06,
+      "loss": 0.8389,
+      "step": 30410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.927991465655189e-06,
+      "loss": 0.8964,
+      "step": 30420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.913174894802347e-06,
+      "loss": 0.729,
+      "step": 30430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898358323949505e-06,
+      "loss": 0.8435,
+      "step": 30440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8835417530966635e-06,
+      "loss": 0.6626,
+      "step": 30450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8687251822438216e-06,
+      "loss": 0.7776,
+      "step": 30460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.85390861139098e-06,
+      "loss": 0.731,
+      "step": 30470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.839092040538138e-06,
+      "loss": 0.7716,
+      "step": 30480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.824275469685297e-06,
+      "loss": 0.8201,
+      "step": 30490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.809458898832455e-06,
+      "loss": 0.8544,
+      "step": 30500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.794642327979613e-06,
+      "loss": 0.8127,
+      "step": 30510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.779825757126771e-06,
+      "loss": 0.7945,
+      "step": 30520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.765009186273929e-06,
+      "loss": 0.7658,
+      "step": 30530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.750192615421087e-06,
+      "loss": 0.709,
+      "step": 30540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.735376044568245e-06,
+      "loss": 0.7834,
+      "step": 30550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.720559473715404e-06,
+      "loss": 0.7525,
+      "step": 30560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705742902862561e-06,
+      "loss": 0.8995,
+      "step": 30570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.69092633200972e-06,
+      "loss": 0.6898,
+      "step": 30580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6761097611568774e-06,
+      "loss": 0.7307,
+      "step": 30590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.661293190304036e-06,
+      "loss": 0.764,
+      "step": 30600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6464766194511944e-06,
+      "loss": 0.7606,
+      "step": 30610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6316600485983525e-06,
+      "loss": 0.8327,
+      "step": 30620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6168434777455114e-06,
+      "loss": 0.7644,
+      "step": 30630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.602026906892669e-06,
+      "loss": 0.6861,
+      "step": 30640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.587210336039828e-06,
+      "loss": 0.7229,
+      "step": 30650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.572393765186985e-06,
+      "loss": 0.79,
+      "step": 30660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.557577194334144e-06,
+      "loss": 0.9027,
+      "step": 30670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.542760623481301e-06,
+      "loss": 0.8741,
+      "step": 30680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.52794405262846e-06,
+      "loss": 0.9217,
+      "step": 30690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.513127481775618e-06,
+      "loss": 0.8175,
+      "step": 30700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.498310910922776e-06,
+      "loss": 0.7815,
+      "step": 30710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.483494340069935e-06,
+      "loss": 0.7731,
+      "step": 30720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.468677769217092e-06,
+      "loss": 0.7689,
+      "step": 30730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.453861198364251e-06,
+      "loss": 0.8046,
+      "step": 30740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.439044627511408e-06,
+      "loss": 0.8386,
+      "step": 30750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.424228056658567e-06,
+      "loss": 0.7341,
+      "step": 30760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.409411485805725e-06,
+      "loss": 0.9079,
+      "step": 30770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3945949149528835e-06,
+      "loss": 0.8967,
+      "step": 30780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3797783441000416e-06,
+      "loss": 1.0217,
+      "step": 30790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3649617732472e-06,
+      "loss": 0.7746,
+      "step": 30800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.350145202394358e-06,
+      "loss": 0.7299,
+      "step": 30810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.335328631541516e-06,
+      "loss": 0.8941,
+      "step": 30820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.320512060688675e-06,
+      "loss": 0.812,
+      "step": 30830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.305695489835833e-06,
+      "loss": 0.9509,
+      "step": 30840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.290878918982991e-06,
+      "loss": 0.779,
+      "step": 30850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.276062348130149e-06,
+      "loss": 0.8534,
+      "step": 30860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.261245777277307e-06,
+      "loss": 0.6726,
+      "step": 30870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.246429206424465e-06,
+      "loss": 0.8623,
+      "step": 30880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.231612635571623e-06,
+      "loss": 0.7388,
+      "step": 30890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.216796064718781e-06,
+      "loss": 0.7825,
+      "step": 30900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.20197949386594e-06,
+      "loss": 0.719,
+      "step": 30910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.187162923013098e-06,
+      "loss": 0.9652,
+      "step": 30920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.172346352160256e-06,
+      "loss": 0.8504,
