diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,55102 @@
+{
+  "best_metric": 3.0536694526672363,
+  "best_model_checkpoint": "/mnt/dolphinfs/hdd_pool/docker/user/hadoop-aipnlp/zengweihao02/SereTOD/SereTOD/SereTOD2022-main/Track2/response/pretrain/all_KB_len_1024_lr_2e_5_bsz_192_seq2seq_1B_ddp_50epoch_warmup_0.1_submit_all_KB_2u/checkpoint-30504",
+  "epoch": 18.0,
+  "global_step": 91512,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9669551534225022e-08,
+      "loss": 9.2657,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9339103068450045e-08,
+      "loss": 9.1388,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.900865460267507e-08,
+      "loss": 9.0747,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.867820613690009e-08,
+      "loss": 8.9678,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.83477576711251e-08,
+      "loss": 8.8178,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1801730920535014e-07,
+      "loss": 8.7617,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3768686073957515e-07,
+      "loss": 8.455,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5735641227380018e-07,
+      "loss": 8.3642,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.770259638080252e-07,
+      "loss": 8.2783,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.966955153422502e-07,
+      "loss": 8.1006,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1636506687647525e-07,
+      "loss": 7.978,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3603461841070028e-07,
+      "loss": 7.8846,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5570416994492526e-07,
+      "loss": 7.7678,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.753737214791503e-07,
+      "loss": 7.6845,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9504327301337533e-07,
+      "loss": 7.6073,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1471282454760036e-07,
+      "loss": 7.4981,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.343823760818254e-07,
+      "loss": 7.3463,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.540519276160504e-07,
+      "loss": 7.3023,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7372147915027535e-07,
+      "loss": 7.2999,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.933910306845004e-07,
+      "loss": 7.1877,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1306058221872547e-07,
+      "loss": 7.0913,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.327301337529505e-07,
+      "loss": 7.1055,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5239968528717553e-07,
+      "loss": 6.9057,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7206923682140056e-07,
+      "loss": 6.8393,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917387883556255e-07,
+      "loss": 6.8125,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.114083398898505e-07,
+      "loss": 6.7601,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.310778914240756e-07,
+      "loss": 6.6865,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.507474429583006e-07,
+      "loss": 6.6376,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.704169944925256e-07,
+      "loss": 6.5192,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.900865460267507e-07,
+      "loss": 6.3433,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.097560975609757e-07,
+      "loss": 6.2774,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.294256490952007e-07,
+      "loss": 6.2367,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.490952006294258e-07,
+      "loss": 6.1516,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.687647521636508e-07,
+      "loss": 6.1676,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.884343036978758e-07,
+      "loss": 6.0194,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.081038552321008e-07,
+      "loss": 5.9296,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.277734067663259e-07,
+      "loss": 5.873,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.474429583005507e-07,
+      "loss": 5.8485,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.671125098347757e-07,
+      "loss": 5.7634,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.867820613690008e-07,
+      "loss": 5.6994,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.064516129032258e-07,
+      "loss": 5.6927,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.261211644374509e-07,
+      "loss": 5.6613,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.45790715971676e-07,
+      "loss": 5.6097,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.65460267505901e-07,
+      "loss": 5.493,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.85129819040126e-07,
+      "loss": 5.4451,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.047993705743511e-07,
+      "loss": 5.3156,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.244689221085761e-07,
+      "loss": 5.2571,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.441384736428011e-07,
+      "loss": 5.1619,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.63808025177026e-07,
+      "loss": 5.1054,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.83477576711251e-07,
+      "loss": 5.0413,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0031471282454761e-06,
+      "loss": 4.9388,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.022816679779701e-06,
+      "loss": 4.9707,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0424862313139262e-06,
+      "loss": 4.8535,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0621557828481511e-06,
+      "loss": 4.8432,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0818253343823763e-06,
+      "loss": 4.7689,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1014948859166012e-06,
+      "loss": 4.7523,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1211644374508263e-06,
+      "loss": 4.7476,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1408339889850512e-06,
+      "loss": 4.694,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1605035405192762e-06,
+      "loss": 4.7041,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1801730920535013e-06,
+      "loss": 4.6129,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1998426435877262e-06,
+      "loss": 4.5952,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 4.6299,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2391817466561763e-06,
+      "loss": 4.5753,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2588512981904014e-06,
+      "loss": 4.4968,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2785208497246262e-06,
+      "loss": 4.5061,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2981904012588515e-06,
+      "loss": 4.421,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3178599527930764e-06,
+      "loss": 4.4725,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3375295043273016e-06,
+      "loss": 4.4675,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3571990558615265e-06,
+      "loss": 4.4363,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3768686073957516e-06,
+      "loss": 4.3907,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3965381589299766e-06,
+      "loss": 4.4394,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4162077104642017e-06,
+      "loss": 4.3606,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4358772619984266e-06,
+      "loss": 4.3973,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4555468135326518e-06,
+      "loss": 4.3232,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4752163650668767e-06,
+      "loss": 4.3427,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4948859166011014e-06,
+      "loss": 4.3518,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5145554681353268e-06,
+      "loss": 4.2749,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5342250196695515e-06,
+      "loss": 4.3374,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5538945712037768e-06,
+      "loss": 4.2984,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5735641227380015e-06,
+      "loss": 4.2897,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5932336742722269e-06,
+      "loss": 4.2405,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 4.2871,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.632572777340677e-06,
+      "loss": 4.2565,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6522423288749019e-06,
+      "loss": 4.2081,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.671911880409127e-06,
+      "loss": 4.2068,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.691581431943352e-06,
+      "loss": 4.1673,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7112509834775767e-06,
+      "loss": 4.2585,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.730920535011802e-06,
+      "loss": 4.1445,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7505900865460267e-06,
+      "loss": 4.2064,
+      "step": 890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.770259638080252e-06,
+      "loss": 4.1744,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7899291896144768e-06,
+      "loss": 4.2387,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8095987411487021e-06,
+      "loss": 4.1399,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8292682926829268e-06,
+      "loss": 4.1244,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8489378442171522e-06,
+      "loss": 4.1568,
+      "step": 940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868607395751377e-06,
+      "loss": 4.154,
+      "step": 950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8882769472856023e-06,
+      "loss": 4.1719,
+      "step": 960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.907946498819827e-06,
+      "loss": 4.104,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.927616050354052e-06,
+      "loss": 4.1177,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9472856018882773e-06,
+      "loss": 4.0864,
+      "step": 990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.966955153422502e-06,
+      "loss": 4.066,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.986624704956727e-06,
+      "loss": 4.0887,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0062942564909522e-06,
+      "loss": 4.0719,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0259638080251774e-06,
+      "loss": 4.0848,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045633359559402e-06,
+      "loss": 4.0547,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0653029110936272e-06,
+      "loss": 4.0397,
+      "step": 1050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0849724626278524e-06,
+      "loss": 4.0075,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.104642014162077e-06,
+      "loss": 4.0453,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1243115656963022e-06,
+      "loss": 4.11,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1439811172305274e-06,
+      "loss": 3.9867,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1636506687647525e-06,
+      "loss": 4.0428,
+      "step": 1100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1833202202989772e-06,
+      "loss": 4.0572,
+      "step": 1110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2029897718332024e-06,
+      "loss": 4.038,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2226593233674275e-06,
+      "loss": 4.0128,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2423288749016526e-06,
+      "loss": 4.0762,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2619984264358773e-06,
+      "loss": 4.0138,
+      "step": 1150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.2816679779701025e-06,
+      "loss": 4.0361,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3013375295043276e-06,
+      "loss": 3.9722,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3210070810385523e-06,
+      "loss": 4.0026,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3406766325727775e-06,
+      "loss": 3.9912,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3603461841070026e-06,
+      "loss": 3.9694,
+      "step": 1200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3800157356412278e-06,
+      "loss": 3.9854,
+      "step": 1210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3996852871754525e-06,
+      "loss": 4.0109,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 3.9949,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 3.9758,
+      "step": 1240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.458693941778128e-06,
+      "loss": 3.9899,
+      "step": 1250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4783634933123526e-06,
+      "loss": 3.9922,
+      "step": 1260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4980330448465777e-06,
+      "loss": 3.947,
+      "step": 1270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.517702596380803e-06,
+      "loss": 3.9814,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.5373721479150276e-06,
+      "loss": 3.9613,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5570416994492523e-06,
+      "loss": 4.0047,
+      "step": 1300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.576711250983478e-06,
+      "loss": 3.9313,
+      "step": 1310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.596380802517703e-06,
+      "loss": 3.9556,
+      "step": 1320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6160503540519277e-06,
+      "loss": 3.9642,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.635719905586153e-06,
+      "loss": 3.9314,
+      "step": 1340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.655389457120378e-06,
+      "loss": 3.9356,
+      "step": 1350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.675059008654603e-06,
+      "loss": 3.9244,
+      "step": 1360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.694728560188828e-06,
+      "loss": 3.8964,
+      "step": 1370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.714398111723053e-06,
+      "loss": 3.973,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7340676632572777e-06,
+      "loss": 3.9137,
+      "step": 1390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7537372147915033e-06,
+      "loss": 3.9207,
+      "step": 1400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.773406766325728e-06,
+      "loss": 3.8746,
+      "step": 1410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.793076317859953e-06,
+      "loss": 3.8698,
+      "step": 1420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.812745869394178e-06,
+      "loss": 3.9087,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8324154209284034e-06,
+      "loss": 3.8707,
+      "step": 1440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.852084972462628e-06,
+      "loss": 3.9122,
+      "step": 1450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8717545239968532e-06,
+      "loss": 3.8523,
+      "step": 1460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.891424075531078e-06,
+      "loss": 3.9097,
+      "step": 1470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9110936270653035e-06,
+      "loss": 3.8648,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9307631785995282e-06,
+      "loss": 3.8786,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9504327301337534e-06,
+      "loss": 3.9296,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.970102281667978e-06,
+      "loss": 3.9163,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.989771833202203e-06,
+      "loss": 3.8863,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0094413847364284e-06,
+      "loss": 3.9165,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0291109362706535e-06,
+      "loss": 3.8519,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.0487804878048782e-06,
+      "loss": 3.8395,
+      "step": 1550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.068450039339103e-06,
+      "loss": 3.7775,
+      "step": 1560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.0881195908733285e-06,
+      "loss": 3.85,
+      "step": 1570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1077891424075536e-06,
+      "loss": 3.8537,
+      "step": 1580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.1274586939417783e-06,
+      "loss": 3.8785,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.147128245476003e-06,
+      "loss": 3.8362,
+      "step": 1600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.166797797010228e-06,
+      "loss": 3.8434,
+      "step": 1610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1864673485444538e-06,
+      "loss": 3.8414,
+      "step": 1620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2061369000786785e-06,
+      "loss": 3.8776,
+      "step": 1630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 3.7993,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2454760031471283e-06,
+      "loss": 3.8442,
+      "step": 1650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.265145554681354e-06,
+      "loss": 3.83,
+      "step": 1660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.2848151062155786e-06,
+      "loss": 3.8087,
+      "step": 1670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3044846577498037e-06,
+      "loss": 3.8019,
+      "step": 1680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3241542092840285e-06,
+      "loss": 3.8287,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.343823760818254e-06,
+      "loss": 3.8088,
+      "step": 1700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3634933123524787e-06,
+      "loss": 3.8469,
+      "step": 1710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.383162863886704e-06,
+      "loss": 3.8237,
+      "step": 1720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4028324154209286e-06,
+      "loss": 3.7912,
+      "step": 1730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4225019669551533e-06,
+      "loss": 3.8102,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.442171518489379e-06,
+      "loss": 3.794,
+      "step": 1750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.461841070023604e-06,
+      "loss": 3.7697,
+      "step": 1760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.4815106215578287e-06,
+      "loss": 3.7966,
+      "step": 1770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5011801730920534e-06,
+      "loss": 3.8054,
+      "step": 1780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.520849724626279e-06,
+      "loss": 3.784,
+      "step": 1790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.540519276160504e-06,
+      "loss": 3.8248,
+      "step": 1800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.560188827694729e-06,
+      "loss": 3.8698,
+      "step": 1810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5798583792289536e-06,
+      "loss": 3.7629,
+      "step": 1820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5995279307631787e-06,
+      "loss": 3.7844,
+      "step": 1830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6191974822974043e-06,
+      "loss": 3.7753,
+      "step": 1840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.638867033831629e-06,
+      "loss": 3.7486,
+      "step": 1850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 3.7569,
+      "step": 1860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.678206136900079e-06,
+      "loss": 3.7559,
+      "step": 1870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6978756884343044e-06,
+      "loss": 3.7439,
+      "step": 1880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.717545239968529e-06,
+      "loss": 3.7771,
+      "step": 1890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.737214791502754e-06,
+      "loss": 3.7389,
+      "step": 1900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.756884343036979e-06,
+      "loss": 3.7672,
+      "step": 1910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7765538945712045e-06,
+      "loss": 3.786,
+      "step": 1920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7962234461054292e-06,
+      "loss": 3.7472,
+      "step": 1930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.815892997639654e-06,
+      "loss": 3.709,
+      "step": 1940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.835562549173879e-06,
+      "loss": 3.7209,
+      "step": 1950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.855232100708104e-06,
+      "loss": 3.767,
+      "step": 1960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.87490165224233e-06,
+      "loss": 3.7491,
+      "step": 1970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8945712037765545e-06,
+      "loss": 3.7415,
+      "step": 1980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.914240755310779e-06,
+      "loss": 3.7236,
+      "step": 1990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.933910306845004e-06,
+      "loss": 3.7032,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9535798583792295e-06,
+      "loss": 3.8092,
+      "step": 2010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.973249409913454e-06,
+      "loss": 3.7298,
+      "step": 2020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.992918961447679e-06,
+      "loss": 3.7428,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0125885129819045e-06,
+      "loss": 3.6921,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 3.7189,
+      "step": 2050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.051927616050355e-06,
+      "loss": 3.7738,
+      "step": 2060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0715971675845795e-06,
+      "loss": 3.7408,
+      "step": 2070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.091266719118804e-06,
+      "loss": 3.7681,
+      "step": 2080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.110936270653029e-06,
+      "loss": 3.8243,
+      "step": 2090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1306058221872545e-06,
+      "loss": 3.7191,
+      "step": 2100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.150275373721479e-06,
+      "loss": 3.7657,
+      "step": 2110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.169944925255705e-06,
+      "loss": 3.771,
+      "step": 2120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1896144767899295e-06,
+      "loss": 3.721,
+      "step": 2130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.209284028324154e-06,
+      "loss": 3.6978,
+      "step": 2140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.22895357985838e-06,
+      "loss": 3.7321,
+      "step": 2150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2486231313926044e-06,
+      "loss": 3.6918,
+      "step": 2160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.268292682926829e-06,
+      "loss": 3.6764,
+      "step": 2170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287962234461055e-06,
+      "loss": 3.6941,
+      "step": 2180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.30763178599528e-06,
+      "loss": 3.6895,
+      "step": 2190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.327301337529505e-06,
+      "loss": 3.7205,
+      "step": 2200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.34697088906373e-06,
+      "loss": 3.702,
+      "step": 2210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3666404405979544e-06,
+      "loss": 3.6531,
+      "step": 2220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.38630999213218e-06,
+      "loss": 3.6964,
+      "step": 2230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.405979543666405e-06,
+      "loss": 3.7065,
+      "step": 2240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.425649095200629e-06,
+      "loss": 3.6752,
+      "step": 2250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.445318646734855e-06,
+      "loss": 3.6683,
+      "step": 2260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.46498819826908e-06,
+      "loss": 3.7344,
+      "step": 2270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.484657749803305e-06,
+      "loss": 3.7057,
+      "step": 2280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.50432730133753e-06,
+      "loss": 3.6546,
+      "step": 2290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.523996852871755e-06,
+      "loss": 3.6802,
+      "step": 2300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.543666404405979e-06,
+      "loss": 3.6627,
+      "step": 2310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.563335955940205e-06,
+      "loss": 3.6907,
+      "step": 2320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.58300550747443e-06,
+      "loss": 3.6717,
+      "step": 2330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.602675059008655e-06,
+      "loss": 3.6795,
+      "step": 2340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.62234461054288e-06,
+      "loss": 3.6535,
+      "step": 2350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.642014162077105e-06,
+      "loss": 3.6423,
+      "step": 2360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.66168371361133e-06,
+      "loss": 3.6411,
+      "step": 2370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.681353265145555e-06,
+      "loss": 3.6351,
+      "step": 2380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.70102281667978e-06,
+      "loss": 3.6791,
+      "step": 2390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.720692368214005e-06,
+      "loss": 3.6282,
+      "step": 2400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.74036191974823e-06,
+      "loss": 3.6498,
+      "step": 2410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7600314712824555e-06,
+      "loss": 3.6583,
+      "step": 2420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.77970102281668e-06,
+      "loss": 3.6926,
+      "step": 2430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.799370574350905e-06,
+      "loss": 3.6803,
+      "step": 2440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8190401258851305e-06,
+      "loss": 3.6405,
+      "step": 2450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 3.671,
+      "step": 2460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.85837922895358e-06,
+      "loss": 3.5974,
+      "step": 2470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 3.6399,
+      "step": 2480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.89771833202203e-06,
+      "loss": 3.7066,
+      "step": 2490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.917387883556256e-06,
+      "loss": 3.6048,
+      "step": 2500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9370574350904805e-06,
+      "loss": 3.6515,
+      "step": 2510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.956726986624705e-06,
+      "loss": 3.6494,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.97639653815893e-06,
+      "loss": 3.6509,
+      "step": 2530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9960660896931555e-06,
+      "loss": 3.6346,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.01573564122738e-06,
+      "loss": 3.6827,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.035405192761606e-06,
+      "loss": 3.6228,
+      "step": 2560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0550747442958305e-06,
+      "loss": 3.6076,
+      "step": 2570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.074744295830055e-06,
+      "loss": 3.6978,
+      "step": 2580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.09441384736428e-06,
+      "loss": 3.6305,
+      "step": 2590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114083398898505e-06,
+      "loss": 3.6758,
+      "step": 2600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.133752950432731e-06,
+      "loss": 3.6434,
+      "step": 2610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.153422501966956e-06,
+      "loss": 3.6553,
+      "step": 2620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1730920535011804e-06,
+      "loss": 3.5696,
+      "step": 2630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.192761605035406e-06,
+      "loss": 3.6377,
+      "step": 2640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.212431156569631e-06,
+      "loss": 3.6475,
+      "step": 2650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2321007081038554e-06,
+      "loss": 3.5999,
+      "step": 2660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.25177025963808e-06,
+      "loss": 3.6661,
+      "step": 2670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.271439811172306e-06,
+      "loss": 3.6036,
+      "step": 2680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.29110936270653e-06,
+      "loss": 3.6243,
+      "step": 2690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.310778914240756e-06,
+      "loss": 3.6249,
+      "step": 2700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.330448465774981e-06,
+      "loss": 3.5835,
+      "step": 2710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.350118017309206e-06,
+      "loss": 3.5837,
+      "step": 2720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.369787568843431e-06,
+      "loss": 3.6285,
+      "step": 2730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.389457120377656e-06,
+      "loss": 3.5927,
+      "step": 2740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.40912667191188e-06,
+      "loss": 3.6039,
+      "step": 2750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.428796223446106e-06,
+      "loss": 3.6034,
+      "step": 2760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.448465774980331e-06,
+      "loss": 3.5793,
+      "step": 2770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.468135326514555e-06,
+      "loss": 3.6462,
+      "step": 2780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.487804878048781e-06,
+      "loss": 3.5696,
+      "step": 2790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.5074744295830065e-06,
+      "loss": 3.5853,
+      "step": 2800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.527143981117231e-06,
+      "loss": 3.555,
+      "step": 2810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.546813532651456e-06,
+      "loss": 3.6268,
+      "step": 2820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.566483084185681e-06,
+      "loss": 3.6226,
+      "step": 2830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.586152635719906e-06,
+      "loss": 3.5737,
+      "step": 2840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.605822187254131e-06,
+      "loss": 3.5802,
+      "step": 2850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.625491738788356e-06,
+      "loss": 3.5624,
+      "step": 2860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.645161290322582e-06,
+      "loss": 3.5406,
+      "step": 2870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.664830841856807e-06,
+      "loss": 3.6064,
+      "step": 2880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.6845003933910315e-06,
+      "loss": 3.564,
+      "step": 2890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.704169944925256e-06,
+      "loss": 3.5618,
+      "step": 2900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.723839496459481e-06,
+      "loss": 3.5909,
+      "step": 2910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7435090479937065e-06,
+      "loss": 3.6188,
+      "step": 2920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.763178599527931e-06,
+      "loss": 3.5531,
+      "step": 2930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.782848151062156e-06,
+      "loss": 3.5678,
+      "step": 2940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.802517702596381e-06,
+      "loss": 3.5839,
+      "step": 2950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.822187254130607e-06,
+      "loss": 3.5184,
+      "step": 2960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.841856805664832e-06,
+      "loss": 3.6194,
+      "step": 2970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.8615263571990565e-06,
+      "loss": 3.6048,
+      "step": 2980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.881195908733281e-06,
+      "loss": 3.5419,
+      "step": 2990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.900865460267507e-06,
+      "loss": 3.5548,
+      "step": 3000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9205350118017315e-06,
+      "loss": 3.5635,
+      "step": 3010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.940204563335956e-06,
+      "loss": 3.5625,
+      "step": 3020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.959874114870181e-06,
+      "loss": 3.5418,
+      "step": 3030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.979543666404406e-06,
+      "loss": 3.5343,
+      "step": 3040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.999213217938632e-06,
+      "loss": 3.5829,
+      "step": 3050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.018882769472857e-06,
+      "loss": 3.6052,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.0385523210070814e-06,
+      "loss": 3.5536,
+      "step": 3070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.058221872541307e-06,
+      "loss": 3.5636,
+      "step": 3080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.077891424075532e-06,
+      "loss": 3.5445,
+      "step": 3090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 3.5567,
+      "step": 3100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.117230527143981e-06,
+      "loss": 3.5037,
+      "step": 3110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.136900078678206e-06,
+      "loss": 3.5135,
+      "step": 3120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.156569630212431e-06,
+      "loss": 3.5011,
+      "step": 3130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.176239181746657e-06,
+      "loss": 3.5445,
+      "step": 3140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.195908733280882e-06,
+      "loss": 3.5671,
+      "step": 3150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.215578284815107e-06,
+      "loss": 3.5517,
+      "step": 3160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.235247836349332e-06,
+      "loss": 3.5507,
+      "step": 3170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254917387883557e-06,
+      "loss": 3.5214,
+      "step": 3180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274586939417781e-06,
+      "loss": 3.5423,
+      "step": 3190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294256490952006e-06,
+      "loss": 3.5576,
+      "step": 3200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313926042486232e-06,
+      "loss": 3.4897,
+      "step": 3210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333595594020456e-06,
+      "loss": 3.4986,
+      "step": 3220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.353265145554682e-06,
+      "loss": 3.5614,
+      "step": 3230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.3729346970889075e-06,
+      "loss": 3.5102,
+      "step": 3240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.392604248623132e-06,
+      "loss": 3.5298,
+      "step": 3250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.412273800157357e-06,
+      "loss": 3.5566,
+      "step": 3260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.431943351691582e-06,
+      "loss": 3.5132,
+      "step": 3270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 3.5227,
+      "step": 3280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.471282454760032e-06,
+      "loss": 3.5358,
+      "step": 3290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.490952006294257e-06,
+      "loss": 3.5533,
+      "step": 3300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.510621557828482e-06,
+      "loss": 3.504,
+      "step": 3310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.530291109362708e-06,
+      "loss": 3.5052,
+      "step": 3320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.5499606608969325e-06,
+      "loss": 3.4848,
+      "step": 3330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.569630212431157e-06,
+      "loss": 3.5255,
+      "step": 3340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.589299763965382e-06,
+      "loss": 3.4982,
+      "step": 3350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.6089693154996075e-06,
+      "loss": 3.4717,
+      "step": 3360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.628638867033832e-06,
+      "loss": 3.5313,
+      "step": 3370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.648308418568057e-06,
+      "loss": 3.4992,
+      "step": 3380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.667977970102282e-06,
+      "loss": 3.4922,
+      "step": 3390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.687647521636508e-06,
+      "loss": 3.5207,
+      "step": 3400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.707317073170733e-06,
+      "loss": 3.4975,
+      "step": 3410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.7269866247049575e-06,
+      "loss": 3.5334,
+      "step": 3420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.746656176239182e-06,
+      "loss": 3.5253,
+      "step": 3430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.766325727773408e-06,
+      "loss": 3.4787,
+      "step": 3440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7859952793076325e-06,
+      "loss": 3.5432,
+      "step": 3450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.805664830841857e-06,
+      "loss": 3.5211,
+      "step": 3460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.825334382376082e-06,
+      "loss": 3.5396,
+      "step": 3470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.845003933910307e-06,
+      "loss": 3.5148,
+      "step": 3480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.864673485444533e-06,
+      "loss": 3.5106,
+      "step": 3490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.884343036978758e-06,
+      "loss": 3.5657,
+      "step": 3500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.9040125885129824e-06,
+      "loss": 3.4741,
+      "step": 3510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.923682140047208e-06,
+      "loss": 3.4687,
+      "step": 3520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.943351691581433e-06,
+      "loss": 3.4902,
+      "step": 3530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.9630212431156574e-06,
+      "loss": 3.4944,
+      "step": 3540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.982690794649882e-06,
+      "loss": 3.5296,
+      "step": 3550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.002360346184107e-06,
+      "loss": 3.4427,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.022029897718332e-06,
+      "loss": 3.4836,
+      "step": 3570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.041699449252558e-06,
+      "loss": 3.5282,
+      "step": 3580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.061369000786783e-06,
+      "loss": 3.4904,
+      "step": 3590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.081038552321008e-06,
+      "loss": 3.5123,
+      "step": 3600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.100708103855233e-06,
+      "loss": 3.4851,
+      "step": 3610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.120377655389458e-06,
+      "loss": 3.5104,
+      "step": 3620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.140047206923682e-06,
+      "loss": 3.4865,
+      "step": 3630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.159716758457907e-06,
+      "loss": 3.4838,
+      "step": 3640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.179386309992133e-06,
+      "loss": 3.4799,
+      "step": 3650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.199055861526357e-06,
+      "loss": 3.4468,
+      "step": 3660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.218725413060583e-06,
+      "loss": 3.4493,
+      "step": 3670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.2383949645948085e-06,
+      "loss": 3.4606,
+      "step": 3680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.258064516129033e-06,
+      "loss": 3.454,
+      "step": 3690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.277734067663258e-06,
+      "loss": 3.4927,
+      "step": 3700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.297403619197483e-06,
+      "loss": 3.4941,
+      "step": 3710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 3.4527,
+      "step": 3720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.336742722265933e-06,
+      "loss": 3.482,
+      "step": 3730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.356412273800158e-06,
+      "loss": 3.4325,
+      "step": 3740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.376081825334382e-06,
+      "loss": 3.493,
+      "step": 3750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.395751376868609e-06,
+      "loss": 3.4876,
+      "step": 3760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.4154209284028335e-06,
+      "loss": 3.4868,
+      "step": 3770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.435090479937058e-06,
+      "loss": 3.4469,
+      "step": 3780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.454760031471283e-06,
+      "loss": 3.4636,
+      "step": 3790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.474429583005508e-06,
+      "loss": 3.477,
+      "step": 3800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.494099134539733e-06,
+      "loss": 3.4457,
+      "step": 3810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.513768686073958e-06,
+      "loss": 3.4261,
+      "step": 3820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.533438237608183e-06,
+      "loss": 3.4647,
+      "step": 3830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.553107789142409e-06,
+      "loss": 3.4426,
+      "step": 3840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.572777340676634e-06,
+      "loss": 3.4723,
+      "step": 3850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.5924468922108585e-06,
+      "loss": 3.4145,
+      "step": 3860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.612116443745083e-06,
+      "loss": 3.5064,
+      "step": 3870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.631785995279308e-06,
+      "loss": 3.471,
+      "step": 3880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.651455546813533e-06,
+      "loss": 3.4526,
+      "step": 3890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.671125098347757e-06,
+      "loss": 3.4661,
+      "step": 3900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.690794649881983e-06,
+      "loss": 3.4499,
+      "step": 3910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.710464201416208e-06,
+      "loss": 3.441,
+      "step": 3920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.730133752950434e-06,
+      "loss": 3.4279,
+      "step": 3930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.74980330448466e-06,
+      "loss": 3.4829,
+      "step": 3940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.769472856018883e-06,
+      "loss": 3.4491,
+      "step": 3950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.789142407553109e-06,
+      "loss": 3.4142,
+      "step": 3960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.808811959087333e-06,
+      "loss": 3.4605,
+      "step": 3970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.828481510621558e-06,
+      "loss": 3.4205,
+      "step": 3980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.848151062155784e-06,
+      "loss": 3.4463,
+      "step": 3990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.867820613690008e-06,
+      "loss": 3.4491,
+      "step": 4000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.887490165224233e-06,
+      "loss": 3.4392,
+      "step": 4010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.907159716758459e-06,
+      "loss": 3.4537,
+      "step": 4020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.926829268292685e-06,
+      "loss": 3.465,
+      "step": 4030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.946498819826908e-06,
+      "loss": 3.4252,
+      "step": 4040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.966168371361134e-06,
+      "loss": 3.4158,
+      "step": 4050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.985837922895358e-06,
+      "loss": 3.4185,
+      "step": 4060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.005507474429583e-06,
+      "loss": 3.4405,
+      "step": 4070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.025177025963809e-06,
+      "loss": 3.4204,
+      "step": 4080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.044846577498033e-06,
+      "loss": 3.4471,
+      "step": 4090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 3.4428,
+      "step": 4100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.084185680566484e-06,
+      "loss": 3.3788,
+      "step": 4110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.10385523210071e-06,
+      "loss": 3.4412,
+      "step": 4120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.123524783634933e-06,
+      "loss": 3.4343,
+      "step": 4130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.143194335169159e-06,
+      "loss": 3.4611,
+      "step": 4140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.162863886703385e-06,
+      "loss": 3.4809,
+      "step": 4150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.182533438237608e-06,
+      "loss": 3.4415,
+      "step": 4160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.202202989771834e-06,
+      "loss": 3.4414,
+      "step": 4170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.221872541306058e-06,
+      "loss": 3.4033,
+      "step": 4180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.241542092840283e-06,
+      "loss": 3.4497,
+      "step": 4190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.261211644374509e-06,
+      "loss": 3.4217,
+      "step": 4200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.280881195908734e-06,
+      "loss": 3.3904,
+      "step": 4210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.300550747442958e-06,
+      "loss": 3.4931,
+      "step": 4220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.320220298977184e-06,
+      "loss": 3.4473,
+      "step": 4230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.33988985051141e-06,
+      "loss": 3.443,
+      "step": 4240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.359559402045633e-06,
+      "loss": 3.4067,
+      "step": 4250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.379228953579859e-06,
+      "loss": 3.4277,
+      "step": 4260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.398898505114084e-06,
+      "loss": 3.435,
+      "step": 4270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.418568056648308e-06,
+      "loss": 3.3978,
+      "step": 4280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.438237608182534e-06,
+      "loss": 3.4154,
+      "step": 4290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.45790715971676e-06,
+      "loss": 3.405,
+      "step": 4300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.477576711250985e-06,
+      "loss": 3.3593,
+      "step": 4310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.497246262785209e-06,
+      "loss": 3.3916,
+      "step": 4320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.516915814319434e-06,
+      "loss": 3.4198,
+      "step": 4330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.536585365853658e-06,
+      "loss": 3.415,
+      "step": 4340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.556254917387884e-06,
+      "loss": 3.4256,
+      "step": 4350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.57592446892211e-06,
+      "loss": 3.4179,
+      "step": 4360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.595594020456335e-06,
+      "loss": 3.4133,
+      "step": 4370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.61526357199056e-06,
+      "loss": 3.3898,
+      "step": 4380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.634933123524784e-06,
+      "loss": 3.4257,
+      "step": 4390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.65460267505901e-06,
+      "loss": 3.4186,
+      "step": 4400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.674272226593234e-06,
+      "loss": 3.3529,
+      "step": 4410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.69394177812746e-06,
+      "loss": 3.4001,
+      "step": 4420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.713611329661685e-06,
+      "loss": 3.3901,
+      "step": 4430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.733280881195909e-06,
+      "loss": 3.3494,
+      "step": 4440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.752950432730134e-06,
+      "loss": 3.4299,
+      "step": 4450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.77261998426436e-06,
+      "loss": 3.3887,
+      "step": 4460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.792289535798586e-06,
+      "loss": 3.4272,
+      "step": 4470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.81195908733281e-06,
+      "loss": 3.4085,
+      "step": 4480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.831628638867035e-06,
+      "loss": 3.3906,
+      "step": 4490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.851298190401259e-06,
+      "loss": 3.403,
+      "step": 4500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.870967741935484e-06,
+      "loss": 3.3728,
+      "step": 4510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.89063729346971e-06,
+      "loss": 3.4064,
+      "step": 4520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.910306845003934e-06,
+      "loss": 3.4116,
+      "step": 4530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.92997639653816e-06,
+      "loss": 3.3489,
+      "step": 4540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.949645948072385e-06,
+      "loss": 3.4465,
+      "step": 4550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.96931549960661e-06,
+      "loss": 3.3944,
+      "step": 4560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.988985051140834e-06,
+      "loss": 3.3866,
+      "step": 4570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.00865460267506e-06,
+      "loss": 3.3671,
+      "step": 4580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.028324154209286e-06,
+      "loss": 3.3888,
+      "step": 4590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.04799370574351e-06,
+      "loss": 3.3679,
+      "step": 4600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.067663257277735e-06,
+      "loss": 3.4261,
+      "step": 4610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.087332808811959e-06,
+      "loss": 3.3394,
+      "step": 4620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.107002360346184e-06,
+      "loss": 3.3798,
+      "step": 4630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.12667191188041e-06,
+      "loss": 3.3296,
+      "step": 4640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.146341463414635e-06,
+      "loss": 3.3878,
+      "step": 4650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.16601101494886e-06,
+      "loss": 3.4094,
+      "step": 4660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.185680566483085e-06,
+      "loss": 3.3979,
+      "step": 4670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.20535011801731e-06,
+      "loss": 3.3383,
+      "step": 4680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.225019669551534e-06,
+      "loss": 3.4225,
+      "step": 4690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.24468922108576e-06,
+      "loss": 3.4002,
+      "step": 4700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.264358772619984e-06,
+      "loss": 3.3844,
+      "step": 4710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.28402832415421e-06,
+      "loss": 3.3253,
+      "step": 4720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.303697875688435e-06,
+      "loss": 3.4584,
+      "step": 4730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.32336742722266e-06,
+      "loss": 3.3855,
+      "step": 4740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.343036978756886e-06,
+      "loss": 3.4037,
+      "step": 4750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.36270653029111e-06,
+      "loss": 3.377,
+      "step": 4760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.382376081825335e-06,
+      "loss": 3.3701,
+      "step": 4770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.40204563335956e-06,
+      "loss": 3.3593,
+      "step": 4780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.421715184893785e-06,
+      "loss": 3.3476,
+      "step": 4790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.44138473642801e-06,
+      "loss": 3.4142,
+      "step": 4800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.461054287962234e-06,
+      "loss": 3.3946,
+      "step": 4810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.48072383949646e-06,
+      "loss": 3.4022,
+      "step": 4820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.500393391030685e-06,
+      "loss": 3.4067,
+      "step": 4830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.520062942564911e-06,
+      "loss": 3.3942,
+      "step": 4840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.539732494099135e-06,
+      "loss": 3.3801,
+      "step": 4850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.55940204563336e-06,
+      "loss": 3.354,
+      "step": 4860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.579071597167586e-06,
+      "loss": 3.3858,
+      "step": 4870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.59874114870181e-06,
+      "loss": 3.3725,
+      "step": 4880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.618410700236035e-06,
+      "loss": 3.3767,
+      "step": 4890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.638080251770261e-06,
+      "loss": 3.417,
+      "step": 4900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.657749803304487e-06,
+      "loss": 3.3679,
+      "step": 4910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 3.318,
+      "step": 4920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.697088906372936e-06,
+      "loss": 3.3397,
+      "step": 4930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.71675845790716e-06,
+      "loss": 3.3753,
+      "step": 4940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.736428009441385e-06,
+      "loss": 3.3676,
+      "step": 4950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 3.386,
+      "step": 4960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.775767112509835e-06,
+      "loss": 3.3541,
+      "step": 4970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.79543666404406e-06,
+      "loss": 3.3501,
+      "step": 4980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.815106215578286e-06,
+      "loss": 3.4148,
+      "step": 4990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.834775767112512e-06,
+      "loss": 3.3436,
+      "step": 5000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.854445318646735e-06,
+      "loss": 3.3512,
+      "step": 5010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.874114870180961e-06,
+      "loss": 3.3778,
+      "step": 5020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.893784421715187e-06,
+      "loss": 3.3236,
+      "step": 5030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.91345397324941e-06,
+      "loss": 3.324,
+      "step": 5040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.933123524783636e-06,
+      "loss": 3.3651,
+      "step": 5050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.95279307631786e-06,
+      "loss": 3.3532,
+      "step": 5060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.972462627852085e-06,
+      "loss": 3.4112,
+      "step": 5070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.992132179386311e-06,
+      "loss": 3.341,
+      "step": 5080
+    },
+    {
+      "epoch": 1.0,
+      "eval_BLEU": 6.4516,
+      "eval_gen_len": 13.5722,
+      "eval_loss": 3.255478858947754,
+      "eval_runtime": 226.9229,
+      "eval_samples_per_second": 47.43,
+      "eval_steps_per_second": 0.189,
+      "step": 5084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0011801730920536e-05,
+      "loss": 3.3662,
+      "step": 5090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.003147128245476e-05,
+      "loss": 3.3356,
+      "step": 5100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0051140833988986e-05,
+      "loss": 3.3143,
+      "step": 5110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0070810385523211e-05,
+      "loss": 3.307,
+      "step": 5120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0090479937057435e-05,
+      "loss": 3.3088,
+      "step": 5130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0110149488591661e-05,
+      "loss": 3.2954,
+      "step": 5140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0129819040125885e-05,
+      "loss": 3.2945,
+      "step": 5150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.014948859166011e-05,
+      "loss": 3.3115,
+      "step": 5160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0169158143194336e-05,
+      "loss": 3.3272,
+      "step": 5170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.018882769472856e-05,
+      "loss": 3.3277,
+      "step": 5180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0208497246262785e-05,
+      "loss": 3.2719,
+      "step": 5190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.022816679779701e-05,
+      "loss": 3.3219,
+      "step": 5200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0247836349331236e-05,
+      "loss": 3.3333,
+      "step": 5210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0267505900865462e-05,
+      "loss": 3.3056,
+      "step": 5220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0287175452399688e-05,
+      "loss": 3.3389,
+      "step": 5230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0306845003933911e-05,
+      "loss": 3.2862,
+      "step": 5240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0326514555468137e-05,
+      "loss": 3.2826,
+      "step": 5250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0346184107002361e-05,
+      "loss": 3.2738,
+      "step": 5260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0365853658536586e-05,
+      "loss": 3.3086,
+      "step": 5270
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0385523210070812e-05,
+      "loss": 3.2969,
+      "step": 5280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0405192761605036e-05,
+      "loss": 3.3023,
+      "step": 5290
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0424862313139261e-05,
+      "loss": 3.3215,
+      "step": 5300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0444531864673485e-05,
+      "loss": 3.3516,
+      "step": 5310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0464201416207711e-05,
+      "loss": 3.2994,
+      "step": 5320
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 3.2808,
+      "step": 5330
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.050354051927616e-05,
+      "loss": 3.3516,
+      "step": 5340
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0523210070810386e-05,
+      "loss": 3.3062,
+      "step": 5350
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0542879622344611e-05,
+      "loss": 3.3032,
+      "step": 5360
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0562549173878835e-05,
+      "loss": 3.3271,
+      "step": 5370
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.058221872541306e-05,
+      "loss": 3.3099,
+      "step": 5380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0601888276947288e-05,
+      "loss": 3.3191,
+      "step": 5390
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0621557828481512e-05,
+      "loss": 3.2991,
+      "step": 5400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0641227380015738e-05,
+      "loss": 3.3429,
+      "step": 5410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0660896931549961e-05,
+      "loss": 3.2812,
+      "step": 5420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0680566483084187e-05,
+      "loss": 3.3372,
+      "step": 5430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0700236034618413e-05,
+      "loss": 3.2994,
+      "step": 5440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0719905586152636e-05,
+      "loss": 3.2952,
+      "step": 5450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0739575137686862e-05,
+      "loss": 3.2838,
+      "step": 5460
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0759244689221088e-05,
+      "loss": 3.2899,
+      "step": 5470
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0778914240755311e-05,
+      "loss": 3.308,
+      "step": 5480
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0798583792289537e-05,
+      "loss": 3.2786,
+      "step": 5490
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.081825334382376e-05,
+      "loss": 3.2968,
+      "step": 5500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0837922895357986e-05,
+      "loss": 3.2391,
+      "step": 5510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0857592446892212e-05,
+      "loss": 3.3058,
+      "step": 5520
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0877261998426436e-05,
+      "loss": 3.2659,
+      "step": 5530
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0896931549960661e-05,
+      "loss": 3.2869,
+      "step": 5540
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0916601101494885e-05,
+      "loss": 3.2869,
+      "step": 5550
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.093627065302911e-05,
+      "loss": 3.2981,
+      "step": 5560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0955940204563338e-05,
+      "loss": 3.2444,
+      "step": 5570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0975609756097562e-05,
+      "loss": 3.2756,
+      "step": 5580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0995279307631787e-05,
+      "loss": 3.2452,
+      "step": 5590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1014948859166013e-05,
+      "loss": 3.3294,
+      "step": 5600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1034618410700237e-05,
+      "loss": 3.3372,
+      "step": 5610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1054287962234462e-05,
+      "loss": 3.2762,
+      "step": 5620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1073957513768688e-05,
+      "loss": 3.2792,
+      "step": 5630
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1093627065302912e-05,
+      "loss": 3.2549,
+      "step": 5640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1113296616837137e-05,
+      "loss": 3.3428,
+      "step": 5650
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1132966168371361e-05,
+      "loss": 3.3078,
+      "step": 5660
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1152635719905587e-05,
+      "loss": 3.332,
+      "step": 5670
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1172305271439812e-05,
+      "loss": 3.2721,
+      "step": 5680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1191974822974036e-05,
+      "loss": 3.332,
+      "step": 5690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1211644374508262e-05,
+      "loss": 3.2787,
+      "step": 5700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1231313926042486e-05,
+      "loss": 3.2874,
+      "step": 5710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1250983477576711e-05,
+      "loss": 3.3135,
+      "step": 5720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1270653029110937e-05,
+      "loss": 3.3044,
+      "step": 5730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 3.2386,
+      "step": 5740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1309992132179388e-05,
+      "loss": 3.2386,
+      "step": 5750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1329661683713614e-05,
+      "loss": 3.2684,
+      "step": 5760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1349331235247837e-05,
+      "loss": 3.2532,
+      "step": 5770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1369000786782063e-05,
+      "loss": 3.3008,
+      "step": 5780
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1388670338316289e-05,
+      "loss": 3.3445,
+      "step": 5790
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1408339889850512e-05,
+      "loss": 3.2447,
+      "step": 5800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1428009441384738e-05,
+      "loss": 3.2995,
+      "step": 5810
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1447678992918962e-05,
+      "loss": 3.22,
+      "step": 5820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1467348544453187e-05,
+      "loss": 3.3058,
+      "step": 5830
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1487018095987413e-05,
+      "loss": 3.2647,
+      "step": 5840
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1506687647521637e-05,
+      "loss": 3.2243,
+      "step": 5850
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1526357199055862e-05,
+      "loss": 3.2505,
+      "step": 5860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1546026750590086e-05,
+      "loss": 3.2714,
+      "step": 5870
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1565696302124312e-05,
+      "loss": 3.2828,
+      "step": 5880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1585365853658537e-05,
+      "loss": 3.2766,
+      "step": 5890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1605035405192761e-05,
+      "loss": 3.293,
+      "step": 5900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1624704956726987e-05,
+      "loss": 3.2671,
+      "step": 5910
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1644374508261214e-05,
+      "loss": 3.2738,
+      "step": 5920
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1664044059795438e-05,
+      "loss": 3.2483,
+      "step": 5930
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1683713611329664e-05,
+      "loss": 3.2937,
+      "step": 5940
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1703383162863889e-05,
+      "loss": 3.2358,
+      "step": 5950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1723052714398113e-05,
+      "loss": 3.2701,
+      "step": 5960
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1742722265932338e-05,
+      "loss": 3.295,
+      "step": 5970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1762391817466562e-05,
+      "loss": 3.2931,
+      "step": 5980
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1782061369000788e-05,
+      "loss": 3.2949,
+      "step": 5990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1801730920535013e-05,
+      "loss": 3.2257,
+      "step": 6000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1821400472069237e-05,
+      "loss": 3.279,
+      "step": 6010
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1841070023603463e-05,
+      "loss": 3.2374,
+      "step": 6020
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1860739575137687e-05,
+      "loss": 3.2497,
+      "step": 6030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1880409126671912e-05,
+      "loss": 3.3213,
+      "step": 6040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1900078678206138e-05,
+      "loss": 3.2804,
+      "step": 6050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1919748229740362e-05,
+      "loss": 3.2553,
+      "step": 6060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1939417781274587e-05,
+      "loss": 3.3241,
+      "step": 6070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1959087332808811e-05,
+      "loss": 3.3012,
+      "step": 6080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1978756884343037e-05,
+      "loss": 3.2165,
+      "step": 6090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1998426435877264e-05,
+      "loss": 3.2303,
+      "step": 6100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.201809598741149e-05,
+      "loss": 3.2656,
+      "step": 6110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2037765538945713e-05,
+      "loss": 3.2523,
+      "step": 6120
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2057435090479939e-05,
+      "loss": 3.2859,
+      "step": 6130
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2077104642014163e-05,
+      "loss": 3.2662,
+      "step": 6140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2096774193548388e-05,
+      "loss": 3.2733,
+      "step": 6150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2116443745082614e-05,
+      "loss": 3.264,
+      "step": 6160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2136113296616838e-05,
+      "loss": 3.2185,
+      "step": 6170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2155782848151063e-05,
+      "loss": 3.3,
+      "step": 6180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2175452399685287e-05,
+      "loss": 3.2113,
+      "step": 6190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 3.2389,
+      "step": 6200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2214791502753738e-05,
+      "loss": 3.2705,
+      "step": 6210
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2234461054287962e-05,
+      "loss": 3.3077,
+      "step": 6220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2254130605822188e-05,
+      "loss": 3.2578,
+      "step": 6230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2273800157356412e-05,
+      "loss": 3.2671,
+      "step": 6240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2293469708890637e-05,
+      "loss": 3.2509,
+      "step": 6250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2313139260424863e-05,
+      "loss": 3.249,
+      "step": 6260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.233280881195909e-05,
+      "loss": 3.3012,
+      "step": 6270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2352478363493314e-05,
+      "loss": 3.2415,
+      "step": 6280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.237214791502754e-05,
+      "loss": 3.2402,
+      "step": 6290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2391817466561763e-05,
+      "loss": 3.2308,
+      "step": 6300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2411487018095989e-05,
+      "loss": 3.2129,
+      "step": 6310
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2431156569630215e-05,
+      "loss": 3.2294,
+      "step": 6320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2450826121164438e-05,
+      "loss": 3.2224,
+      "step": 6330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2470495672698664e-05,
+      "loss": 3.2615,
+      "step": 6340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2490165224232888e-05,
+      "loss": 3.2226,
+      "step": 6350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2509834775767113e-05,
+      "loss": 3.2441,
+      "step": 6360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2529504327301339e-05,
+      "loss": 3.2737,
+      "step": 6370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2549173878835563e-05,
+      "loss": 3.2557,
+      "step": 6380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2568843430369788e-05,
+      "loss": 3.3025,
+      "step": 6390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2588512981904012e-05,
+      "loss": 3.2173,
+      "step": 6400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2608182533438238e-05,
+      "loss": 3.2789,
+      "step": 6410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2627852084972463e-05,
+      "loss": 3.2538,
+      "step": 6420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2647521636506687e-05,
+      "loss": 3.2603,
+      "step": 6430
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2667191188040913e-05,
+      "loss": 3.2838,
+      "step": 6440
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.268686073957514e-05,
+      "loss": 3.2994,
+      "step": 6450
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2706530291109364e-05,
+      "loss": 3.2492,
+      "step": 6460
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.272619984264359e-05,
+      "loss": 3.2277,
+      "step": 6470
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2745869394177815e-05,
+      "loss": 3.2863,
+      "step": 6480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2765538945712039e-05,
+      "loss": 3.2427,
+      "step": 6490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2785208497246264e-05,
+      "loss": 3.2439,
+      "step": 6500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2804878048780488e-05,
+      "loss": 3.1646,
+      "step": 6510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824547600314714e-05,
+      "loss": 3.2549,
+      "step": 6520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.284421715184894e-05,
+      "loss": 3.2259,
+      "step": 6530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2863886703383163e-05,
+      "loss": 3.2502,
+      "step": 6540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2883556254917389e-05,
+      "loss": 3.2327,
+      "step": 6550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 3.23,
+      "step": 6560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2922895357985838e-05,
+      "loss": 3.2352,
+      "step": 6570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2942564909520064e-05,
+      "loss": 3.2511,
+      "step": 6580
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2962234461054288e-05,
+      "loss": 3.2669,
+      "step": 6590
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2981904012588513e-05,
+      "loss": 3.2672,
+      "step": 6600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3001573564122739e-05,
+      "loss": 3.2314,
+      "step": 6610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3021243115656964e-05,
+      "loss": 3.2489,
+      "step": 6620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.304091266719119e-05,
+      "loss": 3.2551,
+      "step": 6630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3060582218725416e-05,
+      "loss": 3.2404,
+      "step": 6640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.308025177025964e-05,
+      "loss": 3.2574,
+      "step": 6650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3099921321793865e-05,
+      "loss": 3.2826,
+      "step": 6660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3119590873328089e-05,
+      "loss": 3.2104,
+      "step": 6670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3139260424862314e-05,
+      "loss": 3.2382,
+      "step": 6680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.315892997639654e-05,
+      "loss": 3.212,
+      "step": 6690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3178599527930764e-05,
+      "loss": 3.1799,
+      "step": 6700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.319826907946499e-05,
+      "loss": 3.2629,
+      "step": 6710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3217938630999215e-05,
+      "loss": 3.259,
+      "step": 6720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3237608182533439e-05,
+      "loss": 3.1751,
+      "step": 6730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3257277734067664e-05,
+      "loss": 3.2497,
+      "step": 6740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3276947285601888e-05,
+      "loss": 3.2474,
+      "step": 6750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3296616837136114e-05,
+      "loss": 3.1967,
+      "step": 6760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.331628638867034e-05,
+      "loss": 3.2157,
+      "step": 6770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3335955940204563e-05,
+      "loss": 3.2092,
+      "step": 6780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3355625491738789e-05,
+      "loss": 3.252,
+      "step": 6790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3375295043273016e-05,
+      "loss": 3.2699,
+      "step": 6800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.339496459480724e-05,
+      "loss": 3.1731,
+      "step": 6810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3414634146341466e-05,
+      "loss": 3.2615,
+      "step": 6820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3434303697875691e-05,
+      "loss": 3.157,
+      "step": 6830
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3453973249409915e-05,
+      "loss": 3.2149,
+      "step": 6840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.347364280094414e-05,
+      "loss": 3.226,
+      "step": 6850
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3493312352478364e-05,
+      "loss": 3.1612,
+      "step": 6860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.351298190401259e-05,
+      "loss": 3.232,
+      "step": 6870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3532651455546815e-05,
+      "loss": 3.1997,
+      "step": 6880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.355232100708104e-05,
+      "loss": 3.2155,
+      "step": 6890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3571990558615265e-05,
+      "loss": 3.2735,
+      "step": 6900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3591660110149489e-05,
+      "loss": 3.2363,
+      "step": 6910
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3611329661683714e-05,
+      "loss": 3.2182,
+      "step": 6920
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.363099921321794e-05,
+      "loss": 3.2111,
+      "step": 6930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3650668764752164e-05,
+      "loss": 3.255,
+      "step": 6940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.367033831628639e-05,
+      "loss": 3.237,
+      "step": 6950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3690007867820613e-05,
+      "loss": 3.2489,
+      "step": 6960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3709677419354839e-05,
+      "loss": 3.2111,
+      "step": 6970
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3729346970889066e-05,
+      "loss": 3.2064,
+      "step": 6980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3749016522423292e-05,
+      "loss": 3.2552,
+      "step": 6990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3768686073957515e-05,
+      "loss": 3.2475,
+      "step": 7000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3788355625491741e-05,
+      "loss": 3.2224,
+      "step": 7010
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3808025177025965e-05,
+      "loss": 3.2198,
+      "step": 7020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.382769472856019e-05,
+      "loss": 3.2584,
+      "step": 7030
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3847364280094416e-05,
+      "loss": 3.257,
+      "step": 7040
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.386703383162864e-05,
+      "loss": 3.2492,
+      "step": 7050
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3886703383162865e-05,
+      "loss": 3.264,
+      "step": 7060
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.390637293469709e-05,
+      "loss": 3.207,
+      "step": 7070
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3926042486231315e-05,
+      "loss": 3.2306,
+      "step": 7080
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.394571203776554e-05,
+      "loss": 3.2242,
+      "step": 7090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3965381589299764e-05,
+      "loss": 3.2216,
+      "step": 7100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.398505114083399e-05,
+      "loss": 3.2287,
+      "step": 7110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4004720692368214e-05,
+      "loss": 3.1977,
+      "step": 7120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.402439024390244e-05,
+      "loss": 3.2438,
+      "step": 7130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4044059795436665e-05,
+      "loss": 3.231,
+      "step": 7140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4063729346970892e-05,
+      "loss": 3.2126,
+      "step": 7150
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4083398898505116e-05,
+      "loss": 3.2821,
+      "step": 7160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4103068450039342e-05,
+      "loss": 3.2244,
+      "step": 7170
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4122738001573565e-05,
+      "loss": 3.249,
+      "step": 7180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4142407553107791e-05,
+      "loss": 3.2064,
+      "step": 7190
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4162077104642017e-05,
+      "loss": 3.2062,
+      "step": 7200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.418174665617624e-05,
+      "loss": 3.1832,
+      "step": 7210
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4201416207710466e-05,
+      "loss": 3.256,
+      "step": 7220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.422108575924469e-05,
+      "loss": 3.267,
+      "step": 7230
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4240755310778915e-05,
+      "loss": 3.1921,
+      "step": 7240
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4260424862313141e-05,
+      "loss": 3.216,
+      "step": 7250
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4280094413847365e-05,
+      "loss": 3.2431,
+      "step": 7260
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.429976396538159e-05,
+      "loss": 3.1827,
+      "step": 7270
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4319433516915814e-05,
+      "loss": 3.2589,
+      "step": 7280
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.433910306845004e-05,
+      "loss": 3.2311,
+      "step": 7290
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4358772619984265e-05,
+      "loss": 3.2191,
+      "step": 7300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.437844217151849e-05,
+      "loss": 3.1649,
+      "step": 7310
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4398111723052715e-05,
+      "loss": 3.1983,
+      "step": 7320
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4417781274586942e-05,
+      "loss": 3.1971,
+      "step": 7330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4437450826121166e-05,
+      "loss": 3.2248,
+      "step": 7340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4457120377655391e-05,
+      "loss": 3.1957,
+      "step": 7350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4476789929189617e-05,
+      "loss": 3.1845,
+      "step": 7360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4496459480723841e-05,
+      "loss": 3.1946,
+      "step": 7370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 3.1942,
+      "step": 7380
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.453579858379229e-05,
+      "loss": 3.2275,
+      "step": 7390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4555468135326516e-05,
+      "loss": 3.2327,
+      "step": 7400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4575137686860741e-05,
+      "loss": 3.1836,
+      "step": 7410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4594807238394965e-05,
+      "loss": 3.1749,
+      "step": 7420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4614476789929191e-05,
+      "loss": 3.1897,
+      "step": 7430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 3.1992,
+      "step": 7440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.465381589299764e-05,
+      "loss": 3.211,
+      "step": 7450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4673485444531866e-05,
+      "loss": 3.1975,
+      "step": 7460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.469315499606609e-05,
+      "loss": 3.1683,
+      "step": 7470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4712824547600315e-05,
+      "loss": 3.231,
+      "step": 7480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.473249409913454e-05,
+      "loss": 3.207,
+      "step": 7490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4752163650668765e-05,
+      "loss": 3.236,
+      "step": 7500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4771833202202992e-05,
+      "loss": 3.2342,
+      "step": 7510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4791502753737218e-05,
+      "loss": 3.1869,
+      "step": 7520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4811172305271441e-05,
+      "loss": 3.1899,
+      "step": 7530
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4830841856805667e-05,
+      "loss": 3.1646,
+      "step": 7540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4850511408339891e-05,
+      "loss": 3.1891,
+      "step": 7550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4870180959874116e-05,
+      "loss": 3.2195,
+      "step": 7560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4889850511408342e-05,
+      "loss": 3.2148,
+      "step": 7570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4909520062942566e-05,
+      "loss": 3.1779,
+      "step": 7580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4929189614476791e-05,
+      "loss": 3.257,
+      "step": 7590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4948859166011015e-05,
+      "loss": 3.2334,
+      "step": 7600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.496852871754524e-05,
+      "loss": 3.2242,
+      "step": 7610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4988198269079466e-05,
+      "loss": 3.1861,
+      "step": 7620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.500786782061369e-05,
+      "loss": 3.2071,
+      "step": 7630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5027537372147916e-05,
+      "loss": 3.1418,
+      "step": 7640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.504720692368214e-05,
+      "loss": 3.2231,
+      "step": 7650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5066876475216365e-05,
+      "loss": 3.1741,
+      "step": 7660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.508654602675059e-05,
+      "loss": 3.1638,
+      "step": 7670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5106215578284818e-05,
+      "loss": 3.1733,
+      "step": 7680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5125885129819042e-05,
+      "loss": 3.2047,
+      "step": 7690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.5145554681353268e-05,
+      "loss": 3.2261,
+      "step": 7700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5165224232887491e-05,
+      "loss": 3.2214,
+      "step": 7710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5184893784421717e-05,
+      "loss": 3.1784,
+      "step": 7720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5204563335955942e-05,
+      "loss": 3.1084,
+      "step": 7730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5224232887490166e-05,
+      "loss": 3.2025,
+      "step": 7740
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.5243902439024392e-05,
+      "loss": 3.1893,
+      "step": 7750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5263571990558616e-05,
+      "loss": 3.2087,
+      "step": 7760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.528324154209284e-05,
+      "loss": 3.1918,
+      "step": 7770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5302911093627067e-05,
+      "loss": 3.1701,
+      "step": 7780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5322580645161292e-05,
+      "loss": 3.2062,
+      "step": 7790
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.5342250196695515e-05,
+      "loss": 3.1886,
+      "step": 7800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.536191974822974e-05,
+      "loss": 3.2148,
+      "step": 7810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5381589299763966e-05,
+      "loss": 3.1718,
+      "step": 7820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.540125885129819e-05,
+      "loss": 3.1608,
+      "step": 7830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.5420928402832417e-05,
+      "loss": 3.2525,
+      "step": 7840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.544059795436664e-05,
+      "loss": 3.1951,
+      "step": 7850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5460267505900868e-05,
+      "loss": 3.2065,
+      "step": 7860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5479937057435094e-05,
+      "loss": 3.1838,
+      "step": 7870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.549960660896932e-05,
+      "loss": 3.19,
+      "step": 7880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.551927616050354e-05,
+      "loss": 3.1829,
+      "step": 7890
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.5538945712037767e-05,
+      "loss": 3.1819,
+      "step": 7900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.5558615263571992e-05,
+      "loss": 3.2,
+      "step": 7910
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.5578284815106218e-05,
+      "loss": 3.1484,
+      "step": 7920
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.5597954366640444e-05,
+      "loss": 3.2176,
+      "step": 7930
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.5617623918174666e-05,
+      "loss": 3.1605,
+      "step": 7940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.563729346970889e-05,
+      "loss": 3.1809,
+      "step": 7950
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5656963021243117e-05,
+      "loss": 3.2037,
+      "step": 7960
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5676632572777342e-05,
+      "loss": 3.1837,
+      "step": 7970
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5696302124311568e-05,
+      "loss": 3.1913,
+      "step": 7980
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.571597167584579e-05,
+      "loss": 3.1783,
+      "step": 7990
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5735641227380016e-05,
+      "loss": 3.169,
+      "step": 8000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.575531077891424e-05,
+      "loss": 3.2182,
+      "step": 8010
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5774980330448467e-05,
+      "loss": 3.1996,
+      "step": 8020
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5794649881982692e-05,
+      "loss": 3.1395,
+      "step": 8030
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5814319433516918e-05,
+      "loss": 3.1939,
+      "step": 8040
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5833988985051144e-05,
+      "loss": 3.1713,
+      "step": 8050
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.585365853658537e-05,
+      "loss": 3.1946,
+      "step": 8060
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.587332808811959e-05,
+      "loss": 3.2095,
+      "step": 8070
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5892997639653817e-05,
+      "loss": 3.2122,
+      "step": 8080
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5912667191188042e-05,
+      "loss": 3.171,
+      "step": 8090
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5932336742722268e-05,
+      "loss": 3.1607,
+      "step": 8100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5952006294256494e-05,
+      "loss": 3.2102,
+      "step": 8110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5971675845790716e-05,
+      "loss": 3.1836,
+      "step": 8120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.599134539732494e-05,
+      "loss": 3.1925,
+      "step": 8130
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6011014948859167e-05,
+      "loss": 3.1723,
+      "step": 8140
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6030684500393392e-05,
+      "loss": 3.204,
+      "step": 8150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6050354051927618e-05,
+      "loss": 3.1313,
+      "step": 8160
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.607002360346184e-05,
+      "loss": 3.224,
+      "step": 8170
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6089693154996066e-05,
+      "loss": 3.2336,
+      "step": 8180
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.610936270653029e-05,
+      "loss": 3.2131,
+      "step": 8190
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 3.1805,
+      "step": 8200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6148701809598742e-05,
+      "loss": 3.1498,
+      "step": 8210
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6168371361132968e-05,
+      "loss": 3.181,
+      "step": 8220
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6188040912667193e-05,
+      "loss": 3.1704,
+      "step": 8230
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.620771046420142e-05,
+      "loss": 3.1509,
+      "step": 8240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6227380015735645e-05,
+      "loss": 3.1824,
+      "step": 8250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6247049567269867e-05,
+      "loss": 3.1833,
+      "step": 8260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6266719118804092e-05,
+      "loss": 3.163,
+      "step": 8270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6286388670338318e-05,
+      "loss": 3.1863,
+      "step": 8280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6306058221872543e-05,
+      "loss": 3.1984,
+      "step": 8290
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.632572777340677e-05,
+      "loss": 3.1714,
+      "step": 8300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.634539732494099e-05,
+      "loss": 3.1382,
+      "step": 8310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6365066876475217e-05,
+      "loss": 3.131,
+      "step": 8320
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6384736428009442e-05,
+      "loss": 3.1752,
+      "step": 8330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6404405979543668e-05,
+      "loss": 3.216,
+      "step": 8340
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6424075531077893e-05,
+      "loss": 3.1728,
+      "step": 8350
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6443745082612116e-05,
+      "loss": 3.1153,
+      "step": 8360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.646341463414634e-05,
+      "loss": 3.193,
+      "step": 8370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6483084185680567e-05,
+      "loss": 3.1629,
+      "step": 8380
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6502753737214792e-05,
+      "loss": 3.2478,
+      "step": 8390
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6522423288749018e-05,
+      "loss": 3.1955,
+      "step": 8400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6542092840283243e-05,
+      "loss": 3.1936,
+      "step": 8410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.656176239181747e-05,
+      "loss": 3.1673,
+      "step": 8420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6581431943351695e-05,
+      "loss": 3.1889,
+      "step": 8430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6601101494885917e-05,
+      "loss": 3.1501,
+      "step": 8440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6620771046420142e-05,
+      "loss": 3.1656,
+      "step": 8450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6640440597954368e-05,
+      "loss": 3.1944,
+      "step": 8460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6660110149488593e-05,
+      "loss": 3.2078,
+      "step": 8470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.667977970102282e-05,
+      "loss": 3.1787,
+      "step": 8480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6699449252557045e-05,
+      "loss": 3.1574,
+      "step": 8490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6719118804091267e-05,
+      "loss": 3.1166,
+      "step": 8500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6738788355625492e-05,
+      "loss": 3.1597,
+      "step": 8510
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6758457907159718e-05,
+      "loss": 3.1147,
+      "step": 8520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6778127458693943e-05,
+      "loss": 3.1914,
+      "step": 8530
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.679779701022817e-05,
+      "loss": 3.1414,
+      "step": 8540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.681746656176239e-05,
+      "loss": 3.1301,
+      "step": 8550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6837136113296617e-05,
+      "loss": 3.1311,
+      "step": 8560
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6856805664830846e-05,
+      "loss": 3.1395,
+      "step": 8570
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6876475216365068e-05,
+      "loss": 3.1607,
+      "step": 8580
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6896144767899293e-05,
+      "loss": 3.1853,
+      "step": 8590
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.691581431943352e-05,
+      "loss": 3.1804,
+      "step": 8600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6935483870967744e-05,
+      "loss": 3.1875,
+      "step": 8610
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.695515342250197e-05,
+      "loss": 3.2255,
+      "step": 8620
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6974822974036192e-05,
+      "loss": 3.2032,
+      "step": 8630
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6994492525570418e-05,
+      "loss": 3.1458,
+      "step": 8640
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7014162077104643e-05,
+      "loss": 3.1697,
+      "step": 8650
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.703383162863887e-05,
+      "loss": 3.1523,
+      "step": 8660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.7053501180173094e-05,
+      "loss": 3.1727,
+      "step": 8670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.7073170731707317e-05,
+      "loss": 3.1465,
+      "step": 8680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.7092840283241542e-05,
+      "loss": 3.157,
+      "step": 8690
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.7112509834775768e-05,
+      "loss": 3.122,
+      "step": 8700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.7132179386309993e-05,
+      "loss": 3.1011,
+      "step": 8710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.715184893784422e-05,
+      "loss": 3.1514,
+      "step": 8720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.717151848937844e-05,
+      "loss": 3.1399,
+      "step": 8730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.719118804091267e-05,
+      "loss": 3.2048,
+      "step": 8740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.7210857592446896e-05,
+      "loss": 3.1481,
+      "step": 8750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.723052714398112e-05,
+      "loss": 3.1829,
+      "step": 8760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.7250196695515343e-05,
+      "loss": 3.2389,
+      "step": 8770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.726986624704957e-05,
+      "loss": 3.1403,
+      "step": 8780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.7289535798583794e-05,
+      "loss": 3.1235,
+      "step": 8790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.730920535011802e-05,
+      "loss": 3.1611,
+      "step": 8800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.7328874901652246e-05,
+      "loss": 3.138,
+      "step": 8810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.7348544453186468e-05,
+      "loss": 3.1561,
+      "step": 8820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7368214004720693e-05,
+      "loss": 3.171,
+      "step": 8830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.738788355625492e-05,
+      "loss": 3.1707,
+      "step": 8840
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7407553107789144e-05,
+      "loss": 3.1642,
+      "step": 8850
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.742722265932337e-05,
+      "loss": 3.1808,
+      "step": 8860
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7446892210857592e-05,
+      "loss": 3.1435,
+      "step": 8870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7466561762391818e-05,
+      "loss": 3.1666,
+      "step": 8880
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7486231313926043e-05,
+      "loss": 3.1183,
+      "step": 8890
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.750590086546027e-05,
+      "loss": 3.1355,
+      "step": 8900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7525570416994494e-05,
+      "loss": 3.1304,
+      "step": 8910
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.754523996852872e-05,
+      "loss": 3.158,
+      "step": 8920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7564909520062946e-05,
+      "loss": 3.1819,
+      "step": 8930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.758457907159717e-05,
+      "loss": 3.1487,
+      "step": 8940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7604248623131393e-05,
+      "loss": 3.1592,
+      "step": 8950
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.762391817466562e-05,
+      "loss": 3.2359,
+      "step": 8960
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7643587726199844e-05,
+      "loss": 3.1754,
+      "step": 8970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.766325727773407e-05,
+      "loss": 3.0982,
+      "step": 8980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7682926829268296e-05,
+      "loss": 3.0819,
+      "step": 8990
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7702596380802518e-05,
+      "loss": 3.1474,
+      "step": 9000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7722265932336743e-05,
+      "loss": 3.1382,
+      "step": 9010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 3.1286,
+      "step": 9020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7761605035405194e-05,
+      "loss": 3.1478,
+      "step": 9030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.778127458693942e-05,
+      "loss": 3.1654,
+      "step": 9040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7800944138473642e-05,
+      "loss": 3.146,
+      "step": 9050
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7820613690007868e-05,
+      "loss": 3.1825,
+      "step": 9060
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7840283241542093e-05,
+      "loss": 3.163,
+      "step": 9070
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.785995279307632e-05,
+      "loss": 3.1081,
+      "step": 9080
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7879622344610544e-05,
+      "loss": 3.1172,
+      "step": 9090
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.789929189614477e-05,
+      "loss": 3.1798,
+      "step": 9100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7918961447678995e-05,
+      "loss": 3.1552,
+      "step": 9110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.793863099921322e-05,
+      "loss": 3.16,
+      "step": 9120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7958300550747447e-05,
+      "loss": 3.1303,
+      "step": 9130
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.797797010228167e-05,
+      "loss": 3.1611,
+      "step": 9140
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7997639653815894e-05,
+      "loss": 3.1065,
+      "step": 9150
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.801730920535012e-05,
+      "loss": 3.1314,
+      "step": 9160
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8036978756884345e-05,
+      "loss": 3.1802,
+      "step": 9170
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.805664830841857e-05,
+      "loss": 3.1244,
+      "step": 9180
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8076317859952793e-05,
+      "loss": 3.2046,
+      "step": 9190
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.809598741148702e-05,
+      "loss": 3.1413,
+      "step": 9200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8115656963021244e-05,
+      "loss": 3.2658,
+      "step": 9210
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.813532651455547e-05,
+      "loss": 3.2212,
+      "step": 9220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8154996066089695e-05,
+      "loss": 3.1532,
+      "step": 9230
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8174665617623918e-05,
+      "loss": 3.1493,
+      "step": 9240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8194335169158143e-05,
+      "loss": 3.1494,
+      "step": 9250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.821400472069237e-05,
+      "loss": 3.2015,
+      "step": 9260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8233674272226594e-05,
+      "loss": 3.217,
+      "step": 9270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.825334382376082e-05,
+      "loss": 3.1972,
+      "step": 9280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8273013375295045e-05,
+      "loss": 3.1276,
+      "step": 9290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.829268292682927e-05,
+      "loss": 3.113,
+      "step": 9300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8312352478363497e-05,
+      "loss": 3.163,
+      "step": 9310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.833202202989772e-05,
+      "loss": 3.1513,
+      "step": 9320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8351691581431944e-05,
+      "loss": 3.0962,
+      "step": 9330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.837136113296617e-05,
+      "loss": 3.1591,
+      "step": 9340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8391030684500395e-05,
+      "loss": 3.1974,
+      "step": 9350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.841070023603462e-05,
+      "loss": 3.1697,
+      "step": 9360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8430369787568843e-05,
+      "loss": 3.1723,
+      "step": 9370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.845003933910307e-05,
+      "loss": 3.1386,
+      "step": 9380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8469708890637294e-05,
+      "loss": 3.12,
+      "step": 9390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.848937844217152e-05,
+      "loss": 3.1467,
+      "step": 9400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8509047993705745e-05,
+      "loss": 3.1388,
+      "step": 9410
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8528717545239968e-05,
+      "loss": 3.195,
+      "step": 9420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 3.1147,
+      "step": 9430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.856805664830842e-05,
+      "loss": 3.139,
+      "step": 9440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8587726199842648e-05,
+      "loss": 3.198,
+      "step": 9450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.860739575137687e-05,
+      "loss": 3.158,
+      "step": 9460
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8627065302911095e-05,
+      "loss": 3.1455,
+      "step": 9470
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.864673485444532e-05,
+      "loss": 3.1386,
+      "step": 9480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8666404405979546e-05,
+      "loss": 3.1766,
+      "step": 9490
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8686073957513772e-05,
+      "loss": 3.134,
+      "step": 9500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8705743509047994e-05,
+      "loss": 3.1505,
+      "step": 9510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.872541306058222e-05,
+      "loss": 3.0932,
+      "step": 9520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8745082612116445e-05,
+      "loss": 3.148,
+      "step": 9530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.876475216365067e-05,
+      "loss": 3.1348,
+      "step": 9540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8784421715184896e-05,
+      "loss": 3.1,
+      "step": 9550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.880409126671912e-05,
+      "loss": 3.1158,
+      "step": 9560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8823760818253344e-05,
+      "loss": 3.1335,
+      "step": 9570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.884343036978757e-05,
+      "loss": 3.1142,
+      "step": 9580
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8863099921321795e-05,
+      "loss": 3.1274,
+      "step": 9590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.888276947285602e-05,
+      "loss": 3.1154,
+      "step": 9600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8902439024390243e-05,
+      "loss": 3.1545,
+      "step": 9610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.892210857592447e-05,
+      "loss": 3.1143,
+      "step": 9620
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8941778127458698e-05,
+      "loss": 3.1142,
+      "step": 9630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.896144767899292e-05,
+      "loss": 3.133,
+      "step": 9640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8981117230527145e-05,
+      "loss": 3.1302,
+      "step": 9650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.900078678206137e-05,
+      "loss": 3.1715,
+      "step": 9660
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9020456333595596e-05,
+      "loss": 3.1131,
+      "step": 9670
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9040125885129822e-05,
+      "loss": 3.1743,
+      "step": 9680
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9059795436664044e-05,
+      "loss": 3.0591,
+      "step": 9690
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.907946498819827e-05,
+      "loss": 3.1452,
+      "step": 9700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9099134539732495e-05,
+      "loss": 3.1396,
+      "step": 9710
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.911880409126672e-05,
+      "loss": 3.1255,
+      "step": 9720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9138473642800946e-05,
+      "loss": 3.1205,
+      "step": 9730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9158143194335172e-05,
+      "loss": 3.1144,
+      "step": 9740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9177812745869394e-05,
+      "loss": 3.1154,
+      "step": 9750
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.919748229740362e-05,
+      "loss": 3.1185,
+      "step": 9760
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9217151848937845e-05,
+      "loss": 3.1229,
+      "step": 9770
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.923682140047207e-05,
+      "loss": 3.0821,
+      "step": 9780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9256490952006296e-05,
+      "loss": 3.1536,
+      "step": 9790
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9276160503540522e-05,
+      "loss": 3.1536,
+      "step": 9800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9295830055074748e-05,
+      "loss": 3.096,
+      "step": 9810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9315499606608973e-05,
+      "loss": 3.098,
+      "step": 9820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9335169158143195e-05,
+      "loss": 3.1006,
+      "step": 9830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 3.1563,
+      "step": 9840
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9374508261211646e-05,
+      "loss": 3.1435,
+      "step": 9850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9394177812745872e-05,
+      "loss": 3.1775,
+      "step": 9860
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9413847364280098e-05,
+      "loss": 3.1356,
+      "step": 9870
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.943351691581432e-05,
+      "loss": 3.1403,
+      "step": 9880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9453186467348545e-05,
+      "loss": 3.1451,
+      "step": 9890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.947285601888277e-05,
+      "loss": 3.2183,
+      "step": 9900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9492525570416996e-05,
+      "loss": 4.6226,
+      "step": 9910
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 4.1217,
+      "step": 9920
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9531864673485444e-05,
+      "loss": 3.5348,
+      "step": 9930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.955153422501967e-05,
+      "loss": 3.2824,
+      "step": 9940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9571203776553895e-05,
+      "loss": 3.1321,
+      "step": 9950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.959087332808812e-05,
+      "loss": 3.1162,
+      "step": 9960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9610542879622346e-05,
+      "loss": 3.1416,
+      "step": 9970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9630212431156572e-05,
+      "loss": 3.1426,
+      "step": 9980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9649881982690797e-05,
+      "loss": 3.1313,
+      "step": 9990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9669551534225023e-05,
+      "loss": 3.1132,
+      "step": 10000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.968922108575925e-05,
+      "loss": 3.0947,
+      "step": 10010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.970889063729347e-05,
+      "loss": 3.1752,
+      "step": 10020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9728560188827696e-05,
+      "loss": 3.1234,
+      "step": 10030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9748229740361922e-05,
+      "loss": 3.1433,
+      "step": 10040
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9767899291896147e-05,
+      "loss": 3.1536,
+      "step": 10050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9787568843430373e-05,
+      "loss": 3.0916,
+      "step": 10060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9807238394964595e-05,
+      "loss": 3.0905,
+      "step": 10070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.982690794649882e-05,
+      "loss": 3.135,
+      "step": 10080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9846577498033046e-05,
+      "loss": 3.1248,
+      "step": 10090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9866247049567272e-05,
+      "loss": 3.1359,
+      "step": 10100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9885916601101497e-05,
+      "loss": 3.1817,
+      "step": 10110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.990558615263572e-05,
+      "loss": 3.0942,
+      "step": 10120
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9925255704169945e-05,
+      "loss": 3.1685,
+      "step": 10130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.994492525570417e-05,
+      "loss": 3.1168,
+      "step": 10140
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9964594807238396e-05,
+      "loss": 3.1437,
+      "step": 10150
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9984264358772622e-05,
+      "loss": 3.1569,
+      "step": 10160
+    },
+    {
+      "epoch": 2.0,
+      "eval_BLEU": 8.2009,
+      "eval_gen_len": 15.0732,
+      "eval_loss": 3.1427183151245117,
+      "eval_runtime": 155.729,
+      "eval_samples_per_second": 69.114,
+      "eval_steps_per_second": 0.276,
+      "step": 10168
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9999562898854796e-05,
+      "loss": 3.1442,
+      "step": 10170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9997377393128773e-05,
+      "loss": 3.0685,
+      "step": 10180
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9995191887402748e-05,
+      "loss": 3.0889,
+      "step": 10190
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9993006381676722e-05,
+      "loss": 3.0139,
+      "step": 10200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9990820875950696e-05,
+      "loss": 3.0516,
+      "step": 10210
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.998863537022467e-05,
+      "loss": 3.0416,
+      "step": 10220
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9986449864498648e-05,
+      "loss": 3.0486,
+      "step": 10230
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9984264358772622e-05,
+      "loss": 3.0776,
+      "step": 10240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9982078853046596e-05,
+      "loss": 3.0309,
+      "step": 10250
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.997989334732057e-05,
+      "loss": 3.0244,
+      "step": 10260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9977707841594548e-05,
+      "loss": 3.0575,
+      "step": 10270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9975522335868522e-05,
+      "loss": 3.0723,
+      "step": 10280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9973336830142496e-05,
+      "loss": 3.0663,
+      "step": 10290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.997115132441647e-05,
+      "loss": 3.0218,
+      "step": 10300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9968965818690448e-05,
+      "loss": 3.0772,
+      "step": 10310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9966780312964422e-05,
+      "loss": 3.0351,
+      "step": 10320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9964594807238396e-05,
+      "loss": 3.0028,
+      "step": 10330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.996240930151237e-05,
+      "loss": 3.0229,
+      "step": 10340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9960223795786348e-05,
+      "loss": 3.0648,
+      "step": 10350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9958038290060322e-05,
+      "loss": 3.0471,
+      "step": 10360
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9955852784334296e-05,
+      "loss": 2.9998,
+      "step": 10370
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.995366727860827e-05,
+      "loss": 3.0413,
+      "step": 10380
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9951481772882245e-05,
+      "loss": 3.0529,
+      "step": 10390
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9949296267156222e-05,
+      "loss": 3.0558,
+      "step": 10400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9947110761430197e-05,
+      "loss": 3.0878,
+      "step": 10410
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.994492525570417e-05,
+      "loss": 3.0594,
+      "step": 10420
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9942739749978145e-05,
+      "loss": 3.0288,
+      "step": 10430
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9940554244252122e-05,
+      "loss": 3.0016,
+      "step": 10440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9938368738526097e-05,
+      "loss": 3.0507,
+      "step": 10450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9936183232800074e-05,
+      "loss": 3.0617,
+      "step": 10460
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9933997727074045e-05,
+      "loss": 3.0386,
+      "step": 10470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9931812221348023e-05,
+      "loss": 3.0727,
+      "step": 10480
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9929626715621997e-05,
+      "loss": 3.064,
+      "step": 10490
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.992744120989597e-05,
+      "loss": 2.9983,
+      "step": 10500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9925255704169945e-05,
+      "loss": 3.0729,
+      "step": 10510
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9923070198443923e-05,
+      "loss": 3.0218,
+      "step": 10520
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9920884692717897e-05,
+      "loss": 3.0482,
+      "step": 10530
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.991869918699187e-05,
+      "loss": 3.042,
+      "step": 10540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9916513681265845e-05,
+      "loss": 3.0306,
+      "step": 10550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.991432817553982e-05,
+      "loss": 3.0377,
+      "step": 10560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9912142669813797e-05,
+      "loss": 3.0116,
+      "step": 10570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.990995716408777e-05,
+      "loss": 3.0412,
+      "step": 10580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.990777165836175e-05,
+      "loss": 3.0444,
+      "step": 10590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.990558615263572e-05,
+      "loss": 3.0106,
+      "step": 10600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9903400646909697e-05,
+      "loss": 3.0345,
+      "step": 10610
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.990121514118367e-05,
+      "loss": 2.9775,
+      "step": 10620
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.989902963545765e-05,
+      "loss": 3.064,
+      "step": 10630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.989684412973162e-05,
+      "loss": 3.0094,
+      "step": 10640
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9894658624005597e-05,
+      "loss": 3.0579,
+      "step": 10650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.989247311827957e-05,
+      "loss": 3.0449,
+      "step": 10660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9890287612553546e-05,
+      "loss": 3.065,
+      "step": 10670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.988810210682752e-05,
+      "loss": 3.0689,
+      "step": 10680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9885916601101497e-05,
+      "loss": 3.0622,
+      "step": 10690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.988373109537547e-05,
+      "loss": 3.0241,
+      "step": 10700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9881545589649446e-05,
+      "loss": 3.0384,
+      "step": 10710
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9879360083923423e-05,
+      "loss": 2.9723,
+      "step": 10720
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9877174578197394e-05,
+      "loss": 3.0361,
+      "step": 10730
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9874989072471372e-05,
+      "loss": 3.0079,
+      "step": 10740
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9872803566745346e-05,
+      "loss": 3.0421,
+      "step": 10750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9870618061019324e-05,
+      "loss": 3.0573,
+      "step": 10760
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9868432555293294e-05,
+      "loss": 3.0499,
+      "step": 10770
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9866247049567272e-05,
+      "loss": 3.0675,
+      "step": 10780
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9864061543841246e-05,
+      "loss": 2.9842,
+      "step": 10790
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9861876038115224e-05,
+      "loss": 3.0765,
+      "step": 10800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9859690532389198e-05,
+      "loss": 3.0598,
+      "step": 10810
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9857505026663172e-05,
+      "loss": 3.0173,
+      "step": 10820
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9855319520937146e-05,
+      "loss": 3.0311,
+      "step": 10830
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.985313401521112e-05,
+      "loss": 3.0337,
+      "step": 10840
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9850948509485098e-05,
+      "loss": 3.0564,
+      "step": 10850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9848763003759072e-05,
+      "loss": 3.0069,
+      "step": 10860
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9846577498033046e-05,
+      "loss": 3.0162,
+      "step": 10870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.984439199230702e-05,
+      "loss": 3.0601,
+      "step": 10880
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9842206486580998e-05,
+      "loss": 2.996,
+      "step": 10890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.984002098085497e-05,
+      "loss": 3.0372,
+      "step": 10900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9837835475128946e-05,
+      "loss": 2.9975,
+      "step": 10910
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.983564996940292e-05,
+      "loss": 3.0339,
+      "step": 10920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9833464463676898e-05,
+      "loss": 3.0624,
+      "step": 10930
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9831278957950872e-05,
+      "loss": 3.0383,
+      "step": 10940
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9829093452224847e-05,
+      "loss": 3.0717,
+      "step": 10950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.982690794649882e-05,
+      "loss": 3.0217,
+      "step": 10960
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.98247224407728e-05,
+      "loss": 3.048,
+      "step": 10970
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9822536935046773e-05,
+      "loss": 3.0045,
+      "step": 10980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9820351429320747e-05,
+      "loss": 3.0558,
+      "step": 10990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.981816592359472e-05,
+      "loss": 3.0578,
+      "step": 11000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9815980417868695e-05,
+      "loss": 3.0745,
+      "step": 11010
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9813794912142673e-05,
+      "loss": 3.0093,
+      "step": 11020
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9811609406416647e-05,
+      "loss": 3.0077,
+      "step": 11030
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.980942390069062e-05,
+      "loss": 2.9504,
+      "step": 11040
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9807238394964595e-05,
+      "loss": 3.026,
+      "step": 11050
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9805052889238573e-05,
+      "loss": 3.0257,
+      "step": 11060
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9802867383512547e-05,
+      "loss": 3.0548,
+      "step": 11070
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.980068187778652e-05,
+      "loss": 3.0473,
+      "step": 11080
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9798496372060495e-05,
+      "loss": 3.0807,
+      "step": 11090
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9796310866334473e-05,
+      "loss": 3.039,
+      "step": 11100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9794125360608447e-05,
+      "loss": 3.0055,
+      "step": 11110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.979193985488242e-05,
+      "loss": 3.0439,
+      "step": 11120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9789754349156395e-05,
+      "loss": 3.0869,
+      "step": 11130
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9787568843430373e-05,
+      "loss": 3.6071,
+      "step": 11140
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9785383337704347e-05,
+      "loss": 5.8187,
+      "step": 11150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.978319783197832e-05,
+      "loss": 4.9683,
+      "step": 11160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9781012326252296e-05,
+      "loss": 3.7634,
+      "step": 11170
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.977882682052627e-05,
+      "loss": 3.1101,
+      "step": 11180
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9776641314800247e-05,
+      "loss": 3.1249,
+      "step": 11190
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.977445580907422e-05,
+      "loss": 3.0732,
+      "step": 11200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9772270303348196e-05,
+      "loss": 3.0718,
+      "step": 11210
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.977008479762217e-05,
+      "loss": 3.0509,
+      "step": 11220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9767899291896147e-05,
+      "loss": 3.0597,
+      "step": 11230
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.976571378617012e-05,
+      "loss": 3.0593,
+      "step": 11240
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9763528280444096e-05,
+      "loss": 3.0651,
+      "step": 11250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.976134277471807e-05,
+      "loss": 3.032,
+      "step": 11260
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9759157268992048e-05,
+      "loss": 3.0044,
+      "step": 11270
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9756971763266022e-05,
+      "loss": 3.0285,
+      "step": 11280
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9754786257539996e-05,
+      "loss": 3.038,
+      "step": 11290
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.975260075181397e-05,
+      "loss": 3.0799,
+      "step": 11300
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9750415246087948e-05,
+      "loss": 2.988,
+      "step": 11310
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9748229740361922e-05,
+      "loss": 2.9852,
+      "step": 11320
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9746044234635896e-05,
+      "loss": 3.0559,
+      "step": 11330
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.974385872890987e-05,
+      "loss": 3.107,
+      "step": 11340
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9741673223183844e-05,
+      "loss": 3.0261,
+      "step": 11350
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9739487717457822e-05,
+      "loss": 3.0103,
+      "step": 11360
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9737302211731796e-05,
+      "loss": 3.0654,
+      "step": 11370
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9735116706005774e-05,
+      "loss": 3.0414,
+      "step": 11380
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9732931200279745e-05,
+      "loss": 3.0769,
+      "step": 11390
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9730745694553722e-05,
+      "loss": 3.0334,
+      "step": 11400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9728560188827696e-05,
+      "loss": 2.9972,
+      "step": 11410
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9726374683101674e-05,
+      "loss": 3.0178,
+      "step": 11420
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9724189177375645e-05,
+      "loss": 3.0048,
+      "step": 11430
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9722003671649622e-05,
+      "loss": 3.0702,
+      "step": 11440
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9719818165923596e-05,
+      "loss": 3.0199,
+      "step": 11450
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.971763266019757e-05,
+      "loss": 3.0431,
+      "step": 11460
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9715447154471545e-05,
+      "loss": 3.0346,
+      "step": 11470
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9713261648745522e-05,
+      "loss": 3.0125,
+      "step": 11480
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9711076143019497e-05,
+      "loss": 3.0099,
+      "step": 11490
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.970889063729347e-05,
+      "loss": 3.0128,
+      "step": 11500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.970670513156745e-05,
+      "loss": 3.0507,
+      "step": 11510
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.970451962584142e-05,
+      "loss": 3.0176,
+      "step": 11520
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9702334120115397e-05,
+      "loss": 3.0267,
+      "step": 11530
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.970014861438937e-05,
+      "loss": 3.1054,
+      "step": 11540
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.969796310866335e-05,
+      "loss": 2.9956,
+      "step": 11550
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.969577760293732e-05,
+      "loss": 3.0139,
+      "step": 11560
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9693592097211297e-05,
+      "loss": 3.0621,
+      "step": 11570
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.969140659148527e-05,
+      "loss": 3.0607,
+      "step": 11580
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.968922108575925e-05,
+      "loss": 3.0595,
+      "step": 11590
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.968703558003322e-05,
+      "loss": 3.0249,
+      "step": 11600