+      "step": 30930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1575297813074144e-06,
+      "loss": 0.809,
+      "step": 30940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1427132104545725e-06,
+      "loss": 0.857,
+      "step": 30950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.127896639601731e-06,
+      "loss": 0.7532,
+      "step": 30960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.113080068748889e-06,
+      "loss": 0.861,
+      "step": 30970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.098263497896048e-06,
+      "loss": 0.8542,
+      "step": 30980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.083446927043205e-06,
+      "loss": 0.707,
+      "step": 30990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.068630356190364e-06,
+      "loss": 0.7639,
+      "step": 31000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.053813785337522e-06,
+      "loss": 0.8398,
+      "step": 31010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.03899721448468e-06,
+      "loss": 0.846,
+      "step": 31020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.024180643631838e-06,
+      "loss": 0.88,
+      "step": 31030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.009364072778996e-06,
+      "loss": 0.8333,
+      "step": 31040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.994547501926155e-06,
+      "loss": 0.8985,
+      "step": 31050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.979730931073312e-06,
+      "loss": 0.7808,
+      "step": 31060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.964914360220471e-06,
+      "loss": 0.9343,
+      "step": 31070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.950097789367628e-06,
+      "loss": 0.8095,
+      "step": 31080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.935281218514787e-06,
+      "loss": 0.9956,
+      "step": 31090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.920464647661945e-06,
+      "loss": 0.8299,
+      "step": 31100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9056480768091035e-06,
+      "loss": 0.8431,
+      "step": 31110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8908315059562616e-06,
+      "loss": 0.8308,
+      "step": 31120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.87601493510342e-06,
+      "loss": 0.7654,
+      "step": 31130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.861198364250579e-06,
+      "loss": 0.8367,
+      "step": 31140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.846381793397736e-06,
+      "loss": 0.8608,
+      "step": 31150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.831565222544895e-06,
+      "loss": 0.8565,
+      "step": 31160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.816748651692052e-06,
+      "loss": 0.7515,
+      "step": 31170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.801932080839211e-06,
+      "loss": 0.7589,
+      "step": 31180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7871155099863694e-06,
+      "loss": 0.7388,
+      "step": 31190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.772298939133527e-06,
+      "loss": 0.8201,
+      "step": 31200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7574823682806856e-06,
+      "loss": 0.7638,
+      "step": 31210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7426657974278432e-06,
+      "loss": 0.8043,
+      "step": 31220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7278492265750017e-06,
+      "loss": 0.747,
+      "step": 31230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7130326557221594e-06,
+      "loss": 0.7445,
+      "step": 31240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.698216084869318e-06,
+      "loss": 0.8245,
+      "step": 31250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6833995140164764e-06,
+      "loss": 0.8488,
+      "step": 31260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6685829431636345e-06,
+      "loss": 0.9464,
+      "step": 31270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6537663723107925e-06,
+      "loss": 0.7752,
+      "step": 31280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6389498014579506e-06,
+      "loss": 0.821,
+      "step": 31290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.624133230605109e-06,
+      "loss": 0.8872,
+      "step": 31300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6093166597522668e-06,
+      "loss": 0.8284,
+      "step": 31310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5945000888994253e-06,
+      "loss": 0.9218,
+      "step": 31320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5796835180465838e-06,
+      "loss": 0.7825,
+      "step": 31330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5648669471937414e-06,
+      "loss": 0.8917,
+      "step": 31340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5500503763409e-06,
+      "loss": 0.6889,
+      "step": 31350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5352338054880576e-06,
+      "loss": 0.8553,
+      "step": 31360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.520417234635216e-06,
+      "loss": 0.858,
+      "step": 31370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.505600663782374e-06,
+      "loss": 0.7792,
+      "step": 31380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4907840929295327e-06,
+      "loss": 0.7633,
+      "step": 31390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.475967522076691e-06,
+      "loss": 0.7359,
+      "step": 31400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.461150951223849e-06,
+      "loss": 0.9595,
+      "step": 31410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4463343803710073e-06,
+      "loss": 0.8637,
+      "step": 31420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.431517809518165e-06,
+      "loss": 0.8134,
+      "step": 31430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4167012386653235e-06,
+      "loss": 0.8257,
+      "step": 31440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.401884667812481e-06,
+      "loss": 0.8745,
+      "step": 31450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3870680969596397e-06,
+      "loss": 0.8975,
+      "step": 31460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.372251526106798e-06,
+      "loss": 0.849,
+      "step": 31470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3574349552539562e-06,
+      "loss": 0.8976,
+      "step": 31480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3426183844011147e-06,
+      "loss": 0.7464,
+      "step": 31490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3278018135482724e-06,
+      "loss": 0.7268,
+      "step": 31500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.312985242695431e-06,
+      "loss": 0.8978,
+      "step": 31510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2981686718425886e-06,
+      "loss": 0.9268,
+      "step": 31520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.283352100989747e-06,
+      "loss": 0.8545,
+      "step": 31530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2685355301369056e-06,
+      "loss": 0.8155,
+      "step": 31540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2537189592840632e-06,
+      "loss": 0.7821,
+      "step": 31550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2389023884312217e-06,
+      "loss": 0.8581,
+      "step": 31560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.22408581757838e-06,
+      "loss": 0.7314,
+      "step": 31570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.209269246725538e-06,
+      "loss": 0.7354,
+      "step": 31580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.194452675872696e-06,
+      "loss": 0.7087,
+      "step": 31590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1796361050198545e-06,
+      "loss": 0.8197,
+      "step": 31600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.164819534167013e-06,
+      "loss": 0.8743,
+      "step": 31610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1500029633141706e-06,
+      "loss": 0.8032,
+      "step": 31620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.135186392461329e-06,
+      "loss": 0.8348,
+      "step": 31630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.120369821608487e-06,
+      "loss": 1.0197,
+      "step": 31640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1055532507556453e-06,
+      "loss": 0.8379,
+      "step": 31650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0907366799028034e-06,
+      "loss": 0.9561,
+      "step": 31660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0759201090499614e-06,
+      "loss": 0.745,
+      "step": 31670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.06110353819712e-06,
+      "loss": 0.8192,
+      "step": 31680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.046286967344278e-06,
+      "loss": 0.7573,
+      "step": 31690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.031470396491436e-06,
+      "loss": 0.7973,
+      "step": 31700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0166538256385946e-06,
+      "loss": 0.8163,
+      "step": 31710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0018372547857527e-06,
+      "loss": 0.6931,
+      "step": 31720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9870206839329108e-06,
+      "loss": 0.7237,
+      "step": 31730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.972204113080069e-06,
+      "loss": 0.8131,
+      "step": 31740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.957387542227227e-06,
+      "loss": 0.7158,
+      "step": 31750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.942570971374385e-06,
+      "loss": 0.9403,
+      "step": 31760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.927754400521543e-06,
+      "loss": 0.9201,
+      "step": 31770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9129378296687016e-06,
+      "loss": 0.9178,
+      "step": 31780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.89812125881586e-06,
+      "loss": 0.9062,
+      "step": 31790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.883304687963018e-06,
+      "loss": 0.8674,
+      "step": 31800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8684881171101763e-06,
+      "loss": 0.7798,
+      "step": 31810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8536715462573343e-06,
+      "loss": 0.7193,
+      "step": 31820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8388549754044924e-06,
+      "loss": 0.7906,
+      "step": 31830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8240384045516505e-06,
+      "loss": 0.8077,
+      "step": 31840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.809221833698809e-06,
+      "loss": 0.7481,
+      "step": 31850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.794405262845967e-06,
+      "loss": 0.8428,
+      "step": 31860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.779588691993125e-06,
+      "loss": 0.7441,
+      "step": 31870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7647721211402832e-06,
+      "loss": 0.7401,
+      "step": 31880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7499555502874417e-06,
+      "loss": 0.869,
+      "step": 31890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7351389794346e-06,
+      "loss": 0.8421,
+      "step": 31900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.720322408581758e-06,
+      "loss": 0.783,
+      "step": 31910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7055058377289164e-06,
+      "loss": 0.8691,
+      "step": 31920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6906892668760745e-06,
+      "loss": 0.6008,
+      "step": 31930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6758726960232326e-06,
+      "loss": 0.8923,
+      "step": 31940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6610561251703906e-06,
+      "loss": 0.8449,
+      "step": 31950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6462395543175487e-06,
+      "loss": 0.7947,
+      "step": 31960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.631422983464707e-06,
+      "loss": 0.7598,
+      "step": 31970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.616606412611865e-06,