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9684850074307197e-05,
+      "loss": 3.0375,
+      "step": 11610
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.968266456858117e-05,
+      "loss": 3.0048,
+      "step": 11620
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9680479062855145e-05,
+      "loss": 3.0558,
+      "step": 11630
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9678293557129123e-05,
+      "loss": 3.0116,
+      "step": 11640
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9676108051403097e-05,
+      "loss": 3.0001,
+      "step": 11650
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.967392254567707e-05,
+      "loss": 3.04,
+      "step": 11660
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9671737039951045e-05,
+      "loss": 3.0551,
+      "step": 11670
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9669551534225023e-05,
+      "loss": 3.0707,
+      "step": 11680
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9667366028498994e-05,
+      "loss": 3.0527,
+      "step": 11690
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.966518052277297e-05,
+      "loss": 3.0491,
+      "step": 11700
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9662995017046946e-05,
+      "loss": 3.0371,
+      "step": 11710
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9660809511320923e-05,
+      "loss": 3.0394,
+      "step": 11720
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9658624005594897e-05,
+      "loss": 2.9753,
+      "step": 11730
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.965643849986887e-05,
+      "loss": 3.0701,
+      "step": 11740
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9654252994142846e-05,
+      "loss": 3.0795,
+      "step": 11750
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9652067488416823e-05,
+      "loss": 3.0596,
+      "step": 11760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9649881982690797e-05,
+      "loss": 3.0527,
+      "step": 11770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.964769647696477e-05,
+      "loss": 3.0577,
+      "step": 11780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9645510971238746e-05,
+      "loss": 3.1012,
+      "step": 11790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.964332546551272e-05,
+      "loss": 3.017,
+      "step": 11800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9641139959786698e-05,
+      "loss": 3.0498,
+      "step": 11810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9638954454060672e-05,
+      "loss": 3.0344,
+      "step": 11820
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9636768948334646e-05,
+      "loss": 3.0551,
+      "step": 11830
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.963458344260862e-05,
+      "loss": 3.0336,
+      "step": 11840
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9632397936882598e-05,
+      "loss": 2.9962,
+      "step": 11850
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9630212431156572e-05,
+      "loss": 3.0015,
+      "step": 11860
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9628026925430546e-05,
+      "loss": 3.0489,
+      "step": 11870
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.962584141970452e-05,
+      "loss": 3.0704,
+      "step": 11880
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9623655913978498e-05,
+      "loss": 2.9986,
+      "step": 11890
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9621470408252472e-05,
+      "loss": 3.0206,
+      "step": 11900
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9619284902526446e-05,
+      "loss": 2.9854,
+      "step": 11910
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.961709939680042e-05,
+      "loss": 3.0421,
+      "step": 11920
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9614913891074398e-05,
+      "loss": 3.0259,
+      "step": 11930
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9612728385348372e-05,
+      "loss": 3.04,
+      "step": 11940
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9610542879622346e-05,
+      "loss": 2.9691,
+      "step": 11950
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.960835737389632e-05,
+      "loss": 3.0336,
+      "step": 11960
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9606171868170295e-05,
+      "loss": 3.0747,
+      "step": 11970
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9603986362444272e-05,
+      "loss": 2.9872,
+      "step": 11980
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9601800856718246e-05,
+      "loss": 3.0264,
+      "step": 11990
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.959961535099222e-05,
+      "loss": 3.02,
+      "step": 12000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9597429845266195e-05,
+      "loss": 3.0291,
+      "step": 12010
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9595244339540172e-05,
+      "loss": 3.019,
+      "step": 12020
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9593058833814147e-05,
+      "loss": 3.0153,
+      "step": 12030
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.959087332808812e-05,
+      "loss": 3.06,
+      "step": 12040
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9588687822362095e-05,
+      "loss": 3.0533,
+      "step": 12050
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9586502316636073e-05,
+      "loss": 3.0372,
+      "step": 12060
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9584316810910047e-05,
+      "loss": 3.0228,
+      "step": 12070
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.958213130518402e-05,
+      "loss": 3.0684,
+      "step": 12080
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9579945799457995e-05,
+      "loss": 3.0683,
+      "step": 12090
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9577760293731973e-05,
+      "loss": 3.0455,
+      "step": 12100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9575574788005947e-05,
+      "loss": 3.0254,
+      "step": 12110
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.957338928227992e-05,
+      "loss": 3.0453,
+      "step": 12120
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9571203776553895e-05,
+      "loss": 2.9986,
+      "step": 12130
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.956901827082787e-05,
+      "loss": 3.0332,
+      "step": 12140
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9566832765101847e-05,
+      "loss": 3.0375,
+      "step": 12150
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.956464725937582e-05,
+      "loss": 3.0319,
+      "step": 12160
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9562461753649795e-05,
+      "loss": 3.0596,
+      "step": 12170
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.956027624792377e-05,
+      "loss": 3.0742,
+      "step": 12180
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9558090742197747e-05,
+      "loss": 3.0026,
+      "step": 12190
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.955590523647172e-05,
+      "loss": 3.0096,
+      "step": 12200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.95537197307457e-05,
+      "loss": 3.0291,
+      "step": 12210
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.955153422501967e-05,
+      "loss": 3.0573,
+      "step": 12220
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9549348719293647e-05,
+      "loss": 2.9988,
+      "step": 12230
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.954716321356762e-05,
+      "loss": 3.0688,
+      "step": 12240
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9544977707841596e-05,
+      "loss": 3.0038,
+      "step": 12250
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.954279220211557e-05,
+      "loss": 3.0779,
+      "step": 12260
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9540606696389547e-05,
+      "loss": 2.9789,
+      "step": 12270
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.953842119066352e-05,
+      "loss": 3.0277,
+      "step": 12280
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9536235684937496e-05,
+      "loss": 3.0086,
+      "step": 12290
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.953405017921147e-05,
+      "loss": 3.0248,
+      "step": 12300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9531864673485444e-05,
+      "loss": 3.0256,
+      "step": 12310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.952967916775942e-05,
+      "loss": 3.0654,
+      "step": 12320
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9527493662033396e-05,
+      "loss": 3.0393,
+      "step": 12330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9525308156307373e-05,
+      "loss": 3.0236,
+      "step": 12340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9523122650581344e-05,
+      "loss": 3.0381,
+      "step": 12350
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9520937144855322e-05,
+      "loss": 2.9981,
+      "step": 12360
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9518751639129296e-05,
+      "loss": 3.0508,
+      "step": 12370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9516566133403274e-05,
+      "loss": 2.9757,
+      "step": 12380
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9514380627677244e-05,
+      "loss": 2.9869,
+      "step": 12390
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 3.0276,
+      "step": 12400
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9510009616225196e-05,
+      "loss": 3.021,
+      "step": 12410
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.950782411049917e-05,
+      "loss": 3.0168,
+      "step": 12420
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9505638604773148e-05,
+      "loss": 3.0263,
+      "step": 12430
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9503453099047122e-05,
+      "loss": 3.0628,
+      "step": 12440
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9501267593321096e-05,
+      "loss": 3.015,
+      "step": 12450
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.949908208759507e-05,
+      "loss": 3.0203,
+      "step": 12460
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9496896581869048e-05,
+      "loss": 3.034,
+      "step": 12470
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.949471107614302e-05,
+      "loss": 2.97,
+      "step": 12480
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9492525570416996e-05,
+      "loss": 3.0483,
+      "step": 12490
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.949034006469097e-05,
+      "loss": 3.0248,
+      "step": 12500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9488154558964948e-05,
+      "loss": 3.0512,
+      "step": 12510
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.948596905323892e-05,
+      "loss": 3.0589,
+      "step": 12520
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9483783547512896e-05,
+      "loss": 3.0859,
+      "step": 12530
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.948159804178687e-05,
+      "loss": 3.0408,
+      "step": 12540
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9479412536060848e-05,
+      "loss": 3.0027,
+      "step": 12550
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9477227030334822e-05,
+      "loss": 3.0338,
+      "step": 12560
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9475041524608797e-05,
+      "loss": 2.9843,
+      "step": 12570
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.947285601888277e-05,
+      "loss": 3.0565,
+      "step": 12580
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9470670513156745e-05,
+      "loss": 3.0695,
+      "step": 12590
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9468485007430723e-05,
+      "loss": 3.0257,
+      "step": 12600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9466299501704697e-05,
+      "loss": 2.9992,
+      "step": 12610
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.946411399597867e-05,
+      "loss": 3.0237,
+      "step": 12620
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9461928490252645e-05,
+      "loss": 3.0031,
+      "step": 12630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9459742984526623e-05,
+      "loss": 2.9882,
+      "step": 12640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9457557478800593e-05,
+      "loss": 3.0002,
+      "step": 12650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.945537197307457e-05,
+      "loss": 2.998,
+      "step": 12660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9453186467348545e-05,
+      "loss": 2.9261,
+      "step": 12670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9451000961622523e-05,
+      "loss": 3.026,
+      "step": 12680
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9448815455896497e-05,
+      "loss": 2.987,
+      "step": 12690
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.944662995017047e-05,
+      "loss": 3.02,
+      "step": 12700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 3.0163,
+      "step": 12710
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9442258938718423e-05,
+      "loss": 3.0031,
+      "step": 12720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9440073432992397e-05,
+      "loss": 3.0242,
+      "step": 12730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.943788792726637e-05,
+      "loss": 3.0295,
+      "step": 12740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9435702421540346e-05,
+      "loss": 2.9974,
+      "step": 12750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.943351691581432e-05,
+      "loss": 2.9755,
+      "step": 12760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9431331410088297e-05,
+      "loss": 3.0078,
+      "step": 12770
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.942914590436227e-05,
+      "loss": 2.9674,
+      "step": 12780
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9426960398636246e-05,
+      "loss": 2.9923,
+      "step": 12790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.942477489291022e-05,
+      "loss": 3.0663,
+      "step": 12800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9422589387184197e-05,
+      "loss": 3.0103,
+      "step": 12810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.942040388145817e-05,
+      "loss": 3.0266,
+      "step": 12820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9418218375732146e-05,
+      "loss": 3.0446,
+      "step": 12830
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.941603287000612e-05,
+      "loss": 3.0689,
+      "step": 12840
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9413847364280098e-05,
+      "loss": 2.9875,
+      "step": 12850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9411661858554072e-05,
+      "loss": 2.9788,
+      "step": 12860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9409476352828046e-05,
+      "loss": 3.0524,
+      "step": 12870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.940729084710202e-05,
+      "loss": 3.0285,
+      "step": 12880
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9405105341375998e-05,
+      "loss": 2.9884,
+      "step": 12890
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9402919835649972e-05,
+      "loss": 3.007,
+      "step": 12900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9400734329923946e-05,
+      "loss": 3.0067,
+      "step": 12910
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.939854882419792e-05,
+      "loss": 2.9855,
+      "step": 12920
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9396363318471894e-05,
+      "loss": 2.9779,
+      "step": 12930
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9394177812745872e-05,
+      "loss": 3.0585,
+      "step": 12940
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9391992307019846e-05,
+      "loss": 3.0277,
+      "step": 12950
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.938980680129382e-05,
+      "loss": 3.0235,
+      "step": 12960
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9387621295567795e-05,
+      "loss": 2.9951,
+      "step": 12970
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9385435789841772e-05,
+      "loss": 2.9608,
+      "step": 12980
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9383250284115746e-05,
+      "loss": 3.0036,
+      "step": 12990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.938106477838972e-05,
+      "loss": 3.0165,
+      "step": 13000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9378879272663695e-05,
+      "loss": 2.9947,
+      "step": 13010
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9376693766937672e-05,
+      "loss": 3.0153,
+      "step": 13020
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9374508261211646e-05,
+      "loss": 2.9955,
+      "step": 13030
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.937232275548562e-05,
+      "loss": 3.0062,
+      "step": 13040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9370137249759595e-05,
+      "loss": 2.9833,
+      "step": 13050
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9367951744033572e-05,
+      "loss": 2.991,
+      "step": 13060
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9365766238307547e-05,
+      "loss": 3.019,
+      "step": 13070
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.936358073258152e-05,
+      "loss": 2.945,
+      "step": 13080
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9361395226855495e-05,
+      "loss": 2.9911,
+      "step": 13090
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.935920972112947e-05,
+      "loss": 3.0482,
+      "step": 13100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9357024215403447e-05,
+      "loss": 3.0531,
+      "step": 13110
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 2.9991,
+      "step": 13120
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.93526532039514e-05,
+      "loss": 3.0452,
+      "step": 13130
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.935046769822537e-05,
+      "loss": 3.0417,
+      "step": 13140
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9348282192499347e-05,
+      "loss": 3.0256,
+      "step": 13150
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.934609668677332e-05,
+      "loss": 3.0279,
+      "step": 13160
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9343911181047295e-05,
+      "loss": 3.0418,
+      "step": 13170
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.934172567532127e-05,
+      "loss": 2.9756,
+      "step": 13180
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9339540169595247e-05,
+      "loss": 3.0187,
+      "step": 13190
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.933735466386922e-05,
+      "loss": 3.0371,
+      "step": 13200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9335169158143195e-05,
+      "loss": 3.0257,
+      "step": 13210
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.933298365241717e-05,
+      "loss": 2.9733,
+      "step": 13220
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9330798146691147e-05,
+      "loss": 2.9765,
+      "step": 13230
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.932861264096512e-05,
+      "loss": 3.0353,
+      "step": 13240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9326427135239095e-05,
+      "loss": 2.9709,
+      "step": 13250
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9324241629513073e-05,
+      "loss": 2.9872,
+      "step": 13260
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9322056123787044e-05,
+      "loss": 2.9699,
+      "step": 13270
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.931987061806102e-05,
+      "loss": 3.0144,
+      "step": 13280
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9317685112334996e-05,
+      "loss": 2.977,
+      "step": 13290
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9315499606608973e-05,
+      "loss": 2.9623,
+      "step": 13300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9313314100882944e-05,
+      "loss": 2.9974,
+      "step": 13310
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.931112859515692e-05,
+      "loss": 3.0023,
+      "step": 13320
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9308943089430896e-05,
+      "loss": 2.9652,
+      "step": 13330
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.930675758370487e-05,
+      "loss": 2.9951,
+      "step": 13340
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9304572077978844e-05,
+      "loss": 2.9464,
+      "step": 13350
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.930238657225282e-05,
+      "loss": 3.0456,
+      "step": 13360
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9300201066526796e-05,
+      "loss": 3.0045,
+      "step": 13370
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.929801556080077e-05,
+      "loss": 2.9953,
+      "step": 13380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9295830055074748e-05,
+      "loss": 2.9615,
+      "step": 13390
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9293644549348722e-05,
+      "loss": 3.0391,
+      "step": 13400
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9291459043622696e-05,
+      "loss": 3.0124,
+      "step": 13410
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.928927353789667e-05,
+      "loss": 2.9935,
+      "step": 13420
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9287088032170648e-05,
+      "loss": 2.9951,
+      "step": 13430
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.928490252644462e-05,
+      "loss": 3.0121,
+      "step": 13440
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9282717020718596e-05,
+      "loss": 2.9859,
+      "step": 13450
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.928053151499257e-05,
+      "loss": 2.9817,
+      "step": 13460
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9278346009266548e-05,
+      "loss": 3.0057,
+      "step": 13470
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9276160503540522e-05,
+      "loss": 3.0064,
+      "step": 13480
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9273974997814496e-05,
+      "loss": 2.9845,
+      "step": 13490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.927178949208847e-05,
+      "loss": 3.0215,
+      "step": 13500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9269603986362448e-05,
+      "loss": 2.9884,
+      "step": 13510
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9267418480636422e-05,
+      "loss": 2.9658,
+      "step": 13520
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9265232974910396e-05,
+      "loss": 2.9447,
+      "step": 13530
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.926304746918437e-05,
+      "loss": 2.9491,
+      "step": 13540
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9260861963458345e-05,
+      "loss": 3.0132,
+      "step": 13550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9258676457732322e-05,
+      "loss": 2.9666,
+      "step": 13560
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9256490952006296e-05,
+      "loss": 3.0345,
+      "step": 13570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.925430544628027e-05,
+      "loss": 2.9831,
+      "step": 13580
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9252119940554245e-05,
+      "loss": 2.978,
+      "step": 13590
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9249934434828222e-05,
+      "loss": 2.9774,
+      "step": 13600
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9247748929102197e-05,
+      "loss": 3.015,
+      "step": 13610
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.924556342337617e-05,
+      "loss": 2.9827,
+      "step": 13620
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9243377917650145e-05,
+      "loss": 2.9804,
+      "step": 13630
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9241192411924122e-05,
+      "loss": 2.9613,
+      "step": 13640
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9239006906198097e-05,
+      "loss": 2.9613,
+      "step": 13650
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.923682140047207e-05,
+      "loss": 2.9802,
+      "step": 13660
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9234635894746045e-05,
+      "loss": 3.0005,
+      "step": 13670
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9232450389020023e-05,
+      "loss": 3.007,
+      "step": 13680
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9230264883293997e-05,
+      "loss": 3.0103,
+      "step": 13690
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.922807937756797e-05,
+      "loss": 3.0328,
+      "step": 13700
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9225893871841945e-05,
+      "loss": 2.969,
+      "step": 13710
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.922370836611592e-05,
+      "loss": 3.0313,
+      "step": 13720
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9221522860389897e-05,
+      "loss": 3.0209,
+      "step": 13730
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.921933735466387e-05,
+      "loss": 3.0143,
+      "step": 13740
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9217151848937845e-05,
+      "loss": 3.0199,
+      "step": 13750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.921496634321182e-05,
+      "loss": 3.0074,
+      "step": 13760
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9212780837485797e-05,
+      "loss": 2.9964,
+      "step": 13770
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.921059533175977e-05,
+      "loss": 3.0207,
+      "step": 13780
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9208409826033745e-05,
+      "loss": 2.9714,
+      "step": 13790
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.920622432030772e-05,
+      "loss": 3.0166,
+      "step": 13800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9204038814581697e-05,
+      "loss": 2.999,
+      "step": 13810
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.920185330885567e-05,
+      "loss": 2.9884,
+      "step": 13820
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9199667803129646e-05,
+      "loss": 2.9915,
+      "step": 13830
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.919748229740362e-05,
+      "loss": 2.9762,
+      "step": 13840
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9195296791677597e-05,
+      "loss": 3.0525,
+      "step": 13850
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.919311128595157e-05,
+      "loss": 2.9786,
+      "step": 13860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9190925780225546e-05,
+      "loss": 2.9851,
+      "step": 13870
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.918874027449952e-05,
+      "loss": 2.9499,
+      "step": 13880
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9186554768773494e-05,
+      "loss": 2.9665,
+      "step": 13890
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.918436926304747e-05,
+      "loss": 3.0159,
+      "step": 13900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9182183757321446e-05,
+      "loss": 3.007,
+      "step": 13910
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.917999825159542e-05,
+      "loss": 2.9574,
+      "step": 13920
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9177812745869394e-05,
+      "loss": 3.0135,
+      "step": 13930
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9175627240143372e-05,
+      "loss": 2.9718,
+      "step": 13940
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9173441734417346e-05,
+      "loss": 3.005,
+      "step": 13950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.917125622869132e-05,
+      "loss": 3.0075,
+      "step": 13960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9169070722965294e-05,
+      "loss": 3.0011,
+      "step": 13970
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9166885217239272e-05,
+      "loss": 2.9276,
+      "step": 13980
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9164699711513246e-05,
+      "loss": 3.0492,
+      "step": 13990
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.916251420578722e-05,
+      "loss": 2.9977,
+      "step": 14000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9160328700061194e-05,
+      "loss": 3.0396,
+      "step": 14010
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9158143194335172e-05,
+      "loss": 2.9914,
+      "step": 14020
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9155957688609146e-05,
+      "loss": 3.0048,
+      "step": 14030
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.915377218288312e-05,
+      "loss": 3.0148,
+      "step": 14040
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9151586677157095e-05,
+      "loss": 2.9814,
+      "step": 14050
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.914940117143107e-05,
+      "loss": 2.9814,
+      "step": 14060
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9147215665705046e-05,
+      "loss": 3.0044,
+      "step": 14070
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.914503015997902e-05,
+      "loss": 2.9926,
+      "step": 14080
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9142844654252998e-05,
+      "loss": 2.9977,
+      "step": 14090
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.914065914852697e-05,
+      "loss": 2.9796,
+      "step": 14100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9138473642800946e-05,
+      "loss": 2.9657,
+      "step": 14110
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.913628813707492e-05,
+      "loss": 2.93,
+      "step": 14120
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9134102631348895e-05,
+      "loss": 2.9999,
+      "step": 14130
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.913191712562287e-05,
+      "loss": 2.9865,
+      "step": 14140
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9129731619896847e-05,
+      "loss": 3.0261,
+      "step": 14150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.912754611417082e-05,
+      "loss": 2.9293,
+      "step": 14160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9125360608444795e-05,
+      "loss": 2.9829,
+      "step": 14170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9123175102718772e-05,
+      "loss": 2.9609,
+      "step": 14180
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9120989596992747e-05,
+      "loss": 3.0027,
+      "step": 14190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.911880409126672e-05,
+      "loss": 2.936,
+      "step": 14200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9116618585540695e-05,
+      "loss": 2.9816,
+      "step": 14210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9114433079814673e-05,
+      "loss": 2.9899,
+      "step": 14220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9112247574088643e-05,
+      "loss": 2.9907,
+      "step": 14230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.911006206836262e-05,
+      "loss": 2.9728,
+      "step": 14240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9107876562636595e-05,
+      "loss": 2.9968,
+      "step": 14250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9105691056910573e-05,
+      "loss": 2.9291,
+      "step": 14260
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9103505551184544e-05,
+      "loss": 3.0314,
+      "step": 14270
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.910132004545852e-05,
+      "loss": 2.9828,
+      "step": 14280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9099134539732495e-05,
+      "loss": 2.9917,
+      "step": 14290
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.909694903400647e-05,
+      "loss": 2.9913,
+      "step": 14300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9094763528280447e-05,
+      "loss": 3.0019,
+      "step": 14310
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.909257802255442e-05,
+      "loss": 2.9943,
+      "step": 14320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9090392516828395e-05,
+      "loss": 3.0194,
+      "step": 14330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.908820701110237e-05,
+      "loss": 2.9922,
+      "step": 14340
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9086021505376347e-05,
+      "loss": 2.9771,
+      "step": 14350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.908383599965032e-05,
+      "loss": 2.9824,
+      "step": 14360
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9081650493924296e-05,
+      "loss": 3.0103,
+      "step": 14370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.907946498819827e-05,
+      "loss": 2.9583,
+      "step": 14380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9077279482472247e-05,
+      "loss": 2.9998,
+      "step": 14390
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9075093976746218e-05,
+      "loss": 2.9724,
+      "step": 14400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9072908471020196e-05,
+      "loss": 2.9619,
+      "step": 14410
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.907072296529417e-05,
+      "loss": 2.9741,
+      "step": 14420
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9068537459568147e-05,
+      "loss": 3.0177,
+      "step": 14430
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.906635195384212e-05,
+      "loss": 3.0082,
+      "step": 14440
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9064166448116096e-05,
+      "loss": 2.9978,
+      "step": 14450
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.906198094239007e-05,
+      "loss": 3.0108,
+      "step": 14460
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9059795436664044e-05,
+      "loss": 2.9988,
+      "step": 14470
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9057609930938022e-05,
+      "loss": 2.9548,
+      "step": 14480
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9055424425211996e-05,
+      "loss": 3.0114,
+      "step": 14490
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.905323891948597e-05,
+      "loss": 3.0075,
+      "step": 14500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9051053413759944e-05,
+      "loss": 2.9721,
+      "step": 14510
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9048867908033922e-05,
+      "loss": 3.0082,
+      "step": 14520
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9046682402307896e-05,
+      "loss": 2.968,
+      "step": 14530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.904449689658187e-05,
+      "loss": 3.0149,
+      "step": 14540
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9042311390855844e-05,
+      "loss": 2.9517,
+      "step": 14550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9040125885129822e-05,
+      "loss": 2.9871,
+      "step": 14560
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9037940379403796e-05,
+      "loss": 2.9882,
+      "step": 14570
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.903575487367777e-05,
+      "loss": 2.9761,
+      "step": 14580
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9033569367951745e-05,
+      "loss": 2.9712,
+      "step": 14590
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9031383862225722e-05,
+      "loss": 2.9699,
+      "step": 14600
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.9029198356499696e-05,
+      "loss": 2.9684,
+      "step": 14610
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.902701285077367e-05,
+      "loss": 2.9845,
+      "step": 14620
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9024827345047645e-05,
+      "loss": 2.9554,
+      "step": 14630
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.902264183932162e-05,
+      "loss": 2.9786,
+      "step": 14640
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9020456333595596e-05,
+      "loss": 2.9823,
+      "step": 14650
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.901827082786957e-05,
+      "loss": 2.8678,
+      "step": 14660
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9016085322143545e-05,
+      "loss": 2.9934,
+      "step": 14670
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.901389981641752e-05,
+      "loss": 2.9729,
+      "step": 14680
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9011714310691497e-05,
+      "loss": 3.0077,
+      "step": 14690
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.900952880496547e-05,
+      "loss": 2.9847,
+      "step": 14700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9007343299239445e-05,
+      "loss": 2.9704,
+      "step": 14710
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.900515779351342e-05,
+      "loss": 2.9809,
+      "step": 14720
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9002972287787397e-05,
+      "loss": 2.9352,
+      "step": 14730
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.900078678206137e-05,
+      "loss": 2.9661,
+      "step": 14740
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8998601276335345e-05,
+      "loss": 2.9743,
+      "step": 14750
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.899641577060932e-05,
+      "loss": 2.9641,
+      "step": 14760
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8994230264883297e-05,
+      "loss": 2.9487,
+      "step": 14770
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.899204475915727e-05,
+      "loss": 3.0136,
+      "step": 14780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8989859253431245e-05,
+      "loss": 3.0356,
+      "step": 14790
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.898767374770522e-05,
+      "loss": 2.9633,
+      "step": 14800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8985488241979197e-05,
+      "loss": 3.0121,
+      "step": 14810
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.898330273625317e-05,
+      "loss": 3.0063,
+      "step": 14820
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8981117230527145e-05,
+      "loss": 2.9585,
+      "step": 14830
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.897893172480112e-05,
+      "loss": 2.9917,
+      "step": 14840
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8976746219075094e-05,
+      "loss": 2.9597,
+      "step": 14850
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.897456071334907e-05,
+      "loss": 3.0024,
+      "step": 14860
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8972375207623045e-05,
+      "loss": 2.9946,
+      "step": 14870
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8970189701897023e-05,
+      "loss": 3.0078,
+      "step": 14880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8968004196170994e-05,
+      "loss": 2.9661,
+      "step": 14890
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.896581869044497e-05,
+      "loss": 2.9787,
+      "step": 14900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8963633184718946e-05,
+      "loss": 3.0382,
+      "step": 14910
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.896144767899292e-05,
+      "loss": 3.0125,
+      "step": 14920
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8959262173266894e-05,
+      "loss": 2.9665,
+      "step": 14930
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.895707666754087e-05,
+      "loss": 2.9477,
+      "step": 14940
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8954891161814846e-05,
+      "loss": 2.9909,
+      "step": 14950
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.895270565608882e-05,
+      "loss": 3.0211,
+      "step": 14960
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8950520150362794e-05,
+      "loss": 2.9625,
+      "step": 14970
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.894833464463677e-05,
+      "loss": 2.9894,
+      "step": 14980
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8946149138910746e-05,
+      "loss": 2.9816,
+      "step": 14990
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.894396363318472e-05,
+      "loss": 2.9483,
+      "step": 15000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8941778127458698e-05,
+      "loss": 2.949,
+      "step": 15010
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.893959262173267e-05,
+      "loss": 2.986,
+      "step": 15020
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8937407116006646e-05,
+      "loss": 2.9313,
+      "step": 15030
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.893522161028062e-05,
+      "loss": 2.9812,
+      "step": 15040
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8933036104554598e-05,
+      "loss": 2.9154,
+      "step": 15050
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.893085059882857e-05,
+      "loss": 2.9492,
+      "step": 15060
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8928665093102546e-05,
+      "loss": 2.9517,
+      "step": 15070
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.892647958737652e-05,
+      "loss": 2.9475,
+      "step": 15080
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8924294081650494e-05,
+      "loss": 2.9801,
+      "step": 15090
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.892210857592447e-05,
+      "loss": 2.9688,
+      "step": 15100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8919923070198446e-05,
+      "loss": 2.9683,
+      "step": 15110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.891773756447242e-05,
+      "loss": 2.9709,
+      "step": 15120
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8915552058746395e-05,
+      "loss": 2.9539,
+      "step": 15130
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8913366553020372e-05,
+      "loss": 2.9683,
+      "step": 15140
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8911181047294346e-05,
+      "loss": 3.0039,
+      "step": 15150
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.890899554156832e-05,
+      "loss": 3.0135,
+      "step": 15160
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8906810035842295e-05,
+      "loss": 2.9249,
+      "step": 15170
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8904624530116272e-05,
+      "loss": 2.971,
+      "step": 15180
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8902439024390243e-05,
+      "loss": 2.9475,
+      "step": 15190
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.890025351866422e-05,
+      "loss": 2.9327,
+      "step": 15200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8898068012938195e-05,
+      "loss": 2.9526,
+      "step": 15210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8895882507212172e-05,
+      "loss": 3.0271,
+      "step": 15220
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8893697001486147e-05,
+      "loss": 3.1039,
+      "step": 15230
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.889151149576012e-05,
+      "loss": 3.0391,
+      "step": 15240
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8889325990034095e-05,
+      "loss": 3.0366,
+      "step": 15250
+    },
+    {
+      "epoch": 3.0,
+      "eval_BLEU": 7.6429,
+      "eval_gen_len": 12.3545,
+      "eval_loss": 3.0847198963165283,
+      "eval_runtime": 156.1394,
+      "eval_samples_per_second": 68.932,
+      "eval_steps_per_second": 0.275,
+      "step": 15252
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.888714048430807e-05,
+      "loss": 2.8745,
+      "step": 15260
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8884954978582047e-05,
+      "loss": 2.8766,
+      "step": 15270
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.888276947285602e-05,
+      "loss": 2.8827,
+      "step": 15280
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8880583967129995e-05,
+      "loss": 2.882,
+      "step": 15290
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.887839846140397e-05,
+      "loss": 2.8814,
+      "step": 15300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8876212955677947e-05,
+      "loss": 2.8163,
+      "step": 15310
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.887402744995192e-05,
+      "loss": 2.9269,
+      "step": 15320
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8871841944225895e-05,
+      "loss": 2.8851,
+      "step": 15330
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.886965643849987e-05,
+      "loss": 2.8412,
+      "step": 15340
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8867470932773847e-05,
+      "loss": 2.8651,
+      "step": 15350
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.886528542704782e-05,
+      "loss": 2.8483,
+      "step": 15360
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8863099921321795e-05,
+      "loss": 2.9415,
+      "step": 15370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.886091441559577e-05,
+      "loss": 2.9485,
+      "step": 15380
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8858728909869747e-05,
+      "loss": 2.9097,
+      "step": 15390
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.885654340414372e-05,
+      "loss": 2.8796,
+      "step": 15400
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8854357898417695e-05,
+      "loss": 2.9197,
+      "step": 15410
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.885217239269167e-05,
+      "loss": 2.9857,
+      "step": 15420
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8849986886965644e-05,
+      "loss": 2.928,
+      "step": 15430
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.884780138123962e-05,
+      "loss": 2.9315,
+      "step": 15440
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8845615875513596e-05,
+      "loss": 2.9323,
+      "step": 15450
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.884343036978757e-05,
+      "loss": 2.9205,
+      "step": 15460
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8841244864061544e-05,
+      "loss": 2.866,
+      "step": 15470
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.883905935833552e-05,
+      "loss": 2.8467,
+      "step": 15480
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8836873852609496e-05,
+      "loss": 2.8937,
+      "step": 15490
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.883468834688347e-05,
+      "loss": 2.888,
+      "step": 15500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8832502841157444e-05,
+      "loss": 2.8923,
+      "step": 15510
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.883031733543142e-05,
+      "loss": 2.9008,
+      "step": 15520
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8828131829705396e-05,
+      "loss": 2.9359,
+      "step": 15530
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.882594632397937e-05,
+      "loss": 2.9028,
+      "step": 15540
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8823760818253344e-05,
+      "loss": 2.9075,
+      "step": 15550
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8821575312527322e-05,
+      "loss": 2.8808,
+      "step": 15560
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8819389806801296e-05,
+      "loss": 2.8815,
+      "step": 15570
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.881720430107527e-05,
+      "loss": 2.8749,
+      "step": 15580
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8815018795349244e-05,
+      "loss": 2.8583,
+      "step": 15590
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.881283328962322e-05,
+      "loss": 2.8803,
+      "step": 15600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8810647783897196e-05,
+      "loss": 2.8965,
+      "step": 15610
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.880846227817117e-05,
+      "loss": 2.9168,
+      "step": 15620
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8806276772445144e-05,
+      "loss": 2.8552,
+      "step": 15630
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.880409126671912e-05,
+      "loss": 2.8854,
+      "step": 15640
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8801905760993096e-05,
+      "loss": 2.8992,
+      "step": 15650
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.879972025526707e-05,
+      "loss": 2.8921,
+      "step": 15660
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8797534749541045e-05,
+      "loss": 2.8551,
+      "step": 15670
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.879534924381502e-05,
+      "loss": 2.8972,
+      "step": 15680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8793163738088996e-05,
+      "loss": 2.9691,
+      "step": 15690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.879097823236297e-05,
+      "loss": 2.9793,
+      "step": 15700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8788792726636945e-05,
+      "loss": 2.942,
+      "step": 15710
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.878660722091092e-05,
+      "loss": 2.9088,
+      "step": 15720
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8784421715184896e-05,
+      "loss": 2.9,
+      "step": 15730
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.878223620945887e-05,
+      "loss": 2.9386,
+      "step": 15740
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8780050703732845e-05,
+      "loss": 2.901,
+      "step": 15750
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.877786519800682e-05,
+      "loss": 2.918,
+      "step": 15760
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8775679692280793e-05,
+      "loss": 2.9024,
+      "step": 15770
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.877349418655477e-05,
+      "loss": 2.9067,
+      "step": 15780
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8771308680828745e-05,
+      "loss": 2.9126,
+      "step": 15790
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.876912317510272e-05,
+      "loss": 2.8803,
+      "step": 15800
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8766937669376693e-05,
+      "loss": 2.8492,
+      "step": 15810
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.876475216365067e-05,
+      "loss": 2.9144,
+      "step": 15820
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8762566657924645e-05,
+      "loss": 2.9336,
+      "step": 15830
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8760381152198623e-05,
+      "loss": 2.9403,
+      "step": 15840
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8758195646472593e-05,
+      "loss": 2.8989,
+      "step": 15850
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.875601014074657e-05,
+      "loss": 2.9521,
+      "step": 15860
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8753824635020545e-05,
+      "loss": 2.952,
+      "step": 15870
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.875163912929452e-05,
+      "loss": 2.9294,
+      "step": 15880
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8749453623568494e-05,
+      "loss": 2.8909,
+      "step": 15890
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.874726811784247e-05,
+      "loss": 2.9077,
+      "step": 15900
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8745082612116445e-05,
+      "loss": 2.933,
+      "step": 15910
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.874289710639042e-05,
+      "loss": 2.8836,
+      "step": 15920
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8740711600664397e-05,
+      "loss": 2.9034,
+      "step": 15930
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.873852609493837e-05,
+      "loss": 2.8931,
+      "step": 15940
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8736340589212345e-05,
+      "loss": 2.9004,
+      "step": 15950
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.873415508348632e-05,
+      "loss": 2.8652,
+      "step": 15960
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8731969577760297e-05,
+      "loss": 2.9508,
+      "step": 15970
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8729784072034268e-05,
+      "loss": 2.9258,
+      "step": 15980
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8727598566308246e-05,
+      "loss": 2.9455,
+      "step": 15990
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.872541306058222e-05,
+      "loss": 2.8791,
+      "step": 16000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8723227554856197e-05,
+      "loss": 2.8702,
+      "step": 16010
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8721042049130168e-05,
+      "loss": 2.8882,
+      "step": 16020
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8718856543404146e-05,
+      "loss": 2.9456,
+      "step": 16030
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.871667103767812e-05,
+      "loss": 2.909,
+      "step": 16040
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8714485531952094e-05,
+      "loss": 2.9027,
+      "step": 16050
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.871230002622607e-05,
+      "loss": 2.8935,
+      "step": 16060
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8710114520500046e-05,
+      "loss": 2.8902,
+      "step": 16070
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.870792901477402e-05,
+      "loss": 2.8698,
+      "step": 16080
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8705743509047994e-05,
+      "loss": 2.8883,
+      "step": 16090
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8703558003321972e-05,
+      "loss": 2.9347,
+      "step": 16100
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8701372497595946e-05,
+      "loss": 2.9096,
+      "step": 16110
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.869918699186992e-05,
+      "loss": 2.9119,
+      "step": 16120
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8697001486143894e-05,
+      "loss": 2.916,
+      "step": 16130
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8694815980417872e-05,
+      "loss": 2.9225,
+      "step": 16140
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8692630474691846e-05,
+      "loss": 2.9471,
+      "step": 16150
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.869044496896582e-05,
+      "loss": 2.9302,
+      "step": 16160
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8688259463239794e-05,
+      "loss": 2.8868,
+      "step": 16170
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8686073957513772e-05,
+      "loss": 2.8795,
+      "step": 16180
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8683888451787746e-05,
+      "loss": 2.8889,
+      "step": 16190
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.868170294606172e-05,
+      "loss": 2.9497,
+      "step": 16200
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8679517440335695e-05,
+      "loss": 2.8759,
+      "step": 16210
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.867733193460967e-05,
+      "loss": 2.8265,
+      "step": 16220
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8675146428883646e-05,
+      "loss": 2.8655,
+      "step": 16230
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.867296092315762e-05,
+      "loss": 2.8337,
+      "step": 16240
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8670775417431595e-05,
+      "loss": 2.9126,
+      "step": 16250
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.866858991170557e-05,
+      "loss": 2.8612,
+      "step": 16260
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8666404405979546e-05,
+      "loss": 2.9019,
+      "step": 16270
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.866421890025352e-05,
+      "loss": 2.9246,
+      "step": 16280
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8662033394527495e-05,
+      "loss": 2.9029,
+      "step": 16290
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.865984788880147e-05,
+      "loss": 2.8582,
+      "step": 16300
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8657662383075447e-05,
+      "loss": 2.8995,
+      "step": 16310
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.865547687734942e-05,
+      "loss": 2.8899,
+      "step": 16320
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8653291371623395e-05,
+      "loss": 2.8637,
+      "step": 16330
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.865110586589737e-05,
+      "loss": 2.8473,
+      "step": 16340
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8648920360171347e-05,
+      "loss": 2.9005,
+      "step": 16350
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.864673485444532e-05,
+      "loss": 2.875,
+      "step": 16360
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8644549348719295e-05,
+      "loss": 2.895,
+      "step": 16370
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.864236384299327e-05,
+      "loss": 2.9258,
+      "step": 16380
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8640178337267243e-05,
+      "loss": 2.81,
+      "step": 16390
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.863799283154122e-05,
+      "loss": 2.8956,
+      "step": 16400
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8635807325815195e-05,
+      "loss": 2.8891,
+      "step": 16410
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.863362182008917e-05,
+      "loss": 2.9097,
+      "step": 16420
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8631436314363144e-05,
+      "loss": 2.8666,
+      "step": 16430
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.862925080863712e-05,
+      "loss": 2.8926,
+      "step": 16440
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8627065302911095e-05,
+      "loss": 2.9089,
+      "step": 16450
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.862487979718507e-05,
+      "loss": 2.9092,
+      "step": 16460
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8622694291459044e-05,
+      "loss": 2.855,
+      "step": 16470
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.862050878573302e-05,
+      "loss": 2.9185,
+      "step": 16480
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8618323280006995e-05,
+      "loss": 2.8375,
+      "step": 16490
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.861613777428097e-05,
+      "loss": 2.876,
+      "step": 16500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8613952268554944e-05,
+      "loss": 2.9058,
+      "step": 16510
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.861176676282892e-05,
+      "loss": 2.9285,
+      "step": 16520
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8609581257102896e-05,
+      "loss": 2.8824,
+      "step": 16530
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.860739575137687e-05,
+      "loss": 2.8348,
+      "step": 16540
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8605210245650844e-05,
+      "loss": 2.8873,
+      "step": 16550
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8603024739924818e-05,
+      "loss": 2.9018,
+      "step": 16560
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8600839234198796e-05,
+      "loss": 2.9074,
+      "step": 16570
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.859865372847277e-05,
+      "loss": 2.8698,
+      "step": 16580
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8596468222746744e-05,
+      "loss": 2.9038,
+      "step": 16590
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.859428271702072e-05,
+      "loss": 2.9069,
+      "step": 16600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8592097211294696e-05,
+      "loss": 2.8578,
+      "step": 16610
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.858991170556867e-05,
+      "loss": 2.8922,
+      "step": 16620
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8587726199842648e-05,
+      "loss": 2.9186,
+      "step": 16630
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.858554069411662e-05,
+      "loss": 2.9116,
+      "step": 16640
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8583355188390596e-05,
+      "loss": 2.8742,
+      "step": 16650
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.858116968266457e-05,
+      "loss": 2.8579,
+      "step": 16660
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8578984176938544e-05,
+      "loss": 2.9083,
+      "step": 16670
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.857679867121252e-05,
+      "loss": 2.9124,
+      "step": 16680
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8574613165486496e-05,
+      "loss": 2.897,
+      "step": 16690
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.857242765976047e-05,
+      "loss": 2.8972,
+      "step": 16700
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8570242154034445e-05,
+      "loss": 2.8541,
+      "step": 16710
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.856805664830842e-05,
+      "loss": 2.8937,
+      "step": 16720
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8565871142582393e-05,
+      "loss": 2.8608,
+      "step": 16730
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.856368563685637e-05,
+      "loss": 2.8976,
+      "step": 16740
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8561500131130345e-05,
+      "loss": 2.9123,
+      "step": 16750
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8559314625404322e-05,
+      "loss": 2.8798,
+      "step": 16760
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8557129119678293e-05,
+      "loss": 2.9287,
+      "step": 16770
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.855494361395227e-05,
+      "loss": 2.8689,
+      "step": 16780
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8552758108226245e-05,
+      "loss": 2.8705,
+      "step": 16790
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8550572602500222e-05,
+      "loss": 2.9026,
+      "step": 16800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 2.9046,
+      "step": 16810
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.854620159104817e-05,
+      "loss": 2.9039,
+      "step": 16820
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8544016085322145e-05,
+      "loss": 2.8441,
+      "step": 16830
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.854183057959612e-05,
+      "loss": 2.895,
+      "step": 16840
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8539645073870097e-05,
+      "loss": 2.9484,
+      "step": 16850
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.853745956814407e-05,
+      "loss": 2.8338,
+      "step": 16860
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8535274062418045e-05,
+      "loss": 2.9235,
+      "step": 16870
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.853308855669202e-05,
+      "loss": 2.8944,
+      "step": 16880
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8530903050965997e-05,
+      "loss": 2.8963,
+      "step": 16890
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8528717545239968e-05,
+      "loss": 2.8618,
+      "step": 16900
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.8526532039513945e-05,
+      "loss": 2.8961,
+      "step": 16910
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.852434653378792e-05,
+      "loss": 2.9016,
+      "step": 16920
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.8522161028061897e-05,
+      "loss": 2.8485,
+      "step": 16930
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.8519975522335868e-05,
+      "loss": 2.8294,
+      "step": 16940
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.8517790016609845e-05,
+      "loss": 2.9425,
+      "step": 16950
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.851560451088382e-05,
+      "loss": 2.9108,
+      "step": 16960
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.8513419005157797e-05,
+      "loss": 2.8691,
+      "step": 16970
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.851123349943177e-05,
+      "loss": 2.8581,
+      "step": 16980
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.8509047993705745e-05,
+      "loss": 2.8798,
+      "step": 16990
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.850686248797972e-05,
+      "loss": 2.8826,
+      "step": 17000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.8504676982253694e-05,
+      "loss": 2.8729,
+      "step": 17010
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.850249147652767e-05,
+      "loss": 2.8908,
+      "step": 17020
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.8500305970801646e-05,
+      "loss": 2.8921,
+      "step": 17030
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.849812046507562e-05,
+      "loss": 2.8862,
+      "step": 17040
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.8495934959349594e-05,
+      "loss": 2.8981,
+      "step": 17050
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.849374945362357e-05,
+      "loss": 2.8933,
+      "step": 17060
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.8491563947897542e-05,
+      "loss": 2.8837,
+      "step": 17070
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.848937844217152e-05,
+      "loss": 2.9294,
+      "step": 17080
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.8487192936445494e-05,
+      "loss": 2.9356,
+      "step": 17090
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.848500743071947e-05,
+      "loss": 2.8625,
+      "step": 17100
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.8482821924993446e-05,
+      "loss": 2.8878,
+      "step": 17110
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.848063641926742e-05,
+      "loss": 2.8915,
+      "step": 17120
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.8478450913541394e-05,
+      "loss": 2.8839,
+      "step": 17130
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.8476265407815372e-05,
+      "loss": 2.8368,
+      "step": 17140
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.8474079902089346e-05,
+      "loss": 2.832,
+      "step": 17150
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.847189439636332e-05,
+      "loss": 2.8857,
+      "step": 17160
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.8469708890637294e-05,
+      "loss": 2.8366,
+      "step": 17170
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.846752338491127e-05,
+      "loss": 2.8455,
+      "step": 17180
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.8465337879185246e-05,
+      "loss": 2.9319,
+      "step": 17190
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.846315237345922e-05,
+      "loss": 2.8457,
+      "step": 17200
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.8460966867733194e-05,
+      "loss": 2.9074,
+      "step": 17210
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.845878136200717e-05,
+      "loss": 2.9139,
+      "step": 17220
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.8456595856281146e-05,
+      "loss": 2.8926,
+      "step": 17230
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.845441035055512e-05,
+      "loss": 2.842,
+      "step": 17240
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.8452224844829095e-05,
+      "loss": 2.8912,
+      "step": 17250
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.845003933910307e-05,
+      "loss": 2.8894,
+      "step": 17260
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.8447853833377046e-05,
+      "loss": 2.8557,
+      "step": 17270
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.844566832765102e-05,
+      "loss": 2.8793,
+      "step": 17280
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.8443482821924995e-05,
+      "loss": 2.8838,
+      "step": 17290
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.844129731619897e-05,
+      "loss": 2.829,
+      "step": 17300
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.8439111810472946e-05,
+      "loss": 2.898,
+      "step": 17310
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.843692630474692e-05,
+      "loss": 2.9329,
+      "step": 17320
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.8434740799020895e-05,
+      "loss": 2.8507,
+      "step": 17330
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.843255529329487e-05,
+      "loss": 2.9182,
+      "step": 17340
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.8430369787568843e-05,
+      "loss": 2.878,
+      "step": 17350
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.842818428184282e-05,
+      "loss": 2.8685,
+      "step": 17360
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.8425998776116795e-05,
+      "loss": 2.836,
+      "step": 17370
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.842381327039077e-05,
+      "loss": 2.8672,
+      "step": 17380
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.8421627764664743e-05,
+      "loss": 2.8415,
+      "step": 17390
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.841944225893872e-05,
+      "loss": 2.8723,
+      "step": 17400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.8417256753212695e-05,
+      "loss": 2.8517,
+      "step": 17410
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.841507124748667e-05,
+      "loss": 2.8422,
+      "step": 17420
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.8412885741760643e-05,
+      "loss": 2.854,
+      "step": 17430
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.841070023603462e-05,
+      "loss": 2.8627,
+      "step": 17440
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.8408514730308595e-05,
+      "loss": 2.8975,
+      "step": 17450
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.840632922458257e-05,
+      "loss": 2.8926,
+      "step": 17460
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.8404143718856544e-05,
+      "loss": 2.8914,
+      "step": 17470
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.840195821313052e-05,
+      "loss": 2.8457,
+      "step": 17480
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.8399772707404495e-05,
+      "loss": 2.8849,
+      "step": 17490
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.839758720167847e-05,
+      "loss": 2.8824,
+      "step": 17500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.8395401695952444e-05,
+      "loss": 2.8172,
+      "step": 17510
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.8393216190226418e-05,
+      "loss": 2.8518,
+      "step": 17520
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.8391030684500395e-05,
+      "loss": 2.8772,
+      "step": 17530
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.838884517877437e-05,
+      "loss": 2.8811,
+      "step": 17540
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.8386659673048347e-05,
+      "loss": 2.8586,
+      "step": 17550
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.8384474167322318e-05,
+      "loss": 2.829,
+      "step": 17560
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.8382288661596296e-05,
+      "loss": 2.8663,
+      "step": 17570
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.838010315587027e-05,
+      "loss": 2.8755,
+      "step": 17580
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.8377917650144247e-05,
+      "loss": 2.8879,
+      "step": 17590
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.8375732144418218e-05,
+      "loss": 2.8552,
+      "step": 17600
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.8373546638692196e-05,
+      "loss": 2.8833,
+      "step": 17610
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.837136113296617e-05,
+      "loss": 2.869,
+      "step": 17620
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.8369175627240144e-05,
+      "loss": 2.8769,
+      "step": 17630
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.8366990121514118e-05,
+      "loss": 2.8684,
+      "step": 17640
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.8364804615788096e-05,
+      "loss": 2.8277,
+      "step": 17650
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.836261911006207e-05,
+      "loss": 2.8478,
+      "step": 17660
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.8360433604336044e-05,
+      "loss": 2.8861,
+      "step": 17670
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.8358248098610022e-05,
+      "loss": 2.89,
+      "step": 17680
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.8356062592883993e-05,
+      "loss": 2.8865,
+      "step": 17690
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.835387708715797e-05,
+      "loss": 2.88,
+      "step": 17700
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.8351691581431944e-05,
+      "loss": 2.9207,
+      "step": 17710
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.8349506075705922e-05,
+      "loss": 2.8805,
+      "step": 17720
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.8347320569979893e-05,
+      "loss": 2.853,
+      "step": 17730
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.834513506425387e-05,
+      "loss": 2.8684,
+      "step": 17740
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.8342949558527844e-05,
+      "loss": 2.8942,
+      "step": 17750
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.8340764052801822e-05,
+      "loss": 2.9436,
+      "step": 17760
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.8338578547075793e-05,
+      "loss": 2.8645,
+      "step": 17770
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.833639304134977e-05,
+      "loss": 2.8929,
+      "step": 17780
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.8334207535623745e-05,
+      "loss": 2.8974,
+      "step": 17790
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.833202202989772e-05,
+      "loss": 2.8453,
+      "step": 17800
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.8329836524171696e-05,
+      "loss": 2.8857,
+      "step": 17810
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.832765101844567e-05,
+      "loss": 2.8627,
+      "step": 17820
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.8325465512719645e-05,
+      "loss": 2.9064,
+      "step": 17830
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.832328000699362e-05,
+      "loss": 2.8613,
+      "step": 17840
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.8321094501267596e-05,
+      "loss": 2.8284,
+      "step": 17850
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.8318908995541567e-05,
+      "loss": 2.8626,
+      "step": 17860
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.8316723489815545e-05,
+      "loss": 2.8434,
+      "step": 17870
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.831453798408952e-05,
+      "loss": 2.877,
+      "step": 17880
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.8312352478363497e-05,
+      "loss": 2.8311,
+      "step": 17890
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.831016697263747e-05,
+      "loss": 2.8676,
+      "step": 17900
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.8307981466911445e-05,
+      "loss": 2.8944,
+      "step": 17910
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.830579596118542e-05,
+      "loss": 2.8816,
+      "step": 17920
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.8303610455459397e-05,
+      "loss": 2.8425,
+      "step": 17930
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.830142494973337e-05,
+      "loss": 2.9297,
+      "step": 17940
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.8299239444007345e-05,
+      "loss": 2.8969,
+      "step": 17950
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.829705393828132e-05,
+      "loss": 2.8911,
+      "step": 17960
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.8294868432555293e-05,
+      "loss": 2.8869,
+      "step": 17970
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.829268292682927e-05,
+      "loss": 2.8635,
+      "step": 17980
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.8290497421103245e-05,
+      "loss": 2.8761,
+      "step": 17990
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.828831191537722e-05,
+      "loss": 2.9002,
+      "step": 18000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.8286126409651194e-05,
+      "loss": 2.8951,
+      "step": 18010
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.828394090392517e-05,
+      "loss": 2.8641,
+      "step": 18020
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.8281755398199145e-05,
+      "loss": 2.8595,
+      "step": 18030
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.827956989247312e-05,
+      "loss": 2.8865,
+      "step": 18040
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.8277384386747094e-05,
+      "loss": 2.8739,
+      "step": 18050
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.827519888102107e-05,
+      "loss": 2.8412,
+      "step": 18060
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.8273013375295045e-05,
+      "loss": 2.8805,
+      "step": 18070
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.827082786956902e-05,
+      "loss": 2.9092,
+      "step": 18080
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.8268642363842994e-05,
+      "loss": 2.891,
+      "step": 18090
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.826645685811697e-05,
+      "loss": 2.8559,
+      "step": 18100
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.8264271352390946e-05,
+      "loss": 2.8824,
+      "step": 18110
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.826208584666492e-05,
+      "loss": 2.9004,
+      "step": 18120
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.8259900340938894e-05,
+      "loss": 2.8837,
+      "step": 18130
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.8257714835212868e-05,
+      "loss": 2.9114,
+      "step": 18140
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.8255529329486846e-05,
+      "loss": 2.8365,
+      "step": 18150
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.825334382376082e-05,
+      "loss": 2.8752,
+      "step": 18160
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.8251158318034794e-05,
+      "loss": 2.9046,
+      "step": 18170
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.8248972812308768e-05,
+      "loss": 2.8736,
+      "step": 18180
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.8246787306582746e-05,
+      "loss": 2.8278,
+      "step": 18190
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.824460180085672e-05,
+      "loss": 2.8578,
+      "step": 18200
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.8242416295130694e-05,
+      "loss": 2.8609,
+      "step": 18210
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.824023078940467e-05,
+      "loss": 2.8825,
+      "step": 18220
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.8238045283678646e-05,
+      "loss": 2.8578,
+      "step": 18230
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.823585977795262e-05,
+      "loss": 2.8413,
+      "step": 18240
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.8233674272226594e-05,
+      "loss": 2.8386,
+      "step": 18250
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.823148876650057e-05,
+      "loss": 2.8559,
+      "step": 18260
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.8229303260774546e-05,
+      "loss": 2.8925,
+      "step": 18270
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.822711775504852e-05,
+      "loss": 2.8679,
+      "step": 18280
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.8224932249322494e-05,
+      "loss": 2.9133,
+      "step": 18290
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.822274674359647e-05,
+      "loss": 2.8623,
+      "step": 18300
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.8220561237870443e-05,
+      "loss": 2.8873,
+      "step": 18310
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.821837573214442e-05,
+      "loss": 2.9342,
+      "step": 18320
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.8216190226418395e-05,
+      "loss": 2.8994,
+      "step": 18330
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.821400472069237e-05,
+      "loss": 2.8464,
+      "step": 18340
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.8211819214966343e-05,
+      "loss": 2.8833,
+      "step": 18350
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.820963370924032e-05,
+      "loss": 2.8501,
+      "step": 18360
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.8207448203514295e-05,
+      "loss": 2.8455,
+      "step": 18370
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.8205262697788272e-05,
+      "loss": 2.8784,
+      "step": 18380
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.8203077192062243e-05,
+      "loss": 2.8849,
+      "step": 18390
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.820089168633622e-05,
+      "loss": 2.8755,
+      "step": 18400
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.8198706180610195e-05,
+      "loss": 2.8787,
+      "step": 18410
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.819652067488417e-05,
+      "loss": 2.8262,
+      "step": 18420
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.8194335169158143e-05,
+      "loss": 2.8125,
+      "step": 18430
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.819214966343212e-05,
+      "loss": 2.838,
+      "step": 18440
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.8189964157706095e-05,
+      "loss": 2.8889,
+      "step": 18450
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.818777865198007e-05,
+      "loss": 2.9004,
+      "step": 18460
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.8185593146254043e-05,
+      "loss": 2.8555,
+      "step": 18470
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.8183407640528017e-05,
+      "loss": 2.8746,
+      "step": 18480
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8181222134801995e-05,
+      "loss": 2.8562,
+      "step": 18490
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.817903662907597e-05,
+      "loss": 2.8923,
+      "step": 18500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8176851123349947e-05,
+      "loss": 2.8952,
+      "step": 18510
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8174665617623918e-05,
+      "loss": 2.8198,
+      "step": 18520
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8172480111897895e-05,
+      "loss": 2.8187,
+      "step": 18530
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.817029460617187e-05,
+      "loss": 2.8965,
+      "step": 18540
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.8168109100445847e-05,
+      "loss": 2.9718,
+      "step": 18550
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.8165923594719818e-05,
+      "loss": 2.8973,
+      "step": 18560
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.8163738088993795e-05,
+      "loss": 2.8522,
+      "step": 18570
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.816155258326777e-05,
+      "loss": 2.8372,
+      "step": 18580
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.8159367077541744e-05,
+      "loss": 2.9145,
+      "step": 18590
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.815718157181572e-05,
+      "loss": 2.8454,
+      "step": 18600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.8154996066089695e-05,
+      "loss": 2.8733,
+      "step": 18610
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.815281056036367e-05,
+      "loss": 2.8959,
+      "step": 18620
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.8150625054637644e-05,
+      "loss": 2.8947,
+      "step": 18630
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.814843954891162e-05,
+      "loss": 2.8681,
+      "step": 18640
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.8146254043185592e-05,
+      "loss": 2.8704,
+      "step": 18650
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.814406853745957e-05,
+      "loss": 2.8873,
+      "step": 18660
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.8141883031733544e-05,
+      "loss": 2.878,
+      "step": 18670
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.813969752600752e-05,
+      "loss": 2.8525,
+      "step": 18680
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.8137512020281492e-05,
+      "loss": 2.8895,
+      "step": 18690
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.813532651455547e-05,
+      "loss": 2.8502,
+      "step": 18700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.8133141008829444e-05,
+      "loss": 2.8733,
+      "step": 18710
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.813095550310342e-05,
+      "loss": 2.8901,
+      "step": 18720
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.8128769997377396e-05,
+      "loss": 2.8886,
+      "step": 18730
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.812658449165137e-05,
+      "loss": 2.8814,
+      "step": 18740
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.8124398985925344e-05,
+      "loss": 2.8584,
+      "step": 18750
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.812221348019932e-05,
+      "loss": 2.8572,
+      "step": 18760
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.8120027974473296e-05,
+      "loss": 2.8767,
+      "step": 18770
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.811784246874727e-05,
+      "loss": 2.8844,
+      "step": 18780
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.8115656963021244e-05,
+      "loss": 2.8953,
+      "step": 18790
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.811347145729522e-05,
+      "loss": 2.8421,
+      "step": 18800
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.8111285951569196e-05,
+      "loss": 2.8207,
+      "step": 18810
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.8109100445843167e-05,
+      "loss": 2.8872,
+      "step": 18820
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.8106914940117144e-05,
+      "loss": 2.866,
+      "step": 18830
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.810472943439112e-05,
+      "loss": 2.8219,
+      "step": 18840
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.8102543928665096e-05,
+      "loss": 2.9253,
+      "step": 18850
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.810035842293907e-05,
+      "loss": 2.8662,
+      "step": 18860
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.8098172917213045e-05,
+      "loss": 2.8892,
+      "step": 18870
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.809598741148702e-05,
+      "loss": 2.8335,
+      "step": 18880
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8093801905760996e-05,
+      "loss": 2.8337,
+      "step": 18890
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.809161640003497e-05,
+      "loss": 2.8512,
+      "step": 18900
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8089430894308945e-05,
+      "loss": 2.8592,
+      "step": 18910
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.808724538858292e-05,
+      "loss": 2.8804,
+      "step": 18920
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8085059882856893e-05,
+      "loss": 2.9188,
+      "step": 18930
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.808287437713087e-05,
+      "loss": 2.8539,
+      "step": 18940
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.8080688871404845e-05,
+      "loss": 2.8788,
+      "step": 18950
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.807850336567882e-05,
+      "loss": 2.8813,
+      "step": 18960
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.8076317859952793e-05,
+      "loss": 2.8503,
+      "step": 18970
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.807413235422677e-05,
+      "loss": 2.8648,
+      "step": 18980
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8071946848500745e-05,
+      "loss": 2.852,
+      "step": 18990
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.806976134277472e-05,
+      "loss": 2.8358,
+      "step": 19000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8067575837048693e-05,
+      "loss": 2.8516,
+      "step": 19010
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.806539033132267e-05,
+      "loss": 2.8665,
+      "step": 19020
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8063204825596645e-05,
+      "loss": 2.8781,
+      "step": 19030
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.806101931987062e-05,
+      "loss": 2.8799,
+      "step": 19040
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8058833814144593e-05,
+      "loss": 2.8706,
+      "step": 19050
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.805664830841857e-05,
+      "loss": 2.8929,
+      "step": 19060
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8054462802692545e-05,
+      "loss": 2.8769,
+      "step": 19070
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.805227729696652e-05,
+      "loss": 2.8376,
+      "step": 19080
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8050091791240494e-05,
+      "loss": 2.9049,
+      "step": 19090
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8047906285514468e-05,
+      "loss": 2.8736,
+      "step": 19100
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8045720779788445e-05,
+      "loss": 2.8249,
+      "step": 19110
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.804353527406242e-05,
+      "loss": 2.8706,
+      "step": 19120
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8041349768336394e-05,
+      "loss": 2.8794,
+      "step": 19130
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8039164262610368e-05,
+      "loss": 2.859,
+      "step": 19140
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8036978756884345e-05,
+      "loss": 2.9096,
+      "step": 19150
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.803479325115832e-05,
+      "loss": 2.8482,
+      "step": 19160
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8032607745432294e-05,
+      "loss": 2.8648,
+      "step": 19170
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8030422239706268e-05,
+      "loss": 2.8389,
+      "step": 19180
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8028236733980246e-05,
+      "loss": 2.8376,
+      "step": 19190
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.802605122825422e-05,
+      "loss": 2.8339,
+      "step": 19200
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8023865722528194e-05,
+      "loss": 2.863,
+      "step": 19210
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8021680216802168e-05,
+      "loss": 2.9116,
+      "step": 19220
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8019494711076146e-05,
+      "loss": 2.8841,
+      "step": 19230
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.801730920535012e-05,
+      "loss": 2.8782,
+      "step": 19240
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8015123699624094e-05,
+      "loss": 2.9027,
+      "step": 19250
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8012938193898068e-05,
+      "loss": 2.9215,
+      "step": 19260
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8010752688172042e-05,
+      "loss": 2.8884,
+      "step": 19270
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.800856718244602e-05,
+      "loss": 2.8602,
+      "step": 19280
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8006381676719994e-05,
+      "loss": 2.8988,
+      "step": 19290
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8004196170993972e-05,
+      "loss": 2.8621,
+      "step": 19300
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8002010665267943e-05,
+      "loss": 2.8257,
+      "step": 19310
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.799982515954192e-05,
+      "loss": 2.8635,
+      "step": 19320
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7997639653815894e-05,
+      "loss": 2.8185,
+      "step": 19330
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7995454148089872e-05,
+      "loss": 2.9036,
+      "step": 19340
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7993268642363843e-05,
+      "loss": 2.8908,
+      "step": 19350
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.799108313663782e-05,
+      "loss": 2.8631,
+      "step": 19360
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7988897630911794e-05,
+      "loss": 2.8737,
+      "step": 19370
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.798671212518577e-05,
+      "loss": 2.8909,
+      "step": 19380
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7984526619459743e-05,
+      "loss": 2.8404,
+      "step": 19390
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.798234111373372e-05,
+      "loss": 2.8242,
+      "step": 19400
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7980155608007695e-05,
+      "loss": 2.8449,
+      "step": 19410
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.797797010228167e-05,
+      "loss": 2.8981,
+      "step": 19420
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7975784596555646e-05,
+      "loss": 2.8165,
+      "step": 19430
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7973599090829617e-05,
+      "loss": 2.912,
+      "step": 19440
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7971413585103595e-05,
+      "loss": 2.8446,
+      "step": 19450
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.796922807937757e-05,
+      "loss": 2.8952,
+      "step": 19460
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7967042573651546e-05,
+      "loss": 2.915,
+      "step": 19470
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7964857067925517e-05,
+      "loss": 2.8422,
+      "step": 19480
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7962671562199495e-05,
+      "loss": 2.8632,
+      "step": 19490
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.796048605647347e-05,
+      "loss": 2.8663,
+      "step": 19500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7958300550747447e-05,
+      "loss": 2.8847,
+      "step": 19510
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7956115045021417e-05,
+      "loss": 2.8637,
+      "step": 19520
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7953929539295395e-05,
+      "loss": 2.9076,
+      "step": 19530
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.795174403356937e-05,
+      "loss": 2.848,
+      "step": 19540
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7949558527843343e-05,
+      "loss": 2.8289,
+      "step": 19550
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.794737302211732e-05,
+      "loss": 2.8627,
+      "step": 19560
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7945187516391295e-05,
+      "loss": 2.8627,
+      "step": 19570
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.794300201066527e-05,
+      "loss": 2.8429,
+      "step": 19580
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7940816504939243e-05,
+      "loss": 2.9194,
+      "step": 19590
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.793863099921322e-05,
+      "loss": 2.8218,
+      "step": 19600
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7936445493487192e-05,
+      "loss": 2.8683,
+      "step": 19610
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.793425998776117e-05,
+      "loss": 2.8698,
+      "step": 19620
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7932074482035144e-05,
+      "loss": 2.8812,
+      "step": 19630
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.792988897630912e-05,
+      "loss": 2.9118,
+      "step": 19640
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7927703470583095e-05,
+      "loss": 2.8334,
+      "step": 19650
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.792551796485707e-05,
+      "loss": 2.896,
+      "step": 19660
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7923332459131044e-05,
+      "loss": 2.8399,
+      "step": 19670
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.792114695340502e-05,
+      "loss": 2.8622,
+      "step": 19680
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7918961447678995e-05,
+      "loss": 2.8437,
+      "step": 19690
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.791677594195297e-05,
+      "loss": 2.8758,
+      "step": 19700
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7914590436226944e-05,
+      "loss": 2.8693,
+      "step": 19710
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7912404930500918e-05,
+      "loss": 2.8458,
+      "step": 19720
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7910219424774896e-05,
+      "loss": 2.8706,
+      "step": 19730
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.790803391904887e-05,
+      "loss": 2.9014,
+      "step": 19740
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7905848413322844e-05,
+      "loss": 2.8776,
+      "step": 19750
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7903662907596818e-05,
+      "loss": 2.8432,
+      "step": 19760
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7901477401870796e-05,
+      "loss": 2.8729,
+      "step": 19770
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.789929189614477e-05,
+      "loss": 2.8983,
+      "step": 19780
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7897106390418744e-05,
+      "loss": 2.8646,
+      "step": 19790
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7894920884692718e-05,
+      "loss": 2.8569,
+      "step": 19800
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7892735378966696e-05,
+      "loss": 2.8237,
+      "step": 19810
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.789054987324067e-05,
+      "loss": 2.8491,
+      "step": 19820
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7888364367514644e-05,
+      "loss": 2.8939,
+      "step": 19830
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 2.8214,
+      "step": 19840
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7883993356062596e-05,
+      "loss": 2.8696,
+      "step": 19850
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.788180785033657e-05,
+      "loss": 2.8669,
+      "step": 19860
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7879622344610544e-05,
+      "loss": 2.9055,
+      "step": 19870
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.787743683888452e-05,
+      "loss": 2.9014,
+      "step": 19880
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7875251333158493e-05,
+      "loss": 2.8588,
+      "step": 19890
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.787306582743247e-05,
+      "loss": 2.8741,
+      "step": 19900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7870880321706444e-05,
+      "loss": 2.8521,
+      "step": 19910
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.786869481598042e-05,
+      "loss": 2.8712,
+      "step": 19920
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7866509310254393e-05,
+      "loss": 2.9317,
+      "step": 19930
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.786432380452837e-05,
+      "loss": 2.886,
+      "step": 19940
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7862138298802345e-05,
+      "loss": 2.8441,
+      "step": 19950
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.785995279307632e-05,
+      "loss": 2.8891,
+      "step": 19960
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7857767287350293e-05,
+      "loss": 2.8685,
+      "step": 19970
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.785558178162427e-05,
+      "loss": 2.8475,
+      "step": 19980
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7853396275898245e-05,
+      "loss": 2.8824,
+      "step": 19990
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.785121077017222e-05,
+      "loss": 2.8656,
+      "step": 20000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7849025264446193e-05,
+      "loss": 2.8353,
+      "step": 20010
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.784683975872017e-05,
+      "loss": 2.8994,
+      "step": 20020
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7844654252994145e-05,
+      "loss": 2.8847,
+      "step": 20030
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.784246874726812e-05,
+      "loss": 2.9077,
+      "step": 20040
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7840283241542093e-05,
+      "loss": 2.8726,
+      "step": 20050
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7838097735816067e-05,
+      "loss": 2.8524,
+      "step": 20060
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7835912230090045e-05,
+      "loss": 2.8532,
+      "step": 20070
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.783372672436402e-05,
+      "loss": 2.8578,
+      "step": 20080
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7831541218637993e-05,
+      "loss": 2.8666,
+      "step": 20090
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7829355712911968e-05,
+      "loss": 2.8798,
+      "step": 20100
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7827170207185945e-05,
+      "loss": 2.8648,
+      "step": 20110
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.782498470145992e-05,
+      "loss": 2.8789,
+      "step": 20120
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7822799195733897e-05,
+      "loss": 2.856,
+      "step": 20130
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7820613690007868e-05,
+      "loss": 2.8824,
+      "step": 20140
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7818428184281845e-05,
+      "loss": 2.8352,
+      "step": 20150
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.781624267855582e-05,
+      "loss": 2.904,
+      "step": 20160
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7814057172829794e-05,
+      "loss": 2.8294,
+      "step": 20170
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7811871667103768e-05,
+      "loss": 2.8865,
+      "step": 20180
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7809686161377745e-05,
+      "loss": 2.82,
+      "step": 20190
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.780750065565172e-05,
+      "loss": 2.8504,
+      "step": 20200
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7805315149925694e-05,
+      "loss": 2.8429,
+      "step": 20210
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7803129644199668e-05,
+      "loss": 2.9098,
+      "step": 20220
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7800944138473642e-05,
+      "loss": 2.8549,
+      "step": 20230
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.779875863274762e-05,
+      "loss": 2.8435,
+      "step": 20240
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7796573127021594e-05,
+      "loss": 2.8386,
+      "step": 20250
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.779438762129557e-05,
+      "loss": 2.8783,
+      "step": 20260
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7792202115569542e-05,
+      "loss": 2.8882,
+      "step": 20270
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.779001660984352e-05,
+      "loss": 2.8437,
+      "step": 20280
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7787831104117494e-05,
+      "loss": 2.8288,
+      "step": 20290
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.778564559839147e-05,
+      "loss": 2.8284,
+      "step": 20300
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7783460092665442e-05,
+      "loss": 2.9072,
+      "step": 20310
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.778127458693942e-05,
+      "loss": 2.8223,
+      "step": 20320
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7779089081213394e-05,
+      "loss": 2.8977,
+      "step": 20330
+    },
+    {
+      "epoch": 4.0,
+      "eval_BLEU": 8.3775,
+      "eval_gen_len": 11.8754,
+      "eval_loss": 3.066272258758545,
+      "eval_runtime": 185.3868,
+      "eval_samples_per_second": 58.057,
+      "eval_steps_per_second": 0.232,
+      "step": 20336
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7776903575487368e-05,
+      "loss": 2.8212,
+      "step": 20340
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7774718069761346e-05,
+      "loss": 2.7501,
+      "step": 20350
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.777253256403532e-05,
+      "loss": 2.7362,
+      "step": 20360
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7770347058309294e-05,
+      "loss": 2.7787,
+      "step": 20370
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.776816155258327e-05,
+      "loss": 2.7568,
+      "step": 20380
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7765976046857246e-05,
+      "loss": 2.7628,
+      "step": 20390
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7763790541131217e-05,
+      "loss": 2.7413,
+      "step": 20400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7761605035405194e-05,
+      "loss": 2.7227,
+      "step": 20410
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.775941952967917e-05,
+      "loss": 2.7528,
+      "step": 20420
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7757234023953146e-05,
+      "loss": 2.7815,
+      "step": 20430
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7755048518227117e-05,
+      "loss": 2.7772,
+      "step": 20440
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7752863012501094e-05,
+      "loss": 2.7655,
+      "step": 20450
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.775067750677507e-05,
+      "loss": 2.7639,
+      "step": 20460
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7748492001049046e-05,
+      "loss": 2.7499,
+      "step": 20470
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.774630649532302e-05,
+      "loss": 2.7701,
+      "step": 20480
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7744120989596995e-05,
+      "loss": 2.7558,
+      "step": 20490
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 2.6984,
+      "step": 20500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7739749978144943e-05,
+      "loss": 2.7557,
+      "step": 20510
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.773756447241892e-05,
+      "loss": 2.7584,
+      "step": 20520
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7735378966692895e-05,
+      "loss": 2.7716,
+      "step": 20530
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.773319346096687e-05,
+      "loss": 2.7852,
+      "step": 20540
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7731007955240843e-05,
+      "loss": 2.7268,
+      "step": 20550
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.772882244951482e-05,
+      "loss": 2.766,
+      "step": 20560
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.772663694378879e-05,
+      "loss": 2.7338,
+      "step": 20570
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.772445143806277e-05,
+      "loss": 2.7299,
+      "step": 20580
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7722265932336743e-05,
+      "loss": 2.7497,
+      "step": 20590
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.772008042661072e-05,
+      "loss": 2.7265,
+      "step": 20600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7717894920884695e-05,
+      "loss": 2.7751,
+      "step": 20610
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.771570941515867e-05,
+      "loss": 2.7414,
+      "step": 20620
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7713523909432643e-05,
+      "loss": 2.769,
+      "step": 20630
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.771133840370662e-05,
+      "loss": 2.772,
+      "step": 20640
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7709152897980595e-05,
+      "loss": 2.7672,
+      "step": 20650
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.770696739225457e-05,
+      "loss": 2.7732,
+      "step": 20660
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7704781886528544e-05,
+      "loss": 2.7508,
+      "step": 20670
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7702596380802518e-05,
+      "loss": 2.7507,
+      "step": 20680
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7700410875076495e-05,
+      "loss": 2.7178,
+      "step": 20690
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.769822536935047e-05,
+      "loss": 2.796,
+      "step": 20700
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7696039863624444e-05,
+      "loss": 2.8035,
+      "step": 20710
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7693854357898418e-05,
+      "loss": 2.7702,
+      "step": 20720
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7691668852172395e-05,
+      "loss": 2.7669,
+      "step": 20730
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.768948334644637e-05,
+      "loss": 2.7438,
+      "step": 20740
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7687297840720344e-05,
+      "loss": 2.7892,
+      "step": 20750
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7685112334994318e-05,
+      "loss": 2.8217,
+      "step": 20760
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7682926829268296e-05,
+      "loss": 2.7499,
+      "step": 20770
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.768074132354227e-05,
+      "loss": 2.7485,
+      "step": 20780
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7678555817816244e-05,
+      "loss": 2.7508,
+      "step": 20790
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7676370312090218e-05,
+      "loss": 2.6944,
+      "step": 20800
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7674184806364196e-05,
+      "loss": 2.7941,
+      "step": 20810
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.767199930063817e-05,
+      "loss": 2.7835,
+      "step": 20820
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7669813794912144e-05,
+      "loss": 2.7689,
+      "step": 20830
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7667628289186118e-05,
+      "loss": 2.7439,
+      "step": 20840
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7665442783460092e-05,
+      "loss": 2.735,
+      "step": 20850
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.766325727773407e-05,
+      "loss": 2.7463,
+      "step": 20860
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7661071772008044e-05,
+      "loss": 2.782,
+      "step": 20870
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.765888626628202e-05,
+      "loss": 2.8072,
+      "step": 20880
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7656700760555993e-05,
+      "loss": 2.7726,
+      "step": 20890
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.765451525482997e-05,
+      "loss": 2.7566,
+      "step": 20900
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7652329749103944e-05,
+      "loss": 2.7815,
+      "step": 20910
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.765014424337792e-05,
+      "loss": 2.7455,
+      "step": 20920
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7647958737651893e-05,
+      "loss": 2.7302,
+      "step": 20930
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.764577323192587e-05,
+      "loss": 2.7739,
+      "step": 20940
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7643587726199844e-05,
+      "loss": 2.7501,
+      "step": 20950
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.764140222047382e-05,
+      "loss": 2.7849,
+      "step": 20960
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7639216714747793e-05,
+      "loss": 2.7598,
+      "step": 20970
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.763703120902177e-05,
+      "loss": 2.7693,
+      "step": 20980
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7634845703295745e-05,
+      "loss": 2.77,
+      "step": 20990
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.763266019756972e-05,
+      "loss": 2.7685,
+      "step": 21000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7630474691843693e-05,
+      "loss": 2.7364,
+      "step": 21010
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7628289186117667e-05,
+      "loss": 2.7436,
+      "step": 21020
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7626103680391645e-05,
+      "loss": 2.7863,
+      "step": 21030
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.762391817466562e-05,
+      "loss": 2.8131,
+      "step": 21040
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7621732668939596e-05,
+      "loss": 2.759,
+      "step": 21050
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7619547163213567e-05,
+      "loss": 2.7462,
+      "step": 21060
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7617361657487545e-05,
+      "loss": 2.7745,
+      "step": 21070
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.761517615176152e-05,
+      "loss": 2.7814,
+      "step": 21080
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7612990646035497e-05,
+      "loss": 2.7539,
+      "step": 21090
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7610805140309467e-05,
+      "loss": 2.7393,
+      "step": 21100
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7608619634583445e-05,
+      "loss": 2.8004,
+      "step": 21110
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.760643412885742e-05,
+      "loss": 2.7874,
+      "step": 21120
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7604248623131393e-05,
+      "loss": 2.7535,
+      "step": 21130
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7602063117405367e-05,
+      "loss": 2.7782,
+      "step": 21140
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7599877611679345e-05,
+      "loss": 2.7503,
+      "step": 21150
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.759769210595332e-05,
+      "loss": 2.7903,
+      "step": 21160
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7595506600227293e-05,
+      "loss": 2.7282,
+      "step": 21170
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.759332109450127e-05,
+      "loss": 2.7537,
+      "step": 21180
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7591135588775242e-05,
+      "loss": 2.7631,
+      "step": 21190
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.758895008304922e-05,
+      "loss": 2.7755,
+      "step": 21200
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7586764577323194e-05,
+      "loss": 2.7681,
+      "step": 21210
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.758457907159717e-05,
+      "loss": 2.7668,
+      "step": 21220
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7582393565871142e-05,
+      "loss": 2.7901,
+      "step": 21230
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.758020806014512e-05,
+      "loss": 2.7668,
+      "step": 21240
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7578022554419094e-05,
+      "loss": 2.7824,
+      "step": 21250
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.757583704869307e-05,
+      "loss": 2.7723,
+      "step": 21260
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7573651542967045e-05,
+      "loss": 2.7442,
+      "step": 21270
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.757146603724102e-05,
+      "loss": 2.7398,
+      "step": 21280
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7569280531514994e-05,
+      "loss": 2.7383,
+      "step": 21290
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7567095025788968e-05,
+      "loss": 2.755,
+      "step": 21300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7564909520062946e-05,
+      "loss": 2.7702,
+      "step": 21310
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.756272401433692e-05,
+      "loss": 2.7265,
+      "step": 21320
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7560538508610894e-05,
+      "loss": 2.7592,
+      "step": 21330
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7558353002884868e-05,
+      "loss": 2.7728,
+      "step": 21340
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7556167497158846e-05,
+      "loss": 2.7823,
+      "step": 21350
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7553981991432816e-05,
+      "loss": 2.7536,
+      "step": 21360
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7551796485706794e-05,
+      "loss": 2.7236,
+      "step": 21370
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7549610979980768e-05,
+      "loss": 2.7527,
+      "step": 21380
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7547425474254746e-05,
+      "loss": 2.8097,
+      "step": 21390
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.754523996852872e-05,
+      "loss": 2.7445,
+      "step": 21400
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7543054462802694e-05,
+      "loss": 2.7496,
+      "step": 21410
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.754086895707667e-05,
+      "loss": 2.779,
+      "step": 21420
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7538683451350646e-05,
+      "loss": 2.8368,
+      "step": 21430
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.753649794562462e-05,
+      "loss": 2.7858,
+      "step": 21440
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7534312439898594e-05,
+      "loss": 2.7446,
+      "step": 21450
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.753212693417257e-05,
+      "loss": 2.8111,
+      "step": 21460
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7529941428446543e-05,
+      "loss": 2.8021,
+      "step": 21470
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.752775592272052e-05,
+      "loss": 2.7527,
+      "step": 21480
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7525570416994494e-05,
+      "loss": 2.7526,
+      "step": 21490
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.752338491126847e-05,
+      "loss": 2.7761,
+      "step": 21500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7521199405542443e-05,
+      "loss": 2.7332,
+      "step": 21510
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.751901389981642e-05,
+      "loss": 2.7851,
+      "step": 21520
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7516828394090395e-05,
+      "loss": 2.7758,
+      "step": 21530
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.751464288836437e-05,
+      "loss": 2.7811,
+      "step": 21540
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7512457382638343e-05,
+      "loss": 2.773,
+      "step": 21550
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.751027187691232e-05,
+      "loss": 2.7945,
+      "step": 21560
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7508086371186295e-05,
+      "loss": 2.7877,
+      "step": 21570
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.750590086546027e-05,
+      "loss": 2.7569,
+      "step": 21580
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7503715359734243e-05,
+      "loss": 2.7885,
+      "step": 21590
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.750152985400822e-05,
+      "loss": 2.7483,
+      "step": 21600
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7499344348282195e-05,
+      "loss": 2.7769,
+      "step": 21610
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.749715884255617e-05,
+      "loss": 2.7713,
+      "step": 21620
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7494973336830143e-05,
+      "loss": 2.7606,
+      "step": 21630
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7492787831104117e-05,
+      "loss": 2.7938,
+      "step": 21640
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7490602325378095e-05,
+      "loss": 2.805,
+      "step": 21650
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.748841681965207e-05,
+      "loss": 2.7876,
+      "step": 21660
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7486231313926043e-05,
+      "loss": 2.7719,
+      "step": 21670
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7484045808200017e-05,
+      "loss": 2.7939,
+      "step": 21680
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7481860302473995e-05,
+      "loss": 2.812,
+      "step": 21690
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.747967479674797e-05,
+      "loss": 2.7789,
+      "step": 21700
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7477489291021943e-05,
+      "loss": 2.7494,
+      "step": 21710
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7475303785295918e-05,
+      "loss": 2.7317,
+      "step": 21720
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7473118279569895e-05,
+      "loss": 2.7582,
+      "step": 21730
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.747093277384387e-05,
+      "loss": 2.7696,
+      "step": 21740
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7468747268117844e-05,
+      "loss": 2.7553,
+      "step": 21750
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7466561762391818e-05,
+      "loss": 2.7488,
+      "step": 21760
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7464376256665795e-05,
+      "loss": 2.7969,
+      "step": 21770
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.746219075093977e-05,
+      "loss": 2.7853,
+      "step": 21780
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7460005245213744e-05,
+      "loss": 2.7321,
+      "step": 21790
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7457819739487718e-05,
+      "loss": 2.7582,
+      "step": 21800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7455634233761692e-05,
+      "loss": 2.7443,
+      "step": 21810
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.745344872803567e-05,
+      "loss": 2.7409,
+      "step": 21820
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7451263222309644e-05,
+      "loss": 2.8343,
+      "step": 21830
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7449077716583618e-05,
+      "loss": 2.7669,
+      "step": 21840
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7446892210857592e-05,
+      "loss": 2.7623,
+      "step": 21850
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.744470670513157e-05,
+      "loss": 2.7923,
+      "step": 21860
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7442521199405544e-05,
+      "loss": 2.7589,
+      "step": 21870
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.744033569367952e-05,
+      "loss": 2.7546,
+      "step": 21880
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7438150187953492e-05,
+      "loss": 2.7775,
+      "step": 21890
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.743596468222747e-05,
+      "loss": 2.7666,
+      "step": 21900
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7433779176501444e-05,
+      "loss": 2.7497,
+      "step": 21910
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7431593670775418e-05,
+      "loss": 2.8062,
+      "step": 21920
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7429408165049392e-05,
+      "loss": 2.7622,
+      "step": 21930
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.742722265932337e-05,
+      "loss": 2.7645,
+      "step": 21940
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7425037153597344e-05,
+      "loss": 2.7424,
+      "step": 21950
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.742285164787132e-05,
+      "loss": 2.7818,
+      "step": 21960
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7420666142145296e-05,
+      "loss": 2.7737,
+      "step": 21970
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7418480636419267e-05,
+      "loss": 2.7544,
+      "step": 21980
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7416295130693244e-05,
+      "loss": 2.762,
+      "step": 21990
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.741410962496722e-05,
+      "loss": 2.7934,
+      "step": 22000
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7411924119241196e-05,
+      "loss": 2.7753,
+      "step": 22010
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7409738613515167e-05,
+      "loss": 2.7869,
+      "step": 22020
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7407553107789144e-05,
+      "loss": 2.7886,
+      "step": 22030
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.740536760206312e-05,
+      "loss": 2.8308,
+      "step": 22040
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7403182096337096e-05,
+      "loss": 2.7911,
+      "step": 22050
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7400996590611067e-05,
+      "loss": 2.7817,
+      "step": 22060
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7398811084885045e-05,
+      "loss": 2.7907,
+      "step": 22070
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.739662557915902e-05,
+      "loss": 2.7945,
+      "step": 22080
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.7394440073432993e-05,
+      "loss": 2.7484,
+      "step": 22090
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.739225456770697e-05,
+      "loss": 2.7741,
+      "step": 22100
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.7390069061980945e-05,
+      "loss": 2.7462,
+      "step": 22110
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.738788355625492e-05,
+      "loss": 2.7215,
+      "step": 22120
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.7385698050528893e-05,
+      "loss": 2.8085,
+      "step": 22130
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.738351254480287e-05,
+      "loss": 2.8071,
+      "step": 22140
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.738132703907684e-05,
+      "loss": 2.7404,
+      "step": 22150
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.737914153335082e-05,
+      "loss": 2.7912,
+      "step": 22160
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.7376956027624793e-05,
+      "loss": 2.7536,
+      "step": 22170
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.737477052189877e-05,
+      "loss": 2.7676,
+      "step": 22180
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.737258501617274e-05,
+      "loss": 2.7425,
+      "step": 22190
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.737039951044672e-05,
+      "loss": 2.8559,
+      "step": 22200
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.7368214004720693e-05,
+      "loss": 2.7105,
+      "step": 22210
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.736602849899467e-05,
+      "loss": 2.784,
+      "step": 22220
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.7363842993268645e-05,
+      "loss": 2.8045,
+      "step": 22230
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.736165748754262e-05,
+      "loss": 2.7551,
+      "step": 22240
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7359471981816593e-05,
+      "loss": 2.8084,
+      "step": 22250
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7357286476090568e-05,
+      "loss": 2.798,
+      "step": 22260
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7355100970364545e-05,
+      "loss": 2.7852,
+      "step": 22270
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.735291546463852e-05,
+      "loss": 2.74,
+      "step": 22280
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7350729958912494e-05,
+      "loss": 2.8008,
+      "step": 22290
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7348544453186468e-05,
+      "loss": 2.7545,
+      "step": 22300
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7346358947460445e-05,
+      "loss": 2.7567,
+      "step": 22310
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.734417344173442e-05,
+      "loss": 2.8066,
+      "step": 22320
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7341987936008394e-05,
+      "loss": 2.7823,
+      "step": 22330
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.7339802430282368e-05,
+      "loss": 2.796,
+      "step": 22340
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7337616924556345e-05,
+      "loss": 2.7871,
+      "step": 22350
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.733543141883032e-05,
+      "loss": 2.782,
+      "step": 22360
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7333245913104294e-05,
+      "loss": 2.7957,
+      "step": 22370
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7331060407378268e-05,
+      "loss": 2.7959,
+      "step": 22380
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7328874901652246e-05,
+      "loss": 2.8183,
+      "step": 22390
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.732668939592622e-05,
+      "loss": 2.7955,
+      "step": 22400
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7324503890200194e-05,
+      "loss": 2.7168,
+      "step": 22410
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7322318384474168e-05,
+      "loss": 2.7677,
+      "step": 22420
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7320132878748142e-05,
+      "loss": 2.8199,
+      "step": 22430
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.731794737302212e-05,
+      "loss": 2.7785,
+      "step": 22440
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7315761867296094e-05,
+      "loss": 2.8233,
+      "step": 22450
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7313576361570068e-05,
+      "loss": 2.7746,
+      "step": 22460
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7311390855844042e-05,
+      "loss": 2.7357,
+      "step": 22470
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.730920535011802e-05,
+      "loss": 2.7563,
+      "step": 22480
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7307019844391994e-05,
+      "loss": 2.7385,
+      "step": 22490
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.730483433866597e-05,
+      "loss": 2.8098,
+      "step": 22500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7302648832939943e-05,
+      "loss": 2.7103,
+      "step": 22510
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.730046332721392e-05,
+      "loss": 2.768,
+      "step": 22520
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7298277821487894e-05,
+      "loss": 2.7973,
+      "step": 22530
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.729609231576187e-05,
+      "loss": 2.7735,
+      "step": 22540
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7293906810035843e-05,
+      "loss": 2.7438,
+      "step": 22550
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.729172130430982e-05,
+      "loss": 2.7622,
+      "step": 22560
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7289535798583794e-05,
+      "loss": 2.7967,
+      "step": 22570
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.728735029285777e-05,
+      "loss": 2.7828,
+      "step": 22580
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7285164787131743e-05,
+      "loss": 2.8031,
+      "step": 22590
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.7282979281405717e-05,
+      "loss": 2.7868,
+      "step": 22600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.7280793775679695e-05,
+      "loss": 2.7849,
+      "step": 22610
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.727860826995367e-05,
+      "loss": 2.7676,
+      "step": 22620
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.7276422764227643e-05,
+      "loss": 2.7852,
+      "step": 22630
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.7274237258501617e-05,
+      "loss": 2.7837,
+      "step": 22640
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.7272051752775595e-05,
+      "loss": 2.7916,
+      "step": 22650
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.726986624704957e-05,
+      "loss": 2.7964,
+      "step": 22660
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.7267680741323546e-05,
+      "loss": 2.8146,
+      "step": 22670
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.7265495235597517e-05,
+      "loss": 2.7533,
+      "step": 22680
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.7263309729871495e-05,
+      "loss": 2.7947,
+      "step": 22690
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.726112422414547e-05,
+      "loss": 2.783,
+      "step": 22700
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.7258938718419443e-05,
+      "loss": 2.7891,
+      "step": 22710
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.7256753212693417e-05,
+      "loss": 2.7629,
+      "step": 22720
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.7254567706967395e-05,
+      "loss": 2.7208,
+      "step": 22730
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.725238220124137e-05,
+      "loss": 2.7616,
+      "step": 22740
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.7250196695515343e-05,
+      "loss": 2.7653,
+      "step": 22750
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.7248011189789318e-05,
+      "loss": 2.7638,
+      "step": 22760
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.724582568406329e-05,
+      "loss": 2.7894,
+      "step": 22770
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.724364017833727e-05,
+      "loss": 2.7909,
+      "step": 22780
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.7241454672611243e-05,
+      "loss": 2.7999,
+      "step": 22790
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.723926916688522e-05,
+      "loss": 2.777,
+      "step": 22800
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.7237083661159192e-05,
+      "loss": 2.7864,
+      "step": 22810
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.723489815543317e-05,
+      "loss": 2.7297,
+      "step": 22820
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.7232712649707144e-05,
+      "loss": 2.7969,
+      "step": 22830
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.723052714398112e-05,
+      "loss": 2.7879,
+      "step": 22840
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.7228341638255092e-05,
+      "loss": 2.7433,
+      "step": 22850
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.722615613252907e-05,
+      "loss": 2.7605,
+      "step": 22860
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.7223970626803044e-05,
+      "loss": 2.8017,
+      "step": 22870
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.7221785121077018e-05,
+      "loss": 2.7489,
+      "step": 22880
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.7219599615350992e-05,
+      "loss": 2.7597,
+      "step": 22890
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.721741410962497e-05,
+      "loss": 2.7449,
+      "step": 22900
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7215228603898944e-05,
+      "loss": 2.7445,
+      "step": 22910
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7213043098172918e-05,
+      "loss": 2.7708,
+      "step": 22920
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7210857592446896e-05,
+      "loss": 2.7925,
+      "step": 22930
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7208672086720866e-05,
+      "loss": 2.7919,
+      "step": 22940
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7206486580994844e-05,
+      "loss": 2.7722,
+      "step": 22950
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.7204301075268818e-05,
+      "loss": 2.7751,
+      "step": 22960
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.7202115569542796e-05,
+      "loss": 2.8083,
+      "step": 22970
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.7199930063816767e-05,
+      "loss": 2.739,
+      "step": 22980
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.7197744558090744e-05,
+      "loss": 2.8252,
+      "step": 22990
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.7195559052364718e-05,
+      "loss": 2.7614,
+      "step": 23000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.7193373546638696e-05,
+      "loss": 2.8066,
+      "step": 23010
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.719118804091267e-05,
+      "loss": 2.7784,
+      "step": 23020
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.7189002535186644e-05,
+      "loss": 2.7469,
+      "step": 23030
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.718681702946062e-05,
+      "loss": 2.7678,
+      "step": 23040
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.7184631523734593e-05,
+      "loss": 2.752,
+      "step": 23050
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.718244601800857e-05,
+      "loss": 2.7448,
+      "step": 23060
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.7180260512282544e-05,
+      "loss": 2.7961,
+      "step": 23070
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.717807500655652e-05,
+      "loss": 2.7517,
+      "step": 23080
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.7175889500830493e-05,
+      "loss": 2.7902,
+      "step": 23090
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.717370399510447e-05,
+      "loss": 2.7687,
+      "step": 23100
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.717151848937844e-05,
+      "loss": 2.7358,
+      "step": 23110
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.716933298365242e-05,
+      "loss": 2.7463,
+      "step": 23120
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7167147477926393e-05,
+      "loss": 2.8112,
+      "step": 23130
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.716496197220037e-05,
+      "loss": 2.7673,
+      "step": 23140
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7162776466474345e-05,
+      "loss": 2.7523,
+      "step": 23150
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.716059096074832e-05,
+      "loss": 2.7632,
+      "step": 23160
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7158405455022293e-05,
+      "loss": 2.7869,
+      "step": 23170
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.715621994929627e-05,
+      "loss": 2.8235,
+      "step": 23180
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7154034443570245e-05,
+      "loss": 2.7623,
+      "step": 23190
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.715184893784422e-05,
+      "loss": 2.7939,
+      "step": 23200
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7149663432118193e-05,
+      "loss": 2.7812,
+      "step": 23210
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7147477926392167e-05,
+      "loss": 2.7874,
+      "step": 23220
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7145292420666145e-05,
+      "loss": 2.8077,
+      "step": 23230
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.714310691494012e-05,
+      "loss": 2.765,
+      "step": 23240
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7140921409214093e-05,
+      "loss": 2.7618,
+      "step": 23250
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.7138735903488067e-05,
+      "loss": 2.7897,
+      "step": 23260
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.7136550397762045e-05,
+      "loss": 2.7727,
+      "step": 23270
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.713436489203602e-05,
+      "loss": 2.7606,
+      "step": 23280
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.7132179386309993e-05,
+      "loss": 2.7664,
+      "step": 23290
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.7129993880583968e-05,
+      "loss": 2.814,
+      "step": 23300
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.7127808374857945e-05,
+      "loss": 2.7685,
+      "step": 23310
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.712562286913192e-05,
+      "loss": 2.7868,
+      "step": 23320
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7123437363405893e-05,
+      "loss": 2.8323,
+      "step": 23330
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7121251857679868e-05,
+      "loss": 2.7782,
+      "step": 23340
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7119066351953845e-05,
+      "loss": 2.7834,
+      "step": 23350
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.711688084622782e-05,
+      "loss": 2.8029,
+      "step": 23360
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.7114695340501794e-05,
+      "loss": 2.8242,
+      "step": 23370
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.7112509834775768e-05,
+      "loss": 2.781,
+      "step": 23380
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.7110324329049742e-05,
+      "loss": 2.8018,
+      "step": 23390
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.710813882332372e-05,
+      "loss": 2.7603,
+      "step": 23400
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.7105953317597694e-05,
+      "loss": 2.8111,
+      "step": 23410
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.7103767811871668e-05,
+      "loss": 2.8208,
+      "step": 23420
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.7101582306145642e-05,
+      "loss": 2.7948,
+      "step": 23430
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.709939680041962e-05,
+      "loss": 2.7985,
+      "step": 23440
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.7097211294693594e-05,
+      "loss": 2.8122,
+      "step": 23450
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.7095025788967568e-05,
+      "loss": 2.7687,
+      "step": 23460
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.7092840283241542e-05,
+      "loss": 2.7851,
+      "step": 23470
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.709065477751552e-05,
+      "loss": 2.7726,
+      "step": 23480
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.7088469271789494e-05,
+      "loss": 2.7347,
+      "step": 23490
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.7086283766063468e-05,
+      "loss": 2.7858,
+      "step": 23500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.7084098260337442e-05,
+      "loss": 2.781,
+      "step": 23510
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.708191275461142e-05,
+      "loss": 2.7408,
+      "step": 23520
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.7079727248885394e-05,
+      "loss": 2.7709,
+      "step": 23530
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.7077541743159368e-05,
+      "loss": 2.7357,
+      "step": 23540
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.7075356237433342e-05,
+      "loss": 2.7737,
+      "step": 23550
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.7073170731707317e-05,
+      "loss": 2.7869,
+      "step": 23560
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.7070985225981294e-05,
+      "loss": 2.7878,
+      "step": 23570
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.706879972025527e-05,
+      "loss": 2.7439,
+      "step": 23580
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.7066614214529243e-05,
+      "loss": 2.8058,
+      "step": 23590
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.7064428708803217e-05,
+      "loss": 2.7805,
+      "step": 23600
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.7062243203077194e-05,
+      "loss": 2.7642,
+      "step": 23610
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.706005769735117e-05,
+      "loss": 2.763,
+      "step": 23620
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.7057872191625143e-05,
+      "loss": 2.7632,
+      "step": 23630
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.7055686685899117e-05,
+      "loss": 2.8417,
+      "step": 23640
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.7053501180173094e-05,
+      "loss": 2.7816,
+      "step": 23650
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.705131567444707e-05,
+      "loss": 2.7768,
+      "step": 23660
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.7049130168721043e-05,
+      "loss": 2.7957,
+      "step": 23670
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.7046944662995017e-05,
+      "loss": 2.7904,
+      "step": 23680
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.7044759157268995e-05,
+      "loss": 2.7469,
+      "step": 23690
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.704257365154297e-05,
+      "loss": 2.7553,
+      "step": 23700
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.7040388145816943e-05,
+      "loss": 2.7744,
+      "step": 23710
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.703820264009092e-05,
+      "loss": 2.7554,
+      "step": 23720
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.703601713436489e-05,
+      "loss": 2.8427,
+      "step": 23730
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.703383162863887e-05,
+      "loss": 2.7977,
+      "step": 23740
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.7031646122912843e-05,
+      "loss": 2.7691,
+      "step": 23750
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.702946061718682e-05,
+      "loss": 2.7784,
+      "step": 23760
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.702727511146079e-05,
+      "loss": 2.7636,
+      "step": 23770
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.702508960573477e-05,
+      "loss": 2.7685,
+      "step": 23780
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.7022904100008743e-05,
+      "loss": 2.7575,
+      "step": 23790
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.702071859428272e-05,
+      "loss": 2.7698,
+      "step": 23800
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.701853308855669e-05,
+      "loss": 2.7705,
+      "step": 23810
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.701634758283067e-05,
+      "loss": 2.7782,
+      "step": 23820
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.7014162077104643e-05,
+      "loss": 2.8105,
+      "step": 23830
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.7011976571378618e-05,
+      "loss": 2.755,
+      "step": 23840
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.7009791065652595e-05,
+      "loss": 2.738,
+      "step": 23850
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.700760555992657e-05,
+      "loss": 2.7496,
+      "step": 23860
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.7005420054200543e-05,
+      "loss": 2.8034,
+      "step": 23870
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.7003234548474518e-05,
+      "loss": 2.7505,
+      "step": 23880
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.7001049042748495e-05,
+      "loss": 2.8121,
+      "step": 23890
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6998863537022466e-05,
+      "loss": 2.7478,
+      "step": 23900
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6996678031296444e-05,
+      "loss": 2.804,
+      "step": 23910
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6994492525570418e-05,
+      "loss": 2.7843,
+      "step": 23920
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6992307019844395e-05,
+      "loss": 2.7593,
+      "step": 23930
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6990121514118366e-05,
+      "loss": 2.7819,
+      "step": 23940
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6987936008392344e-05,
+      "loss": 2.7449,
+      "step": 23950
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6985750502666318e-05,
+      "loss": 2.7569,
+      "step": 23960
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6983564996940295e-05,
+      "loss": 2.803,
+      "step": 23970
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.698137949121427e-05,
+      "loss": 2.7996,
+      "step": 23980
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6979193985488244e-05,
+      "loss": 2.7805,
+      "step": 23990
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6977008479762218e-05,
+      "loss": 2.8239,
+      "step": 24000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6974822974036192e-05,
+      "loss": 2.7677,
+      "step": 24010
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.697263746831017e-05,
+      "loss": 2.8192,
+      "step": 24020
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6970451962584144e-05,
+      "loss": 2.7946,
+      "step": 24030
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6968266456858118e-05,
+      "loss": 2.7887,
+      "step": 24040
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6966080951132092e-05,
+      "loss": 2.7845,
+      "step": 24050
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.696389544540607e-05,
+      "loss": 2.7913,
+      "step": 24060
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6961709939680044e-05,
+      "loss": 2.8116,
+      "step": 24070
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6959524433954018e-05,
+      "loss": 2.7525,
+      "step": 24080
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6957338928227992e-05,
+      "loss": 2.7787,
+      "step": 24090
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.695515342250197e-05,
+      "loss": 2.8093,
+      "step": 24100
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6952967916775944e-05,
+      "loss": 2.7184,
+      "step": 24110
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.695078241104992e-05,
+      "loss": 2.7737,
+      "step": 24120
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.6948596905323893e-05,
+      "loss": 2.796,
+      "step": 24130
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.694641139959787e-05,
+      "loss": 2.8096,
+      "step": 24140
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.6944225893871844e-05,
+      "loss": 2.7932,
+      "step": 24150
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.694204038814582e-05,
+      "loss": 2.7887,
+      "step": 24160
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.6939854882419793e-05,
+      "loss": 2.784,
+      "step": 24170
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6937669376693767e-05,
+      "loss": 2.7874,
+      "step": 24180
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6935483870967744e-05,
+      "loss": 2.7735,
+      "step": 24190
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.693329836524172e-05,
+      "loss": 2.7948,
+      "step": 24200
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6931112859515693e-05,
+      "loss": 2.7867,
+      "step": 24210
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6928927353789667e-05,
+      "loss": 2.7586,
+      "step": 24220
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6926741848063645e-05,
+      "loss": 2.7819,
+      "step": 24230
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.692455634233762e-05,
+      "loss": 2.7631,
+      "step": 24240
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6922370836611593e-05,
+      "loss": 2.7825,
+      "step": 24250
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6920185330885567e-05,
+      "loss": 2.6973,
+      "step": 24260
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6917999825159545e-05,
+      "loss": 2.7529,
+      "step": 24270
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.691581431943352e-05,
+      "loss": 2.8214,
+      "step": 24280
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.6913628813707493e-05,
+      "loss": 2.801,
+      "step": 24290
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.6911443307981467e-05,
+      "loss": 2.7794,
+      "step": 24300
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.6909257802255445e-05,
+      "loss": 2.784,
+      "step": 24310
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.690707229652942e-05,
+      "loss": 2.7872,
+      "step": 24320
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6904886790803393e-05,
+      "loss": 2.7856,
+      "step": 24330
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6902701285077367e-05,
+      "loss": 2.7269,
+      "step": 24340
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.690051577935134e-05,
+      "loss": 2.7542,
+      "step": 24350
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.689833027362532e-05,
+      "loss": 2.8096,
+      "step": 24360
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6896144767899293e-05,
+      "loss": 2.7775,
+      "step": 24370
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6893959262173268e-05,
+      "loss": 2.7776,
+      "step": 24380
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6891773756447242e-05,
+      "loss": 2.7807,
+      "step": 24390
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.688958825072122e-05,
+      "loss": 2.79,
+      "step": 24400
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6887402744995193e-05,
+      "loss": 2.7682,
+      "step": 24410
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6885217239269168e-05,
+      "loss": 2.7753,
+      "step": 24420
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.6883031733543142e-05,
+      "loss": 2.791,
+      "step": 24430
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.688084622781712e-05,
+      "loss": 2.8123,
+      "step": 24440
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.6878660722091094e-05,
+      "loss": 2.7724,
+      "step": 24450
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.6876475216365068e-05,
+      "loss": 2.7327,
+      "step": 24460
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.6874289710639042e-05,
+      "loss": 2.7728,
+      "step": 24470
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.687210420491302e-05,
+      "loss": 2.7394,
+      "step": 24480
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6869918699186994e-05,
+      "loss": 2.7847,
+      "step": 24490
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6867733193460968e-05,
+      "loss": 2.793,
+      "step": 24500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6865547687734942e-05,
+      "loss": 2.7484,
+      "step": 24510
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6863362182008916e-05,
+      "loss": 2.8304,
+      "step": 24520
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6861176676282894e-05,
+      "loss": 2.7687,
+      "step": 24530
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6858991170556868e-05,
+      "loss": 2.7954,
+      "step": 24540
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6856805664830846e-05,
+      "loss": 2.7322,
+      "step": 24550
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6854620159104816e-05,
+      "loss": 2.7765,
+      "step": 24560
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6852434653378794e-05,
+      "loss": 2.8048,
+      "step": 24570
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6850249147652768e-05,
+      "loss": 2.7927,
+      "step": 24580
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6848063641926742e-05,
+      "loss": 2.7579,
+      "step": 24590
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6845878136200717e-05,
+      "loss": 2.8082,
+      "step": 24600
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6843692630474694e-05,
+      "loss": 2.7654,
+      "step": 24610
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.684150712474867e-05,
+      "loss": 2.7747,
+      "step": 24620
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6839321619022643e-05,
+      "loss": 2.7451,
+      "step": 24630
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.6837136113296617e-05,
+      "loss": 2.8232,
+      "step": 24640
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.6834950607570594e-05,
+      "loss": 2.818,
+      "step": 24650
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.683276510184457e-05,
+      "loss": 2.7592,
+      "step": 24660
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.6830579596118543e-05,
+      "loss": 2.7792,
+      "step": 24670
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.682839409039252e-05,
+      "loss": 2.7791,
+      "step": 24680
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.682620858466649e-05,
+      "loss": 2.7739,
+      "step": 24690
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.682402307894047e-05,
+      "loss": 2.7982,
+      "step": 24700
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.6821837573214443e-05,
+      "loss": 2.7829,
+      "step": 24710
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.681965206748842e-05,
+      "loss": 2.7517,
+      "step": 24720
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.681746656176239e-05,
+      "loss": 2.7314,
+      "step": 24730
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.681528105603637e-05,
+      "loss": 2.76,
+      "step": 24740
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6813095550310343e-05,
+      "loss": 2.7675,
+      "step": 24750
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6810910044584317e-05,
+      "loss": 2.8262,
+      "step": 24760
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6808724538858295e-05,
+      "loss": 2.8027,
+      "step": 24770
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.680653903313227e-05,
+      "loss": 2.7996,
+      "step": 24780
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6804353527406243e-05,
+      "loss": 2.7227,
+      "step": 24790
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6802168021680217e-05,
+      "loss": 2.8057,
+      "step": 24800
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6799982515954195e-05,
+      "loss": 2.7778,
+      "step": 24810
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.679779701022817e-05,
+      "loss": 2.8073,
+      "step": 24820
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6795611504502143e-05,
+      "loss": 2.8057,
+      "step": 24830
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.6793425998776117e-05,
+      "loss": 2.7553,
+      "step": 24840
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.6791240493050095e-05,
+      "loss": 2.7508,
+      "step": 24850
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.6789054987324066e-05,
+      "loss": 2.7749,
+      "step": 24860
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.6786869481598043e-05,
+      "loss": 2.7528,
+      "step": 24870
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.6784683975872017e-05,
+      "loss": 2.7463,
+      "step": 24880
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6782498470145995e-05,
+      "loss": 2.7742,
+      "step": 24890
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.678031296441997e-05,
+      "loss": 2.739,
+      "step": 24900
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6778127458693943e-05,
+      "loss": 2.7847,
+      "step": 24910
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6775941952967918e-05,
+      "loss": 2.7831,
+      "step": 24920
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6773756447241892e-05,
+      "loss": 2.7599,
+      "step": 24930
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.677157094151587e-05,
+      "loss": 2.7317,
+      "step": 24940
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6769385435789844e-05,
+      "loss": 2.7929,
+      "step": 24950
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6767199930063818e-05,
+      "loss": 2.7609,
+      "step": 24960
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6765014424337792e-05,
+      "loss": 2.7739,
+      "step": 24970
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.676282891861177e-05,
+      "loss": 2.8176,
+      "step": 24980
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6760643412885744e-05,
+      "loss": 2.7215,
+      "step": 24990
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6758457907159718e-05,
+      "loss": 2.7884,
+      "step": 25000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6756272401433692e-05,
+      "loss": 2.819,
+      "step": 25010
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.675408689570767e-05,
+      "loss": 2.7952,
+      "step": 25020
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6751901389981644e-05,
+      "loss": 2.7411,
+      "step": 25030
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6749715884255618e-05,
+      "loss": 2.7123,
+      "step": 25040
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6747530378529592e-05,
+      "loss": 2.7536,
+      "step": 25050
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.674534487280357e-05,
+      "loss": 2.7732,
+      "step": 25060
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6743159367077544e-05,
+      "loss": 2.7924,
+      "step": 25070
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6740973861351518e-05,
+      "loss": 2.7555,
+      "step": 25080
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6738788355625492e-05,
+      "loss": 2.764,
+      "step": 25090
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.673660284989947e-05,
+      "loss": 2.7306,
+      "step": 25100
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6734417344173444e-05,
+      "loss": 2.7791,
+      "step": 25110
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6732231838447418e-05,
+      "loss": 2.7242,
+      "step": 25120
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6730046332721392e-05,
+      "loss": 2.7609,
+      "step": 25130
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6727860826995367e-05,
+      "loss": 2.8009,
+      "step": 25140
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6725675321269344e-05,
+      "loss": 2.8399,
+      "step": 25150
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.672348981554332e-05,
+      "loss": 2.765,
+      "step": 25160
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6721304309817293e-05,
+      "loss": 2.7607,
+      "step": 25170
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6719118804091267e-05,
+      "loss": 2.7711,
+      "step": 25180
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6716933298365244e-05,
+      "loss": 2.7712,
+      "step": 25190
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.671474779263922e-05,
+      "loss": 2.8238,
+      "step": 25200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6712562286913193e-05,
+      "loss": 2.7618,
+      "step": 25210
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6710376781187167e-05,
+      "loss": 2.7575,
+      "step": 25220
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6708191275461144e-05,
+      "loss": 2.7628,
+      "step": 25230
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.670600576973512e-05,
+      "loss": 2.8114,
+      "step": 25240
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6703820264009093e-05,
+      "loss": 2.7696,
+      "step": 25250
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6701634758283067e-05,
+      "loss": 2.7974,
+      "step": 25260
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6699449252557045e-05,
+      "loss": 2.7761,
+      "step": 25270
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.669726374683102e-05,
+      "loss": 2.7324,
+      "step": 25280
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6695078241104993e-05,
+      "loss": 2.7322,
+      "step": 25290
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6692892735378967e-05,
+      "loss": 2.7345,
+      "step": 25300
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.669070722965294e-05,
+      "loss": 2.7626,
+      "step": 25310
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.668852172392692e-05,
+      "loss": 2.8103,
+      "step": 25320
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6686336218200893e-05,
+      "loss": 2.8019,
+      "step": 25330
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6684150712474867e-05,
+      "loss": 2.7805,
+      "step": 25340
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.668196520674884e-05,
+      "loss": 2.8168,
+      "step": 25350
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.667977970102282e-05,
+      "loss": 2.76,
+      "step": 25360
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6677594195296793e-05,
+      "loss": 2.7775,
+      "step": 25370
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6675408689570767e-05,
+      "loss": 2.7784,
+      "step": 25380
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.667322318384474e-05,
+      "loss": 2.7641,
+      "step": 25390
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.667103767811872e-05,
+      "loss": 2.8025,
+      "step": 25400
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6668852172392693e-05,
+      "loss": 2.7691,
+      "step": 25410
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 2.7357,
+      "step": 25420
+    },
+    {
+      "epoch": 5.0,
+      "eval_BLEU": 8.191,
+      "eval_gen_len": 11.9758,
+      "eval_loss": 3.0567257404327393,
+      "eval_runtime": 186.5794,
+      "eval_samples_per_second": 57.686,
+      "eval_steps_per_second": 0.23,
+      "step": 25420
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.666448116094064e-05,
+      "loss": 2.6357,
+      "step": 25430
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.666229565521462e-05,
+      "loss": 2.6185,
+      "step": 25440
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6660110149488593e-05,
+      "loss": 2.6912,
+      "step": 25450
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6657924643762568e-05,
+      "loss": 2.678,
+      "step": 25460
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6655739138036545e-05,
+      "loss": 2.664,
+      "step": 25470
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6653553632310516e-05,
+      "loss": 2.6482,
+      "step": 25480
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6651368126584494e-05,
+      "loss": 2.6508,
+      "step": 25490
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6649182620858468e-05,
+      "loss": 2.657,
+      "step": 25500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6646997115132445e-05,
+      "loss": 2.6783,
+      "step": 25510
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6644811609406416e-05,
+      "loss": 2.6683,
+      "step": 25520
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6642626103680394e-05,
+      "loss": 2.6444,
+      "step": 25530
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6640440597954368e-05,
+      "loss": 2.6428,
+      "step": 25540
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6638255092228342e-05,
+      "loss": 2.6204,
+      "step": 25550
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6636069586502316e-05,
+      "loss": 2.6408,
+      "step": 25560
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6633884080776294e-05,
+      "loss": 2.6593,
+      "step": 25570
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6631698575050268e-05,
+      "loss": 2.7163,
+      "step": 25580
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6629513069324242e-05,
+      "loss": 2.6165,
+      "step": 25590
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.662732756359822e-05,
+      "loss": 2.6364,
+      "step": 25600
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6625142057872194e-05,
+      "loss": 2.6844,
+      "step": 25610
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6622956552146168e-05,
+      "loss": 2.6954,
+      "step": 25620
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6620771046420142e-05,
+      "loss": 2.6622,
+      "step": 25630
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.661858554069412e-05,
+      "loss": 2.7043,
+      "step": 25640
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.661640003496809e-05,
+      "loss": 2.6382,
+      "step": 25650
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.6614214529242068e-05,
+      "loss": 2.653,
+      "step": 25660
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.6612029023516042e-05,
+      "loss": 2.6154,
+      "step": 25670
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.660984351779002e-05,
+      "loss": 2.6319,
+      "step": 25680
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.660765801206399e-05,
+      "loss": 2.6246,
+      "step": 25690
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.660547250633797e-05,
+      "loss": 2.6704,
+      "step": 25700
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6603287000611943e-05,
+      "loss": 2.6643,
+      "step": 25710
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6601101494885917e-05,
+      "loss": 2.6456,
+      "step": 25720
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6598915989159894e-05,
+      "loss": 2.6625,
+      "step": 25730
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.659673048343387e-05,
+      "loss": 2.67,
+      "step": 25740
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6594544977707843e-05,
+      "loss": 2.7306,
+      "step": 25750
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6592359471981817e-05,
+      "loss": 2.6891,
+      "step": 25760
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6590173966255794e-05,
+      "loss": 2.6972,
+      "step": 25770
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.658798846052977e-05,
+      "loss": 2.6892,
+      "step": 25780
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6585802954803743e-05,
+      "loss": 2.6963,
+      "step": 25790
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6583617449077717e-05,
+      "loss": 2.7013,
+      "step": 25800
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6581431943351695e-05,
+      "loss": 2.68,
+      "step": 25810
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.657924643762567e-05,
+      "loss": 2.6983,
+      "step": 25820
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6577060931899643e-05,
+      "loss": 2.6673,
+      "step": 25830
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6574875426173617e-05,
+      "loss": 2.6692,
+      "step": 25840
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6572689920447595e-05,
+      "loss": 2.6734,
+      "step": 25850
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.657050441472157e-05,
+      "loss": 2.6804,
+      "step": 25860
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6568318908995543e-05,
+      "loss": 2.7136,
+      "step": 25870
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6566133403269517e-05,
+      "loss": 2.67,
+      "step": 25880
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.656394789754349e-05,
+      "loss": 2.679,
+      "step": 25890
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.656176239181747e-05,
+      "loss": 2.724,
+      "step": 25900
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.6559576886091443e-05,
+      "loss": 2.6311,
+      "step": 25910
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.6557391380365417e-05,
+      "loss": 2.6875,
+      "step": 25920
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.655520587463939e-05,
+      "loss": 2.6941,
+      "step": 25930
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.655302036891337e-05,
+      "loss": 2.6436,
+      "step": 25940
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.6550834863187343e-05,
+      "loss": 2.6826,
+      "step": 25950
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6548649357461317e-05,
+      "loss": 2.695,
+      "step": 25960
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.654646385173529e-05,
+      "loss": 2.6578,
+      "step": 25970
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.654427834600927e-05,
+      "loss": 2.6935,
+      "step": 25980
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6542092840283243e-05,
+      "loss": 2.6486,
+      "step": 25990
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6539907334557218e-05,
+      "loss": 2.6679,
+      "step": 26000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6537721828831192e-05,
+      "loss": 2.6522,
+      "step": 26010
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.653553632310517e-05,
+      "loss": 2.6771,
+      "step": 26020
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6533350817379144e-05,
+      "loss": 2.6913,
+      "step": 26030
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6531165311653118e-05,
+      "loss": 2.6476,
+      "step": 26040
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6528979805927092e-05,
+      "loss": 2.6609,
+      "step": 26050
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6526794300201066e-05,
+      "loss": 2.6387,
+      "step": 26060
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6524608794475044e-05,
+      "loss": 2.6765,
+      "step": 26070
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6522423288749018e-05,
+      "loss": 2.63,
+      "step": 26080
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6520237783022992e-05,
+      "loss": 2.7107,
+      "step": 26090
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6518052277296966e-05,
+      "loss": 2.5994,
+      "step": 26100
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6515866771570944e-05,
+      "loss": 2.6406,
+      "step": 26110
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6513681265844918e-05,
+      "loss": 2.7045,
+      "step": 26120
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6511495760118892e-05,
+      "loss": 2.6698,
+      "step": 26130
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6509310254392866e-05,
+      "loss": 2.6545,
+      "step": 26140
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6507124748666844e-05,
+      "loss": 2.7076,
+      "step": 26150
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6504939242940818e-05,
+      "loss": 2.6467,
+      "step": 26160
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6502753737214792e-05,
+      "loss": 2.6891,
+      "step": 26170
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6500568231488766e-05,
+      "loss": 2.6945,
+      "step": 26180
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6498382725762744e-05,
+      "loss": 2.655,
+      "step": 26190
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6496197220036718e-05,
+      "loss": 2.6567,
+      "step": 26200
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6494011714310692e-05,
+      "loss": 2.6609,
+      "step": 26210
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6491826208584667e-05,
+      "loss": 2.6721,
+      "step": 26220
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.648964070285864e-05,
+      "loss": 2.6933,
+      "step": 26230
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.648745519713262e-05,
+      "loss": 2.6502,
+      "step": 26240
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6485269691406593e-05,
+      "loss": 2.6405,
+      "step": 26250
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.6483084185680567e-05,
+      "loss": 2.6772,
+      "step": 26260
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.648089867995454e-05,
+      "loss": 2.6823,
+      "step": 26270
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.647871317422852e-05,
+      "loss": 2.6719,
+      "step": 26280
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.6476527668502493e-05,
+      "loss": 2.6658,
+      "step": 26290
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.647434216277647e-05,
+      "loss": 2.7213,
+      "step": 26300
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.647215665705044e-05,
+      "loss": 2.7098,
+      "step": 26310
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.646997115132442e-05,
+      "loss": 2.6631,
+      "step": 26320
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6467785645598393e-05,
+      "loss": 2.7199,
+      "step": 26330
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6465600139872367e-05,
+      "loss": 2.6586,
+      "step": 26340
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.646341463414634e-05,
+      "loss": 2.6979,
+      "step": 26350
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.646122912842032e-05,
+      "loss": 2.6701,
+      "step": 26360
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6459043622694293e-05,
+      "loss": 2.689,
+      "step": 26370
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6456858116968267e-05,
+      "loss": 2.678,
+      "step": 26380
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6454672611242245e-05,
+      "loss": 2.6844,
+      "step": 26390
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.645248710551622e-05,
+      "loss": 2.6318,
+      "step": 26400
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6450301599790193e-05,
+      "loss": 2.6742,
+      "step": 26410
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6448116094064167e-05,
+      "loss": 2.6861,
+      "step": 26420
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6445930588338145e-05,
+      "loss": 2.7035,
+      "step": 26430
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6443745082612116e-05,
+      "loss": 2.6581,
+      "step": 26440
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6441559576886093e-05,
+      "loss": 2.7208,
+      "step": 26450
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6439374071160067e-05,
+      "loss": 2.7025,
+      "step": 26460
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.6437188565434045e-05,
+      "loss": 2.6837,
+      "step": 26470
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.6435003059708016e-05,
+      "loss": 2.6736,
+      "step": 26480
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.6432817553981993e-05,
+      "loss": 2.6794,
+      "step": 26490
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.6430632048255968e-05,
+      "loss": 2.6793,
+      "step": 26500
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.642844654252994e-05,
+      "loss": 2.6554,
+      "step": 26510
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.642626103680392e-05,
+      "loss": 2.6879,
+      "step": 26520
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.6424075531077893e-05,
+      "loss": 2.6535,
+      "step": 26530
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.6421890025351868e-05,
+      "loss": 2.6795,
+      "step": 26540
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.6419704519625842e-05,
+      "loss": 2.7067,
+      "step": 26550
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.641751901389982e-05,
+      "loss": 2.67,
+      "step": 26560
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.6415333508173794e-05,
+      "loss": 2.6714,
+      "step": 26570
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.6413148002447768e-05,
+      "loss": 2.6747,
+      "step": 26580
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.6410962496721742e-05,
+      "loss": 2.6983,
+      "step": 26590
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.640877699099572e-05,
+      "loss": 2.686,
+      "step": 26600
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.640659148526969e-05,
+      "loss": 2.609,
+      "step": 26610
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.6404405979543668e-05,
+      "loss": 2.6975,
+      "step": 26620
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.6402220473817642e-05,
+      "loss": 2.6644,
+      "step": 26630
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.640003496809162e-05,
+      "loss": 2.6608,
+      "step": 26640
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.6397849462365594e-05,
+      "loss": 2.6951,
+      "step": 26650
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.6395663956639568e-05,
+      "loss": 2.6277,
+      "step": 26660
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.6393478450913542e-05,
+      "loss": 2.6953,
+      "step": 26670
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.6391292945187516e-05,
+      "loss": 2.6728,
+      "step": 26680
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.6389107439461494e-05,
+      "loss": 2.6265,
+      "step": 26690
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.6386921933735468e-05,
+      "loss": 2.7181,
+      "step": 26700
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.6384736428009442e-05,
+      "loss": 2.6614,
+      "step": 26710
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.6382550922283417e-05,
+      "loss": 2.6502,
+      "step": 26720
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.6380365416557394e-05,
+      "loss": 2.6608,
+      "step": 26730
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.6378179910831368e-05,
+      "loss": 2.7288,
+      "step": 26740
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.6375994405105342e-05,
+      "loss": 2.6478,
+      "step": 26750
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.6373808899379317e-05,
+      "loss": 2.6867,
+      "step": 26760
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.6371623393653294e-05,
+      "loss": 2.6808,
+      "step": 26770
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.636943788792727e-05,
+      "loss": 2.6391,
+      "step": 26780
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.6367252382201243e-05,
+      "loss": 2.6177,
+      "step": 26790
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.6365066876475217e-05,
+      "loss": 2.6973,
+      "step": 26800
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.6362881370749194e-05,
+      "loss": 2.711,
+      "step": 26810
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.636069586502317e-05,
+      "loss": 2.6709,
+      "step": 26820
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.6358510359297143e-05,
+      "loss": 2.6912,
+      "step": 26830
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.6356324853571117e-05,
+      "loss": 2.6529,
+      "step": 26840
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.635413934784509e-05,
+      "loss": 2.6695,
+      "step": 26850
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.635195384211907e-05,
+      "loss": 2.6545,
+      "step": 26860
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.6349768336393043e-05,
+      "loss": 2.6544,
+      "step": 26870
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.6347582830667017e-05,
+      "loss": 2.6844,
+      "step": 26880
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.634539732494099e-05,
+      "loss": 2.688,
+      "step": 26890
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.634321181921497e-05,
+      "loss": 2.6427,
+      "step": 26900
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.6341026313488943e-05,
+      "loss": 2.6628,
+      "step": 26910
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.6338840807762917e-05,
+      "loss": 2.691,
+      "step": 26920
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.633665530203689e-05,
+      "loss": 2.6852,
+      "step": 26930
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.633446979631087e-05,
+      "loss": 2.7024,
+      "step": 26940
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.6332284290584843e-05,
+      "loss": 2.6748,
+      "step": 26950
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.6330098784858817e-05,
+      "loss": 2.6727,
+      "step": 26960
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.632791327913279e-05,
+      "loss": 2.657,
+      "step": 26970
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.632572777340677e-05,
+      "loss": 2.6744,
+      "step": 26980
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.6323542267680743e-05,
+      "loss": 2.6277,
+      "step": 26990
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.6321356761954717e-05,
+      "loss": 2.6394,
+      "step": 27000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.631917125622869e-05,
+      "loss": 2.6933,
+      "step": 27010
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.6316985750502666e-05,
+      "loss": 2.6956,
+      "step": 27020
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.6314800244776643e-05,
+      "loss": 2.6711,
+      "step": 27030
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.6312614739050618e-05,
+      "loss": 2.7053,
+      "step": 27040
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.6310429233324592e-05,
+      "loss": 2.6913,
+      "step": 27050
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.6308243727598566e-05,
+      "loss": 2.6713,
+      "step": 27060
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.6306058221872543e-05,
+      "loss": 2.6956,
+      "step": 27070
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.6303872716146518e-05,
+      "loss": 2.6737,
+      "step": 27080
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.6301687210420495e-05,
+      "loss": 2.6211,
+      "step": 27090
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.6299501704694466e-05,
+      "loss": 2.6727,
+      "step": 27100
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.6297316198968444e-05,
+      "loss": 2.6977,
+      "step": 27110
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.6295130693242418e-05,
+      "loss": 2.7078,
+      "step": 27120
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6292945187516392e-05,
+      "loss": 2.6913,
+      "step": 27130
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6290759681790366e-05,
+      "loss": 2.665,
+      "step": 27140
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6288574176064344e-05,
+      "loss": 2.6781,
+      "step": 27150
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6286388670338318e-05,
+      "loss": 2.6818,
+      "step": 27160
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.6284203164612292e-05,
+      "loss": 2.668,
+      "step": 27170
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.6282017658886266e-05,
+      "loss": 2.7075,
+      "step": 27180
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.627983215316024e-05,
+      "loss": 2.6739,
+      "step": 27190
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.6277646647434218e-05,
+      "loss": 2.6937,
+      "step": 27200
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.6275461141708192e-05,
+      "loss": 2.6691,
+      "step": 27210
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.627327563598217e-05,
+      "loss": 2.6226,
+      "step": 27220
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.627109013025614e-05,
+      "loss": 2.7431,
+      "step": 27230
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.6268904624530118e-05,
+      "loss": 2.6768,
+      "step": 27240
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.6266719118804092e-05,
+      "loss": 2.6855,
+      "step": 27250
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.626453361307807e-05,
+      "loss": 2.745,
+      "step": 27260
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.626234810735204e-05,
+      "loss": 2.7517,
+      "step": 27270
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 2.6714,
+      "step": 27280
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.6257977095899992e-05,
+      "loss": 2.7071,
+      "step": 27290
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.6255791590173967e-05,
+      "loss": 2.6605,
+      "step": 27300
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.625360608444794e-05,
+      "loss": 2.7063,
+      "step": 27310
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.625142057872192e-05,
+      "loss": 2.6624,
+      "step": 27320
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.6249235072995893e-05,
+      "loss": 2.7205,
+      "step": 27330
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.6247049567269867e-05,
+      "loss": 2.6821,
+      "step": 27340
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.6244864061543844e-05,
+      "loss": 2.7085,
+      "step": 27350
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.6242678555817815e-05,
+      "loss": 2.6859,
+      "step": 27360
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.6240493050091793e-05,
+      "loss": 2.6647,
+      "step": 27370
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.6238307544365767e-05,
+      "loss": 2.709,
+      "step": 27380
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.6236122038639744e-05,
+      "loss": 2.6668,
+      "step": 27390
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.6233936532913715e-05,
+      "loss": 2.6948,
+      "step": 27400
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.6231751027187693e-05,
+      "loss": 2.7105,
+      "step": 27410
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.6229565521461667e-05,
+      "loss": 2.6934,
+      "step": 27420
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.6227380015735645e-05,
+      "loss": 2.7013,
+      "step": 27430
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.622519451000962e-05,
+      "loss": 2.692,
+      "step": 27440
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.6223009004283593e-05,
+      "loss": 2.7054,
+      "step": 27450
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.6220823498557567e-05,
+      "loss": 2.6841,
+      "step": 27460
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.621863799283154e-05,
+      "loss": 2.7187,
+      "step": 27470
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.621645248710552e-05,
+      "loss": 2.749,
+      "step": 27480
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.6214266981379493e-05,
+      "loss": 2.6888,
+      "step": 27490
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.6212081475653467e-05,
+      "loss": 2.6899,
+      "step": 27500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.620989596992744e-05,
+      "loss": 2.6608,
+      "step": 27510
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.620771046420142e-05,
+      "loss": 2.6704,
+      "step": 27520
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.6205524958475393e-05,
+      "loss": 2.7056,
+      "step": 27530
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.6203339452749367e-05,
+      "loss": 2.7051,
+      "step": 27540
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.620115394702334e-05,
+      "loss": 2.6728,
+      "step": 27550
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.619896844129732e-05,
+      "loss": 2.6822,
+      "step": 27560
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.6196782935571293e-05,
+      "loss": 2.6225,
+      "step": 27570
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.6194597429845268e-05,
+      "loss": 2.7202,
+      "step": 27580
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.6192411924119242e-05,
+      "loss": 2.648,
+      "step": 27590
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.619022641839322e-05,
+      "loss": 2.6815,
+      "step": 27600
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.6188040912667193e-05,
+      "loss": 2.7277,
+      "step": 27610
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.6185855406941168e-05,
+      "loss": 2.6975,
+      "step": 27620
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.6183669901215142e-05,
+      "loss": 2.7085,
+      "step": 27630
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6181484395489116e-05,
+      "loss": 2.6559,
+      "step": 27640
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6179298889763094e-05,
+      "loss": 2.712,
+      "step": 27650
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6177113384037068e-05,
+      "loss": 2.6606,
+      "step": 27660
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6174927878311042e-05,
+      "loss": 2.6865,
+      "step": 27670
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.6172742372585016e-05,
+      "loss": 2.7143,
+      "step": 27680
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.6170556866858994e-05,
+      "loss": 2.6835,
+      "step": 27690
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.6168371361132968e-05,
+      "loss": 2.6957,
+      "step": 27700
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.6166185855406942e-05,
+      "loss": 2.6901,
+      "step": 27710
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.6164000349680916e-05,
+      "loss": 2.6883,
+      "step": 27720
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.6161814843954894e-05,
+      "loss": 2.7177,
+      "step": 27730
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6159629338228868e-05,
+      "loss": 2.6702,
+      "step": 27740
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6157443832502842e-05,
+      "loss": 2.6759,
+      "step": 27750
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6155258326776816e-05,
+      "loss": 2.6941,
+      "step": 27760
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6153072821050794e-05,
+      "loss": 2.7082,
+      "step": 27770
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.6150887315324768e-05,
+      "loss": 2.7008,
+      "step": 27780
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.6148701809598742e-05,
+      "loss": 2.6882,
+      "step": 27790
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.6146516303872717e-05,
+      "loss": 2.6823,
+      "step": 27800
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.614433079814669e-05,
+      "loss": 2.6877,
+      "step": 27810
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.6142145292420668e-05,
+      "loss": 2.7264,
+      "step": 27820
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.6139959786694642e-05,
+      "loss": 2.7155,
+      "step": 27830
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.6137774280968617e-05,
+      "loss": 2.7228,
+      "step": 27840
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.613558877524259e-05,
+      "loss": 2.7131,
+      "step": 27850
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.613340326951657e-05,
+      "loss": 2.691,
+      "step": 27860
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.6131217763790543e-05,
+      "loss": 2.6585,
+      "step": 27870
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 2.6957,
+      "step": 27880
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.612684675233849e-05,
+      "loss": 2.6409,
+      "step": 27890
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.612466124661247e-05,
+      "loss": 2.6162,
+      "step": 27900
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.6122475740886443e-05,
+      "loss": 2.6799,
+      "step": 27910
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.6120290235160417e-05,
+      "loss": 2.7195,
+      "step": 27920
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.611810472943439e-05,
+      "loss": 2.7006,
+      "step": 27930
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.611591922370837e-05,
+      "loss": 2.7158,
+      "step": 27940
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.6113733717982343e-05,
+      "loss": 2.6824,
+      "step": 27950
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.6111548212256317e-05,
+      "loss": 2.6819,
+      "step": 27960
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.610936270653029e-05,
+      "loss": 2.6688,
+      "step": 27970
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.6107177200804265e-05,
+      "loss": 2.6651,
+      "step": 27980
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.6104991695078243e-05,
+      "loss": 2.7169,
+      "step": 27990
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.6102806189352217e-05,
+      "loss": 2.6962,
+      "step": 28000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.610062068362619e-05,
+      "loss": 2.7184,
+      "step": 28010
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.6098435177900166e-05,
+      "loss": 2.6582,
+      "step": 28020
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.6096249672174143e-05,
+      "loss": 2.6501,
+      "step": 28030
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.6094064166448117e-05,
+      "loss": 2.7097,
+      "step": 28040
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.6091878660722095e-05,
+      "loss": 2.686,
+      "step": 28050
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.6089693154996066e-05,
+      "loss": 2.7049,
+      "step": 28060
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.6087507649270043e-05,
+      "loss": 2.6945,
+      "step": 28070
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.6085322143544017e-05,
+      "loss": 2.7069,
+      "step": 28080
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.608313663781799e-05,
+      "loss": 2.6839,
+      "step": 28090
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.6080951132091966e-05,
+      "loss": 2.6718,
+      "step": 28100
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.6078765626365943e-05,
+      "loss": 2.6753,
+      "step": 28110
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.6076580120639918e-05,
+      "loss": 2.6571,
+      "step": 28120
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.6074394614913892e-05,
+      "loss": 2.7609,
+      "step": 28130
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.607220910918787e-05,
+      "loss": 2.6542,
+      "step": 28140
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.607002360346184e-05,
+      "loss": 2.6949,
+      "step": 28150
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.6067838097735818e-05,
+      "loss": 2.6721,
+      "step": 28160
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.6065652592009792e-05,
+      "loss": 2.6693,
+      "step": 28170
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.606346708628377e-05,
+      "loss": 2.7059,
+      "step": 28180
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.606128158055774e-05,
+      "loss": 2.719,
+      "step": 28190
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.6059096074831718e-05,
+      "loss": 2.6695,
+      "step": 28200
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.6056910569105692e-05,
+      "loss": 2.6975,
+      "step": 28210
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.605472506337967e-05,
+      "loss": 2.6574,
+      "step": 28220
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.605253955765364e-05,
+      "loss": 2.6944,
+      "step": 28230
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.6050354051927618e-05,
+      "loss": 2.6589,
+      "step": 28240
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.6048168546201592e-05,
+      "loss": 2.7356,
+      "step": 28250
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.6045983040475566e-05,
+      "loss": 2.7257,
+      "step": 28260
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.6043797534749544e-05,
+      "loss": 2.6964,
+      "step": 28270
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.6041612029023518e-05,
+      "loss": 2.6734,
+      "step": 28280
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.6039426523297492e-05,
+      "loss": 2.665,
+      "step": 28290
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6037241017571466e-05,
+      "loss": 2.6692,
+      "step": 28300
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6035055511845444e-05,
+      "loss": 2.7201,
+      "step": 28310
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6032870006119415e-05,
+      "loss": 2.6698,
+      "step": 28320
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6030684500393392e-05,
+      "loss": 2.7182,
+      "step": 28330
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.6028498994667367e-05,
+      "loss": 2.6899,
+      "step": 28340
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.6026313488941344e-05,
+      "loss": 2.7433,
+      "step": 28350
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.6024127983215315e-05,
+      "loss": 2.6523,
+      "step": 28360
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.6021942477489292e-05,
+      "loss": 2.6963,
+      "step": 28370
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.6019756971763267e-05,
+      "loss": 2.7128,
+      "step": 28380
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.6017571466037244e-05,
+      "loss": 2.718,
+      "step": 28390
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.601538596031122e-05,
+      "loss": 2.6929,
+      "step": 28400
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.6013200454585193e-05,
+      "loss": 2.708,
+      "step": 28410
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.6011014948859167e-05,
+      "loss": 2.6534,
+      "step": 28420
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.600882944313314e-05,
+      "loss": 2.685,
+      "step": 28430
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.600664393740712e-05,
+      "loss": 2.7163,
+      "step": 28440
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.6004458431681093e-05,
+      "loss": 2.6845,
+      "step": 28450
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.6002272925955067e-05,
+      "loss": 2.6832,
+      "step": 28460
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.600008742022904e-05,
+      "loss": 2.6245,
+      "step": 28470
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.599790191450302e-05,
+      "loss": 2.7287,
+      "step": 28480
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.5995716408776993e-05,
+      "loss": 2.7243,
+      "step": 28490
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.5993530903050967e-05,
+      "loss": 2.7154,
+      "step": 28500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.599134539732494e-05,
+      "loss": 2.7267,
+      "step": 28510
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.598915989159892e-05,
+      "loss": 2.7157,
+      "step": 28520
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.5986974385872893e-05,
+      "loss": 2.7054,
+      "step": 28530
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.5984788880146867e-05,
+      "loss": 2.6931,
+      "step": 28540
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.598260337442084e-05,
+      "loss": 2.6913,
+      "step": 28550
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.598041786869482e-05,
+      "loss": 2.6892,
+      "step": 28560
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.5978232362968793e-05,
+      "loss": 2.7091,
+      "step": 28570
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.5976046857242767e-05,
+      "loss": 2.6742,
+      "step": 28580
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.597386135151674e-05,
+      "loss": 2.7168,
+      "step": 28590
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.5971675845790716e-05,
+      "loss": 2.6777,
+      "step": 28600
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.5969490340064693e-05,
+      "loss": 2.6968,
+      "step": 28610
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.5967304834338667e-05,
+      "loss": 2.6947,
+      "step": 28620
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.596511932861264e-05,
+      "loss": 2.7249,
+      "step": 28630
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.5962933822886616e-05,
+      "loss": 2.6958,
+      "step": 28640
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.5960748317160593e-05,
+      "loss": 2.7021,
+      "step": 28650
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.5958562811434568e-05,
+      "loss": 2.6504,
+      "step": 28660
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.5956377305708542e-05,
+      "loss": 2.7123,
+      "step": 28670
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.5954191799982516e-05,
+      "loss": 2.7121,
+      "step": 28680
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.5952006294256494e-05,
+      "loss": 2.666,
+      "step": 28690
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.5949820788530468e-05,
+      "loss": 2.7558,
+      "step": 28700
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.5947635282804442e-05,
+      "loss": 2.7051,
+      "step": 28710
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.5945449777078416e-05,
+      "loss": 2.7549,
+      "step": 28720
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.5943264271352394e-05,
+      "loss": 2.69,
+      "step": 28730
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.5941078765626368e-05,
+      "loss": 2.6906,
+      "step": 28740
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.5938893259900342e-05,
+      "loss": 2.6783,
+      "step": 28750
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.5936707754174316e-05,
+      "loss": 2.6955,
+      "step": 28760
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.593452224844829e-05,
+      "loss": 2.6539,
+      "step": 28770
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.5932336742722268e-05,
+      "loss": 2.666,
+      "step": 28780
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.5930151236996242e-05,
+      "loss": 2.6936,
+      "step": 28790
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.5927965731270216e-05,
+      "loss": 2.6815,
+      "step": 28800
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.592578022554419e-05,
+      "loss": 2.7332,
+      "step": 28810
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5923594719818168e-05,
+      "loss": 2.7124,
+      "step": 28820
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5921409214092142e-05,
+      "loss": 2.7405,
+      "step": 28830
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.591922370836612e-05,
+      "loss": 2.7417,
+      "step": 28840
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.591703820264009e-05,
+      "loss": 2.6838,
+      "step": 28850
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.5914852696914068e-05,
+      "loss": 2.6815,
+      "step": 28860
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.5912667191188042e-05,
+      "loss": 2.7301,
+      "step": 28870
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.5910481685462017e-05,
+      "loss": 2.6821,
+      "step": 28880
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.590829617973599e-05,
+      "loss": 2.6826,
+      "step": 28890
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.590611067400997e-05,
+      "loss": 2.6872,
+      "step": 28900
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.5903925168283943e-05,
+      "loss": 2.6676,
+      "step": 28910
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.5901739662557917e-05,
+      "loss": 2.7176,
+      "step": 28920
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.589955415683189e-05,
+      "loss": 2.6745,
+      "step": 28930
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.5897368651105865e-05,
+      "loss": 2.6646,
+      "step": 28940
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.5895183145379843e-05,
+      "loss": 2.711,
+      "step": 28950
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.5892997639653817e-05,
+      "loss": 2.6994,
+      "step": 28960
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.5890812133927794e-05,
+      "loss": 2.6778,
+      "step": 28970
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.5888626628201765e-05,
+      "loss": 2.7069,
+      "step": 28980
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.5886441122475743e-05,
+      "loss": 2.7179,
+      "step": 28990
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.5884255616749717e-05,
+      "loss": 2.6878,
+      "step": 29000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.5882070111023695e-05,
+      "loss": 2.7002,
+      "step": 29010
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.5879884605297665e-05,
+      "loss": 2.7117,
+      "step": 29020
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.5877699099571643e-05,
+      "loss": 2.6739,
+      "step": 29030
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.5875513593845617e-05,
+      "loss": 2.6845,
+      "step": 29040
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.587332808811959e-05,
+      "loss": 2.6556,
+      "step": 29050
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.5871142582393565e-05,
+      "loss": 2.6881,
+      "step": 29060
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.5868957076667543e-05,
+      "loss": 2.7454,
+      "step": 29070
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.5866771570941517e-05,
+      "loss": 2.723,
+      "step": 29080
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.586458606521549e-05,
+      "loss": 2.6906,
+      "step": 29090
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.586240055948947e-05,
+      "loss": 2.7008,
+      "step": 29100
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 2.7148,
+      "step": 29110
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.5858029548037417e-05,
+      "loss": 2.6734,
+      "step": 29120
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.585584404231139e-05,
+      "loss": 2.6893,
+      "step": 29130
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.585365853658537e-05,
+      "loss": 2.7162,
+      "step": 29140
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.585147303085934e-05,
+      "loss": 2.6747,
+      "step": 29150
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.5849287525133317e-05,
+      "loss": 2.6864,
+      "step": 29160
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.584710201940729e-05,
+      "loss": 2.6739,
+      "step": 29170
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.584491651368127e-05,
+      "loss": 2.7054,
+      "step": 29180
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.5842731007955243e-05,
+      "loss": 2.7277,
+      "step": 29190
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.5840545502229218e-05,
+      "loss": 2.7265,
+      "step": 29200
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.5838359996503192e-05,
+      "loss": 2.6819,
+      "step": 29210
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.5836174490777166e-05,
+      "loss": 2.6453,
+      "step": 29220
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.5833988985051144e-05,
+      "loss": 2.6941,
+      "step": 29230
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.5831803479325118e-05,
+      "loss": 2.7098,
+      "step": 29240
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.5829617973599092e-05,
+      "loss": 2.7153,
+      "step": 29250
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.5827432467873066e-05,
+      "loss": 2.6503,
+      "step": 29260
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.5825246962147044e-05,
+      "loss": 2.7163,
+      "step": 29270
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.5823061456421014e-05,
+      "loss": 2.6823,
+      "step": 29280
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.5820875950694992e-05,
+      "loss": 2.711,
+      "step": 29290
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.5818690444968966e-05,
+      "loss": 2.6842,
+      "step": 29300
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.5816504939242944e-05,
+      "loss": 2.7211,
+      "step": 29310
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.5814319433516918e-05,
+      "loss": 2.7152,
+      "step": 29320
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.5812133927790892e-05,
+      "loss": 2.6793,
+      "step": 29330
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.5809948422064866e-05,
+      "loss": 2.6998,
+      "step": 29340
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.5807762916338844e-05,
+      "loss": 2.7023,
+      "step": 29350
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.5805577410612818e-05,
+      "loss": 2.676,
+      "step": 29360
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.5803391904886792e-05,
+      "loss": 2.7137,
+      "step": 29370
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.5801206399160766e-05,
+      "loss": 2.6863,
+      "step": 29380
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.579902089343474e-05,
+      "loss": 2.7101,
+      "step": 29390
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.5796835387708718e-05,
+      "loss": 2.6502,
+      "step": 29400
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.5794649881982692e-05,
+      "loss": 2.7421,
+      "step": 29410
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.5792464376256667e-05,
+      "loss": 2.7018,
+      "step": 29420
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.579027887053064e-05,
+      "loss": 2.7169,
+      "step": 29430
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.578809336480462e-05,
+      "loss": 2.6502,
+      "step": 29440
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.5785907859078593e-05,
+      "loss": 2.7062,
+      "step": 29450
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.5783722353352567e-05,
+      "loss": 2.6564,
+      "step": 29460
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.578153684762654e-05,
+      "loss": 2.6477,
+      "step": 29470
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.577935134190052e-05,
+      "loss": 2.7082,
+      "step": 29480
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.5777165836174493e-05,
+      "loss": 2.6483,
+      "step": 29490
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.5774980330448467e-05,
+      "loss": 2.7137,
+      "step": 29500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.577279482472244e-05,
+      "loss": 2.7363,
+      "step": 29510
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.577060931899642e-05,
+      "loss": 2.6873,
+      "step": 29520
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.5768423813270393e-05,
+      "loss": 2.7138,
+      "step": 29530
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.5766238307544367e-05,
+      "loss": 2.6987,
+      "step": 29540
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.576405280181834e-05,
+      "loss": 2.6863,
+      "step": 29550
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.5761867296092315e-05,
+      "loss": 2.7251,
+      "step": 29560
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.5759681790366293e-05,
+      "loss": 2.7209,
+      "step": 29570
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.5757496284640267e-05,
+      "loss": 2.6825,
+      "step": 29580
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.575531077891424e-05,
+      "loss": 2.6753,
+      "step": 29590
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.5753125273188215e-05,
+      "loss": 2.6745,
+      "step": 29600
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.5750939767462193e-05,
+      "loss": 2.699,
+      "step": 29610
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.5748754261736167e-05,
+      "loss": 2.6955,
+      "step": 29620
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.574656875601014e-05,
+      "loss": 2.7181,
+      "step": 29630
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.5744383250284116e-05,
+      "loss": 2.7086,
+      "step": 29640
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.5742197744558093e-05,
+      "loss": 2.7056,
+      "step": 29650
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.5740012238832067e-05,
+      "loss": 2.7422,
+      "step": 29660
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.573782673310604e-05,
+      "loss": 2.6778,
+      "step": 29670
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.5735641227380016e-05,
+      "loss": 2.6923,
+      "step": 29680
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.5733455721653993e-05,
+      "loss": 2.7096,
+      "step": 29690
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.5731270215927967e-05,
+      "loss": 2.6777,
+      "step": 29700
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.572908471020194e-05,
+      "loss": 2.6766,
+      "step": 29710
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.5726899204475916e-05,
+      "loss": 2.7197,
+      "step": 29720
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.572471369874989e-05,
+      "loss": 2.6655,
+      "step": 29730
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.5722528193023868e-05,
+      "loss": 2.6838,
+      "step": 29740
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.5720342687297842e-05,
+      "loss": 2.6887,
+      "step": 29750
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.5718157181571816e-05,
+      "loss": 2.6767,
+      "step": 29760
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.571597167584579e-05,
+      "loss": 2.6805,
+      "step": 29770
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.5713786170119768e-05,
+      "loss": 2.646,
+      "step": 29780
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.5711600664393742e-05,
+      "loss": 2.7148,
+      "step": 29790
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.570941515866772e-05,
+      "loss": 2.7143,
+      "step": 29800
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.570722965294169e-05,
+      "loss": 2.7234,
+      "step": 29810
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.5705044147215668e-05,
+      "loss": 2.6808,
+      "step": 29820
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.5702858641489642e-05,
+      "loss": 2.6892,
+      "step": 29830
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.5700673135763616e-05,
+      "loss": 2.7095,
+      "step": 29840
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.569848763003759e-05,
+      "loss": 2.7087,
+      "step": 29850
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.5696302124311568e-05,
+      "loss": 2.729,
+      "step": 29860
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.5694116618585542e-05,
+      "loss": 2.7089,
+      "step": 29870
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.5691931112859516e-05,
+      "loss": 2.6711,
+      "step": 29880
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.5689745607133494e-05,
+      "loss": 2.7054,
+      "step": 29890
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.5687560101407465e-05,
+      "loss": 2.7181,
+      "step": 29900
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.5685374595681442e-05,
+      "loss": 2.6827,
+      "step": 29910
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.5683189089955416e-05,
+      "loss": 2.6772,
+      "step": 29920
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.5681003584229394e-05,
+      "loss": 2.6787,
+      "step": 29930
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.5678818078503365e-05,
+      "loss": 2.7155,
+      "step": 29940
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.5676632572777342e-05,
+      "loss": 2.6899,
+      "step": 29950
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.5674447067051317e-05,
+      "loss": 2.7087,
+      "step": 29960
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.5672261561325294e-05,
+      "loss": 2.6831,
+      "step": 29970
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.5670076055599265e-05,
+      "loss": 2.6951,
+      "step": 29980
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.5667890549873243e-05,
+      "loss": 2.6756,
+      "step": 29990
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.5665705044147217e-05,
+      "loss": 2.697,
+      "step": 30000
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.566351953842119e-05,
+      "loss": 2.7056,
+      "step": 30010
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.566133403269517e-05,
+      "loss": 2.6975,
+      "step": 30020
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.5659148526969143e-05,
+      "loss": 2.6952,
+      "step": 30030
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.5656963021243117e-05,
+      "loss": 2.6876,
+      "step": 30040
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.565477751551709e-05,
+      "loss": 2.6947,
+      "step": 30050
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.565259200979107e-05,
+      "loss": 2.7057,
+      "step": 30060
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.565040650406504e-05,
+      "loss": 2.7469,
+      "step": 30070
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.5648220998339017e-05,
+      "loss": 2.6848,
+      "step": 30080
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.564603549261299e-05,
+      "loss": 2.7172,
+      "step": 30090
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.564384998688697e-05,
+      "loss": 2.6843,
+      "step": 30100
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.564166448116094e-05,
+      "loss": 2.7449,
+      "step": 30110
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.5639478975434917e-05,
+      "loss": 2.7256,
+      "step": 30120
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.563729346970889e-05,
+      "loss": 2.6912,
+      "step": 30130
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.563510796398287e-05,
+      "loss": 2.6828,
+      "step": 30140
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.5632922458256843e-05,
+      "loss": 2.7274,
+      "step": 30150
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.5630736952530817e-05,
+      "loss": 2.6888,
+      "step": 30160
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.562855144680479e-05,
+      "loss": 2.6695,
+      "step": 30170
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.5626365941078766e-05,
+      "loss": 2.6886,
+      "step": 30180
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.5624180435352743e-05,
+      "loss": 2.7291,
+      "step": 30190
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.5621994929626717e-05,
+      "loss": 2.7182,
+      "step": 30200
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.561980942390069e-05,
+      "loss": 2.6905,
+      "step": 30210
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.5617623918174666e-05,
+      "loss": 2.7128,
+      "step": 30220
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.5615438412448643e-05,
+      "loss": 2.6937,
+      "step": 30230
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.5613252906722617e-05,
+      "loss": 2.6856,
+      "step": 30240
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.561106740099659e-05,
+      "loss": 2.673,
+      "step": 30250
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.5608881895270566e-05,
+      "loss": 2.7238,
+      "step": 30260
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.5606696389544543e-05,
+      "loss": 2.6898,
+      "step": 30270
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.5604510883818518e-05,
+      "loss": 2.7081,
+      "step": 30280
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.5602325378092492e-05,
+      "loss": 2.7019,
+      "step": 30290
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.5600139872366466e-05,
+      "loss": 2.6547,
+      "step": 30300
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.5597954366640444e-05,
+      "loss": 2.6929,
+      "step": 30310
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.5595768860914418e-05,
+      "loss": 2.6743,
+      "step": 30320
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.5593583355188392e-05,
+      "loss": 2.7134,
+      "step": 30330
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.5591397849462366e-05,
+      "loss": 2.6683,
+      "step": 30340
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.558921234373634e-05,
+      "loss": 2.7483,
+      "step": 30350
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.5587026838010318e-05,
+      "loss": 2.7083,
+      "step": 30360
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.5584841332284292e-05,
+      "loss": 2.7068,
+      "step": 30370
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.5582655826558266e-05,
+      "loss": 2.7363,
+      "step": 30380
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.558047032083224e-05,
+      "loss": 2.6928,
+      "step": 30390
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.5578284815106218e-05,
+      "loss": 2.6889,
+      "step": 30400
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.5576099309380192e-05,
+      "loss": 2.7422,
+      "step": 30410
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.5573913803654166e-05,
+      "loss": 2.6814,
+      "step": 30420
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.557172829792814e-05,
+      "loss": 2.6938,
+      "step": 30430
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.5569542792202118e-05,
+      "loss": 2.7051,
+      "step": 30440
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.5567357286476092e-05,
+      "loss": 2.6972,
+      "step": 30450
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.5565171780750067e-05,
+      "loss": 2.684,
+      "step": 30460
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.556298627502404e-05,
+      "loss": 2.7011,
+      "step": 30470
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.5560800769298018e-05,
+      "loss": 2.7039,
+      "step": 30480
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.5558615263571992e-05,
+      "loss": 2.7019,
+      "step": 30490
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.5556429757845967e-05,
+      "loss": 2.7291,
+      "step": 30500
+    },
+    {
+      "epoch": 6.0,
+      "eval_BLEU": 8.6439,
+      "eval_gen_len": 11.8513,
+      "eval_loss": 3.0536694526672363,
+      "eval_runtime": 154.4128,
+      "eval_samples_per_second": 69.703,
+      "eval_steps_per_second": 0.278,
+      "step": 30504
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.555424425211994e-05,
+      "loss": 2.6893,
+      "step": 30510
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.5552058746393915e-05,
+      "loss": 2.5668,
+      "step": 30520
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.5549873240667893e-05,
+      "loss": 2.5935,
+      "step": 30530
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.5547687734941867e-05,
+      "loss": 2.6039,
+      "step": 30540
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.554550222921584e-05,
+      "loss": 2.5169,
+      "step": 30550
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.5543316723489815e-05,
+      "loss": 2.6159,
+      "step": 30560
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.5541131217763793e-05,
+      "loss": 2.5625,
+      "step": 30570
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.5538945712037767e-05,
+      "loss": 2.591,
+      "step": 30580
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.5536760206311744e-05,
+      "loss": 2.5815,
+      "step": 30590
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.5534574700585715e-05,
+      "loss": 2.5914,
+      "step": 30600
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.5532389194859693e-05,
+      "loss": 2.563,
+      "step": 30610
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.5530203689133667e-05,
+      "loss": 2.5695,
+      "step": 30620
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.552801818340764e-05,
+      "loss": 2.6274,
+      "step": 30630
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.5525832677681615e-05,
+      "loss": 2.5349,
+      "step": 30640
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.5523647171955593e-05,
+      "loss": 2.5182,
+      "step": 30650
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.5521461666229567e-05,
+      "loss": 2.5828,
+      "step": 30660
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.551927616050354e-05,
+      "loss": 2.5859,
+      "step": 30670
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.5517090654777516e-05,
+      "loss": 2.5937,
+      "step": 30680
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.551490514905149e-05,
+      "loss": 2.5915,
+      "step": 30690
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.5512719643325467e-05,
+      "loss": 2.6078,
+      "step": 30700
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.551053413759944e-05,
+      "loss": 2.6222,
+      "step": 30710
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.550834863187342e-05,
+      "loss": 2.5946,
+      "step": 30720
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.550616312614739e-05,
+      "loss": 2.5913,
+      "step": 30730
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.5503977620421367e-05,
+      "loss": 2.5471,
+      "step": 30740
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.550179211469534e-05,
+      "loss": 2.6044,
+      "step": 30750
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.549960660896932e-05,
+      "loss": 2.5936,
+      "step": 30760
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.549742110324329e-05,
+      "loss": 2.5719,
+      "step": 30770
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.5495235597517268e-05,
+      "loss": 2.5848,
+      "step": 30780
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.5493050091791242e-05,
+      "loss": 2.5822,
+      "step": 30790
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.5490864586065216e-05,
+      "loss": 2.6236,
+      "step": 30800
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.548867908033919e-05,
+      "loss": 2.596,
+      "step": 30810
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.5486493574613168e-05,
+      "loss": 2.5733,
+      "step": 30820
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.5484308068887142e-05,
+      "loss": 2.5748,
+      "step": 30830
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.5482122563161116e-05,
+      "loss": 2.646,
+      "step": 30840
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.5479937057435094e-05,
+      "loss": 2.5439,
+      "step": 30850
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.5477751551709064e-05,
+      "loss": 2.6058,
+      "step": 30860
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.5475566045983042e-05,
+      "loss": 2.5831,
+      "step": 30870
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.5473380540257016e-05,
+      "loss": 2.576,
+      "step": 30880
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.5471195034530994e-05,
+      "loss": 2.5962,
+      "step": 30890
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.5469009528804965e-05,
+      "loss": 2.5854,
+      "step": 30900
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.5466824023078942e-05,
+      "loss": 2.5655,
+      "step": 30910
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.5464638517352916e-05,
+      "loss": 2.5542,
+      "step": 30920
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.5462453011626894e-05,
+      "loss": 2.6011,
+      "step": 30930
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.5460267505900868e-05,
+      "loss": 2.6225,
+      "step": 30940
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.5458082000174842e-05,
+      "loss": 2.5965,
+      "step": 30950
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.5455896494448816e-05,
+      "loss": 2.6071,
+      "step": 30960
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.545371098872279e-05,
+      "loss": 2.6076,
+      "step": 30970
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.5451525482996768e-05,
+      "loss": 2.597,
+      "step": 30980
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.5449339977270742e-05,
+      "loss": 2.6053,
+      "step": 30990
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.5447154471544717e-05,
+      "loss": 2.582,
+      "step": 31000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.544496896581869e-05,
+      "loss": 2.596,
+      "step": 31010
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.5442783460092668e-05,
+      "loss": 2.6125,
+      "step": 31020
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.544059795436664e-05,
+      "loss": 2.5755,
+      "step": 31030
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.5438412448640617e-05,
+      "loss": 2.5712,
+      "step": 31040
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.543622694291459e-05,
+      "loss": 2.5513,
+      "step": 31050
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.543404143718857e-05,
+      "loss": 2.6029,
+      "step": 31060
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.5431855931462543e-05,
+      "loss": 2.5938,
+      "step": 31070
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.5429670425736517e-05,
+      "loss": 2.6025,
+      "step": 31080
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.542748492001049e-05,
+      "loss": 2.6077,
+      "step": 31090
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.542529941428447e-05,
+      "loss": 2.5881,
+      "step": 31100
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.5423113908558443e-05,
+      "loss": 2.6175,
+      "step": 31110
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.5420928402832417e-05,
+      "loss": 2.583,
+      "step": 31120
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.541874289710639e-05,
+      "loss": 2.589,
+      "step": 31130
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5416557391380365e-05,
+      "loss": 2.6108,
+      "step": 31140
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5414371885654343e-05,
+      "loss": 2.5956,
+      "step": 31150
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5412186379928317e-05,
+      "loss": 2.6045,
+      "step": 31160
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.541000087420229e-05,
+      "loss": 2.5899,
+      "step": 31170
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5407815368476265e-05,
+      "loss": 2.6315,
+      "step": 31180
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.5405629862750243e-05,
+      "loss": 2.5994,
+      "step": 31190
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.5403444357024217e-05,
+      "loss": 2.5663,
+      "step": 31200
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.540125885129819e-05,
+      "loss": 2.6066,
+      "step": 31210
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.5399073345572166e-05,
+      "loss": 2.5098,
+      "step": 31220
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.5396887839846143e-05,
+      "loss": 2.5241,
+      "step": 31230
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.5394702334120117e-05,
+      "loss": 2.5713,
+      "step": 31240
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.539251682839409e-05,
+      "loss": 2.5911,
+      "step": 31250
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.5390331322668066e-05,
+      "loss": 2.5775,
+      "step": 31260
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.5388145816942043e-05,
+      "loss": 2.6095,
+      "step": 31270
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.5385960311216017e-05,
+      "loss": 2.6124,
+      "step": 31280
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.538377480548999e-05,
+      "loss": 2.5892,
+      "step": 31290
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.5381589299763966e-05,
+      "loss": 2.5544,
+      "step": 31300
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.537940379403794e-05,
+      "loss": 2.6397,
+      "step": 31310
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.5377218288311918e-05,
+      "loss": 2.5889,
+      "step": 31320
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.5375032782585892e-05,
+      "loss": 2.5945,
+      "step": 31330
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.5372847276859866e-05,
+      "loss": 2.5734,
+      "step": 31340
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.537066177113384e-05,
+      "loss": 2.562,
+      "step": 31350
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.5368476265407818e-05,
+      "loss": 2.6013,
+      "step": 31360
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.5366290759681792e-05,
+      "loss": 2.6025,
+      "step": 31370
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.5364105253955766e-05,
+      "loss": 2.6185,
+      "step": 31380
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.536191974822974e-05,
+      "loss": 2.5785,
+      "step": 31390
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.5359734242503718e-05,
+      "loss": 2.6332,
+      "step": 31400
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.5357548736777692e-05,
+      "loss": 2.5816,
+      "step": 31410
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.5355363231051666e-05,
+      "loss": 2.6131,
+      "step": 31420
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.535317772532564e-05,
+      "loss": 2.6223,
+      "step": 31430
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.5350992219599618e-05,
+      "loss": 2.6216,
+      "step": 31440
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.5348806713873592e-05,
+      "loss": 2.6491,
+      "step": 31450
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.5346621208147566e-05,
+      "loss": 2.6171,
+      "step": 31460
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.534443570242154e-05,
+      "loss": 2.6019,
+      "step": 31470
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.5342250196695515e-05,
+      "loss": 2.5421,
+      "step": 31480
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.5340064690969492e-05,
+      "loss": 2.615,
+      "step": 31490
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5337879185243466e-05,
+      "loss": 2.5931,
+      "step": 31500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5335693679517444e-05,
+      "loss": 2.6341,
+      "step": 31510
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5333508173791415e-05,
+      "loss": 2.6381,
+      "step": 31520
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5331322668065392e-05,
+      "loss": 2.6212,
+      "step": 31530
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.5329137162339367e-05,
+      "loss": 2.6163,
+      "step": 31540
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.5326951656613344e-05,
+      "loss": 2.6099,
+      "step": 31550
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.5324766150887315e-05,
+      "loss": 2.6053,
+      "step": 31560
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.5322580645161292e-05,
+      "loss": 2.5873,
+      "step": 31570
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.5320395139435267e-05,
+      "loss": 2.5772,
+      "step": 31580
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.531820963370924e-05,
+      "loss": 2.6517,
+      "step": 31590
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.5316024127983215e-05,
+      "loss": 2.6138,
+      "step": 31600
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.5313838622257193e-05,
+      "loss": 2.5795,
+      "step": 31610
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.5311653116531167e-05,
+      "loss": 2.6123,
+      "step": 31620
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.530946761080514e-05,
+      "loss": 2.6051,
+      "step": 31630
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.530728210507912e-05,
+      "loss": 2.6101,
+      "step": 31640
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.530509659935309e-05,
+      "loss": 2.6365,
+      "step": 31650
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.5302911093627067e-05,
+      "loss": 2.5738,
+      "step": 31660
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.530072558790104e-05,
+      "loss": 2.603,
+      "step": 31670
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.529854008217502e-05,
+      "loss": 2.5868,
+      "step": 31680
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.529635457644899e-05,
+      "loss": 2.6051,
+      "step": 31690
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.5294169070722967e-05,
+      "loss": 2.5938,
+      "step": 31700
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.529198356499694e-05,
+      "loss": 2.6028,
+      "step": 31710
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.528979805927092e-05,
+      "loss": 2.6033,
+      "step": 31720
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.528761255354489e-05,
+      "loss": 2.5845,
+      "step": 31730
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.5285427047818867e-05,
+      "loss": 2.5823,
+      "step": 31740
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.528324154209284e-05,
+      "loss": 2.5863,
+      "step": 31750
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.5281056036366816e-05,
+      "loss": 2.6163,
+      "step": 31760
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.5278870530640793e-05,
+      "loss": 2.6221,
+      "step": 31770
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.5276685024914767e-05,
+      "loss": 2.6145,
+      "step": 31780
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.527449951918874e-05,
+      "loss": 2.5809,
+      "step": 31790
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.5272314013462716e-05,
+      "loss": 2.5908,
+      "step": 31800
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.5270128507736693e-05,
+      "loss": 2.6273,
+      "step": 31810
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.5267943002010664e-05,
+      "loss": 2.5544,
+      "step": 31820
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.526575749628464e-05,
+      "loss": 2.6117,
+      "step": 31830
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.5263571990558616e-05,
+      "loss": 2.5955,
+      "step": 31840
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.5261386484832593e-05,
+      "loss": 2.5981,
+      "step": 31850
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5259200979106568e-05,
+      "loss": 2.5976,
+      "step": 31860
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5257015473380542e-05,
+      "loss": 2.6729,
+      "step": 31870
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5254829967654516e-05,
+      "loss": 2.6169,
+      "step": 31880
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5252644461928492e-05,
+      "loss": 2.5788,
+      "step": 31890
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.5250458956202468e-05,
+      "loss": 2.6427,
+      "step": 31900
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.524827345047644e-05,
+      "loss": 2.5881,
+      "step": 31910
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.5246087944750416e-05,
+      "loss": 2.5992,
+      "step": 31920
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.5243902439024392e-05,
+      "loss": 2.6011,
+      "step": 31930
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.5241716933298368e-05,
+      "loss": 2.5996,
+      "step": 31940
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.523953142757234e-05,
+      "loss": 2.605,
+      "step": 31950
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5237345921846316e-05,
+      "loss": 2.6201,
+      "step": 31960
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5235160416120292e-05,
+      "loss": 2.5692,
+      "step": 31970
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5232974910394266e-05,
+      "loss": 2.5911,
+      "step": 31980
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5230789404668242e-05,
+      "loss": 2.6366,
+      "step": 31990
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.5228603898942216e-05,
+      "loss": 2.6029,
+      "step": 32000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.522641839321619e-05,
+      "loss": 2.5814,
+      "step": 32010
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.5224232887490166e-05,
+      "loss": 2.6234,
+      "step": 32020
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.5222047381764142e-05,
+      "loss": 2.6365,
+      "step": 32030
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.5219861876038116e-05,
+      "loss": 2.6255,
+      "step": 32040
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.521767637031209e-05,
+      "loss": 2.5849,
+      "step": 32050
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.5215490864586066e-05,
+      "loss": 2.619,
+      "step": 32060
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.5213305358860042e-05,
+      "loss": 2.6113,
+      "step": 32070
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.5211119853134015e-05,
+      "loss": 2.6116,
+      "step": 32080
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.520893434740799e-05,
+      "loss": 2.598,
+      "step": 32090
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.5206748841681967e-05,
+      "loss": 2.633,
+      "step": 32100
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.5204563335955942e-05,
+      "loss": 2.5799,
+      "step": 32110
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.5202377830229917e-05,
+      "loss": 2.5921,
+      "step": 32120
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.5200192324503891e-05,
+      "loss": 2.598,
+      "step": 32130
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.5198006818777867e-05,
+      "loss": 2.591,
+      "step": 32140
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.5195821313051841e-05,
+      "loss": 2.6589,
+      "step": 32150
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.5193635807325817e-05,
+      "loss": 2.6434,
+      "step": 32160
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.5191450301599791e-05,
+      "loss": 2.6607,
+      "step": 32170
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.5189264795873765e-05,
+      "loss": 2.6076,
+      "step": 32180
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.5187079290147741e-05,
+      "loss": 2.6248,
+      "step": 32190
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.5184893784421717e-05,
+      "loss": 2.6444,
+      "step": 32200
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5182708278695693e-05,
+      "loss": 2.6055,
+      "step": 32210
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5180522772969665e-05,
+      "loss": 2.6122,
+      "step": 32220
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5178337267243641e-05,
+      "loss": 2.5585,
+      "step": 32230
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5176151761517617e-05,
+      "loss": 2.6452,
+      "step": 32240
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.5173966255791593e-05,
+      "loss": 2.6061,
+      "step": 32250
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.5171780750065565e-05,
+      "loss": 2.6233,
+      "step": 32260
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.5169595244339541e-05,
+      "loss": 2.5831,
+      "step": 32270
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.5167409738613517e-05,
+      "loss": 2.6298,
+      "step": 32280
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.5165224232887491e-05,
+      "loss": 2.6221,
+      "step": 32290
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.5163038727161466e-05,
+      "loss": 2.5895,
+      "step": 32300
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.5160853221435441e-05,
+      "loss": 2.5932,
+      "step": 32310
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.5158667715709416e-05,
+      "loss": 2.5974,
+      "step": 32320
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.5156482209983391e-05,
+      "loss": 2.6365,
+      "step": 32330
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.5154296704257367e-05,
+      "loss": 2.6135,
+      "step": 32340
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.515211119853134e-05,
+      "loss": 2.6135,
+      "step": 32350
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5149925692805316e-05,
+      "loss": 2.6162,
+      "step": 32360
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5147740187079292e-05,
+      "loss": 2.6367,
+      "step": 32370
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5145554681353268e-05,
+      "loss": 2.5796,
+      "step": 32380
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.514336917562724e-05,
+      "loss": 2.614,
+      "step": 32390
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5141183669901216e-05,
+      "loss": 2.6407,
+      "step": 32400
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.5138998164175192e-05,
+      "loss": 2.6075,
+      "step": 32410
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.5136812658449168e-05,
+      "loss": 2.6275,
+      "step": 32420
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.513462715272314e-05,
+      "loss": 2.5828,
+      "step": 32430
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.5132441646997116e-05,
+      "loss": 2.6462,
+      "step": 32440
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.5130256141271092e-05,
+      "loss": 2.6243,
+      "step": 32450
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.5128070635545066e-05,
+      "loss": 2.6349,
+      "step": 32460
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.5125885129819042e-05,
+      "loss": 2.5627,
+      "step": 32470
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.5123699624093016e-05,
+      "loss": 2.6123,
+      "step": 32480
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.512151411836699e-05,
+      "loss": 2.6044,
+      "step": 32490
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.5119328612640966e-05,
+      "loss": 2.6071,
+      "step": 32500
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.5117143106914942e-05,
+      "loss": 2.6064,
+      "step": 32510
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.5114957601188916e-05,
+      "loss": 2.6012,
+      "step": 32520
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.511277209546289e-05,
+      "loss": 2.6589,
+      "step": 32530
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.5110586589736866e-05,
+      "loss": 2.5767,
+      "step": 32540
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.5108401084010842e-05,
+      "loss": 2.5975,
+      "step": 32550
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.5106215578284818e-05,
+      "loss": 2.6151,
+      "step": 32560
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.510403007255879e-05,
+      "loss": 2.5946,
+      "step": 32570
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.5101844566832766e-05,
+      "loss": 2.6027,
+      "step": 32580
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.5099659061106742e-05,
+      "loss": 2.6178,
+      "step": 32590
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.5097473555380717e-05,
+      "loss": 2.6368,
+      "step": 32600
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.509528804965469e-05,
+      "loss": 2.5823,
+      "step": 32610
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.5093102543928667e-05,
+      "loss": 2.6481,
+      "step": 32620
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.509091703820264e-05,
+      "loss": 2.662,
+      "step": 32630
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.5088731532476617e-05,
+      "loss": 2.6078,
+      "step": 32640
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.508654602675059e-05,
+      "loss": 2.6395,
+      "step": 32650
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.5084360521024565e-05,
+      "loss": 2.5914,
+      "step": 32660
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.5082175015298541e-05,
+      "loss": 2.6003,
+      "step": 32670
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.5079989509572517e-05,
+      "loss": 2.6428,
+      "step": 32680
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.5077804003846493e-05,
+      "loss": 2.6336,
+      "step": 32690
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.5075618498120465e-05,
+      "loss": 2.6092,
+      "step": 32700
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.5073432992394441e-05,
+      "loss": 2.6182,
+      "step": 32710
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.5071247486668417e-05,
+      "loss": 2.6012,
+      "step": 32720
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.5069061980942393e-05,
+      "loss": 2.6307,
+      "step": 32730
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.5066876475216365e-05,
+      "loss": 2.6429,
+      "step": 32740
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.5064690969490341e-05,
+      "loss": 2.5729,
+      "step": 32750
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.5062505463764317e-05,
+      "loss": 2.6112,
+      "step": 32760
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5060319958038291e-05,
+      "loss": 2.6311,
+      "step": 32770
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5058134452312265e-05,
+      "loss": 2.5973,
+      "step": 32780
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5055948946586241e-05,
+      "loss": 2.6019,
+      "step": 32790
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5053763440860215e-05,
+      "loss": 2.591,
+      "step": 32800
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.5051577935134191e-05,
+      "loss": 2.6395,
+      "step": 32810
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.5049392429408167e-05,
+      "loss": 2.5735,
+      "step": 32820
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.504720692368214e-05,
+      "loss": 2.6187,
+      "step": 32830
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.5045021417956116e-05,
+      "loss": 2.6072,
+      "step": 32840
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.5042835912230091e-05,
+      "loss": 2.6374,
+      "step": 32850
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.5040650406504067e-05,
+      "loss": 2.5883,
+      "step": 32860
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.503846490077804e-05,
+      "loss": 2.6258,
+      "step": 32870
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.5036279395052016e-05,
+      "loss": 2.5859,
+      "step": 32880
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.5034093889325992e-05,
+      "loss": 2.6485,
+      "step": 32890
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.5031908383599967e-05,
+      "loss": 2.6498,
+      "step": 32900
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.5029722877873942e-05,
+      "loss": 2.6307,
+      "step": 32910
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5027537372147916e-05,
+      "loss": 2.6315,
+      "step": 32920
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5025351866421892e-05,
+      "loss": 2.6209,
+      "step": 32930
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5023166360695866e-05,
+      "loss": 2.6045,
+      "step": 32940
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5020980854969842e-05,
+      "loss": 2.5958,
+      "step": 32950
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.5018795349243816e-05,
+      "loss": 2.6407,
+      "step": 32960
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.501660984351779e-05,
+      "loss": 2.6318,
+      "step": 32970
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.5014424337791766e-05,
+      "loss": 2.5757,
+      "step": 32980
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.5012238832065742e-05,
+      "loss": 2.6215,
+      "step": 32990
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.5010053326339716e-05,
+      "loss": 2.6649,
+      "step": 33000
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.500786782061369e-05,
+      "loss": 2.6063,
+      "step": 33010
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.5005682314887666e-05,
+      "loss": 2.6027,
+      "step": 33020
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.5003496809161642e-05,
+      "loss": 2.5765,
+      "step": 33030
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.5001311303435618e-05,
+      "loss": 2.5726,
+      "step": 33040
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.499912579770959e-05,
+      "loss": 2.6326,
+      "step": 33050
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.4996940291983566e-05,
+      "loss": 2.5924,
+      "step": 33060
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.4994754786257542e-05,
+      "loss": 2.5754,
+      "step": 33070
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.4992569280531516e-05,
+      "loss": 2.6022,
+      "step": 33080
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.499038377480549e-05,
+      "loss": 2.5938,
+      "step": 33090
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.4988198269079466e-05,
+      "loss": 2.6308,
+      "step": 33100
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.498601276335344e-05,
+      "loss": 2.6153,
+      "step": 33110
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.4983827257627416e-05,
+      "loss": 2.6176,
+      "step": 33120
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.498164175190139e-05,
+      "loss": 2.6342,
+      "step": 33130
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.4979456246175365e-05,
+      "loss": 2.6356,
+      "step": 33140
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.497727074044934e-05,
+      "loss": 2.6179,
+      "step": 33150
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.4975085234723317e-05,
+      "loss": 2.6489,
+      "step": 33160
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.4972899728997292e-05,
+      "loss": 2.633,
+      "step": 33170
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.4970714223271265e-05,
+      "loss": 2.6358,
+      "step": 33180
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.496852871754524e-05,
+      "loss": 2.592,
+      "step": 33190
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.4966343211819217e-05,
+      "loss": 2.5681,
+      "step": 33200
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.4964157706093193e-05,
+      "loss": 2.5947,
+      "step": 33210
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.4961972200367165e-05,
+      "loss": 2.6064,
+      "step": 33220
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.4959786694641141e-05,
+      "loss": 2.5823,
+      "step": 33230
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.4957601188915117e-05,
+      "loss": 2.6206,
+      "step": 33240
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.4955415683189091e-05,
+      "loss": 2.6013,
+      "step": 33250
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.4953230177463067e-05,
+      "loss": 2.6479,
+      "step": 33260
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.4951044671737041e-05,
+      "loss": 2.6167,
+      "step": 33270
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.4948859166011015e-05,
+      "loss": 2.636,
+      "step": 33280
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.4946673660284991e-05,
+      "loss": 2.6297,
+      "step": 33290
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.4944488154558967e-05,
+      "loss": 2.6,
+      "step": 33300
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.494230264883294e-05,
+      "loss": 2.6368,
+      "step": 33310
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.4940117143106915e-05,
+      "loss": 2.6111,
+      "step": 33320
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.4937931637380891e-05,
+      "loss": 2.6299,
+      "step": 33330
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.4935746131654867e-05,
+      "loss": 2.6376,
+      "step": 33340
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.493356062592884e-05,
+      "loss": 2.6551,
+      "step": 33350
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.4931375120202816e-05,
+      "loss": 2.5399,
+      "step": 33360
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.4929189614476791e-05,
+      "loss": 2.5891,
+      "step": 33370
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.4927004108750767e-05,
+      "loss": 2.6158,
+      "step": 33380
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.4924818603024741e-05,
+      "loss": 2.5922,
+      "step": 33390
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.4922633097298716e-05,
+      "loss": 2.6117,
+      "step": 33400
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.4920447591572692e-05,
+      "loss": 2.597,
+      "step": 33410
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.4918262085846666e-05,
+      "loss": 2.6396,
+      "step": 33420
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.4916076580120642e-05,
+      "loss": 2.6054,
+      "step": 33430
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.4913891074394616e-05,
+      "loss": 2.6752,
+      "step": 33440
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.491170556866859e-05,
+      "loss": 2.6349,
+      "step": 33450
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.4909520062942566e-05,
+      "loss": 2.631,
+      "step": 33460
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.4907334557216542e-05,
+      "loss": 2.6229,
+      "step": 33470
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.4905149051490514e-05,
+      "loss": 2.6413,
+      "step": 33480
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.490296354576449e-05,
+      "loss": 2.6281,
+      "step": 33490
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.4900778040038466e-05,
+      "loss": 2.5888,
+      "step": 33500
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.4898592534312442e-05,
+      "loss": 2.6379,
+      "step": 33510
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.4896407028586418e-05,
+      "loss": 2.6313,
+      "step": 33520
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.489422152286039e-05,
+      "loss": 2.638,
+      "step": 33530
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.4892036017134366e-05,
+      "loss": 2.6049,
+      "step": 33540
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.4889850511408342e-05,
+      "loss": 2.6297,
+      "step": 33550
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.4887665005682316e-05,
+      "loss": 2.6268,
+      "step": 33560
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.488547949995629e-05,
+      "loss": 2.6443,
+      "step": 33570
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.4883293994230266e-05,
+      "loss": 2.632,
+      "step": 33580
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.488110848850424e-05,
+      "loss": 2.612,
+      "step": 33590
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.4878922982778216e-05,
+      "loss": 2.6212,
+      "step": 33600
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.4876737477052192e-05,
+      "loss": 2.6637,
+      "step": 33610
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.4874551971326165e-05,
+      "loss": 2.6463,
+      "step": 33620
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.487236646560014e-05,
+      "loss": 2.6343,
+      "step": 33630
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.4870180959874116e-05,
+      "loss": 2.6287,
+      "step": 33640
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.4867995454148092e-05,
+      "loss": 2.6083,
+      "step": 33650
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.4865809948422065e-05,
+      "loss": 2.6211,
+      "step": 33660
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.486362444269604e-05,
+      "loss": 2.6332,
+      "step": 33670
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.4861438936970017e-05,
+      "loss": 2.6078,
+      "step": 33680
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.4859253431243992e-05,
+      "loss": 2.6526,
+      "step": 33690
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.4857067925517965e-05,
+      "loss": 2.6003,
+      "step": 33700
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.485488241979194e-05,
+      "loss": 2.6187,
+      "step": 33710
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.4852696914065917e-05,
+      "loss": 2.6121,
+      "step": 33720
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.4850511408339891e-05,
+      "loss": 2.6837,
+      "step": 33730
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.4848325902613867e-05,
+      "loss": 2.597,
+      "step": 33740
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.4846140396887841e-05,
+      "loss": 2.6039,
+      "step": 33750
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.4843954891161815e-05,
+      "loss": 2.6504,
+      "step": 33760
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.4841769385435791e-05,
+      "loss": 2.6404,
+      "step": 33770
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.4839583879709767e-05,
+      "loss": 2.618,
+      "step": 33780
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.483739837398374e-05,
+      "loss": 2.6058,
+      "step": 33790
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.4835212868257715e-05,
+      "loss": 2.6195,
+      "step": 33800
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.4833027362531691e-05,
+      "loss": 2.6338,
+      "step": 33810
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.4830841856805667e-05,
+      "loss": 2.6157,
+      "step": 33820
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.482865635107964e-05,
+      "loss": 2.624,
+      "step": 33830
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.4826470845353615e-05,
+      "loss": 2.6231,
+      "step": 33840
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.4824285339627591e-05,
+      "loss": 2.6257,
+      "step": 33850
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.4822099833901567e-05,
+      "loss": 2.5956,
+      "step": 33860
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.4819914328175541e-05,
+      "loss": 2.6496,
+      "step": 33870
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.4817728822449515e-05,
+      "loss": 2.6139,
+      "step": 33880
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.4815543316723491e-05,
+      "loss": 2.5952,
+      "step": 33890
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.4813357810997466e-05,
+      "loss": 2.5977,
+      "step": 33900
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.4811172305271441e-05,
+      "loss": 2.6271,
+      "step": 33910