+      "loss": 0.8125,
+      "step": 31980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6017898417590234e-06,
+      "loss": 0.8404,
+      "step": 31990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.586973270906182e-06,
+      "loss": 0.8872,
+      "step": 32000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.57215670005334e-06,
+      "loss": 0.9069,
+      "step": 32010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.557340129200498e-06,
+      "loss": 0.7658,
+      "step": 32020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.542523558347656e-06,
+      "loss": 0.9541,
+      "step": 32030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.527706987494814e-06,
+      "loss": 0.8887,
+      "step": 32040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5128904166419723e-06,
+      "loss": 0.9671,
+      "step": 32050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4980738457891308e-06,
+      "loss": 0.8185,
+      "step": 32060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.483257274936289e-06,
+      "loss": 0.8478,
+      "step": 32070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.468440704083447e-06,
+      "loss": 0.7373,
+      "step": 32080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.453624133230605e-06,
+      "loss": 0.7813,
+      "step": 32090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4388075623777635e-06,
+      "loss": 0.8227,
+      "step": 32100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4239909915249216e-06,
+      "loss": 0.734,
+      "step": 32110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4091744206720797e-06,
+      "loss": 0.8542,
+      "step": 32120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.394357849819238e-06,
+      "loss": 0.7181,
+      "step": 32130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3795412789663963e-06,
+      "loss": 0.8298,
+      "step": 32140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3647247081135543e-06,
+      "loss": 0.8034,
+      "step": 32150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3499081372607124e-06,
+      "loss": 0.8093,
+      "step": 32160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3350915664078705e-06,
+      "loss": 0.8805,
+      "step": 32170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3202749955550286e-06,
+      "loss": 0.8055,
+      "step": 32180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.305458424702187e-06,
+      "loss": 0.8558,
+      "step": 32190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.290641853849345e-06,
+      "loss": 0.8407,
+      "step": 32200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2758252829965037e-06,
+      "loss": 0.9348,
+      "step": 32210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2610087121436617e-06,
+      "loss": 0.779,
+      "step": 32220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.24619214129082e-06,
+      "loss": 0.6843,
+      "step": 32230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.231375570437978e-06,
+      "loss": 0.8221,
+      "step": 32240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.216558999585136e-06,
+      "loss": 0.8315,
+      "step": 32250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.201742428732294e-06,
+      "loss": 0.7218,
+      "step": 32260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1869258578794526e-06,
+      "loss": 0.9209,
+      "step": 32270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1721092870266106e-06,
+      "loss": 0.8556,
+      "step": 32280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1572927161737687e-06,
+      "loss": 0.9266,
+      "step": 32290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1424761453209272e-06,
+      "loss": 0.8504,
+      "step": 32300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1276595744680853e-06,
+      "loss": 0.7773,
+      "step": 32310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1128430036152434e-06,
+      "loss": 0.878,
+      "step": 32320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0980264327624015e-06,
+      "loss": 0.8129,
+      "step": 32330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08320986190956e-06,
+      "loss": 0.8973,
+      "step": 32340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.068393291056718e-06,
+      "loss": 0.9417,
+      "step": 32350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.053576720203876e-06,
+      "loss": 0.7483,
+      "step": 32360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.038760149351034e-06,
+      "loss": 0.6655,
+      "step": 32370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0239435784981923e-06,
+      "loss": 0.9379,
+      "step": 32380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0091270076453504e-06,
+      "loss": 0.8425,
+      "step": 32390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.994310436792509e-06,
+      "loss": 0.827,
+      "step": 32400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9794938659396674e-06,
+      "loss": 0.8824,
+      "step": 32410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9646772950868255e-06,
+      "loss": 0.8184,
+      "step": 32420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9498607242339835e-06,
+      "loss": 0.7644,
+      "step": 32430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9350441533811416e-06,
+      "loss": 0.8675,
+      "step": 32440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9202275825282997e-06,
+      "loss": 0.787,
+      "step": 32450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9054110116754578e-06,
+      "loss": 0.8345,
+      "step": 32460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.890594440822616e-06,
+      "loss": 0.8468,
+      "step": 32470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8757778699697744e-06,
+      "loss": 0.7413,
+      "step": 32480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8609612991169326e-06,
+      "loss": 0.7738,
+      "step": 32490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8461447282640907e-06,