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.4808986799545416e-05,
+      "loss": 2.636,
+      "step": 33920
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.480680129381939e-05,
+      "loss": 2.6314,
+      "step": 33930
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.4804615788093366e-05,
+      "loss": 2.6509,
+      "step": 33940
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.4802430282367342e-05,
+      "loss": 2.6021,
+      "step": 33950
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.4800244776641317e-05,
+      "loss": 2.626,
+      "step": 33960
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.479805927091529e-05,
+      "loss": 2.6161,
+      "step": 33970
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.4795873765189266e-05,
+      "loss": 2.6381,
+      "step": 33980
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.4793688259463242e-05,
+      "loss": 2.5932,
+      "step": 33990
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.4791502753737218e-05,
+      "loss": 2.6396,
+      "step": 34000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.478931724801119e-05,
+      "loss": 2.6349,
+      "step": 34010
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.4787131742285166e-05,
+      "loss": 2.6267,
+      "step": 34020
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.4784946236559142e-05,
+      "loss": 2.6076,
+      "step": 34030
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.4782760730833116e-05,
+      "loss": 2.6119,
+      "step": 34040
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.478057522510709e-05,
+      "loss": 2.5942,
+      "step": 34050
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.4778389719381066e-05,
+      "loss": 2.6165,
+      "step": 34060
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.477620421365504e-05,
+      "loss": 2.5984,
+      "step": 34070
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.4774018707929016e-05,
+      "loss": 2.6729,
+      "step": 34080
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.4771833202202992e-05,
+      "loss": 2.6227,
+      "step": 34090
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.4769647696476964e-05,
+      "loss": 2.5982,
+      "step": 34100
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.476746219075094e-05,
+      "loss": 2.5755,
+      "step": 34110
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.4765276685024916e-05,
+      "loss": 2.6405,
+      "step": 34120
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.4763091179298892e-05,
+      "loss": 2.5881,
+      "step": 34130
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.4760905673572865e-05,
+      "loss": 2.6277,
+      "step": 34140
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.475872016784684e-05,
+      "loss": 2.5629,
+      "step": 34150
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.4756534662120816e-05,
+      "loss": 2.6408,
+      "step": 34160
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.4754349156394792e-05,
+      "loss": 2.6156,
+      "step": 34170
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.4752163650668765e-05,
+      "loss": 2.6346,
+      "step": 34180
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.474997814494274e-05,
+      "loss": 2.5913,
+      "step": 34190
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.4747792639216716e-05,
+      "loss": 2.6276,
+      "step": 34200
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.474560713349069e-05,
+      "loss": 2.5968,
+      "step": 34210
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.4743421627764667e-05,
+      "loss": 2.6021,
+      "step": 34220
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.474123612203864e-05,
+      "loss": 2.5918,
+      "step": 34230
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.4739050616312615e-05,
+      "loss": 2.62,
+      "step": 34240
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.473686511058659e-05,
+      "loss": 2.5585,
+      "step": 34250
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.4734679604860567e-05,
+      "loss": 2.6474,
+      "step": 34260
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.473249409913454e-05,
+      "loss": 2.6146,
+      "step": 34270
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.4730308593408515e-05,
+      "loss": 2.6583,
+      "step": 34280
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.4728123087682491e-05,
+      "loss": 2.6051,
+      "step": 34290
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.4725937581956467e-05,
+      "loss": 2.6074,
+      "step": 34300
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.4723752076230443e-05,
+      "loss": 2.5894,
+      "step": 34310
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.4721566570504415e-05,
+      "loss": 2.6071,
+      "step": 34320
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.4719381064778391e-05,
+      "loss": 2.6209,
+      "step": 34330
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.4717195559052367e-05,
+      "loss": 2.6249,
+      "step": 34340
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.4715010053326341e-05,
+      "loss": 2.6227,
+      "step": 34350
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.4712824547600315e-05,
+      "loss": 2.6114,
+      "step": 34360
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.4710639041874291e-05,
+      "loss": 2.6217,
+      "step": 34370
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.4708453536148265e-05,
+      "loss": 2.6438,
+      "step": 34380
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.4706268030422241e-05,
+      "loss": 2.6287,
+      "step": 34390
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.4704082524696215e-05,
+      "loss": 2.6399,
+      "step": 34400
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.470189701897019e-05,
+      "loss": 2.5831,
+      "step": 34410
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.4699711513244166e-05,
+      "loss": 2.596,
+      "step": 34420
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.4697526007518141e-05,
+      "loss": 2.614,
+      "step": 34430
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.4695340501792117e-05,
+      "loss": 2.641,
+      "step": 34440
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.469315499606609e-05,
+      "loss": 2.6436,
+      "step": 34450
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.4690969490340066e-05,
+      "loss": 2.6171,
+      "step": 34460
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.4688783984614042e-05,
+      "loss": 2.6153,
+      "step": 34470
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.4686598478888017e-05,
+      "loss": 2.5999,
+      "step": 34480
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.468441297316199e-05,
+      "loss": 2.6392,
+      "step": 34490
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.4682227467435966e-05,
+      "loss": 2.6508,
+      "step": 34500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.4680041961709942e-05,
+      "loss": 2.6191,
+      "step": 34510
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.4677856455983916e-05,
+      "loss": 2.5885,
+      "step": 34520
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.467567095025789e-05,
+      "loss": 2.6344,
+      "step": 34530
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.4673485444531866e-05,
+      "loss": 2.6037,
+      "step": 34540
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.467129993880584e-05,
+      "loss": 2.6453,
+      "step": 34550
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.4669114433079816e-05,
+      "loss": 2.5907,
+      "step": 34560
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.4666928927353792e-05,
+      "loss": 2.6248,
+      "step": 34570
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.4664743421627764e-05,
+      "loss": 2.6316,
+      "step": 34580
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.466255791590174e-05,
+      "loss": 2.6469,
+      "step": 34590
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.4660372410175716e-05,
+      "loss": 2.6059,
+      "step": 34600
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.4658186904449692e-05,
+      "loss": 2.6477,
+      "step": 34610
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.4656001398723664e-05,
+      "loss": 2.6439,
+      "step": 34620
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.465381589299764e-05,
+      "loss": 2.6199,
+      "step": 34630
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.4651630387271616e-05,
+      "loss": 2.5966,
+      "step": 34640
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.4649444881545592e-05,
+      "loss": 2.6561,
+      "step": 34650
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.4647259375819566e-05,
+      "loss": 2.5673,
+      "step": 34660
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.464507387009354e-05,
+      "loss": 2.6121,
+      "step": 34670
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.4642888364367516e-05,
+      "loss": 2.6601,
+      "step": 34680
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.464070285864149e-05,
+      "loss": 2.6633,
+      "step": 34690
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.4638517352915466e-05,
+      "loss": 2.623,
+      "step": 34700
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.463633184718944e-05,
+      "loss": 2.6028,
+      "step": 34710
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 2.6275,
+      "step": 34720
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.463196083573739e-05,
+      "loss": 2.6307,
+      "step": 34730
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.4629775330011367e-05,
+      "loss": 2.6287,
+      "step": 34740
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.4627589824285339e-05,
+      "loss": 2.6256,
+      "step": 34750
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.4625404318559315e-05,
+      "loss": 2.5943,
+      "step": 34760
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.462321881283329e-05,
+      "loss": 2.6189,
+      "step": 34770
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.4621033307107267e-05,
+      "loss": 2.6212,
+      "step": 34780
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.4618847801381243e-05,
+      "loss": 2.6245,
+      "step": 34790
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4616662295655215e-05,
+      "loss": 2.6287,
+      "step": 34800
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4614476789929191e-05,
+      "loss": 2.6168,
+      "step": 34810
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4612291284203167e-05,
+      "loss": 2.6186,
+      "step": 34820
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4610105778477141e-05,
+      "loss": 2.6604,
+      "step": 34830
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4607920272751115e-05,
+      "loss": 2.5996,
+      "step": 34840
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4605734767025091e-05,
+      "loss": 2.5828,
+      "step": 34850
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.4603549261299065e-05,
+      "loss": 2.6335,
+      "step": 34860
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.4601363755573041e-05,
+      "loss": 2.6441,
+      "step": 34870
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.4599178249847015e-05,
+      "loss": 2.5901,
+      "step": 34880
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.459699274412099e-05,
+      "loss": 2.639,
+      "step": 34890
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.4594807238394965e-05,
+      "loss": 2.6418,
+      "step": 34900
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.4592621732668941e-05,
+      "loss": 2.5845,
+      "step": 34910
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.4590436226942917e-05,
+      "loss": 2.6043,
+      "step": 34920
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.458825072121689e-05,
+      "loss": 2.6193,
+      "step": 34930
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.4586065215490865e-05,
+      "loss": 2.614,
+      "step": 34940
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.4583879709764841e-05,
+      "loss": 2.6349,
+      "step": 34950
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.4581694204038817e-05,
+      "loss": 2.5811,
+      "step": 34960
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.457950869831279e-05,
+      "loss": 2.6227,
+      "step": 34970
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.4577323192586766e-05,
+      "loss": 2.6436,
+      "step": 34980
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.4575137686860741e-05,
+      "loss": 2.6282,
+      "step": 34990
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.4572952181134716e-05,
+      "loss": 2.6268,
+      "step": 35000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.4570766675408692e-05,
+      "loss": 2.6641,
+      "step": 35010
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.4568581169682666e-05,
+      "loss": 2.6137,
+      "step": 35020
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.456639566395664e-05,
+      "loss": 2.6243,
+      "step": 35030
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.4564210158230616e-05,
+      "loss": 2.653,
+      "step": 35040
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.4562024652504592e-05,
+      "loss": 2.624,
+      "step": 35050
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.4559839146778564e-05,
+      "loss": 2.6536,
+      "step": 35060
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.455765364105254e-05,
+      "loss": 2.6064,
+      "step": 35070
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.4555468135326516e-05,
+      "loss": 2.623,
+      "step": 35080
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.4553282629600492e-05,
+      "loss": 2.6102,
+      "step": 35090
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.4551097123874464e-05,
+      "loss": 2.6418,
+      "step": 35100
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.454891161814844e-05,
+      "loss": 2.6235,
+      "step": 35110
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.4546726112422416e-05,
+      "loss": 2.6788,
+      "step": 35120
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.4544540606696392e-05,
+      "loss": 2.5969,
+      "step": 35130
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.4542355100970366e-05,
+      "loss": 2.6423,
+      "step": 35140
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.454016959524434e-05,
+      "loss": 2.6263,
+      "step": 35150
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.4537984089518316e-05,
+      "loss": 2.5802,
+      "step": 35160
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.453579858379229e-05,
+      "loss": 2.6163,
+      "step": 35170
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.4533613078066266e-05,
+      "loss": 2.6096,
+      "step": 35180
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.453142757234024e-05,
+      "loss": 2.6179,
+      "step": 35190
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.4529242066614215e-05,
+      "loss": 2.6601,
+      "step": 35200
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.452705656088819e-05,
+      "loss": 2.6415,
+      "step": 35210
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.4524871055162166e-05,
+      "loss": 2.6442,
+      "step": 35220
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.4522685549436139e-05,
+      "loss": 2.6202,
+      "step": 35230
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.4520500043710115e-05,
+      "loss": 2.6264,
+      "step": 35240
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.451831453798409e-05,
+      "loss": 2.6059,
+      "step": 35250
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 2.6807,
+      "step": 35260
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.4513943526532042e-05,
+      "loss": 2.598,
+      "step": 35270
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.4511758020806015e-05,
+      "loss": 2.6172,
+      "step": 35280
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.450957251507999e-05,
+      "loss": 2.5839,
+      "step": 35290
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.4507387009353967e-05,
+      "loss": 2.6561,
+      "step": 35300
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.450520150362794e-05,
+      "loss": 2.6488,
+      "step": 35310
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.4503015997901915e-05,
+      "loss": 2.591,
+      "step": 35320
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.450083049217589e-05,
+      "loss": 2.6123,
+      "step": 35330
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.4498644986449865e-05,
+      "loss": 2.5991,
+      "step": 35340
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.4496459480723841e-05,
+      "loss": 2.6449,
+      "step": 35350
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.4494273974997817e-05,
+      "loss": 2.6123,
+      "step": 35360
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.449208846927179e-05,
+      "loss": 2.6561,
+      "step": 35370
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.4489902963545765e-05,
+      "loss": 2.6046,
+      "step": 35380
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.4487717457819741e-05,
+      "loss": 2.5955,
+      "step": 35390
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.4485531952093717e-05,
+      "loss": 2.6376,
+      "step": 35400
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.448334644636769e-05,
+      "loss": 2.6336,
+      "step": 35410
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.4481160940641665e-05,
+      "loss": 2.6394,
+      "step": 35420
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.4478975434915641e-05,
+      "loss": 2.6285,
+      "step": 35430
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.4476789929189617e-05,
+      "loss": 2.5992,
+      "step": 35440
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.447460442346359e-05,
+      "loss": 2.6536,
+      "step": 35450
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.4472418917737565e-05,
+      "loss": 2.6103,
+      "step": 35460
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.4470233412011541e-05,
+      "loss": 2.6202,
+      "step": 35470
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.4468047906285515e-05,
+      "loss": 2.6883,
+      "step": 35480
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.4465862400559491e-05,
+      "loss": 2.6072,
+      "step": 35490
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.4463676894833466e-05,
+      "loss": 2.5903,
+      "step": 35500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.446149138910744e-05,
+      "loss": 2.6403,
+      "step": 35510
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.4459305883381416e-05,
+      "loss": 2.6233,
+      "step": 35520
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.4457120377655391e-05,
+      "loss": 2.6379,
+      "step": 35530
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.4454934871929364e-05,
+      "loss": 2.635,
+      "step": 35540
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.445274936620334e-05,
+      "loss": 2.6423,
+      "step": 35550
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.4450563860477316e-05,
+      "loss": 2.6317,
+      "step": 35560
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.4448378354751292e-05,
+      "loss": 2.6303,
+      "step": 35570
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.4446192849025264e-05,
+      "loss": 2.6299,
+      "step": 35580
+    },
+    {
+      "epoch": 7.0,
+      "eval_BLEU": 9.0004,
+      "eval_gen_len": 11.7704,
+      "eval_loss": 3.065833568572998,
+      "eval_runtime": 173.6882,
+      "eval_samples_per_second": 61.967,
+      "eval_steps_per_second": 0.248,
+      "step": 35588
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.444400734329924e-05,
+      "loss": 2.6224,
+      "step": 35590
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.4441821837573216e-05,
+      "loss": 2.4841,
+      "step": 35600
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.4439636331847192e-05,
+      "loss": 2.5085,
+      "step": 35610
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4437450826121166e-05,
+      "loss": 2.5028,
+      "step": 35620
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.443526532039514e-05,
+      "loss": 2.5097,
+      "step": 35630
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4433079814669116e-05,
+      "loss": 2.5293,
+      "step": 35640
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.443089430894309e-05,
+      "loss": 2.5198,
+      "step": 35650
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4428708803217066e-05,
+      "loss": 2.5108,
+      "step": 35660
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.442652329749104e-05,
+      "loss": 2.4726,
+      "step": 35670
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.4424337791765014e-05,
+      "loss": 2.5113,
+      "step": 35680
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.442215228603899e-05,
+      "loss": 2.5007,
+      "step": 35690
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.4419966780312966e-05,
+      "loss": 2.4987,
+      "step": 35700
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.4417781274586942e-05,
+      "loss": 2.5225,
+      "step": 35710
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.4415595768860915e-05,
+      "loss": 2.4973,
+      "step": 35720
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.441341026313489e-05,
+      "loss": 2.5233,
+      "step": 35730
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.4411224757408866e-05,
+      "loss": 2.4959,
+      "step": 35740
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.4409039251682842e-05,
+      "loss": 2.528,
+      "step": 35750
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.4406853745956815e-05,
+      "loss": 2.5215,
+      "step": 35760
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.440466824023079e-05,
+      "loss": 2.5047,
+      "step": 35770
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.4402482734504766e-05,
+      "loss": 2.5225,
+      "step": 35780
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.440029722877874e-05,
+      "loss": 2.518,
+      "step": 35790
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.4398111723052715e-05,
+      "loss": 2.5393,
+      "step": 35800
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.439592621732669e-05,
+      "loss": 2.5102,
+      "step": 35810
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4393740711600665e-05,
+      "loss": 2.5211,
+      "step": 35820
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.439155520587464e-05,
+      "loss": 2.5038,
+      "step": 35830
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4389369700148617e-05,
+      "loss": 2.5336,
+      "step": 35840
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4387184194422589e-05,
+      "loss": 2.5199,
+      "step": 35850
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4384998688696565e-05,
+      "loss": 2.5313,
+      "step": 35860
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4382813182970541e-05,
+      "loss": 2.5155,
+      "step": 35870
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4380627677244517e-05,
+      "loss": 2.4767,
+      "step": 35880
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.437844217151849e-05,
+      "loss": 2.5239,
+      "step": 35890
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4376256665792465e-05,
+      "loss": 2.5088,
+      "step": 35900
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4374071160066441e-05,
+      "loss": 2.4999,
+      "step": 35910
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4371885654340417e-05,
+      "loss": 2.5143,
+      "step": 35920
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.436970014861439e-05,
+      "loss": 2.5523,
+      "step": 35930
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4367514642888365e-05,
+      "loss": 2.5432,
+      "step": 35940
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4365329137162341e-05,
+      "loss": 2.5429,
+      "step": 35950
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4363143631436315e-05,
+      "loss": 2.5062,
+      "step": 35960
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4360958125710291e-05,
+      "loss": 2.5387,
+      "step": 35970
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4358772619984265e-05,
+      "loss": 2.5468,
+      "step": 35980
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.435658711425824e-05,
+      "loss": 2.554,
+      "step": 35990
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4354401608532215e-05,
+      "loss": 2.5055,
+      "step": 36000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4352216102806191e-05,
+      "loss": 2.5176,
+      "step": 36010
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4350030597080164e-05,
+      "loss": 2.5042,
+      "step": 36020
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.434784509135414e-05,
+      "loss": 2.5234,
+      "step": 36030
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4345659585628116e-05,
+      "loss": 2.4873,
+      "step": 36040
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4343474079902091e-05,
+      "loss": 2.5171,
+      "step": 36050
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4341288574176067e-05,
+      "loss": 2.5179,
+      "step": 36060
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.433910306845004e-05,
+      "loss": 2.4781,
+      "step": 36070
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4336917562724016e-05,
+      "loss": 2.502,
+      "step": 36080
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4334732056997992e-05,
+      "loss": 2.5169,
+      "step": 36090
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4332546551271966e-05,
+      "loss": 2.5228,
+      "step": 36100
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.433036104554594e-05,
+      "loss": 2.5278,
+      "step": 36110
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4328175539819916e-05,
+      "loss": 2.5654,
+      "step": 36120
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.432599003409389e-05,
+      "loss": 2.5622,
+      "step": 36130
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4323804528367866e-05,
+      "loss": 2.5083,
+      "step": 36140
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.432161902264184e-05,
+      "loss": 2.5181,
+      "step": 36150
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4319433516915814e-05,
+      "loss": 2.4952,
+      "step": 36160
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.431724801118979e-05,
+      "loss": 2.5516,
+      "step": 36170
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4315062505463766e-05,
+      "loss": 2.527,
+      "step": 36180
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4312876999737742e-05,
+      "loss": 2.5103,
+      "step": 36190
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4310691494011714e-05,
+      "loss": 2.5282,
+      "step": 36200
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.430850598828569e-05,
+      "loss": 2.507,
+      "step": 36210
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4306320482559666e-05,
+      "loss": 2.5624,
+      "step": 36220
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4304134976833642e-05,
+      "loss": 2.5301,
+      "step": 36230
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4301949471107614e-05,
+      "loss": 2.516,
+      "step": 36240
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.429976396538159e-05,
+      "loss": 2.5368,
+      "step": 36250
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4297578459655566e-05,
+      "loss": 2.5121,
+      "step": 36260
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.429539295392954e-05,
+      "loss": 2.5353,
+      "step": 36270
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4293207448203515e-05,
+      "loss": 2.5057,
+      "step": 36280
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.429102194247749e-05,
+      "loss": 2.4848,
+      "step": 36290
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4288836436751465e-05,
+      "loss": 2.5518,
+      "step": 36300
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.428665093102544e-05,
+      "loss": 2.5586,
+      "step": 36310
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4284465425299416e-05,
+      "loss": 2.5392,
+      "step": 36320
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4282279919573389e-05,
+      "loss": 2.5024,
+      "step": 36330
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4280094413847365e-05,
+      "loss": 2.542,
+      "step": 36340
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.427790890812134e-05,
+      "loss": 2.5446,
+      "step": 36350
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4275723402395317e-05,
+      "loss": 2.4862,
+      "step": 36360
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4273537896669289e-05,
+      "loss": 2.4777,
+      "step": 36370
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4271352390943265e-05,
+      "loss": 2.4811,
+      "step": 36380
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.426916688521724e-05,
+      "loss": 2.5255,
+      "step": 36390
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4266981379491217e-05,
+      "loss": 2.4866,
+      "step": 36400
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4264795873765191e-05,
+      "loss": 2.4717,
+      "step": 36410
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4262610368039165e-05,
+      "loss": 2.5218,
+      "step": 36420
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4260424862313141e-05,
+      "loss": 2.5255,
+      "step": 36430
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4258239356587115e-05,
+      "loss": 2.5181,
+      "step": 36440
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4256053850861091e-05,
+      "loss": 2.5323,
+      "step": 36450
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4253868345135065e-05,
+      "loss": 2.524,
+      "step": 36460
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.425168283940904e-05,
+      "loss": 2.5041,
+      "step": 36470
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4249497333683015e-05,
+      "loss": 2.4976,
+      "step": 36480
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4247311827956991e-05,
+      "loss": 2.4975,
+      "step": 36490
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4245126322230964e-05,
+      "loss": 2.4999,
+      "step": 36500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.424294081650494e-05,
+      "loss": 2.5149,
+      "step": 36510
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4240755310778915e-05,
+      "loss": 2.5422,
+      "step": 36520
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4238569805052891e-05,
+      "loss": 2.5297,
+      "step": 36530
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4236384299326867e-05,
+      "loss": 2.5599,
+      "step": 36540
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.423419879360084e-05,
+      "loss": 2.5279,
+      "step": 36550
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4232013287874816e-05,
+      "loss": 2.5264,
+      "step": 36560
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4229827782148791e-05,
+      "loss": 2.521,
+      "step": 36570
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.4227642276422766e-05,
+      "loss": 2.5583,
+      "step": 36580
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.422545677069674e-05,
+      "loss": 2.5408,
+      "step": 36590
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.4223271264970716e-05,
+      "loss": 2.5542,
+      "step": 36600
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.422108575924469e-05,
+      "loss": 2.577,
+      "step": 36610
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.4218900253518666e-05,
+      "loss": 2.5228,
+      "step": 36620
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.421671474779264e-05,
+      "loss": 2.5537,
+      "step": 36630
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.4214529242066614e-05,
+      "loss": 2.507,
+      "step": 36640
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.421234373634059e-05,
+      "loss": 2.5079,
+      "step": 36650
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.4210158230614566e-05,
+      "loss": 2.5668,
+      "step": 36660
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.4207972724888542e-05,
+      "loss": 2.5629,
+      "step": 36670
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.4205787219162514e-05,
+      "loss": 2.5107,
+      "step": 36680
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.420360171343649e-05,
+      "loss": 2.5197,
+      "step": 36690
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.4201416207710466e-05,
+      "loss": 2.5266,
+      "step": 36700
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.4199230701984442e-05,
+      "loss": 2.5082,
+      "step": 36710
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.4197045196258414e-05,
+      "loss": 2.5107,
+      "step": 36720
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.419485969053239e-05,
+      "loss": 2.5285,
+      "step": 36730
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.4192674184806366e-05,
+      "loss": 2.5408,
+      "step": 36740
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.419048867908034e-05,
+      "loss": 2.5581,
+      "step": 36750
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.4188303173354316e-05,
+      "loss": 2.5208,
+      "step": 36760
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.418611766762829e-05,
+      "loss": 2.517,
+      "step": 36770
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.4183932161902265e-05,
+      "loss": 2.5283,
+      "step": 36780
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.418174665617624e-05,
+      "loss": 2.5346,
+      "step": 36790
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.4179561150450216e-05,
+      "loss": 2.5434,
+      "step": 36800
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.4177375644724189e-05,
+      "loss": 2.5218,
+      "step": 36810
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.4175190138998165e-05,
+      "loss": 2.5012,
+      "step": 36820
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.417300463327214e-05,
+      "loss": 2.5297,
+      "step": 36830
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.4170819127546116e-05,
+      "loss": 2.5275,
+      "step": 36840
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.4168633621820089e-05,
+      "loss": 2.5569,
+      "step": 36850
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.4166448116094065e-05,
+      "loss": 2.5619,
+      "step": 36860
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.416426261036804e-05,
+      "loss": 2.5522,
+      "step": 36870
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.4162077104642017e-05,
+      "loss": 2.5694,
+      "step": 36880
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.415989159891599e-05,
+      "loss": 2.5272,
+      "step": 36890
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.4157706093189965e-05,
+      "loss": 2.5297,
+      "step": 36900
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.415552058746394e-05,
+      "loss": 2.5262,
+      "step": 36910
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.4153335081737915e-05,
+      "loss": 2.5228,
+      "step": 36920
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.415114957601189e-05,
+      "loss": 2.6,
+      "step": 36930
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.4148964070285865e-05,
+      "loss": 2.5689,
+      "step": 36940
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.414677856455984e-05,
+      "loss": 2.5453,
+      "step": 36950
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.4144593058833815e-05,
+      "loss": 2.5183,
+      "step": 36960
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.4142407553107791e-05,
+      "loss": 2.5504,
+      "step": 36970
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.4140222047381767e-05,
+      "loss": 2.5513,
+      "step": 36980
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.413803654165574e-05,
+      "loss": 2.5312,
+      "step": 36990
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.4135851035929715e-05,
+      "loss": 2.5559,
+      "step": 37000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.4133665530203691e-05,
+      "loss": 2.5568,
+      "step": 37010
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.4131480024477667e-05,
+      "loss": 2.5698,
+      "step": 37020
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.412929451875164e-05,
+      "loss": 2.5388,
+      "step": 37030
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.4127109013025615e-05,
+      "loss": 2.5198,
+      "step": 37040
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.4124923507299591e-05,
+      "loss": 2.5142,
+      "step": 37050
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.4122738001573565e-05,
+      "loss": 2.5702,
+      "step": 37060
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.412055249584754e-05,
+      "loss": 2.5634,
+      "step": 37070
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.4118366990121515e-05,
+      "loss": 2.5483,
+      "step": 37080
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.411618148439549e-05,
+      "loss": 2.5568,
+      "step": 37090
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.4113995978669466e-05,
+      "loss": 2.4919,
+      "step": 37100
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.4111810472943441e-05,
+      "loss": 2.5414,
+      "step": 37110
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.4109624967217414e-05,
+      "loss": 2.5132,
+      "step": 37120
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.410743946149139e-05,
+      "loss": 2.5057,
+      "step": 37130
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.4105253955765366e-05,
+      "loss": 2.5076,
+      "step": 37140
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.4103068450039342e-05,
+      "loss": 2.5347,
+      "step": 37150
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.4100882944313314e-05,
+      "loss": 2.5492,
+      "step": 37160
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.409869743858729e-05,
+      "loss": 2.5162,
+      "step": 37170
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.4096511932861266e-05,
+      "loss": 2.5624,
+      "step": 37180
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.4094326427135242e-05,
+      "loss": 2.5913,
+      "step": 37190
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.4092140921409214e-05,
+      "loss": 2.5151,
+      "step": 37200
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.408995541568319e-05,
+      "loss": 2.5652,
+      "step": 37210
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.4087769909957166e-05,
+      "loss": 2.5148,
+      "step": 37220
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.408558440423114e-05,
+      "loss": 2.5703,
+      "step": 37230
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.4083398898505116e-05,
+      "loss": 2.5385,
+      "step": 37240
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.408121339277909e-05,
+      "loss": 2.5583,
+      "step": 37250
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.4079027887053064e-05,
+      "loss": 2.5318,
+      "step": 37260
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.407684238132704e-05,
+      "loss": 2.5601,
+      "step": 37270
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.4074656875601016e-05,
+      "loss": 2.5456,
+      "step": 37280
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.4072471369874989e-05,
+      "loss": 2.5444,
+      "step": 37290
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.4070285864148964e-05,
+      "loss": 2.5223,
+      "step": 37300
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.406810035842294e-05,
+      "loss": 2.5586,
+      "step": 37310
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.4065914852696916e-05,
+      "loss": 2.5585,
+      "step": 37320
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.4063729346970892e-05,
+      "loss": 2.5264,
+      "step": 37330
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.4061543841244865e-05,
+      "loss": 2.5765,
+      "step": 37340
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.405935833551884e-05,
+      "loss": 2.5331,
+      "step": 37350
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.4057172829792816e-05,
+      "loss": 2.4986,
+      "step": 37360
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.405498732406679e-05,
+      "loss": 2.5208,
+      "step": 37370
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.4052801818340765e-05,
+      "loss": 2.4853,
+      "step": 37380
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.405061631261474e-05,
+      "loss": 2.5645,
+      "step": 37390
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.4048430806888715e-05,
+      "loss": 2.5415,
+      "step": 37400
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.404624530116269e-05,
+      "loss": 2.5206,
+      "step": 37410
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.4044059795436665e-05,
+      "loss": 2.5347,
+      "step": 37420
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.4041874289710639e-05,
+      "loss": 2.5066,
+      "step": 37430
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.4039688783984615e-05,
+      "loss": 2.5569,
+      "step": 37440
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.403750327825859e-05,
+      "loss": 2.5551,
+      "step": 37450
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.4035317772532567e-05,
+      "loss": 2.5812,
+      "step": 37460
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.403313226680654e-05,
+      "loss": 2.5655,
+      "step": 37470
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.4030946761080515e-05,
+      "loss": 2.4953,
+      "step": 37480
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.4028761255354491e-05,
+      "loss": 2.5439,
+      "step": 37490
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.4026575749628467e-05,
+      "loss": 2.5352,
+      "step": 37500
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.402439024390244e-05,
+      "loss": 2.5252,
+      "step": 37510
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.4022204738176415e-05,
+      "loss": 2.5614,
+      "step": 37520
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.4020019232450391e-05,
+      "loss": 2.5331,
+      "step": 37530
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.4017833726724365e-05,
+      "loss": 2.5411,
+      "step": 37540
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.401564822099834e-05,
+      "loss": 2.5302,
+      "step": 37550
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.4013462715272315e-05,
+      "loss": 2.5202,
+      "step": 37560
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.401127720954629e-05,
+      "loss": 2.5696,
+      "step": 37570
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.4009091703820265e-05,
+      "loss": 2.5431,
+      "step": 37580
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.4006906198094241e-05,
+      "loss": 2.5304,
+      "step": 37590
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.4004720692368214e-05,
+      "loss": 2.5238,
+      "step": 37600
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.400253518664219e-05,
+      "loss": 2.5492,
+      "step": 37610
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.4000349680916165e-05,
+      "loss": 2.5142,
+      "step": 37620
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.3998164175190141e-05,
+      "loss": 2.5758,
+      "step": 37630
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.3995978669464114e-05,
+      "loss": 2.5542,
+      "step": 37640
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.399379316373809e-05,
+      "loss": 2.5638,
+      "step": 37650
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.3991607658012066e-05,
+      "loss": 2.5781,
+      "step": 37660
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.3989422152286041e-05,
+      "loss": 2.5101,
+      "step": 37670
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.3987236646560016e-05,
+      "loss": 2.5692,
+      "step": 37680
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.398505114083399e-05,
+      "loss": 2.5309,
+      "step": 37690
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.3982865635107966e-05,
+      "loss": 2.5694,
+      "step": 37700
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.398068012938194e-05,
+      "loss": 2.5596,
+      "step": 37710
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.3978494623655916e-05,
+      "loss": 2.514,
+      "step": 37720
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.397630911792989e-05,
+      "loss": 2.5132,
+      "step": 37730
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.3974123612203864e-05,
+      "loss": 2.492,
+      "step": 37740
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.397193810647784e-05,
+      "loss": 2.5231,
+      "step": 37750
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.3969752600751816e-05,
+      "loss": 2.5348,
+      "step": 37760
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.3967567095025788e-05,
+      "loss": 2.5541,
+      "step": 37770
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.3965381589299764e-05,
+      "loss": 2.5441,
+      "step": 37780
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.396319608357374e-05,
+      "loss": 2.5602,
+      "step": 37790
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.3961010577847716e-05,
+      "loss": 2.5492,
+      "step": 37800
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.3958825072121692e-05,
+      "loss": 2.5634,
+      "step": 37810
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.3956639566395664e-05,
+      "loss": 2.5281,
+      "step": 37820
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.395445406066964e-05,
+      "loss": 2.5744,
+      "step": 37830
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.3952268554943616e-05,
+      "loss": 2.5509,
+      "step": 37840
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.395008304921759e-05,
+      "loss": 2.5545,
+      "step": 37850
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.3947897543491565e-05,
+      "loss": 2.5784,
+      "step": 37860
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.394571203776554e-05,
+      "loss": 2.5314,
+      "step": 37870
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.3943526532039515e-05,
+      "loss": 2.5681,
+      "step": 37880
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.394134102631349e-05,
+      "loss": 2.5292,
+      "step": 37890
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.3939155520587465e-05,
+      "loss": 2.5455,
+      "step": 37900
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.3936970014861439e-05,
+      "loss": 2.5524,
+      "step": 37910
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.3934784509135415e-05,
+      "loss": 2.5342,
+      "step": 37920
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.393259900340939e-05,
+      "loss": 2.5592,
+      "step": 37930
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.3930413497683366e-05,
+      "loss": 2.5203,
+      "step": 37940
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.3928227991957339e-05,
+      "loss": 2.5879,
+      "step": 37950
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.3926042486231315e-05,
+      "loss": 2.5612,
+      "step": 37960
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.392385698050529e-05,
+      "loss": 2.5754,
+      "step": 37970
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.3921671474779267e-05,
+      "loss": 2.4902,
+      "step": 37980
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.3919485969053239e-05,
+      "loss": 2.5336,
+      "step": 37990
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.3917300463327215e-05,
+      "loss": 2.5132,
+      "step": 38000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.3915114957601191e-05,
+      "loss": 2.5622,
+      "step": 38010
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.3912929451875165e-05,
+      "loss": 2.4987,
+      "step": 38020
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.3910743946149141e-05,
+      "loss": 2.5373,
+      "step": 38030
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.3908558440423115e-05,
+      "loss": 2.4906,
+      "step": 38040
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.390637293469709e-05,
+      "loss": 2.5254,
+      "step": 38050
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.3904187428971065e-05,
+      "loss": 2.5428,
+      "step": 38060
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.3902001923245041e-05,
+      "loss": 2.5474,
+      "step": 38070
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.3899816417519014e-05,
+      "loss": 2.5379,
+      "step": 38080
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.389763091179299e-05,
+      "loss": 2.5436,
+      "step": 38090
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.3895445406066965e-05,
+      "loss": 2.5172,
+      "step": 38100
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.3893259900340941e-05,
+      "loss": 2.5458,
+      "step": 38110
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.3891074394614914e-05,
+      "loss": 2.5251,
+      "step": 38120
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 2.5518,
+      "step": 38130
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.3886703383162865e-05,
+      "loss": 2.5226,
+      "step": 38140
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.3884517877436841e-05,
+      "loss": 2.5316,
+      "step": 38150
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.3882332371710815e-05,
+      "loss": 2.5447,
+      "step": 38160
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.388014686598479e-05,
+      "loss": 2.5181,
+      "step": 38170
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.3877961360258766e-05,
+      "loss": 2.5396,
+      "step": 38180
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.387577585453274e-05,
+      "loss": 2.5492,
+      "step": 38190
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.3873590348806716e-05,
+      "loss": 2.5663,
+      "step": 38200
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.387140484308069e-05,
+      "loss": 2.5616,
+      "step": 38210
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.3869219337354664e-05,
+      "loss": 2.5561,
+      "step": 38220
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.386703383162864e-05,
+      "loss": 2.5633,
+      "step": 38230
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.3864848325902616e-05,
+      "loss": 2.5586,
+      "step": 38240
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.3862662820176588e-05,
+      "loss": 2.5632,
+      "step": 38250
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.3860477314450564e-05,
+      "loss": 2.502,
+      "step": 38260
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.385829180872454e-05,
+      "loss": 2.5806,
+      "step": 38270
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.3856106302998516e-05,
+      "loss": 2.5621,
+      "step": 38280
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.3853920797272492e-05,
+      "loss": 2.5303,
+      "step": 38290
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.3851735291546464e-05,
+      "loss": 2.5661,
+      "step": 38300
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.384954978582044e-05,
+      "loss": 2.57,
+      "step": 38310
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.3847364280094416e-05,
+      "loss": 2.5483,
+      "step": 38320
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.384517877436839e-05,
+      "loss": 2.558,
+      "step": 38330
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.3842993268642364e-05,
+      "loss": 2.571,
+      "step": 38340
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.384080776291634e-05,
+      "loss": 2.5514,
+      "step": 38350
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.3838622257190314e-05,
+      "loss": 2.5586,
+      "step": 38360
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.383643675146429e-05,
+      "loss": 2.55,
+      "step": 38370
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.3834251245738266e-05,
+      "loss": 2.5739,
+      "step": 38380
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.3832065740012239e-05,
+      "loss": 2.5244,
+      "step": 38390
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.3829880234286215e-05,
+      "loss": 2.5436,
+      "step": 38400
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.382769472856019e-05,
+      "loss": 2.5402,
+      "step": 38410
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.3825509222834166e-05,
+      "loss": 2.5408,
+      "step": 38420
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.3823323717108139e-05,
+      "loss": 2.5415,
+      "step": 38430
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.3821138211382115e-05,
+      "loss": 2.5447,
+      "step": 38440
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.381895270565609e-05,
+      "loss": 2.5787,
+      "step": 38450
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.3816767199930066e-05,
+      "loss": 2.5809,
+      "step": 38460
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.3814581694204039e-05,
+      "loss": 2.5703,
+      "step": 38470
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.3812396188478015e-05,
+      "loss": 2.5923,
+      "step": 38480
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.381021068275199e-05,
+      "loss": 2.5796,
+      "step": 38490
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.3808025177025965e-05,
+      "loss": 2.5552,
+      "step": 38500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.380583967129994e-05,
+      "loss": 2.5744,
+      "step": 38510
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.3803654165573915e-05,
+      "loss": 2.5865,
+      "step": 38520
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.3801468659847889e-05,
+      "loss": 2.5409,
+      "step": 38530
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.3799283154121865e-05,
+      "loss": 2.5716,
+      "step": 38540
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.3797097648395841e-05,
+      "loss": 2.5732,
+      "step": 38550
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.3794912142669813e-05,
+      "loss": 2.546,
+      "step": 38560
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.379272663694379e-05,
+      "loss": 2.5463,
+      "step": 38570
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.3790541131217765e-05,
+      "loss": 2.5725,
+      "step": 38580
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.3788355625491741e-05,
+      "loss": 2.5567,
+      "step": 38590
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.3786170119765714e-05,
+      "loss": 2.5459,
+      "step": 38600
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.378398461403969e-05,
+      "loss": 2.5209,
+      "step": 38610
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.3781799108313665e-05,
+      "loss": 2.5391,
+      "step": 38620
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.3779613602587641e-05,
+      "loss": 2.5603,
+      "step": 38630
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.3777428096861615e-05,
+      "loss": 2.5445,
+      "step": 38640
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.377524259113559e-05,
+      "loss": 2.5582,
+      "step": 38650
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.3773057085409565e-05,
+      "loss": 2.5686,
+      "step": 38660
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.377087157968354e-05,
+      "loss": 2.5295,
+      "step": 38670
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.3768686073957515e-05,
+      "loss": 2.5831,
+      "step": 38680
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.376650056823149e-05,
+      "loss": 2.5572,
+      "step": 38690
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.3764315062505464e-05,
+      "loss": 2.5216,
+      "step": 38700
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.376212955677944e-05,
+      "loss": 2.5412,
+      "step": 38710
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.3759944051053416e-05,
+      "loss": 2.4929,
+      "step": 38720
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.3757758545327391e-05,
+      "loss": 2.5791,
+      "step": 38730
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.3755573039601364e-05,
+      "loss": 2.539,
+      "step": 38740
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.375338753387534e-05,
+      "loss": 2.5551,
+      "step": 38750
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.3751202028149316e-05,
+      "loss": 2.5351,
+      "step": 38760
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.3749016522423292e-05,
+      "loss": 2.5854,
+      "step": 38770
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.3746831016697264e-05,
+      "loss": 2.5122,
+      "step": 38780
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.374464551097124e-05,
+      "loss": 2.5106,
+      "step": 38790
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.3742460005245216e-05,
+      "loss": 2.5786,
+      "step": 38800
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.374027449951919e-05,
+      "loss": 2.5285,
+      "step": 38810
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.3738088993793164e-05,
+      "loss": 2.5715,
+      "step": 38820
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.373590348806714e-05,
+      "loss": 2.5268,
+      "step": 38830
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.3733717982341114e-05,
+      "loss": 2.5826,
+      "step": 38840
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.373153247661509e-05,
+      "loss": 2.5556,
+      "step": 38850
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.3729346970889066e-05,
+      "loss": 2.5434,
+      "step": 38860
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.3727161465163039e-05,
+      "loss": 2.5863,
+      "step": 38870
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.3724975959437014e-05,
+      "loss": 2.5344,
+      "step": 38880
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.372279045371099e-05,
+      "loss": 2.5645,
+      "step": 38890
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.3720604947984966e-05,
+      "loss": 2.5848,
+      "step": 38900
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.3718419442258939e-05,
+      "loss": 2.5786,
+      "step": 38910
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.3716233936532915e-05,
+      "loss": 2.5825,
+      "step": 38920
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.371404843080689e-05,
+      "loss": 2.5727,
+      "step": 38930
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.3711862925080866e-05,
+      "loss": 2.4969,
+      "step": 38940
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.3709677419354839e-05,
+      "loss": 2.5269,
+      "step": 38950
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.3707491913628815e-05,
+      "loss": 2.5429,
+      "step": 38960
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.370530640790279e-05,
+      "loss": 2.5291,
+      "step": 38970
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.3703120902176765e-05,
+      "loss": 2.5456,
+      "step": 38980
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.370093539645074e-05,
+      "loss": 2.5558,
+      "step": 38990
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.3698749890724715e-05,
+      "loss": 2.5571,
+      "step": 39000
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.3696564384998689e-05,
+      "loss": 2.5444,
+      "step": 39010
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.3694378879272665e-05,
+      "loss": 2.5251,
+      "step": 39020
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.369219337354664e-05,
+      "loss": 2.5454,
+      "step": 39030
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.3690007867820613e-05,
+      "loss": 2.5734,
+      "step": 39040
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.3687822362094589e-05,
+      "loss": 2.5536,
+      "step": 39050
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.3685636856368565e-05,
+      "loss": 2.5656,
+      "step": 39060
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.368345135064254e-05,
+      "loss": 2.5816,
+      "step": 39070
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.3681265844916517e-05,
+      "loss": 2.5818,
+      "step": 39080
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.367908033919049e-05,
+      "loss": 2.5197,
+      "step": 39090
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.3676894833464465e-05,
+      "loss": 2.5272,
+      "step": 39100
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.3674709327738441e-05,
+      "loss": 2.5312,
+      "step": 39110
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.3672523822012415e-05,
+      "loss": 2.5506,
+      "step": 39120
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.367033831628639e-05,
+      "loss": 2.5436,
+      "step": 39130
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.3668152810560365e-05,
+      "loss": 2.5879,
+      "step": 39140
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.366596730483434e-05,
+      "loss": 2.5237,
+      "step": 39150
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.3663781799108315e-05,
+      "loss": 2.5303,
+      "step": 39160
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.366159629338229e-05,
+      "loss": 2.5355,
+      "step": 39170
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.3659410787656264e-05,
+      "loss": 2.5655,
+      "step": 39180
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.365722528193024e-05,
+      "loss": 2.5309,
+      "step": 39190
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.3655039776204215e-05,
+      "loss": 2.5929,
+      "step": 39200
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.3652854270478191e-05,
+      "loss": 2.5436,
+      "step": 39210
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.3650668764752164e-05,
+      "loss": 2.5541,
+      "step": 39220
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.364848325902614e-05,
+      "loss": 2.5347,
+      "step": 39230
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.3646297753300116e-05,
+      "loss": 2.5585,
+      "step": 39240
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.3644112247574091e-05,
+      "loss": 2.5443,
+      "step": 39250
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.3641926741848064e-05,
+      "loss": 2.5965,
+      "step": 39260
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.363974123612204e-05,
+      "loss": 2.5247,
+      "step": 39270
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.3637555730396016e-05,
+      "loss": 2.5816,
+      "step": 39280
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.363537022466999e-05,
+      "loss": 2.5158,
+      "step": 39290
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.3633184718943964e-05,
+      "loss": 2.5713,
+      "step": 39300
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.363099921321794e-05,
+      "loss": 2.5705,
+      "step": 39310
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.3628813707491914e-05,
+      "loss": 2.5205,
+      "step": 39320
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.362662820176589e-05,
+      "loss": 2.5297,
+      "step": 39330
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.3624442696039866e-05,
+      "loss": 2.5626,
+      "step": 39340
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.3622257190313838e-05,
+      "loss": 2.5477,
+      "step": 39350
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.3620071684587814e-05,
+      "loss": 2.5747,
+      "step": 39360
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.361788617886179e-05,
+      "loss": 2.5544,
+      "step": 39370
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.3615700673135766e-05,
+      "loss": 2.5989,
+      "step": 39380
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.3613515167409738e-05,
+      "loss": 2.5797,
+      "step": 39390
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.3611329661683714e-05,
+      "loss": 2.5425,
+      "step": 39400
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.360914415595769e-05,
+      "loss": 2.5431,
+      "step": 39410
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.3606958650231666e-05,
+      "loss": 2.6174,
+      "step": 39420
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.360477314450564e-05,
+      "loss": 2.562,
+      "step": 39430
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.3602587638779614e-05,
+      "loss": 2.5449,
+      "step": 39440
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.360040213305359e-05,
+      "loss": 2.5601,
+      "step": 39450
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.3598216627327565e-05,
+      "loss": 2.5744,
+      "step": 39460
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.359603112160154e-05,
+      "loss": 2.5659,
+      "step": 39470
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.3593845615875515e-05,
+      "loss": 2.5749,
+      "step": 39480
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.3591660110149489e-05,
+      "loss": 2.5241,
+      "step": 39490
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.3589474604423465e-05,
+      "loss": 2.565,
+      "step": 39500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.358728909869744e-05,
+      "loss": 2.5835,
+      "step": 39510
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.3585103592971413e-05,
+      "loss": 2.5654,
+      "step": 39520
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.3582918087245389e-05,
+      "loss": 2.5357,
+      "step": 39530
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.3580732581519365e-05,
+      "loss": 2.5237,
+      "step": 39540
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.357854707579334e-05,
+      "loss": 2.5608,
+      "step": 39550
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.3576361570067315e-05,
+      "loss": 2.5642,
+      "step": 39560
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.3574176064341289e-05,
+      "loss": 2.5318,
+      "step": 39570
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.3571990558615265e-05,
+      "loss": 2.5641,
+      "step": 39580
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.356980505288924e-05,
+      "loss": 2.5702,
+      "step": 39590
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.3567619547163215e-05,
+      "loss": 2.574,
+      "step": 39600
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.356543404143719e-05,
+      "loss": 2.5938,
+      "step": 39610
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.3563248535711165e-05,
+      "loss": 2.5658,
+      "step": 39620
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.356106302998514e-05,
+      "loss": 2.5748,
+      "step": 39630
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.3558877524259115e-05,
+      "loss": 2.5349,
+      "step": 39640
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.355669201853309e-05,
+      "loss": 2.5368,
+      "step": 39650
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.3554506512807063e-05,
+      "loss": 2.5712,
+      "step": 39660
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.355232100708104e-05,
+      "loss": 2.5148,
+      "step": 39670
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.3550135501355015e-05,
+      "loss": 2.5524,
+      "step": 39680
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.3547949995628991e-05,
+      "loss": 2.5985,
+      "step": 39690
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.3545764489902964e-05,
+      "loss": 2.5456,
+      "step": 39700
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.354357898417694e-05,
+      "loss": 2.5466,
+      "step": 39710
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.3541393478450915e-05,
+      "loss": 2.5731,
+      "step": 39720
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.3539207972724891e-05,
+      "loss": 2.5857,
+      "step": 39730
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.3537022466998864e-05,
+      "loss": 2.5808,
+      "step": 39740
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.353483696127284e-05,
+      "loss": 2.5644,
+      "step": 39750
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.3532651455546815e-05,
+      "loss": 2.5716,
+      "step": 39760
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.353046594982079e-05,
+      "loss": 2.5504,
+      "step": 39770
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.3528280444094766e-05,
+      "loss": 2.5527,
+      "step": 39780
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.352609493836874e-05,
+      "loss": 2.5941,
+      "step": 39790
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.3523909432642714e-05,
+      "loss": 2.55,
+      "step": 39800
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.352172392691669e-05,
+      "loss": 2.5534,
+      "step": 39810
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.3519538421190666e-05,
+      "loss": 2.5547,
+      "step": 39820
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.3517352915464638e-05,
+      "loss": 2.57,
+      "step": 39830
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.3515167409738614e-05,
+      "loss": 2.5815,
+      "step": 39840
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.351298190401259e-05,
+      "loss": 2.5635,
+      "step": 39850
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.3510796398286566e-05,
+      "loss": 2.5601,
+      "step": 39860
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.3508610892560538e-05,
+      "loss": 2.5349,
+      "step": 39870
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.3506425386834514e-05,
+      "loss": 2.5803,
+      "step": 39880
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.350423988110849e-05,
+      "loss": 2.5406,
+      "step": 39890
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.3502054375382466e-05,
+      "loss": 2.5634,
+      "step": 39900
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.349986886965644e-05,
+      "loss": 2.6059,
+      "step": 39910
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.3497683363930414e-05,
+      "loss": 2.5432,
+      "step": 39920
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.349549785820439e-05,
+      "loss": 2.5674,
+      "step": 39930
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3493312352478364e-05,
+      "loss": 2.605,
+      "step": 39940
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.349112684675234e-05,
+      "loss": 2.5072,
+      "step": 39950
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3488941341026314e-05,
+      "loss": 2.5404,
+      "step": 39960
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3486755835300289e-05,
+      "loss": 2.5455,
+      "step": 39970
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3484570329574264e-05,
+      "loss": 2.6123,
+      "step": 39980
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.348238482384824e-05,
+      "loss": 2.5431,
+      "step": 39990
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3480199318122213e-05,
+      "loss": 2.5244,
+      "step": 40000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3478013812396189e-05,
+      "loss": 2.602,
+      "step": 40010
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3475828306670165e-05,
+      "loss": 2.5414,
+      "step": 40020
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.347364280094414e-05,
+      "loss": 2.561,
+      "step": 40030
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.3471457295218115e-05,
+      "loss": 2.5707,
+      "step": 40040
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.3469271789492089e-05,
+      "loss": 2.538,
+      "step": 40050
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.3467086283766065e-05,
+      "loss": 2.548,
+      "step": 40060
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.346490077804004e-05,
+      "loss": 2.5497,
+      "step": 40070
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.3462715272314015e-05,
+      "loss": 2.5714,
+      "step": 40080
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.3460529766587989e-05,
+      "loss": 2.5057,
+      "step": 40090
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.3458344260861965e-05,
+      "loss": 2.5338,
+      "step": 40100
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.3456158755135939e-05,
+      "loss": 2.5227,
+      "step": 40110
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.3453973249409915e-05,
+      "loss": 2.6036,
+      "step": 40120
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.345178774368389e-05,
+      "loss": 2.5807,
+      "step": 40130
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.3449602237957863e-05,
+      "loss": 2.5681,
+      "step": 40140
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.344741673223184e-05,
+      "loss": 2.6026,
+      "step": 40150
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.3445231226505815e-05,
+      "loss": 2.521,
+      "step": 40160
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.3443045720779791e-05,
+      "loss": 2.5182,
+      "step": 40170
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.3440860215053763e-05,
+      "loss": 2.518,
+      "step": 40180
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.343867470932774e-05,
+      "loss": 2.5604,
+      "step": 40190
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.3436489203601715e-05,
+      "loss": 2.5197,
+      "step": 40200
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.3434303697875691e-05,
+      "loss": 2.5413,
+      "step": 40210
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.3432118192149664e-05,
+      "loss": 2.5634,
+      "step": 40220
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.342993268642364e-05,
+      "loss": 2.5717,
+      "step": 40230
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.3427747180697615e-05,
+      "loss": 2.5423,
+      "step": 40240
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.342556167497159e-05,
+      "loss": 2.5838,
+      "step": 40250
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.3423376169245565e-05,
+      "loss": 2.6037,
+      "step": 40260
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.342119066351954e-05,
+      "loss": 2.4972,
+      "step": 40270
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.3419005157793514e-05,
+      "loss": 2.5352,
+      "step": 40280
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.341681965206749e-05,
+      "loss": 2.5481,
+      "step": 40290
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.3414634146341466e-05,
+      "loss": 2.5654,
+      "step": 40300
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.3412448640615438e-05,
+      "loss": 2.5087,
+      "step": 40310
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.3410263134889414e-05,
+      "loss": 2.5286,
+      "step": 40320
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.340807762916339e-05,
+      "loss": 2.5972,
+      "step": 40330
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.3405892123437366e-05,
+      "loss": 2.5778,
+      "step": 40340
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.3403706617711338e-05,
+      "loss": 2.5675,
+      "step": 40350
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.3401521111985314e-05,
+      "loss": 2.5723,
+      "step": 40360
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.339933560625929e-05,
+      "loss": 2.5396,
+      "step": 40370
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.3397150100533266e-05,
+      "loss": 2.5423,
+      "step": 40380
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.339496459480724e-05,
+      "loss": 2.5374,
+      "step": 40390
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.3392779089081214e-05,
+      "loss": 2.5438,
+      "step": 40400
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.339059358335519e-05,
+      "loss": 2.5406,
+      "step": 40410
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.3388408077629164e-05,
+      "loss": 2.5385,
+      "step": 40420
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.338622257190314e-05,
+      "loss": 2.5538,
+      "step": 40430
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.3384037066177114e-05,
+      "loss": 2.5829,
+      "step": 40440
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.3381851560451088e-05,
+      "loss": 2.5591,
+      "step": 40450
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.3379666054725064e-05,
+      "loss": 2.5631,
+      "step": 40460
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.337748054899904e-05,
+      "loss": 2.513,
+      "step": 40470
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.3375295043273016e-05,
+      "loss": 2.5719,
+      "step": 40480
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.3373109537546989e-05,
+      "loss": 2.53,
+      "step": 40490
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.3370924031820964e-05,
+      "loss": 2.5631,
+      "step": 40500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.336873852609494e-05,
+      "loss": 2.5091,
+      "step": 40510
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.3366553020368915e-05,
+      "loss": 2.5647,
+      "step": 40520
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.3364367514642889e-05,
+      "loss": 2.5622,
+      "step": 40530
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.3362182008916865e-05,
+      "loss": 2.5645,
+      "step": 40540
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.335999650319084e-05,
+      "loss": 2.573,
+      "step": 40550
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.3357810997464815e-05,
+      "loss": 2.5635,
+      "step": 40560
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.3355625491738789e-05,
+      "loss": 2.593,
+      "step": 40570
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.3353439986012765e-05,
+      "loss": 2.5343,
+      "step": 40580
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.3351254480286739e-05,
+      "loss": 2.5791,
+      "step": 40590
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.3349068974560715e-05,
+      "loss": 2.5331,
+      "step": 40600
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.334688346883469e-05,
+      "loss": 2.555,
+      "step": 40610
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.3344697963108663e-05,
+      "loss": 2.4872,
+      "step": 40620
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.3342512457382639e-05,
+      "loss": 2.5625,
+      "step": 40630
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.3340326951656615e-05,
+      "loss": 2.5583,
+      "step": 40640
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.333814144593059e-05,
+      "loss": 2.5503,
+      "step": 40650
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.3335955940204563e-05,
+      "loss": 2.5542,
+      "step": 40660
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.3333770434478539e-05,
+      "loss": 2.56,
+      "step": 40670
+    },
+    {
+      "epoch": 8.0,
+      "eval_BLEU": 8.8298,
+      "eval_gen_len": 11.5468,
+      "eval_loss": 3.071040630340576,
+      "eval_runtime": 143.1231,
+      "eval_samples_per_second": 75.201,
+      "eval_steps_per_second": 0.3,
+      "step": 40672
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.3331584928752515e-05,
+      "loss": 2.5244,
+      "step": 40680
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.332939942302649e-05,
+      "loss": 2.4291,
+      "step": 40690
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.3327213917300463e-05,
+      "loss": 2.435,
+      "step": 40700
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.332502841157444e-05,
+      "loss": 2.4779,
+      "step": 40710
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.3322842905848415e-05,
+      "loss": 2.4722,
+      "step": 40720
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.332065740012239e-05,
+      "loss": 2.4588,
+      "step": 40730
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.3318471894396365e-05,
+      "loss": 2.4455,
+      "step": 40740
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.331628638867034e-05,
+      "loss": 2.4593,
+      "step": 40750
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.3314100882944314e-05,
+      "loss": 2.4615,
+      "step": 40760
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.331191537721829e-05,
+      "loss": 2.4603,
+      "step": 40770
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.3309729871492265e-05,
+      "loss": 2.4149,
+      "step": 40780
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.3307544365766238e-05,
+      "loss": 2.4883,
+      "step": 40790
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.3305358860040214e-05,
+      "loss": 2.4222,
+      "step": 40800
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.330317335431419e-05,
+      "loss": 2.4734,
+      "step": 40810
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.3300987848588165e-05,
+      "loss": 2.4018,
+      "step": 40820
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.329880234286214e-05,
+      "loss": 2.4397,
+      "step": 40830
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.3296616837136114e-05,
+      "loss": 2.4651,
+      "step": 40840
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.329443133141009e-05,
+      "loss": 2.4452,
+      "step": 40850
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.3292245825684066e-05,
+      "loss": 2.4091,
+      "step": 40860
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.329006031995804e-05,
+      "loss": 2.4396,
+      "step": 40870
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.3287874814232014e-05,
+      "loss": 2.4695,
+      "step": 40880
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.328568930850599e-05,
+      "loss": 2.3993,
+      "step": 40890
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.3283503802779964e-05,
+      "loss": 2.4478,
+      "step": 40900
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.328131829705394e-05,
+      "loss": 2.4653,
+      "step": 40910
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.3279132791327914e-05,
+      "loss": 2.4562,
+      "step": 40920
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.3276947285601888e-05,
+      "loss": 2.4466,
+      "step": 40930
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.3274761779875864e-05,
+      "loss": 2.4938,
+      "step": 40940
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.327257627414984e-05,
+      "loss": 2.4569,
+      "step": 40950
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.3270390768423816e-05,
+      "loss": 2.415,
+      "step": 40960
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.3268205262697788e-05,
+      "loss": 2.4543,
+      "step": 40970
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.3266019756971764e-05,
+      "loss": 2.4564,
+      "step": 40980
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.326383425124574e-05,
+      "loss": 2.4678,
+      "step": 40990
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.3261648745519714e-05,
+      "loss": 2.4728,
+      "step": 41000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.3259463239793689e-05,
+      "loss": 2.3958,
+      "step": 41010
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.3257277734067664e-05,
+      "loss": 2.4905,
+      "step": 41020
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.325509222834164e-05,
+      "loss": 2.4439,
+      "step": 41030
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.3252906722615614e-05,
+      "loss": 2.4523,
+      "step": 41040
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.3250721216889589e-05,
+      "loss": 2.4442,
+      "step": 41050
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.3248535711163565e-05,
+      "loss": 2.4172,
+      "step": 41060
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.3246350205437539e-05,
+      "loss": 2.4424,
+      "step": 41070
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.3244164699711515e-05,
+      "loss": 2.4784,
+      "step": 41080
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.324197919398549e-05,
+      "loss": 2.4481,
+      "step": 41090
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.3239793688259463e-05,
+      "loss": 2.4489,
+      "step": 41100
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.3237608182533439e-05,
+      "loss": 2.4444,
+      "step": 41110
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.3235422676807415e-05,
+      "loss": 2.4742,
+      "step": 41120
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.323323717108139e-05,
+      "loss": 2.4792,
+      "step": 41130
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.3231051665355363e-05,
+      "loss": 2.495,
+      "step": 41140
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.3228866159629339e-05,
+      "loss": 2.4579,
+      "step": 41150
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.3226680653903315e-05,
+      "loss": 2.4575,
+      "step": 41160
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.3224495148177289e-05,
+      "loss": 2.4667,
+      "step": 41170
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.3222309642451265e-05,
+      "loss": 2.4065,
+      "step": 41180
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.3220124136725239e-05,
+      "loss": 2.4286,
+      "step": 41190
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.3217938630999215e-05,
+      "loss": 2.4371,
+      "step": 41200
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.321575312527319e-05,
+      "loss": 2.4464,
+      "step": 41210
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.3213567619547165e-05,
+      "loss": 2.4251,
+      "step": 41220
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.321138211382114e-05,
+      "loss": 2.4124,
+      "step": 41230
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.3209196608095113e-05,
+      "loss": 2.4681,
+      "step": 41240
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.320701110236909e-05,
+      "loss": 2.4376,
+      "step": 41250
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.3204825596643065e-05,
+      "loss": 2.4836,
+      "step": 41260
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.3202640090917038e-05,
+      "loss": 2.4204,
+      "step": 41270
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.3200454585191014e-05,
+      "loss": 2.4716,
+      "step": 41280
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.319826907946499e-05,
+      "loss": 2.4092,
+      "step": 41290
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.3196083573738965e-05,
+      "loss": 2.4242,
+      "step": 41300
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.319389806801294e-05,
+      "loss": 2.4679,
+      "step": 41310
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.3191712562286914e-05,
+      "loss": 2.4279,
+      "step": 41320
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.318952705656089e-05,
+      "loss": 2.4459,
+      "step": 41330
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.3187341550834864e-05,
+      "loss": 2.462,
+      "step": 41340
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.318515604510884e-05,
+      "loss": 2.4338,
+      "step": 41350
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.3182970539382814e-05,
+      "loss": 2.4665,
+      "step": 41360
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.318078503365679e-05,
+      "loss": 2.4804,
+      "step": 41370
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.3178599527930764e-05,
+      "loss": 2.4296,
+      "step": 41380
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.317641402220474e-05,
+      "loss": 2.4249,
+      "step": 41390
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.3174228516478714e-05,
+      "loss": 2.4358,
+      "step": 41400
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 2.4705,
+      "step": 41410
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.3169857505026664e-05,
+      "loss": 2.4674,
+      "step": 41420
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.316767199930064e-05,
+      "loss": 2.4438,
+      "step": 41430
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.3165486493574616e-05,
+      "loss": 2.4576,
+      "step": 41440
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.3163300987848588e-05,
+      "loss": 2.4735,
+      "step": 41450
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.3161115482122564e-05,
+      "loss": 2.4733,
+      "step": 41460
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.315892997639654e-05,
+      "loss": 2.4487,
+      "step": 41470
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.3156744470670514e-05,
+      "loss": 2.4743,
+      "step": 41480
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.3154558964944488e-05,
+      "loss": 2.4562,
+      "step": 41490
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.3152373459218464e-05,
+      "loss": 2.4615,
+      "step": 41500
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.315018795349244e-05,
+      "loss": 2.4085,
+      "step": 41510
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.3148002447766414e-05,
+      "loss": 2.4637,
+      "step": 41520
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.314581694204039e-05,
+      "loss": 2.466,
+      "step": 41530
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.3143631436314364e-05,
+      "loss": 2.4737,
+      "step": 41540
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.3141445930588339e-05,
+      "loss": 2.4161,
+      "step": 41550
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.3139260424862314e-05,
+      "loss": 2.4411,
+      "step": 41560
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.313707491913629e-05,
+      "loss": 2.4513,
+      "step": 41570
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.3134889413410263e-05,
+      "loss": 2.4214,
+      "step": 41580
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.3132703907684239e-05,
+      "loss": 2.513,
+      "step": 41590
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.3130518401958215e-05,
+      "loss": 2.5047,
+      "step": 41600
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.312833289623219e-05,
+      "loss": 2.4692,
+      "step": 41610
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.3126147390506163e-05,
+      "loss": 2.4573,
+      "step": 41620
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.3123961884780139e-05,
+      "loss": 2.4932,
+      "step": 41630
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.3121776379054115e-05,
+      "loss": 2.4827,
+      "step": 41640
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.3119590873328089e-05,
+      "loss": 2.4608,
+      "step": 41650
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.3117405367602065e-05,
+      "loss": 2.4263,
+      "step": 41660
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.3115219861876039e-05,
+      "loss": 2.4498,
+      "step": 41670
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.3113034356150015e-05,
+      "loss": 2.4441,
+      "step": 41680
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.3110848850423989e-05,
+      "loss": 2.4688,
+      "step": 41690
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.3108663344697965e-05,
+      "loss": 2.4443,
+      "step": 41700
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.3106477838971939e-05,
+      "loss": 2.468,
+      "step": 41710
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.3104292333245913e-05,
+      "loss": 2.4601,
+      "step": 41720
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.3102106827519889e-05,
+      "loss": 2.4775,
+      "step": 41730
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.3099921321793865e-05,
+      "loss": 2.4344,
+      "step": 41740
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.3097735816067837e-05,
+      "loss": 2.4516,
+      "step": 41750
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.3095550310341813e-05,
+      "loss": 2.5035,
+      "step": 41760
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.309336480461579e-05,
+      "loss": 2.4589,
+      "step": 41770
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.3091179298889765e-05,
+      "loss": 2.4364,
+      "step": 41780
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.308899379316374e-05,
+      "loss": 2.4597,
+      "step": 41790
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.3086808287437713e-05,
+      "loss": 2.4587,
+      "step": 41800
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.308462278171169e-05,
+      "loss": 2.4758,
+      "step": 41810
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.3082437275985664e-05,
+      "loss": 2.4514,
+      "step": 41820
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.308025177025964e-05,
+      "loss": 2.4572,
+      "step": 41830
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.3078066264533614e-05,
+      "loss": 2.4735,
+      "step": 41840
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.307588075880759e-05,
+      "loss": 2.4865,
+      "step": 41850
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.3073695253081564e-05,
+      "loss": 2.4613,
+      "step": 41860
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.307150974735554e-05,
+      "loss": 2.4588,
+      "step": 41870
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.3069324241629515e-05,
+      "loss": 2.4596,
+      "step": 41880
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.3067138735903488e-05,
+      "loss": 2.5088,
+      "step": 41890
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.3064953230177464e-05,
+      "loss": 2.4274,
+      "step": 41900
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.306276772445144e-05,
+      "loss": 2.49,
+      "step": 41910
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.3060582218725416e-05,
+      "loss": 2.5,
+      "step": 41920
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.3058396712999388e-05,
+      "loss": 2.4349,
+      "step": 41930
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.3056211207273364e-05,
+      "loss": 2.4443,
+      "step": 41940
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.305402570154734e-05,
+      "loss": 2.4665,
+      "step": 41950
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.3051840195821314e-05,
+      "loss": 2.4605,
+      "step": 41960
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.3049654690095288e-05,
+      "loss": 2.4642,
+      "step": 41970
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.3047469184369264e-05,
+      "loss": 2.483,
+      "step": 41980
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.3045283678643238e-05,
+      "loss": 2.4783,
+      "step": 41990
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.3043098172917214e-05,
+      "loss": 2.4875,
+      "step": 42000
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.304091266719119e-05,
+      "loss": 2.4542,
+      "step": 42010
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.3038727161465164e-05,
+      "loss": 2.4658,
+      "step": 42020
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.3036541655739138e-05,
+      "loss": 2.5052,
+      "step": 42030
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.3034356150013114e-05,
+      "loss": 2.4628,
+      "step": 42040
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.303217064428709e-05,
+      "loss": 2.4529,
+      "step": 42050
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.3029985138561063e-05,
+      "loss": 2.486,
+      "step": 42060
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.3027799632835038e-05,
+      "loss": 2.4785,
+      "step": 42070
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.3025614127109014e-05,
+      "loss": 2.5229,
+      "step": 42080
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.302342862138299e-05,
+      "loss": 2.4421,
+      "step": 42090
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.3021243115656964e-05,
+      "loss": 2.4798,
+      "step": 42100
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.3019057609930939e-05,
+      "loss": 2.4981,
+      "step": 42110
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.3016872104204914e-05,
+      "loss": 2.4835,
+      "step": 42120
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.3014686598478889e-05,
+      "loss": 2.4596,
+      "step": 42130
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.3012501092752865e-05,
+      "loss": 2.4698,
+      "step": 42140
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.3010315587026839e-05,
+      "loss": 2.4716,
+      "step": 42150
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 2.4887,
+      "step": 42160
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.3005944575574789e-05,
+      "loss": 2.4935,
+      "step": 42170
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.3003759069848765e-05,
+      "loss": 2.4669,
+      "step": 42180
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.3001573564122739e-05,
+      "loss": 2.4605,
+      "step": 42190
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.2999388058396713e-05,
+      "loss": 2.4757,
+      "step": 42200
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.2997202552670689e-05,
+      "loss": 2.489,
+      "step": 42210
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.2995017046944665e-05,
+      "loss": 2.4648,
+      "step": 42220
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.299283154121864e-05,
+      "loss": 2.4713,
+      "step": 42230
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.2990646035492613e-05,
+      "loss": 2.513,
+      "step": 42240
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.2988460529766589e-05,
+      "loss": 2.4681,
+      "step": 42250
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.2986275024040565e-05,
+      "loss": 2.4589,
+      "step": 42260
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.2984089518314539e-05,
+      "loss": 2.4535,
+      "step": 42270
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.2981904012588513e-05,
+      "loss": 2.461,
+      "step": 42280
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.297971850686249e-05,
+      "loss": 2.4417,
+      "step": 42290
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.2977533001136463e-05,
+      "loss": 2.5008,
+      "step": 42300
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.297534749541044e-05,
+      "loss": 2.4894,
+      "step": 42310
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.2973161989684413e-05,
+      "loss": 2.5118,
+      "step": 42320
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.297097648395839e-05,
+      "loss": 2.4908,
+      "step": 42330
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.2968790978232363e-05,
+      "loss": 2.4762,
+      "step": 42340
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.296660547250634e-05,
+      "loss": 2.4928,
+      "step": 42350
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.2964419966780315e-05,
+      "loss": 2.4593,
+      "step": 42360
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.2962234461054288e-05,
+      "loss": 2.4596,
+      "step": 42370
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.2960048955328264e-05,
+      "loss": 2.4796,
+      "step": 42380
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.295786344960224e-05,
+      "loss": 2.4855,
+      "step": 42390
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.2955677943876215e-05,
+      "loss": 2.4708,
+      "step": 42400
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.2953492438150188e-05,
+      "loss": 2.5057,
+      "step": 42410
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.2951306932424164e-05,
+      "loss": 2.4662,
+      "step": 42420
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.294912142669814e-05,
+      "loss": 2.4686,
+      "step": 42430
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.2946935920972114e-05,
+      "loss": 2.4572,
+      "step": 42440
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.294475041524609e-05,
+      "loss": 2.5164,
+      "step": 42450
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.2942564909520064e-05,
+      "loss": 2.464,
+      "step": 42460
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.2940379403794038e-05,
+      "loss": 2.4656,
+      "step": 42470
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.2938193898068014e-05,
+      "loss": 2.4616,
+      "step": 42480
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.293600839234199e-05,
+      "loss": 2.5035,
+      "step": 42490
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.2933822886615964e-05,
+      "loss": 2.5017,
+      "step": 42500
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.2931637380889938e-05,
+      "loss": 2.4962,
+      "step": 42510
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.2929451875163914e-05,
+      "loss": 2.488,
+      "step": 42520
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.292726636943789e-05,
+      "loss": 2.5274,
+      "step": 42530
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.2925080863711862e-05,
+      "loss": 2.4817,
+      "step": 42540
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.2922895357985838e-05,
+      "loss": 2.4397,
+      "step": 42550
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.2920709852259814e-05,
+      "loss": 2.5199,
+      "step": 42560
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.291852434653379e-05,
+      "loss": 2.5012,
+      "step": 42570
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.2916338840807764e-05,
+      "loss": 2.4903,
+      "step": 42580
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.2914153335081738e-05,
+      "loss": 2.4921,
+      "step": 42590
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.2911967829355714e-05,
+      "loss": 2.4633,
+      "step": 42600
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.2909782323629689e-05,
+      "loss": 2.4523,
+      "step": 42610
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.2907596817903664e-05,
+      "loss": 2.4761,
+      "step": 42620
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.2905411312177639e-05,
+      "loss": 2.4658,
+      "step": 42630
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 2.4899,
+      "step": 42640
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.2901040300725589e-05,
+      "loss": 2.4277,
+      "step": 42650
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.2898854794999565e-05,
+      "loss": 2.4554,
+      "step": 42660
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.2896669289273539e-05,
+      "loss": 2.4784,
+      "step": 42670
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.2894483783547513e-05,
+      "loss": 2.4274,
+      "step": 42680
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.2892298277821489e-05,
+      "loss": 2.4795,
+      "step": 42690
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.2890112772095465e-05,
+      "loss": 2.4862,
+      "step": 42700
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.288792726636944e-05,
+      "loss": 2.5051,
+      "step": 42710
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.2885741760643413e-05,
+      "loss": 2.4733,
+      "step": 42720
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.2883556254917389e-05,
+      "loss": 2.4607,
+      "step": 42730
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.2881370749191365e-05,
+      "loss": 2.5016,
+      "step": 42740
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.2879185243465339e-05,
+      "loss": 2.4369,
+      "step": 42750
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.2876999737739313e-05,
+      "loss": 2.4667,
+      "step": 42760
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.2874814232013289e-05,
+      "loss": 2.467,
+      "step": 42770
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.2872628726287263e-05,
+      "loss": 2.4596,
+      "step": 42780
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.2870443220561239e-05,
+      "loss": 2.5095,
+      "step": 42790
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.2868257714835215e-05,
+      "loss": 2.4801,
+      "step": 42800
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.2866072209109189e-05,
+      "loss": 2.4619,
+      "step": 42810
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.2863886703383163e-05,
+      "loss": 2.5099,
+      "step": 42820
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.286170119765714e-05,
+      "loss": 2.4839,
+      "step": 42830
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.2859515691931115e-05,
+      "loss": 2.5083,
+      "step": 42840
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.2857330186205088e-05,
+      "loss": 2.4919,
+      "step": 42850
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.2855144680479063e-05,
+      "loss": 2.4888,
+      "step": 42860
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.285295917475304e-05,
+      "loss": 2.5096,
+      "step": 42870
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.2850773669027015e-05,
+      "loss": 2.4886,
+      "step": 42880
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.2848588163300988e-05,
+      "loss": 2.4592,
+      "step": 42890
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.2846402657574964e-05,
+      "loss": 2.4751,
+      "step": 42900
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.284421715184894e-05,
+      "loss": 2.5024,
+      "step": 42910
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.2842031646122914e-05,
+      "loss": 2.4556,
+      "step": 42920
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.283984614039689e-05,
+      "loss": 2.5063,
+      "step": 42930
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.2837660634670864e-05,
+      "loss": 2.4767,
+      "step": 42940
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.2835475128944838e-05,
+      "loss": 2.4587,
+      "step": 42950
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.2833289623218814e-05,
+      "loss": 2.4666,
+      "step": 42960
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.283110411749279e-05,
+      "loss": 2.4934,
+      "step": 42970
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.2828918611766764e-05,
+      "loss": 2.4533,
+      "step": 42980
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.2826733106040738e-05,
+      "loss": 2.5107,
+      "step": 42990
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.2824547600314714e-05,
+      "loss": 2.4615,
+      "step": 43000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.282236209458869e-05,
+      "loss": 2.4875,
+      "step": 43010
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.2820176588862662e-05,
+      "loss": 2.4925,
+      "step": 43020
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.2817991083136638e-05,
+      "loss": 2.4936,
+      "step": 43030
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.2815805577410614e-05,
+      "loss": 2.5148,
+      "step": 43040
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.281362007168459e-05,
+      "loss": 2.4711,
+      "step": 43050
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.2811434565958564e-05,
+      "loss": 2.481,
+      "step": 43060
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.2809249060232538e-05,
+      "loss": 2.4505,
+      "step": 43070
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.2807063554506514e-05,
+      "loss": 2.46,
+      "step": 43080
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.2804878048780488e-05,
+      "loss": 2.5176,
+      "step": 43090
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.2802692543054464e-05,
+      "loss": 2.5343,
+      "step": 43100
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.2800507037328438e-05,
+      "loss": 2.4633,
+      "step": 43110
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.2798321531602413e-05,
+      "loss": 2.4615,
+      "step": 43120
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.2796136025876388e-05,
+      "loss": 2.4823,
+      "step": 43130
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.2793950520150364e-05,
+      "loss": 2.4773,
+      "step": 43140
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.279176501442434e-05,
+      "loss": 2.4794,
+      "step": 43150
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.2789579508698313e-05,
+      "loss": 2.4529,
+      "step": 43160
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.2787394002972289e-05,
+      "loss": 2.4958,
+      "step": 43170
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.2785208497246264e-05,
+      "loss": 2.478,
+      "step": 43180
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.278302299152024e-05,
+      "loss": 2.4833,
+      "step": 43190
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.2780837485794213e-05,
+      "loss": 2.5014,
+      "step": 43200
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.2778651980068189e-05,
+      "loss": 2.465,
+      "step": 43210
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.2776466474342165e-05,
+      "loss": 2.4791,
+      "step": 43220
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.2774280968616139e-05,
+      "loss": 2.5181,
+      "step": 43230
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.2772095462890113e-05,
+      "loss": 2.5013,
+      "step": 43240
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.2769909957164089e-05,
+      "loss": 2.502,
+      "step": 43250
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.2767724451438063e-05,
+      "loss": 2.5019,
+      "step": 43260
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.2765538945712039e-05,
+      "loss": 2.4651,
+      "step": 43270
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.2763353439986015e-05,
+      "loss": 2.5098,
+      "step": 43280
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.2761167934259987e-05,
+      "loss": 2.5008,
+      "step": 43290
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.2758982428533963e-05,
+      "loss": 2.5145,
+      "step": 43300
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.2756796922807939e-05,
+      "loss": 2.4384,
+      "step": 43310
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.2754611417081915e-05,
+      "loss": 2.4382,
+      "step": 43320
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.2752425911355887e-05,
+      "loss": 2.4732,
+      "step": 43330
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.2750240405629863e-05,
+      "loss": 2.4708,
+      "step": 43340
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.274805489990384e-05,
+      "loss": 2.4476,
+      "step": 43350
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.2745869394177815e-05,
+      "loss": 2.4803,
+      "step": 43360
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.2743683888451788e-05,
+      "loss": 2.5274,
+      "step": 43370
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.2741498382725763e-05,
+      "loss": 2.5144,
+      "step": 43380
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.273931287699974e-05,
+      "loss": 2.4613,
+      "step": 43390
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.2737127371273713e-05,
+      "loss": 2.4746,
+      "step": 43400
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.273494186554769e-05,
+      "loss": 2.5141,
+      "step": 43410
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.2732756359821664e-05,
+      "loss": 2.4418,
+      "step": 43420
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.2730570854095638e-05,
+      "loss": 2.4895,
+      "step": 43430
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.2728385348369614e-05,
+      "loss": 2.492,
+      "step": 43440
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.272619984264359e-05,
+      "loss": 2.4551,
+      "step": 43450
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.2724014336917564e-05,
+      "loss": 2.52,
+      "step": 43460
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.2721828831191538e-05,
+      "loss": 2.5009,
+      "step": 43470
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.2719643325465514e-05,
+      "loss": 2.4474,
+      "step": 43480
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.271745781973949e-05,
+      "loss": 2.49,
+      "step": 43490
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.2715272314013465e-05,
+      "loss": 2.4618,
+      "step": 43500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.2713086808287438e-05,
+      "loss": 2.4851,
+      "step": 43510
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.2710901302561414e-05,
+      "loss": 2.474,
+      "step": 43520
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.270871579683539e-05,
+      "loss": 2.4843,
+      "step": 43530
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.2706530291109364e-05,
+      "loss": 2.4568,
+      "step": 43540
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.2704344785383338e-05,
+      "loss": 2.4925,
+      "step": 43550
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.2702159279657314e-05,
+      "loss": 2.5166,
+      "step": 43560
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.2699973773931288e-05,
+      "loss": 2.521,
+      "step": 43570
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.2697788268205264e-05,
+      "loss": 2.4662,
+      "step": 43580
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.2695602762479238e-05,
+      "loss": 2.4506,
+      "step": 43590
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.2693417256753212e-05,
+      "loss": 2.495,
+      "step": 43600
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.2691231751027188e-05,
+      "loss": 2.4718,
+      "step": 43610
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.2689046245301164e-05,
+      "loss": 2.4678,
+      "step": 43620
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.268686073957514e-05,
+      "loss": 2.4706,
+      "step": 43630
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.2684675233849113e-05,
+      "loss": 2.4897,
+      "step": 43640
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.2682489728123088e-05,
+      "loss": 2.4546,
+      "step": 43650
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.2680304222397064e-05,
+      "loss": 2.4723,
+      "step": 43660
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.267811871667104e-05,
+      "loss": 2.4834,
+      "step": 43670
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.2675933210945013e-05,
+      "loss": 2.5042,
+      "step": 43680
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.2673747705218989e-05,
+      "loss": 2.4689,
+      "step": 43690
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.2671562199492964e-05,
+      "loss": 2.4732,
+      "step": 43700
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.2669376693766939e-05,
+      "loss": 2.5022,
+      "step": 43710
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.2667191188040913e-05,
+      "loss": 2.4948,
+      "step": 43720
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.2665005682314889e-05,
+      "loss": 2.4831,
+      "step": 43730
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.2662820176588863e-05,
+      "loss": 2.4652,
+      "step": 43740
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.2660634670862839e-05,
+      "loss": 2.5193,
+      "step": 43750
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.2658449165136815e-05,
+      "loss": 2.4437,
+      "step": 43760
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.2656263659410787e-05,
+      "loss": 2.47,
+      "step": 43770
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.2654078153684763e-05,
+      "loss": 2.5198,
+      "step": 43780
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.2651892647958739e-05,
+      "loss": 2.5158,
+      "step": 43790
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.2649707142232715e-05,
+      "loss": 2.4674,
+      "step": 43800
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.2647521636506687e-05,
+      "loss": 2.4932,
+      "step": 43810
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.2645336130780663e-05,
+      "loss": 2.493,
+      "step": 43820
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.2643150625054639e-05,
+      "loss": 2.4713,
+      "step": 43830
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.2640965119328615e-05,
+      "loss": 2.4783,
+      "step": 43840
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.2638779613602589e-05,
+      "loss": 2.5396,
+      "step": 43850
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.2636594107876563e-05,
+      "loss": 2.5044,
+      "step": 43860
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.2634408602150539e-05,
+      "loss": 2.4672,
+      "step": 43870
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.2632223096424513e-05,
+      "loss": 2.5103,
+      "step": 43880
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.263003759069849e-05,
+      "loss": 2.4657,
+      "step": 43890
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.2627852084972463e-05,
+      "loss": 2.4935,
+      "step": 43900
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.2625666579246438e-05,
+      "loss": 2.5235,
+      "step": 43910
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.2623481073520413e-05,
+      "loss": 2.459,
+      "step": 43920
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.262129556779439e-05,
+      "loss": 2.5261,
+      "step": 43930
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.2619110062068363e-05,
+      "loss": 2.4929,
+      "step": 43940
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.2616924556342338e-05,
+      "loss": 2.4563,
+      "step": 43950
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.2614739050616314e-05,
+      "loss": 2.5099,
+      "step": 43960
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.261255354489029e-05,
+      "loss": 2.4954,
+      "step": 43970
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.2610368039164265e-05,
+      "loss": 2.5145,
+      "step": 43980
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.2608182533438238e-05,
+      "loss": 2.4634,
+      "step": 43990
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.2605997027712214e-05,
+      "loss": 2.4691,
+      "step": 44000
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.260381152198619e-05,
+      "loss": 2.5317,
+      "step": 44010
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.2601626016260164e-05,
+      "loss": 2.4624,
+      "step": 44020
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.2599440510534138e-05,
+      "loss": 2.4976,
+      "step": 44030
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.2597255004808114e-05,
+      "loss": 2.5295,
+      "step": 44040
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.2595069499082088e-05,
+      "loss": 2.4842,
+      "step": 44050
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.2592883993356064e-05,
+      "loss": 2.5402,
+      "step": 44060
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.2590698487630038e-05,
+      "loss": 2.4658,
+      "step": 44070
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.2588512981904012e-05,
+      "loss": 2.4895,
+      "step": 44080
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.2586327476177988e-05,
+      "loss": 2.5062,
+      "step": 44090
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.2584141970451964e-05,
+      "loss": 2.4952,
+      "step": 44100
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.258195646472594e-05,
+      "loss": 2.4878,
+      "step": 44110
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.2579770958999912e-05,
+      "loss": 2.4536,
+      "step": 44120
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.2577585453273888e-05,
+      "loss": 2.5087,
+      "step": 44130
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.2575399947547864e-05,
+      "loss": 2.488,
+      "step": 44140
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.257321444182184e-05,
+      "loss": 2.4891,
+      "step": 44150
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.2571028936095812e-05,
+      "loss": 2.4994,
+      "step": 44160
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.2568843430369788e-05,
+      "loss": 2.4864,
+      "step": 44170
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.2566657924643764e-05,
+      "loss": 2.4768,
+      "step": 44180
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.2564472418917738e-05,
+      "loss": 2.512,
+      "step": 44190
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.2562286913191714e-05,
+      "loss": 2.4902,
+      "step": 44200
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.2560101407465688e-05,
+      "loss": 2.4681,
+      "step": 44210
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.2557915901739663e-05,
+      "loss": 2.4949,
+      "step": 44220
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.2555730396013639e-05,
+      "loss": 2.5698,
+      "step": 44230
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.2553544890287614e-05,
+      "loss": 2.5075,
+      "step": 44240
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.2551359384561587e-05,
+      "loss": 2.4799,
+      "step": 44250
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2549173878835563e-05,
+      "loss": 2.5206,
+      "step": 44260
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2546988373109539e-05,
+      "loss": 2.5049,
+      "step": 44270
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2544802867383515e-05,
+      "loss": 2.544,
+      "step": 44280
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2542617361657487e-05,
+      "loss": 2.4547,
+      "step": 44290
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2540431855931463e-05,
+      "loss": 2.4887,
+      "step": 44300
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.2538246350205439e-05,
+      "loss": 2.4501,
+      "step": 44310
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.2536060844479415e-05,
+      "loss": 2.5024,
+      "step": 44320
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.2533875338753389e-05,
+      "loss": 2.5269,
+      "step": 44330
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.2531689833027363e-05,
+      "loss": 2.4999,
+      "step": 44340
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.2529504327301339e-05,
+      "loss": 2.4928,
+      "step": 44350
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.2527318821575313e-05,
+      "loss": 2.4995,
+      "step": 44360
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.2525133315849289e-05,
+      "loss": 2.4683,
+      "step": 44370
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.2522947810123263e-05,
+      "loss": 2.5345,
+      "step": 44380
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.2520762304397237e-05,
+      "loss": 2.4679,
+      "step": 44390
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.2518576798671213e-05,
+      "loss": 2.4993,
+      "step": 44400
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.2516391292945189e-05,
+      "loss": 2.4938,
+      "step": 44410
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.2514205787219162e-05,
+      "loss": 2.4963,
+      "step": 44420
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.2512020281493138e-05,
+      "loss": 2.4603,
+      "step": 44430
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.2509834775767113e-05,
+      "loss": 2.4888,
+      "step": 44440
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.250764927004109e-05,
+      "loss": 2.4955,
+      "step": 44450
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.2505463764315065e-05,
+      "loss": 2.5324,
+      "step": 44460
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.2503278258589038e-05,
+      "loss": 2.5092,
+      "step": 44470
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.2501092752863014e-05,
+      "loss": 2.4921,
+      "step": 44480
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.249890724713699e-05,
+      "loss": 2.4876,
+      "step": 44490
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.2496721741410964e-05,
+      "loss": 2.4974,
+      "step": 44500
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.2494536235684938e-05,
+      "loss": 2.4906,
+      "step": 44510
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.2492350729958914e-05,
+      "loss": 2.5267,
+      "step": 44520
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.2490165224232888e-05,
+      "loss": 2.4607,
+      "step": 44530
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.2487979718506864e-05,
+      "loss": 2.4915,
+      "step": 44540
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.248579421278084e-05,
+      "loss": 2.5057,
+      "step": 44550
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.2483608707054812e-05,
+      "loss": 2.4876,
+      "step": 44560
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.2481423201328788e-05,
+      "loss": 2.5262,
+      "step": 44570
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.2479237695602764e-05,
+      "loss": 2.5113,
+      "step": 44580
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.247705218987674e-05,
+      "loss": 2.5457,
+      "step": 44590
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.2474866684150712e-05,
+      "loss": 2.4836,
+      "step": 44600
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.2472681178424688e-05,
+      "loss": 2.4903,
+      "step": 44610
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.2470495672698664e-05,
+      "loss": 2.4604,
+      "step": 44620
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.246831016697264e-05,
+      "loss": 2.5147,
+      "step": 44630
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.2466124661246612e-05,
+      "loss": 2.4942,
+      "step": 44640
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.2463939155520588e-05,
+      "loss": 2.5242,
+      "step": 44650
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.2461753649794564e-05,
+      "loss": 2.5184,
+      "step": 44660
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.2459568144068538e-05,
+      "loss": 2.4869,
+      "step": 44670
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.2457382638342514e-05,
+      "loss": 2.4754,
+      "step": 44680
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.2455197132616488e-05,
+      "loss": 2.5062,
+      "step": 44690
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.2453011626890463e-05,
+      "loss": 2.5116,
+      "step": 44700
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.2450826121164438e-05,
+      "loss": 2.5137,
+      "step": 44710
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.2448640615438414e-05,
+      "loss": 2.4434,
+      "step": 44720
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.2446455109712387e-05,
+      "loss": 2.5027,
+      "step": 44730
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.2444269603986363e-05,
+      "loss": 2.4524,
+      "step": 44740
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.2442084098260339e-05,
+      "loss": 2.46,
+      "step": 44750
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.2439898592534314e-05,
+      "loss": 2.4893,
+      "step": 44760
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.2437713086808287e-05,
+      "loss": 2.483,
+      "step": 44770
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.2435527581082263e-05,
+      "loss": 2.4976,
+      "step": 44780
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.2433342075356239e-05,
+      "loss": 2.4985,
+      "step": 44790
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.2431156569630215e-05,
+      "loss": 2.5001,
+      "step": 44800
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.2428971063904189e-05,
+      "loss": 2.5147,
+      "step": 44810
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.2426785558178163e-05,
+      "loss": 2.4987,
+      "step": 44820
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.2424600052452139e-05,
+      "loss": 2.5484,
+      "step": 44830
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.2422414546726113e-05,
+      "loss": 2.46,
+      "step": 44840
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.2420229041000089e-05,
+      "loss": 2.457,
+      "step": 44850
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.2418043535274063e-05,
+      "loss": 2.4982,
+      "step": 44860
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.2415858029548037e-05,
+      "loss": 2.5081,
+      "step": 44870
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.2413672523822013e-05,
+      "loss": 2.4966,
+      "step": 44880
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.2411487018095989e-05,
+      "loss": 2.4847,
+      "step": 44890
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.2409301512369965e-05,
+      "loss": 2.504,
+      "step": 44900
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.2407116006643937e-05,
+      "loss": 2.509,
+      "step": 44910
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.2404930500917913e-05,
+      "loss": 2.5259,
+      "step": 44920
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.2402744995191889e-05,
+      "loss": 2.5532,
+      "step": 44930
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.2400559489465865e-05,
+      "loss": 2.5279,
+      "step": 44940
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.2398373983739837e-05,
+      "loss": 2.5098,
+      "step": 44950
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.2396188478013813e-05,
+      "loss": 2.489,
+      "step": 44960
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.239400297228779e-05,
+      "loss": 2.4913,
+      "step": 44970
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.2391817466561763e-05,
+      "loss": 2.5,
+      "step": 44980
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.2389631960835738e-05,
+      "loss": 2.5299,
+      "step": 44990
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.2387446455109713e-05,
+      "loss": 2.4872,
+      "step": 45000
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.2385260949383688e-05,
+      "loss": 2.5128,
+      "step": 45010
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.2383075443657664e-05,
+      "loss": 2.5207,
+      "step": 45020
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.238088993793164e-05,
+      "loss": 2.4855,
+      "step": 45030
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.2378704432205612e-05,
+      "loss": 2.4896,
+      "step": 45040
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.2376518926479588e-05,
+      "loss": 2.5336,
+      "step": 45050
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.2374333420753564e-05,
+      "loss": 2.5071,
+      "step": 45060
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.237214791502754e-05,
+      "loss": 2.4995,
+      "step": 45070
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.2369962409301512e-05,
+      "loss": 2.4752,
+      "step": 45080
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.2367776903575488e-05,
+      "loss": 2.5381,
+      "step": 45090
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.2365591397849464e-05,
+      "loss": 2.4732,
+      "step": 45100
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.236340589212344e-05,
+      "loss": 2.5376,
+      "step": 45110
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.2361220386397412e-05,
+      "loss": 2.48,
+      "step": 45120
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.2359034880671388e-05,
+      "loss": 2.5188,
+      "step": 45130
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.2356849374945364e-05,
+      "loss": 2.502,
+      "step": 45140
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.2354663869219338e-05,
+      "loss": 2.4991,
+      "step": 45150
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.2352478363493314e-05,
+      "loss": 2.4888,
+      "step": 45160
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.2350292857767288e-05,
+      "loss": 2.4828,
+      "step": 45170
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.2348107352041262e-05,
+      "loss": 2.4843,
+      "step": 45180
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.2345921846315238e-05,
+      "loss": 2.5373,
+      "step": 45190
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.2343736340589214e-05,
+      "loss": 2.524,
+      "step": 45200
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.2341550834863187e-05,
+      "loss": 2.4565,
+      "step": 45210
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.2339365329137162e-05,
+      "loss": 2.4803,
+      "step": 45220
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.2337179823411138e-05,
+      "loss": 2.4984,
+      "step": 45230
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.2334994317685114e-05,
+      "loss": 2.4885,
+      "step": 45240
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.233280881195909e-05,
+      "loss": 2.4889,
+      "step": 45250
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.2330623306233063e-05,
+      "loss": 2.4989,
+      "step": 45260
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.2328437800507038e-05,
+      "loss": 2.4863,
+      "step": 45270
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.2326252294781014e-05,
+      "loss": 2.4971,
+      "step": 45280
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.2324066789054989e-05,
+      "loss": 2.5027,
+      "step": 45290
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.2321881283328963e-05,
+      "loss": 2.4951,
+      "step": 45300
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.2319695777602939e-05,
+      "loss": 2.5107,
+      "step": 45310
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.2317510271876913e-05,
+      "loss": 2.5119,
+      "step": 45320
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.2315324766150889e-05,
+      "loss": 2.4963,
+      "step": 45330
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.2313139260424863e-05,
+      "loss": 2.5043,
+      "step": 45340
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.2310953754698837e-05,
+      "loss": 2.535,
+      "step": 45350
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.2308768248972813e-05,
+      "loss": 2.4443,
+      "step": 45360
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.2306582743246789e-05,
+      "loss": 2.4835,
+      "step": 45370
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.2304397237520765e-05,
+      "loss": 2.501,
+      "step": 45380
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.2302211731794737e-05,
+      "loss": 2.4823,
+      "step": 45390
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.2300026226068713e-05,
+      "loss": 2.4683,
+      "step": 45400
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.2297840720342689e-05,
+      "loss": 2.4925,
+      "step": 45410
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.2295655214616665e-05,
+      "loss": 2.5152,
+      "step": 45420
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.2293469708890637e-05,
+      "loss": 2.4869,
+      "step": 45430
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.2291284203164613e-05,
+      "loss": 2.4825,
+      "step": 45440
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.2289098697438589e-05,
+      "loss": 2.5151,
+      "step": 45450
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.2286913191712563e-05,
+      "loss": 2.5023,
+      "step": 45460
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.2284727685986537e-05,
+      "loss": 2.4894,
+      "step": 45470
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.2282542180260513e-05,
+      "loss": 2.5197,
+      "step": 45480
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.2280356674534487e-05,
+      "loss": 2.4959,
+      "step": 45490
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.2278171168808463e-05,
+      "loss": 2.5122,
+      "step": 45500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.227598566308244e-05,
+      "loss": 2.5289,
+      "step": 45510
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.2273800157356412e-05,
+      "loss": 2.5064,
+      "step": 45520
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.2271614651630388e-05,
+      "loss": 2.4883,
+      "step": 45530
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.2269429145904363e-05,
+      "loss": 2.5001,
+      "step": 45540
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.226724364017834e-05,
+      "loss": 2.5137,
+      "step": 45550
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.2265058134452312e-05,
+      "loss": 2.4972,
+      "step": 45560
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.2262872628726288e-05,
+      "loss": 2.498,
+      "step": 45570
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.2260687123000264e-05,
+      "loss": 2.4865,
+      "step": 45580
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.225850161727424e-05,
+      "loss": 2.528,
+      "step": 45590
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.2256316111548214e-05,
+      "loss": 2.5004,
+      "step": 45600
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.2254130605822188e-05,
+      "loss": 2.4404,
+      "step": 45610
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.2251945100096164e-05,
+      "loss": 2.5205,
+      "step": 45620
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.2249759594370138e-05,
+      "loss": 2.5167,
+      "step": 45630
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.2247574088644114e-05,
+      "loss": 2.4769,
+      "step": 45640
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.2245388582918088e-05,
+      "loss": 2.4993,
+      "step": 45650
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.2243203077192062e-05,
+      "loss": 2.5089,
+      "step": 45660
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.2241017571466038e-05,
+      "loss": 2.4514,
+      "step": 45670
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.2238832065740014e-05,
+      "loss": 2.5046,
+      "step": 45680
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.2236646560013986e-05,
+      "loss": 2.513,
+      "step": 45690
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.2234461054287962e-05,
+      "loss": 2.5151,
+      "step": 45700
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.2232275548561938e-05,
+      "loss": 2.51,
+      "step": 45710
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.2230090042835914e-05,
+      "loss": 2.5319,
+      "step": 45720
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.222790453710989e-05,
+      "loss": 2.4828,
+      "step": 45730
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2225719031383862e-05,
+      "loss": 2.5011,
+      "step": 45740
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2223533525657838e-05,
+      "loss": 2.4846,
+      "step": 45750
+    },
+    {
+      "epoch": 9.0,
+      "eval_BLEU": 9.9118,
+      "eval_gen_len": 12.7213,
+      "eval_loss": 3.0921640396118164,
+      "eval_runtime": 175.415,
+      "eval_samples_per_second": 61.357,
+      "eval_steps_per_second": 0.245,
+      "step": 45756
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2221348019931814e-05,
+      "loss": 2.4813,
+      "step": 45760
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2219162514205788e-05,
+      "loss": 2.3777,
+      "step": 45770
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.2216977008479763e-05,
+      "loss": 2.4228,
+      "step": 45780
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.2214791502753738e-05,
+      "loss": 2.3717,
+      "step": 45790
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.2212605997027713e-05,
+      "loss": 2.4298,
+      "step": 45800
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.2210420491301688e-05,
+      "loss": 2.4072,
+      "step": 45810
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.2208234985575663e-05,
+      "loss": 2.3738,
+      "step": 45820
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.2206049479849637e-05,
+      "loss": 2.3733,
+      "step": 45830
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.2203863974123613e-05,
+      "loss": 2.3693,
+      "step": 45840
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.2201678468397589e-05,
+      "loss": 2.3663,
+      "step": 45850
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.2199492962671564e-05,
+      "loss": 2.4098,
+      "step": 45860
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.2197307456945537e-05,
+      "loss": 2.3959,
+      "step": 45870
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 2.3636,
+      "step": 45880
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.2192936445493489e-05,
+      "loss": 2.4149,
+      "step": 45890
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.2190750939767465e-05,
+      "loss": 2.3642,
+      "step": 45900
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.2188565434041437e-05,
+      "loss": 2.4003,
+      "step": 45910
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.2186379928315413e-05,
+      "loss": 2.4006,
+      "step": 45920
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.2184194422589389e-05,
+      "loss": 2.3737,
+      "step": 45930
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.2182008916863363e-05,
+      "loss": 2.3734,
+      "step": 45940
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.2179823411137339e-05,
+      "loss": 2.3278,
+      "step": 45950
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.2177637905411313e-05,
+      "loss": 2.3789,
+      "step": 45960
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.2175452399685287e-05,
+      "loss": 2.4195,
+      "step": 45970
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.2173266893959263e-05,
+      "loss": 2.3596,
+      "step": 45980
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.2171081388233239e-05,
+      "loss": 2.3837,
+      "step": 45990
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.2168895882507212e-05,
+      "loss": 2.396,
+      "step": 46000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.2166710376781187e-05,
+      "loss": 2.3924,
+      "step": 46010
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.2164524871055163e-05,
+      "loss": 2.4112,
+      "step": 46020
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.216233936532914e-05,
+      "loss": 2.381,
+      "step": 46030
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.2160153859603112e-05,
+      "loss": 2.3495,
+      "step": 46040
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.2157968353877088e-05,
+      "loss": 2.3992,
+      "step": 46050
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.2155782848151063e-05,
+      "loss": 2.3609,
+      "step": 46060
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.215359734242504e-05,
+      "loss": 2.3895,
+      "step": 46070
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.2151411836699013e-05,
+      "loss": 2.3916,
+      "step": 46080
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.2149226330972988e-05,
+      "loss": 2.3315,
+      "step": 46090
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.2147040825246964e-05,
+      "loss": 2.3883,
+      "step": 46100
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.2144855319520938e-05,
+      "loss": 2.3737,
+      "step": 46110
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.2142669813794914e-05,
+      "loss": 2.3947,
+      "step": 46120
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.2140484308068888e-05,
+      "loss": 2.3788,
+      "step": 46130
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.2138298802342862e-05,
+      "loss": 2.4287,
+      "step": 46140
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.2136113296616838e-05,
+      "loss": 2.3776,
+      "step": 46150
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.2133927790890814e-05,
+      "loss": 2.4274,
+      "step": 46160
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.2131742285164786e-05,
+      "loss": 2.4036,
+      "step": 46170
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.2129556779438762e-05,
+      "loss": 2.3717,
+      "step": 46180
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.2127371273712738e-05,
+      "loss": 2.3409,
+      "step": 46190
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.2125185767986714e-05,
+      "loss": 2.3705,
+      "step": 46200
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.212300026226069e-05,
+      "loss": 2.3506,
+      "step": 46210
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.2120814756534662e-05,
+      "loss": 2.3945,
+      "step": 46220
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.2118629250808638e-05,
+      "loss": 2.4363,
+      "step": 46230
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.2116443745082614e-05,
+      "loss": 2.3863,
+      "step": 46240
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.2114258239356588e-05,
+      "loss": 2.3745,
+      "step": 46250
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.2112072733630562e-05,
+      "loss": 2.4071,
+      "step": 46260
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.2109887227904538e-05,
+      "loss": 2.3894,
+      "step": 46270
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.2107701722178512e-05,
+      "loss": 2.3928,
+      "step": 46280
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.2105516216452488e-05,
+      "loss": 2.4355,
+      "step": 46290
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.2103330710726464e-05,
+      "loss": 2.4027,
+      "step": 46300
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.2101145205000437e-05,
+      "loss": 2.3941,
+      "step": 46310
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.2098959699274413e-05,
+      "loss": 2.3724,
+      "step": 46320
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.2096774193548388e-05,
+      "loss": 2.405,
+      "step": 46330
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.2094588687822364e-05,
+      "loss": 2.3813,
+      "step": 46340
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.2092403182096337e-05,
+      "loss": 2.3757,
+      "step": 46350
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.2090217676370313e-05,
+      "loss": 2.3767,
+      "step": 46360
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.2088032170644289e-05,
+      "loss": 2.3797,
+      "step": 46370
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.2085846664918264e-05,
+      "loss": 2.3868,
+      "step": 46380
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.2083661159192237e-05,
+      "loss": 2.4107,
+      "step": 46390
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.2081475653466213e-05,
+      "loss": 2.4362,
+      "step": 46400
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.2079290147740189e-05,
+      "loss": 2.3511,
+      "step": 46410
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.2077104642014163e-05,
+      "loss": 2.4527,
+      "step": 46420
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.2074919136288139e-05,
+      "loss": 2.4013,
+      "step": 46430
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.2072733630562113e-05,
+      "loss": 2.4164,
+      "step": 46440
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.2070548124836087e-05,
+      "loss": 2.4052,
+      "step": 46450
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.2068362619110063e-05,
+      "loss": 2.3933,
+      "step": 46460
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.2066177113384039e-05,
+      "loss": 2.3926,
+      "step": 46470
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.2063991607658011e-05,
+      "loss": 2.3772,
+      "step": 46480
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.2061806101931987e-05,
+      "loss": 2.4259,
+      "step": 46490
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.2059620596205963e-05,
+      "loss": 2.3819,
+      "step": 46500
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.2057435090479939e-05,
+      "loss": 2.4004,
+      "step": 46510
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.2055249584753912e-05,
+      "loss": 2.3816,
+      "step": 46520
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.2053064079027887e-05,
+      "loss": 2.4253,
+      "step": 46530
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.2050878573301863e-05,
+      "loss": 2.3998,
+      "step": 46540
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.2048693067575839e-05,
+      "loss": 2.4159,
+      "step": 46550
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.2046507561849813e-05,
+      "loss": 2.4035,
+      "step": 46560
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.2044322056123788e-05,
+      "loss": 2.4125,
+      "step": 46570
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.2042136550397763e-05,
+      "loss": 2.4053,
+      "step": 46580
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.2039951044671738e-05,
+      "loss": 2.4067,
+      "step": 46590
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.2037765538945713e-05,
+      "loss": 2.4042,
+      "step": 46600
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.2035580033219688e-05,
+      "loss": 2.4207,
+      "step": 46610
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.2033394527493662e-05,
+      "loss": 2.4281,
+      "step": 46620
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.2031209021767638e-05,
+      "loss": 2.3955,
+      "step": 46630
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.2029023516041614e-05,
+      "loss": 2.3827,
+      "step": 46640
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.202683801031559e-05,
+      "loss": 2.376,
+      "step": 46650
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.2024652504589562e-05,
+      "loss": 2.4172,
+      "step": 46660
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.2022466998863538e-05,
+      "loss": 2.4021,
+      "step": 46670
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.2020281493137514e-05,
+      "loss": 2.3996,
+      "step": 46680
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.201809598741149e-05,
+      "loss": 2.4151,
+      "step": 46690
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.2015910481685462e-05,
+      "loss": 2.4289,
+      "step": 46700
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.2013724975959438e-05,
+      "loss": 2.4041,
+      "step": 46710
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.2011539470233414e-05,
+      "loss": 2.4486,
+      "step": 46720
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.2009353964507388e-05,
+      "loss": 2.4233,
+      "step": 46730
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.2007168458781362e-05,
+      "loss": 2.4054,
+      "step": 46740
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.2004982953055338e-05,
+      "loss": 2.419,
+      "step": 46750
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.2002797447329312e-05,
+      "loss": 2.3522,
+      "step": 46760
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.2000611941603288e-05,
+      "loss": 2.4039,
+      "step": 46770
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.1998426435877264e-05,
+      "loss": 2.3906,
+      "step": 46780
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.1996240930151237e-05,
+      "loss": 2.4244,
+      "step": 46790
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.1994055424425212e-05,
+      "loss": 2.4155,
+      "step": 46800
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.1991869918699188e-05,
+      "loss": 2.3881,
+      "step": 46810
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.1989684412973164e-05,
+      "loss": 2.4168,
+      "step": 46820
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.1987498907247137e-05,
+      "loss": 2.4507,
+      "step": 46830
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.1985313401521113e-05,
+      "loss": 2.4074,
+      "step": 46840
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.1983127895795088e-05,
+      "loss": 2.383,
+      "step": 46850
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.1980942390069064e-05,
+      "loss": 2.3896,
+      "step": 46860
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.1978756884343037e-05,
+      "loss": 2.3786,
+      "step": 46870
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.1976571378617013e-05,
+      "loss": 2.451,
+      "step": 46880
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.1974385872890989e-05,
+      "loss": 2.4026,
+      "step": 46890
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.1972200367164963e-05,
+      "loss": 2.4184,
+      "step": 46900
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.1970014861438939e-05,
+      "loss": 2.4028,
+      "step": 46910
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.1967829355712913e-05,
+      "loss": 2.4578,
+      "step": 46920
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.1965643849986887e-05,
+      "loss": 2.4036,
+      "step": 46930
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.1963458344260863e-05,
+      "loss": 2.42,
+      "step": 46940
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.1961272838534839e-05,
+      "loss": 2.4087,
+      "step": 46950
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.1959087332808811e-05,
+      "loss": 2.4123,
+      "step": 46960
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.1956901827082787e-05,
+      "loss": 2.403,
+      "step": 46970
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.1954716321356763e-05,
+      "loss": 2.3849,
+      "step": 46980
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.1952530815630739e-05,
+      "loss": 2.3763,
+      "step": 46990
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.1950345309904715e-05,
+      "loss": 2.3732,
+      "step": 47000
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.1948159804178687e-05,
+      "loss": 2.3952,
+      "step": 47010
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.1945974298452663e-05,
+      "loss": 2.4249,
+      "step": 47020
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.1943788792726639e-05,
+      "loss": 2.3942,
+      "step": 47030
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.1941603287000613e-05,
+      "loss": 2.4129,
+      "step": 47040
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.1939417781274587e-05,
+      "loss": 2.4246,
+      "step": 47050
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.1937232275548563e-05,
+      "loss": 2.3883,
+      "step": 47060
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.1935046769822537e-05,
+      "loss": 2.4082,
+      "step": 47070
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.1932861264096513e-05,
+      "loss": 2.3658,
+      "step": 47080
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.1930675758370487e-05,
+      "loss": 2.4228,
+      "step": 47090
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.1928490252644462e-05,
+      "loss": 2.4052,
+      "step": 47100
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.1926304746918438e-05,
+      "loss": 2.4125,
+      "step": 47110
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.1924119241192413e-05,
+      "loss": 2.3791,
+      "step": 47120
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.192193373546639e-05,
+      "loss": 2.4352,
+      "step": 47130
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.1919748229740362e-05,
+      "loss": 2.4567,
+      "step": 47140
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.1917562724014338e-05,
+      "loss": 2.4341,
+      "step": 47150
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.1915377218288314e-05,
+      "loss": 2.4179,
+      "step": 47160
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.191319171256229e-05,
+      "loss": 2.3957,
+      "step": 47170
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.1911006206836262e-05,
+      "loss": 2.4032,
+      "step": 47180
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.1908820701110238e-05,
+      "loss": 2.3906,
+      "step": 47190
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.1906635195384214e-05,
+      "loss": 2.3927,
+      "step": 47200
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.1904449689658188e-05,
+      "loss": 2.3742,
+      "step": 47210
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.1902264183932164e-05,
+      "loss": 2.4024,
+      "step": 47220
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.1900078678206138e-05,
+      "loss": 2.4055,
+      "step": 47230
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.1897893172480112e-05,
+      "loss": 2.4258,
+      "step": 47240
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.1895707666754088e-05,
+      "loss": 2.3655,
+      "step": 47250
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.1893522161028064e-05,
+      "loss": 2.3892,
+      "step": 47260
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.1891336655302036e-05,
+      "loss": 2.4145,
+      "step": 47270
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.1889151149576012e-05,
+      "loss": 2.4302,
+      "step": 47280
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.1886965643849988e-05,
+      "loss": 2.4151,
+      "step": 47290
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.1884780138123964e-05,
+      "loss": 2.4365,
+      "step": 47300
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.1882594632397936e-05,
+      "loss": 2.3854,
+      "step": 47310
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.1880409126671912e-05,
+      "loss": 2.4235,
+      "step": 47320
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.1878223620945888e-05,
+      "loss": 2.4238,
+      "step": 47330
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.1876038115219864e-05,
+      "loss": 2.4377,
+      "step": 47340
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.1873852609493838e-05,
+      "loss": 2.3822,
+      "step": 47350
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.1871667103767812e-05,
+      "loss": 2.4074,
+      "step": 47360
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.1869481598041788e-05,
+      "loss": 2.411,
+      "step": 47370
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.1867296092315763e-05,
+      "loss": 2.4052,
+      "step": 47380
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.1865110586589738e-05,
+      "loss": 2.428,
+      "step": 47390
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.1862925080863713e-05,
+      "loss": 2.454,
+      "step": 47400
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.1860739575137687e-05,
+      "loss": 2.3928,
+      "step": 47410
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.1858554069411663e-05,
+      "loss": 2.4088,
+      "step": 47420
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.1856368563685639e-05,
+      "loss": 2.4657,
+      "step": 47430
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.1854183057959611e-05,
+      "loss": 2.4092,
+      "step": 47440
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.1851997552233587e-05,
+      "loss": 2.4054,
+      "step": 47450
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.1849812046507563e-05,
+      "loss": 2.424,
+      "step": 47460
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.1847626540781539e-05,
+      "loss": 2.4514,
+      "step": 47470
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.1845441035055515e-05,
+      "loss": 2.4046,
+      "step": 47480
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.1843255529329487e-05,
+      "loss": 2.4333,
+      "step": 47490
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.1841070023603463e-05,
+      "loss": 2.4243,
+      "step": 47500
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.1838884517877439e-05,
+      "loss": 2.4189,
+      "step": 47510
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.1836699012151413e-05,
+      "loss": 2.3826,
+      "step": 47520
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.1834513506425387e-05,
+      "loss": 2.4025,
+      "step": 47530
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.1832328000699363e-05,
+      "loss": 2.4425,
+      "step": 47540
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.1830142494973337e-05,
+      "loss": 2.4578,
+      "step": 47550
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.1827956989247313e-05,
+      "loss": 2.4262,
+      "step": 47560
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.1825771483521289e-05,
+      "loss": 2.4101,
+      "step": 47570
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.1823585977795261e-05,
+      "loss": 2.4122,
+      "step": 47580
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.1821400472069237e-05,
+      "loss": 2.3998,
+      "step": 47590
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.1819214966343213e-05,
+      "loss": 2.415,
+      "step": 47600
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.1817029460617189e-05,
+      "loss": 2.4668,
+      "step": 47610
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.1814843954891162e-05,
+      "loss": 2.3881,
+      "step": 47620
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.1812658449165137e-05,
+      "loss": 2.444,
+      "step": 47630
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.1810472943439113e-05,
+      "loss": 2.399,
+      "step": 47640
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.180828743771309e-05,
+      "loss": 2.4097,
+      "step": 47650
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.1806101931987062e-05,
+      "loss": 2.4395,
+      "step": 47660
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.1803916426261038e-05,
+      "loss": 2.4087,
+      "step": 47670
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.1801730920535013e-05,
+      "loss": 2.3867,
+      "step": 47680
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.1799545414808988e-05,
+      "loss": 2.3898,
+      "step": 47690
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.1797359909082964e-05,
+      "loss": 2.4384,
+      "step": 47700
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.1795174403356938e-05,
+      "loss": 2.4373,
+      "step": 47710
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.1792988897630912e-05,
+      "loss": 2.4069,
+      "step": 47720
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.1790803391904888e-05,
+      "loss": 2.4349,
+      "step": 47730
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.1788617886178864e-05,
+      "loss": 2.4377,
+      "step": 47740
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.1786432380452836e-05,
+      "loss": 2.4142,
+      "step": 47750
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.1784246874726812e-05,
+      "loss": 2.4185,
+      "step": 47760
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.1782061369000788e-05,
+      "loss": 2.4544,
+      "step": 47770
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.1779875863274764e-05,
+      "loss": 2.416,
+      "step": 47780
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.1777690357548736e-05,
+      "loss": 2.4492,
+      "step": 47790
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.1775504851822712e-05,
+      "loss": 2.394,
+      "step": 47800
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.1773319346096688e-05,
+      "loss": 2.4126,
+      "step": 47810
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.1771133840370664e-05,
+      "loss": 2.4004,
+      "step": 47820
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.1768948334644638e-05,
+      "loss": 2.4345,
+      "step": 47830
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.1766762828918612e-05,
+      "loss": 2.4241,
+      "step": 47840
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.1764577323192588e-05,
+      "loss": 2.3998,
+      "step": 47850
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.1762391817466562e-05,
+      "loss": 2.4166,
+      "step": 47860
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.1760206311740538e-05,
+      "loss": 2.4379,
+      "step": 47870
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.1758020806014512e-05,
+      "loss": 2.4389,
+      "step": 47880
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.1755835300288487e-05,
+      "loss": 2.4343,
+      "step": 47890
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.1753649794562462e-05,
+      "loss": 2.4272,
+      "step": 47900
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.1751464288836438e-05,
+      "loss": 2.4154,
+      "step": 47910
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.1749278783110414e-05,
+      "loss": 2.4231,
+      "step": 47920
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.1747093277384387e-05,
+      "loss": 2.4039,
+      "step": 47930
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.1744907771658363e-05,
+      "loss": 2.4209,
+      "step": 47940
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.1742722265932338e-05,
+      "loss": 2.4361,
+      "step": 47950
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.1740536760206314e-05,
+      "loss": 2.4155,
+      "step": 47960
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.1738351254480287e-05,
+      "loss": 2.3704,
+      "step": 47970
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.1736165748754263e-05,
+      "loss": 2.4513,
+      "step": 47980
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.1733980243028239e-05,
+      "loss": 2.4097,
+      "step": 47990
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.1731794737302213e-05,
+      "loss": 2.4353,
+      "step": 48000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.1729609231576187e-05,
+      "loss": 2.4358,
+      "step": 48010
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.1727423725850163e-05,
+      "loss": 2.4076,
+      "step": 48020
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.1725238220124137e-05,
+      "loss": 2.4217,
+      "step": 48030
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.1723052714398113e-05,
+      "loss": 2.4088,
+      "step": 48040
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.1720867208672089e-05,
+      "loss": 2.4872,
+      "step": 48050
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.1718681702946061e-05,
+      "loss": 2.4307,
+      "step": 48060
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.1716496197220037e-05,
+      "loss": 2.4529,
+      "step": 48070
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.1714310691494013e-05,
+      "loss": 2.3683,
+      "step": 48080
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.1712125185767989e-05,
+      "loss": 2.4266,
+      "step": 48090
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.1709939680041961e-05,
+      "loss": 2.4561,
+      "step": 48100
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.1707754174315937e-05,
+      "loss": 2.4313,
+      "step": 48110
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.1705568668589913e-05,
+      "loss": 2.4055,
+      "step": 48120
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.1703383162863889e-05,
+      "loss": 2.4431,
+      "step": 48130
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.1701197657137862e-05,
+      "loss": 2.4258,
+      "step": 48140
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.1699012151411837e-05,
+      "loss": 2.4432,
+      "step": 48150
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.1696826645685813e-05,
+      "loss": 2.4427,
+      "step": 48160
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.1694641139959787e-05,
+      "loss": 2.4087,
+      "step": 48170
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.1692455634233763e-05,
+      "loss": 2.3971,
+      "step": 48180
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.1690270128507738e-05,
+      "loss": 2.3939,
+      "step": 48190
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.1688084622781712e-05,
+      "loss": 2.4475,
+      "step": 48200
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.1685899117055688e-05,
+      "loss": 2.3942,
+      "step": 48210
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.1683713611329664e-05,
+      "loss": 2.4053,
+      "step": 48220
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.1681528105603636e-05,
+      "loss": 2.4313,
+      "step": 48230
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.1679342599877612e-05,
+      "loss": 2.4448,
+      "step": 48240
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.1677157094151588e-05,
+      "loss": 2.4157,
+      "step": 48250
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.1674971588425564e-05,
+      "loss": 2.3971,
+      "step": 48260
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.167278608269954e-05,
+      "loss": 2.4295,
+      "step": 48270
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.1670600576973512e-05,
+      "loss": 2.4339,
+      "step": 48280
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.1668415071247488e-05,
+      "loss": 2.4025,
+      "step": 48290
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.1666229565521464e-05,
+      "loss": 2.4564,
+      "step": 48300
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.1664044059795438e-05,
+      "loss": 2.4145,
+      "step": 48310
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.1661858554069412e-05,
+      "loss": 2.4418,
+      "step": 48320
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.1659673048343388e-05,
+      "loss": 2.4298,
+      "step": 48330
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.1657487542617362e-05,
+      "loss": 2.4303,
+      "step": 48340
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.1655302036891338e-05,
+      "loss": 2.4382,
+      "step": 48350
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.1653116531165312e-05,
+      "loss": 2.3953,
+      "step": 48360
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.1650931025439286e-05,
+      "loss": 2.4054,
+      "step": 48370
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.1648745519713262e-05,
+      "loss": 2.4234,
+      "step": 48380
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.1646560013987238e-05,
+      "loss": 2.4772,
+      "step": 48390
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.1644374508261214e-05,
+      "loss": 2.4361,
+      "step": 48400
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.1642189002535187e-05,
+      "loss": 2.4193,
+      "step": 48410
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.1640003496809162e-05,
+      "loss": 2.4843,
+      "step": 48420
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.1637817991083138e-05,
+      "loss": 2.3969,
+      "step": 48430
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.1635632485357114e-05,
+      "loss": 2.4302,
+      "step": 48440
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.1633446979631087e-05,
+      "loss": 2.4195,
+      "step": 48450
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.1631261473905063e-05,
+      "loss": 2.4588,
+      "step": 48460
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.1629075968179038e-05,
+      "loss": 2.4303,
+      "step": 48470
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.1626890462453013e-05,
+      "loss": 2.4198,
+      "step": 48480
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.1624704956726987e-05,
+      "loss": 2.4111,
+      "step": 48490
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.1622519451000963e-05,
+      "loss": 2.4148,
+      "step": 48500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.1620333945274937e-05,
+      "loss": 2.4216,
+      "step": 48510
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.1618148439548913e-05,
+      "loss": 2.4353,
+      "step": 48520
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.1615962933822889e-05,
+      "loss": 2.4563,
+      "step": 48530
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.1613777428096861e-05,
+      "loss": 2.4263,
+      "step": 48540
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.1611591922370837e-05,
+      "loss": 2.426,
+      "step": 48550
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.1609406416644813e-05,
+      "loss": 2.4218,
+      "step": 48560
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.1607220910918789e-05,
+      "loss": 2.4626,
+      "step": 48570
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.1605035405192761e-05,
+      "loss": 2.4242,
+      "step": 48580
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.1602849899466737e-05,
+      "loss": 2.4438,
+      "step": 48590
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.1600664393740713e-05,
+      "loss": 2.4109,
+      "step": 48600
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.1598478888014689e-05,
+      "loss": 2.4201,
+      "step": 48610
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.1596293382288663e-05,
+      "loss": 2.4388,
+      "step": 48620
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.1594107876562637e-05,
+      "loss": 2.4244,
+      "step": 48630
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.1591922370836613e-05,
+      "loss": 2.4781,
+      "step": 48640
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.1589736865110587e-05,
+      "loss": 2.3933,
+      "step": 48650
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.1587551359384563e-05,
+      "loss": 2.441,
+      "step": 48660
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.1585365853658537e-05,
+      "loss": 2.4568,
+      "step": 48670
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.1583180347932512e-05,
+      "loss": 2.4407,
+      "step": 48680
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.1580994842206487e-05,
+      "loss": 2.4194,
+      "step": 48690
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.1578809336480463e-05,
+      "loss": 2.411,
+      "step": 48700
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.1576623830754436e-05,
+      "loss": 2.3839,
+      "step": 48710
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.1574438325028412e-05,
+      "loss": 2.421,
+      "step": 48720
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.1572252819302388e-05,
+      "loss": 2.4309,
+      "step": 48730
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.1570067313576363e-05,
+      "loss": 2.4489,
+      "step": 48740
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.156788180785034e-05,
+      "loss": 2.4732,
+      "step": 48750
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.1565696302124312e-05,
+      "loss": 2.4262,
+      "step": 48760
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.1563510796398288e-05,
+      "loss": 2.4285,
+      "step": 48770
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.1561325290672264e-05,
+      "loss": 2.4679,
+      "step": 48780
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.1559139784946238e-05,
+      "loss": 2.4389,
+      "step": 48790
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.1556954279220212e-05,
+      "loss": 2.4097,
+      "step": 48800
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.1554768773494188e-05,
+      "loss": 2.4037,
+      "step": 48810
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.1552583267768162e-05,
+      "loss": 2.3698,
+      "step": 48820
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.1550397762042138e-05,
+      "loss": 2.3875,
+      "step": 48830
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1548212256316112e-05,
+      "loss": 2.4367,
+      "step": 48840
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1546026750590086e-05,
+      "loss": 2.4459,
+      "step": 48850
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1543841244864062e-05,
+      "loss": 2.4435,
+      "step": 48860
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1541655739138038e-05,
+      "loss": 2.4464,
+      "step": 48870
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1539470233412014e-05,
+      "loss": 2.4314,
+      "step": 48880
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1537284727685986e-05,
+      "loss": 2.4219,
+      "step": 48890
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1535099221959962e-05,
+      "loss": 2.4611,
+      "step": 48900
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1532913716233938e-05,
+      "loss": 2.4308,
+      "step": 48910
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1530728210507914e-05,
+      "loss": 2.3951,
+      "step": 48920
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1528542704781887e-05,
+      "loss": 2.4231,
+      "step": 48930
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1526357199055862e-05,
+      "loss": 2.4382,
+      "step": 48940
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1524171693329838e-05,
+      "loss": 2.4484,
+      "step": 48950
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1521986187603812e-05,
+      "loss": 2.4433,
+      "step": 48960
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1519800681877788e-05,
+      "loss": 2.4028,
+      "step": 48970
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1517615176151763e-05,
+      "loss": 2.4472,
+      "step": 48980
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1515429670425737e-05,
+      "loss": 2.4005,
+      "step": 48990
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1513244164699713e-05,
+      "loss": 2.4062,
+      "step": 49000
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1511058658973688e-05,
+      "loss": 2.4267,
+      "step": 49010
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1508873153247661e-05,
+      "loss": 2.4293,
+      "step": 49020
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1506687647521637e-05,
+      "loss": 2.4526,
+      "step": 49030
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1504502141795613e-05,
+      "loss": 2.4143,
+      "step": 49040
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1502316636069589e-05,
+      "loss": 2.4146,
+      "step": 49050
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1500131130343561e-05,
+      "loss": 2.4391,
+      "step": 49060
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1497945624617537e-05,
+      "loss": 2.4652,
+      "step": 49070
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1495760118891513e-05,
+      "loss": 2.4181,
+      "step": 49080
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1493574613165489e-05,
+      "loss": 2.3942,
+      "step": 49090
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1491389107439463e-05,
+      "loss": 2.4437,
+      "step": 49100
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1489203601713437e-05,
+      "loss": 2.39,
+      "step": 49110
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1487018095987413e-05,
+      "loss": 2.4037,
+      "step": 49120
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1484832590261387e-05,
+      "loss": 2.3659,
+      "step": 49130
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1482647084535363e-05,
+      "loss": 2.4487,
+      "step": 49140
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1480461578809337e-05,
+      "loss": 2.4631,
+      "step": 49150
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1478276073083311e-05,
+      "loss": 2.4274,
+      "step": 49160
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1476090567357287e-05,
+      "loss": 2.4317,
+      "step": 49170
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1473905061631263e-05,
+      "loss": 2.4371,
+      "step": 49180
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.1471719555905236e-05,
+      "loss": 2.4211,
+      "step": 49190
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.1469534050179212e-05,
+      "loss": 2.4315,
+      "step": 49200
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.1467348544453187e-05,
+      "loss": 2.4197,
+      "step": 49210
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.1465163038727163e-05,
+      "loss": 2.4497,
+      "step": 49220
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.146297753300114e-05,
+      "loss": 2.4127,
+      "step": 49230
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.1460792027275112e-05,
+      "loss": 2.4027,
+      "step": 49240
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.1458606521549088e-05,
+      "loss": 2.4382,
+      "step": 49250
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.1456421015823063e-05,
+      "loss": 2.4708,
+      "step": 49260
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.1454235510097038e-05,
+      "loss": 2.4099,
+      "step": 49270
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.1452050004371012e-05,
+      "loss": 2.4302,
+      "step": 49280
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.1449864498644988e-05,
+      "loss": 2.4322,
+      "step": 49290
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.1447678992918962e-05,
+      "loss": 2.4389,
+      "step": 49300
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.1445493487192938e-05,
+      "loss": 2.4872,
+      "step": 49310
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.1443307981466914e-05,
+      "loss": 2.3886,
+      "step": 49320
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.1441122475740886e-05,
+      "loss": 2.4346,
+      "step": 49330
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.1438936970014862e-05,
+      "loss": 2.4495,
+      "step": 49340
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.1436751464288838e-05,
+      "loss": 2.4013,
+      "step": 49350
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.1434565958562814e-05,
+      "loss": 2.4348,
+      "step": 49360
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.1432380452836786e-05,
+      "loss": 2.4511,
+      "step": 49370
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.1430194947110762e-05,
+      "loss": 2.4564,
+      "step": 49380
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.1428009441384738e-05,
+      "loss": 2.4314,
+      "step": 49390
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.1425823935658714e-05,
+      "loss": 2.4336,
+      "step": 49400
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.1423638429932686e-05,
+      "loss": 2.4291,
+      "step": 49410
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.1421452924206662e-05,
+      "loss": 2.3889,
+      "step": 49420
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.1419267418480638e-05,
+      "loss": 2.4116,
+      "step": 49430
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.1417081912754612e-05,
+      "loss": 2.3943,
+      "step": 49440
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.1414896407028588e-05,
+      "loss": 2.4486,
+      "step": 49450
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.1412710901302562e-05,
+      "loss": 2.4301,
+      "step": 49460
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.1410525395576537e-05,
+      "loss": 2.4077,
+      "step": 49470
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.1408339889850512e-05,
+      "loss": 2.4461,
+      "step": 49480
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.1406154384124488e-05,
+      "loss": 2.4099,
+      "step": 49490
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.140396887839846e-05,
+      "loss": 2.4625,
+      "step": 49500
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.1401783372672437e-05,
+      "loss": 2.4254,
+      "step": 49510
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.1399597866946413e-05,
+      "loss": 2.4583,
+      "step": 49520
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.1397412361220388e-05,
+      "loss": 2.4251,
+      "step": 49530
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.1395226855494361e-05,
+      "loss": 2.3826,
+      "step": 49540
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.1393041349768337e-05,
+      "loss": 2.4536,
+      "step": 49550
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.1390855844042313e-05,
+      "loss": 2.4417,
+      "step": 49560
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.1388670338316289e-05,
+      "loss": 2.4361,
+      "step": 49570
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.1386484832590263e-05,
+      "loss": 2.3946,
+      "step": 49580
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.1384299326864237e-05,
+      "loss": 2.4705,
+      "step": 49590
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.1382113821138213e-05,
+      "loss": 2.4169,
+      "step": 49600
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.1379928315412187e-05,
+      "loss": 2.4187,
+      "step": 49610
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.1377742809686163e-05,
+      "loss": 2.3735,
+      "step": 49620
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.1375557303960137e-05,
+      "loss": 2.4349,
+      "step": 49630
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.1373371798234111e-05,
+      "loss": 2.4766,
+      "step": 49640
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1371186292508087e-05,
+      "loss": 2.4362,
+      "step": 49650
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1369000786782063e-05,
+      "loss": 2.4531,
+      "step": 49660
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1366815281056039e-05,
+      "loss": 2.4437,
+      "step": 49670
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1364629775330011e-05,
+      "loss": 2.4363,
+      "step": 49680
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1362444269603987e-05,
+      "loss": 2.431,
+      "step": 49690
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1360258763877963e-05,
+      "loss": 2.4393,
+      "step": 49700
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1358073258151939e-05,
+      "loss": 2.4352,
+      "step": 49710
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1355887752425911e-05,
+      "loss": 2.4413,
+      "step": 49720
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1353702246699887e-05,
+      "loss": 2.4619,
+      "step": 49730
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1351516740973863e-05,
+      "loss": 2.4003,
+      "step": 49740
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.1349331235247837e-05,
+      "loss": 2.4511,
+      "step": 49750
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.1347145729521812e-05,
+      "loss": 2.4292,
+      "step": 49760
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.1344960223795787e-05,
+      "loss": 2.4097,
+      "step": 49770
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.1342774718069762e-05,
+      "loss": 2.451,
+      "step": 49780
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.1340589212343738e-05,
+      "loss": 2.4201,
+      "step": 49790
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.1338403706617713e-05,
+      "loss": 2.4507,
+      "step": 49800
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.1336218200891686e-05,
+      "loss": 2.4887,
+      "step": 49810
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.1334032695165662e-05,
+      "loss": 2.4519,
+      "step": 49820
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.1331847189439638e-05,
+      "loss": 2.4438,
+      "step": 49830
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.1329661683713614e-05,
+      "loss": 2.4425,
+      "step": 49840
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.1327476177987586e-05,
+      "loss": 2.4364,
+      "step": 49850
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.1325290672261562e-05,
+      "loss": 2.4329,
+      "step": 49860
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.1323105166535538e-05,
+      "loss": 2.4372,
+      "step": 49870
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.1320919660809514e-05,
+      "loss": 2.4316,
+      "step": 49880
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.1318734155083486e-05,
+      "loss": 2.462,
+      "step": 49890
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.1316548649357462e-05,
+      "loss": 2.4742,
+      "step": 49900
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.1314363143631438e-05,
+      "loss": 2.4322,
+      "step": 49910
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.1312177637905412e-05,
+      "loss": 2.442,
+      "step": 49920
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.1309992132179388e-05,
+      "loss": 2.4257,
+      "step": 49930
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.1307806626453362e-05,
+      "loss": 2.4245,
+      "step": 49940
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.1305621120727336e-05,
+      "loss": 2.4165,
+      "step": 49950
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.1303435615001312e-05,
+      "loss": 2.4528,
+      "step": 49960
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.1301250109275288e-05,
+      "loss": 2.4591,
+      "step": 49970
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.129906460354926e-05,
+      "loss": 2.471,
+      "step": 49980
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.1296879097823236e-05,
+      "loss": 2.4502,
+      "step": 49990
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.1294693592097212e-05,
+      "loss": 2.451,
+      "step": 50000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.1292508086371188e-05,
+      "loss": 2.4614,
+      "step": 50010
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 2.4233,
+      "step": 50020
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.1288137074919137e-05,
+      "loss": 2.4499,
+      "step": 50030
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.1285951569193112e-05,
+      "loss": 2.4535,
+      "step": 50040
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.1283766063467088e-05,
+      "loss": 2.4462,
+      "step": 50050
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.1281580557741063e-05,
+      "loss": 2.4736,
+      "step": 50060
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.1279395052015037e-05,
+      "loss": 2.4461,
+      "step": 50070
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.1277209546289013e-05,
+      "loss": 2.4538,
+      "step": 50080
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.1275024040562987e-05,
+      "loss": 2.4032,
+      "step": 50090
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.1272838534836963e-05,
+      "loss": 2.4437,
+      "step": 50100
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.1270653029110937e-05,
+      "loss": 2.4695,
+      "step": 50110
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.1268467523384911e-05,
+      "loss": 2.4282,
+      "step": 50120
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.1266282017658887e-05,
+      "loss": 2.4339,
+      "step": 50130
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.1264096511932863e-05,
+      "loss": 2.4385,
+      "step": 50140
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.1261911006206839e-05,
+      "loss": 2.4438,
+      "step": 50150
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.1259725500480811e-05,
+      "loss": 2.4677,
+      "step": 50160
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.1257539994754787e-05,
+      "loss": 2.4306,
+      "step": 50170
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.1255354489028763e-05,
+      "loss": 2.4894,
+      "step": 50180
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.1253168983302739e-05,
+      "loss": 2.4644,
+      "step": 50190
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.1250983477576711e-05,
+      "loss": 2.4262,
+      "step": 50200
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.1248797971850687e-05,
+      "loss": 2.4336,
+      "step": 50210
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.1246612466124663e-05,
+      "loss": 2.3984,
+      "step": 50220
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.1244426960398637e-05,
+      "loss": 2.457,
+      "step": 50230
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.1242241454672611e-05,
+      "loss": 2.442,
+      "step": 50240
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.1240055948946587e-05,
+      "loss": 2.4483,
+      "step": 50250
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.1237870443220561e-05,
+      "loss": 2.4903,
+      "step": 50260
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.1235684937494537e-05,
+      "loss": 2.3746,
+      "step": 50270
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.1233499431768513e-05,
+      "loss": 2.4378,
+      "step": 50280
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.1231313926042486e-05,
+      "loss": 2.4357,
+      "step": 50290
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.1229128420316462e-05,
+      "loss": 2.3893,
+      "step": 50300
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.1226942914590438e-05,
+      "loss": 2.453,
+      "step": 50310
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.1224757408864413e-05,
+      "loss": 2.4528,
+      "step": 50320
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.1222571903138386e-05,
+      "loss": 2.3869,
+      "step": 50330
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.1220386397412362e-05,
+      "loss": 2.4518,
+      "step": 50340
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.1218200891686338e-05,
+      "loss": 2.4582,
+      "step": 50350
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.1216015385960314e-05,
+      "loss": 2.4602,
+      "step": 50360
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.1213829880234288e-05,
+      "loss": 2.4229,
+      "step": 50370
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.1211644374508262e-05,
+      "loss": 2.4422,
+      "step": 50380
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.1209458868782238e-05,
+      "loss": 2.4588,
+      "step": 50390
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.1207273363056212e-05,
+      "loss": 2.4649,
+      "step": 50400
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.1205087857330188e-05,
+      "loss": 2.4808,
+      "step": 50410
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.1202902351604162e-05,
+      "loss": 2.4271,
+      "step": 50420
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.1200716845878136e-05,
+      "loss": 2.4281,
+      "step": 50430
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.1198531340152112e-05,
+      "loss": 2.4455,
+      "step": 50440
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.1196345834426088e-05,
+      "loss": 2.4344,
+      "step": 50450
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.119416032870006e-05,
+      "loss": 2.427,
+      "step": 50460
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.1191974822974036e-05,
+      "loss": 2.4277,
+      "step": 50470
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.1189789317248012e-05,
+      "loss": 2.4038,
+      "step": 50480
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.1187603811521988e-05,
+      "loss": 2.4264,
+      "step": 50490
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.1185418305795962e-05,
+      "loss": 2.4519,
+      "step": 50500
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.1183232800069936e-05,
+      "loss": 2.3947,
+      "step": 50510
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.1181047294343912e-05,
+      "loss": 2.4235,
+      "step": 50520
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.1178861788617888e-05,
+      "loss": 2.4875,
+      "step": 50530
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.1176676282891862e-05,
+      "loss": 2.4587,
+      "step": 50540
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.1174490777165837e-05,
+      "loss": 2.443,
+      "step": 50550
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.1172305271439812e-05,
+      "loss": 2.4465,
+      "step": 50560
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.1170119765713787e-05,
+      "loss": 2.3973,
+      "step": 50570
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.1167934259987763e-05,
+      "loss": 2.4286,
+      "step": 50580
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.1165748754261737e-05,
+      "loss": 2.4676,
+      "step": 50590
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.1163563248535711e-05,
+      "loss": 2.4595,
+      "step": 50600
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.1161377742809687e-05,
+      "loss": 2.4702,
+      "step": 50610
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.1159192237083663e-05,
+      "loss": 2.447,
+      "step": 50620
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.1157006731357639e-05,
+      "loss": 2.4643,
+      "step": 50630
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.1154821225631611e-05,
+      "loss": 2.4319,
+      "step": 50640
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.1152635719905587e-05,
+      "loss": 2.4165,
+      "step": 50650
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.1150450214179563e-05,
+      "loss": 2.4526,
+      "step": 50660
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.1148264708453539e-05,
+      "loss": 2.4039,
+      "step": 50670
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.1146079202727511e-05,
+      "loss": 2.4385,
+      "step": 50680
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.1143893697001487e-05,
+      "loss": 2.4231,
+      "step": 50690
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.1141708191275463e-05,
+      "loss": 2.4503,
+      "step": 50700
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.1139522685549437e-05,
+      "loss": 2.4361,
+      "step": 50710
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.1137337179823413e-05,
+      "loss": 2.4792,
+      "step": 50720
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.1135151674097387e-05,
+      "loss": 2.4404,
+      "step": 50730
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.1132966168371361e-05,
+      "loss": 2.4559,
+      "step": 50740
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.1130780662645337e-05,
+      "loss": 2.4383,
+      "step": 50750
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.1128595156919313e-05,
+      "loss": 2.4656,
+      "step": 50760
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.1126409651193286e-05,
+      "loss": 2.4252,
+      "step": 50770
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.1124224145467261e-05,
+      "loss": 2.4456,
+      "step": 50780
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.1122038639741237e-05,
+      "loss": 2.4074,
+      "step": 50790
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.1119853134015213e-05,
+      "loss": 2.4492,
+      "step": 50800
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.1117667628289186e-05,
+      "loss": 2.4589,
+      "step": 50810
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1115482122563162e-05,
+      "loss": 2.4341,
+      "step": 50820
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1113296616837137e-05,
+      "loss": 2.4317,
+      "step": 50830
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 2.4385,
+      "step": 50840
+    },
+    {
+      "epoch": 10.0,
+      "eval_BLEU": 9.0563,
+      "eval_gen_len": 11.4229,
+      "eval_loss": 3.114694833755493,
+      "eval_runtime": 149.2271,
+      "eval_samples_per_second": 72.125,
+      "eval_steps_per_second": 0.288,
+      "step": 50840
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1108925605385088e-05,
+      "loss": 2.3201,
+      "step": 50850
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.1106740099659062e-05,
+      "loss": 2.3289,
+      "step": 50860
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1104554593933038e-05,
+      "loss": 2.3354,
+      "step": 50870
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1102369088207012e-05,
+      "loss": 2.3151,
+      "step": 50880
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1100183582480988e-05,
+      "loss": 2.3262,
+      "step": 50890
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1097998076754962e-05,
+      "loss": 2.3512,
+      "step": 50900
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.1095812571028936e-05,
+      "loss": 2.3557,
+      "step": 50910
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.1093627065302912e-05,
+      "loss": 2.3312,
+      "step": 50920
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.1091441559576888e-05,
+      "loss": 2.3116,
+      "step": 50930
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.108925605385086e-05,
+      "loss": 2.309,
+      "step": 50940
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.1087070548124836e-05,
+      "loss": 2.3013,
+      "step": 50950
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.1084885042398812e-05,
+      "loss": 2.322,
+      "step": 50960
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.1082699536672788e-05,
+      "loss": 2.318,
+      "step": 50970
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.1080514030946762e-05,
+      "loss": 2.3743,
+      "step": 50980
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.1078328525220736e-05,
+      "loss": 2.3145,
+      "step": 50990
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.1076143019494712e-05,
+      "loss": 2.3165,
+      "step": 51000
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.1073957513768688e-05,
+      "loss": 2.336,
+      "step": 51010
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.1071772008042662e-05,
+      "loss": 2.3361,
+      "step": 51020
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.1069586502316636e-05,
+      "loss": 2.301,
+      "step": 51030
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.1067400996590612e-05,
+      "loss": 2.2711,
+      "step": 51040
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.1065215490864586e-05,
+      "loss": 2.3134,
+      "step": 51050
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.1063029985138562e-05,
+      "loss": 2.3506,
+      "step": 51060
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.1060844479412538e-05,
+      "loss": 2.3202,
+      "step": 51070
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.105865897368651e-05,
+      "loss": 2.3202,
+      "step": 51080
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.1056473467960487e-05,
+      "loss": 2.2915,
+      "step": 51090
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.1054287962234462e-05,
+      "loss": 2.3317,
+      "step": 51100
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.1052102456508438e-05,
+      "loss": 2.3137,
+      "step": 51110
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.104991695078241e-05,
+      "loss": 2.3515,
+      "step": 51120
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.1047731445056387e-05,
+      "loss": 2.3145,
+      "step": 51130
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.1045545939330363e-05,
+      "loss": 2.3342,
+      "step": 51140
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.1043360433604338e-05,
+      "loss": 2.3332,
+      "step": 51150
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.1041174927878311e-05,
+      "loss": 2.3458,
+      "step": 51160
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.1038989422152287e-05,
+      "loss": 2.3137,
+      "step": 51170
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.1036803916426263e-05,
+      "loss": 2.3219,
+      "step": 51180
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.1034618410700237e-05,
+      "loss": 2.2953,
+      "step": 51190
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.1032432904974213e-05,
+      "loss": 2.3442,
+      "step": 51200
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.1030247399248187e-05,
+      "loss": 2.3413,
+      "step": 51210
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.1028061893522161e-05,
+      "loss": 2.3538,
+      "step": 51220
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.1025876387796137e-05,
+      "loss": 2.3541,
+      "step": 51230
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.1023690882070113e-05,
+      "loss": 2.3053,
+      "step": 51240
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.1021505376344085e-05,
+      "loss": 2.3385,
+      "step": 51250
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.1019319870618061e-05,
+      "loss": 2.3557,
+      "step": 51260
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.1017134364892037e-05,
+      "loss": 2.3407,
+      "step": 51270
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.1014948859166013e-05,
+      "loss": 2.3741,
+      "step": 51280
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.1012763353439986e-05,
+      "loss": 2.3435,
+      "step": 51290
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.1010577847713961e-05,
+      "loss": 2.3589,
+      "step": 51300
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.1008392341987937e-05,
+      "loss": 2.3428,
+      "step": 51310
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.1006206836261913e-05,
+      "loss": 2.3063,
+      "step": 51320
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.1004021330535887e-05,
+      "loss": 2.3378,
+      "step": 51330
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.1001835824809862e-05,
+      "loss": 2.3899,
+      "step": 51340
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.0999650319083837e-05,
+      "loss": 2.3322,
+      "step": 51350
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.0997464813357812e-05,
+      "loss": 2.3127,
+      "step": 51360
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.0995279307631787e-05,
+      "loss": 2.335,
+      "step": 51370
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.0993093801905762e-05,
+      "loss": 2.2761,
+      "step": 51380
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.0990908296179736e-05,
+      "loss": 2.3032,
+      "step": 51390
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.0988722790453712e-05,
+      "loss": 2.362,
+      "step": 51400
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.0986537284727688e-05,
+      "loss": 2.346,
+      "step": 51410
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.0984351779001663e-05,
+      "loss": 2.3285,
+      "step": 51420
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.0982166273275636e-05,
+      "loss": 2.3859,
+      "step": 51430
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.0979980767549612e-05,
+      "loss": 2.3458,
+      "step": 51440
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.0977795261823588e-05,
+      "loss": 2.3807,
+      "step": 51450
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.0975609756097562e-05,
+      "loss": 2.3617,
+      "step": 51460
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.0973424250371536e-05,
+      "loss": 2.359,
+      "step": 51470
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.0971238744645512e-05,
+      "loss": 2.3454,
+      "step": 51480
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.0969053238919488e-05,
+      "loss": 2.3726,
+      "step": 51490
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.0966867733193462e-05,
+      "loss": 2.3204,
+      "step": 51500
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.0964682227467436e-05,
+      "loss": 2.3569,
+      "step": 51510
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.0962496721741412e-05,
+      "loss": 2.3478,
+      "step": 51520
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.0960311216015386e-05,
+      "loss": 2.3688,
+      "step": 51530
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.0958125710289362e-05,
+      "loss": 2.3343,
+      "step": 51540
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.0955940204563338e-05,
+      "loss": 2.353,
+      "step": 51550
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.095375469883731e-05,
+      "loss": 2.3455,
+      "step": 51560
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.0951569193111286e-05,
+      "loss": 2.3676,
+      "step": 51570
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.0949383687385262e-05,
+      "loss": 2.3309,
+      "step": 51580
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.0947198181659238e-05,
+      "loss": 2.3353,
+      "step": 51590
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.094501267593321e-05,
+      "loss": 2.3535,
+      "step": 51600
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.0942827170207187e-05,
+      "loss": 2.3358,
+      "step": 51610
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.0940641664481162e-05,
+      "loss": 2.332,
+      "step": 51620
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.0938456158755137e-05,
+      "loss": 2.3727,
+      "step": 51630
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.093627065302911e-05,
+      "loss": 2.3708,
+      "step": 51640
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.0934085147303087e-05,
+      "loss": 2.2742,
+      "step": 51650
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.0931899641577063e-05,
+      "loss": 2.3525,
+      "step": 51660
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.0929714135851037e-05,
+      "loss": 2.3474,
+      "step": 51670
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.0927528630125013e-05,
+      "loss": 2.3508,
+      "step": 51680
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.0925343124398987e-05,
+      "loss": 2.3426,
+      "step": 51690
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.0923157618672961e-05,
+      "loss": 2.3127,
+      "step": 51700
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.0920972112946937e-05,
+      "loss": 2.3515,
+      "step": 51710
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.0918786607220913e-05,
+      "loss": 2.3226,
+      "step": 51720
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.0916601101494885e-05,
+      "loss": 2.3827,
+      "step": 51730
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.0914415595768861e-05,
+      "loss": 2.3649,
+      "step": 51740
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.0912230090042837e-05,
+      "loss": 2.3168,
+      "step": 51750
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.0910044584316813e-05,
+      "loss": 2.349,
+      "step": 51760
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.0907859078590787e-05,
+      "loss": 2.3387,
+      "step": 51770
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.0905673572864761e-05,
+      "loss": 2.3456,
+      "step": 51780
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.0903488067138737e-05,
+      "loss": 2.3646,
+      "step": 51790
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.0901302561412713e-05,
+      "loss": 2.3469,
+      "step": 51800
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.0899117055686687e-05,
+      "loss": 2.3336,
+      "step": 51810
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.0896931549960661e-05,
+      "loss": 2.3827,
+      "step": 51820
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.0894746044234637e-05,
+      "loss": 2.3565,
+      "step": 51830
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.0892560538508611e-05,
+      "loss": 2.3773,
+      "step": 51840
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.0890375032782587e-05,
+      "loss": 2.3601,
+      "step": 51850
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.0888189527056561e-05,
+      "loss": 2.3536,
+      "step": 51860
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.0886004021330536e-05,
+      "loss": 2.3734,
+      "step": 51870
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.0883818515604512e-05,
+      "loss": 2.3611,
+      "step": 51880
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.0881633009878487e-05,
+      "loss": 2.3731,
+      "step": 51890
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.0879447504152463e-05,
+      "loss": 2.3673,
+      "step": 51900
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.0877261998426436e-05,
+      "loss": 2.3517,
+      "step": 51910
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.0875076492700412e-05,
+      "loss": 2.3746,
+      "step": 51920
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.0872890986974388e-05,
+      "loss": 2.3272,
+      "step": 51930
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.0870705481248362e-05,
+      "loss": 2.369,
+      "step": 51940
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.0868519975522336e-05,
+      "loss": 2.3339,
+      "step": 51950
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.0866334469796312e-05,
+      "loss": 2.3666,
+      "step": 51960
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.0864148964070288e-05,
+      "loss": 2.3851,
+      "step": 51970
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.0861963458344262e-05,
+      "loss": 2.3651,
+      "step": 51980
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.0859777952618236e-05,
+      "loss": 2.3228,
+      "step": 51990
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.0857592446892212e-05,
+      "loss": 2.363,
+      "step": 52000
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.0855406941166186e-05,
+      "loss": 2.3832,
+      "step": 52010
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.0853221435440162e-05,
+      "loss": 2.3443,
+      "step": 52020
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.0851035929714138e-05,
+      "loss": 2.3686,
+      "step": 52030
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.084885042398811e-05,
+      "loss": 2.2976,
+      "step": 52040
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.0846664918262086e-05,
+      "loss": 2.3398,
+      "step": 52050
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.0844479412536062e-05,
+      "loss": 2.3321,
+      "step": 52060
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.0842293906810038e-05,
+      "loss": 2.345,
+      "step": 52070
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.084010840108401e-05,
+      "loss": 2.3607,
+      "step": 52080
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.0837922895357986e-05,
+      "loss": 2.356,
+      "step": 52090
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.0835737389631962e-05,
+      "loss": 2.3903,
+      "step": 52100
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.0833551883905936e-05,
+      "loss": 2.3525,
+      "step": 52110
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.0831366378179912e-05,
+      "loss": 2.3581,
+      "step": 52120
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.0829180872453886e-05,
+      "loss": 2.3745,
+      "step": 52130
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.0826995366727862e-05,
+      "loss": 2.3661,
+      "step": 52140
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.0824809861001837e-05,
+      "loss": 2.3648,
+      "step": 52150
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.0822624355275812e-05,
+      "loss": 2.2754,
+      "step": 52160
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.0820438849549787e-05,
+      "loss": 2.3578,
+      "step": 52170
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.081825334382376e-05,
+      "loss": 2.3564,
+      "step": 52180
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.0816067838097737e-05,
+      "loss": 2.3403,
+      "step": 52190
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.0813882332371713e-05,
+      "loss": 2.3633,
+      "step": 52200
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.0811696826645685e-05,
+      "loss": 2.3269,
+      "step": 52210
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.0809511320919661e-05,
+      "loss": 2.317,
+      "step": 52220
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.0807325815193637e-05,
+      "loss": 2.3785,
+      "step": 52230
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.0805140309467613e-05,
+      "loss": 2.3536,
+      "step": 52240
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.0802954803741587e-05,
+      "loss": 2.3342,
+      "step": 52250
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.0800769298015561e-05,
+      "loss": 2.3878,
+      "step": 52260
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.0798583792289537e-05,
+      "loss": 2.3509,
+      "step": 52270
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.0796398286563511e-05,
+      "loss": 2.3907,
+      "step": 52280
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.0794212780837487e-05,
+      "loss": 2.3528,
+      "step": 52290
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.0792027275111461e-05,
+      "loss": 2.3882,
+      "step": 52300
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.0789841769385437e-05,
+      "loss": 2.3573,
+      "step": 52310
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.0787656263659411e-05,
+      "loss": 2.3505,
+      "step": 52320
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.0785470757933387e-05,
+      "loss": 2.3367,
+      "step": 52330
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.0783285252207363e-05,
+      "loss": 2.3246,
+      "step": 52340
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.0781099746481336e-05,
+      "loss": 2.3719,
+      "step": 52350
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.0778914240755311e-05,
+      "loss": 2.3331,
+      "step": 52360
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.0776728735029287e-05,
+      "loss": 2.3885,
+      "step": 52370
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.0774543229303263e-05,
+      "loss": 2.3789,
+      "step": 52380
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.0772357723577236e-05,
+      "loss": 2.3349,
+      "step": 52390
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.0770172217851212e-05,
+      "loss": 2.3632,
+      "step": 52400
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.0767986712125187e-05,
+      "loss": 2.3421,
+      "step": 52410
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.0765801206399162e-05,
+      "loss": 2.3543,
+      "step": 52420
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.0763615700673136e-05,
+      "loss": 2.388,
+      "step": 52430
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.0761430194947112e-05,
+      "loss": 2.3987,
+      "step": 52440
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.0759244689221088e-05,
+      "loss": 2.3644,
+      "step": 52450
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.0757059183495062e-05,
+      "loss": 2.3637,
+      "step": 52460
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.0754873677769038e-05,
+      "loss": 2.3176,
+      "step": 52470
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 2.3772,
+      "step": 52480
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.0750502666316986e-05,
+      "loss": 2.3526,
+      "step": 52490
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.0748317160590962e-05,
+      "loss": 2.3427,
+      "step": 52500
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.0746131654864938e-05,
+      "loss": 2.3342,
+      "step": 52510
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.074394614913891e-05,
+      "loss": 2.3685,
+      "step": 52520
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.0741760643412886e-05,
+      "loss": 2.3488,
+      "step": 52530
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.0739575137686862e-05,
+      "loss": 2.3166,
+      "step": 52540
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.0737389631960838e-05,
+      "loss": 2.3734,
+      "step": 52550
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.073520412623481e-05,
+      "loss": 2.3169,
+      "step": 52560
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.0733018620508786e-05,
+      "loss": 2.373,
+      "step": 52570
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.0730833114782762e-05,
+      "loss": 2.3759,
+      "step": 52580
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.0728647609056736e-05,
+      "loss": 2.3154,
+      "step": 52590
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.0726462103330712e-05,
+      "loss": 2.3565,
+      "step": 52600
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.0724276597604686e-05,
+      "loss": 2.3862,
+      "step": 52610
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.0722091091878662e-05,
+      "loss": 2.3872,
+      "step": 52620
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.0719905586152636e-05,
+      "loss": 2.3476,
+      "step": 52630
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.0717720080426612e-05,
+      "loss": 2.3911,
+      "step": 52640
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.0715534574700586e-05,
+      "loss": 2.3932,
+      "step": 52650
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.071334906897456e-05,
+      "loss": 2.3533,
+      "step": 52660
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.0711163563248537e-05,
+      "loss": 2.3659,
+      "step": 52670
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.0708978057522512e-05,
+      "loss": 2.3437,
+      "step": 52680
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.0706792551796488e-05,
+      "loss": 2.3624,
+      "step": 52690
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.070460704607046e-05,
+      "loss": 2.3914,
+      "step": 52700
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.0702421540344437e-05,
+      "loss": 2.3772,
+      "step": 52710
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.0700236034618413e-05,
+      "loss": 2.355,
+      "step": 52720
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.0698050528892387e-05,
+      "loss": 2.3965,
+      "step": 52730
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.0695865023166361e-05,
+      "loss": 2.3949,
+      "step": 52740
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.0693679517440337e-05,
+      "loss": 2.3771,
+      "step": 52750
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.0691494011714311e-05,
+      "loss": 2.3832,
+      "step": 52760
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.0689308505988287e-05,
+      "loss": 2.3349,
+      "step": 52770
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.0687123000262261e-05,
+      "loss": 2.3468,
+      "step": 52780
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.0684937494536237e-05,
+      "loss": 2.3782,
+      "step": 52790
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.0682751988810211e-05,
+      "loss": 2.3695,
+      "step": 52800
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.0680566483084187e-05,
+      "loss": 2.3869,
+      "step": 52810
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.0678380977358163e-05,
+      "loss": 2.4096,
+      "step": 52820
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.0676195471632135e-05,
+      "loss": 2.341,
+      "step": 52830
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.0674009965906111e-05,
+      "loss": 2.3255,
+      "step": 52840
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.0671824460180087e-05,
+      "loss": 2.3565,
+      "step": 52850
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.0669638954454063e-05,
+      "loss": 2.3698,
+      "step": 52860
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.0667453448728035e-05,
+      "loss": 2.3814,
+      "step": 52870
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.0665267943002011e-05,
+      "loss": 2.3569,
+      "step": 52880
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.0663082437275987e-05,
+      "loss": 2.3421,
+      "step": 52890
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.0660896931549961e-05,
+      "loss": 2.3946,
+      "step": 52900
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.0658711425823936e-05,
+      "loss": 2.3368,
+      "step": 52910
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.0656525920097911e-05,
+      "loss": 2.3626,
+      "step": 52920
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.0654340414371886e-05,
+      "loss": 2.3734,
+      "step": 52930
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.0652154908645862e-05,
+      "loss": 2.3724,
+      "step": 52940
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.0649969402919837e-05,
+      "loss": 2.3933,
+      "step": 52950
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.0647783897193812e-05,
+      "loss": 2.3827,
+      "step": 52960
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.0645598391467786e-05,
+      "loss": 2.3707,
+      "step": 52970
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.0643412885741762e-05,
+      "loss": 2.3589,
+      "step": 52980
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.0641227380015738e-05,
+      "loss": 2.3728,
+      "step": 52990
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.063904187428971e-05,
+      "loss": 2.3658,
+      "step": 53000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.0636856368563686e-05,
+      "loss": 2.3556,
+      "step": 53010
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.0634670862837662e-05,
+      "loss": 2.3479,
+      "step": 53020
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.0632485357111638e-05,
+      "loss": 2.3877,
+      "step": 53030
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.0630299851385612e-05,
+      "loss": 2.3785,
+      "step": 53040
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.0628114345659586e-05,
+      "loss": 2.3709,
+      "step": 53050
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.0625928839933562e-05,
+      "loss": 2.3846,
+      "step": 53060
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.0623743334207536e-05,
+      "loss": 2.3461,
+      "step": 53070
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.0621557828481512e-05,
+      "loss": 2.3652,
+      "step": 53080
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.0619372322755486e-05,
+      "loss": 2.4101,
+      "step": 53090
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.0617186817029462e-05,
+      "loss": 2.3669,
+      "step": 53100
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.0615001311303436e-05,
+      "loss": 2.3619,
+      "step": 53110
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.0612815805577412e-05,
+      "loss": 2.3953,
+      "step": 53120
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.0610630299851386e-05,
+      "loss": 2.3338,
+      "step": 53130
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.060844479412536e-05,
+      "loss": 2.3962,
+      "step": 53140
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.0606259288399336e-05,
+      "loss": 2.3744,
+      "step": 53150
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.0604073782673312e-05,
+      "loss": 2.3407,
+      "step": 53160
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.0601888276947288e-05,
+      "loss": 2.3687,
+      "step": 53170
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.059970277122126e-05,
+      "loss": 2.3901,
+      "step": 53180
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.0597517265495236e-05,
+      "loss": 2.3891,
+      "step": 53190
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.0595331759769212e-05,
+      "loss": 2.3367,
+      "step": 53200
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.0593146254043187e-05,
+      "loss": 2.3845,
+      "step": 53210
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.059096074831716e-05,
+      "loss": 2.3903,
+      "step": 53220
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.0588775242591137e-05,
+      "loss": 2.409,
+      "step": 53230
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.058658973686511e-05,
+      "loss": 2.3806,
+      "step": 53240
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.0584404231139087e-05,
+      "loss": 2.3787,
+      "step": 53250
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.058221872541306e-05,
+      "loss": 2.3454,
+      "step": 53260
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.0580033219687037e-05,
+      "loss": 2.3811,
+      "step": 53270
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.0577847713961011e-05,
+      "loss": 2.3693,
+      "step": 53280
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.0575662208234987e-05,
+      "loss": 2.3675,
+      "step": 53290
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.0573476702508963e-05,
+      "loss": 2.3588,
+      "step": 53300
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.0571291196782935e-05,
+      "loss": 2.3736,
+      "step": 53310
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.0569105691056911e-05,
+      "loss": 2.3412,
+      "step": 53320
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.0566920185330887e-05,
+      "loss": 2.3758,
+      "step": 53330
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.0564734679604863e-05,
+      "loss": 2.3476,
+      "step": 53340
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.0562549173878835e-05,
+      "loss": 2.326,
+      "step": 53350
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.0560363668152811e-05,
+      "loss": 2.3634,
+      "step": 53360
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.0558178162426787e-05,
+      "loss": 2.388,
+      "step": 53370
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.0555992656700761e-05,
+      "loss": 2.3835,
+      "step": 53380
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.0553807150974737e-05,
+      "loss": 2.3767,
+      "step": 53390
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.0551621645248711e-05,
+      "loss": 2.3752,
+      "step": 53400
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.0549436139522685e-05,
+      "loss": 2.4097,
+      "step": 53410
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.0547250633796661e-05,
+      "loss": 2.3465,
+      "step": 53420
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.0545065128070637e-05,
+      "loss": 2.369,
+      "step": 53430
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.0542879622344611e-05,
+      "loss": 2.3992,
+      "step": 53440
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.0540694116618586e-05,
+      "loss": 2.3779,
+      "step": 53450
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.0538508610892561e-05,
+      "loss": 2.3477,
+      "step": 53460
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.0536323105166537e-05,
+      "loss": 2.325,
+      "step": 53470
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.053413759944051e-05,
+      "loss": 2.3798,
+      "step": 53480
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.0531952093714486e-05,
+      "loss": 2.4024,
+      "step": 53490
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.0529766587988462e-05,
+      "loss": 2.3669,
+      "step": 53500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.0527581082262437e-05,
+      "loss": 2.3465,
+      "step": 53510
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.0525395576536412e-05,
+      "loss": 2.4095,
+      "step": 53520
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.0523210070810386e-05,
+      "loss": 2.3706,
+      "step": 53530
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.0521024565084362e-05,
+      "loss": 2.3386,
+      "step": 53540
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.0518839059358336e-05,
+      "loss": 2.3499,
+      "step": 53550
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.0516653553632312e-05,
+      "loss": 2.3719,
+      "step": 53560
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.0514468047906286e-05,
+      "loss": 2.3734,
+      "step": 53570
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.051228254218026e-05,
+      "loss": 2.4134,
+      "step": 53580
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.0510097036454236e-05,
+      "loss": 2.4188,
+      "step": 53590
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.0507911530728212e-05,
+      "loss": 2.3922,
+      "step": 53600
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.0505726025002186e-05,
+      "loss": 2.3521,
+      "step": 53610
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.050354051927616e-05,
+      "loss": 2.3268,
+      "step": 53620
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.0501355013550136e-05,
+      "loss": 2.3479,
+      "step": 53630
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.0499169507824112e-05,
+      "loss": 2.3988,
+      "step": 53640
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.0496984002098088e-05,
+      "loss": 2.3901,
+      "step": 53650
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.049479849637206e-05,
+      "loss": 2.371,
+      "step": 53660
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.0492612990646036e-05,
+      "loss": 2.3296,
+      "step": 53670
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.0490427484920012e-05,
+      "loss": 2.344,
+      "step": 53680
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.0488241979193986e-05,
+      "loss": 2.3754,
+      "step": 53690
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.048605647346796e-05,
+      "loss": 2.3942,
+      "step": 53700
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 2.3719,
+      "step": 53710
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.048168546201591e-05,
+      "loss": 2.4134,
+      "step": 53720
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.0479499956289886e-05,
+      "loss": 2.3626,
+      "step": 53730
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.0477314450563862e-05,
+      "loss": 2.3868,
+      "step": 53740
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.0475128944837837e-05,
+      "loss": 2.3668,
+      "step": 53750
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.047294343911181e-05,
+      "loss": 2.4059,
+      "step": 53760
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.0470757933385787e-05,
+      "loss": 2.3943,
+      "step": 53770
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.0468572427659762e-05,
+      "loss": 2.3944,
+      "step": 53780
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.0466386921933735e-05,
+      "loss": 2.393,
+      "step": 53790
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.0464201416207711e-05,
+      "loss": 2.3931,
+      "step": 53800
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.0462015910481687e-05,
+      "loss": 2.3694,
+      "step": 53810
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.0459830404755663e-05,
+      "loss": 2.3898,
+      "step": 53820
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.0457644899029635e-05,
+      "loss": 2.4304,
+      "step": 53830
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.0455459393303611e-05,
+      "loss": 2.4087,
+      "step": 53840
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.0453273887577587e-05,
+      "loss": 2.3982,
+      "step": 53850
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.0451088381851561e-05,
+      "loss": 2.3686,
+      "step": 53860
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.0448902876125537e-05,
+      "loss": 2.362,
+      "step": 53870
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.0446717370399511e-05,
+      "loss": 2.3878,
+      "step": 53880
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.0444531864673485e-05,
+      "loss": 2.3258,
+      "step": 53890
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.0442346358947461e-05,
+      "loss": 2.3561,
+      "step": 53900
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.0440160853221437e-05,
+      "loss": 2.3829,
+      "step": 53910
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.0437975347495411e-05,
+      "loss": 2.3419,
+      "step": 53920
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.0435789841769385e-05,
+      "loss": 2.3393,
+      "step": 53930
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.0433604336043361e-05,
+      "loss": 2.3813,
+      "step": 53940
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.0431418830317337e-05,
+      "loss": 2.3907,
+      "step": 53950
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.042923332459131e-05,
+      "loss": 2.3779,
+      "step": 53960
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.0427047818865286e-05,
+      "loss": 2.3678,
+      "step": 53970
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.0424862313139261e-05,
+      "loss": 2.4033,
+      "step": 53980
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.0422676807413237e-05,
+      "loss": 2.3585,
+      "step": 53990
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.0420491301687211e-05,
+      "loss": 2.3536,
+      "step": 54000
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.0418305795961186e-05,
+      "loss": 2.3469,
+      "step": 54010
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.0416120290235162e-05,
+      "loss": 2.3642,
+      "step": 54020
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.0413934784509136e-05,
+      "loss": 2.429,
+      "step": 54030
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.0411749278783112e-05,
+      "loss": 2.3755,
+      "step": 54040
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.0409563773057086e-05,
+      "loss": 2.3669,
+      "step": 54050
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.040737826733106e-05,
+      "loss": 2.362,
+      "step": 54060
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.0405192761605036e-05,
+      "loss": 2.3784,
+      "step": 54070
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.0403007255879012e-05,
+      "loss": 2.3487,
+      "step": 54080
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.0400821750152988e-05,
+      "loss": 2.3605,
+      "step": 54090
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.039863624442696e-05,
+      "loss": 2.4214,
+      "step": 54100
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.0396450738700936e-05,
+      "loss": 2.3726,
+      "step": 54110
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.0394265232974912e-05,
+      "loss": 2.3956,
+      "step": 54120
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.0392079727248888e-05,
+      "loss": 2.3415,
+      "step": 54130
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.038989422152286e-05,
+      "loss": 2.3521,
+      "step": 54140
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.0387708715796836e-05,
+      "loss": 2.3647,
+      "step": 54150
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.0385523210070812e-05,
+      "loss": 2.3375,
+      "step": 54160
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.0383337704344786e-05,
+      "loss": 2.3868,
+      "step": 54170
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.038115219861876e-05,
+      "loss": 2.4179,
+      "step": 54180
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.0378966692892736e-05,
+      "loss": 2.4035,
+      "step": 54190
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.037678118716671e-05,
+      "loss": 2.3229,
+      "step": 54200
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.0374595681440686e-05,
+      "loss": 2.3504,
+      "step": 54210
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.0372410175714662e-05,
+      "loss": 2.3483,
+      "step": 54220
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.0370224669988635e-05,
+      "loss": 2.3558,
+      "step": 54230
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.036803916426261e-05,
+      "loss": 2.3636,
+      "step": 54240
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.0365853658536586e-05,
+      "loss": 2.3384,
+      "step": 54250
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.0363668152810562e-05,
+      "loss": 2.386,
+      "step": 54260
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.0361482647084535e-05,
+      "loss": 2.3865,
+      "step": 54270
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.035929714135851e-05,
+      "loss": 2.3949,
+      "step": 54280
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.0357111635632487e-05,
+      "loss": 2.3786,
+      "step": 54290
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.0354926129906462e-05,
+      "loss": 2.3553,
+      "step": 54300
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.0352740624180435e-05,
+      "loss": 2.3321,
+      "step": 54310
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.035055511845441e-05,
+      "loss": 2.3868,
+      "step": 54320
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.0348369612728387e-05,
+      "loss": 2.3705,
+      "step": 54330
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.0346184107002361e-05,
+      "loss": 2.3948,
+      "step": 54340
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.0343998601276337e-05,
+      "loss": 2.3888,
+      "step": 54350
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.0341813095550311e-05,
+      "loss": 2.3941,
+      "step": 54360
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.0339627589824285e-05,
+      "loss": 2.3824,
+      "step": 54370
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.0337442084098261e-05,
+      "loss": 2.3653,
+      "step": 54380
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.0335256578372237e-05,
+      "loss": 2.378,
+      "step": 54390
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.0333071072646211e-05,
+      "loss": 2.3861,
+      "step": 54400
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.0330885566920185e-05,
+      "loss": 2.374,
+      "step": 54410
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.0328700061194161e-05,
+      "loss": 2.3855,
+      "step": 54420
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.0326514555468137e-05,
+      "loss": 2.3704,
+      "step": 54430
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.0324329049742113e-05,
+      "loss": 2.3686,
+      "step": 54440
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.0322143544016085e-05,
+      "loss": 2.4152,
+      "step": 54450
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.0319958038290061e-05,
+      "loss": 2.3886,
+      "step": 54460
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.0317772532564037e-05,
+      "loss": 2.3688,
+      "step": 54470
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.0315587026838011e-05,
+      "loss": 2.3766,
+      "step": 54480
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.0313401521111986e-05,
+      "loss": 2.3719,
+      "step": 54490
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.0311216015385961e-05,
+      "loss": 2.371,
+      "step": 54500
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.0309030509659936e-05,
+      "loss": 2.3902,
+      "step": 54510
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.0306845003933911e-05,
+      "loss": 2.3605,
+      "step": 54520
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.0304659498207886e-05,
+      "loss": 2.3922,
+      "step": 54530
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.030247399248186e-05,
+      "loss": 2.3613,
+      "step": 54540
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.0300288486755836e-05,
+      "loss": 2.389,
+      "step": 54550
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.0298102981029812e-05,
+      "loss": 2.3918,
+      "step": 54560
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.0295917475303787e-05,
+      "loss": 2.3496,
+      "step": 54570
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.029373196957776e-05,
+      "loss": 2.3842,
+      "step": 54580
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.0291546463851736e-05,
+      "loss": 2.4314,
+      "step": 54590
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.0289360958125712e-05,
+      "loss": 2.3768,
+      "step": 54600
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.0287175452399688e-05,
+      "loss": 2.3364,
+      "step": 54610
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.028498994667366e-05,
+      "loss": 2.3597,
+      "step": 54620
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.0282804440947636e-05,
+      "loss": 2.3848,
+      "step": 54630
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.0280618935221612e-05,
+      "loss": 2.3923,
+      "step": 54640
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.0278433429495586e-05,
+      "loss": 2.4084,
+      "step": 54650
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.027624792376956e-05,
+      "loss": 2.3838,
+      "step": 54660
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.0274062418043536e-05,
+      "loss": 2.3758,
+      "step": 54670
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.027187691231751e-05,
+      "loss": 2.3733,
+      "step": 54680
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.0269691406591486e-05,
+      "loss": 2.3692,
+      "step": 54690
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.0267505900865462e-05,
+      "loss": 2.3617,
+      "step": 54700
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.0265320395139435e-05,
+      "loss": 2.3686,
+      "step": 54710
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.026313488941341e-05,
+      "loss": 2.43,
+      "step": 54720
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.0260949383687386e-05,
+      "loss": 2.3551,
+      "step": 54730
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.0258763877961362e-05,
+      "loss": 2.3885,
+      "step": 54740
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.0256578372235335e-05,
+      "loss": 2.3701,
+      "step": 54750
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.025439286650931e-05,
+      "loss": 2.3619,
+      "step": 54760
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.0252207360783286e-05,
+      "loss": 2.4078,
+      "step": 54770
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.0250021855057262e-05,
+      "loss": 2.3737,
+      "step": 54780
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.0247836349331236e-05,
+      "loss": 2.4066,
+      "step": 54790
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.024565084360521e-05,
+      "loss": 2.3743,
+      "step": 54800
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.0243465337879187e-05,
+      "loss": 2.4094,
+      "step": 54810
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.024127983215316e-05,
+      "loss": 2.3467,
+      "step": 54820
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.0239094326427137e-05,
+      "loss": 2.3973,
+      "step": 54830
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.023690882070111e-05,
+      "loss": 2.3949,
+      "step": 54840
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.0234723314975085e-05,
+      "loss": 2.3508,
+      "step": 54850
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.023253780924906e-05,
+      "loss": 2.351,
+      "step": 54860
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.0230352303523037e-05,
+      "loss": 2.3832,
+      "step": 54870
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.022816679779701e-05,
+      "loss": 2.4085,
+      "step": 54880
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.0225981292070985e-05,
+      "loss": 2.3704,
+      "step": 54890
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.0223795786344961e-05,
+      "loss": 2.4019,
+      "step": 54900
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.0221610280618937e-05,
+      "loss": 2.3778,
+      "step": 54910
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.0219424774892913e-05,
+      "loss": 2.3798,
+      "step": 54920
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.0217239269166885e-05,
+      "loss": 2.3764,
+      "step": 54930
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.0215053763440861e-05,
+      "loss": 2.3937,
+      "step": 54940
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.0212868257714837e-05,
+      "loss": 2.3745,
+      "step": 54950
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.0210682751988811e-05,
+      "loss": 2.3776,
+      "step": 54960
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.0208497246262785e-05,
+      "loss": 2.37,
+      "step": 54970
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.0206311740536761e-05,
+      "loss": 2.4195,
+      "step": 54980
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.0204126234810735e-05,
+      "loss": 2.386,
+      "step": 54990
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.0201940729084711e-05,
+      "loss": 2.3975,
+      "step": 55000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.0199755223358685e-05,
+      "loss": 2.3957,
+      "step": 55010
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.019756971763266e-05,
+      "loss": 2.3473,
+      "step": 55020
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.0195384211906636e-05,
+      "loss": 2.3888,
+      "step": 55030
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.0193198706180611e-05,
+      "loss": 2.3545,
+      "step": 55040
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.0191013200454587e-05,
+      "loss": 2.3649,
+      "step": 55050
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.018882769472856e-05,
+      "loss": 2.365,
+      "step": 55060
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.0186642189002536e-05,
+      "loss": 2.3803,
+      "step": 55070
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.0184456683276512e-05,
+      "loss": 2.3567,
+      "step": 55080
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.0182271177550487e-05,
+      "loss": 2.3893,
+      "step": 55090
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.018008567182446e-05,
+      "loss": 2.3679,
+      "step": 55100
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.0177900166098436e-05,
+      "loss": 2.3597,
+      "step": 55110
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.0175714660372412e-05,
+      "loss": 2.4045,
+      "step": 55120
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.0173529154646386e-05,
+      "loss": 2.4232,
+      "step": 55130
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.0171343648920362e-05,
+      "loss": 2.4119,
+      "step": 55140
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.0169158143194336e-05,
+      "loss": 2.4062,
+      "step": 55150
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.016697263746831e-05,
+      "loss": 2.4175,
+      "step": 55160
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.0164787131742286e-05,
+      "loss": 2.3678,
+      "step": 55170
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.0162601626016262e-05,
+      "loss": 2.3691,
+      "step": 55180
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.0160416120290234e-05,
+      "loss": 2.3534,
+      "step": 55190
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.015823061456421e-05,
+      "loss": 2.3663,
+      "step": 55200
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.0156045108838186e-05,
+      "loss": 2.3916,
+      "step": 55210
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.0153859603112162e-05,
+      "loss": 2.3674,
+      "step": 55220
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.0151674097386134e-05,
+      "loss": 2.3826,
+      "step": 55230
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.014948859166011e-05,
+      "loss": 2.3619,
+      "step": 55240
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.0147303085934086e-05,
+      "loss": 2.3554,
+      "step": 55250
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.0145117580208062e-05,
+      "loss": 2.3656,
+      "step": 55260
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.0142932074482036e-05,
+      "loss": 2.351,
+      "step": 55270
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.014074656875601e-05,
+      "loss": 2.3715,
+      "step": 55280
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.0138561063029986e-05,
+      "loss": 2.3685,
+      "step": 55290
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.013637555730396e-05,
+      "loss": 2.4164,
+      "step": 55300
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.0134190051577936e-05,
+      "loss": 2.3969,
+      "step": 55310
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.013200454585191e-05,
+      "loss": 2.4069,
+      "step": 55320
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.0129819040125885e-05,
+      "loss": 2.3707,
+      "step": 55330
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.012763353439986e-05,
+      "loss": 2.3941,
+      "step": 55340
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.0125448028673837e-05,
+      "loss": 2.3591,
+      "step": 55350
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.0123262522947809e-05,
+      "loss": 2.3772,
+      "step": 55360
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.0121077017221785e-05,
+      "loss": 2.3774,
+      "step": 55370
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.011889151149576e-05,
+      "loss": 2.4247,
+      "step": 55380
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.0116706005769737e-05,
+      "loss": 2.3789,
+      "step": 55390
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.0114520500043713e-05,
+      "loss": 2.3599,
+      "step": 55400
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.0112334994317685e-05,
+      "loss": 2.3925,
+      "step": 55410
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.0110149488591661e-05,
+      "loss": 2.3701,
+      "step": 55420
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.0107963982865637e-05,
+      "loss": 2.3917,
+      "step": 55430
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.0105778477139611e-05,
+      "loss": 2.3776,
+      "step": 55440
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.0103592971413585e-05,
+      "loss": 2.3945,
+      "step": 55450
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.0101407465687561e-05,
+      "loss": 2.3718,
+      "step": 55460
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.0099221959961535e-05,
+      "loss": 2.3679,
+      "step": 55470
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.0097036454235511e-05,
+      "loss": 2.3343,
+      "step": 55480
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.0094850948509487e-05,
+      "loss": 2.3689,
+      "step": 55490
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.009266544278346e-05,
+      "loss": 2.3532,
+      "step": 55500
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.0090479937057435e-05,
+      "loss": 2.3699,
+      "step": 55510
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.0088294431331411e-05,
+      "loss": 2.37,
+      "step": 55520
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.0086108925605387e-05,
+      "loss": 2.4252,
+      "step": 55530
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.008392341987936e-05,
+      "loss": 2.3945,
+      "step": 55540
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.0081737914153335e-05,
+      "loss": 2.3972,
+      "step": 55550
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.0079552408427311e-05,
+      "loss": 2.3617,
+      "step": 55560
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.0077366902701287e-05,
+      "loss": 2.3598,
+      "step": 55570
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.007518139697526e-05,
+      "loss": 2.3236,
+      "step": 55580
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.0072995891249236e-05,
+      "loss": 2.403,
+      "step": 55590
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.0070810385523211e-05,
+      "loss": 2.3907,
+      "step": 55600
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.0068624879797186e-05,
+      "loss": 2.414,
+      "step": 55610
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.0066439374071162e-05,
+      "loss": 2.3358,
+      "step": 55620
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.0064253868345136e-05,
+      "loss": 2.384,
+      "step": 55630
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.006206836261911e-05,
+      "loss": 2.4157,
+      "step": 55640
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.0059882856893086e-05,
+      "loss": 2.4128,
+      "step": 55650
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.0057697351167062e-05,
+      "loss": 2.3963,
+      "step": 55660
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.0055511845441034e-05,
+      "loss": 2.3954,
+      "step": 55670
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.005332633971501e-05,
+      "loss": 2.3565,
+      "step": 55680
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.0051140833988986e-05,
+      "loss": 2.3602,
+      "step": 55690
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.0048955328262962e-05,
+      "loss": 2.3431,
+      "step": 55700
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.0046769822536934e-05,
+      "loss": 2.3663,
+      "step": 55710
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.004458431681091e-05,
+      "loss": 2.402,
+      "step": 55720
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.0042398811084886e-05,
+      "loss": 2.4359,
+      "step": 55730
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.0040213305358862e-05,
+      "loss": 2.3856,
+      "step": 55740
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.0038027799632836e-05,
+      "loss": 2.4025,
+      "step": 55750
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.003584229390681e-05,
+      "loss": 2.3938,
+      "step": 55760
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.0033656788180786e-05,
+      "loss": 2.3579,
+      "step": 55770
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.003147128245476e-05,
+      "loss": 2.3865,
+      "step": 55780
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.0029285776728736e-05,
+      "loss": 2.4001,
+      "step": 55790
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.002710027100271e-05,
+      "loss": 2.3618,
+      "step": 55800
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.0024914765276685e-05,
+      "loss": 2.3575,
+      "step": 55810
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.002272925955066e-05,
+      "loss": 2.4174,
+      "step": 55820
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.0020543753824636e-05,
+      "loss": 2.3767,
+      "step": 55830
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.0018358248098612e-05,
+      "loss": 2.3773,
+      "step": 55840
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.0016172742372585e-05,
+      "loss": 2.404,
+      "step": 55850
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.001398723664656e-05,
+      "loss": 2.3711,
+      "step": 55860
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.0011801730920536e-05,
+      "loss": 2.3973,
+      "step": 55870
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.0009616225194512e-05,
+      "loss": 2.3736,
+      "step": 55880
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.0007430719468485e-05,
+      "loss": 2.4031,
+      "step": 55890
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.000524521374246e-05,
+      "loss": 2.4345,
+      "step": 55900
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0003059708016437e-05,
+      "loss": 2.3738,
+      "step": 55910
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.000087420229041e-05,
+      "loss": 2.4076,
+      "step": 55920
+    },
+    {
+      "epoch": 11.0,
+      "eval_BLEU": 9.6223,
+      "eval_gen_len": 12.1131,
+      "eval_loss": 3.148531198501587,
+      "eval_runtime": 151.7671,
+      "eval_samples_per_second": 70.918,
+      "eval_steps_per_second": 0.283,
+      "step": 55924
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 9.998688696564387e-06,
+      "loss": 2.2791,
+      "step": 55930
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 9.996503190838361e-06,
+      "loss": 2.2488,
+      "step": 55940
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.994317685112335e-06,
+      "loss": 2.3049,
+      "step": 55950
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.992132179386311e-06,
+      "loss": 2.2827,
+      "step": 55960
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.989946673660285e-06,
+      "loss": 2.2865,
+      "step": 55970
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.987761167934261e-06,
+      "loss": 2.2886,
+      "step": 55980
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.985575662208235e-06,
+      "loss": 2.2564,
+      "step": 55990
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.983390156482211e-06,
+      "loss": 2.3044,
+      "step": 56000
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.981204650756185e-06,
+      "loss": 2.2495,
+      "step": 56010
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.979019145030161e-06,
+      "loss": 2.2758,
+      "step": 56020
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.976833639304135e-06,
+      "loss": 2.2627,
+      "step": 56030
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.974648133578111e-06,
+      "loss": 2.2584,
+      "step": 56040
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.972462627852085e-06,
+      "loss": 2.2796,
+      "step": 56050
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.970277122126061e-06,
+      "loss": 2.2542,
+      "step": 56060
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.968091616400037e-06,
+      "loss": 2.261,
+      "step": 56070
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.965906110674011e-06,
+      "loss": 2.329,
+      "step": 56080
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.963720604947985e-06,
+      "loss": 2.3204,
+      "step": 56090
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.961535099221961e-06,
+      "loss": 2.2495,
+      "step": 56100
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.959349593495936e-06,