+      "loss": 0.7922,
+      "step": 32500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8313281574112488e-06,
+      "loss": 0.8119,
+      "step": 32510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8165115865584069e-06,
+      "loss": 0.8329,
+      "step": 32520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8016950157055652e-06,
+      "loss": 0.6763,
+      "step": 32530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7868784448527233e-06,
+      "loss": 0.6897,
+      "step": 32540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7720618739998818e-06,
+      "loss": 0.854,
+      "step": 32550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7572453031470398e-06,
+      "loss": 0.775,
+      "step": 32560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.742428732294198e-06,
+      "loss": 0.9592,
+      "step": 32570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.727612161441356e-06,
+      "loss": 0.7021,
+      "step": 32580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7127955905885143e-06,
+      "loss": 0.8669,
+      "step": 32590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6979790197356724e-06,
+      "loss": 0.9422,
+      "step": 32600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6831624488828304e-06,
+      "loss": 0.7729,
+      "step": 32610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.668345878029989e-06,
+      "loss": 0.8819,
+      "step": 32620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.653529307177147e-06,
+      "loss": 0.8327,
+      "step": 32630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6387127363243053e-06,
+      "loss": 0.7456,
+      "step": 32640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6238961654714634e-06,
+      "loss": 0.8397,
+      "step": 32650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6090795946186215e-06,
+      "loss": 0.8461,
+      "step": 32660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5942630237657796e-06,
+      "loss": 0.8236,
+      "step": 32670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5794464529129378e-06,
+      "loss": 0.9109,
+      "step": 32680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5646298820600961e-06,
+      "loss": 0.7619,
+      "step": 32690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5498133112072542e-06,
+      "loss": 0.7171,
+      "step": 32700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5349967403544125e-06,
+      "loss": 0.7824,
+      "step": 32710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5201801695015706e-06,
+      "loss": 0.8839,
+      "step": 32720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5053635986487287e-06,
+      "loss": 0.7877,
+      "step": 32730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4905470277958872e-06,
+      "loss": 0.7949,
+      "step": 32740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4757304569430452e-06,
+      "loss": 0.7999,
+      "step": 32750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4609138860902033e-06,
+      "loss": 0.8508,
+      "step": 32760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4460973152373614e-06,
+      "loss": 0.8006,
+      "step": 32770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4312807443845197e-06,
+      "loss": 0.8227,
+      "step": 32780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.416464173531678e-06,
+      "loss": 0.9299,
+      "step": 32790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.401647602678836e-06,
+      "loss": 0.8562,
+      "step": 32800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3868310318259944e-06,
+      "loss": 0.8914,
+      "step": 32810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3720144609731524e-06,
+      "loss": 0.8771,
+      "step": 32820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3571978901203105e-06,
+      "loss": 0.8683,
+      "step": 32830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3423813192674688e-06,
+      "loss": 0.8071,
+      "step": 32840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.327564748414627e-06,
+      "loss": 0.7712,
+      "step": 32850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3127481775617852e-06,
+      "loss": 0.6798,
+      "step": 32860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2979316067089433e-06,
+      "loss": 0.863,
+      "step": 32870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2831150358561016e-06,
+      "loss": 0.879,
+      "step": 32880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2682984650032596e-06,
+      "loss": 0.9034,
+      "step": 32890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.253481894150418e-06,
+      "loss": 0.8587,
+      "step": 32900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.238665323297576e-06,
+      "loss": 0.773,
+      "step": 32910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2238487524447343e-06,
+      "loss": 0.8008,
+      "step": 32920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2090321815918924e-06,
+      "loss": 0.7931,
+      "step": 32930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1942156107390507e-06,
+      "loss": 0.7081,
+      "step": 32940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.179399039886209e-06,
+      "loss": 0.808,
+      "step": 32950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.164582469033367e-06,
+      "loss": 0.8978,
+      "step": 32960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1497658981805251e-06,
+      "loss": 0.8346,
+      "step": 32970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1349493273276832e-06,
+      "loss": 0.7151,
+      "step": 32980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1201327564748415e-06,
+      "loss": 0.6956,
+      "step": 32990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1053161856219998e-06,
+      "loss": 0.8311,
+      "step": 33000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 33746,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 4.97334915956736e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}