+      "loss": 2.2583,
+      "step": 56110
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.95716408776991e-06,
+      "loss": 2.2675,
+      "step": 56120
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.954978582043886e-06,
+      "loss": 2.2932,
+      "step": 56130
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.95279307631786e-06,
+      "loss": 2.2972,
+      "step": 56140
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.950607570591836e-06,
+      "loss": 2.2441,
+      "step": 56150
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 9.94842206486581e-06,
+      "loss": 2.3104,
+      "step": 56160
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 9.946236559139786e-06,
+      "loss": 2.3001,
+      "step": 56170
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 9.94405105341376e-06,
+      "loss": 2.2543,
+      "step": 56180
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 9.941865547687736e-06,
+      "loss": 2.3297,
+      "step": 56190
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 9.939680041961712e-06,
+      "loss": 2.2763,
+      "step": 56200
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 9.937494536235686e-06,
+      "loss": 2.255,
+      "step": 56210
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 9.935309030509662e-06,
+      "loss": 2.3196,
+      "step": 56220
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 9.933123524783636e-06,
+      "loss": 2.2702,
+      "step": 56230
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 9.930938019057612e-06,
+      "loss": 2.2481,
+      "step": 56240
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 9.928752513331586e-06,
+      "loss": 2.3006,
+      "step": 56250
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 9.92656700760556e-06,
+      "loss": 2.2741,
+      "step": 56260
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 9.924381501879536e-06,
+      "loss": 2.2955,
+      "step": 56270
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 9.92219599615351e-06,
+      "loss": 2.2867,
+      "step": 56280
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 9.920010490427484e-06,
+      "loss": 2.2914,
+      "step": 56290
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 9.91782498470146e-06,
+      "loss": 2.3044,
+      "step": 56300
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 9.915639478975436e-06,
+      "loss": 2.2856,
+      "step": 56310
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 9.91345397324941e-06,
+      "loss": 2.2618,
+      "step": 56320
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 9.911268467523386e-06,
+      "loss": 2.2759,
+      "step": 56330
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 9.90908296179736e-06,
+      "loss": 2.2742,
+      "step": 56340
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 9.906897456071336e-06,
+      "loss": 2.2825,
+      "step": 56350
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 9.90471195034531e-06,
+      "loss": 2.2763,
+      "step": 56360
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 9.902526444619286e-06,
+      "loss": 2.2731,
+      "step": 56370
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 9.90034093889326e-06,
+      "loss": 2.283,
+      "step": 56380
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 9.898155433167236e-06,
+      "loss": 2.231,
+      "step": 56390
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 9.89596992744121e-06,
+      "loss": 2.299,
+      "step": 56400
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 9.893784421715187e-06,
+      "loss": 2.2745,
+      "step": 56410
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 9.89159891598916e-06,
+      "loss": 2.2725,
+      "step": 56420
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 9.889413410263135e-06,
+      "loss": 2.3286,
+      "step": 56430
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 9.88722790453711e-06,
+      "loss": 2.3102,
+      "step": 56440
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 9.885042398811085e-06,
+      "loss": 2.2856,
+      "step": 56450
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 9.88285689308506e-06,
+      "loss": 2.315,
+      "step": 56460
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 9.880671387359035e-06,
+      "loss": 2.2627,
+      "step": 56470
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 9.878485881633011e-06,
+      "loss": 2.2745,
+      "step": 56480
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 9.876300375906985e-06,
+      "loss": 2.2944,
+      "step": 56490
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 9.874114870180961e-06,
+      "loss": 2.3095,
+      "step": 56500
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 9.871929364454935e-06,
+      "loss": 2.3187,
+      "step": 56510
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 9.869743858728911e-06,
+      "loss": 2.2992,
+      "step": 56520
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 9.867558353002887e-06,
+      "loss": 2.2898,
+      "step": 56530
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 9.865372847276861e-06,
+      "loss": 2.2914,
+      "step": 56540
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 9.863187341550837e-06,
+      "loss": 2.2733,
+      "step": 56550
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 9.861001835824811e-06,
+      "loss": 2.2704,
+      "step": 56560
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 9.858816330098785e-06,
+      "loss": 2.2983,
+      "step": 56570
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 9.856630824372761e-06,
+      "loss": 2.3032,
+      "step": 56580
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 9.854445318646735e-06,
+      "loss": 2.2651,
+      "step": 56590
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 9.85225981292071e-06,
+      "loss": 2.2697,
+      "step": 56600
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 9.850074307194685e-06,
+      "loss": 2.3033,
+      "step": 56610
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 9.84788880146866e-06,
+      "loss": 2.2845,
+      "step": 56620
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 9.845703295742636e-06,
+      "loss": 2.2939,
+      "step": 56630
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 9.84351779001661e-06,
+      "loss": 2.2849,
+      "step": 56640
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 9.841332284290586e-06,
+      "loss": 2.3172,
+      "step": 56650
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 9.839146778564561e-06,
+      "loss": 2.2906,
+      "step": 56660
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 9.836961272838536e-06,
+      "loss": 2.3067,
+      "step": 56670
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 9.834775767112512e-06,
+      "loss": 2.2743,
+      "step": 56680
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 9.832590261386486e-06,
+      "loss": 2.3025,
+      "step": 56690
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 9.830404755660462e-06,
+      "loss": 2.2631,
+      "step": 56700
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 9.828219249934436e-06,
+      "loss": 2.2771,
+      "step": 56710
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 9.826033744208412e-06,
+      "loss": 2.2831,
+      "step": 56720
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 9.823848238482386e-06,
+      "loss": 2.3133,
+      "step": 56730
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 9.82166273275636e-06,
+      "loss": 2.3012,
+      "step": 56740
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 9.819477227030336e-06,
+      "loss": 2.2953,
+      "step": 56750
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 9.81729172130431e-06,
+      "loss": 2.2783,
+      "step": 56760
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 9.815106215578286e-06,
+      "loss": 2.3116,
+      "step": 56770
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 9.81292070985226e-06,
+      "loss": 2.3048,
+      "step": 56780
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 9.810735204126236e-06,
+      "loss": 2.2813,
+      "step": 56790
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 9.80854969840021e-06,
+      "loss": 2.2969,
+      "step": 56800
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 9.806364192674186e-06,
+      "loss": 2.3016,
+      "step": 56810
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 9.80417868694816e-06,
+      "loss": 2.3311,
+      "step": 56820
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 9.801993181222136e-06,
+      "loss": 2.296,
+      "step": 56830
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 9.79980767549611e-06,
+      "loss": 2.2704,
+      "step": 56840
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 9.797622169770086e-06,
+      "loss": 2.2819,
+      "step": 56850
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 9.79543666404406e-06,
+      "loss": 2.2854,
+      "step": 56860
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 9.793251158318036e-06,
+      "loss": 2.319,
+      "step": 56870
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 9.79106565259201e-06,
+      "loss": 2.2881,
+      "step": 56880
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 9.788880146865986e-06,
+      "loss": 2.2651,
+      "step": 56890
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 9.78669464113996e-06,
+      "loss": 2.299,
+      "step": 56900
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 9.784509135413935e-06,
+      "loss": 2.3155,
+      "step": 56910
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 9.78232362968791e-06,
+      "loss": 2.3111,
+      "step": 56920
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 9.780138123961885e-06,
+      "loss": 2.2908,
+      "step": 56930
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 9.77795261823586e-06,
+      "loss": 2.3132,
+      "step": 56940
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 9.775767112509835e-06,
+      "loss": 2.3192,
+      "step": 56950
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 9.77358160678381e-06,
+      "loss": 2.2685,
+      "step": 56960
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 9.771396101057785e-06,
+      "loss": 2.2852,
+      "step": 56970
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 9.76921059533176e-06,
+      "loss": 2.2788,
+      "step": 56980
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 9.767025089605735e-06,
+      "loss": 2.3161,
+      "step": 56990
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 9.76483958387971e-06,
+      "loss": 2.2986,
+      "step": 57000
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 9.762654078153687e-06,
+      "loss": 2.3182,
+      "step": 57010
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 9.760468572427661e-06,
+      "loss": 2.3013,
+      "step": 57020
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 9.758283066701637e-06,
+      "loss": 2.3077,
+      "step": 57030
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 2.282,
+      "step": 57040
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 9.753912055249585e-06,
+      "loss": 2.2914,
+      "step": 57050
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 9.751726549523561e-06,
+      "loss": 2.2815,
+      "step": 57060
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 9.749541043797535e-06,
+      "loss": 2.2935,
+      "step": 57070
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 9.74735553807151e-06,
+      "loss": 2.3066,
+      "step": 57080
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 9.745170032345485e-06,
+      "loss": 2.2728,
+      "step": 57090
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 9.74298452661946e-06,
+      "loss": 2.3364,
+      "step": 57100
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 9.740799020893435e-06,
+      "loss": 2.2924,
+      "step": 57110
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 9.738613515167411e-06,
+      "loss": 2.3075,
+      "step": 57120
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 9.736428009441385e-06,
+      "loss": 2.2857,
+      "step": 57130
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 9.734242503715361e-06,
+      "loss": 2.3112,
+      "step": 57140
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 9.732056997989335e-06,
+      "loss": 2.3397,
+      "step": 57150
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 9.729871492263311e-06,
+      "loss": 2.2978,
+      "step": 57160
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 9.727685986537286e-06,
+      "loss": 2.3066,
+      "step": 57170
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 9.725500480811261e-06,
+      "loss": 2.2907,
+      "step": 57180
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 9.723314975085236e-06,
+      "loss": 2.2821,
+      "step": 57190
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 9.721129469359211e-06,
+      "loss": 2.331,
+      "step": 57200
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 9.718943963633186e-06,
+      "loss": 2.2955,
+      "step": 57210
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 9.71675845790716e-06,
+      "loss": 2.3217,
+      "step": 57220
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 9.714572952181136e-06,
+      "loss": 2.3,
+      "step": 57230
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 9.71238744645511e-06,
+      "loss": 2.2992,
+      "step": 57240
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 9.710201940729086e-06,
+      "loss": 2.3061,
+      "step": 57250
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 9.70801643500306e-06,
+      "loss": 2.3201,
+      "step": 57260
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 9.705830929277036e-06,
+      "loss": 2.3363,
+      "step": 57270
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 9.70364542355101e-06,
+      "loss": 2.2773,
+      "step": 57280
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 9.701459917824986e-06,
+      "loss": 2.3034,
+      "step": 57290
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 9.69927441209896e-06,
+      "loss": 2.2959,
+      "step": 57300
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 9.697088906372936e-06,
+      "loss": 2.2687,
+      "step": 57310
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 9.69490340064691e-06,
+      "loss": 2.2473,
+      "step": 57320
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 9.692717894920886e-06,
+      "loss": 2.3201,
+      "step": 57330
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 9.69053238919486e-06,
+      "loss": 2.2596,
+      "step": 57340
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 9.688346883468836e-06,
+      "loss": 2.3032,
+      "step": 57350
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 9.68616137774281e-06,
+      "loss": 2.3132,
+      "step": 57360
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 9.683975872016786e-06,
+      "loss": 2.325,
+      "step": 57370
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 9.68179036629076e-06,
+      "loss": 2.2975,
+      "step": 57380
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 9.679604860564735e-06,
+      "loss": 2.3182,
+      "step": 57390
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 2.354,
+      "step": 57400
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 9.675233849112685e-06,
+      "loss": 2.276,
+      "step": 57410
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 9.67304834338666e-06,
+      "loss": 2.3425,
+      "step": 57420
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 9.670862837660635e-06,
+      "loss": 2.3222,
+      "step": 57430
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 9.66867733193461e-06,
+      "loss": 2.3229,
+      "step": 57440
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 9.666491826208585e-06,
+      "loss": 2.3047,
+      "step": 57450
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 9.66430632048256e-06,
+      "loss": 2.2981,
+      "step": 57460
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 9.662120814756536e-06,
+      "loss": 2.2792,
+      "step": 57470
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 9.65993530903051e-06,
+      "loss": 2.2983,
+      "step": 57480
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 9.657749803304487e-06,
+      "loss": 2.2935,
+      "step": 57490
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 9.65556429757846e-06,
+      "loss": 2.3202,
+      "step": 57500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 9.653378791852435e-06,
+      "loss": 2.3396,
+      "step": 57510
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 9.65119328612641e-06,
+      "loss": 2.3338,
+      "step": 57520
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 9.649007780400385e-06,
+      "loss": 2.3213,
+      "step": 57530
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 9.646822274674361e-06,
+      "loss": 2.3269,
+      "step": 57540
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 9.644636768948335e-06,
+      "loss": 2.2964,
+      "step": 57550
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 9.64245126322231e-06,
+      "loss": 2.3247,
+      "step": 57560
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 9.640265757496285e-06,
+      "loss": 2.2922,
+      "step": 57570
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 9.638080251770261e-06,
+      "loss": 2.2906,
+      "step": 57580
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 9.635894746044235e-06,
+      "loss": 2.2759,
+      "step": 57590
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 9.633709240318211e-06,
+      "loss": 2.2917,
+      "step": 57600
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 9.631523734592185e-06,
+      "loss": 2.2989,
+      "step": 57610
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 9.629338228866161e-06,
+      "loss": 2.3106,
+      "step": 57620
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 9.627152723140135e-06,
+      "loss": 2.3134,
+      "step": 57630
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 9.624967217414111e-06,
+      "loss": 2.3042,
+      "step": 57640
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 9.622781711688085e-06,
+      "loss": 2.2809,
+      "step": 57650
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 9.620596205962061e-06,
+      "loss": 2.2867,
+      "step": 57660
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 9.618410700236035e-06,
+      "loss": 2.3191,
+      "step": 57670
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 9.616225194510011e-06,
+      "loss": 2.3289,
+      "step": 57680
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 9.614039688783985e-06,
+      "loss": 2.3422,
+      "step": 57690
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 9.61185418305796e-06,
+      "loss": 2.3236,
+      "step": 57700
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 9.609668677331936e-06,
+      "loss": 2.3326,
+      "step": 57710
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 9.60748317160591e-06,
+      "loss": 2.277,
+      "step": 57720
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 9.605297665879886e-06,
+      "loss": 2.2737,
+      "step": 57730
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 9.60311216015386e-06,
+      "loss": 2.315,
+      "step": 57740
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 9.600926654427836e-06,
+      "loss": 2.3411,
+      "step": 57750
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 9.59874114870181e-06,
+      "loss": 2.3241,
+      "step": 57760
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 9.596555642975786e-06,
+      "loss": 2.3692,
+      "step": 57770
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 9.59437013724976e-06,
+      "loss": 2.3232,
+      "step": 57780
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 9.592184631523736e-06,
+      "loss": 2.2825,
+      "step": 57790
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 9.58999912579771e-06,
+      "loss": 2.266,
+      "step": 57800
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 9.587813620071686e-06,
+      "loss": 2.3605,
+      "step": 57810
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 9.58562811434566e-06,
+      "loss": 2.3215,
+      "step": 57820
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 9.583442608619636e-06,
+      "loss": 2.3167,
+      "step": 57830
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 9.58125710289361e-06,
+      "loss": 2.2842,
+      "step": 57840
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 9.579071597167586e-06,
+      "loss": 2.3048,
+      "step": 57850
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 9.57688609144156e-06,
+      "loss": 2.2609,
+      "step": 57860
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 9.574700585715534e-06,
+      "loss": 2.2917,
+      "step": 57870
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 9.57251507998951e-06,
+      "loss": 2.3161,
+      "step": 57880
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 9.570329574263484e-06,
+      "loss": 2.3083,
+      "step": 57890
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 9.56814406853746e-06,
+      "loss": 2.3211,
+      "step": 57900
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 9.565958562811434e-06,
+      "loss": 2.391,
+      "step": 57910
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 9.56377305708541e-06,
+      "loss": 2.3005,
+      "step": 57920
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 9.561587551359386e-06,
+      "loss": 2.2862,
+      "step": 57930
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 9.55940204563336e-06,
+      "loss": 2.3257,
+      "step": 57940
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 9.557216539907336e-06,
+      "loss": 2.3154,
+      "step": 57950
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 9.55503103418131e-06,
+      "loss": 2.3085,
+      "step": 57960
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 9.552845528455286e-06,
+      "loss": 2.3246,
+      "step": 57970
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 9.55066002272926e-06,
+      "loss": 2.2782,
+      "step": 57980
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 9.548474517003235e-06,
+      "loss": 2.3267,
+      "step": 57990
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 9.54628901127721e-06,
+      "loss": 2.3301,
+      "step": 58000
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 9.544103505551185e-06,
+      "loss": 2.298,
+      "step": 58010
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 9.54191799982516e-06,
+      "loss": 2.322,
+      "step": 58020
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 9.539732494099135e-06,
+      "loss": 2.2983,
+      "step": 58030
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 9.537546988373109e-06,
+      "loss": 2.2842,
+      "step": 58040
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 9.535361482647085e-06,
+      "loss": 2.2755,
+      "step": 58050
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 9.53317597692106e-06,
+      "loss": 2.3083,
+      "step": 58060
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 9.530990471195035e-06,
+      "loss": 2.3387,
+      "step": 58070
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 9.528804965469011e-06,
+      "loss": 2.3208,
+      "step": 58080
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 9.526619459742985e-06,
+      "loss": 2.3221,
+      "step": 58090
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 9.524433954016961e-06,
+      "loss": 2.3028,
+      "step": 58100
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 9.522248448290935e-06,
+      "loss": 2.2892,
+      "step": 58110
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 9.520062942564911e-06,
+      "loss": 2.3639,
+      "step": 58120
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 9.517877436838885e-06,
+      "loss": 2.2762,
+      "step": 58130
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 9.515691931112861e-06,
+      "loss": 2.3225,
+      "step": 58140
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 9.513506425386835e-06,
+      "loss": 2.2802,
+      "step": 58150
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 9.51132091966081e-06,
+      "loss": 2.3103,
+      "step": 58160
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 9.509135413934785e-06,
+      "loss": 2.3177,
+      "step": 58170
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 9.50694990820876e-06,
+      "loss": 2.2894,
+      "step": 58180
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 9.504764402482735e-06,
+      "loss": 2.3216,
+      "step": 58190
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 9.50257889675671e-06,
+      "loss": 2.3557,
+      "step": 58200
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 9.500393391030685e-06,
+      "loss": 2.2833,
+      "step": 58210
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 9.49820788530466e-06,
+      "loss": 2.3263,
+      "step": 58220
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 9.496022379578636e-06,
+      "loss": 2.326,
+      "step": 58230
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 9.49383687385261e-06,
+      "loss": 2.3287,
+      "step": 58240
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 9.491651368126586e-06,
+      "loss": 2.3371,
+      "step": 58250
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 9.48946586240056e-06,
+      "loss": 2.284,
+      "step": 58260
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 9.487280356674536e-06,
+      "loss": 2.3407,
+      "step": 58270
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 9.485094850948512e-06,
+      "loss": 2.2817,
+      "step": 58280
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 9.482909345222486e-06,
+      "loss": 2.2882,
+      "step": 58290
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 9.48072383949646e-06,
+      "loss": 2.3466,
+      "step": 58300
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 9.478538333770436e-06,
+      "loss": 2.2884,
+      "step": 58310
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 9.47635282804441e-06,
+      "loss": 2.3561,
+      "step": 58320
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 9.474167322318386e-06,
+      "loss": 2.315,
+      "step": 58330
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 9.47198181659236e-06,
+      "loss": 2.3283,
+      "step": 58340
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 9.469796310866334e-06,
+      "loss": 2.2934,
+      "step": 58350
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 9.46761080514031e-06,
+      "loss": 2.3423,
+      "step": 58360
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 9.465425299414284e-06,
+      "loss": 2.2943,
+      "step": 58370
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 9.46323979368826e-06,
+      "loss": 2.3098,
+      "step": 58380
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 9.461054287962234e-06,
+      "loss": 2.3216,
+      "step": 58390
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 9.45886878223621e-06,
+      "loss": 2.3356,
+      "step": 58400
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 9.456683276510186e-06,
+      "loss": 2.327,
+      "step": 58410
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 9.45449777078416e-06,
+      "loss": 2.3289,
+      "step": 58420
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 9.452312265058136e-06,
+      "loss": 2.3262,
+      "step": 58430
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 9.45012675933211e-06,
+      "loss": 2.3169,
+      "step": 58440
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 9.447941253606086e-06,
+      "loss": 2.3321,
+      "step": 58450
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 9.44575574788006e-06,
+      "loss": 2.3246,
+      "step": 58460
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 9.443570242154035e-06,
+      "loss": 2.2968,
+      "step": 58470
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 9.44138473642801e-06,
+      "loss": 2.3172,
+      "step": 58480
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 9.439199230701985e-06,
+      "loss": 2.3297,
+      "step": 58490
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 9.43701372497596e-06,
+      "loss": 2.3387,
+      "step": 58500
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 9.434828219249935e-06,
+      "loss": 2.3107,
+      "step": 58510
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 9.43264271352391e-06,
+      "loss": 2.3029,
+      "step": 58520
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 9.430457207797885e-06,
+      "loss": 2.3184,
+      "step": 58530
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 9.42827170207186e-06,
+      "loss": 2.2997,
+      "step": 58540
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 9.426086196345835e-06,
+      "loss": 2.3241,
+      "step": 58550
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 9.42390069061981e-06,
+      "loss": 2.316,
+      "step": 58560
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 9.421715184893785e-06,
+      "loss": 2.3225,
+      "step": 58570
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 9.41952967916776e-06,
+      "loss": 2.35,
+      "step": 58580
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 9.417344173441735e-06,
+      "loss": 2.3114,
+      "step": 58590
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 9.41515866771571e-06,
+      "loss": 2.2958,
+      "step": 58600
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 9.412973161989685e-06,
+      "loss": 2.3369,
+      "step": 58610
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 9.410787656263661e-06,
+      "loss": 2.3121,
+      "step": 58620
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 9.408602150537635e-06,
+      "loss": 2.3355,
+      "step": 58630
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 9.40641664481161e-06,
+      "loss": 2.3203,
+      "step": 58640
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 9.404231139085585e-06,
+      "loss": 2.325,
+      "step": 58650
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 9.40204563335956e-06,
+      "loss": 2.3558,
+      "step": 58660
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 9.399860127633535e-06,
+      "loss": 2.3208,
+      "step": 58670
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 9.39767462190751e-06,
+      "loss": 2.3455,
+      "step": 58680
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 9.395489116181485e-06,
+      "loss": 2.3308,
+      "step": 58690
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 9.39330361045546e-06,
+      "loss": 2.2981,
+      "step": 58700
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 9.391118104729435e-06,
+      "loss": 2.2907,
+      "step": 58710
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 9.38893259900341e-06,
+      "loss": 2.2743,
+      "step": 58720
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 9.386747093277385e-06,
+      "loss": 2.3142,
+      "step": 58730
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 9.38456158755136e-06,
+      "loss": 2.3157,
+      "step": 58740
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 9.382376081825335e-06,
+      "loss": 2.3006,
+      "step": 58750
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 9.380190576099311e-06,
+      "loss": 2.28,
+      "step": 58760
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 9.378005070373286e-06,
+      "loss": 2.3192,
+      "step": 58770
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 9.37581956464726e-06,
+      "loss": 2.2705,
+      "step": 58780
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 9.373634058921236e-06,
+      "loss": 2.3603,
+      "step": 58790
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 9.37144855319521e-06,
+      "loss": 2.2958,
+      "step": 58800
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 9.369263047469186e-06,
+      "loss": 2.336,
+      "step": 58810
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 9.36707754174316e-06,
+      "loss": 2.308,
+      "step": 58820
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 9.364892036017134e-06,
+      "loss": 2.3418,
+      "step": 58830
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 9.36270653029111e-06,
+      "loss": 2.3114,
+      "step": 58840
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 9.360521024565084e-06,
+      "loss": 2.3758,
+      "step": 58850
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 9.35833551883906e-06,
+      "loss": 2.3291,
+      "step": 58860
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 9.356150013113036e-06,
+      "loss": 2.3073,
+      "step": 58870
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 9.35396450738701e-06,
+      "loss": 2.3498,
+      "step": 58880
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 9.351779001660986e-06,
+      "loss": 2.3539,
+      "step": 58890
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 9.34959349593496e-06,
+      "loss": 2.3656,
+      "step": 58900
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 9.347407990208936e-06,
+      "loss": 2.319,
+      "step": 58910
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 9.34522248448291e-06,
+      "loss": 2.2961,
+      "step": 58920
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 9.343036978756886e-06,
+      "loss": 2.2974,
+      "step": 58930
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 9.34085147303086e-06,
+      "loss": 2.3123,
+      "step": 58940
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 9.338665967304834e-06,
+      "loss": 2.3171,
+      "step": 58950
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 9.33648046157881e-06,
+      "loss": 2.2585,
+      "step": 58960
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 9.334294955852784e-06,
+      "loss": 2.2749,
+      "step": 58970
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 9.33210945012676e-06,
+      "loss": 2.3446,
+      "step": 58980
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 9.329923944400735e-06,
+      "loss": 2.3251,
+      "step": 58990
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 9.32773843867471e-06,
+      "loss": 2.3536,
+      "step": 59000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 9.325552932948685e-06,
+      "loss": 2.327,
+      "step": 59010
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 9.32336742722266e-06,
+      "loss": 2.3511,
+      "step": 59020
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 9.321181921496635e-06,
+      "loss": 2.317,
+      "step": 59030
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 9.31899641577061e-06,
+      "loss": 2.3203,
+      "step": 59040
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 9.316810910044585e-06,
+      "loss": 2.3264,
+      "step": 59050
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 9.31462540431856e-06,
+      "loss": 2.3114,
+      "step": 59060
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 9.312439898592535e-06,
+      "loss": 2.3338,
+      "step": 59070
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 9.31025439286651e-06,
+      "loss": 2.3143,
+      "step": 59080
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 9.308068887140485e-06,
+      "loss": 2.3193,
+      "step": 59090
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 9.30588338141446e-06,
+      "loss": 2.3373,
+      "step": 59100
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 9.303697875688435e-06,
+      "loss": 2.3507,
+      "step": 59110
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 9.301512369962409e-06,
+      "loss": 2.2828,
+      "step": 59120
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 9.299326864236385e-06,
+      "loss": 2.2982,
+      "step": 59130
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 9.29714135851036e-06,
+      "loss": 2.3367,
+      "step": 59140
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 9.294955852784335e-06,
+      "loss": 2.3079,
+      "step": 59150
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 9.29277034705831e-06,
+      "loss": 2.3685,
+      "step": 59160
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 9.290584841332285e-06,
+      "loss": 2.3226,
+      "step": 59170
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 9.28839933560626e-06,
+      "loss": 2.3244,
+      "step": 59180
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 9.286213829880235e-06,
+      "loss": 2.2733,
+      "step": 59190
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 9.28402832415421e-06,
+      "loss": 2.28,
+      "step": 59200
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 9.281842818428185e-06,
+      "loss": 2.315,
+      "step": 59210
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 9.279657312702161e-06,
+      "loss": 2.3013,
+      "step": 59220
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 9.277471806976135e-06,
+      "loss": 2.3423,
+      "step": 59230
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 9.275286301250111e-06,
+      "loss": 2.3658,
+      "step": 59240
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 9.273100795524085e-06,
+      "loss": 2.3162,
+      "step": 59250
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 9.27091528979806e-06,
+      "loss": 2.3093,
+      "step": 59260
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 9.268729784072035e-06,
+      "loss": 2.2908,
+      "step": 59270
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 9.26654427834601e-06,
+      "loss": 2.3461,
+      "step": 59280
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 9.264358772619984e-06,
+      "loss": 2.3261,
+      "step": 59290
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 9.26217326689396e-06,
+      "loss": 2.3079,
+      "step": 59300
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 9.259987761167934e-06,
+      "loss": 2.2546,
+      "step": 59310
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 9.25780225544191e-06,
+      "loss": 2.3527,
+      "step": 59320
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 9.255616749715886e-06,
+      "loss": 2.3315,
+      "step": 59330
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 9.25343124398986e-06,
+      "loss": 2.3188,
+      "step": 59340
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 9.251245738263836e-06,
+      "loss": 2.3013,
+      "step": 59350
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 9.24906023253781e-06,
+      "loss": 2.3001,
+      "step": 59360
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 9.246874726811786e-06,
+      "loss": 2.3367,
+      "step": 59370
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 9.24468922108576e-06,
+      "loss": 2.297,
+      "step": 59380
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 9.242503715359736e-06,
+      "loss": 2.3279,
+      "step": 59390
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 9.24031820963371e-06,
+      "loss": 2.3174,
+      "step": 59400
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 9.238132703907686e-06,
+      "loss": 2.3325,
+      "step": 59410
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 9.23594719818166e-06,
+      "loss": 2.3484,
+      "step": 59420
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 9.233761692455634e-06,
+      "loss": 2.3487,
+      "step": 59430
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 9.23157618672961e-06,
+      "loss": 2.2837,
+      "step": 59440
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 9.229390681003584e-06,
+      "loss": 2.3118,
+      "step": 59450
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 9.22720517527756e-06,
+      "loss": 2.3209,
+      "step": 59460
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 9.225019669551534e-06,
+      "loss": 2.3499,
+      "step": 59470
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 9.22283416382551e-06,
+      "loss": 2.3225,
+      "step": 59480
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 9.220648658099484e-06,
+      "loss": 2.3531,
+      "step": 59490
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 9.21846315237346e-06,
+      "loss": 2.2686,
+      "step": 59500
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 9.216277646647434e-06,
+      "loss": 2.3073,
+      "step": 59510
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 9.21409214092141e-06,
+      "loss": 2.3294,
+      "step": 59520
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 9.211906635195385e-06,
+      "loss": 2.3203,
+      "step": 59530
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 9.20972112946936e-06,
+      "loss": 2.3422,
+      "step": 59540
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 9.207535623743335e-06,
+      "loss": 2.3012,
+      "step": 59550
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 9.20535011801731e-06,
+      "loss": 2.3284,
+      "step": 59560
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 9.203164612291285e-06,
+      "loss": 2.3376,
+      "step": 59570
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 9.20097910656526e-06,
+      "loss": 2.3104,
+      "step": 59580
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 9.198793600839235e-06,
+      "loss": 2.3137,
+      "step": 59590
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 9.196608095113209e-06,
+      "loss": 2.311,
+      "step": 59600
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 9.194422589387185e-06,
+      "loss": 2.3538,
+      "step": 59610
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 9.192237083661159e-06,
+      "loss": 2.3582,
+      "step": 59620
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 9.190051577935135e-06,
+      "loss": 2.3504,
+      "step": 59630
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 9.187866072209109e-06,
+      "loss": 2.3256,
+      "step": 59640
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 9.185680566483085e-06,
+      "loss": 2.2677,
+      "step": 59650
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 9.183495060757059e-06,
+      "loss": 2.3534,
+      "step": 59660
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 9.181309555031035e-06,
+      "loss": 2.3057,
+      "step": 59670
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 9.179124049305011e-06,
+      "loss": 2.2978,
+      "step": 59680
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 9.176938543578985e-06,
+      "loss": 2.3061,
+      "step": 59690
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 9.174753037852961e-06,
+      "loss": 2.3497,
+      "step": 59700
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 9.172567532126935e-06,
+      "loss": 2.3576,
+      "step": 59710
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 9.170382026400911e-06,
+      "loss": 2.3548,
+      "step": 59720
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 9.168196520674885e-06,
+      "loss": 2.2963,
+      "step": 59730
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 9.16601101494886e-06,
+      "loss": 2.329,
+      "step": 59740
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 9.163825509222835e-06,
+      "loss": 2.275,
+      "step": 59750
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 9.16164000349681e-06,
+      "loss": 2.3271,
+      "step": 59760
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 9.159454497770784e-06,
+      "loss": 2.3351,
+      "step": 59770
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 9.15726899204476e-06,
+      "loss": 2.3153,
+      "step": 59780
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 9.155083486318735e-06,
+      "loss": 2.3058,
+      "step": 59790
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 9.15289798059271e-06,
+      "loss": 2.3492,
+      "step": 59800
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 9.150712474866685e-06,
+      "loss": 2.3042,
+      "step": 59810
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 9.14852696914066e-06,
+      "loss": 2.3319,
+      "step": 59820
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 9.146341463414635e-06,
+      "loss": 2.3554,
+      "step": 59830
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 9.14415595768861e-06,
+      "loss": 2.3522,
+      "step": 59840
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 9.141970451962586e-06,
+      "loss": 2.2745,
+      "step": 59850
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 9.13978494623656e-06,
+      "loss": 2.3011,
+      "step": 59860
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 9.137599440510536e-06,
+      "loss": 2.3274,
+      "step": 59870
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 9.13541393478451e-06,
+      "loss": 2.3165,
+      "step": 59880
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 9.133228429058486e-06,
+      "loss": 2.3302,
+      "step": 59890
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 9.13104292333246e-06,
+      "loss": 2.2995,
+      "step": 59900
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 9.128857417606434e-06,
+      "loss": 2.3332,
+      "step": 59910
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 9.12667191188041e-06,
+      "loss": 2.2941,
+      "step": 59920
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 9.124486406154384e-06,
+      "loss": 2.3123,
+      "step": 59930
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 9.12230090042836e-06,
+      "loss": 2.3147,
+      "step": 59940
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 9.120115394702334e-06,
+      "loss": 2.304,
+      "step": 59950
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 9.11792988897631e-06,
+      "loss": 2.3327,
+      "step": 59960
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 9.115744383250284e-06,
+      "loss": 2.3086,
+      "step": 59970
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 9.11355887752426e-06,
+      "loss": 2.3426,
+      "step": 59980
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 9.111373371798234e-06,
+      "loss": 2.3602,
+      "step": 59990
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 9.10918786607221e-06,
+      "loss": 2.3333,
+      "step": 60000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 9.107002360346184e-06,
+      "loss": 2.2793,
+      "step": 60010
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 9.10481685462016e-06,
+      "loss": 2.3256,
+      "step": 60020
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 9.102631348894136e-06,
+      "loss": 2.3323,
+      "step": 60030
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 9.10044584316811e-06,
+      "loss": 2.3408,
+      "step": 60040
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 9.098260337442084e-06,
+      "loss": 2.3436,
+      "step": 60050
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 9.09607483171606e-06,
+      "loss": 2.3103,
+      "step": 60060
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 9.093889325990035e-06,
+      "loss": 2.2744,
+      "step": 60070
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 9.091703820264009e-06,
+      "loss": 2.3204,
+      "step": 60080
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 9.089518314537985e-06,
+      "loss": 2.3132,
+      "step": 60090
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 9.087332808811959e-06,
+      "loss": 2.3444,
+      "step": 60100
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 9.085147303085935e-06,
+      "loss": 2.3362,
+      "step": 60110
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 9.082961797359909e-06,
+      "loss": 2.319,
+      "step": 60120
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 9.080776291633885e-06,
+      "loss": 2.3179,
+      "step": 60130
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 9.07859078590786e-06,
+      "loss": 2.3305,
+      "step": 60140
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 9.076405280181835e-06,
+      "loss": 2.3555,
+      "step": 60150
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 9.07421977445581e-06,
+      "loss": 2.3684,
+      "step": 60160
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 9.072034268729785e-06,
+      "loss": 2.3237,
+      "step": 60170
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 9.06984876300376e-06,
+      "loss": 2.3215,
+      "step": 60180
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 9.067663257277735e-06,
+      "loss": 2.3083,
+      "step": 60190
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 9.06547775155171e-06,
+      "loss": 2.3298,
+      "step": 60200
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 9.063292245825685e-06,
+      "loss": 2.3362,
+      "step": 60210
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 9.06110674009966e-06,
+      "loss": 2.348,
+      "step": 60220
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 9.058921234373635e-06,
+      "loss": 2.3768,
+      "step": 60230
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 9.05673572864761e-06,
+      "loss": 2.3708,
+      "step": 60240
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 9.054550222921583e-06,
+      "loss": 2.3349,
+      "step": 60250
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 9.05236471719556e-06,
+      "loss": 2.3762,
+      "step": 60260
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 9.050179211469535e-06,
+      "loss": 2.3308,
+      "step": 60270
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 9.04799370574351e-06,
+      "loss": 2.3549,
+      "step": 60280
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 9.045808200017485e-06,
+      "loss": 2.3271,
+      "step": 60290
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 9.04362269429146e-06,
+      "loss": 2.3517,
+      "step": 60300
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 9.041437188565435e-06,
+      "loss": 2.3355,
+      "step": 60310
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 9.03925168283941e-06,
+      "loss": 2.2994,
+      "step": 60320
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 9.037066177113385e-06,
+      "loss": 2.3581,
+      "step": 60330
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 9.03488067138736e-06,
+      "loss": 2.3318,
+      "step": 60340
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 9.032695165661335e-06,
+      "loss": 2.3727,
+      "step": 60350
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 9.03050965993531e-06,
+      "loss": 2.3593,
+      "step": 60360
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 9.028324154209286e-06,
+      "loss": 2.3243,
+      "step": 60370
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 9.02613864848326e-06,
+      "loss": 2.3238,
+      "step": 60380
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 9.023953142757234e-06,
+      "loss": 2.3328,
+      "step": 60390
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 9.02176763703121e-06,
+      "loss": 2.3444,
+      "step": 60400
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 9.019582131305184e-06,
+      "loss": 2.2744,
+      "step": 60410
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 9.01739662557916e-06,
+      "loss": 2.3104,
+      "step": 60420
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 9.015211119853134e-06,
+      "loss": 2.3618,
+      "step": 60430
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 9.01302561412711e-06,
+      "loss": 2.3045,
+      "step": 60440
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 9.010840108401084e-06,
+      "loss": 2.3183,
+      "step": 60450
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 9.00865460267506e-06,
+      "loss": 2.3847,
+      "step": 60460
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 9.006469096949034e-06,
+      "loss": 2.336,
+      "step": 60470
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 9.00428359122301e-06,
+      "loss": 2.3386,
+      "step": 60480
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 9.002098085496986e-06,
+      "loss": 2.3321,
+      "step": 60490
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 8.99991257977096e-06,
+      "loss": 2.3302,
+      "step": 60500
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 8.997727074044936e-06,
+      "loss": 2.3273,
+      "step": 60510
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 8.99554156831891e-06,
+      "loss": 2.3416,
+      "step": 60520
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 8.993356062592884e-06,
+      "loss": 2.3541,
+      "step": 60530
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 8.99117055686686e-06,
+      "loss": 2.3457,
+      "step": 60540
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 8.988985051140834e-06,
+      "loss": 2.3725,
+      "step": 60550
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 8.986799545414809e-06,
+      "loss": 2.4035,
+      "step": 60560
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 8.984614039688784e-06,
+      "loss": 2.3355,
+      "step": 60570
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 8.982428533962759e-06,
+      "loss": 2.3636,
+      "step": 60580
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 8.980243028236735e-06,
+      "loss": 2.363,
+      "step": 60590
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 8.978057522510709e-06,
+      "loss": 2.3164,
+      "step": 60600
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 8.975872016784685e-06,
+      "loss": 2.342,
+      "step": 60610
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 8.97368651105866e-06,
+      "loss": 2.3204,
+      "step": 60620
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 8.971501005332635e-06,
+      "loss": 2.3194,
+      "step": 60630
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 8.96931549960661e-06,
+      "loss": 2.3516,
+      "step": 60640
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 8.967129993880585e-06,
+      "loss": 2.2977,
+      "step": 60650
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 8.96494448815456e-06,
+      "loss": 2.3186,
+      "step": 60660
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 8.962758982428535e-06,
+      "loss": 2.3272,
+      "step": 60670
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 8.96057347670251e-06,
+      "loss": 2.3344,
+      "step": 60680
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 8.958387970976485e-06,
+      "loss": 2.3375,
+      "step": 60690
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 8.956202465250459e-06,
+      "loss": 2.3261,
+      "step": 60700
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 8.954016959524435e-06,
+      "loss": 2.3213,
+      "step": 60710
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 8.951831453798409e-06,
+      "loss": 2.3442,
+      "step": 60720
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 8.949645948072385e-06,
+      "loss": 2.3425,
+      "step": 60730
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 8.947460442346359e-06,
+      "loss": 2.3528,
+      "step": 60740
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 8.945274936620335e-06,
+      "loss": 2.3212,
+      "step": 60750
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 8.94308943089431e-06,
+      "loss": 2.349,
+      "step": 60760
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 8.940903925168285e-06,
+      "loss": 2.3244,
+      "step": 60770
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 8.93871841944226e-06,
+      "loss": 2.3672,
+      "step": 60780
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 8.936532913716235e-06,
+      "loss": 2.3492,
+      "step": 60790
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 8.93434740799021e-06,
+      "loss": 2.2972,
+      "step": 60800
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 8.932161902264185e-06,
+      "loss": 2.3324,
+      "step": 60810
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 8.92997639653816e-06,
+      "loss": 2.3408,
+      "step": 60820
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 8.927790890812135e-06,
+      "loss": 2.3217,
+      "step": 60830
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 8.92560538508611e-06,
+      "loss": 2.3138,
+      "step": 60840
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 8.923419879360085e-06,
+      "loss": 2.2842,
+      "step": 60850
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 8.92123437363406e-06,
+      "loss": 2.2836,
+      "step": 60860
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 8.919048867908034e-06,
+      "loss": 2.3197,
+      "step": 60870
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 8.91686336218201e-06,
+      "loss": 2.3677,
+      "step": 60880
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 8.914677856455984e-06,
+      "loss": 2.345,
+      "step": 60890
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 8.91249235072996e-06,
+      "loss": 2.3386,
+      "step": 60900
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 8.910306845003934e-06,
+      "loss": 2.3285,
+      "step": 60910
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 8.90812133927791e-06,
+      "loss": 2.3374,
+      "step": 60920
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 8.905935833551884e-06,
+      "loss": 2.3312,
+      "step": 60930
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 8.90375032782586e-06,
+      "loss": 2.3118,
+      "step": 60940
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 8.901564822099834e-06,
+      "loss": 2.3548,
+      "step": 60950
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 8.89937931637381e-06,
+      "loss": 2.2886,
+      "step": 60960
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 8.897193810647786e-06,
+      "loss": 2.3603,
+      "step": 60970
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 8.89500830492176e-06,
+      "loss": 2.3124,
+      "step": 60980
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 8.892822799195736e-06,
+      "loss": 2.2951,
+      "step": 60990
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 8.89063729346971e-06,
+      "loss": 2.3181,
+      "step": 61000
+    },
+    {
+      "epoch": 12.0,
+      "eval_BLEU": 9.4709,
+      "eval_gen_len": 11.9557,
+      "eval_loss": 3.180349588394165,
+      "eval_runtime": 195.4926,
+      "eval_samples_per_second": 55.056,
+      "eval_steps_per_second": 0.22,
+      "step": 61008
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 8.888451787743684e-06,
+      "loss": 2.2747,
+      "step": 61010
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 8.88626628201766e-06,
+      "loss": 2.2398,
+      "step": 61020
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 8.884080776291634e-06,
+      "loss": 2.2102,
+      "step": 61030
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.881895270565608e-06,
+      "loss": 2.2131,
+      "step": 61040
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.879709764839584e-06,
+      "loss": 2.2585,
+      "step": 61050
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.877524259113558e-06,
+      "loss": 2.2185,
+      "step": 61060
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.875338753387534e-06,
+      "loss": 2.2593,
+      "step": 61070
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.87315324766151e-06,
+      "loss": 2.2816,
+      "step": 61080
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.870967741935484e-06,
+      "loss": 2.229,
+      "step": 61090
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.86878223620946e-06,
+      "loss": 2.1952,
+      "step": 61100
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.866596730483434e-06,
+      "loss": 2.2549,
+      "step": 61110
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.86441122475741e-06,
+      "loss": 2.2544,
+      "step": 61120
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.862225719031385e-06,
+      "loss": 2.2302,
+      "step": 61130
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.86004021330536e-06,
+      "loss": 2.2339,
+      "step": 61140
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.857854707579335e-06,
+      "loss": 2.2358,
+      "step": 61150
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.85566920185331e-06,
+      "loss": 2.2249,
+      "step": 61160
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.853483696127285e-06,
+      "loss": 2.2446,
+      "step": 61170
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.851298190401259e-06,
+      "loss": 2.2043,
+      "step": 61180
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.849112684675235e-06,
+      "loss": 2.2392,
+      "step": 61190
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.846927178949209e-06,
+      "loss": 2.2152,
+      "step": 61200
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.844741673223185e-06,
+      "loss": 2.2001,
+      "step": 61210
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.842556167497159e-06,
+      "loss": 2.2565,
+      "step": 61220
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.840370661771135e-06,
+      "loss": 2.265,
+      "step": 61230
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.838185156045109e-06,
+      "loss": 2.2655,
+      "step": 61240
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.835999650319085e-06,
+      "loss": 2.2397,
+      "step": 61250
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.833814144593059e-06,
+      "loss": 2.2694,
+      "step": 61260
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.831628638867035e-06,
+      "loss": 2.2328,
+      "step": 61270
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.82944313314101e-06,
+      "loss": 2.2483,
+      "step": 61280
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 8.827257627414985e-06,
+      "loss": 2.2094,
+      "step": 61290
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 8.82507212168896e-06,
+      "loss": 2.2101,
+      "step": 61300
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 8.822886615962935e-06,
+      "loss": 2.2636,
+      "step": 61310
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 8.82070111023691e-06,
+      "loss": 2.223,
+      "step": 61320
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 8.818515604510885e-06,
+      "loss": 2.272,
+      "step": 61330
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 8.81633009878486e-06,
+      "loss": 2.2614,
+      "step": 61340
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 8.814144593058834e-06,
+      "loss": 2.2212,
+      "step": 61350
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 8.81195908733281e-06,
+      "loss": 2.2602,
+      "step": 61360
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 8.809773581606784e-06,
+      "loss": 2.2304,
+      "step": 61370
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 8.80758807588076e-06,
+      "loss": 2.2316,
+      "step": 61380
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 8.805402570154734e-06,
+      "loss": 2.238,
+      "step": 61390
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 8.80321706442871e-06,
+      "loss": 2.2998,
+      "step": 61400
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 8.801031558702684e-06,
+      "loss": 2.2749,
+      "step": 61410
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 8.79884605297666e-06,
+      "loss": 2.237,
+      "step": 61420
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 8.796660547250635e-06,
+      "loss": 2.2708,
+      "step": 61430
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 8.79447504152461e-06,
+      "loss": 2.2241,
+      "step": 61440
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 8.792289535798586e-06,
+      "loss": 2.2594,
+      "step": 61450
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 8.79010403007256e-06,
+      "loss": 2.2466,
+      "step": 61460
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 8.787918524346536e-06,
+      "loss": 2.2252,
+      "step": 61470
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 8.78573301862051e-06,
+      "loss": 2.2498,
+      "step": 61480
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 8.783547512894484e-06,
+      "loss": 2.267,
+      "step": 61490
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 8.78136200716846e-06,
+      "loss": 2.2227,
+      "step": 61500
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 8.779176501442434e-06,
+      "loss": 2.2408,
+      "step": 61510
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 8.776990995716408e-06,
+      "loss": 2.2232,
+      "step": 61520
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 8.774805489990384e-06,
+      "loss": 2.2575,
+      "step": 61530
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 8.77261998426436e-06,
+      "loss": 2.2558,
+      "step": 61540
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 8.770434478538334e-06,
+      "loss": 2.2176,
+      "step": 61550
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 8.76824897281231e-06,
+      "loss": 2.2636,
+      "step": 61560
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 8.766063467086284e-06,
+      "loss": 2.2494,
+      "step": 61570
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 8.76387796136026e-06,
+      "loss": 2.2431,
+      "step": 61580
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 8.761692455634234e-06,
+      "loss": 2.2713,
+      "step": 61590
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 8.75950694990821e-06,
+      "loss": 2.2735,
+      "step": 61600
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 8.757321444182184e-06,
+      "loss": 2.2426,
+      "step": 61610
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 8.75513593845616e-06,
+      "loss": 2.2293,
+      "step": 61620
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 8.752950432730134e-06,
+      "loss": 2.2109,
+      "step": 61630
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 8.75076492700411e-06,
+      "loss": 2.2396,
+      "step": 61640
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 8.748579421278084e-06,
+      "loss": 2.253,
+      "step": 61650
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 8.746393915552059e-06,
+      "loss": 2.2312,
+      "step": 61660
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 8.744208409826035e-06,
+      "loss": 2.2501,
+      "step": 61670
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 8.742022904100009e-06,
+      "loss": 2.271,
+      "step": 61680
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 8.739837398373985e-06,
+      "loss": 2.2474,
+      "step": 61690
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 8.737651892647959e-06,
+      "loss": 2.2849,
+      "step": 61700
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 8.735466386921935e-06,
+      "loss": 2.2254,
+      "step": 61710
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 8.733280881195909e-06,
+      "loss": 2.2602,
+      "step": 61720
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 8.731095375469885e-06,
+      "loss": 2.2525,
+      "step": 61730
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 8.728909869743859e-06,
+      "loss": 2.2396,
+      "step": 61740
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 8.726724364017835e-06,
+      "loss": 2.2355,
+      "step": 61750
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 8.724538858291809e-06,
+      "loss": 2.2222,
+      "step": 61760
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 8.722353352565785e-06,
+      "loss": 2.2621,
+      "step": 61770
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 8.72016784683976e-06,
+      "loss": 2.2346,
+      "step": 61780
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 8.717982341113735e-06,
+      "loss": 2.2222,
+      "step": 61790
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 8.715796835387709e-06,
+      "loss": 2.2187,
+      "step": 61800
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 8.713611329661685e-06,
+      "loss": 2.2579,
+      "step": 61810
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 8.71142582393566e-06,
+      "loss": 2.2198,
+      "step": 61820
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 8.709240318209633e-06,
+      "loss": 2.2514,
+      "step": 61830
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 8.70705481248361e-06,
+      "loss": 2.2629,
+      "step": 61840
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 8.704869306757583e-06,
+      "loss": 2.2386,
+      "step": 61850
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 8.70268380103156e-06,
+      "loss": 2.2034,
+      "step": 61860
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 8.700498295305533e-06,
+      "loss": 2.2455,
+      "step": 61870
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 8.69831278957951e-06,
+      "loss": 2.2653,
+      "step": 61880
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 8.696127283853485e-06,
+      "loss": 2.2409,
+      "step": 61890
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 8.69394177812746e-06,
+      "loss": 2.2612,
+      "step": 61900
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 8.691756272401435e-06,
+      "loss": 2.2747,
+      "step": 61910
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 8.68957076667541e-06,
+      "loss": 2.2315,
+      "step": 61920
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 8.687385260949385e-06,
+      "loss": 2.2143,
+      "step": 61930
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 8.68519975522336e-06,
+      "loss": 2.2349,
+      "step": 61940
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 8.683014249497335e-06,
+      "loss": 2.252,
+      "step": 61950
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 8.68082874377131e-06,
+      "loss": 2.2525,
+      "step": 61960
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 8.678643238045284e-06,
+      "loss": 2.2374,
+      "step": 61970
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 8.67645773231926e-06,
+      "loss": 2.2609,
+      "step": 61980
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 8.674272226593234e-06,
+      "loss": 2.2261,
+      "step": 61990
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 8.67208672086721e-06,
+      "loss": 2.2538,
+      "step": 62000
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 8.669901215141184e-06,
+      "loss": 2.2357,
+      "step": 62010
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 8.66771570941516e-06,
+      "loss": 2.2417,
+      "step": 62020
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 8.665530203689134e-06,
+      "loss": 2.2614,
+      "step": 62030
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 8.66334469796311e-06,
+      "loss": 2.2486,
+      "step": 62040
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 8.661159192237084e-06,
+      "loss": 2.2184,
+      "step": 62050
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 8.65897368651106e-06,
+      "loss": 2.2478,
+      "step": 62060
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 8.656788180785034e-06,
+      "loss": 2.227,
+      "step": 62070
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 8.65460267505901e-06,
+      "loss": 2.2761,
+      "step": 62080
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 8.652417169332984e-06,
+      "loss": 2.2685,
+      "step": 62090
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 8.65023166360696e-06,
+      "loss": 2.2609,
+      "step": 62100
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 8.648046157880934e-06,
+      "loss": 2.285,
+      "step": 62110
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 8.64586065215491e-06,
+      "loss": 2.2676,
+      "step": 62120
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 8.643675146428884e-06,
+      "loss": 2.2998,
+      "step": 62130
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 8.641489640702858e-06,
+      "loss": 2.2624,
+      "step": 62140
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 8.639304134976834e-06,
+      "loss": 2.3022,
+      "step": 62150
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 8.637118629250809e-06,
+      "loss": 2.2665,
+      "step": 62160
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 8.634933123524784e-06,
+      "loss": 2.252,
+      "step": 62170
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 8.632747617798759e-06,
+      "loss": 2.2106,
+      "step": 62180
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 8.630562112072735e-06,
+      "loss": 2.2482,
+      "step": 62190
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 8.628376606346709e-06,
+      "loss": 2.2693,
+      "step": 62200
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 8.626191100620685e-06,
+      "loss": 2.2613,
+      "step": 62210
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 8.624005594894659e-06,
+      "loss": 2.2035,
+      "step": 62220
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 8.621820089168635e-06,
+      "loss": 2.2413,
+      "step": 62230
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 8.61963458344261e-06,
+      "loss": 2.2369,
+      "step": 62240
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 8.617449077716585e-06,
+      "loss": 2.2694,
+      "step": 62250
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 8.61526357199056e-06,
+      "loss": 2.2816,
+      "step": 62260
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 8.613078066264535e-06,
+      "loss": 2.2366,
+      "step": 62270
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 8.610892560538509e-06,
+      "loss": 2.2445,
+      "step": 62280
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 8.608707054812485e-06,
+      "loss": 2.2924,
+      "step": 62290
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 8.606521549086459e-06,
+      "loss": 2.248,
+      "step": 62300
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 8.604336043360433e-06,
+      "loss": 2.2445,
+      "step": 62310
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 8.602150537634409e-06,
+      "loss": 2.2494,
+      "step": 62320
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 8.599965031908383e-06,
+      "loss": 2.2647,
+      "step": 62330
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 8.597779526182359e-06,
+      "loss": 2.2258,
+      "step": 62340
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 8.595594020456335e-06,
+      "loss": 2.2726,
+      "step": 62350
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 8.59340851473031e-06,
+      "loss": 2.2439,
+      "step": 62360
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 8.591223009004285e-06,
+      "loss": 2.2204,
+      "step": 62370
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 8.58903750327826e-06,
+      "loss": 2.2536,
+      "step": 62380
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 8.586851997552235e-06,
+      "loss": 2.2085,
+      "step": 62390
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 8.58466649182621e-06,
+      "loss": 2.255,
+      "step": 62400
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 8.582480986100185e-06,
+      "loss": 2.2583,
+      "step": 62410
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 8.58029548037416e-06,
+      "loss": 2.2701,
+      "step": 62420
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 8.578109974648135e-06,
+      "loss": 2.2683,
+      "step": 62430
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 8.57592446892211e-06,
+      "loss": 2.2494,
+      "step": 62440
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 8.573738963196084e-06,
+      "loss": 2.2707,
+      "step": 62450
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 8.57155345747006e-06,
+      "loss": 2.2714,
+      "step": 62460
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 8.569367951744034e-06,
+      "loss": 2.2885,
+      "step": 62470
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 8.56718244601801e-06,
+      "loss": 2.2867,
+      "step": 62480
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 8.564996940291984e-06,
+      "loss": 2.242,
+      "step": 62490
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 8.56281143456596e-06,
+      "loss": 2.2066,
+      "step": 62500
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 8.560625928839934e-06,
+      "loss": 2.2223,
+      "step": 62510
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 8.55844042311391e-06,
+      "loss": 2.2707,
+      "step": 62520
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 8.556254917387884e-06,
+      "loss": 2.2614,
+      "step": 62530
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 8.55406941166186e-06,
+      "loss": 2.2243,
+      "step": 62540
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 8.551883905935834e-06,
+      "loss": 2.298,
+      "step": 62550
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 8.54969840020981e-06,
+      "loss": 2.2665,
+      "step": 62560
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 8.547512894483784e-06,
+      "loss": 2.2909,
+      "step": 62570
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 8.54532738875776e-06,
+      "loss": 2.2448,
+      "step": 62580
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 8.543141883031734e-06,
+      "loss": 2.2442,
+      "step": 62590
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 8.54095637730571e-06,
+      "loss": 2.2621,
+      "step": 62600
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.538770871579684e-06,
+      "loss": 2.2862,
+      "step": 62610
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.536585365853658e-06,
+      "loss": 2.2182,
+      "step": 62620
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.534399860127634e-06,
+      "loss": 2.233,
+      "step": 62630
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.532214354401608e-06,
+      "loss": 2.2494,
+      "step": 62640
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.530028848675584e-06,
+      "loss": 2.2403,
+      "step": 62650
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.527843342949558e-06,
+      "loss": 2.2509,
+      "step": 62660
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 8.525657837223534e-06,
+      "loss": 2.2696,
+      "step": 62670
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 8.523472331497509e-06,
+      "loss": 2.2155,
+      "step": 62680
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 8.521286825771484e-06,
+      "loss": 2.2582,
+      "step": 62690
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 8.51910132004546e-06,
+      "loss": 2.2387,
+      "step": 62700
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 8.516915814319434e-06,
+      "loss": 2.2415,
+      "step": 62710
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 8.51473030859341e-06,
+      "loss": 2.2485,
+      "step": 62720
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 8.512544802867385e-06,
+      "loss": 2.2701,
+      "step": 62730
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 8.51035929714136e-06,
+      "loss": 2.2562,
+      "step": 62740
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 8.508173791415335e-06,
+      "loss": 2.2922,
+      "step": 62750
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 8.505988285689309e-06,
+      "loss": 2.3103,
+      "step": 62760
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 8.503802779963285e-06,
+      "loss": 2.2575,
+      "step": 62770
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 8.501617274237259e-06,
+      "loss": 2.2487,
+      "step": 62780
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 8.499431768511233e-06,
+      "loss": 2.2779,
+      "step": 62790
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 8.497246262785209e-06,
+      "loss": 2.2311,
+      "step": 62800
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 8.495060757059183e-06,
+      "loss": 2.257,
+      "step": 62810
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 8.492875251333159e-06,
+      "loss": 2.2833,
+      "step": 62820
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 8.490689745607135e-06,
+      "loss": 2.2408,
+      "step": 62830
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 8.488504239881109e-06,
+      "loss": 2.2584,
+      "step": 62840
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 8.486318734155085e-06,
+      "loss": 2.2334,
+      "step": 62850
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 8.484133228429059e-06,
+      "loss": 2.2586,
+      "step": 62860
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 8.481947722703035e-06,
+      "loss": 2.2807,
+      "step": 62870
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 8.479762216977009e-06,
+      "loss": 2.2618,
+      "step": 62880
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 8.477576711250985e-06,
+      "loss": 2.2366,
+      "step": 62890
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 8.47539120552496e-06,
+      "loss": 2.2857,
+      "step": 62900
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 8.473205699798935e-06,
+      "loss": 2.311,
+      "step": 62910
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 8.47102019407291e-06,
+      "loss": 2.2578,
+      "step": 62920
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 8.468834688346883e-06,
+      "loss": 2.2747,
+      "step": 62930
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 8.46664918262086e-06,
+      "loss": 2.2732,
+      "step": 62940
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 8.464463676894834e-06,
+      "loss": 2.2814,
+      "step": 62950
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 8.46227817116881e-06,
+      "loss": 2.261,
+      "step": 62960
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 8.460092665442784e-06,
+      "loss": 2.259,
+      "step": 62970
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 8.45790715971676e-06,
+      "loss": 2.2443,
+      "step": 62980
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 8.455721653990734e-06,
+      "loss": 2.2805,
+      "step": 62990
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 8.45353614826471e-06,
+      "loss": 2.2296,
+      "step": 63000
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 8.451350642538684e-06,
+      "loss": 2.2366,
+      "step": 63010
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 8.44916513681266e-06,
+      "loss": 2.2255,
+      "step": 63020
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 8.446979631086634e-06,
+      "loss": 2.2773,
+      "step": 63030
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 8.44479412536061e-06,
+      "loss": 2.2378,
+      "step": 63040
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 8.442608619634584e-06,
+      "loss": 2.2593,
+      "step": 63050
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 8.44042311390856e-06,
+      "loss": 2.2697,
+      "step": 63060
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 8.438237608182534e-06,
+      "loss": 2.2406,
+      "step": 63070
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 8.43605210245651e-06,
+      "loss": 2.2619,
+      "step": 63080
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 8.433866596730484e-06,
+      "loss": 2.2642,
+      "step": 63090
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 8.431681091004458e-06,
+      "loss": 2.2154,
+      "step": 63100
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 8.429495585278434e-06,
+      "loss": 2.2579,
+      "step": 63110
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 8.427310079552408e-06,
+      "loss": 2.25,
+      "step": 63120
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 8.425124573826384e-06,
+      "loss": 2.3187,
+      "step": 63130
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 8.422939068100358e-06,
+      "loss": 2.2596,
+      "step": 63140
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 8.420753562374334e-06,
+      "loss": 2.2698,
+      "step": 63150
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 8.418568056648308e-06,
+      "loss": 2.2582,
+      "step": 63160
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 8.416382550922284e-06,
+      "loss": 2.2625,
+      "step": 63170
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 8.41419704519626e-06,
+      "loss": 2.2723,
+      "step": 63180
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 8.412011539470234e-06,
+      "loss": 2.2687,
+      "step": 63190
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 8.40982603374421e-06,
+      "loss": 2.2769,
+      "step": 63200
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 8.407640528018184e-06,
+      "loss": 2.2549,
+      "step": 63210
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 8.405455022292159e-06,
+      "loss": 2.2553,
+      "step": 63220
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 8.403269516566134e-06,
+      "loss": 2.2829,
+      "step": 63230
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 8.401084010840109e-06,
+      "loss": 2.2946,
+      "step": 63240
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 8.398898505114084e-06,
+      "loss": 2.2741,
+      "step": 63250
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 8.396712999388059e-06,
+      "loss": 2.2675,
+      "step": 63260
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 8.394527493662033e-06,
+      "loss": 2.2605,
+      "step": 63270
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 8.392341987936009e-06,
+      "loss": 2.2449,
+      "step": 63280
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 8.390156482209985e-06,
+      "loss": 2.2795,
+      "step": 63290
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 8.387970976483959e-06,
+      "loss": 2.2357,
+      "step": 63300
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 8.385785470757935e-06,
+      "loss": 2.2676,
+      "step": 63310
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 8.383599965031909e-06,
+      "loss": 2.2801,
+      "step": 63320
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 8.381414459305885e-06,
+      "loss": 2.2825,
+      "step": 63330
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 8.379228953579859e-06,
+      "loss": 2.265,
+      "step": 63340
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 8.377043447853835e-06,
+      "loss": 2.2767,
+      "step": 63350
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 8.374857942127809e-06,
+      "loss": 2.2849,
+      "step": 63360
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 8.372672436401785e-06,
+      "loss": 2.3138,
+      "step": 63370
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 8.370486930675759e-06,
+      "loss": 2.2812,
+      "step": 63380
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 8.368301424949735e-06,
+      "loss": 2.2604,
+      "step": 63390
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 8.366115919223709e-06,
+      "loss": 2.2657,
+      "step": 63400
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 8.363930413497683e-06,
+      "loss": 2.2659,
+      "step": 63410
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 8.36174490777166e-06,
+      "loss": 2.2901,
+      "step": 63420
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 8.359559402045633e-06,
+      "loss": 2.2392,
+      "step": 63430
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 8.35737389631961e-06,
+      "loss": 2.2284,
+      "step": 63440
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 8.355188390593583e-06,
+      "loss": 2.2638,
+      "step": 63450
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 8.35300288486756e-06,
+      "loss": 2.2908,
+      "step": 63460
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 8.350817379141533e-06,
+      "loss": 2.268,
+      "step": 63470
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 8.34863187341551e-06,
+      "loss": 2.2559,
+      "step": 63480
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 8.346446367689484e-06,
+      "loss": 2.3025,
+      "step": 63490
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 8.34426086196346e-06,
+      "loss": 2.2724,
+      "step": 63500
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 8.342075356237434e-06,
+      "loss": 2.2651,
+      "step": 63510
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 8.33988985051141e-06,
+      "loss": 2.2544,
+      "step": 63520
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 8.337704344785384e-06,
+      "loss": 2.2925,
+      "step": 63530
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 8.33551883905936e-06,
+      "loss": 2.2502,
+      "step": 63540
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 2.2664,
+      "step": 63550
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 8.33114782760731e-06,
+      "loss": 2.2574,
+      "step": 63560
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 8.328962321881284e-06,
+      "loss": 2.2883,
+      "step": 63570
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 8.326776816155258e-06,
+      "loss": 2.2428,
+      "step": 63580
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 8.324591310429234e-06,
+      "loss": 2.2784,
+      "step": 63590
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 8.322405804703208e-06,
+      "loss": 2.2629,
+      "step": 63600
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 8.320220298977184e-06,
+      "loss": 2.2543,
+      "step": 63610
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 8.318034793251158e-06,
+      "loss": 2.2821,
+      "step": 63620
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 8.315849287525134e-06,
+      "loss": 2.2572,
+      "step": 63630
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 8.31366378179911e-06,
+      "loss": 2.2839,
+      "step": 63640
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 8.311478276073084e-06,
+      "loss": 2.2605,
+      "step": 63650
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 8.30929277034706e-06,
+      "loss": 2.2659,
+      "step": 63660
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 8.307107264621034e-06,
+      "loss": 2.2791,
+      "step": 63670
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 8.30492175889501e-06,
+      "loss": 2.26,
+      "step": 63680
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 8.302736253168984e-06,
+      "loss": 2.2494,
+      "step": 63690
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 8.300550747442958e-06,
+      "loss": 2.293,
+      "step": 63700
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 8.298365241716934e-06,
+      "loss": 2.2483,
+      "step": 63710
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 8.296179735990908e-06,
+      "loss": 2.2731,
+      "step": 63720
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 8.293994230264884e-06,
+      "loss": 2.3046,
+      "step": 63730
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 8.291808724538858e-06,
+      "loss": 2.2931,
+      "step": 63740
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 8.289623218812834e-06,
+      "loss": 2.2577,
+      "step": 63750
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 8.287437713086809e-06,
+      "loss": 2.2851,
+      "step": 63760
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 8.285252207360784e-06,
+      "loss": 2.2986,
+      "step": 63770
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 8.283066701634759e-06,
+      "loss": 2.2843,
+      "step": 63780
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 8.280881195908734e-06,
+      "loss": 2.2387,
+      "step": 63790
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 8.278695690182709e-06,
+      "loss": 2.2648,
+      "step": 63800
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 8.276510184456685e-06,
+      "loss": 2.3054,
+      "step": 63810
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 8.274324678730659e-06,
+      "loss": 2.2876,
+      "step": 63820
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.272139173004635e-06,
+      "loss": 2.2626,
+      "step": 63830
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.269953667278609e-06,
+      "loss": 2.2653,
+      "step": 63840
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.267768161552585e-06,
+      "loss": 2.2928,
+      "step": 63850
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.265582655826559e-06,
+      "loss": 2.2712,
+      "step": 63860
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.263397150100533e-06,
+      "loss": 2.2455,
+      "step": 63870
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.261211644374509e-06,
+      "loss": 2.2644,
+      "step": 63880
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 8.259026138648483e-06,
+      "loss": 2.3113,
+      "step": 63890
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 8.256840632922459e-06,
+      "loss": 2.2599,
+      "step": 63900
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 8.254655127196433e-06,
+      "loss": 2.3088,
+      "step": 63910
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 8.252469621470409e-06,
+      "loss": 2.2943,
+      "step": 63920
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 8.250284115744383e-06,
+      "loss": 2.2743,
+      "step": 63930
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 8.248098610018359e-06,
+      "loss": 2.313,
+      "step": 63940
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 8.245913104292333e-06,
+      "loss": 2.282,
+      "step": 63950
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 8.24372759856631e-06,
+      "loss": 2.308,
+      "step": 63960
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 8.241542092840283e-06,
+      "loss": 2.2583,
+      "step": 63970
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 8.23935658711426e-06,
+      "loss": 2.2433,
+      "step": 63980
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 8.237171081388235e-06,
+      "loss": 2.3097,
+      "step": 63990
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 8.23498557566221e-06,
+      "loss": 2.2451,
+      "step": 64000
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 8.232800069936183e-06,
+      "loss": 2.2635,
+      "step": 64010
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 8.23061456421016e-06,
+      "loss": 2.2903,
+      "step": 64020
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 8.228429058484134e-06,
+      "loss": 2.2702,
+      "step": 64030
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 8.22624355275811e-06,
+      "loss": 2.2683,
+      "step": 64040
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 8.224058047032084e-06,
+      "loss": 2.2874,
+      "step": 64050
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 8.221872541306058e-06,
+      "loss": 2.2652,
+      "step": 64060
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 8.219687035580034e-06,
+      "loss": 2.2914,
+      "step": 64070
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 8.217501529854008e-06,
+      "loss": 2.2687,
+      "step": 64080
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 8.215316024127984e-06,
+      "loss": 2.2969,
+      "step": 64090
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 8.21313051840196e-06,
+      "loss": 2.2656,
+      "step": 64100
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 8.210945012675934e-06,
+      "loss": 2.2568,
+      "step": 64110
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 8.20875950694991e-06,
+      "loss": 2.2975,
+      "step": 64120
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 8.206574001223884e-06,
+      "loss": 2.2511,
+      "step": 64130
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 8.20438849549786e-06,
+      "loss": 2.2672,
+      "step": 64140
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 8.202202989771834e-06,
+      "loss": 2.2806,
+      "step": 64150
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 8.20001748404581e-06,
+      "loss": 2.2817,
+      "step": 64160
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 8.197831978319784e-06,
+      "loss": 2.3157,
+      "step": 64170
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 8.195646472593758e-06,
+      "loss": 2.212,
+      "step": 64180
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 8.193460966867734e-06,
+      "loss": 2.31,
+      "step": 64190
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 8.191275461141708e-06,
+      "loss": 2.2429,
+      "step": 64200
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 8.189089955415684e-06,
+      "loss": 2.2848,
+      "step": 64210
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 8.186904449689658e-06,
+      "loss": 2.2978,
+      "step": 64220
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 8.184718943963634e-06,
+      "loss": 2.2703,
+      "step": 64230
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 8.182533438237608e-06,
+      "loss": 2.2686,
+      "step": 64240
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 8.180347932511584e-06,
+      "loss": 2.2616,
+      "step": 64250
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 8.178162426785558e-06,
+      "loss": 2.2887,
+      "step": 64260
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 8.175976921059534e-06,
+      "loss": 2.2359,
+      "step": 64270
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 8.173791415333509e-06,
+      "loss": 2.2669,
+      "step": 64280
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 8.171605909607484e-06,
+      "loss": 2.255,
+      "step": 64290
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 8.169420403881459e-06,
+      "loss": 2.2688,
+      "step": 64300
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 8.167234898155434e-06,
+      "loss": 2.2912,
+      "step": 64310
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 8.165049392429409e-06,
+      "loss": 2.2541,
+      "step": 64320
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 8.162863886703385e-06,
+      "loss": 2.2694,
+      "step": 64330
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 8.160678380977359e-06,
+      "loss": 2.245,
+      "step": 64340
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 8.158492875251333e-06,
+      "loss": 2.2697,
+      "step": 64350
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 8.156307369525309e-06,
+      "loss": 2.2237,
+      "step": 64360
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 8.154121863799283e-06,
+      "loss": 2.2593,
+      "step": 64370
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 8.151936358073259e-06,
+      "loss": 2.2588,
+      "step": 64380
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 8.149750852347233e-06,
+      "loss": 2.2679,
+      "step": 64390
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 8.147565346621209e-06,
+      "loss": 2.3063,
+      "step": 64400
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 8.145379840895183e-06,
+      "loss": 2.2871,
+      "step": 64410
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 8.143194335169159e-06,
+      "loss": 2.295,
+      "step": 64420
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 8.141008829443133e-06,
+      "loss": 2.2595,
+      "step": 64430
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 8.138823323717109e-06,
+      "loss": 2.297,
+      "step": 64440
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 8.136637817991085e-06,
+      "loss": 2.2801,
+      "step": 64450
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 8.134452312265059e-06,
+      "loss": 2.265,
+      "step": 64460
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 8.132266806539035e-06,
+      "loss": 2.2641,
+      "step": 64470
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 2.2666,
+      "step": 64480
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 8.127895795086983e-06,
+      "loss": 2.2756,
+      "step": 64490
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 8.12571028936096e-06,
+      "loss": 2.2726,
+      "step": 64500
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 8.123524783634933e-06,
+      "loss": 2.2361,
+      "step": 64510
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 8.121339277908908e-06,
+      "loss": 2.2955,
+      "step": 64520
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 8.119153772182883e-06,
+      "loss": 2.2483,
+      "step": 64530
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 8.116968266456858e-06,
+      "loss": 2.3049,
+      "step": 64540
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 8.114782760730834e-06,
+      "loss": 2.2586,
+      "step": 64550
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 8.11259725500481e-06,
+      "loss": 2.2708,
+      "step": 64560
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 8.110411749278784e-06,
+      "loss": 2.2452,
+      "step": 64570
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 8.10822624355276e-06,
+      "loss": 2.2621,
+      "step": 64580
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 8.106040737826734e-06,
+      "loss": 2.3095,
+      "step": 64590
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 8.10385523210071e-06,
+      "loss": 2.275,
+      "step": 64600
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 8.101669726374684e-06,
+      "loss": 2.2892,
+      "step": 64610
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 8.09948422064866e-06,
+      "loss": 2.2866,
+      "step": 64620
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 8.097298714922634e-06,
+      "loss": 2.2628,
+      "step": 64630
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 8.09511320919661e-06,
+      "loss": 2.2926,
+      "step": 64640
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 8.092927703470584e-06,
+      "loss": 2.2765,
+      "step": 64650
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 8.090742197744558e-06,
+      "loss": 2.2759,
+      "step": 64660
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 8.088556692018534e-06,
+      "loss": 2.2861,
+      "step": 64670
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 8.086371186292508e-06,
+      "loss": 2.2632,
+      "step": 64680
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 8.084185680566484e-06,
+      "loss": 2.2674,
+      "step": 64690
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 8.082000174840458e-06,
+      "loss": 2.2538,
+      "step": 64700
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 8.079814669114434e-06,
+      "loss": 2.289,
+      "step": 64710
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 8.077629163388408e-06,
+      "loss": 2.2838,
+      "step": 64720
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 8.075443657662384e-06,
+      "loss": 2.3179,
+      "step": 64730
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 8.073258151936358e-06,
+      "loss": 2.2978,
+      "step": 64740
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 8.071072646210334e-06,
+      "loss": 2.2776,
+      "step": 64750
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 8.068887140484308e-06,
+      "loss": 2.317,
+      "step": 64760
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 8.066701634758284e-06,
+      "loss": 2.2479,
+      "step": 64770
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 2.2531,
+      "step": 64780
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 8.062330623306234e-06,
+      "loss": 2.2565,
+      "step": 64790
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 8.060145117580208e-06,
+      "loss": 2.2722,
+      "step": 64800
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 8.057959611854184e-06,
+      "loss": 2.2721,
+      "step": 64810
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 8.055774106128159e-06,
+      "loss": 2.2779,
+      "step": 64820
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 8.053588600402133e-06,
+      "loss": 2.3067,
+      "step": 64830
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 8.051403094676109e-06,
+      "loss": 2.3004,
+      "step": 64840
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 8.049217588950083e-06,
+      "loss": 2.2601,
+      "step": 64850
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 8.047032083224059e-06,
+      "loss": 2.2748,
+      "step": 64860
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 8.044846577498033e-06,
+      "loss": 2.2536,
+      "step": 64870
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 8.042661071772009e-06,
+      "loss": 2.2572,
+      "step": 64880
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 8.040475566045983e-06,
+      "loss": 2.2766,
+      "step": 64890
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 8.038290060319959e-06,
+      "loss": 2.2596,
+      "step": 64900
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 8.036104554593935e-06,
+      "loss": 2.275,
+      "step": 64910
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 8.033919048867909e-06,
+      "loss": 2.2327,
+      "step": 64920
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 8.031733543141885e-06,
+      "loss": 2.2914,
+      "step": 64930
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 8.029548037415859e-06,
+      "loss": 2.3015,
+      "step": 64940
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 8.027362531689835e-06,
+      "loss": 2.2749,
+      "step": 64950
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 8.025177025963809e-06,
+      "loss": 2.27,
+      "step": 64960
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 8.022991520237783e-06,
+      "loss": 2.2733,
+      "step": 64970
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 8.020806014511759e-06,
+      "loss": 2.218,
+      "step": 64980
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 8.018620508785733e-06,
+      "loss": 2.2628,
+      "step": 64990
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 8.016435003059707e-06,
+      "loss": 2.2849,
+      "step": 65000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 8.014249497333683e-06,
+      "loss": 2.2772,
+      "step": 65010
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 8.012063991607657e-06,
+      "loss": 2.2825,
+      "step": 65020
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 8.009878485881633e-06,
+      "loss": 2.2678,
+      "step": 65030
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 8.00769298015561e-06,
+      "loss": 2.2936,
+      "step": 65040
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 8.005507474429583e-06,
+      "loss": 2.3031,
+      "step": 65050
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 8.00332196870356e-06,
+      "loss": 2.2984,
+      "step": 65060
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 8.001136462977533e-06,
+      "loss": 2.2875,
+      "step": 65070
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 7.99895095725151e-06,
+      "loss": 2.2675,
+      "step": 65080
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 7.996765451525484e-06,
+      "loss": 2.2584,
+      "step": 65090
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 7.99457994579946e-06,
+      "loss": 2.2844,
+      "step": 65100
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.992394440073434e-06,
+      "loss": 2.2379,
+      "step": 65110
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.99020893434741e-06,
+      "loss": 2.2748,
+      "step": 65120
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.988023428621384e-06,
+      "loss": 2.2501,
+      "step": 65130
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.985837922895358e-06,
+      "loss": 2.2675,
+      "step": 65140
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.983652417169334e-06,
+      "loss": 2.2822,
+      "step": 65150
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 7.981466911443308e-06,
+      "loss": 2.3297,
+      "step": 65160
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 7.979281405717284e-06,
+      "loss": 2.2931,
+      "step": 65170
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 7.977095899991258e-06,
+      "loss": 2.2695,
+      "step": 65180
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 7.974910394265234e-06,
+      "loss": 2.2623,
+      "step": 65190
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 7.972724888539208e-06,
+      "loss": 2.3125,
+      "step": 65200
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.970539382813184e-06,
+      "loss": 2.2521,
+      "step": 65210
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.968353877087158e-06,
+      "loss": 2.2916,
+      "step": 65220
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.966168371361134e-06,
+      "loss": 2.2936,
+      "step": 65230
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.963982865635108e-06,
+      "loss": 2.2573,
+      "step": 65240
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.961797359909084e-06,
+      "loss": 2.2902,
+      "step": 65250
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 7.95961185418306e-06,
+      "loss": 2.2704,
+      "step": 65260
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 7.957426348457034e-06,
+      "loss": 2.303,
+      "step": 65270
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 7.955240842731008e-06,
+      "loss": 2.3288,
+      "step": 65280
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 7.953055337004984e-06,
+      "loss": 2.2925,
+      "step": 65290
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 7.950869831278958e-06,
+      "loss": 2.2555,
+      "step": 65300
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.948684325552933e-06,
+      "loss": 2.2494,
+      "step": 65310
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.946498819826908e-06,
+      "loss": 2.287,
+      "step": 65320
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.944313314100883e-06,
+      "loss": 2.2715,
+      "step": 65330
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.942127808374858e-06,
+      "loss": 2.2564,
+      "step": 65340
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.939942302648833e-06,
+      "loss": 2.28,
+      "step": 65350
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 7.937756796922809e-06,
+      "loss": 2.2648,
+      "step": 65360
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 7.935571291196783e-06,
+      "loss": 2.2396,
+      "step": 65370
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 7.933385785470759e-06,
+      "loss": 2.2736,
+      "step": 65380
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 7.931200279744734e-06,
+      "loss": 2.3189,
+      "step": 65390
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 7.929014774018709e-06,
+      "loss": 2.2778,
+      "step": 65400
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.926829268292685e-06,
+      "loss": 2.2974,
+      "step": 65410
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.924643762566659e-06,
+      "loss": 2.2448,
+      "step": 65420
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.922458256840635e-06,
+      "loss": 2.3008,
+      "step": 65430
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.920272751114609e-06,
+      "loss": 2.2693,
+      "step": 65440
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.918087245388583e-06,
+      "loss": 2.2812,
+      "step": 65450
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 7.915901739662559e-06,
+      "loss": 2.2727,
+      "step": 65460
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 7.913716233936533e-06,
+      "loss": 2.2697,
+      "step": 65470
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 7.911530728210507e-06,
+      "loss": 2.2317,
+      "step": 65480
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 7.909345222484483e-06,
+      "loss": 2.2662,
+      "step": 65490
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 7.907159716758459e-06,
+      "loss": 2.3287,
+      "step": 65500
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.904974211032433e-06,
+      "loss": 2.2669,
+      "step": 65510
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.902788705306409e-06,
+      "loss": 2.2938,
+      "step": 65520
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.900603199580383e-06,
+      "loss": 2.2691,
+      "step": 65530
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.898417693854359e-06,
+      "loss": 2.2666,
+      "step": 65540
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.896232188128333e-06,
+      "loss": 2.3037,
+      "step": 65550
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 7.89404668240231e-06,
+      "loss": 2.281,
+      "step": 65560
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 7.891861176676283e-06,
+      "loss": 2.2544,
+      "step": 65570
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 7.88967567095026e-06,
+      "loss": 2.2481,
+      "step": 65580
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 7.887490165224233e-06,
+      "loss": 2.2456,
+      "step": 65590
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 7.88530465949821e-06,
+      "loss": 2.2591,
+      "step": 65600
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 7.883119153772183e-06,
+      "loss": 2.2811,
+      "step": 65610
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 7.880933648046158e-06,
+      "loss": 2.2909,
+      "step": 65620
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 7.878748142320134e-06,
+      "loss": 2.2702,
+      "step": 65630
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 7.876562636594108e-06,
+      "loss": 2.2685,
+      "step": 65640
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 7.874377130868084e-06,
+      "loss": 2.3319,
+      "step": 65650
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 7.872191625142058e-06,
+      "loss": 2.2863,
+      "step": 65660
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 7.870006119416034e-06,
+      "loss": 2.2823,
+      "step": 65670
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 7.867820613690008e-06,
+      "loss": 2.2876,
+      "step": 65680
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 7.865635107963984e-06,
+      "loss": 2.2775,
+      "step": 65690
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 7.863449602237958e-06,
+      "loss": 2.2588,
+      "step": 65700
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 7.861264096511934e-06,
+      "loss": 2.2901,
+      "step": 65710
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 7.859078590785908e-06,
+      "loss": 2.2779,
+      "step": 65720
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 7.856893085059884e-06,
+      "loss": 2.2904,
+      "step": 65730
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 7.85470757933386e-06,
+      "loss": 2.2663,
+      "step": 65740
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 7.852522073607834e-06,
+      "loss": 2.3104,
+      "step": 65750
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 7.850336567881808e-06,
+      "loss": 2.2906,
+      "step": 65760
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 7.848151062155784e-06,
+      "loss": 2.2615,
+      "step": 65770
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 7.845965556429758e-06,
+      "loss": 2.302,
+      "step": 65780
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 7.843780050703732e-06,
+      "loss": 2.2975,
+      "step": 65790
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 7.841594544977708e-06,
+      "loss": 2.277,
+      "step": 65800
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 7.839409039251682e-06,
+      "loss": 2.2698,
+      "step": 65810
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 7.837223533525658e-06,
+      "loss": 2.2669,
+      "step": 65820
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 7.835038027799632e-06,
+      "loss": 2.2667,
+      "step": 65830
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 7.832852522073608e-06,
+      "loss": 2.2495,
+      "step": 65840
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 7.830667016347584e-06,
+      "loss": 2.2922,
+      "step": 65850
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 7.828481510621558e-06,
+      "loss": 2.2599,
+      "step": 65860
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 7.826296004895534e-06,
+      "loss": 2.2825,
+      "step": 65870
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 7.824110499169508e-06,
+      "loss": 2.2881,
+      "step": 65880
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 7.821924993443484e-06,
+      "loss": 2.2544,
+      "step": 65890
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 7.819739487717459e-06,
+      "loss": 2.296,
+      "step": 65900
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 7.817553981991434e-06,
+      "loss": 2.2799,
+      "step": 65910
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 7.815368476265409e-06,
+      "loss": 2.2667,
+      "step": 65920
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 7.813182970539383e-06,
+      "loss": 2.3111,
+      "step": 65930
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 7.810997464813359e-06,
+      "loss": 2.3256,
+      "step": 65940
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 7.808811959087333e-06,
+      "loss": 2.2825,
+      "step": 65950
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 7.806626453361309e-06,
+      "loss": 2.2609,
+      "step": 65960
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 7.804440947635283e-06,
+      "loss": 2.2703,
+      "step": 65970
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 7.802255441909259e-06,
+      "loss": 2.3007,
+      "step": 65980
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 7.800069936183233e-06,
+      "loss": 2.3096,
+      "step": 65990
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 7.797884430457209e-06,
+      "loss": 2.3021,
+      "step": 66000
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 2.3104,
+      "step": 66010
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 7.793513419005159e-06,
+      "loss": 2.338,
+      "step": 66020
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 7.791327913279133e-06,
+      "loss": 2.2988,
+      "step": 66030
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 7.789142407553109e-06,
+      "loss": 2.2931,
+      "step": 66040
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 7.786956901827083e-06,
+      "loss": 2.3322,
+      "step": 66050
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 7.784771396101059e-06,
+      "loss": 2.2736,
+      "step": 66060
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.782585890375033e-06,
+      "loss": 2.2985,
+      "step": 66070
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.780400384649009e-06,
+      "loss": 2.2867,
+      "step": 66080
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.778214878922983e-06,
+      "loss": 2.2873,
+      "step": 66090
+    },
+    {
+      "epoch": 13.0,
+      "eval_BLEU": 10.1845,
+      "eval_gen_len": 12.3905,
+      "eval_loss": 3.2065279483795166,
+      "eval_runtime": 145.3086,
+      "eval_samples_per_second": 74.07,
+      "eval_steps_per_second": 0.296,
+      "step": 66092
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.776029373196957e-06,
+      "loss": 2.1958,
+      "step": 66100
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.773843867470933e-06,
+      "loss": 2.2172,
+      "step": 66110
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.771658361744908e-06,
+      "loss": 2.2055,
+      "step": 66120
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.769472856018883e-06,
+      "loss": 2.2114,
+      "step": 66130
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.767287350292858e-06,
+      "loss": 2.1948,
+      "step": 66140
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.765101844566834e-06,
+      "loss": 2.1998,
+      "step": 66150
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.762916338840808e-06,
+      "loss": 2.1951,
+      "step": 66160
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.760730833114784e-06,
+      "loss": 2.1728,
+      "step": 66170
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.758545327388758e-06,
+      "loss": 2.2108,
+      "step": 66180
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.756359821662734e-06,
+      "loss": 2.2181,
+      "step": 66190
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.75417431593671e-06,
+      "loss": 2.1897,
+      "step": 66200
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 7.751988810210684e-06,
+      "loss": 2.1549,
+      "step": 66210
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.74980330448466e-06,
+      "loss": 2.2214,
+      "step": 66220
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.747617798758634e-06,
+      "loss": 2.2168,
+      "step": 66230
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.745432293032608e-06,
+      "loss": 2.2106,
+      "step": 66240
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.743246787306584e-06,
+      "loss": 2.1959,
+      "step": 66250
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 7.741061281580558e-06,
+      "loss": 2.1666,
+      "step": 66260
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.738875775854532e-06,
+      "loss": 2.2058,
+      "step": 66270
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.736690270128508e-06,
+      "loss": 2.1582,
+      "step": 66280
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.734504764402482e-06,
+      "loss": 2.1893,
+      "step": 66290
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.732319258676458e-06,
+      "loss": 2.2173,
+      "step": 66300
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.730133752950434e-06,
+      "loss": 2.1981,
+      "step": 66310
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 7.727948247224408e-06,
+      "loss": 2.206,
+      "step": 66320
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 7.725762741498384e-06,
+      "loss": 2.2053,
+      "step": 66330
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 7.723577235772358e-06,
+      "loss": 2.1886,
+      "step": 66340
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 7.721391730046334e-06,
+      "loss": 2.1871,
+      "step": 66350
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 7.719206224320308e-06,
+      "loss": 2.2053,
+      "step": 66360
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 7.717020718594284e-06,
+      "loss": 2.19,
+      "step": 66370
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 7.714835212868258e-06,
+      "loss": 2.2454,
+      "step": 66380
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 7.712649707142234e-06,
+      "loss": 2.1856,
+      "step": 66390
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 7.710464201416208e-06,
+      "loss": 2.1922,
+      "step": 66400
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 7.708278695690183e-06,
+      "loss": 2.1857,
+      "step": 66410
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 7.706093189964159e-06,
+      "loss": 2.1953,
+      "step": 66420
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 7.703907684238133e-06,
+      "loss": 2.2275,
+      "step": 66430
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 7.701722178512109e-06,
+      "loss": 2.1967,
+      "step": 66440
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 7.699536672786083e-06,
+      "loss": 2.1805,
+      "step": 66450
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 7.697351167060059e-06,
+      "loss": 2.1725,
+      "step": 66460
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 7.695165661334033e-06,
+      "loss": 2.182,
+      "step": 66470
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 7.692980155608009e-06,
+      "loss": 2.1757,
+      "step": 66480
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 7.690794649881983e-06,
+      "loss": 2.1702,
+      "step": 66490
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 7.688609144155959e-06,
+      "loss": 2.1671,
+      "step": 66500
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 7.686423638429933e-06,
+      "loss": 2.1856,
+      "step": 66510
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 7.684238132703909e-06,
+      "loss": 2.1954,
+      "step": 66520
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 7.682052626977883e-06,
+      "loss": 2.1574,
+      "step": 66530
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 7.679867121251859e-06,
+      "loss": 2.2044,
+      "step": 66540
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 7.677681615525833e-06,
+      "loss": 2.1805,
+      "step": 66550
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 7.675496109799809e-06,
+      "loss": 2.2081,
+      "step": 66560
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 7.673310604073783e-06,
+      "loss": 2.1994,
+      "step": 66570
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 7.671125098347757e-06,
+      "loss": 2.2156,
+      "step": 66580
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 7.668939592621733e-06,
+      "loss": 2.1989,
+      "step": 66590
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 7.666754086895707e-06,
+      "loss": 2.2227,
+      "step": 66600
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 7.664568581169683e-06,
+      "loss": 2.1773,
+      "step": 66610
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 7.662383075443657e-06,
+      "loss": 2.2031,
+      "step": 66620
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 7.660197569717633e-06,
+      "loss": 2.1559,
+      "step": 66630
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 7.658012063991608e-06,
+      "loss": 2.194,
+      "step": 66640
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 7.655826558265583e-06,
+      "loss": 2.1866,
+      "step": 66650
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 7.65364105253956e-06,
+      "loss": 2.2143,
+      "step": 66660
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 7.651455546813533e-06,
+      "loss": 2.2024,
+      "step": 66670
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 7.64927004108751e-06,
+      "loss": 2.1992,
+      "step": 66680
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 7.647084535361484e-06,
+      "loss": 2.1575,
+      "step": 66690
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 7.64489902963546e-06,
+      "loss": 2.1875,
+      "step": 66700
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 7.642713523909434e-06,
+      "loss": 2.1924,
+      "step": 66710
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 7.640528018183408e-06,
+      "loss": 2.2207,
+      "step": 66720
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 7.638342512457384e-06,
+      "loss": 2.2054,
+      "step": 66730
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 7.636157006731358e-06,
+      "loss": 2.1805,
+      "step": 66740
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 7.633971501005332e-06,
+      "loss": 2.2049,
+      "step": 66750
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 7.631785995279308e-06,
+      "loss": 2.2409,
+      "step": 66760
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 7.629600489553284e-06,
+      "loss": 2.2121,
+      "step": 66770
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 7.627414983827258e-06,
+      "loss": 2.1868,
+      "step": 66780
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 7.625229478101234e-06,
+      "loss": 2.2322,
+      "step": 66790
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 7.623043972375208e-06,
+      "loss": 2.1973,
+      "step": 66800
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 7.620858466649184e-06,
+      "loss": 2.2301,
+      "step": 66810
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 7.618672960923158e-06,
+      "loss": 2.205,
+      "step": 66820
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 7.616487455197133e-06,
+      "loss": 2.1999,
+      "step": 66830
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 7.614301949471108e-06,
+      "loss": 2.1948,
+      "step": 66840
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 7.612116443745083e-06,
+      "loss": 2.2238,
+      "step": 66850
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 7.609930938019058e-06,
+      "loss": 2.1939,
+      "step": 66860
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 7.607745432293033e-06,
+      "loss": 2.2212,
+      "step": 66870
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 7.6055599265670074e-06,
+      "loss": 2.2147,
+      "step": 66880
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 7.603374420840983e-06,
+      "loss": 2.2215,
+      "step": 66890
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 7.601188915114958e-06,
+      "loss": 2.2094,
+      "step": 66900
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 7.599003409388933e-06,
+      "loss": 2.1909,
+      "step": 66910
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 7.596817903662908e-06,
+      "loss": 2.2183,
+      "step": 66920
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 7.594632397936883e-06,
+      "loss": 2.2141,
+      "step": 66930
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 7.5924468922108585e-06,
+      "loss": 2.1967,
+      "step": 66940
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 7.590261386484833e-06,
+      "loss": 2.1834,
+      "step": 66950
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 7.5880758807588085e-06,
+      "loss": 2.1905,
+      "step": 66960
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 7.585890375032783e-06,
+      "loss": 2.2088,
+      "step": 66970
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 7.583704869306759e-06,
+      "loss": 2.1827,
+      "step": 66980
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 7.581519363580733e-06,
+      "loss": 2.2131,
+      "step": 66990
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 7.579333857854708e-06,
+      "loss": 2.1827,
+      "step": 67000
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 7.577148352128684e-06,
+      "loss": 2.2508,
+      "step": 67010
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 7.574962846402658e-06,
+      "loss": 2.1967,
+      "step": 67020
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 7.572777340676634e-06,
+      "loss": 2.2155,
+      "step": 67030
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 7.570591834950608e-06,
+      "loss": 2.2006,
+      "step": 67040
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 7.568406329224584e-06,
+      "loss": 2.2222,
+      "step": 67050
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 7.566220823498558e-06,
+      "loss": 2.1743,
+      "step": 67060
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 7.564035317772533e-06,
+      "loss": 2.2299,
+      "step": 67070
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 7.561849812046508e-06,
+      "loss": 2.2082,
+      "step": 67080
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 7.559664306320483e-06,
+      "loss": 2.2089,
+      "step": 67090
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 7.557478800594458e-06,
+      "loss": 2.1787,
+      "step": 67100
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 7.555293294868433e-06,
+      "loss": 2.1948,
+      "step": 67110
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 7.553107789142409e-06,
+      "loss": 2.1733,
+      "step": 67120
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 7.550922283416383e-06,
+      "loss": 2.2439,
+      "step": 67130
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 7.548736777690358e-06,
+      "loss": 2.199,
+      "step": 67140
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 7.546551271964333e-06,
+      "loss": 2.2226,
+      "step": 67150
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 7.544365766238308e-06,
+      "loss": 2.1857,
+      "step": 67160
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 7.5421802605122825e-06,
+      "loss": 2.2186,
+      "step": 67170
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 7.539994754786258e-06,
+      "loss": 2.1585,
+      "step": 67180
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 7.5378092490602326e-06,
+      "loss": 2.235,
+      "step": 67190
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 7.5356237433342084e-06,
+      "loss": 2.2082,
+      "step": 67200
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 7.533438237608183e-06,
+      "loss": 2.1757,
+      "step": 67210
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 7.5312527318821585e-06,
+      "loss": 2.2314,
+      "step": 67220
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 7.529067226156133e-06,
+      "loss": 2.2168,
+      "step": 67230
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 7.526881720430108e-06,
+      "loss": 2.2148,
+      "step": 67240
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 7.524696214704084e-06,
+      "loss": 2.1675,
+      "step": 67250
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 7.522510708978058e-06,
+      "loss": 2.2038,
+      "step": 67260
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 7.520325203252034e-06,
+      "loss": 2.1698,
+      "step": 67270
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 7.518139697526008e-06,
+      "loss": 2.2132,
+      "step": 67280
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 7.515954191799984e-06,
+      "loss": 2.2034,
+      "step": 67290
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 7.513768686073958e-06,
+      "loss": 2.1715,
+      "step": 67300
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 7.511583180347933e-06,
+      "loss": 2.201,
+      "step": 67310
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 7.509397674621908e-06,
+      "loss": 2.2106,
+      "step": 67320
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 7.507212168895883e-06,
+      "loss": 2.2405,
+      "step": 67330
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 7.505026663169858e-06,
+      "loss": 2.1732,
+      "step": 67340
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 7.502841157443833e-06,
+      "loss": 2.1913,
+      "step": 67350
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 7.500655651717809e-06,
+      "loss": 2.2356,
+      "step": 67360
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 7.498470145991783e-06,
+      "loss": 2.1935,
+      "step": 67370
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 7.496284640265758e-06,
+      "loss": 2.1917,
+      "step": 67380
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 7.494099134539733e-06,
+      "loss": 2.1866,
+      "step": 67390
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 7.491913628813708e-06,
+      "loss": 2.2025,
+      "step": 67400
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 7.489728123087682e-06,
+      "loss": 2.1529,
+      "step": 67410
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 7.487542617361658e-06,
+      "loss": 2.1989,
+      "step": 67420
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 7.4853571116356325e-06,
+      "loss": 2.2196,
+      "step": 67430
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 7.483171605909608e-06,
+      "loss": 2.2465,
+      "step": 67440
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 7.4809861001835825e-06,
+      "loss": 2.2097,
+      "step": 67450
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 7.478800594457558e-06,
+      "loss": 2.1874,
+      "step": 67460
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 7.4766150887315335e-06,
+      "loss": 2.2153,
+      "step": 67470
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 7.474429583005508e-06,
+      "loss": 2.2068,
+      "step": 67480
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 7.4722440772794835e-06,
+      "loss": 2.1884,
+      "step": 67490
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 7.470058571553458e-06,
+      "loss": 2.2171,
+      "step": 67500
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 7.467873065827434e-06,
+      "loss": 2.1989,
+      "step": 67510
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 7.465687560101408e-06,
+      "loss": 2.2155,
+      "step": 67520
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 7.463502054375384e-06,
+      "loss": 2.2144,
+      "step": 67530
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 7.461316548649358e-06,
+      "loss": 2.2098,
+      "step": 67540
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 7.459131042923333e-06,
+      "loss": 2.1879,
+      "step": 67550
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 7.456945537197308e-06,
+      "loss": 2.1982,
+      "step": 67560
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 7.454760031471283e-06,
+      "loss": 2.2249,
+      "step": 67570
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 7.452574525745257e-06,
+      "loss": 2.2549,
+      "step": 67580
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 7.450389020019233e-06,
+      "loss": 2.2025,
+      "step": 67590
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 7.448203514293209e-06,
+      "loss": 2.1867,
+      "step": 67600
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 7.446018008567183e-06,
+      "loss": 2.2134,
+      "step": 67610
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 7.443832502841158e-06,
+      "loss": 2.1988,
+      "step": 67620
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 7.441646997115133e-06,
+      "loss": 2.1956,
+      "step": 67630
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 7.439461491389108e-06,
+      "loss": 2.2115,
+      "step": 67640
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 7.437275985663082e-06,
+      "loss": 2.2331,
+      "step": 67650
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 7.435090479937058e-06,
+      "loss": 2.2037,
+      "step": 67660
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 7.432904974211032e-06,
+      "loss": 2.2276,
+      "step": 67670
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 7.430719468485008e-06,
+      "loss": 2.1993,
+      "step": 67680
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 7.4285339627589825e-06,
+      "loss": 2.1835,
+      "step": 67690
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 7.426348457032958e-06,
+      "loss": 2.1946,
+      "step": 67700
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 7.424162951306933e-06,
+      "loss": 2.1611,
+      "step": 67710
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 7.4219774455809076e-06,
+      "loss": 2.1769,
+      "step": 67720
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 7.4197919398548834e-06,
+      "loss": 2.1908,
+      "step": 67730
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 7.417606434128858e-06,
+      "loss": 2.1828,
+      "step": 67740
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 7.4154209284028335e-06,
+      "loss": 2.2165,
+      "step": 67750
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 7.413235422676808e-06,
+      "loss": 2.2074,
+      "step": 67760
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 7.4110499169507836e-06,
+      "loss": 2.1848,
+      "step": 67770
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 7.408864411224758e-06,
+      "loss": 2.1948,
+      "step": 67780
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 7.406678905498733e-06,
+      "loss": 2.2002,
+      "step": 67790
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 7.404493399772708e-06,
+      "loss": 2.2086,
+      "step": 67800
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 7.402307894046683e-06,
+      "loss": 2.2251,
+      "step": 67810
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 7.400122388320659e-06,
+      "loss": 2.1731,
+      "step": 67820
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 7.397936882594633e-06,
+      "loss": 2.2281,
+      "step": 67830
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 7.395751376868609e-06,
+      "loss": 2.1787,
+      "step": 67840
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 7.393565871142583e-06,
+      "loss": 2.2129,
+      "step": 67850
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 7.391380365416558e-06,
+      "loss": 2.2369,
+      "step": 67860
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 7.389194859690533e-06,
+      "loss": 2.2166,
+      "step": 67870
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 7.387009353964508e-06,
+      "loss": 2.1948,
+      "step": 67880
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 7.384823848238482e-06,
+      "loss": 2.2509,
+      "step": 67890
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 7.382638342512458e-06,
+      "loss": 2.208,
+      "step": 67900
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 7.380452836786432e-06,
+      "loss": 2.1584,
+      "step": 67910
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 7.378267331060408e-06,
+      "loss": 2.1977,
+      "step": 67920
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 7.376081825334382e-06,
+      "loss": 2.2146,
+      "step": 67930
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 7.373896319608358e-06,
+      "loss": 2.2413,
+      "step": 67940
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 7.371710813882333e-06,
+      "loss": 2.237,
+      "step": 67950
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 7.3695253081563075e-06,
+      "loss": 2.2784,
+      "step": 67960
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 7.367339802430283e-06,
+      "loss": 2.2456,
+      "step": 67970
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 7.3651542967042575e-06,
+      "loss": 2.2076,
+      "step": 67980
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 7.362968790978233e-06,
+      "loss": 2.1959,
+      "step": 67990
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 7.360783285252208e-06,
+      "loss": 2.2227,
+      "step": 68000
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 7.3585977795261835e-06,
+      "loss": 2.2314,
+      "step": 68010
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 7.356412273800158e-06,
+      "loss": 2.1862,
+      "step": 68020
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 7.354226768074133e-06,
+      "loss": 2.2316,
+      "step": 68030
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 7.352041262348108e-06,
+      "loss": 2.1928,
+      "step": 68040
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 7.349855756622083e-06,
+      "loss": 2.1972,
+      "step": 68050
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 7.347670250896059e-06,
+      "loss": 2.1644,
+      "step": 68060
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 7.345484745170033e-06,
+      "loss": 2.2259,
+      "step": 68070
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 7.343299239444009e-06,
+      "loss": 2.1895,
+      "step": 68080
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 7.341113733717983e-06,
+      "loss": 2.2104,
+      "step": 68090
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 7.338928227991958e-06,
+      "loss": 2.246,
+      "step": 68100
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 7.336742722265933e-06,
+      "loss": 2.1927,
+      "step": 68110
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 7.334557216539908e-06,
+      "loss": 2.2217,
+      "step": 68120
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 7.332371710813882e-06,
+      "loss": 2.2399,
+      "step": 68130
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 7.330186205087858e-06,
+      "loss": 2.2079,
+      "step": 68140
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 7.328000699361832e-06,
+      "loss": 2.2288,
+      "step": 68150
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 7.325815193635808e-06,
+      "loss": 2.1916,
+      "step": 68160
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 7.323629687909783e-06,
+      "loss": 2.2101,
+      "step": 68170
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 7.321444182183758e-06,
+      "loss": 2.2053,
+      "step": 68180
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 7.319258676457733e-06,
+      "loss": 2.2278,
+      "step": 68190
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 2.2144,
+      "step": 68200
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 7.314887665005683e-06,
+      "loss": 2.2121,
+      "step": 68210
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 7.3127021592796574e-06,
+      "loss": 2.2244,
+      "step": 68220
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 7.310516653553633e-06,
+      "loss": 2.2508,
+      "step": 68230
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 7.3083311478276075e-06,
+      "loss": 2.2243,
+      "step": 68240
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 7.306145642101583e-06,
+      "loss": 2.2065,
+      "step": 68250
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 7.3039601363755576e-06,
+      "loss": 2.2303,
+      "step": 68260
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 7.301774630649533e-06,
+      "loss": 2.1835,
+      "step": 68270
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 7.299589124923508e-06,
+      "loss": 2.2181,
+      "step": 68280
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 7.297403619197483e-06,
+      "loss": 2.229,
+      "step": 68290
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 7.2952181134714585e-06,
+      "loss": 2.2086,
+      "step": 68300
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 7.293032607745433e-06,
+      "loss": 2.1653,
+      "step": 68310
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 7.290847102019409e-06,
+      "loss": 2.2228,
+      "step": 68320
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 7.288661596293383e-06,
+      "loss": 2.2085,
+      "step": 68330
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 7.286476090567358e-06,
+      "loss": 2.2089,
+      "step": 68340
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 7.284290584841333e-06,
+      "loss": 2.2441,
+      "step": 68350
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 7.282105079115308e-06,
+      "loss": 2.1877,
+      "step": 68360
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 7.279919573389282e-06,
+      "loss": 2.2212,
+      "step": 68370
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 7.277734067663258e-06,
+      "loss": 2.2408,
+      "step": 68380
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 7.275548561937232e-06,
+      "loss": 2.2089,
+      "step": 68390
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 7.273363056211208e-06,
+      "loss": 2.2161,
+      "step": 68400
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 7.271177550485183e-06,
+      "loss": 2.2318,
+      "step": 68410
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 7.268992044759158e-06,
+      "loss": 2.1898,
+      "step": 68420
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 7.266806539033133e-06,
+      "loss": 2.2194,
+      "step": 68430
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 7.264621033307107e-06,
+      "loss": 2.2329,
+      "step": 68440
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 7.262435527581083e-06,
+      "loss": 2.1799,
+      "step": 68450
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 7.260250021855057e-06,
+      "loss": 2.2379,
+      "step": 68460
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 7.258064516129033e-06,
+      "loss": 2.2122,
+      "step": 68470
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 7.255879010403007e-06,
+      "loss": 2.2308,
+      "step": 68480
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 7.253693504676983e-06,
+      "loss": 2.1915,
+      "step": 68490
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 7.2515079989509575e-06,
+      "loss": 2.2174,
+      "step": 68500
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 7.2493224932249325e-06,
+      "loss": 2.2285,
+      "step": 68510
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 7.247136987498908e-06,
+      "loss": 2.2166,
+      "step": 68520
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 7.244951481772883e-06,
+      "loss": 2.2328,
+      "step": 68530
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 7.2427659760468585e-06,
+      "loss": 2.2152,
+      "step": 68540
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 7.240580470320833e-06,
+      "loss": 2.2335,
+      "step": 68550
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 7.2383949645948085e-06,
+      "loss": 2.1828,
+      "step": 68560
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 7.236209458868783e-06,
+      "loss": 2.2796,
+      "step": 68570
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 7.234023953142758e-06,
+      "loss": 2.2136,
+      "step": 68580
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 7.231838447416733e-06,
+      "loss": 2.2346,
+      "step": 68590
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 7.229652941690708e-06,
+      "loss": 2.1964,
+      "step": 68600
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 7.227467435964682e-06,
+      "loss": 2.2396,
+      "step": 68610
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 7.225281930238658e-06,
+      "loss": 2.2066,
+      "step": 68620
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 7.223096424512632e-06,
+      "loss": 2.2209,
+      "step": 68630
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 7.220910918786608e-06,
+      "loss": 2.2539,
+      "step": 68640
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 7.218725413060583e-06,
+      "loss": 2.2511,
+      "step": 68650
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 7.216539907334558e-06,
+      "loss": 2.2092,
+      "step": 68660
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 7.214354401608533e-06,
+      "loss": 2.231,
+      "step": 68670
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 7.212168895882507e-06,
+      "loss": 2.2194,
+      "step": 68680
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 7.209983390156483e-06,
+      "loss": 2.1943,
+      "step": 68690
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 7.207797884430457e-06,
+      "loss": 2.1934,
+      "step": 68700
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 7.205612378704433e-06,
+      "loss": 2.2167,
+      "step": 68710
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 7.203426872978407e-06,
+      "loss": 2.1955,
+      "step": 68720
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 7.201241367252383e-06,
+      "loss": 2.1781,
+      "step": 68730
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 7.199055861526357e-06,
+      "loss": 2.2411,
+      "step": 68740
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 7.1968703558003324e-06,
+      "loss": 2.2071,
+      "step": 68750
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 7.194684850074308e-06,
+      "loss": 2.1798,
+      "step": 68760
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 7.1924993443482825e-06,
+      "loss": 2.2106,
+      "step": 68770
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 7.190313838622258e-06,
+      "loss": 2.2306,
+      "step": 68780
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 7.1881283328962326e-06,
+      "loss": 2.2498,
+      "step": 68790
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 7.1859428271702084e-06,
+      "loss": 2.2058,
+      "step": 68800
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 7.183757321444183e-06,
+      "loss": 2.1934,
+      "step": 68810
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 7.181571815718158e-06,
+      "loss": 2.2165,
+      "step": 68820
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 7.179386309992133e-06,
+      "loss": 2.2407,
+      "step": 68830
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 7.177200804266108e-06,
+      "loss": 2.2292,
+      "step": 68840
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 7.175015298540082e-06,
+      "loss": 2.2143,
+      "step": 68850
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 7.172829792814058e-06,
+      "loss": 2.2215,
+      "step": 68860
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 7.170644287088034e-06,
+      "loss": 2.2037,
+      "step": 68870
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 7.168458781362008e-06,
+      "loss": 2.2064,
+      "step": 68880
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 7.166273275635983e-06,
+      "loss": 2.2346,
+      "step": 68890
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 7.164087769909958e-06,
+      "loss": 2.2288,
+      "step": 68900
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 7.161902264183933e-06,
+      "loss": 2.2461,
+      "step": 68910
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 7.159716758457907e-06,
+      "loss": 2.1824,
+      "step": 68920
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 7.157531252731883e-06,
+      "loss": 2.2277,
+      "step": 68930
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 7.155345747005857e-06,
+      "loss": 2.196,
+      "step": 68940
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 7.153160241279833e-06,
+      "loss": 2.2128,
+      "step": 68950
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 7.150974735553807e-06,
+      "loss": 2.1971,
+      "step": 68960
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 7.148789229827783e-06,
+      "loss": 2.2615,
+      "step": 68970
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 7.146603724101757e-06,
+      "loss": 2.2481,
+      "step": 68980
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 7.144418218375732e-06,
+      "loss": 2.2125,
+      "step": 68990
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 7.142232712649708e-06,
+      "loss": 2.2483,
+      "step": 69000
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 7.140047206923682e-06,
+      "loss": 2.2528,
+      "step": 69010
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 7.137861701197658e-06,
+      "loss": 2.2826,
+      "step": 69020
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 7.1356761954716325e-06,
+      "loss": 2.2032,
+      "step": 69030
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 7.133490689745608e-06,
+      "loss": 2.2154,
+      "step": 69040
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 7.1313051840195825e-06,
+      "loss": 2.194,
+      "step": 69050
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 7.1291196782935576e-06,
+      "loss": 2.2367,
+      "step": 69060
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 7.126934172567533e-06,
+      "loss": 2.1794,
+      "step": 69070
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 7.124748666841508e-06,
+      "loss": 2.2691,
+      "step": 69080
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 7.122563161115482e-06,
+      "loss": 2.2405,
+      "step": 69090
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 7.120377655389458e-06,
+      "loss": 2.2445,
+      "step": 69100
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 7.1181921496634336e-06,
+      "loss": 2.2457,
+      "step": 69110
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 7.116006643937408e-06,
+      "loss": 2.2352,
+      "step": 69120
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 7.113821138211383e-06,
+      "loss": 2.1906,
+      "step": 69130
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 7.111635632485358e-06,
+      "loss": 2.2194,
+      "step": 69140
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 7.109450126759333e-06,
+      "loss": 2.2501,
+      "step": 69150
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 7.107264621033307e-06,
+      "loss": 2.2109,
+      "step": 69160
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 7.105079115307283e-06,
+      "loss": 2.2397,
+      "step": 69170
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 7.102893609581257e-06,
+      "loss": 2.197,
+      "step": 69180
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 7.100708103855233e-06,
+      "loss": 2.2347,
+      "step": 69190
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 7.098522598129207e-06,
+      "loss": 2.2283,
+      "step": 69200
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 7.096337092403183e-06,
+      "loss": 2.2343,
+      "step": 69210
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 7.094151586677158e-06,
+      "loss": 2.2391,
+      "step": 69220
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 7.091966080951132e-06,
+      "loss": 2.2356,
+      "step": 69230
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 7.089780575225108e-06,
+      "loss": 2.2553,
+      "step": 69240
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 7.087595069499082e-06,
+      "loss": 2.2308,
+      "step": 69250
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 7.085409563773058e-06,
+      "loss": 2.2214,
+      "step": 69260
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 7.083224058047032e-06,
+      "loss": 2.2015,
+      "step": 69270
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 7.081038552321008e-06,
+      "loss": 2.2429,
+      "step": 69280
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 7.0788530465949824e-06,
+      "loss": 2.2293,
+      "step": 69290
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 7.0766675408689575e-06,
+      "loss": 2.2039,
+      "step": 69300
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 7.0744820351429325e-06,
+      "loss": 2.2364,
+      "step": 69310
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 7.0722965294169075e-06,
+      "loss": 2.2612,
+      "step": 69320
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 7.070111023690883e-06,
+      "loss": 2.2245,
+      "step": 69330
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 7.067925517964858e-06,
+      "loss": 2.2394,
+      "step": 69340
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 7.0657400122388335e-06,
+      "loss": 2.2631,
+      "step": 69350
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 7.063554506512808e-06,
+      "loss": 2.1879,
+      "step": 69360
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 7.061369000786783e-06,
+      "loss": 2.2243,
+      "step": 69370
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 7.059183495060758e-06,
+      "loss": 2.2608,
+      "step": 69380
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 7.056997989334733e-06,
+      "loss": 2.2171,
+      "step": 69390
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 7.054812483608707e-06,
+      "loss": 2.2271,
+      "step": 69400
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 7.052626977882683e-06,
+      "loss": 2.2481,
+      "step": 69410
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 7.050441472156657e-06,
+      "loss": 2.2162,
+      "step": 69420
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 7.048255966430633e-06,
+      "loss": 2.2385,
+      "step": 69430
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 7.046070460704607e-06,
+      "loss": 2.2177,
+      "step": 69440
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 7.043884954978583e-06,
+      "loss": 2.1907,
+      "step": 69450
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 7.041699449252558e-06,
+      "loss": 2.2015,
+      "step": 69460
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 7.039513943526532e-06,
+      "loss": 2.2615,
+      "step": 69470
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 7.037328437800508e-06,
+      "loss": 2.2436,
+      "step": 69480
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 7.035142932074482e-06,
+      "loss": 2.254,
+      "step": 69490
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 7.032957426348458e-06,
+      "loss": 2.2727,
+      "step": 69500
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 7.030771920622432e-06,
+      "loss": 2.2396,
+      "step": 69510
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 7.028586414896408e-06,
+      "loss": 2.2432,
+      "step": 69520
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 7.026400909170382e-06,
+      "loss": 2.1848,
+      "step": 69530
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 7.024215403444357e-06,
+      "loss": 2.2666,
+      "step": 69540
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 7.022029897718332e-06,
+      "loss": 2.271,
+      "step": 69550
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 7.0198443919923075e-06,
+      "loss": 2.2395,
+      "step": 69560
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 7.017658886266283e-06,
+      "loss": 2.2124,
+      "step": 69570
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 7.0154733805402575e-06,
+      "loss": 2.2422,
+      "step": 69580
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 7.013287874814233e-06,
+      "loss": 2.2337,
+      "step": 69590
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 7.011102369088208e-06,
+      "loss": 2.2882,
+      "step": 69600
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 7.008916863362183e-06,
+      "loss": 2.2278,
+      "step": 69610
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 7.006731357636158e-06,
+      "loss": 2.2106,
+      "step": 69620
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 7.004545851910133e-06,
+      "loss": 2.2141,
+      "step": 69630
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 7.002360346184107e-06,
+      "loss": 2.2108,
+      "step": 69640
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 7.000174840458083e-06,
+      "loss": 2.2518,
+      "step": 69650
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 6.997989334732057e-06,
+      "loss": 2.1953,
+      "step": 69660
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 6.995803829006033e-06,
+      "loss": 2.2354,
+      "step": 69670
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 6.993618323280008e-06,
+      "loss": 2.2121,
+      "step": 69680
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 6.991432817553983e-06,
+      "loss": 2.2204,
+      "step": 69690
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 6.989247311827958e-06,
+      "loss": 2.2511,
+      "step": 69700
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 6.987061806101932e-06,
+      "loss": 2.2192,
+      "step": 69710
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 6.984876300375908e-06,
+      "loss": 2.2486,
+      "step": 69720
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 6.982690794649882e-06,
+      "loss": 2.2261,
+      "step": 69730
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 6.980505288923858e-06,
+      "loss": 2.2493,
+      "step": 69740
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 6.978319783197832e-06,
+      "loss": 2.2042,
+      "step": 69750
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 6.976134277471808e-06,
+      "loss": 2.2166,
+      "step": 69760
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 6.973948771745782e-06,
+      "loss": 2.2517,
+      "step": 69770
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 6.971763266019757e-06,
+      "loss": 2.2181,
+      "step": 69780
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 6.969577760293732e-06,
+      "loss": 2.2159,
+      "step": 69790
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 6.967392254567707e-06,
+      "loss": 2.2282,
+      "step": 69800
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 6.965206748841683e-06,
+      "loss": 2.2382,
+      "step": 69810
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 6.9630212431156574e-06,
+      "loss": 2.2156,
+      "step": 69820
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 6.960835737389633e-06,
+      "loss": 2.2034,
+      "step": 69830
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 6.9586502316636075e-06,
+      "loss": 2.2681,
+      "step": 69840
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 6.9564647259375825e-06,
+      "loss": 2.2225,
+      "step": 69850
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 6.9542792202115576e-06,
+      "loss": 2.2778,
+      "step": 69860
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 6.952093714485533e-06,
+      "loss": 2.2332,
+      "step": 69870
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 6.949908208759507e-06,
+      "loss": 2.2176,
+      "step": 69880
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 6.947722703033483e-06,
+      "loss": 2.2721,
+      "step": 69890
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 6.945537197307457e-06,
+      "loss": 2.2615,
+      "step": 69900
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 6.943351691581433e-06,
+      "loss": 2.2159,
+      "step": 69910
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 6.941166185855408e-06,
+      "loss": 2.1947,
+      "step": 69920
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 6.938980680129383e-06,
+      "loss": 2.2757,
+      "step": 69930
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 6.936795174403358e-06,
+      "loss": 2.2624,
+      "step": 69940
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 6.934609668677332e-06,
+      "loss": 2.2626,
+      "step": 69950
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 6.932424162951308e-06,
+      "loss": 2.2254,
+      "step": 69960
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 6.930238657225282e-06,
+      "loss": 2.2751,
+      "step": 69970
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 6.928053151499258e-06,
+      "loss": 2.2408,
+      "step": 69980
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 6.925867645773232e-06,
+      "loss": 2.2288,
+      "step": 69990
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 6.923682140047208e-06,
+      "loss": 2.215,
+      "step": 70000
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 6.921496634321182e-06,
+      "loss": 2.22,
+      "step": 70010
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 6.919311128595157e-06,
+      "loss": 2.2501,
+      "step": 70020
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 6.917125622869133e-06,
+      "loss": 2.2196,
+      "step": 70030
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 6.914940117143107e-06,
+      "loss": 2.2515,
+      "step": 70040
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 6.912754611417083e-06,
+      "loss": 2.2706,
+      "step": 70050
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 6.910569105691057e-06,
+      "loss": 2.2282,
+      "step": 70060
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 6.908383599965033e-06,
+      "loss": 2.2497,
+      "step": 70070
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 6.906198094239007e-06,
+      "loss": 2.2779,
+      "step": 70080
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 6.9040125885129824e-06,
+      "loss": 2.2229,
+      "step": 70090
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 6.9018270827869575e-06,
+      "loss": 2.2151,
+      "step": 70100
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 6.8996415770609325e-06,
+      "loss": 2.2259,
+      "step": 70110
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 6.897456071334907e-06,
+      "loss": 2.1885,
+      "step": 70120
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 6.8952705656088826e-06,
+      "loss": 2.2323,
+      "step": 70130
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 6.893085059882857e-06,
+      "loss": 2.2381,
+      "step": 70140
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 6.890899554156833e-06,
+      "loss": 2.2239,
+      "step": 70150
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 6.888714048430808e-06,
+      "loss": 2.2394,
+      "step": 70160
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 6.886528542704783e-06,
+      "loss": 2.2242,
+      "step": 70170
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 6.884343036978758e-06,
+      "loss": 2.228,
+      "step": 70180
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 6.882157531252732e-06,
+      "loss": 2.2249,
+      "step": 70190
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 6.879972025526708e-06,
+      "loss": 2.2542,
+      "step": 70200
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 6.877786519800682e-06,
+      "loss": 2.2212,
+      "step": 70210
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 6.875601014074658e-06,
+      "loss": 2.2594,
+      "step": 70220
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 6.873415508348632e-06,
+      "loss": 2.2476,
+      "step": 70230
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 6.871230002622608e-06,
+      "loss": 2.241,
+      "step": 70240
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 6.869044496896582e-06,
+      "loss": 2.2176,
+      "step": 70250
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 6.866858991170557e-06,
+      "loss": 2.2606,
+      "step": 70260
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 6.864673485444533e-06,
+      "loss": 2.2221,
+      "step": 70270
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 6.862487979718507e-06,
+      "loss": 2.2624,
+      "step": 70280
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 6.860302473992483e-06,
+      "loss": 2.2283,
+      "step": 70290
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 6.858116968266457e-06,
+      "loss": 2.2314,
+      "step": 70300
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 6.855931462540433e-06,
+      "loss": 2.2492,
+      "step": 70310
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 6.853745956814407e-06,
+      "loss": 2.2238,
+      "step": 70320
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 6.851560451088382e-06,
+      "loss": 2.1925,
+      "step": 70330
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 6.849374945362357e-06,
+      "loss": 2.1914,
+      "step": 70340
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 6.847189439636332e-06,
+      "loss": 2.2229,
+      "step": 70350
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 6.845003933910307e-06,
+      "loss": 2.2198,
+      "step": 70360
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 6.8428184281842825e-06,
+      "loss": 2.2623,
+      "step": 70370
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 6.840632922458258e-06,
+      "loss": 2.2648,
+      "step": 70380
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 6.8384474167322325e-06,
+      "loss": 2.2147,
+      "step": 70390
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 6.8362619110062076e-06,
+      "loss": 2.2403,
+      "step": 70400
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 6.834076405280183e-06,
+      "loss": 2.2275,
+      "step": 70410
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 6.831890899554158e-06,
+      "loss": 2.2438,
+      "step": 70420
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 6.829705393828132e-06,
+      "loss": 2.1818,
+      "step": 70430
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 6.827519888102108e-06,
+      "loss": 2.2114,
+      "step": 70440
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 6.825334382376082e-06,
+      "loss": 2.2436,
+      "step": 70450
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 6.823148876650058e-06,
+      "loss": 2.2169,
+      "step": 70460
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 6.820963370924032e-06,
+      "loss": 2.1971,
+      "step": 70470
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 6.818777865198008e-06,
+      "loss": 2.2402,
+      "step": 70480
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 6.816592359471982e-06,
+      "loss": 2.2666,
+      "step": 70490
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 6.814406853745957e-06,
+      "loss": 2.1854,
+      "step": 70500
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 6.812221348019933e-06,
+      "loss": 2.2245,
+      "step": 70510
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 6.810035842293907e-06,
+      "loss": 2.2337,
+      "step": 70520
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 6.807850336567883e-06,
+      "loss": 2.2234,
+      "step": 70530
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 6.805664830841857e-06,
+      "loss": 2.2265,
+      "step": 70540
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 6.803479325115833e-06,
+      "loss": 2.2052,
+      "step": 70550
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 6.801293819389807e-06,
+      "loss": 2.2439,
+      "step": 70560
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 6.799108313663782e-06,
+      "loss": 2.1987,
+      "step": 70570
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 6.796922807937757e-06,
+      "loss": 2.2248,
+      "step": 70580
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 6.794737302211732e-06,
+      "loss": 2.2383,
+      "step": 70590
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 6.7925517964857065e-06,
+      "loss": 2.2197,
+      "step": 70600
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 6.790366290759682e-06,
+      "loss": 2.2396,
+      "step": 70610
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 6.7881807850336574e-06,
+      "loss": 2.2336,
+      "step": 70620
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 6.7859952793076325e-06,
+      "loss": 2.2096,
+      "step": 70630
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 6.7838097735816075e-06,
+      "loss": 2.2296,
+      "step": 70640
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 6.7816242678555825e-06,
+      "loss": 2.2478,
+      "step": 70650
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 6.7794387621295576e-06,
+      "loss": 2.1818,
+      "step": 70660
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 6.777253256403532e-06,
+      "loss": 2.2816,
+      "step": 70670
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 6.775067750677508e-06,
+      "loss": 2.2586,
+      "step": 70680
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 6.772882244951482e-06,
+      "loss": 2.2299,
+      "step": 70690
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 6.770696739225458e-06,
+      "loss": 2.2229,
+      "step": 70700
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 6.768511233499432e-06,
+      "loss": 2.267,
+      "step": 70710
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 6.766325727773408e-06,
+      "loss": 2.2497,
+      "step": 70720
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 6.764140222047383e-06,
+      "loss": 2.223,
+      "step": 70730
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 6.761954716321357e-06,
+      "loss": 2.2431,
+      "step": 70740
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 6.759769210595333e-06,
+      "loss": 2.2586,
+      "step": 70750
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 6.757583704869307e-06,
+      "loss": 2.2443,
+      "step": 70760
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 6.755398199143283e-06,
+      "loss": 2.2562,
+      "step": 70770
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 6.753212693417257e-06,
+      "loss": 2.2417,
+      "step": 70780
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 6.751027187691233e-06,
+      "loss": 2.2193,
+      "step": 70790
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 6.748841681965207e-06,
+      "loss": 2.2265,
+      "step": 70800
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 6.746656176239182e-06,
+      "loss": 2.249,
+      "step": 70810
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 6.744470670513157e-06,
+      "loss": 2.253,
+      "step": 70820
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 6.742285164787132e-06,
+      "loss": 2.2383,
+      "step": 70830
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 6.7400996590611064e-06,
+      "loss": 2.2514,
+      "step": 70840
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 6.737914153335082e-06,
+      "loss": 2.23,
+      "step": 70850
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 6.735728647609057e-06,
+      "loss": 2.2322,
+      "step": 70860
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 6.733543141883032e-06,
+      "loss": 2.2719,
+      "step": 70870
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 6.731357636157007e-06,
+      "loss": 2.2239,
+      "step": 70880
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 6.7291721304309824e-06,
+      "loss": 2.2131,
+      "step": 70890
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 6.7269866247049575e-06,
+      "loss": 2.2323,
+      "step": 70900
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 6.724801118978932e-06,
+      "loss": 2.2673,
+      "step": 70910
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 6.7226156132529075e-06,
+      "loss": 2.2173,
+      "step": 70920
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 6.720430107526882e-06,
+      "loss": 2.2673,
+      "step": 70930
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 6.718244601800858e-06,
+      "loss": 2.2414,
+      "step": 70940
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 6.716059096074832e-06,
+      "loss": 2.2432,
+      "step": 70950
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 6.713873590348808e-06,
+      "loss": 2.2044,
+      "step": 70960
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 6.711688084622783e-06,
+      "loss": 2.2254,
+      "step": 70970
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 6.709502578896757e-06,
+      "loss": 2.2407,
+      "step": 70980
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 6.707317073170733e-06,
+      "loss": 2.2449,
+      "step": 70990
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 6.705131567444707e-06,
+      "loss": 2.2386,
+      "step": 71000
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 6.702946061718683e-06,
+      "loss": 2.2456,
+      "step": 71010
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 6.700760555992657e-06,
+      "loss": 2.2473,
+      "step": 71020
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 6.698575050266633e-06,
+      "loss": 2.2463,
+      "step": 71030
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 6.696389544540607e-06,
+      "loss": 2.229,
+      "step": 71040
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 6.694204038814582e-06,
+      "loss": 2.2225,
+      "step": 71050
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 6.692018533088557e-06,
+      "loss": 2.2369,
+      "step": 71060
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 6.689833027362532e-06,
+      "loss": 2.1908,
+      "step": 71070
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 6.687647521636508e-06,
+      "loss": 2.2123,
+      "step": 71080
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 6.685462015910482e-06,
+      "loss": 2.2293,
+      "step": 71090
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 6.683276510184457e-06,
+      "loss": 2.2659,
+      "step": 71100
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 6.681091004458432e-06,
+      "loss": 2.2342,
+      "step": 71110
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 6.678905498732407e-06,
+      "loss": 2.197,
+      "step": 71120
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 6.676719993006382e-06,
+      "loss": 2.2664,
+      "step": 71130
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 6.674534487280357e-06,
+      "loss": 2.2067,
+      "step": 71140
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 6.6723489815543316e-06,
+      "loss": 2.2088,
+      "step": 71150
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 6.6701634758283074e-06,
+      "loss": 2.2746,
+      "step": 71160
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 6.667977970102282e-06,
+      "loss": 2.2552,
+      "step": 71170
+    },
+    {
+      "epoch": 14.0,
+      "eval_BLEU": 9.8647,
+      "eval_gen_len": 11.9744,
+      "eval_loss": 3.2441341876983643,
+      "eval_runtime": 163.2403,
+      "eval_samples_per_second": 65.933,
+      "eval_steps_per_second": 0.263,
+      "step": 71176
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 6.6657924643762575e-06,
+      "loss": 2.1917,
+      "step": 71180
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 6.663606958650232e-06,
+      "loss": 2.1732,
+      "step": 71190
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 6.6614214529242076e-06,
+      "loss": 2.1495,
+      "step": 71200
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.659235947198183e-06,
+      "loss": 2.1532,
+      "step": 71210
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.657050441472157e-06,
+      "loss": 2.1747,
+      "step": 71220
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.654864935746133e-06,
+      "loss": 2.1893,
+      "step": 71230
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.652679430020107e-06,
+      "loss": 2.1474,
+      "step": 71240
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 6.650493924294083e-06,
+      "loss": 2.1716,
+      "step": 71250
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.648308418568057e-06,
+      "loss": 2.1803,
+      "step": 71260
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.646122912842033e-06,
+      "loss": 2.1742,
+      "step": 71270
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.643937407116007e-06,
+      "loss": 2.1429,
+      "step": 71280
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.641751901389982e-06,
+      "loss": 2.116,
+      "step": 71290
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.639566395663957e-06,
+      "loss": 2.157,
+      "step": 71300
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.637380889937932e-06,
+      "loss": 2.1416,
+      "step": 71310
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.635195384211908e-06,
+      "loss": 2.1385,
+      "step": 71320
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.633009878485882e-06,
+      "loss": 2.1621,
+      "step": 71330
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.630824372759857e-06,
+      "loss": 2.1596,
+      "step": 71340
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 6.628638867033832e-06,
+      "loss": 2.1514,
+      "step": 71350
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.626453361307807e-06,
+      "loss": 2.1168,
+      "step": 71360
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.624267855581782e-06,
+      "loss": 2.1494,
+      "step": 71370
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.622082349855757e-06,
+      "loss": 2.1696,
+      "step": 71380
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.6198968441297315e-06,
+      "loss": 2.1544,
+      "step": 71390
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 6.617711338403707e-06,
+      "loss": 2.159,
+      "step": 71400
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.6155258326776815e-06,
+      "loss": 2.1507,
+      "step": 71410
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.613340326951657e-06,
+      "loss": 2.1267,
+      "step": 71420
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.6111548212256325e-06,
+      "loss": 2.1304,
+      "step": 71430
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.6089693154996075e-06,
+      "loss": 2.1447,
+      "step": 71440
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.6067838097735825e-06,
+      "loss": 2.1413,
+      "step": 71450
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 6.604598304047557e-06,
+      "loss": 2.1552,
+      "step": 71460
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 6.602412798321533e-06,
+      "loss": 2.1402,
+      "step": 71470
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 6.600227292595507e-06,
+      "loss": 2.1582,
+      "step": 71480
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 6.598041786869483e-06,
+      "loss": 2.1684,
+      "step": 71490
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 6.595856281143457e-06,
+      "loss": 2.1755,
+      "step": 71500
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 6.593670775417432e-06,
+      "loss": 2.1785,
+      "step": 71510
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 6.591485269691407e-06,
+      "loss": 2.1313,
+      "step": 71520
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 6.589299763965382e-06,
+      "loss": 2.1459,
+      "step": 71530
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 6.587114258239357e-06,
+      "loss": 2.1577,
+      "step": 71540
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 6.584928752513332e-06,
+      "loss": 2.1885,
+      "step": 71550
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 6.582743246787308e-06,
+      "loss": 2.1729,
+      "step": 71560
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 6.580557741061282e-06,
+      "loss": 2.1251,
+      "step": 71570
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 6.578372235335257e-06,
+      "loss": 2.1816,
+      "step": 71580
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 6.576186729609232e-06,
+      "loss": 2.1716,
+      "step": 71590
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 6.574001223883207e-06,
+      "loss": 2.1505,
+      "step": 71600
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 6.571815718157182e-06,
+      "loss": 2.1644,
+      "step": 71610
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 6.569630212431157e-06,
+      "loss": 2.1106,
+      "step": 71620
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 6.567444706705131e-06,
+      "loss": 2.148,
+      "step": 71630
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 6.565259200979107e-06,
+      "loss": 2.1336,
+      "step": 71640
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 6.5630736952530815e-06,
+      "loss": 2.1427,
+      "step": 71650
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 6.560888189527057e-06,
+      "loss": 2.1787,
+      "step": 71660
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 6.558702683801032e-06,
+      "loss": 2.1799,
+      "step": 71670
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 6.556517178075007e-06,
+      "loss": 2.1362,
+      "step": 71680
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 6.5543316723489824e-06,
+      "loss": 2.1641,
+      "step": 71690
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 6.552146166622957e-06,
+      "loss": 2.1595,
+      "step": 71700
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 6.5499606608969325e-06,
+      "loss": 2.1697,
+      "step": 71710
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 6.547775155170907e-06,
+      "loss": 2.1493,
+      "step": 71720
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 6.5455896494448826e-06,
+      "loss": 2.1482,
+      "step": 71730
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 6.543404143718857e-06,
+      "loss": 2.1481,
+      "step": 71740
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 6.541218637992832e-06,
+      "loss": 2.1594,
+      "step": 71750
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 6.539033132266807e-06,
+      "loss": 2.1341,
+      "step": 71760
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 6.536847626540782e-06,
+      "loss": 2.1415,
+      "step": 71770
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 6.534662120814758e-06,
+      "loss": 2.1325,
+      "step": 71780
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 6.532476615088732e-06,
+      "loss": 2.1617,
+      "step": 71790
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 6.530291109362708e-06,
+      "loss": 2.1265,
+      "step": 71800
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 6.528105603636682e-06,
+      "loss": 2.1613,
+      "step": 71810
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 6.525920097910657e-06,
+      "loss": 2.1248,
+      "step": 71820
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 6.523734592184632e-06,
+      "loss": 2.1464,
+      "step": 71830
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 6.521549086458607e-06,
+      "loss": 2.1574,
+      "step": 71840
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 6.519363580732582e-06,
+      "loss": 2.1586,
+      "step": 71850
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 6.517178075006557e-06,
+      "loss": 2.1403,
+      "step": 71860
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 6.514992569280531e-06,
+      "loss": 2.1586,
+      "step": 71870
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 6.512807063554507e-06,
+      "loss": 2.1644,
+      "step": 71880
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 6.510621557828482e-06,
+      "loss": 2.1338,
+      "step": 71890
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 6.508436052102457e-06,
+      "loss": 2.1121,
+      "step": 71900
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 6.506250546376432e-06,
+      "loss": 2.1313,
+      "step": 71910
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 2.1471,
+      "step": 71920
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 6.501879534924382e-06,
+      "loss": 2.1648,
+      "step": 71930
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 6.4996940291983565e-06,
+      "loss": 2.1665,
+      "step": 71940
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 6.497508523472332e-06,
+      "loss": 2.149,
+      "step": 71950
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 6.495323017746307e-06,
+      "loss": 2.1666,
+      "step": 71960
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 6.4931375120202825e-06,
+      "loss": 2.172,
+      "step": 71970
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 6.490952006294257e-06,
+      "loss": 2.171,
+      "step": 71980
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 6.488766500568232e-06,
+      "loss": 2.1351,
+      "step": 71990
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 6.486580994842207e-06,
+      "loss": 2.1951,
+      "step": 72000
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 6.484395489116182e-06,
+      "loss": 2.1821,
+      "step": 72010
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 6.482209983390158e-06,
+      "loss": 2.1166,
+      "step": 72020
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 6.480024477664132e-06,
+      "loss": 2.1646,
+      "step": 72030
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 6.477838971938108e-06,
+      "loss": 2.172,
+      "step": 72040
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 6.475653466212082e-06,
+      "loss": 2.2132,
+      "step": 72050
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 6.473467960486057e-06,
+      "loss": 2.1341,
+      "step": 72060
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 6.471282454760032e-06,
+      "loss": 2.1297,
+      "step": 72070
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 6.469096949034007e-06,
+      "loss": 2.1303,
+      "step": 72080
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 6.466911443307982e-06,
+      "loss": 2.1367,
+      "step": 72090
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 6.464725937581957e-06,
+      "loss": 2.1892,
+      "step": 72100
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 6.462540431855931e-06,
+      "loss": 2.1485,
+      "step": 72110
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 6.460354926129907e-06,
+      "loss": 2.1484,
+      "step": 72120
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 6.458169420403882e-06,
+      "loss": 2.1709,
+      "step": 72130
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 6.455983914677857e-06,
+      "loss": 2.1599,
+      "step": 72140
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 6.453798408951832e-06,
+      "loss": 2.1484,
+      "step": 72150
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 2.1627,
+      "step": 72160
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 6.449427397499782e-06,
+      "loss": 2.1807,
+      "step": 72170
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 6.4472418917737564e-06,
+      "loss": 2.1773,
+      "step": 72180
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 6.445056386047732e-06,
+      "loss": 2.1777,
+      "step": 72190
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 6.4428708803217065e-06,
+      "loss": 2.1709,
+      "step": 72200
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 6.440685374595682e-06,
+      "loss": 2.1881,
+      "step": 72210
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 6.4384998688696566e-06,
+      "loss": 2.1639,
+      "step": 72220
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 6.436314363143632e-06,
+      "loss": 2.1555,
+      "step": 72230
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 6.4341288574176075e-06,
+      "loss": 2.2046,
+      "step": 72240
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 6.431943351691582e-06,
+      "loss": 2.1927,
+      "step": 72250
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 6.4297578459655575e-06,
+      "loss": 2.1447,
+      "step": 72260
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 6.427572340239532e-06,
+      "loss": 2.1747,
+      "step": 72270
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 6.425386834513508e-06,
+      "loss": 2.1746,
+      "step": 72280
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 6.423201328787482e-06,
+      "loss": 2.1962,
+      "step": 72290
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 6.421015823061457e-06,
+      "loss": 2.1951,
+      "step": 72300
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 6.418830317335432e-06,
+      "loss": 2.1731,
+      "step": 72310
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 6.416644811609407e-06,
+      "loss": 2.1558,
+      "step": 72320
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 6.414459305883382e-06,
+      "loss": 2.1789,
+      "step": 72330
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 6.412273800157357e-06,
+      "loss": 2.1704,
+      "step": 72340
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 6.410088294431331e-06,
+      "loss": 2.1814,
+      "step": 72350
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 6.407902788705307e-06,
+      "loss": 2.1941,
+      "step": 72360
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 6.405717282979282e-06,
+      "loss": 2.2071,
+      "step": 72370
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 6.403531777253257e-06,
+      "loss": 2.2166,
+      "step": 72380
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 6.401346271527232e-06,
+      "loss": 2.1744,
+      "step": 72390
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 6.399160765801206e-06,
+      "loss": 2.1696,
+      "step": 72400
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 6.396975260075182e-06,
+      "loss": 2.1473,
+      "step": 72410
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 6.394789754349156e-06,
+      "loss": 2.1613,
+      "step": 72420
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 6.392604248623132e-06,
+      "loss": 2.1718,
+      "step": 72430
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 6.390418742897106e-06,
+      "loss": 2.1646,
+      "step": 72440
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 6.388233237171082e-06,
+      "loss": 2.1511,
+      "step": 72450
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 6.3860477314450565e-06,
+      "loss": 2.1993,
+      "step": 72460
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 6.3838622257190315e-06,
+      "loss": 2.1493,
+      "step": 72470
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 6.381676719993007e-06,
+      "loss": 2.2049,
+      "step": 72480
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 6.379491214266982e-06,
+      "loss": 2.1909,
+      "step": 72490
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 6.3773057085409575e-06,
+      "loss": 2.1591,
+      "step": 72500
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 6.375120202814932e-06,
+      "loss": 2.1716,
+      "step": 72510
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 6.3729346970889075e-06,
+      "loss": 2.1669,
+      "step": 72520
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 6.370749191362882e-06,
+      "loss": 2.1538,
+      "step": 72530
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 6.368563685636857e-06,
+      "loss": 2.1762,
+      "step": 72540
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 6.366378179910832e-06,
+      "loss": 2.1597,
+      "step": 72550
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 6.364192674184807e-06,
+      "loss": 2.1941,
+      "step": 72560
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 6.362007168458782e-06,
+      "loss": 2.1725,
+      "step": 72570
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 6.359821662732757e-06,
+      "loss": 2.1364,
+      "step": 72580
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 6.357636157006733e-06,
+      "loss": 2.1862,
+      "step": 72590
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 6.355450651280707e-06,
+      "loss": 2.1429,
+      "step": 72600
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 6.353265145554682e-06,
+      "loss": 2.1649,
+      "step": 72610
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 6.351079639828657e-06,
+      "loss": 2.1745,
+      "step": 72620
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 6.348894134102632e-06,
+      "loss": 2.1626,
+      "step": 72630
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 6.346708628376606e-06,
+      "loss": 2.1701,
+      "step": 72640
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 6.344523122650582e-06,
+      "loss": 2.1931,
+      "step": 72650
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 6.342337616924556e-06,
+      "loss": 2.13,
+      "step": 72660
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 6.340152111198532e-06,
+      "loss": 2.1891,
+      "step": 72670
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 6.337966605472506e-06,
+      "loss": 2.1318,
+      "step": 72680
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 6.335781099746482e-06,
+      "loss": 2.1295,
+      "step": 72690
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 6.333595594020456e-06,
+      "loss": 2.1629,
+      "step": 72700
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 6.3314100882944314e-06,
+      "loss": 2.1542,
+      "step": 72710
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 6.329224582568407e-06,
+      "loss": 2.1499,
+      "step": 72720
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 6.3270390768423815e-06,
+      "loss": 2.1624,
+      "step": 72730
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 6.324853571116357e-06,
+      "loss": 2.1556,
+      "step": 72740
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 6.3226680653903316e-06,
+      "loss": 2.1527,
+      "step": 72750
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 6.3204825596643074e-06,
+      "loss": 2.1589,
+      "step": 72760
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 6.318297053938282e-06,
+      "loss": 2.1808,
+      "step": 72770
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 6.316111548212257e-06,
+      "loss": 2.1638,
+      "step": 72780
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 6.313926042486232e-06,
+      "loss": 2.1816,
+      "step": 72790
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 6.311740536760207e-06,
+      "loss": 2.1986,
+      "step": 72800
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 6.309555031034182e-06,
+      "loss": 2.187,
+      "step": 72810
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 6.307369525308157e-06,
+      "loss": 2.2036,
+      "step": 72820
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 6.305184019582133e-06,
+      "loss": 2.1627,
+      "step": 72830
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 6.302998513856107e-06,
+      "loss": 2.1762,
+      "step": 72840
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 6.300813008130082e-06,
+      "loss": 2.1518,
+      "step": 72850
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 6.298627502404057e-06,
+      "loss": 2.1702,
+      "step": 72860
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 6.296441996678032e-06,
+      "loss": 2.1601,
+      "step": 72870
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 6.294256490952006e-06,
+      "loss": 2.1616,
+      "step": 72880
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 6.292070985225982e-06,
+      "loss": 2.1604,
+      "step": 72890
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 6.289885479499956e-06,
+      "loss": 2.1326,
+      "step": 72900
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 6.287699973773932e-06,
+      "loss": 2.1634,
+      "step": 72910
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 6.285514468047906e-06,
+      "loss": 2.1849,
+      "step": 72920
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 6.283328962321882e-06,
+      "loss": 2.1158,
+      "step": 72930
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 6.281143456595857e-06,
+      "loss": 2.1504,
+      "step": 72940
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 6.278957950869831e-06,
+      "loss": 2.1523,
+      "step": 72950
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 6.276772445143807e-06,
+      "loss": 2.1494,
+      "step": 72960
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 6.274586939417781e-06,
+      "loss": 2.184,
+      "step": 72970
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 6.272401433691757e-06,
+      "loss": 2.1946,
+      "step": 72980
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 6.2702159279657315e-06,
+      "loss": 2.1585,
+      "step": 72990
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 6.268030422239707e-06,
+      "loss": 2.2099,
+      "step": 73000
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 6.2658449165136815e-06,
+      "loss": 2.151,
+      "step": 73010
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 6.2636594107876566e-06,
+      "loss": 2.1975,
+      "step": 73020
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 6.261473905061632e-06,
+      "loss": 2.1731,
+      "step": 73030
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 6.259288399335607e-06,
+      "loss": 2.1916,
+      "step": 73040
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 6.257102893609581e-06,
+      "loss": 2.1735,
+      "step": 73050
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 6.254917387883557e-06,
+      "loss": 2.2131,
+      "step": 73060
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 6.2527318821575326e-06,
+      "loss": 2.1647,
+      "step": 73070
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 6.250546376431507e-06,
+      "loss": 2.1885,
+      "step": 73080
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 6.248360870705482e-06,
+      "loss": 2.1716,
+      "step": 73090
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 6.246175364979457e-06,
+      "loss": 2.1974,
+      "step": 73100
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 6.243989859253432e-06,
+      "loss": 2.1709,
+      "step": 73110
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 6.241804353527406e-06,
+      "loss": 2.1705,
+      "step": 73120
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 6.239618847801382e-06,
+      "loss": 2.2058,
+      "step": 73130
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 6.237433342075356e-06,
+      "loss": 2.1556,
+      "step": 73140
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 6.235247836349332e-06,
+      "loss": 2.1934,
+      "step": 73150
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 6.233062330623306e-06,
+      "loss": 2.1584,
+      "step": 73160
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 6.230876824897282e-06,
+      "loss": 2.1738,
+      "step": 73170
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 6.228691319171257e-06,
+      "loss": 2.1986,
+      "step": 73180
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 6.226505813445231e-06,
+      "loss": 2.1943,
+      "step": 73190
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 6.224320307719207e-06,
+      "loss": 2.1543,
+      "step": 73200
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 6.222134801993181e-06,
+      "loss": 2.1712,
+      "step": 73210
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 6.219949296267157e-06,
+      "loss": 2.1591,
+      "step": 73220
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 6.217763790541131e-06,
+      "loss": 2.1714,
+      "step": 73230
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 6.215578284815107e-06,
+      "loss": 2.154,
+      "step": 73240
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 6.2133927790890814e-06,
+      "loss": 2.1801,
+      "step": 73250
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 6.2112072733630565e-06,
+      "loss": 2.1247,
+      "step": 73260
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 6.2090217676370315e-06,
+      "loss": 2.1541,
+      "step": 73270
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 6.2068362619110065e-06,
+      "loss": 2.1687,
+      "step": 73280
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 6.204650756184982e-06,
+      "loss": 2.2033,
+      "step": 73290
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 6.202465250458957e-06,
+      "loss": 2.1933,
+      "step": 73300
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 6.2002797447329325e-06,
+      "loss": 2.1815,
+      "step": 73310
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 6.198094239006907e-06,
+      "loss": 2.1461,
+      "step": 73320
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 6.195908733280882e-06,
+      "loss": 2.1864,
+      "step": 73330
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 6.193723227554857e-06,
+      "loss": 2.1696,
+      "step": 73340
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 6.191537721828832e-06,
+      "loss": 2.1479,
+      "step": 73350
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 6.189352216102806e-06,
+      "loss": 2.1548,
+      "step": 73360
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 6.187166710376782e-06,
+      "loss": 2.1946,
+      "step": 73370
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 6.184981204650756e-06,
+      "loss": 2.185,
+      "step": 73380
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 6.182795698924732e-06,
+      "loss": 2.1857,
+      "step": 73390
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 6.180610193198706e-06,
+      "loss": 2.1645,
+      "step": 73400
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 6.178424687472682e-06,
+      "loss": 2.1683,
+      "step": 73410
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 6.176239181746657e-06,
+      "loss": 2.1743,
+      "step": 73420
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 6.174053676020631e-06,
+      "loss": 2.1847,
+      "step": 73430
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 6.171868170294607e-06,
+      "loss": 2.1872,
+      "step": 73440
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 6.169682664568581e-06,
+      "loss": 2.1698,
+      "step": 73450
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 6.167497158842557e-06,
+      "loss": 2.1984,
+      "step": 73460
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 6.165311653116531e-06,
+      "loss": 2.1589,
+      "step": 73470
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 6.163126147390507e-06,
+      "loss": 2.1808,
+      "step": 73480
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 6.160940641664481e-06,
+      "loss": 2.2112,
+      "step": 73490
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 6.158755135938456e-06,
+      "loss": 2.1917,
+      "step": 73500
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 6.156569630212431e-06,
+      "loss": 2.1885,
+      "step": 73510
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 6.1543841244864065e-06,
+      "loss": 2.1827,
+      "step": 73520
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 6.152198618760382e-06,
+      "loss": 2.1762,
+      "step": 73530
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 6.1500131130343565e-06,
+      "loss": 2.1739,
+      "step": 73540
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 6.147827607308332e-06,
+      "loss": 2.1771,
+      "step": 73550
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 6.145642101582307e-06,
+      "loss": 2.2171,
+      "step": 73560
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 6.143456595856282e-06,
+      "loss": 2.1892,
+      "step": 73570
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 6.141271090130257e-06,
+      "loss": 2.1792,
+      "step": 73580
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 6.139085584404232e-06,
+      "loss": 2.1814,
+      "step": 73590
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 6.136900078678206e-06,
+      "loss": 2.1882,
+      "step": 73600
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 6.134714572952182e-06,
+      "loss": 2.1706,
+      "step": 73610
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 6.132529067226156e-06,
+      "loss": 2.197,
+      "step": 73620
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 6.130343561500132e-06,
+      "loss": 2.179,
+      "step": 73630
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 6.128158055774107e-06,
+      "loss": 2.1732,
+      "step": 73640
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 6.125972550048082e-06,
+      "loss": 2.1842,
+      "step": 73650
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 6.123787044322057e-06,
+      "loss": 2.1802,
+      "step": 73660
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 6.121601538596031e-06,
+      "loss": 2.2113,
+      "step": 73670
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 6.119416032870007e-06,
+      "loss": 2.1625,
+      "step": 73680
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 6.117230527143981e-06,
+      "loss": 2.1844,
+      "step": 73690
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 6.115045021417957e-06,
+      "loss": 2.1613,
+      "step": 73700
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 6.112859515691931e-06,
+      "loss": 2.1713,
+      "step": 73710
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 6.110674009965907e-06,
+      "loss": 2.2027,
+      "step": 73720
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 6.108488504239881e-06,
+      "loss": 2.1212,
+      "step": 73730
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 6.106302998513856e-06,
+      "loss": 2.1899,
+      "step": 73740
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 6.104117492787831e-06,
+      "loss": 2.1815,
+      "step": 73750
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 6.101931987061806e-06,
+      "loss": 2.2028,
+      "step": 73760
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 6.099746481335782e-06,
+      "loss": 2.1812,
+      "step": 73770
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 2.1672,
+      "step": 73780
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 6.095375469883732e-06,
+      "loss": 2.2186,
+      "step": 73790
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 6.0931899641577065e-06,
+      "loss": 2.1444,
+      "step": 73800
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 6.0910044584316815e-06,
+      "loss": 2.2193,
+      "step": 73810
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 6.0888189527056566e-06,
+      "loss": 2.2018,
+      "step": 73820
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 6.086633446979632e-06,
+      "loss": 2.2073,
+      "step": 73830
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 6.084447941253606e-06,
+      "loss": 2.1879,
+      "step": 73840
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 6.082262435527582e-06,
+      "loss": 2.1879,
+      "step": 73850
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 6.080076929801556e-06,
+      "loss": 2.1751,
+      "step": 73860
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 6.077891424075532e-06,
+      "loss": 2.1864,
+      "step": 73870
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 6.075705918349507e-06,
+      "loss": 2.2189,
+      "step": 73880
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 6.073520412623482e-06,
+      "loss": 2.1287,
+      "step": 73890
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 6.071334906897457e-06,
+      "loss": 2.1979,
+      "step": 73900
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 6.069149401171431e-06,
+      "loss": 2.1545,
+      "step": 73910
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 6.066963895445407e-06,
+      "loss": 2.2054,
+      "step": 73920
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 6.064778389719381e-06,
+      "loss": 2.1754,
+      "step": 73930
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 6.062592883993357e-06,
+      "loss": 2.1615,
+      "step": 73940
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 6.060407378267331e-06,
+      "loss": 2.1886,
+      "step": 73950
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 6.058221872541307e-06,
+      "loss": 2.1654,
+      "step": 73960
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 6.056036366815281e-06,
+      "loss": 2.1679,
+      "step": 73970
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 6.053850861089256e-06,
+      "loss": 2.1902,
+      "step": 73980
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 6.051665355363232e-06,
+      "loss": 2.1652,
+      "step": 73990
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 6.049479849637206e-06,
+      "loss": 2.2329,
+      "step": 74000
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 6.047294343911182e-06,
+      "loss": 2.1701,
+      "step": 74010
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 6.045108838185156e-06,
+      "loss": 2.1798,
+      "step": 74020
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 6.042923332459132e-06,
+      "loss": 2.1868,
+      "step": 74030
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 6.040737826733106e-06,
+      "loss": 2.145,
+      "step": 74040
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 6.0385523210070814e-06,
+      "loss": 2.1897,
+      "step": 74050
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 6.0363668152810565e-06,
+      "loss": 2.1734,
+      "step": 74060
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 6.0341813095550315e-06,
+      "loss": 2.1714,
+      "step": 74070
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 6.031995803829006e-06,
+      "loss": 2.1887,
+      "step": 74080
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 6.0298102981029816e-06,
+      "loss": 2.1687,
+      "step": 74090
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 6.027624792376956e-06,
+      "loss": 2.1719,
+      "step": 74100
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 6.025439286650932e-06,
+      "loss": 2.1767,
+      "step": 74110
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 6.023253780924907e-06,
+      "loss": 2.1467,
+      "step": 74120
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 6.021068275198882e-06,
+      "loss": 2.1759,
+      "step": 74130
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 6.018882769472857e-06,
+      "loss": 2.1917,
+      "step": 74140
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 6.016697263746831e-06,
+      "loss": 2.1788,
+      "step": 74150
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 6.014511758020807e-06,
+      "loss": 2.1814,
+      "step": 74160
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 6.012326252294781e-06,
+      "loss": 2.2126,
+      "step": 74170
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 6.010140746568757e-06,
+      "loss": 2.1639,
+      "step": 74180
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 6.007955240842731e-06,
+      "loss": 2.2205,
+      "step": 74190
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 6.005769735116707e-06,
+      "loss": 2.1608,
+      "step": 74200
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 6.003584229390681e-06,
+      "loss": 2.1798,
+      "step": 74210
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 6.001398723664656e-06,
+      "loss": 2.1966,
+      "step": 74220
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 5.999213217938632e-06,
+      "loss": 2.1879,
+      "step": 74230
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 5.997027712212606e-06,
+      "loss": 2.1402,
+      "step": 74240
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 5.994842206486582e-06,
+      "loss": 2.2115,
+      "step": 74250
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 5.992656700760556e-06,
+      "loss": 2.1572,
+      "step": 74260
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 5.990471195034532e-06,
+      "loss": 2.1716,
+      "step": 74270
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 5.988285689308506e-06,
+      "loss": 2.1937,
+      "step": 74280
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 5.986100183582481e-06,
+      "loss": 2.1312,
+      "step": 74290
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 5.983914677856456e-06,
+      "loss": 2.1894,
+      "step": 74300
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 5.981729172130431e-06,
+      "loss": 2.2036,
+      "step": 74310
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 5.979543666404406e-06,
+      "loss": 2.2091,
+      "step": 74320
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 5.9773581606783815e-06,
+      "loss": 2.1771,
+      "step": 74330
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 5.975172654952357e-06,
+      "loss": 2.1738,
+      "step": 74340
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 5.9729871492263315e-06,
+      "loss": 2.1903,
+      "step": 74350
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 5.9708016435003066e-06,
+      "loss": 2.1037,
+      "step": 74360
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 5.968616137774282e-06,
+      "loss": 2.1992,
+      "step": 74370
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 5.966430632048257e-06,
+      "loss": 2.2314,
+      "step": 74380
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 5.964245126322231e-06,
+      "loss": 2.1845,
+      "step": 74390
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 5.962059620596207e-06,
+      "loss": 2.1666,
+      "step": 74400
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 5.959874114870181e-06,
+      "loss": 2.2213,
+      "step": 74410
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 5.957688609144157e-06,
+      "loss": 2.1738,
+      "step": 74420
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 5.955503103418131e-06,
+      "loss": 2.166,
+      "step": 74430
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 5.953317597692107e-06,
+      "loss": 2.2176,
+      "step": 74440
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 5.951132091966082e-06,
+      "loss": 2.1893,
+      "step": 74450
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 5.948946586240056e-06,
+      "loss": 2.1848,
+      "step": 74460
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 5.946761080514032e-06,
+      "loss": 2.1554,
+      "step": 74470
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 5.944575574788006e-06,
+      "loss": 2.1548,
+      "step": 74480
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 5.942390069061982e-06,
+      "loss": 2.1834,
+      "step": 74490
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 5.940204563335956e-06,
+      "loss": 2.1661,
+      "step": 74500
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 5.938019057609932e-06,
+      "loss": 2.1834,
+      "step": 74510
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 5.935833551883906e-06,
+      "loss": 2.199,
+      "step": 74520
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 5.933648046157881e-06,
+      "loss": 2.2007,
+      "step": 74530
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 5.931462540431856e-06,
+      "loss": 2.177,
+      "step": 74540
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 5.929277034705831e-06,
+      "loss": 2.1696,
+      "step": 74550
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 5.9270915289798055e-06,
+      "loss": 2.2037,
+      "step": 74560
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 5.924906023253781e-06,
+      "loss": 2.1666,
+      "step": 74570
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 5.922720517527757e-06,
+      "loss": 2.206,
+      "step": 74580
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 5.9205350118017315e-06,
+      "loss": 2.1731,
+      "step": 74590
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 5.9183495060757065e-06,
+      "loss": 2.1796,
+      "step": 74600
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 5.9161640003496815e-06,
+      "loss": 2.1799,
+      "step": 74610
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 5.9139784946236566e-06,
+      "loss": 2.1889,
+      "step": 74620
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 5.911792988897631e-06,
+      "loss": 2.1762,
+      "step": 74630
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 5.909607483171607e-06,
+      "loss": 2.194,
+      "step": 74640
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 5.907421977445581e-06,
+      "loss": 2.1829,
+      "step": 74650
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 5.905236471719557e-06,
+      "loss": 2.2291,
+      "step": 74660
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 5.903050965993531e-06,
+      "loss": 2.1508,
+      "step": 74670
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 5.900865460267507e-06,
+      "loss": 2.1914,
+      "step": 74680
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 5.898679954541482e-06,
+      "loss": 2.184,
+      "step": 74690
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 5.896494448815456e-06,
+      "loss": 2.1687,
+      "step": 74700
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 5.894308943089432e-06,
+      "loss": 2.1754,
+      "step": 74710
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 5.892123437363406e-06,
+      "loss": 2.193,
+      "step": 74720
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 5.889937931637382e-06,
+      "loss": 2.2096,
+      "step": 74730
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 5.887752425911356e-06,
+      "loss": 2.1959,
+      "step": 74740
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 5.885566920185332e-06,
+      "loss": 2.1737,
+      "step": 74750
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 5.883381414459306e-06,
+      "loss": 2.2138,
+      "step": 74760
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 5.881195908733281e-06,
+      "loss": 2.2063,
+      "step": 74770
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 5.879010403007256e-06,
+      "loss": 2.1778,
+      "step": 74780
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 5.876824897281231e-06,
+      "loss": 2.2038,
+      "step": 74790
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 5.874639391555207e-06,
+      "loss": 2.1598,
+      "step": 74800
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 5.872453885829181e-06,
+      "loss": 2.1918,
+      "step": 74810
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 5.870268380103157e-06,
+      "loss": 2.2071,
+      "step": 74820
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 5.868082874377131e-06,
+      "loss": 2.2004,
+      "step": 74830
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 5.865897368651106e-06,
+      "loss": 2.2143,
+      "step": 74840
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 5.8637118629250814e-06,
+      "loss": 2.1926,
+      "step": 74850
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 5.8615263571990565e-06,
+      "loss": 2.1955,
+      "step": 74860
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 5.859340851473031e-06,
+      "loss": 2.2184,
+      "step": 74870
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 5.8571553457470065e-06,
+      "loss": 2.1843,
+      "step": 74880
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 5.854969840020981e-06,
+      "loss": 2.2042,
+      "step": 74890
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 5.852784334294957e-06,
+      "loss": 2.1538,
+      "step": 74900
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 5.850598828568931e-06,
+      "loss": 2.1884,
+      "step": 74910
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 5.848413322842907e-06,
+      "loss": 2.1924,
+      "step": 74920
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 5.846227817116882e-06,
+      "loss": 2.1583,
+      "step": 74930
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 5.844042311390856e-06,
+      "loss": 2.2141,
+      "step": 74940
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 5.841856805664832e-06,
+      "loss": 2.1974,
+      "step": 74950
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 5.839671299938806e-06,
+      "loss": 2.1886,
+      "step": 74960
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 5.837485794212782e-06,
+      "loss": 2.1586,
+      "step": 74970
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 5.835300288486756e-06,
+      "loss": 2.217,
+      "step": 74980
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 5.833114782760732e-06,
+      "loss": 2.2321,
+      "step": 74990
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 5.830929277034706e-06,
+      "loss": 2.1814,
+      "step": 75000
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 5.828743771308681e-06,
+      "loss": 2.1816,
+      "step": 75010
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 5.826558265582656e-06,
+      "loss": 2.1841,
+      "step": 75020
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 5.824372759856631e-06,
+      "loss": 2.186,
+      "step": 75030
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 5.822187254130607e-06,
+      "loss": 2.195,
+      "step": 75040
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 5.820001748404581e-06,
+      "loss": 2.2109,
+      "step": 75050
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 5.817816242678557e-06,
+      "loss": 2.1821,
+      "step": 75060
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 5.815630736952531e-06,
+      "loss": 2.1666,
+      "step": 75070
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 5.813445231226506e-06,
+      "loss": 2.1807,
+      "step": 75080
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 5.811259725500481e-06,
+      "loss": 2.177,
+      "step": 75090
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 5.809074219774456e-06,
+      "loss": 2.2144,
+      "step": 75100
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 5.8068887140484306e-06,
+      "loss": 2.2242,
+      "step": 75110
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 5.8047032083224064e-06,
+      "loss": 2.2169,
+      "step": 75120
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 5.802517702596381e-06,
+      "loss": 2.19,
+      "step": 75130
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 5.8003321968703565e-06,
+      "loss": 2.191,
+      "step": 75140
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 5.7981466911443315e-06,
+      "loss": 2.1855,
+      "step": 75150
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 5.7959611854183066e-06,
+      "loss": 2.2023,
+      "step": 75160
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 5.793775679692282e-06,
+      "loss": 2.2007,
+      "step": 75170
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 5.791590173966256e-06,
+      "loss": 2.1404,
+      "step": 75180
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 5.789404668240232e-06,
+      "loss": 2.1689,
+      "step": 75190
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 5.787219162514206e-06,
+      "loss": 2.2026,
+      "step": 75200
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 5.785033656788182e-06,
+      "loss": 2.1847,
+      "step": 75210
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 5.782848151062156e-06,
+      "loss": 2.1797,
+      "step": 75220
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 5.780662645336132e-06,
+      "loss": 2.1563,
+      "step": 75230
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 5.778477139610106e-06,
+      "loss": 2.1675,
+      "step": 75240
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 5.776291633884081e-06,
+      "loss": 2.1941,
+      "step": 75250
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 5.774106128158056e-06,
+      "loss": 2.1922,
+      "step": 75260
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 5.771920622432031e-06,
+      "loss": 2.172,
+      "step": 75270
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 5.769735116706007e-06,
+      "loss": 2.2114,
+      "step": 75280
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 5.767549610979981e-06,
+      "loss": 2.2361,
+      "step": 75290
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 5.765364105253957e-06,
+      "loss": 2.1992,
+      "step": 75300
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 5.763178599527931e-06,
+      "loss": 2.1937,
+      "step": 75310
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 5.760993093801906e-06,
+      "loss": 2.1961,
+      "step": 75320
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 5.758807588075881e-06,
+      "loss": 2.2218,
+      "step": 75330
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 5.756622082349856e-06,
+      "loss": 2.2252,
+      "step": 75340
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 5.7544365766238305e-06,
+      "loss": 2.1706,
+      "step": 75350
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 5.752251070897806e-06,
+      "loss": 2.176,
+      "step": 75360
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 5.7500655651717805e-06,
+      "loss": 2.2188,
+      "step": 75370
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 5.747880059445756e-06,
+      "loss": 2.2005,
+      "step": 75380
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 5.7456945537197315e-06,
+      "loss": 2.1818,
+      "step": 75390
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 5.7435090479937065e-06,
+      "loss": 2.2098,
+      "step": 75400
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 5.7413235422676815e-06,
+      "loss": 2.2104,
+      "step": 75410
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 5.739138036541656e-06,
+      "loss": 2.2019,
+      "step": 75420
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.736952530815632e-06,
+      "loss": 2.1768,
+      "step": 75430
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.734767025089606e-06,
+      "loss": 2.2258,
+      "step": 75440
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.732581519363582e-06,
+      "loss": 2.1748,
+      "step": 75450
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.730396013637556e-06,
+      "loss": 2.1768,
+      "step": 75460
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.728210507911532e-06,
+      "loss": 2.2018,
+      "step": 75470
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 5.726025002185506e-06,
+      "loss": 2.1752,
+      "step": 75480
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 5.723839496459481e-06,
+      "loss": 2.1871,
+      "step": 75490
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 5.721653990733457e-06,
+      "loss": 2.1494,
+      "step": 75500
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 5.719468485007431e-06,
+      "loss": 2.1832,
+      "step": 75510
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 5.717282979281407e-06,
+      "loss": 2.1703,
+      "step": 75520
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 5.715097473555381e-06,
+      "loss": 2.2018,
+      "step": 75530
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 5.712911967829357e-06,
+      "loss": 2.1808,
+      "step": 75540
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 5.710726462103331e-06,
+      "loss": 2.1694,
+      "step": 75550
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 5.708540956377306e-06,
+      "loss": 2.1966,
+      "step": 75560
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 5.706355450651281e-06,
+      "loss": 2.1503,
+      "step": 75570
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 5.704169944925256e-06,
+      "loss": 2.1852,
+      "step": 75580
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 5.70198443919923e-06,
+      "loss": 2.1867,
+      "step": 75590
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 5.699798933473206e-06,
+      "loss": 2.1917,
+      "step": 75600
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 5.6976134277471805e-06,
+      "loss": 2.2057,
+      "step": 75610
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 5.695427922021156e-06,
+      "loss": 2.1647,
+      "step": 75620
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 5.693242416295131e-06,
+      "loss": 2.2015,
+      "step": 75630
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 5.691056910569106e-06,
+      "loss": 2.2039,
+      "step": 75640
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 5.6888714048430814e-06,
+      "loss": 2.2004,
+      "step": 75650
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 5.686685899117056e-06,
+      "loss": 2.1745,
+      "step": 75660
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 5.6845003933910315e-06,
+      "loss": 2.1816,
+      "step": 75670
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 5.682314887665006e-06,
+      "loss": 2.193,
+      "step": 75680
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 5.6801293819389816e-06,
+      "loss": 2.1604,
+      "step": 75690
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 5.677943876212956e-06,
+      "loss": 2.1788,
+      "step": 75700
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 5.675758370486932e-06,
+      "loss": 2.225,
+      "step": 75710
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 5.673572864760906e-06,
+      "loss": 2.198,
+      "step": 75720
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 5.671387359034881e-06,
+      "loss": 2.1418,
+      "step": 75730
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 5.669201853308857e-06,
+      "loss": 2.1694,
+      "step": 75740
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 5.667016347582831e-06,
+      "loss": 2.1774,
+      "step": 75750
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 5.664830841856807e-06,
+      "loss": 2.182,
+      "step": 75760
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 5.662645336130781e-06,
+      "loss": 2.1888,
+      "step": 75770
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 5.660459830404757e-06,
+      "loss": 2.1942,
+      "step": 75780
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 5.658274324678731e-06,
+      "loss": 2.1882,
+      "step": 75790
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 5.656088818952706e-06,
+      "loss": 2.181,
+      "step": 75800
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 5.653903313226681e-06,
+      "loss": 2.2285,
+      "step": 75810
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 5.651717807500656e-06,
+      "loss": 2.1582,
+      "step": 75820
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 5.64953230177463e-06,
+      "loss": 2.1985,
+      "step": 75830
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 5.647346796048606e-06,
+      "loss": 2.1984,
+      "step": 75840
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 5.645161290322582e-06,
+      "loss": 2.2092,
+      "step": 75850
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 5.642975784596556e-06,
+      "loss": 2.1695,
+      "step": 75860
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 5.640790278870531e-06,
+      "loss": 2.1666,
+      "step": 75870
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 5.638604773144506e-06,
+      "loss": 2.1593,
+      "step": 75880
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 5.636419267418481e-06,
+      "loss": 2.237,
+      "step": 75890
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 5.6342337616924555e-06,
+      "loss": 2.204,
+      "step": 75900
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 5.632048255966431e-06,
+      "loss": 2.1898,
+      "step": 75910
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 5.629862750240406e-06,
+      "loss": 2.1855,
+      "step": 75920
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 5.6276772445143815e-06,
+      "loss": 2.1594,
+      "step": 75930
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 5.625491738788356e-06,
+      "loss": 2.1792,
+      "step": 75940
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 5.6233062330623315e-06,
+      "loss": 2.1603,
+      "step": 75950
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 5.621120727336306e-06,
+      "loss": 2.1915,
+      "step": 75960
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 5.618935221610281e-06,
+      "loss": 2.1946,
+      "step": 75970
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 5.616749715884257e-06,
+      "loss": 2.1982,
+      "step": 75980
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 5.614564210158231e-06,
+      "loss": 2.1805,
+      "step": 75990
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 5.612378704432207e-06,
+      "loss": 2.1396,
+      "step": 76000
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 5.610193198706181e-06,
+      "loss": 2.1784,
+      "step": 76010
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 5.608007692980157e-06,
+      "loss": 2.2234,
+      "step": 76020
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 5.605822187254131e-06,
+      "loss": 2.2006,
+      "step": 76030
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 5.603636681528106e-06,
+      "loss": 2.1936,
+      "step": 76040
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 5.601451175802081e-06,
+      "loss": 2.222,
+      "step": 76050
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 5.599265670076056e-06,
+      "loss": 2.1948,
+      "step": 76060
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 5.59708016435003e-06,
+      "loss": 2.183,
+      "step": 76070
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 5.594894658624006e-06,
+      "loss": 2.1845,
+      "step": 76080
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 5.592709152897981e-06,
+      "loss": 2.1592,
+      "step": 76090
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 5.590523647171956e-06,
+      "loss": 2.1728,
+      "step": 76100
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 5.588338141445931e-06,
+      "loss": 2.1789,
+      "step": 76110
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 5.586152635719906e-06,
+      "loss": 2.1678,
+      "step": 76120
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 5.583967129993881e-06,
+      "loss": 2.2247,
+      "step": 76130
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 5.5817816242678554e-06,
+      "loss": 2.2061,
+      "step": 76140
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 5.579596118541831e-06,
+      "loss": 2.219,
+      "step": 76150
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 5.5774106128158055e-06,
+      "loss": 2.1669,
+      "step": 76160
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 5.575225107089781e-06,
+      "loss": 2.2007,
+      "step": 76170
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 5.5730396013637556e-06,
+      "loss": 2.1918,
+      "step": 76180
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 5.5708540956377314e-06,
+      "loss": 2.2329,
+      "step": 76190
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 5.5686685899117065e-06,
+      "loss": 2.159,
+      "step": 76200
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 5.566483084185681e-06,
+      "loss": 2.1927,
+      "step": 76210
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 5.5642975784596565e-06,
+      "loss": 2.1968,
+      "step": 76220
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 5.562112072733631e-06,
+      "loss": 2.2404,
+      "step": 76230
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 5.559926567007607e-06,
+      "loss": 2.2142,
+      "step": 76240
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 5.557741061281581e-06,
+      "loss": 2.1817,
+      "step": 76250
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 2.1679,
+      "step": 76260
+    },
+    {
+      "epoch": 15.0,
+      "eval_BLEU": 10.2901,
+      "eval_gen_len": 12.209,
+      "eval_loss": 3.2753219604492188,
+      "eval_runtime": 139.7961,
+      "eval_samples_per_second": 76.991,
+      "eval_steps_per_second": 0.308,
+      "step": 76260
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 5.553370049829531e-06,
+      "loss": 2.112,
+      "step": 76270
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 5.551184544103506e-06,
+      "loss": 2.1229,
+      "step": 76280
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.548999038377481e-06,
+      "loss": 2.1341,
+      "step": 76290
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.546813532651456e-06,
+      "loss": 2.0995,
+      "step": 76300
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.54462802692543e-06,
+      "loss": 2.0781,
+      "step": 76310
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.542442521199406e-06,
+      "loss": 2.1098,
+      "step": 76320
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.540257015473381e-06,
+      "loss": 2.0866,
+      "step": 76330
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.538071509747356e-06,
+      "loss": 2.1079,
+      "step": 76340
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.535886004021331e-06,
+      "loss": 2.1654,
+      "step": 76350
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.533700498295306e-06,
+      "loss": 2.1253,
+      "step": 76360
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.531514992569281e-06,
+      "loss": 2.1357,
+      "step": 76370
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 5.529329486843255e-06,
+      "loss": 2.1083,
+      "step": 76380
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.527143981117231e-06,
+      "loss": 2.0703,
+      "step": 76390
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.524958475391205e-06,
+      "loss": 2.1316,
+      "step": 76400
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.522772969665181e-06,
+      "loss": 2.1445,
+      "step": 76410
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.5205874639391555e-06,
+      "loss": 2.1389,
+      "step": 76420
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 5.518401958213131e-06,
+      "loss": 2.078,
+      "step": 76430
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.516216452487106e-06,
+      "loss": 2.0606,
+      "step": 76440
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.514030946761081e-06,
+      "loss": 2.1224,
+      "step": 76450
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.5118454410350565e-06,
+      "loss": 2.1204,
+      "step": 76460
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.509659935309031e-06,
+      "loss": 2.0814,
+      "step": 76470
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.5074744295830065e-06,
+      "loss": 2.1348,
+      "step": 76480
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 5.505288923856981e-06,
+      "loss": 2.0928,
+      "step": 76490
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 5.503103418130957e-06,
+      "loss": 2.0896,
+      "step": 76500
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 5.500917912404931e-06,
+      "loss": 2.0889,
+      "step": 76510
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 5.498732406678906e-06,
+      "loss": 2.1358,
+      "step": 76520
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 5.496546900952881e-06,
+      "loss": 2.1356,
+      "step": 76530
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 5.494361395226856e-06,
+      "loss": 2.0872,
+      "step": 76540
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 5.492175889500832e-06,
+      "loss": 2.1306,
+      "step": 76550
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 5.489990383774806e-06,
+      "loss": 2.1065,
+      "step": 76560
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 5.487804878048781e-06,
+      "loss": 2.096,
+      "step": 76570
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 5.485619372322756e-06,
+      "loss": 2.1341,
+      "step": 76580
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 5.483433866596731e-06,
+      "loss": 2.095,
+      "step": 76590
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 5.481248360870706e-06,
+      "loss": 2.1137,
+      "step": 76600
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 5.479062855144681e-06,
+      "loss": 2.1182,
+      "step": 76610
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 5.476877349418655e-06,
+      "loss": 2.1403,
+      "step": 76620
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 5.474691843692631e-06,
+      "loss": 2.1606,
+      "step": 76630
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 5.472506337966605e-06,
+      "loss": 2.1195,
+      "step": 76640
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 5.470320832240581e-06,
+      "loss": 2.1499,
+      "step": 76650
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 5.468135326514555e-06,
+      "loss": 2.1152,
+      "step": 76660
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 5.465949820788531e-06,
+      "loss": 2.1411,
+      "step": 76670
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 5.463764315062506e-06,
+      "loss": 2.0988,
+      "step": 76680
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 5.4615788093364805e-06,
+      "loss": 2.0833,
+      "step": 76690
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 5.459393303610456e-06,
+      "loss": 2.0895,
+      "step": 76700
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 5.4572077978844306e-06,
+      "loss": 2.1159,
+      "step": 76710
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 5.4550222921584064e-06,
+      "loss": 2.1135,
+      "step": 76720
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 5.452836786432381e-06,
+      "loss": 2.1161,
+      "step": 76730
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 5.4506512807063565e-06,
+      "loss": 2.1105,
+      "step": 76740
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 5.448465774980331e-06,
+      "loss": 2.1136,
+      "step": 76750
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 5.446280269254306e-06,
+      "loss": 2.0847,
+      "step": 76760
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 5.444094763528281e-06,
+      "loss": 2.1395,
+      "step": 76770
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 5.441909257802256e-06,
+      "loss": 2.1246,
+      "step": 76780
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 5.439723752076232e-06,
+      "loss": 2.1439,
+      "step": 76790
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 5.437538246350206e-06,
+      "loss": 2.1564,
+      "step": 76800
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 5.435352740624181e-06,
+      "loss": 2.0779,
+      "step": 76810
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 5.433167234898156e-06,
+      "loss": 2.1395,
+      "step": 76820
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 5.430981729172131e-06,
+      "loss": 2.1465,
+      "step": 76830
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 5.428796223446106e-06,
+      "loss": 2.132,
+      "step": 76840
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 5.426610717720081e-06,
+      "loss": 2.0723,
+      "step": 76850
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 5.424425211994055e-06,
+      "loss": 2.1204,
+      "step": 76860
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 5.422239706268031e-06,
+      "loss": 2.1018,
+      "step": 76870
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 5.420054200542005e-06,
+      "loss": 2.1104,
+      "step": 76880
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 5.417868694815981e-06,
+      "loss": 2.1102,
+      "step": 76890
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 5.415683189089956e-06,
+      "loss": 2.1145,
+      "step": 76900
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 5.413497683363931e-06,
+      "loss": 2.1106,
+      "step": 76910
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 5.411312177637906e-06,
+      "loss": 2.1391,
+      "step": 76920
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 5.40912667191188e-06,
+      "loss": 2.1133,
+      "step": 76930
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 5.406941166185856e-06,
+      "loss": 2.1374,
+      "step": 76940
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 5.4047556604598305e-06,
+      "loss": 2.1305,
+      "step": 76950
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 5.402570154733806e-06,
+      "loss": 2.1083,
+      "step": 76960
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 5.4003846490077805e-06,
+      "loss": 2.1175,
+      "step": 76970
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 5.3981991432817556e-06,
+      "loss": 2.0949,
+      "step": 76980
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 5.396013637555731e-06,
+      "loss": 2.1405,
+      "step": 76990
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 5.393828131829706e-06,
+      "loss": 2.1416,
+      "step": 77000
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 5.3916426261036815e-06,
+      "loss": 2.141,
+      "step": 77010
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 5.389457120377656e-06,
+      "loss": 2.1406,
+      "step": 77020
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 5.3872716146516316e-06,
+      "loss": 2.0813,
+      "step": 77030
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 5.385086108925606e-06,
+      "loss": 2.1094,
+      "step": 77040
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 5.382900603199581e-06,
+      "loss": 2.101,
+      "step": 77050
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 5.380715097473556e-06,
+      "loss": 2.1279,
+      "step": 77060
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 5.378529591747531e-06,
+      "loss": 2.1021,
+      "step": 77070
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 5.376344086021506e-06,
+      "loss": 2.1249,
+      "step": 77080
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 5.374158580295481e-06,
+      "loss": 2.1187,
+      "step": 77090
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 5.371973074569455e-06,
+      "loss": 2.1093,
+      "step": 77100
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 5.369787568843431e-06,
+      "loss": 2.1262,
+      "step": 77110
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 5.367602063117405e-06,
+      "loss": 2.1261,
+      "step": 77120
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 5.365416557391381e-06,
+      "loss": 2.1354,
+      "step": 77130
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 5.363231051665356e-06,
+      "loss": 2.0934,
+      "step": 77140
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 5.361045545939331e-06,
+      "loss": 2.1351,
+      "step": 77150
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 5.358860040213306e-06,
+      "loss": 2.073,
+      "step": 77160
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 5.35667453448728e-06,
+      "loss": 2.1041,
+      "step": 77170
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 5.354489028761256e-06,
+      "loss": 2.0998,
+      "step": 77180
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 5.35230352303523e-06,
+      "loss": 2.1124,
+      "step": 77190
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 5.350118017309206e-06,
+      "loss": 2.1477,
+      "step": 77200
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 5.3479325115831804e-06,
+      "loss": 2.1505,
+      "step": 77210
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 5.3457470058571555e-06,
+      "loss": 2.0996,
+      "step": 77220
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 5.3435615001311305e-06,
+      "loss": 2.1289,
+      "step": 77230
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 5.3413759944051055e-06,
+      "loss": 2.0921,
+      "step": 77240
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 5.339190488679081e-06,
+      "loss": 2.1224,
+      "step": 77250
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 5.337004982953056e-06,
+      "loss": 2.1418,
+      "step": 77260
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 5.3348194772270315e-06,
+      "loss": 2.1361,
+      "step": 77270
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 5.332633971501006e-06,
+      "loss": 2.1383,
+      "step": 77280
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 5.330448465774981e-06,
+      "loss": 2.116,
+      "step": 77290
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 5.328262960048956e-06,
+      "loss": 2.1555,
+      "step": 77300
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 5.326077454322931e-06,
+      "loss": 2.1096,
+      "step": 77310
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 5.323891948596906e-06,
+      "loss": 2.1161,
+      "step": 77320
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 5.321706442870881e-06,
+      "loss": 2.1215,
+      "step": 77330
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 5.319520937144855e-06,
+      "loss": 2.1235,
+      "step": 77340
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 5.317335431418831e-06,
+      "loss": 2.1494,
+      "step": 77350
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 5.315149925692806e-06,
+      "loss": 2.1399,
+      "step": 77360
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 5.312964419966781e-06,
+      "loss": 2.1277,
+      "step": 77370
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 5.310778914240756e-06,
+      "loss": 2.1634,
+      "step": 77380
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 5.308593408514731e-06,
+      "loss": 2.1346,
+      "step": 77390
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 5.306407902788706e-06,
+      "loss": 2.1027,
+      "step": 77400
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 5.30422239706268e-06,
+      "loss": 2.1219,
+      "step": 77410
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 5.302036891336656e-06,
+      "loss": 2.121,
+      "step": 77420
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 5.29985138561063e-06,
+      "loss": 2.1544,
+      "step": 77430
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 5.297665879884606e-06,
+      "loss": 2.1337,
+      "step": 77440
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 5.29548037415858e-06,
+      "loss": 2.0937,
+      "step": 77450
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 5.293294868432555e-06,
+      "loss": 2.1468,
+      "step": 77460
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 5.29110936270653e-06,
+      "loss": 2.126,
+      "step": 77470
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 5.2889238569805055e-06,
+      "loss": 2.0856,
+      "step": 77480
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 5.286738351254481e-06,
+      "loss": 2.1603,
+      "step": 77490
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 5.2845528455284555e-06,
+      "loss": 2.1403,
+      "step": 77500
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 5.282367339802431e-06,
+      "loss": 2.1342,
+      "step": 77510
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 5.280181834076406e-06,
+      "loss": 2.1425,
+      "step": 77520
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 5.277996328350381e-06,
+      "loss": 2.1286,
+      "step": 77530
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 5.275810822624356e-06,
+      "loss": 2.1292,
+      "step": 77540
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 5.273625316898331e-06,
+      "loss": 2.1123,
+      "step": 77550
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 5.271439811172306e-06,
+      "loss": 2.1035,
+      "step": 77560
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 5.269254305446281e-06,
+      "loss": 2.1451,
+      "step": 77570
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 5.267068799720255e-06,
+      "loss": 2.1584,
+      "step": 77580
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 5.264883293994231e-06,
+      "loss": 2.1363,
+      "step": 77590
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 5.262697788268206e-06,
+      "loss": 2.1132,
+      "step": 77600
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 5.260512282542181e-06,
+      "loss": 2.1466,
+      "step": 77610
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 5.258326776816156e-06,
+      "loss": 2.0921,
+      "step": 77620
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 5.25614127109013e-06,
+      "loss": 2.1282,
+      "step": 77630
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 5.253955765364106e-06,
+      "loss": 2.1317,
+      "step": 77640
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 5.25177025963808e-06,
+      "loss": 2.1438,
+      "step": 77650
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 5.249584753912056e-06,
+      "loss": 2.1668,
+      "step": 77660
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 5.24739924818603e-06,
+      "loss": 2.1274,
+      "step": 77670
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 5.245213742460006e-06,
+      "loss": 2.1615,
+      "step": 77680
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 5.24302823673398e-06,
+      "loss": 2.1135,
+      "step": 77690
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 5.240842731007955e-06,
+      "loss": 2.1109,
+      "step": 77700
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 5.238657225281931e-06,
+      "loss": 2.1111,
+      "step": 77710
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 5.236471719555905e-06,
+      "loss": 2.1059,
+      "step": 77720
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 5.234286213829881e-06,
+      "loss": 2.1298,
+      "step": 77730
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 5.2321007081038554e-06,
+      "loss": 2.0992,
+      "step": 77740
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 5.229915202377831e-06,
+      "loss": 2.1251,
+      "step": 77750
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 5.2277296966518055e-06,
+      "loss": 2.1642,
+      "step": 77760
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 5.2255441909257805e-06,
+      "loss": 2.1506,
+      "step": 77770
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 5.2233586851997556e-06,
+      "loss": 2.1527,
+      "step": 77780
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 5.221173179473731e-06,
+      "loss": 2.1123,
+      "step": 77790
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 5.218987673747706e-06,
+      "loss": 2.173,
+      "step": 77800
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 5.216802168021681e-06,
+      "loss": 2.1329,
+      "step": 77810
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 5.214616662295655e-06,
+      "loss": 2.1452,
+      "step": 77820
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 5.212431156569631e-06,
+      "loss": 2.1289,
+      "step": 77830
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 5.210245650843606e-06,
+      "loss": 2.1523,
+      "step": 77840
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 5.208060145117581e-06,
+      "loss": 2.1203,
+      "step": 77850
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 5.205874639391556e-06,
+      "loss": 2.1401,
+      "step": 77860
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 5.20368913366553e-06,
+      "loss": 2.1178,
+      "step": 77870
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 5.201503627939506e-06,
+      "loss": 2.0932,
+      "step": 77880
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 5.19931812221348e-06,
+      "loss": 2.1196,
+      "step": 77890
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 5.197132616487456e-06,
+      "loss": 2.1228,
+      "step": 77900
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 5.19494711076143e-06,
+      "loss": 2.1513,
+      "step": 77910
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 5.192761605035406e-06,
+      "loss": 2.1355,
+      "step": 77920
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 5.19057609930938e-06,
+      "loss": 2.1633,
+      "step": 77930
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 5.188390593583355e-06,
+      "loss": 2.1482,
+      "step": 77940
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 5.186205087857331e-06,
+      "loss": 2.1351,
+      "step": 77950
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 5.184019582131305e-06,
+      "loss": 2.1153,
+      "step": 77960
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 5.181834076405281e-06,
+      "loss": 2.1214,
+      "step": 77970
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 5.179648570679255e-06,
+      "loss": 2.1236,
+      "step": 77980
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 5.177463064953231e-06,
+      "loss": 2.0955,
+      "step": 77990
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 5.175277559227205e-06,
+      "loss": 2.1159,
+      "step": 78000
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 5.1730920535011804e-06,
+      "loss": 2.1475,
+      "step": 78010
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 5.1709065477751555e-06,
+      "loss": 2.1619,
+      "step": 78020
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 5.1687210420491305e-06,
+      "loss": 2.1631,
+      "step": 78030
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 5.1665355363231055e-06,
+      "loss": 2.1613,
+      "step": 78040
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 5.1643500305970806e-06,
+      "loss": 2.1531,
+      "step": 78050
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 5.1621645248710564e-06,
+      "loss": 2.1056,
+      "step": 78060
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 5.159979019145031e-06,
+      "loss": 2.152,
+      "step": 78070
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 5.157793513419006e-06,
+      "loss": 2.108,
+      "step": 78080
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 5.155608007692981e-06,
+      "loss": 2.1098,
+      "step": 78090
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 5.153422501966956e-06,
+      "loss": 2.1398,
+      "step": 78100
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 5.15123699624093e-06,
+      "loss": 2.1595,
+      "step": 78110
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 5.149051490514906e-06,
+      "loss": 2.1753,
+      "step": 78120
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 5.14686598478888e-06,
+      "loss": 2.1468,
+      "step": 78130
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 5.144680479062856e-06,
+      "loss": 2.1247,
+      "step": 78140
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 5.14249497333683e-06,
+      "loss": 2.1411,
+      "step": 78150
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 5.140309467610806e-06,
+      "loss": 2.102,
+      "step": 78160
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 5.13812396188478e-06,
+      "loss": 2.1498,
+      "step": 78170
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 5.135938456158755e-06,
+      "loss": 2.1081,
+      "step": 78180
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 5.133752950432731e-06,
+      "loss": 2.1317,
+      "step": 78190
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 5.131567444706705e-06,
+      "loss": 2.1731,
+      "step": 78200
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 5.129381938980681e-06,
+      "loss": 2.1421,
+      "step": 78210
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 5.127196433254655e-06,
+      "loss": 2.1336,
+      "step": 78220
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 5.125010927528631e-06,
+      "loss": 2.1536,
+      "step": 78230
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 5.122825421802605e-06,
+      "loss": 2.1296,
+      "step": 78240
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 5.12063991607658e-06,
+      "loss": 2.1532,
+      "step": 78250
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 5.118454410350555e-06,
+      "loss": 2.1555,
+      "step": 78260
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 5.11626890462453e-06,
+      "loss": 2.1232,
+      "step": 78270
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 5.114083398898505e-06,
+      "loss": 2.1678,
+      "step": 78280
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 5.1118978931724805e-06,
+      "loss": 2.1599,
+      "step": 78290
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 5.109712387446456e-06,
+      "loss": 2.1136,
+      "step": 78300
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 2.1618,
+      "step": 78310
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 5.1053413759944056e-06,
+      "loss": 2.1274,
+      "step": 78320
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 5.103155870268381e-06,
+      "loss": 2.1319,
+      "step": 78330
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 5.100970364542356e-06,
+      "loss": 2.127,
+      "step": 78340
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 5.09878485881633e-06,
+      "loss": 2.1198,
+      "step": 78350
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 5.096599353090306e-06,
+      "loss": 2.1323,
+      "step": 78360
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 5.09441384736428e-06,
+      "loss": 2.1712,
+      "step": 78370
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 5.092228341638256e-06,
+      "loss": 2.17,
+      "step": 78380
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 5.09004283591223e-06,
+      "loss": 2.1608,
+      "step": 78390
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 5.087857330186206e-06,
+      "loss": 2.1022,
+      "step": 78400
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 5.085671824460181e-06,
+      "loss": 2.1107,
+      "step": 78410
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 5.083486318734155e-06,
+      "loss": 2.1294,
+      "step": 78420
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 5.081300813008131e-06,
+      "loss": 2.1177,
+      "step": 78430
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 5.079115307282105e-06,
+      "loss": 2.145,
+      "step": 78440
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 5.076929801556081e-06,
+      "loss": 2.1182,
+      "step": 78450
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 5.074744295830055e-06,
+      "loss": 2.1817,
+      "step": 78460
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 5.072558790104031e-06,
+      "loss": 2.1422,
+      "step": 78470
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 5.070373284378005e-06,
+      "loss": 2.1446,
+      "step": 78480
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 5.06818777865198e-06,
+      "loss": 2.1506,
+      "step": 78490
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 5.066002272925955e-06,
+      "loss": 2.1451,
+      "step": 78500
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 5.06381676719993e-06,
+      "loss": 2.1769,
+      "step": 78510
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 5.0616312614739045e-06,
+      "loss": 2.1525,
+      "step": 78520
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 5.05944575574788e-06,
+      "loss": 2.1173,
+      "step": 78530
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 5.057260250021856e-06,
+      "loss": 2.1173,
+      "step": 78540
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 5.0550747442958305e-06,
+      "loss": 2.1402,
+      "step": 78550
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 5.0528892385698055e-06,
+      "loss": 2.1395,
+      "step": 78560
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 5.0507037328437805e-06,
+      "loss": 2.1309,
+      "step": 78570
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 5.0485182271177556e-06,
+      "loss": 2.1449,
+      "step": 78580
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 5.04633272139173e-06,
+      "loss": 2.1111,
+      "step": 78590
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 5.044147215665706e-06,
+      "loss": 2.1265,
+      "step": 78600
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 5.04196170993968e-06,
+      "loss": 2.1176,
+      "step": 78610
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 5.039776204213656e-06,
+      "loss": 2.1524,
+      "step": 78620
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 5.03759069848763e-06,
+      "loss": 2.1537,
+      "step": 78630
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 5.035405192761606e-06,
+      "loss": 2.1652,
+      "step": 78640
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 5.033219687035581e-06,
+      "loss": 2.1411,
+      "step": 78650
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 5.031034181309555e-06,
+      "loss": 2.1468,
+      "step": 78660
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 5.028848675583531e-06,
+      "loss": 2.1296,
+      "step": 78670
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 5.026663169857505e-06,
+      "loss": 2.1189,
+      "step": 78680
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 5.024477664131481e-06,
+      "loss": 2.1414,
+      "step": 78690
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 5.022292158405455e-06,
+      "loss": 2.1601,
+      "step": 78700
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 5.020106652679431e-06,
+      "loss": 2.1311,
+      "step": 78710
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 5.017921146953405e-06,
+      "loss": 2.1226,
+      "step": 78720
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 5.01573564122738e-06,
+      "loss": 2.1454,
+      "step": 78730
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 5.013550135501355e-06,
+      "loss": 2.1287,
+      "step": 78740
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 5.01136462977533e-06,
+      "loss": 2.1182,
+      "step": 78750
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 5.009179124049306e-06,
+      "loss": 2.1354,
+      "step": 78760
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 5.00699361832328e-06,
+      "loss": 2.1328,
+      "step": 78770
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 5.004808112597256e-06,
+      "loss": 2.1334,
+      "step": 78780
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 5.00262260687123e-06,
+      "loss": 2.1213,
+      "step": 78790
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 5.000437101145205e-06,
+      "loss": 2.1511,
+      "step": 78800
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 4.9982515954191804e-06,
+      "loss": 2.14,
+      "step": 78810
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 4.9960660896931555e-06,
+      "loss": 2.1529,
+      "step": 78820
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.9938805839671305e-06,
+      "loss": 2.172,
+      "step": 78830
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.9916950782411055e-06,
+      "loss": 2.143,
+      "step": 78840
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.9895095725150806e-06,
+      "loss": 2.1547,
+      "step": 78850
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.987324066789056e-06,
+      "loss": 2.1117,
+      "step": 78860
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.985138561063031e-06,
+      "loss": 2.1398,
+      "step": 78870
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.982953055337006e-06,
+      "loss": 2.1515,
+      "step": 78880
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.980767549610981e-06,
+      "loss": 2.1383,
+      "step": 78890
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.978582043884955e-06,
+      "loss": 2.1332,
+      "step": 78900
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.97639653815893e-06,
+      "loss": 2.1421,
+      "step": 78910
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.974211032432905e-06,
+      "loss": 2.133,
+      "step": 78920
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.97202552670688e-06,
+      "loss": 2.1473,
+      "step": 78930
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.969840020980856e-06,
+      "loss": 2.1259,
+      "step": 78940
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.967654515254831e-06,
+      "loss": 2.1213,
+      "step": 78950
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.965469009528806e-06,
+      "loss": 2.1706,
+      "step": 78960
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.96328350380278e-06,
+      "loss": 2.1544,
+      "step": 78970
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.961097998076755e-06,
+      "loss": 2.1306,
+      "step": 78980
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.95891249235073e-06,
+      "loss": 2.1381,
+      "step": 78990
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.956726986624705e-06,
+      "loss": 2.1381,
+      "step": 79000
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.95454148089868e-06,
+      "loss": 2.1477,
+      "step": 79010
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.952355975172655e-06,
+      "loss": 2.1463,
+      "step": 79020
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.95017046944663e-06,
+      "loss": 2.1381,
+      "step": 79030
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.947984963720605e-06,
+      "loss": 2.1371,
+      "step": 79040
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.94579945799458e-06,
+      "loss": 2.1057,
+      "step": 79050
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.943613952268555e-06,
+      "loss": 2.112,
+      "step": 79060
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.94142844654253e-06,
+      "loss": 2.1142,
+      "step": 79070
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.9392429408165054e-06,
+      "loss": 2.156,
+      "step": 79080
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.9370574350904805e-06,
+      "loss": 2.1158,
+      "step": 79090
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.9348719293644555e-06,
+      "loss": 2.17,
+      "step": 79100
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.9326864236384305e-06,
+      "loss": 2.1069,
+      "step": 79110
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.9305009179124056e-06,
+      "loss": 2.1355,
+      "step": 79120
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.928315412186381e-06,
+      "loss": 2.1385,
+      "step": 79130
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.926129906460355e-06,
+      "loss": 2.148,
+      "step": 79140
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.92394440073433e-06,
+      "loss": 2.1195,
+      "step": 79150
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.921758895008305e-06,
+      "loss": 2.1483,
+      "step": 79160
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.919573389282281e-06,
+      "loss": 2.1331,
+      "step": 79170
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.917387883556256e-06,
+      "loss": 2.1173,
+      "step": 79180
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.915202377830231e-06,
+      "loss": 2.1883,
+      "step": 79190
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.913016872104206e-06,
+      "loss": 2.1333,
+      "step": 79200
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.91083136637818e-06,
+      "loss": 2.1108,
+      "step": 79210
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.908645860652155e-06,
+      "loss": 2.1203,
+      "step": 79220
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.90646035492613e-06,
+      "loss": 2.1447,
+      "step": 79230
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.904274849200105e-06,
+      "loss": 2.1244,
+      "step": 79240
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.90208934347408e-06,
+      "loss": 2.1756,
+      "step": 79250
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.899903837748055e-06,
+      "loss": 2.1364,
+      "step": 79260
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.89771833202203e-06,
+      "loss": 2.1428,
+      "step": 79270
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.895532826296005e-06,
+      "loss": 2.101,
+      "step": 79280
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.89334732056998e-06,
+      "loss": 2.143,
+      "step": 79290
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.891161814843955e-06,
+      "loss": 2.1415,
+      "step": 79300
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.88897630911793e-06,
+      "loss": 2.1229,
+      "step": 79310
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.886790803391905e-06,
+      "loss": 2.1577,
+      "step": 79320
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.88460529766588e-06,
+      "loss": 2.1207,
+      "step": 79330
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.882419791939855e-06,
+      "loss": 2.1203,
+      "step": 79340
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.8802342862138305e-06,
+      "loss": 2.2118,
+      "step": 79350
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 2.1409,
+      "step": 79360
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.8758632747617805e-06,
+      "loss": 2.1624,
+      "step": 79370
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.873677769035755e-06,
+      "loss": 2.1091,
+      "step": 79380
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.87149226330973e-06,
+      "loss": 2.1429,
+      "step": 79390
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.869306757583706e-06,
+      "loss": 2.1307,
+      "step": 79400
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.867121251857681e-06,
+      "loss": 2.1263,
+      "step": 79410
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.864935746131656e-06,
+      "loss": 2.1321,
+      "step": 79420
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.862750240405631e-06,
+      "loss": 2.1494,
+      "step": 79430
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.860564734679606e-06,
+      "loss": 2.1241,
+      "step": 79440
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.85837922895358e-06,
+      "loss": 2.1526,
+      "step": 79450
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.856193723227555e-06,
+      "loss": 2.1099,
+      "step": 79460
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.85400821750153e-06,
+      "loss": 2.1371,
+      "step": 79470
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.851822711775505e-06,
+      "loss": 2.1429,
+      "step": 79480
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.84963720604948e-06,
+      "loss": 2.124,
+      "step": 79490
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.847451700323455e-06,
+      "loss": 2.1337,
+      "step": 79500
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.84526619459743e-06,
+      "loss": 2.1443,
+      "step": 79510
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.843080688871405e-06,
+      "loss": 2.1234,
+      "step": 79520
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.84089518314538e-06,
+      "loss": 2.1381,
+      "step": 79530
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 2.1287,
+      "step": 79540
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.83652417169333e-06,
+      "loss": 2.1519,
+      "step": 79550
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.834338665967305e-06,
+      "loss": 2.1278,
+      "step": 79560
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.83215316024128e-06,
+      "loss": 2.155,
+      "step": 79570
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.829967654515255e-06,
+      "loss": 2.1576,
+      "step": 79580
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.82778214878923e-06,
+      "loss": 2.1718,
+      "step": 79590
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.825596643063205e-06,
+      "loss": 2.1076,
+      "step": 79600
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.8234111373371804e-06,
+      "loss": 2.1403,
+      "step": 79610
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.821225631611155e-06,
+      "loss": 2.1362,
+      "step": 79620
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.8190401258851305e-06,
+      "loss": 2.1671,
+      "step": 79630
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.8168546201591055e-06,
+      "loss": 2.1632,
+      "step": 79640
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.8146691144330806e-06,
+      "loss": 2.1443,
+      "step": 79650
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.812483608707056e-06,
+      "loss": 2.1378,
+      "step": 79660
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.810298102981031e-06,
+      "loss": 2.1252,
+      "step": 79670
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.808112597255006e-06,
+      "loss": 2.1271,
+      "step": 79680
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.80592709152898e-06,
+      "loss": 2.1466,
+      "step": 79690
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.803741585802955e-06,
+      "loss": 2.1534,
+      "step": 79700
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.80155608007693e-06,
+      "loss": 2.1371,
+      "step": 79710
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.799370574350905e-06,
+      "loss": 2.1926,
+      "step": 79720
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.79718506862488e-06,
+      "loss": 2.1633,
+      "step": 79730
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.794999562898855e-06,
+      "loss": 2.1473,
+      "step": 79740
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.79281405717283e-06,
+      "loss": 2.1492,
+      "step": 79750
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.790628551446805e-06,
+      "loss": 2.1509,
+      "step": 79760
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.78844304572078e-06,
+      "loss": 2.1678,
+      "step": 79770
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.786257539994755e-06,
+      "loss": 2.1235,
+      "step": 79780
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.78407203426873e-06,
+      "loss": 2.1388,
+      "step": 79790
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.781886528542705e-06,
+      "loss": 2.2168,
+      "step": 79800
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.77970102281668e-06,
+      "loss": 2.1128,
+      "step": 79810
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.777515517090655e-06,
+      "loss": 2.1315,
+      "step": 79820
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.77533001136463e-06,
+      "loss": 2.1492,
+      "step": 79830
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.773144505638605e-06,
+      "loss": 2.1855,
+      "step": 79840
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.77095899991258e-06,
+      "loss": 2.1253,
+      "step": 79850
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.7687734941865545e-06,
+      "loss": 2.1164,
+      "step": 79860
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.76658798846053e-06,
+      "loss": 2.1526,
+      "step": 79870
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.7644024827345054e-06,
+      "loss": 2.1489,
+      "step": 79880
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.7622169770084805e-06,
+      "loss": 2.1582,
+      "step": 79890
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.7600314712824555e-06,
+      "loss": 2.112,
+      "step": 79900
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.7578459655564305e-06,
+      "loss": 2.1281,
+      "step": 79910
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.755660459830405e-06,
+      "loss": 2.1513,
+      "step": 79920
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.75347495410438e-06,
+      "loss": 2.1624,
+      "step": 79930
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.751289448378355e-06,
+      "loss": 2.1418,
+      "step": 79940
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.74910394265233e-06,
+      "loss": 2.1232,
+      "step": 79950
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.746918436926305e-06,
+      "loss": 2.1575,
+      "step": 79960
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.74473293120028e-06,
+      "loss": 2.1776,
+      "step": 79970
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.742547425474256e-06,
+      "loss": 2.1185,
+      "step": 79980
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.74036191974823e-06,
+      "loss": 2.1778,
+      "step": 79990
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.738176414022205e-06,
+      "loss": 2.145,
+      "step": 80000
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.73599090829618e-06,
+      "loss": 2.1517,
+      "step": 80010
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.733805402570155e-06,
+      "loss": 2.1655,
+      "step": 80020
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.73161989684413e-06,
+      "loss": 2.1254,
+      "step": 80030
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.729434391118105e-06,
+      "loss": 2.1689,
+      "step": 80040
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.72724888539208e-06,
+      "loss": 2.1478,
+      "step": 80050
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.725063379666055e-06,
+      "loss": 2.1444,
+      "step": 80060
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.72287787394003e-06,
+      "loss": 2.1511,
+      "step": 80070
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.720692368214005e-06,
+      "loss": 2.1346,
+      "step": 80080
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.71850686248798e-06,
+      "loss": 2.1554,
+      "step": 80090
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.716321356761955e-06,
+      "loss": 2.168,
+      "step": 80100
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.71413585103593e-06,
+      "loss": 2.163,
+      "step": 80110
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.711950345309905e-06,
+      "loss": 2.1558,
+      "step": 80120
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.70976483958388e-06,
+      "loss": 2.1466,
+      "step": 80130
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.707579333857855e-06,
+      "loss": 2.1529,
+      "step": 80140
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.7053938281318304e-06,
+      "loss": 2.1564,
+      "step": 80150
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.703208322405805e-06,
+      "loss": 2.1818,
+      "step": 80160
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.70102281667978e-06,
+      "loss": 2.1648,
+      "step": 80170
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.698837310953755e-06,
+      "loss": 2.1433,
+      "step": 80180
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.69665180522773e-06,
+      "loss": 2.1117,
+      "step": 80190
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.694466299501705e-06,
+      "loss": 2.1681,
+      "step": 80200
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.69228079377568e-06,
+      "loss": 2.1619,
+      "step": 80210
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.690095288049656e-06,
+      "loss": 2.1204,
+      "step": 80220
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.68790978232363e-06,
+      "loss": 2.1679,
+      "step": 80230
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.685724276597605e-06,
+      "loss": 2.1288,
+      "step": 80240
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.68353877087158e-06,
+      "loss": 2.1576,
+      "step": 80250
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.681353265145555e-06,
+      "loss": 2.1491,
+      "step": 80260
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.67916775941953e-06,
+      "loss": 2.1412,
+      "step": 80270
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.676982253693505e-06,
+      "loss": 2.1391,
+      "step": 80280
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.67479674796748e-06,
+      "loss": 2.1758,
+      "step": 80290
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.672611242241455e-06,
+      "loss": 2.1148,
+      "step": 80300
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.67042573651543e-06,
+      "loss": 2.1558,
+      "step": 80310
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.668240230789405e-06,
+      "loss": 2.1539,
+      "step": 80320
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.66605472506338e-06,
+      "loss": 2.1432,
+      "step": 80330
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.663869219337355e-06,
+      "loss": 2.0968,
+      "step": 80340
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.66168371361133e-06,
+      "loss": 2.1558,
+      "step": 80350
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.659498207885305e-06,
+      "loss": 2.139,
+      "step": 80360
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.65731270215928e-06,
+      "loss": 2.1375,
+      "step": 80370
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.655127196433255e-06,
+      "loss": 2.1329,
+      "step": 80380
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.65294169070723e-06,
+      "loss": 2.1418,
+      "step": 80390
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.6507561849812045e-06,
+      "loss": 2.1887,
+      "step": 80400
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.64857067925518e-06,
+      "loss": 2.1515,
+      "step": 80410
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.646385173529155e-06,
+      "loss": 2.2106,
+      "step": 80420
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.64419966780313e-06,
+      "loss": 2.144,
+      "step": 80430
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.642014162077105e-06,
+      "loss": 2.1841,
+      "step": 80440
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.6398286563510806e-06,
+      "loss": 2.1693,
+      "step": 80450
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.637643150625056e-06,
+      "loss": 2.1408,
+      "step": 80460
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.63545764489903e-06,
+      "loss": 2.1475,
+      "step": 80470
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.633272139173005e-06,
+      "loss": 2.1676,
+      "step": 80480
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.63108663344698e-06,
+      "loss": 2.1341,
+      "step": 80490
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.628901127720955e-06,
+      "loss": 2.1349,
+      "step": 80500
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.62671562199493e-06,
+      "loss": 2.1818,
+      "step": 80510
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.624530116268905e-06,
+      "loss": 2.1204,
+      "step": 80520
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.62234461054288e-06,
+      "loss": 2.1218,
+      "step": 80530
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.620159104816855e-06,
+      "loss": 2.1548,
+      "step": 80540
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.61797359909083e-06,
+      "loss": 2.1679,
+      "step": 80550
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.615788093364805e-06,
+      "loss": 2.1582,
+      "step": 80560
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.61360258763878e-06,
+      "loss": 2.1515,
+      "step": 80570
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.611417081912755e-06,
+      "loss": 2.1327,
+      "step": 80580
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.60923157618673e-06,
+      "loss": 2.1517,
+      "step": 80590
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.607046070460705e-06,
+      "loss": 2.1688,
+      "step": 80600
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.60486056473468e-06,
+      "loss": 2.1247,
+      "step": 80610
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.602675059008655e-06,
+      "loss": 2.1452,
+      "step": 80620
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.60048955328263e-06,
+      "loss": 2.1586,
+      "step": 80630
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.5983040475566045e-06,
+      "loss": 2.1791,
+      "step": 80640
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.5961185418305795e-06,
+      "loss": 2.1462,
+      "step": 80650
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.5939330361045545e-06,
+      "loss": 2.1781,
+      "step": 80660
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.5917475303785296e-06,
+      "loss": 2.1556,
+      "step": 80670
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.5895620246525054e-06,
+      "loss": 2.1186,
+      "step": 80680
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.5873765189264805e-06,
+      "loss": 2.1831,
+      "step": 80690
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.5851910132004555e-06,
+      "loss": 2.1691,
+      "step": 80700
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.58300550747443e-06,
+      "loss": 2.1479,
+      "step": 80710
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.580820001748405e-06,
+      "loss": 2.1493,
+      "step": 80720
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.57863449602238e-06,
+      "loss": 2.128,
+      "step": 80730
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.576448990296355e-06,
+      "loss": 2.1759,
+      "step": 80740
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.57426348457033e-06,
+      "loss": 2.1374,
+      "step": 80750
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.572077978844305e-06,
+      "loss": 2.1504,
+      "step": 80760
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.56989247311828e-06,
+      "loss": 2.1611,
+      "step": 80770
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.567706967392255e-06,
+      "loss": 2.1627,
+      "step": 80780
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.56552146166623e-06,
+      "loss": 2.1522,
+      "step": 80790
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.563335955940205e-06,
+      "loss": 2.1796,
+      "step": 80800
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.56115045021418e-06,
+      "loss": 2.1274,
+      "step": 80810
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.558964944488155e-06,
+      "loss": 2.1548,
+      "step": 80820
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.55677943876213e-06,
+      "loss": 2.1744,
+      "step": 80830
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.554593933036105e-06,
+      "loss": 2.1616,
+      "step": 80840
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.55240842731008e-06,
+      "loss": 2.1523,
+      "step": 80850
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.550222921584055e-06,
+      "loss": 2.1048,
+      "step": 80860
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.54803741585803e-06,
+      "loss": 2.1354,
+      "step": 80870
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.545851910132004e-06,
+      "loss": 2.1633,
+      "step": 80880
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.543666404405979e-06,
+      "loss": 2.122,
+      "step": 80890
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.5414808986799544e-06,
+      "loss": 2.1135,
+      "step": 80900
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.53929539295393e-06,
+      "loss": 2.1258,
+      "step": 80910
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.537109887227905e-06,
+      "loss": 2.1344,
+      "step": 80920
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.53492438150188e-06,
+      "loss": 2.1636,
+      "step": 80930
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.532738875775855e-06,
+      "loss": 2.1467,
+      "step": 80940
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.53055337004983e-06,
+      "loss": 2.1697,
+      "step": 80950
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.528367864323805e-06,
+      "loss": 2.1232,
+      "step": 80960
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.52618235859778e-06,
+      "loss": 2.1892,
+      "step": 80970
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.523996852871755e-06,
+      "loss": 2.1546,
+      "step": 80980
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.52181134714573e-06,
+      "loss": 2.1202,
+      "step": 80990
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.519625841419705e-06,
+      "loss": 2.1728,
+      "step": 81000
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.51744033569368e-06,
+      "loss": 2.1855,
+      "step": 81010
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.515254829967655e-06,
+      "loss": 2.1627,
+      "step": 81020
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.51306932424163e-06,
+      "loss": 2.158,
+      "step": 81030
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.510883818515605e-06,
+      "loss": 2.1538,
+      "step": 81040
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.50869831278958e-06,
+      "loss": 2.1534,
+      "step": 81050
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.506512807063555e-06,
+      "loss": 2.1091,
+      "step": 81060
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.50432730133753e-06,
+      "loss": 2.1634,
+      "step": 81070
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.502141795611505e-06,
+      "loss": 2.182,
+      "step": 81080
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.49995628988548e-06,
+      "loss": 2.2079,
+      "step": 81090
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.497770784159455e-06,
+      "loss": 2.1634,
+      "step": 81100
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.49558527843343e-06,
+      "loss": 2.1339,
+      "step": 81110
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.493399772707404e-06,
+      "loss": 2.1505,
+      "step": 81120
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.491214266981379e-06,
+      "loss": 2.1907,
+      "step": 81130
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.489028761255354e-06,
+      "loss": 2.1576,
+      "step": 81140
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.48684325552933e-06,
+      "loss": 2.1455,
+      "step": 81150
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.484657749803305e-06,
+      "loss": 2.1456,
+      "step": 81160
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.48247224407728e-06,
+      "loss": 2.1884,
+      "step": 81170
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.480286738351255e-06,
+      "loss": 2.1741,
+      "step": 81180
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.4781012326252295e-06,
+      "loss": 2.1507,
+      "step": 81190
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.4759157268992045e-06,
+      "loss": 2.1549,
+      "step": 81200
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.4737302211731796e-06,
+      "loss": 2.164,
+      "step": 81210
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.471544715447155e-06,
+      "loss": 2.1578,
+      "step": 81220
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.46935920972113e-06,
+      "loss": 2.1512,
+      "step": 81230
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.467173703995105e-06,
+      "loss": 2.1933,
+      "step": 81240
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.46498819826908e-06,
+      "loss": 2.1134,
+      "step": 81250
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.462802692543055e-06,
+      "loss": 2.1754,
+      "step": 81260
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.46061718681703e-06,
+      "loss": 2.1421,
+      "step": 81270
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.458431681091005e-06,
+      "loss": 2.1691,
+      "step": 81280
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.45624617536498e-06,
+      "loss": 2.149,
+      "step": 81290
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.454060669638955e-06,
+      "loss": 2.1727,
+      "step": 81300
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.45187516391293e-06,
+      "loss": 2.16,
+      "step": 81310
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.449689658186905e-06,
+      "loss": 2.1559,
+      "step": 81320
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.44750415246088e-06,
+      "loss": 2.1807,
+      "step": 81330
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.445318646734855e-06,
+      "loss": 2.1686,
+      "step": 81340
+    },
+    {
+      "epoch": 16.0,
+      "eval_BLEU": 10.3011,
+      "eval_gen_len": 12.0853,
+      "eval_loss": 3.312498092651367,
+      "eval_runtime": 137.4635,
+      "eval_samples_per_second": 78.297,
+      "eval_steps_per_second": 0.313,
+      "step": 81344
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.44313314100883e-06,
+      "loss": 2.1124,
+      "step": 81350
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.440947635282804e-06,
+      "loss": 2.1107,
+      "step": 81360
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.438762129556779e-06,
+      "loss": 2.0799,
+      "step": 81370
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.436576623830755e-06,
+      "loss": 2.0496,
+      "step": 81380
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.43439111810473e-06,
+      "loss": 2.0918,
+      "step": 81390
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.432205612378705e-06,
+      "loss": 2.0939,
+      "step": 81400
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.43002010665268e-06,
+      "loss": 2.0827,
+      "step": 81410
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.427834600926655e-06,
+      "loss": 2.0494,
+      "step": 81420
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.425649095200629e-06,
+      "loss": 2.0652,
+      "step": 81430
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.4234635894746045e-06,
+      "loss": 2.0912,
+      "step": 81440
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.4212780837485795e-06,
+      "loss": 2.0709,
+      "step": 81450
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.4190925780225545e-06,
+      "loss": 2.068,
+      "step": 81460
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.4169070722965295e-06,
+      "loss": 2.0855,
+      "step": 81470
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.414721566570505e-06,
+      "loss": 2.0817,
+      "step": 81480
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.41253606084448e-06,
+      "loss": 2.1062,
+      "step": 81490
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.410350555118455e-06,
+      "loss": 2.0845,
+      "step": 81500
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.40816504939243e-06,
+      "loss": 2.1002,
+      "step": 81510
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.405979543666405e-06,
+      "loss": 2.048,
+      "step": 81520
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.40379403794038e-06,
+      "loss": 2.0929,
+      "step": 81530
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.401608532214355e-06,
+      "loss": 2.1186,
+      "step": 81540
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.39942302648833e-06,
+      "loss": 2.0836,
+      "step": 81550
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.397237520762305e-06,
+      "loss": 2.0702,
+      "step": 81560
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.39505201503628e-06,
+      "loss": 2.0787,
+      "step": 81570
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.392866509310255e-06,
+      "loss": 2.0937,
+      "step": 81580
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.39068100358423e-06,
+      "loss": 2.0837,
+      "step": 81590
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.388495497858204e-06,
+      "loss": 2.0487,
+      "step": 81600
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.38630999213218e-06,
+      "loss": 2.0743,
+      "step": 81610
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.384124486406155e-06,
+      "loss": 2.0726,
+      "step": 81620
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.38193898068013e-06,
+      "loss": 2.1013,
+      "step": 81630
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.379753474954105e-06,
+      "loss": 2.0781,
+      "step": 81640
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.37756796922808e-06,
+      "loss": 2.083,
+      "step": 81650
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.375382463502055e-06,
+      "loss": 2.086,
+      "step": 81660
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.373196957776029e-06,
+      "loss": 2.1331,
+      "step": 81670
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.371011452050004e-06,
+      "loss": 2.0947,
+      "step": 81680
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.368825946323979e-06,
+      "loss": 2.0784,
+      "step": 81690
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.3666404405979544e-06,
+      "loss": 2.1003,
+      "step": 81700
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.3644549348719295e-06,
+      "loss": 2.1255,
+      "step": 81710
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.3622694291459045e-06,
+      "loss": 2.1109,
+      "step": 81720
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.36008392341988e-06,
+      "loss": 2.1057,
+      "step": 81730
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.3578984176938546e-06,
+      "loss": 2.0988,
+      "step": 81740
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.35571291196783e-06,
+      "loss": 2.0825,
+      "step": 81750
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.353527406241805e-06,
+      "loss": 2.0953,
+      "step": 81760
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.35134190051578e-06,
+      "loss": 2.0958,
+      "step": 81770
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.349156394789755e-06,
+      "loss": 2.1099,
+      "step": 81780
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.34697088906373e-06,
+      "loss": 2.0947,
+      "step": 81790
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.344785383337705e-06,
+      "loss": 2.1333,
+      "step": 81800
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.34259987761168e-06,
+      "loss": 2.1015,
+      "step": 81810
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.340414371885655e-06,
+      "loss": 2.0625,
+      "step": 81820
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.33822886615963e-06,
+      "loss": 2.0846,
+      "step": 81830
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.336043360433605e-06,
+      "loss": 2.1007,
+      "step": 81840
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.33385785470758e-06,
+      "loss": 2.108,
+      "step": 81850
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.331672348981555e-06,
+      "loss": 2.0964,
+      "step": 81860
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.32948684325553e-06,
+      "loss": 2.0817,
+      "step": 81870
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.327301337529505e-06,
+      "loss": 2.0914,
+      "step": 81880
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.32511583180348e-06,
+      "loss": 2.1044,
+      "step": 81890
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.322930326077455e-06,
+      "loss": 2.0625,
+      "step": 81900
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.320744820351429e-06,
+      "loss": 2.081,
+      "step": 81910
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.318559314625404e-06,
+      "loss": 2.1044,
+      "step": 81920
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.316373808899379e-06,
+      "loss": 2.0666,
+      "step": 81930
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.314188303173354e-06,
+      "loss": 2.0983,
+      "step": 81940
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.312002797447329e-06,
+      "loss": 2.0904,
+      "step": 81950
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.309817291721305e-06,
+      "loss": 2.1193,
+      "step": 81960
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.30763178599528e-06,
+      "loss": 2.0926,
+      "step": 81970
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.3054462802692545e-06,
+      "loss": 2.0792,
+      "step": 81980
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.3032607745432295e-06,
+      "loss": 2.0788,
+      "step": 81990
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.3010752688172045e-06,
+      "loss": 2.128,
+      "step": 82000
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.2988897630911796e-06,
+      "loss": 2.0796,
+      "step": 82010
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.296704257365155e-06,
+      "loss": 2.0998,
+      "step": 82020
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.29451875163913e-06,
+      "loss": 2.0777,
+      "step": 82030
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.292333245913105e-06,
+      "loss": 2.0949,
+      "step": 82040
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.29014774018708e-06,
+      "loss": 2.1237,
+      "step": 82050
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.287962234461055e-06,
+      "loss": 2.1068,
+      "step": 82060
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.28577672873503e-06,
+      "loss": 2.0873,
+      "step": 82070
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.283591223009005e-06,
+      "loss": 2.0554,
+      "step": 82080
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.28140571728298e-06,
+      "loss": 2.0742,
+      "step": 82090
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.279220211556955e-06,
+      "loss": 2.0916,
+      "step": 82100
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.27703470583093e-06,
+      "loss": 2.098,
+      "step": 82110
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.274849200104905e-06,
+      "loss": 2.1037,
+      "step": 82120
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.27266369437888e-06,
+      "loss": 2.0971,
+      "step": 82130
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.270478188652855e-06,
+      "loss": 2.0797,
+      "step": 82140
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.268292682926829e-06,
+      "loss": 2.0785,
+      "step": 82150
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.266107177200804e-06,
+      "loss": 2.1133,
+      "step": 82160
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.263921671474779e-06,
+      "loss": 2.0744,
+      "step": 82170
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.261736165748754e-06,
+      "loss": 2.0693,
+      "step": 82180
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.25955066002273e-06,
+      "loss": 2.1461,
+      "step": 82190
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.257365154296705e-06,
+      "loss": 2.0706,
+      "step": 82200
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.25517964857068e-06,
+      "loss": 2.0883,
+      "step": 82210
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.252994142844654e-06,
+      "loss": 2.1184,
+      "step": 82220
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.250808637118629e-06,
+      "loss": 2.0625,
+      "step": 82230
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.2486231313926044e-06,
+      "loss": 2.0901,
+      "step": 82240
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.2464376256665795e-06,
+      "loss": 2.098,
+      "step": 82250
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.2442521199405545e-06,
+      "loss": 2.0833,
+      "step": 82260
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.2420666142145295e-06,
+      "loss": 2.1231,
+      "step": 82270
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.2398811084885046e-06,
+      "loss": 2.0919,
+      "step": 82280
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.23769560276248e-06,
+      "loss": 2.0966,
+      "step": 82290
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.235510097036455e-06,
+      "loss": 2.0907,
+      "step": 82300
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.23332459131043e-06,
+      "loss": 2.143,
+      "step": 82310
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.231139085584405e-06,
+      "loss": 2.1029,
+      "step": 82320
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.22895357985838e-06,
+      "loss": 2.1215,
+      "step": 82330
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.226768074132355e-06,
+      "loss": 2.1217,
+      "step": 82340
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.22458256840633e-06,
+      "loss": 2.1145,
+      "step": 82350
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.222397062680305e-06,
+      "loss": 2.0539,
+      "step": 82360
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.22021155695428e-06,
+      "loss": 2.1088,
+      "step": 82370
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.218026051228255e-06,
+      "loss": 2.1196,
+      "step": 82380
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.215840545502229e-06,
+      "loss": 2.0962,
+      "step": 82390
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.213655039776204e-06,
+      "loss": 2.0784,
+      "step": 82400
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.211469534050179e-06,
+      "loss": 2.0756,
+      "step": 82410
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.209284028324154e-06,
+      "loss": 2.1079,
+      "step": 82420
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.20709852259813e-06,
+      "loss": 2.1303,
+      "step": 82430
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.204913016872105e-06,
+      "loss": 2.0799,
+      "step": 82440
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.202727511146079e-06,
+      "loss": 2.0689,
+      "step": 82450
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.200542005420054e-06,
+      "loss": 2.0988,
+      "step": 82460
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.198356499694029e-06,
+      "loss": 2.1182,
+      "step": 82470
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.196170993968004e-06,
+      "loss": 2.1249,
+      "step": 82480
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.193985488241979e-06,
+      "loss": 2.0869,
+      "step": 82490
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.1917999825159544e-06,
+      "loss": 2.0757,
+      "step": 82500
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.1896144767899295e-06,
+      "loss": 2.0786,
+      "step": 82510
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.1874289710639045e-06,
+      "loss": 2.1008,
+      "step": 82520
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.1852434653378795e-06,
+      "loss": 2.1389,
+      "step": 82530
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.1830579596118546e-06,
+      "loss": 2.1252,
+      "step": 82540
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.18087245388583e-06,
+      "loss": 2.1267,
+      "step": 82550
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.178686948159805e-06,
+      "loss": 2.1393,
+      "step": 82560
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.17650144243378e-06,
+      "loss": 2.093,
+      "step": 82570
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.174315936707755e-06,
+      "loss": 2.0854,
+      "step": 82580
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.17213043098173e-06,
+      "loss": 2.0753,
+      "step": 82590
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.169944925255705e-06,
+      "loss": 2.0995,
+      "step": 82600
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.16775941952968e-06,
+      "loss": 2.068,
+      "step": 82610
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.165573913803655e-06,
+      "loss": 2.0899,
+      "step": 82620
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.163388408077629e-06,
+      "loss": 2.1194,
+      "step": 82630
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.161202902351604e-06,
+      "loss": 2.1119,
+      "step": 82640
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.159017396625579e-06,
+      "loss": 2.1108,
+      "step": 82650
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.156831890899555e-06,
+      "loss": 2.0838,
+      "step": 82660
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.15464638517353e-06,
+      "loss": 2.0884,
+      "step": 82670
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.152460879447505e-06,
+      "loss": 2.0923,
+      "step": 82680
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.150275373721479e-06,
+      "loss": 2.0744,
+      "step": 82690
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.148089867995454e-06,
+      "loss": 2.0822,
+      "step": 82700
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.145904362269429e-06,
+      "loss": 2.0996,
+      "step": 82710
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.143718856543404e-06,
+      "loss": 2.1491,
+      "step": 82720
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.141533350817379e-06,
+      "loss": 2.0575,
+      "step": 82730
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.139347845091354e-06,
+      "loss": 2.0908,
+      "step": 82740
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.137162339365329e-06,
+      "loss": 2.0345,
+      "step": 82750
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.134976833639304e-06,
+      "loss": 2.0923,
+      "step": 82760
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.1327913279132794e-06,
+      "loss": 2.1071,
+      "step": 82770
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.1306058221872545e-06,
+      "loss": 2.1353,
+      "step": 82780
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.1284203164612295e-06,
+      "loss": 2.132,
+      "step": 82790
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.1262348107352045e-06,
+      "loss": 2.0936,
+      "step": 82800
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.1240493050091796e-06,
+      "loss": 2.1023,
+      "step": 82810
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.121863799283155e-06,
+      "loss": 2.0943,
+      "step": 82820
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.11967829355713e-06,
+      "loss": 2.0859,
+      "step": 82830
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.117492787831105e-06,
+      "loss": 2.0833,
+      "step": 82840
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.11530728210508e-06,
+      "loss": 2.0729,
+      "step": 82850
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.113121776379055e-06,
+      "loss": 2.0976,
+      "step": 82860
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.110936270653029e-06,
+      "loss": 2.1211,
+      "step": 82870
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.108750764927004e-06,
+      "loss": 2.0794,
+      "step": 82880
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.10656525920098e-06,
+      "loss": 2.0484,
+      "step": 82890
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.104379753474955e-06,
+      "loss": 2.095,
+      "step": 82900
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.10219424774893e-06,
+      "loss": 2.0867,
+      "step": 82910
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.100008742022905e-06,
+      "loss": 2.1074,
+      "step": 82920
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.097823236296879e-06,
+      "loss": 2.1486,
+      "step": 82930
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.095637730570854e-06,
+      "loss": 2.0841,
+      "step": 82940
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.093452224844829e-06,
+      "loss": 2.1209,
+      "step": 82950
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.091266719118804e-06,
+      "loss": 2.0602,
+      "step": 82960
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.089081213392779e-06,
+      "loss": 2.0937,
+      "step": 82970
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.086895707666754e-06,
+      "loss": 2.1104,
+      "step": 82980
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.084710201940729e-06,
+      "loss": 2.0917,
+      "step": 82990
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.082524696214704e-06,
+      "loss": 2.1008,
+      "step": 83000
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.080339190488679e-06,
+      "loss": 2.1306,
+      "step": 83010
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.078153684762654e-06,
+      "loss": 2.1098,
+      "step": 83020
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.075968179036629e-06,
+      "loss": 2.0976,
+      "step": 83030
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.0737826733106044e-06,
+      "loss": 2.12,
+      "step": 83040
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.0715971675845795e-06,
+      "loss": 2.0756,
+      "step": 83050
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.0694116618585545e-06,
+      "loss": 2.0692,
+      "step": 83060
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.0672261561325295e-06,
+      "loss": 2.1351,
+      "step": 83070
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.0650406504065046e-06,
+      "loss": 2.0737,
+      "step": 83080
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.06285514468048e-06,
+      "loss": 2.1086,
+      "step": 83090
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.060669638954454e-06,
+      "loss": 2.0876,
+      "step": 83100
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.058484133228429e-06,
+      "loss": 2.1273,
+      "step": 83110
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.056298627502405e-06,
+      "loss": 2.0987,
+      "step": 83120
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.05411312177638e-06,
+      "loss": 2.1161,
+      "step": 83130
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.051927616050355e-06,
+      "loss": 2.0871,
+      "step": 83140
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.04974211032433e-06,
+      "loss": 2.1122,
+      "step": 83150
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.047556604598305e-06,
+      "loss": 2.1067,
+      "step": 83160
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.045371098872279e-06,
+      "loss": 2.0546,
+      "step": 83170
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.043185593146254e-06,
+      "loss": 2.094,
+      "step": 83180
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.041000087420229e-06,
+      "loss": 2.0894,
+      "step": 83190
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.038814581694204e-06,
+      "loss": 2.0875,
+      "step": 83200
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.036629075968179e-06,
+      "loss": 2.1011,
+      "step": 83210
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.034443570242154e-06,
+      "loss": 2.1152,
+      "step": 83220
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 2.0885,
+      "step": 83230
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.030072558790104e-06,
+      "loss": 2.0748,
+      "step": 83240
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.027887053064079e-06,
+      "loss": 2.0866,
+      "step": 83250
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.025701547338054e-06,
+      "loss": 2.0913,
+      "step": 83260
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.023516041612029e-06,
+      "loss": 2.0962,
+      "step": 83270
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.021330535886004e-06,
+      "loss": 2.0771,
+      "step": 83280
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.019145030159979e-06,
+      "loss": 2.0957,
+      "step": 83290
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.016959524433954e-06,
+      "loss": 2.0881,
+      "step": 83300
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.0147740187079295e-06,
+      "loss": 2.1256,
+      "step": 83310
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.0125885129819045e-06,
+      "loss": 2.0901,
+      "step": 83320
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.0104030072558795e-06,
+      "loss": 2.108,
+      "step": 83330
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.008217501529854e-06,
+      "loss": 2.1109,
+      "step": 83340
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.006031995803829e-06,
+      "loss": 2.0871,
+      "step": 83350
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.003846490077805e-06,
+      "loss": 2.1146,
+      "step": 83360
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.00166098435178e-06,
+      "loss": 2.12,
+      "step": 83370
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.999475478625755e-06,
+      "loss": 2.0654,
+      "step": 83380
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.99728997289973e-06,
+      "loss": 2.1163,
+      "step": 83390
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.995104467173705e-06,
+      "loss": 2.1079,
+      "step": 83400
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.992918961447679e-06,
+      "loss": 2.0934,
+      "step": 83410
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.990733455721654e-06,
+      "loss": 2.1183,
+      "step": 83420
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.988547949995629e-06,
+      "loss": 2.0853,
+      "step": 83430
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.986362444269604e-06,
+      "loss": 2.1074,
+      "step": 83440
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.984176938543579e-06,
+      "loss": 2.0637,
+      "step": 83450
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.981991432817554e-06,
+      "loss": 2.092,
+      "step": 83460
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.97980592709153e-06,
+      "loss": 2.0983,
+      "step": 83470
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.977620421365504e-06,
+      "loss": 2.1128,
+      "step": 83480
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.975434915639479e-06,
+      "loss": 2.0516,
+      "step": 83490
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.973249409913454e-06,
+      "loss": 2.0736,
+      "step": 83500
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.971063904187429e-06,
+      "loss": 2.0786,
+      "step": 83510
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.968878398461404e-06,
+      "loss": 2.113,
+      "step": 83520
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.966692892735379e-06,
+      "loss": 2.083,
+      "step": 83530
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.964507387009354e-06,
+      "loss": 2.1085,
+      "step": 83540
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.962321881283329e-06,
+      "loss": 2.065,
+      "step": 83550
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.960136375557304e-06,
+      "loss": 2.1153,
+      "step": 83560
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.9579508698312794e-06,
+      "loss": 2.1047,
+      "step": 83570
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.955765364105254e-06,
+      "loss": 2.0949,
+      "step": 83580
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.9535798583792295e-06,
+      "loss": 2.1015,
+      "step": 83590
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.9513943526532045e-06,
+      "loss": 2.096,
+      "step": 83600
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.9492088469271796e-06,
+      "loss": 2.0676,
+      "step": 83610
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.947023341201155e-06,
+      "loss": 2.0903,
+      "step": 83620
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.94483783547513e-06,
+      "loss": 2.0875,
+      "step": 83630
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.942652329749105e-06,
+      "loss": 2.0957,
+      "step": 83640
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.940466824023079e-06,
+      "loss": 2.0915,
+      "step": 83650
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.938281318297054e-06,
+      "loss": 2.0807,
+      "step": 83660
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.936095812571029e-06,
+      "loss": 2.0816,
+      "step": 83670
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.933910306845004e-06,
+      "loss": 2.1296,
+      "step": 83680
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.931724801118979e-06,
+      "loss": 2.1241,
+      "step": 83690
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.929539295392954e-06,
+      "loss": 2.0738,
+      "step": 83700
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.92735378966693e-06,
+      "loss": 2.0996,
+      "step": 83710
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.925168283940904e-06,
+      "loss": 2.0788,
+      "step": 83720
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.922982778214879e-06,
+      "loss": 2.1028,
+      "step": 83730
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.920797272488854e-06,
+      "loss": 2.0656,
+      "step": 83740
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.918611766762829e-06,
+      "loss": 2.0969,
+      "step": 83750
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.916426261036804e-06,
+      "loss": 2.0982,
+      "step": 83760
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.914240755310779e-06,
+      "loss": 2.1139,
+      "step": 83770
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.912055249584754e-06,
+      "loss": 2.0786,
+      "step": 83780
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.909869743858729e-06,
+      "loss": 2.1317,
+      "step": 83790
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.907684238132704e-06,
+      "loss": 2.0766,
+      "step": 83800
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.905498732406679e-06,
+      "loss": 2.1229,
+      "step": 83810
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.903313226680654e-06,
+      "loss": 2.0653,
+      "step": 83820
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.901127720954629e-06,
+      "loss": 2.0987,
+      "step": 83830
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.8989422152286044e-06,
+      "loss": 2.131,
+      "step": 83840
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.8967567095025795e-06,
+      "loss": 2.1122,
+      "step": 83850
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.8945712037765545e-06,
+      "loss": 2.1343,
+      "step": 83860
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.8923856980505295e-06,
+      "loss": 2.1032,
+      "step": 83870
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.8902001923245046e-06,
+      "loss": 2.0909,
+      "step": 83880
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.888014686598479e-06,
+      "loss": 2.1035,
+      "step": 83890
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.885829180872454e-06,
+      "loss": 2.1001,
+      "step": 83900
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.883643675146429e-06,
+      "loss": 2.1049,
+      "step": 83910
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.881458169420404e-06,
+      "loss": 2.0872,
+      "step": 83920
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.879272663694379e-06,
+      "loss": 2.0864,
+      "step": 83930
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.877087157968355e-06,
+      "loss": 2.1293,
+      "step": 83940
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.87490165224233e-06,
+      "loss": 2.1079,
+      "step": 83950
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.872716146516304e-06,
+      "loss": 2.1453,
+      "step": 83960
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.870530640790279e-06,
+      "loss": 2.1252,
+      "step": 83970
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.868345135064254e-06,
+      "loss": 2.117,
+      "step": 83980
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.866159629338229e-06,
+      "loss": 2.1272,
+      "step": 83990
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.863974123612204e-06,
+      "loss": 2.0834,
+      "step": 84000
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.861788617886179e-06,
+      "loss": 2.072,
+      "step": 84010
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.859603112160154e-06,
+      "loss": 2.1005,
+      "step": 84020
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.857417606434129e-06,
+      "loss": 2.143,
+      "step": 84030
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.855232100708104e-06,
+      "loss": 2.0798,
+      "step": 84040
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.853046594982079e-06,
+      "loss": 2.0929,
+      "step": 84050
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.850861089256054e-06,
+      "loss": 2.1417,
+      "step": 84060
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.848675583530029e-06,
+      "loss": 2.1019,
+      "step": 84070
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.846490077804004e-06,
+      "loss": 2.1067,
+      "step": 84080
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.844304572077979e-06,
+      "loss": 2.1242,
+      "step": 84090
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.842119066351954e-06,
+      "loss": 2.1448,
+      "step": 84100
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.8399335606259294e-06,
+      "loss": 2.1226,
+      "step": 84110
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.8377480548999045e-06,
+      "loss": 2.1003,
+      "step": 84120
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.835562549173879e-06,
+      "loss": 2.0904,
+      "step": 84130
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.833377043447854e-06,
+      "loss": 2.0891,
+      "step": 84140
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.831191537721829e-06,
+      "loss": 2.1078,
+      "step": 84150
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.829006031995804e-06,
+      "loss": 2.0742,
+      "step": 84160
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.82682052626978e-06,
+      "loss": 2.0798,
+      "step": 84170
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.824635020543755e-06,
+      "loss": 2.0806,
+      "step": 84180
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.82244951481773e-06,
+      "loss": 2.1559,
+      "step": 84190
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.820264009091704e-06,
+      "loss": 2.1111,
+      "step": 84200
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.818078503365679e-06,
+      "loss": 2.1045,
+      "step": 84210
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.815892997639654e-06,
+      "loss": 2.1374,
+      "step": 84220
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.813707491913629e-06,
+      "loss": 2.1395,
+      "step": 84230
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.811521986187604e-06,
+      "loss": 2.1205,
+      "step": 84240
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.809336480461579e-06,
+      "loss": 2.102,
+      "step": 84250
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.807150974735554e-06,
+      "loss": 2.1213,
+      "step": 84260
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.804965469009529e-06,
+      "loss": 2.1274,
+      "step": 84270
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.8027799632835037e-06,
+      "loss": 2.0954,
+      "step": 84280
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.800594457557479e-06,
+      "loss": 2.1115,
+      "step": 84290
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.798408951831454e-06,
+      "loss": 2.1163,
+      "step": 84300
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.7962234461054292e-06,
+      "loss": 2.1056,
+      "step": 84310
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.7940379403794043e-06,
+      "loss": 2.0999,
+      "step": 84320
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.7918524346533793e-06,
+      "loss": 2.117,
+      "step": 84330
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.789666928927354e-06,
+      "loss": 2.1445,
+      "step": 84340
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.787481423201329e-06,
+      "loss": 2.1134,
+      "step": 84350
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.785295917475304e-06,
+      "loss": 2.0932,
+      "step": 84360
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.783110411749279e-06,
+      "loss": 2.1202,
+      "step": 84370
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.780924906023254e-06,
+      "loss": 2.0772,
+      "step": 84380
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.778739400297229e-06,
+      "loss": 2.1097,
+      "step": 84390
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.7765538945712045e-06,
+      "loss": 2.0969,
+      "step": 84400
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.774368388845179e-06,
+      "loss": 2.1279,
+      "step": 84410
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.772182883119154e-06,
+      "loss": 2.1061,
+      "step": 84420
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.769997377393129e-06,
+      "loss": 2.095,
+      "step": 84430
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.7678118716671042e-06,
+      "loss": 2.1059,
+      "step": 84440
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.7656263659410793e-06,
+      "loss": 2.1021,
+      "step": 84450
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.763440860215054e-06,
+      "loss": 2.1243,
+      "step": 84460
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.761255354489029e-06,
+      "loss": 2.0981,
+      "step": 84470
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.759069848763004e-06,
+      "loss": 2.0727,
+      "step": 84480
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.756884343036979e-06,
+      "loss": 2.1061,
+      "step": 84490
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.754698837310954e-06,
+      "loss": 2.0988,
+      "step": 84500
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.752513331584929e-06,
+      "loss": 2.1211,
+      "step": 84510
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.7503278258589045e-06,
+      "loss": 2.0772,
+      "step": 84520
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.748142320132879e-06,
+      "loss": 2.1137,
+      "step": 84530
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.745956814406854e-06,
+      "loss": 2.1131,
+      "step": 84540
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.743771308680829e-06,
+      "loss": 2.0953,
+      "step": 84550
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.741585802954804e-06,
+      "loss": 2.1008,
+      "step": 84560
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.739400297228779e-06,
+      "loss": 2.1201,
+      "step": 84570
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.737214791502754e-06,
+      "loss": 2.121,
+      "step": 84580
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.735029285776729e-06,
+      "loss": 2.0951,
+      "step": 84590
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.732843780050704e-06,
+      "loss": 2.1114,
+      "step": 84600
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.730658274324679e-06,
+      "loss": 2.1376,
+      "step": 84610
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.728472768598654e-06,
+      "loss": 2.1007,
+      "step": 84620
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.7262872628726286e-06,
+      "loss": 2.1091,
+      "step": 84630
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.7241017571466044e-06,
+      "loss": 2.1549,
+      "step": 84640
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.721916251420579e-06,
+      "loss": 2.0858,
+      "step": 84650
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.719730745694554e-06,
+      "loss": 2.1081,
+      "step": 84660
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.717545239968529e-06,
+      "loss": 2.1112,
+      "step": 84670
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.715359734242504e-06,
+      "loss": 2.1075,
+      "step": 84680
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.713174228516479e-06,
+      "loss": 2.1024,
+      "step": 84690
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.7109887227904538e-06,
+      "loss": 2.1046,
+      "step": 84700
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.708803217064429e-06,
+      "loss": 2.0855,
+      "step": 84710
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.706617711338404e-06,
+      "loss": 2.0705,
+      "step": 84720
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.704432205612379e-06,
+      "loss": 2.1146,
+      "step": 84730
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.702246699886354e-06,
+      "loss": 2.1145,
+      "step": 84740
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.7000611941603294e-06,
+      "loss": 2.1019,
+      "step": 84750
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.6978756884343044e-06,
+      "loss": 2.1011,
+      "step": 84760
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.695690182708279e-06,
+      "loss": 2.0847,
+      "step": 84770
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.693504676982254e-06,
+      "loss": 2.1379,
+      "step": 84780
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.691319171256229e-06,
+      "loss": 2.0906,
+      "step": 84790
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.689133665530204e-06,
+      "loss": 2.0918,
+      "step": 84800
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.686948159804179e-06,
+      "loss": 2.1246,
+      "step": 84810
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.6847626540781537e-06,
+      "loss": 2.1033,
+      "step": 84820
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.6825771483521288e-06,
+      "loss": 2.0733,
+      "step": 84830
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.680391642626104e-06,
+      "loss": 2.109,
+      "step": 84840
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.678206136900079e-06,
+      "loss": 2.1097,
+      "step": 84850
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.676020631174054e-06,
+      "loss": 2.0872,
+      "step": 84860
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.6738351254480293e-06,
+      "loss": 2.0994,
+      "step": 84870
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.6716496197220043e-06,
+      "loss": 2.1265,
+      "step": 84880
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.669464113995979e-06,
+      "loss": 2.0963,
+      "step": 84890
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.667278608269954e-06,
+      "loss": 2.1514,
+      "step": 84900
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.665093102543929e-06,
+      "loss": 2.1637,
+      "step": 84910
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.662907596817904e-06,
+      "loss": 2.0785,
+      "step": 84920
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.660722091091879e-06,
+      "loss": 2.1367,
+      "step": 84930
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 2.1162,
+      "step": 84940
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.6563510796398287e-06,
+      "loss": 2.1131,
+      "step": 84950
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.6541655739138038e-06,
+      "loss": 2.152,
+      "step": 84960
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.6519800681877788e-06,
+      "loss": 2.1146,
+      "step": 84970
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.649794562461754e-06,
+      "loss": 2.1372,
+      "step": 84980
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.6476090567357293e-06,
+      "loss": 2.1178,
+      "step": 84990
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.6454235510097043e-06,
+      "loss": 2.1041,
+      "step": 85000
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.643238045283679e-06,
+      "loss": 2.0767,
+      "step": 85010
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.641052539557654e-06,
+      "loss": 2.1085,
+      "step": 85020
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.638867033831629e-06,
+      "loss": 2.0725,
+      "step": 85030
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.636681528105604e-06,
+      "loss": 2.1221,
+      "step": 85040
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.634496022379579e-06,
+      "loss": 2.1114,
+      "step": 85050
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.6323105166535536e-06,
+      "loss": 2.1075,
+      "step": 85060
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.6301250109275287e-06,
+      "loss": 2.0726,
+      "step": 85070
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.6279395052015037e-06,
+      "loss": 2.0802,
+      "step": 85080
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.6257539994754787e-06,
+      "loss": 2.1094,
+      "step": 85090
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.623568493749454e-06,
+      "loss": 2.0672,
+      "step": 85100
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.6213829880234292e-06,
+      "loss": 2.1053,
+      "step": 85110
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.6191974822974043e-06,
+      "loss": 2.1056,
+      "step": 85120
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.617011976571379e-06,
+      "loss": 2.1284,
+      "step": 85130
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.614826470845354e-06,
+      "loss": 2.1073,
+      "step": 85140
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.612640965119329e-06,
+      "loss": 2.1268,
+      "step": 85150
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.610455459393304e-06,
+      "loss": 2.1078,
+      "step": 85160
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.608269953667279e-06,
+      "loss": 2.1311,
+      "step": 85170
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.6060844479412536e-06,
+      "loss": 2.0743,
+      "step": 85180
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.6038989422152286e-06,
+      "loss": 2.0927,
+      "step": 85190
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.6017134364892037e-06,
+      "loss": 2.093,
+      "step": 85200
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.5995279307631787e-06,
+      "loss": 2.128,
+      "step": 85210
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.597342425037154e-06,
+      "loss": 2.1288,
+      "step": 85220
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.595156919311129e-06,
+      "loss": 2.1428,
+      "step": 85230
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.5929714135851042e-06,
+      "loss": 2.078,
+      "step": 85240
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.590785907859079e-06,
+      "loss": 2.0987,
+      "step": 85250
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.588600402133054e-06,
+      "loss": 2.0974,
+      "step": 85260
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.586414896407029e-06,
+      "loss": 2.1155,
+      "step": 85270
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.584229390681004e-06,
+      "loss": 2.1151,
+      "step": 85280
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.582043884954979e-06,
+      "loss": 2.0991,
+      "step": 85290
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.5798583792289536e-06,
+      "loss": 2.0618,
+      "step": 85300
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.5776728735029286e-06,
+      "loss": 2.1363,
+      "step": 85310
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.5754873677769036e-06,
+      "loss": 2.0955,
+      "step": 85320
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.5733018620508787e-06,
+      "loss": 2.119,
+      "step": 85330
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.571116356324854e-06,
+      "loss": 2.1102,
+      "step": 85340
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.568930850598829e-06,
+      "loss": 2.091,
+      "step": 85350
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.566745344872804e-06,
+      "loss": 2.1611,
+      "step": 85360
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.5645598391467788e-06,
+      "loss": 2.1267,
+      "step": 85370
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.562374333420754e-06,
+      "loss": 2.1377,
+      "step": 85380
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.560188827694729e-06,
+      "loss": 2.1036,
+      "step": 85390
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.558003321968704e-06,
+      "loss": 2.0846,
+      "step": 85400
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.555817816242679e-06,
+      "loss": 2.1348,
+      "step": 85410
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.5536323105166535e-06,
+      "loss": 2.1073,
+      "step": 85420
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.5514468047906285e-06,
+      "loss": 2.1417,
+      "step": 85430
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.5492612990646036e-06,
+      "loss": 2.1354,
+      "step": 85440
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.547075793338579e-06,
+      "loss": 2.1189,
+      "step": 85450
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.544890287612554e-06,
+      "loss": 2.1004,
+      "step": 85460
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.542704781886529e-06,
+      "loss": 2.125,
+      "step": 85470
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.540519276160504e-06,
+      "loss": 2.1037,
+      "step": 85480
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.5383337704344787e-06,
+      "loss": 2.1189,
+      "step": 85490
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.5361482647084538e-06,
+      "loss": 2.073,
+      "step": 85500
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.533962758982429e-06,
+      "loss": 2.0507,
+      "step": 85510
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.531777253256404e-06,
+      "loss": 2.134,
+      "step": 85520
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.529591747530379e-06,
+      "loss": 2.1137,
+      "step": 85530
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.5274062418043535e-06,
+      "loss": 2.1582,
+      "step": 85540
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.5252207360783285e-06,
+      "loss": 2.0842,
+      "step": 85550
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.5230352303523035e-06,
+      "loss": 2.1035,
+      "step": 85560
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.520849724626279e-06,
+      "loss": 2.1015,
+      "step": 85570
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.518664218900254e-06,
+      "loss": 2.1031,
+      "step": 85580
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.516478713174229e-06,
+      "loss": 2.0793,
+      "step": 85590
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.514293207448204e-06,
+      "loss": 2.1013,
+      "step": 85600
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.5121077017221787e-06,
+      "loss": 2.0917,
+      "step": 85610
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.5099221959961537e-06,
+      "loss": 2.1185,
+      "step": 85620
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.5077366902701288e-06,
+      "loss": 2.0605,
+      "step": 85630
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.505551184544104e-06,
+      "loss": 2.1105,
+      "step": 85640
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.503365678818079e-06,
+      "loss": 2.1016,
+      "step": 85650
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.5011801730920534e-06,
+      "loss": 2.1028,
+      "step": 85660
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.4989946673660285e-06,
+      "loss": 2.086,
+      "step": 85670
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.496809161640004e-06,
+      "loss": 2.1091,
+      "step": 85680
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.494623655913979e-06,
+      "loss": 2.1211,
+      "step": 85690
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.492438150187954e-06,
+      "loss": 2.1118,
+      "step": 85700
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.490252644461929e-06,
+      "loss": 2.1409,
+      "step": 85710
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.488067138735904e-06,
+      "loss": 2.127,
+      "step": 85720
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.4858816330098787e-06,
+      "loss": 2.1319,
+      "step": 85730
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.4836961272838537e-06,
+      "loss": 2.1026,
+      "step": 85740
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.4815106215578287e-06,
+      "loss": 2.1083,
+      "step": 85750
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.4793251158318037e-06,
+      "loss": 2.1289,
+      "step": 85760
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.4771396101057788e-06,
+      "loss": 2.1156,
+      "step": 85770
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.4749541043797534e-06,
+      "loss": 2.1032,
+      "step": 85780
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.4727685986537284e-06,
+      "loss": 2.146,
+      "step": 85790
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.470583092927704e-06,
+      "loss": 2.092,
+      "step": 85800
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.468397587201679e-06,
+      "loss": 2.1242,
+      "step": 85810
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.466212081475654e-06,
+      "loss": 2.1231,
+      "step": 85820
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.464026575749629e-06,
+      "loss": 2.0882,
+      "step": 85830
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.461841070023604e-06,
+      "loss": 2.1145,
+      "step": 85840
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.4596555642975786e-06,
+      "loss": 2.1264,
+      "step": 85850
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.4574700585715536e-06,
+      "loss": 2.0844,
+      "step": 85860
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.4552845528455287e-06,
+      "loss": 2.1046,
+      "step": 85870
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.4530990471195037e-06,
+      "loss": 2.1348,
+      "step": 85880
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.4509135413934787e-06,
+      "loss": 2.081,
+      "step": 85890
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.4487280356674533e-06,
+      "loss": 2.0852,
+      "step": 85900
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.4465425299414284e-06,
+      "loss": 2.1197,
+      "step": 85910
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.444357024215404e-06,
+      "loss": 2.1168,
+      "step": 85920
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.442171518489379e-06,
+      "loss": 2.1384,
+      "step": 85930
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.439986012763354e-06,
+      "loss": 2.0864,
+      "step": 85940
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.437800507037329e-06,
+      "loss": 2.0784,
+      "step": 85950
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.435615001311304e-06,
+      "loss": 2.1201,
+      "step": 85960
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.4334294955852786e-06,
+      "loss": 2.1152,
+      "step": 85970
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.4312439898592536e-06,
+      "loss": 2.0937,
+      "step": 85980
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.4290584841332286e-06,
+      "loss": 2.1119,
+      "step": 85990
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.4268729784072037e-06,
+      "loss": 2.1084,
+      "step": 86000
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.4246874726811787e-06,
+      "loss": 2.0817,
+      "step": 86010
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.4225019669551533e-06,
+      "loss": 2.0498,
+      "step": 86020
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.420316461229129e-06,
+      "loss": 2.1205,
+      "step": 86030
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.4181309555031038e-06,
+      "loss": 2.1468,
+      "step": 86040
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.415945449777079e-06,
+      "loss": 2.0991,
+      "step": 86050
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.413759944051054e-06,
+      "loss": 2.1297,
+      "step": 86060
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.411574438325029e-06,
+      "loss": 2.1218,
+      "step": 86070
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.409388932599004e-06,
+      "loss": 2.1299,
+      "step": 86080
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.4072034268729785e-06,
+      "loss": 2.1141,
+      "step": 86090
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.4050179211469536e-06,
+      "loss": 2.0741,
+      "step": 86100
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.4028324154209286e-06,
+      "loss": 2.1377,
+      "step": 86110
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.4006469096949036e-06,
+      "loss": 2.1052,
+      "step": 86120
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.3984614039688786e-06,
+      "loss": 2.1311,
+      "step": 86130
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.3962758982428533e-06,
+      "loss": 2.0885,
+      "step": 86140
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.3940903925168287e-06,
+      "loss": 2.0953,
+      "step": 86150
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.3919048867908037e-06,
+      "loss": 2.1138,
+      "step": 86160
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.3897193810647788e-06,
+      "loss": 2.0895,
+      "step": 86170
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.387533875338754e-06,
+      "loss": 2.0862,
+      "step": 86180
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.385348369612729e-06,
+      "loss": 2.138,
+      "step": 86190
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.383162863886704e-06,
+      "loss": 2.096,
+      "step": 86200
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.3809773581606785e-06,
+      "loss": 2.1486,
+      "step": 86210
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.3787918524346535e-06,
+      "loss": 2.1434,
+      "step": 86220
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.3766063467086285e-06,
+      "loss": 2.1226,
+      "step": 86230
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.3744208409826036e-06,
+      "loss": 2.1082,
+      "step": 86240
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.3722353352565786e-06,
+      "loss": 2.1546,
+      "step": 86250
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.3700498295305532e-06,
+      "loss": 2.1333,
+      "step": 86260
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.3678643238045287e-06,
+      "loss": 2.118,
+      "step": 86270
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.3656788180785037e-06,
+      "loss": 2.1214,
+      "step": 86280
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.3634933123524787e-06,
+      "loss": 2.0967,
+      "step": 86290
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.3613078066264538e-06,
+      "loss": 2.0992,
+      "step": 86300
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.359122300900429e-06,
+      "loss": 2.1758,
+      "step": 86310
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.356936795174404e-06,
+      "loss": 2.1211,
+      "step": 86320
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.3547512894483784e-06,
+      "loss": 2.1594,
+      "step": 86330
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.3525657837223535e-06,
+      "loss": 2.1335,
+      "step": 86340
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.3503802779963285e-06,
+      "loss": 2.1151,
+      "step": 86350
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.3481947722703035e-06,
+      "loss": 2.1009,
+      "step": 86360
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.3460092665442786e-06,
+      "loss": 2.0985,
+      "step": 86370
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.343823760818254e-06,
+      "loss": 2.1154,
+      "step": 86380
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.3416382550922286e-06,
+      "loss": 2.1167,
+      "step": 86390
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.3394527493662037e-06,
+      "loss": 2.1075,
+      "step": 86400
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.3372672436401787e-06,
+      "loss": 2.1214,
+      "step": 86410
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.3350817379141537e-06,
+      "loss": 2.1343,
+      "step": 86420
+    },
+    {
+      "epoch": 17.0,
+      "eval_BLEU": 10.3314,
+      "eval_gen_len": 12.1336,
+      "eval_loss": 3.3399696350097656,
+      "eval_runtime": 153.0971,
+      "eval_samples_per_second": 70.302,
+      "eval_steps_per_second": 0.281,
+      "step": 86428
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.3328962321881288e-06,
+      "loss": 2.1072,
+      "step": 86430
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.3307107264621038e-06,
+      "loss": 2.0531,
+      "step": 86440
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.3285252207360784e-06,
+      "loss": 2.0477,
+      "step": 86450
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.3263397150100534e-06,
+      "loss": 2.075,
+      "step": 86460
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.3241542092840285e-06,
+      "loss": 2.0585,
+      "step": 86470
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.3219687035580035e-06,
+      "loss": 2.0786,
+      "step": 86480
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.3197831978319785e-06,
+      "loss": 2.0603,
+      "step": 86490
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.317597692105954e-06,
+      "loss": 2.0663,
+      "step": 86500
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.3154121863799286e-06,
+      "loss": 2.0412,
+      "step": 86510
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.3132266806539036e-06,
+      "loss": 2.0859,
+      "step": 86520
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.3110411749278786e-06,
+      "loss": 2.0404,
+      "step": 86530
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.3088556692018537e-06,
+      "loss": 2.0379,
+      "step": 86540
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.3066701634758287e-06,
+      "loss": 2.0892,
+      "step": 86550
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.3044846577498037e-06,
+      "loss": 2.0679,
+      "step": 86560
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.3022991520237784e-06,
+      "loss": 2.064,
+      "step": 86570
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.3001136462977534e-06,
+      "loss": 2.0799,
+      "step": 86580
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.2979281405717284e-06,
+      "loss": 2.0391,
+      "step": 86590
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.2957426348457034e-06,
+      "loss": 2.0584,
+      "step": 86600
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.2935571291196785e-06,
+      "loss": 2.0525,
+      "step": 86610
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.291371623393654e-06,
+      "loss": 2.0739,
+      "step": 86620
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.2891861176676285e-06,
+      "loss": 2.049,
+      "step": 86630
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.2870006119416036e-06,
+      "loss": 2.0424,
+      "step": 86640
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.2848151062155786e-06,
+      "loss": 2.0945,
+      "step": 86650
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2826296004895536e-06,
+      "loss": 2.0579,
+      "step": 86660
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2804440947635287e-06,
+      "loss": 2.0503,
+      "step": 86670
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2782585890375037e-06,
+      "loss": 2.0508,
+      "step": 86680
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2760730833114783e-06,
+      "loss": 2.0484,
+      "step": 86690
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2738875775854533e-06,
+      "loss": 2.0911,
+      "step": 86700
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2717020718594284e-06,
+      "loss": 2.0893,
+      "step": 86710
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2695165661334034e-06,
+      "loss": 2.0284,
+      "step": 86720
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.267331060407379e-06,
+      "loss": 2.0428,
+      "step": 86730
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.265145554681354e-06,
+      "loss": 2.0792,
+      "step": 86740
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2629600489553285e-06,
+      "loss": 2.0756,
+      "step": 86750
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.2607745432293035e-06,
+      "loss": 2.0661,
+      "step": 86760
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.2585890375032786e-06,
+      "loss": 2.0433,
+      "step": 86770
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.2564035317772536e-06,
+      "loss": 2.0633,
+      "step": 86780
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.2542180260512286e-06,
+      "loss": 2.0485,
+      "step": 86790
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 2.0745,
+      "step": 86800
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2498470145991783e-06,
+      "loss": 2.0691,
+      "step": 86810
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2476615088731533e-06,
+      "loss": 2.0849,
+      "step": 86820
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2454760031471283e-06,
+      "loss": 2.0788,
+      "step": 86830
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2432904974211034e-06,
+      "loss": 2.0383,
+      "step": 86840
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.241104991695079e-06,
+      "loss": 2.0554,
+      "step": 86850
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.238919485969054e-06,
+      "loss": 2.078,
+      "step": 86860
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2367339802430285e-06,
+      "loss": 2.0506,
+      "step": 86870
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2345484745170035e-06,
+      "loss": 2.0537,
+      "step": 86880
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2323629687909785e-06,
+      "loss": 2.0833,
+      "step": 86890
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2301774630649535e-06,
+      "loss": 2.077,
+      "step": 86900
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2279919573389286e-06,
+      "loss": 2.0712,
+      "step": 86910
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 2.0615,
+      "step": 86920
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.2236209458868782e-06,
+      "loss": 2.0763,
+      "step": 86930
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.2214354401608533e-06,
+      "loss": 2.0364,
+      "step": 86940
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.2192499344348283e-06,
+      "loss": 2.0446,
+      "step": 86950
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.2170644287088037e-06,
+      "loss": 2.0858,
+      "step": 86960
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2148789229827788e-06,
+      "loss": 2.0331,
+      "step": 86970
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.212693417256754e-06,
+      "loss": 2.0502,
+      "step": 86980
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2105079115307284e-06,
+      "loss": 2.0532,
+      "step": 86990
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2083224058047034e-06,
+      "loss": 2.085,
+      "step": 87000
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2061369000786785e-06,
+      "loss": 2.1007,
+      "step": 87010
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.2039513943526535e-06,
+      "loss": 2.0978,
+      "step": 87020
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.2017658886266285e-06,
+      "loss": 2.0448,
+      "step": 87030
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.199580382900603e-06,
+      "loss": 2.0088,
+      "step": 87040
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.197394877174578e-06,
+      "loss": 2.0595,
+      "step": 87050
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.195209371448553e-06,
+      "loss": 2.0563,
+      "step": 87060
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.1930238657225282e-06,
+      "loss": 2.0529,
+      "step": 87070
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.1908383599965037e-06,
+      "loss": 2.0791,
+      "step": 87080
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.1886528542704787e-06,
+      "loss": 2.0653,
+      "step": 87090
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.1864673485444538e-06,
+      "loss": 2.0598,
+      "step": 87100
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.1842818428184284e-06,
+      "loss": 2.052,
+      "step": 87110
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.1820963370924034e-06,
+      "loss": 2.0648,
+      "step": 87120
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.1799108313663784e-06,
+      "loss": 2.0472,
+      "step": 87130
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.1777253256403535e-06,
+      "loss": 2.0392,
+      "step": 87140
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.1755398199143285e-06,
+      "loss": 2.04,
+      "step": 87150
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.173354314188303e-06,
+      "loss": 2.0699,
+      "step": 87160
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.171168808462278e-06,
+      "loss": 2.0558,
+      "step": 87170
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.168983302736253e-06,
+      "loss": 2.0831,
+      "step": 87180
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.166797797010228e-06,
+      "loss": 2.06,
+      "step": 87190
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.1646122912842037e-06,
+      "loss": 2.0293,
+      "step": 87200
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.1624267855581787e-06,
+      "loss": 2.0328,
+      "step": 87210
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.1602412798321537e-06,
+      "loss": 2.0596,
+      "step": 87220
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.1580557741061283e-06,
+      "loss": 2.0727,
+      "step": 87230
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.1558702683801034e-06,
+      "loss": 2.0674,
+      "step": 87240
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.1536847626540784e-06,
+      "loss": 2.0362,
+      "step": 87250
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.1514992569280534e-06,
+      "loss": 2.0951,
+      "step": 87260
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.1493137512020285e-06,
+      "loss": 2.0829,
+      "step": 87270
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.147128245476003e-06,
+      "loss": 2.0716,
+      "step": 87280
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.144942739749978e-06,
+      "loss": 2.0936,
+      "step": 87290
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.142757234023953e-06,
+      "loss": 2.0418,
+      "step": 87300
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.1405717282979286e-06,
+      "loss": 2.0902,
+      "step": 87310
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.1383862225719036e-06,
+      "loss": 2.0592,
+      "step": 87320
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.1362007168458786e-06,
+      "loss": 2.0517,
+      "step": 87330
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.1340152111198537e-06,
+      "loss": 2.02,
+      "step": 87340
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.1318297053938283e-06,
+      "loss": 2.0509,
+      "step": 87350
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.1296441996678033e-06,
+      "loss": 2.0904,
+      "step": 87360
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.1274586939417783e-06,
+      "loss": 2.0749,
+      "step": 87370
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.1252731882157534e-06,
+      "loss": 2.0731,
+      "step": 87380
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.1230876824897284e-06,
+      "loss": 2.0423,
+      "step": 87390
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.120902176763703e-06,
+      "loss": 2.0841,
+      "step": 87400
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.118716671037678e-06,
+      "loss": 2.0682,
+      "step": 87410
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.116531165311653e-06,
+      "loss": 2.0977,
+      "step": 87420
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.1143456595856285e-06,
+      "loss": 2.0904,
+      "step": 87430
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.1121601538596036e-06,
+      "loss": 2.0844,
+      "step": 87440
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.1099746481335786e-06,
+      "loss": 2.0386,
+      "step": 87450
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.1077891424075536e-06,
+      "loss": 2.0652,
+      "step": 87460
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.1056036366815282e-06,
+      "loss": 2.0188,
+      "step": 87470
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.1034181309555033e-06,
+      "loss": 2.1138,
+      "step": 87480
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.1012326252294783e-06,
+      "loss": 2.0771,
+      "step": 87490
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.0990471195034533e-06,
+      "loss": 2.0563,
+      "step": 87500
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.0968616137774284e-06,
+      "loss": 2.056,
+      "step": 87510
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.094676108051403e-06,
+      "loss": 2.0347,
+      "step": 87520
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.092490602325378e-06,
+      "loss": 2.0776,
+      "step": 87530
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.090305096599353e-06,
+      "loss": 2.0858,
+      "step": 87540
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.0881195908733285e-06,
+      "loss": 2.1008,
+      "step": 87550
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.0859340851473035e-06,
+      "loss": 2.0574,
+      "step": 87560
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.0837485794212786e-06,
+      "loss": 2.0709,
+      "step": 87570
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.0815630736952536e-06,
+      "loss": 2.0854,
+      "step": 87580
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.079377567969228e-06,
+      "loss": 2.074,
+      "step": 87590
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.0771920622432032e-06,
+      "loss": 2.0465,
+      "step": 87600
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.0750065565171783e-06,
+      "loss": 2.0535,
+      "step": 87610
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.0728210507911533e-06,
+      "loss": 2.0063,
+      "step": 87620
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.0706355450651283e-06,
+      "loss": 2.059,
+      "step": 87630
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.068450039339103e-06,
+      "loss": 2.0926,
+      "step": 87640
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.066264533613078e-06,
+      "loss": 2.0926,
+      "step": 87650
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.0640790278870534e-06,
+      "loss": 2.0959,
+      "step": 87660
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.0618935221610284e-06,
+      "loss": 2.0743,
+      "step": 87670
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.0597080164350035e-06,
+      "loss": 2.0898,
+      "step": 87680
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.0575225107089785e-06,
+      "loss": 2.062,
+      "step": 87690
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.0553370049829535e-06,
+      "loss": 2.0759,
+      "step": 87700
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.053151499256928e-06,
+      "loss": 2.0696,
+      "step": 87710
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.050965993530903e-06,
+      "loss": 2.0695,
+      "step": 87720
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.0487804878048782e-06,
+      "loss": 2.0624,
+      "step": 87730
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.0465949820788532e-06,
+      "loss": 2.0501,
+      "step": 87740
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.0444094763528283e-06,
+      "loss": 2.0955,
+      "step": 87750
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.042223970626803e-06,
+      "loss": 2.0538,
+      "step": 87760
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.040038464900778e-06,
+      "loss": 2.07,
+      "step": 87770
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.0378529591747534e-06,
+      "loss": 2.0895,
+      "step": 87780
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.0356674534487284e-06,
+      "loss": 2.058,
+      "step": 87790
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.0334819477227034e-06,
+      "loss": 2.0746,
+      "step": 87800
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.0312964419966785e-06,
+      "loss": 2.0617,
+      "step": 87810
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.0291109362706535e-06,
+      "loss": 2.062,
+      "step": 87820
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.026925430544628e-06,
+      "loss": 2.0834,
+      "step": 87830
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.024739924818603e-06,
+      "loss": 2.0687,
+      "step": 87840
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.022554419092578e-06,
+      "loss": 2.0368,
+      "step": 87850
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.020368913366553e-06,
+      "loss": 2.0476,
+      "step": 87860
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.0181834076405282e-06,
+      "loss": 2.0537,
+      "step": 87870
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.015997901914503e-06,
+      "loss": 2.0458,
+      "step": 87880
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.013812396188478e-06,
+      "loss": 2.08,
+      "step": 87890
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.0116268904624533e-06,
+      "loss": 2.0393,
+      "step": 87900
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.0094413847364284e-06,
+      "loss": 2.0533,
+      "step": 87910
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.0072558790104034e-06,
+      "loss": 2.0965,
+      "step": 87920
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.0050703732843784e-06,
+      "loss": 2.0744,
+      "step": 87930
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.0028848675583535e-06,
+      "loss": 2.0166,
+      "step": 87940
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.000699361832328e-06,
+      "loss": 2.1,
+      "step": 87950
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 2.998513856106303e-06,
+      "loss": 2.0622,
+      "step": 87960
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 2.996328350380278e-06,
+      "loss": 2.0742,
+      "step": 87970
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 2.994142844654253e-06,
+      "loss": 2.0583,
+      "step": 87980
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 2.991957338928228e-06,
+      "loss": 2.093,
+      "step": 87990
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 2.989771833202203e-06,
+      "loss": 2.088,
+      "step": 88000
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 2.9875863274761787e-06,
+      "loss": 2.0861,
+      "step": 88010
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 2.9854008217501533e-06,
+      "loss": 2.0997,
+      "step": 88020
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.9832153160241283e-06,
+      "loss": 2.0634,
+      "step": 88030
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.9810298102981034e-06,
+      "loss": 2.0481,
+      "step": 88040
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.9788443045720784e-06,
+      "loss": 2.0602,
+      "step": 88050
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.9766587988460534e-06,
+      "loss": 2.1039,
+      "step": 88060
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.974473293120028e-06,
+      "loss": 2.0483,
+      "step": 88070
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.972287787394003e-06,
+      "loss": 2.0772,
+      "step": 88080
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 2.970102281667978e-06,
+      "loss": 2.0541,
+      "step": 88090
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 2.967916775941953e-06,
+      "loss": 2.0515,
+      "step": 88100
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 2.965731270215928e-06,
+      "loss": 2.1084,
+      "step": 88110
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 2.9635457644899028e-06,
+      "loss": 2.0696,
+      "step": 88120
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 2.9613602587638786e-06,
+      "loss": 2.0602,
+      "step": 88130
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 2.9591747530378532e-06,
+      "loss": 2.0867,
+      "step": 88140
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 2.9569892473118283e-06,
+      "loss": 2.0584,
+      "step": 88150
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 2.9548037415858033e-06,
+      "loss": 2.0942,
+      "step": 88160
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 2.9526182358597783e-06,
+      "loss": 2.0413,
+      "step": 88170
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 2.9504327301337534e-06,
+      "loss": 2.0953,
+      "step": 88180
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 2.948247224407728e-06,
+      "loss": 2.0516,
+      "step": 88190
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 2.946061718681703e-06,
+      "loss": 2.0283,
+      "step": 88200
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 2.943876212955678e-06,
+      "loss": 2.0708,
+      "step": 88210
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 2.941690707229653e-06,
+      "loss": 2.0753,
+      "step": 88220
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 2.939505201503628e-06,
+      "loss": 2.0562,
+      "step": 88230
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 2.9373196957776036e-06,
+      "loss": 2.0705,
+      "step": 88240
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 2.9351341900515786e-06,
+      "loss": 2.0471,
+      "step": 88250
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 2.932948684325553e-06,
+      "loss": 2.0468,
+      "step": 88260
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 2.9307631785995282e-06,
+      "loss": 2.0692,
+      "step": 88270
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 2.9285776728735033e-06,
+      "loss": 2.0319,
+      "step": 88280
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.9263921671474783e-06,
+      "loss": 2.0443,
+      "step": 88290
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.9242066614214533e-06,
+      "loss": 2.0623,
+      "step": 88300
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.922021155695428e-06,
+      "loss": 2.0327,
+      "step": 88310
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.919835649969403e-06,
+      "loss": 2.0604,
+      "step": 88320
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.917650144243378e-06,
+      "loss": 2.0606,
+      "step": 88330
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 2.915464638517353e-06,
+      "loss": 2.0466,
+      "step": 88340
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 2.913279132791328e-06,
+      "loss": 2.1028,
+      "step": 88350
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 2.9110936270653035e-06,
+      "loss": 2.0818,
+      "step": 88360
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 2.9089081213392786e-06,
+      "loss": 2.0691,
+      "step": 88370
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 2.906722615613253e-06,
+      "loss": 2.0787,
+      "step": 88380
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.904537109887228e-06,
+      "loss": 2.0868,
+      "step": 88390
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.9023516041612032e-06,
+      "loss": 2.0684,
+      "step": 88400
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.9001660984351783e-06,
+      "loss": 2.0224,
+      "step": 88410
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.8979805927091533e-06,
+      "loss": 2.0396,
+      "step": 88420
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.895795086983128e-06,
+      "loss": 2.0675,
+      "step": 88430
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 2.893609581257103e-06,
+      "loss": 2.0819,
+      "step": 88440
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 2.891424075531078e-06,
+      "loss": 2.0585,
+      "step": 88450
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 2.889238569805053e-06,
+      "loss": 2.0673,
+      "step": 88460
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 2.887053064079028e-06,
+      "loss": 2.072,
+      "step": 88470
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 2.8848675583530035e-06,
+      "loss": 2.0314,
+      "step": 88480
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 2.8826820526269785e-06,
+      "loss": 2.0612,
+      "step": 88490
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 2.880496546900953e-06,
+      "loss": 2.0851,
+      "step": 88500
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 2.878311041174928e-06,
+      "loss": 2.0631,
+      "step": 88510
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 2.876125535448903e-06,
+      "loss": 2.1165,
+      "step": 88520
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 2.873940029722878e-06,
+      "loss": 2.0779,
+      "step": 88530
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 2.8717545239968532e-06,
+      "loss": 2.0494,
+      "step": 88540
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 2.869569018270828e-06,
+      "loss": 2.0665,
+      "step": 88550
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 2.867383512544803e-06,
+      "loss": 2.0653,
+      "step": 88560
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 2.865198006818778e-06,
+      "loss": 2.0595,
+      "step": 88570
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 2.863012501092753e-06,
+      "loss": 2.0837,
+      "step": 88580
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 2.8608269953667284e-06,
+      "loss": 2.0771,
+      "step": 88590
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 2.8586414896407034e-06,
+      "loss": 2.0711,
+      "step": 88600
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 2.8564559839146785e-06,
+      "loss": 2.1238,
+      "step": 88610
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 2.854270478188653e-06,
+      "loss": 2.0988,
+      "step": 88620
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 2.852084972462628e-06,
+      "loss": 2.0872,
+      "step": 88630
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.849899466736603e-06,
+      "loss": 2.0543,
+      "step": 88640
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.847713961010578e-06,
+      "loss": 2.064,
+      "step": 88650
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.845528455284553e-06,
+      "loss": 2.0772,
+      "step": 88660
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.843342949558528e-06,
+      "loss": 2.0752,
+      "step": 88670
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.841157443832503e-06,
+      "loss": 2.0686,
+      "step": 88680
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.838971938106478e-06,
+      "loss": 2.0839,
+      "step": 88690
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 2.836786432380453e-06,
+      "loss": 2.0428,
+      "step": 88700
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 2.8346009266544284e-06,
+      "loss": 2.0549,
+      "step": 88710
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 2.8324154209284034e-06,
+      "loss": 2.0719,
+      "step": 88720
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 2.8302299152023784e-06,
+      "loss": 2.119,
+      "step": 88730
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 2.828044409476353e-06,
+      "loss": 2.0651,
+      "step": 88740
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 2.825858903750328e-06,
+      "loss": 2.0792,
+      "step": 88750
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 2.823673398024303e-06,
+      "loss": 2.0612,
+      "step": 88760
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 2.821487892298278e-06,
+      "loss": 2.0649,
+      "step": 88770
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 2.819302386572253e-06,
+      "loss": 2.0515,
+      "step": 88780
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 2.8171168808462278e-06,
+      "loss": 2.0437,
+      "step": 88790
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 2.814931375120203e-06,
+      "loss": 2.0726,
+      "step": 88800
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 2.812745869394178e-06,
+      "loss": 2.0787,
+      "step": 88810
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 2.810560363668153e-06,
+      "loss": 2.1144,
+      "step": 88820
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 2.8083748579421283e-06,
+      "loss": 2.0874,
+      "step": 88830
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 2.8061893522161033e-06,
+      "loss": 2.07,
+      "step": 88840
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 2.8040038464900784e-06,
+      "loss": 2.0823,
+      "step": 88850
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 2.801818340764053e-06,
+      "loss": 2.115,
+      "step": 88860
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 2.799632835038028e-06,
+      "loss": 2.0549,
+      "step": 88870
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 2.797447329312003e-06,
+      "loss": 2.1138,
+      "step": 88880
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 2.795261823585978e-06,
+      "loss": 2.0801,
+      "step": 88890
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 2.793076317859953e-06,
+      "loss": 2.1019,
+      "step": 88900
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 2.7908908121339277e-06,
+      "loss": 2.0708,
+      "step": 88910
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 2.7887053064079028e-06,
+      "loss": 2.0355,
+      "step": 88920
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 2.7865198006818778e-06,
+      "loss": 2.1087,
+      "step": 88930
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 2.7843342949558532e-06,
+      "loss": 2.0729,
+      "step": 88940
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 2.7821487892298283e-06,
+      "loss": 2.0666,
+      "step": 88950
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 2.7799632835038033e-06,
+      "loss": 2.0289,
+      "step": 88960
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 2.0921,
+      "step": 88970
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 2.775592272051753e-06,
+      "loss": 2.0833,
+      "step": 88980
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 2.773406766325728e-06,
+      "loss": 2.0888,
+      "step": 88990
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 2.771221260599703e-06,
+      "loss": 2.0425,
+      "step": 89000
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 2.769035754873678e-06,
+      "loss": 2.0471,
+      "step": 89010
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 2.766850249147653e-06,
+      "loss": 2.0814,
+      "step": 89020
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 2.7646647434216277e-06,
+      "loss": 2.0763,
+      "step": 89030
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 2.7624792376956027e-06,
+      "loss": 2.0806,
+      "step": 89040
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 2.7602937319695777e-06,
+      "loss": 2.106,
+      "step": 89050
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 2.758108226243553e-06,
+      "loss": 2.0818,
+      "step": 89060
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 2.7559227205175282e-06,
+      "loss": 2.0725,
+      "step": 89070
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 2.7537372147915033e-06,
+      "loss": 2.0867,
+      "step": 89080
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 2.7515517090654783e-06,
+      "loss": 2.0959,
+      "step": 89090
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 2.749366203339453e-06,
+      "loss": 2.0567,
+      "step": 89100
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 2.747180697613428e-06,
+      "loss": 2.0721,
+      "step": 89110
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 2.744995191887403e-06,
+      "loss": 2.0749,
+      "step": 89120
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 2.742809686161378e-06,
+      "loss": 2.0657,
+      "step": 89130
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 2.740624180435353e-06,
+      "loss": 2.0787,
+      "step": 89140
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 2.7384386747093276e-06,
+      "loss": 2.0608,
+      "step": 89150
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 2.7362531689833027e-06,
+      "loss": 2.0475,
+      "step": 89160
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 2.7340676632572777e-06,
+      "loss": 2.07,
+      "step": 89170
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 2.731882157531253e-06,
+      "loss": 2.0598,
+      "step": 89180
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 2.729696651805228e-06,
+      "loss": 2.1182,
+      "step": 89190
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 2.7275111460792032e-06,
+      "loss": 2.0435,
+      "step": 89200
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 2.7253256403531782e-06,
+      "loss": 2.0855,
+      "step": 89210
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 2.723140134627153e-06,
+      "loss": 2.0032,
+      "step": 89220
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 2.720954628901128e-06,
+      "loss": 2.0714,
+      "step": 89230
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 2.718769123175103e-06,
+      "loss": 2.0876,
+      "step": 89240
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 2.716583617449078e-06,
+      "loss": 2.0768,
+      "step": 89250
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 2.714398111723053e-06,
+      "loss": 2.0564,
+      "step": 89260
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 2.7122126059970276e-06,
+      "loss": 2.0621,
+      "step": 89270
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 2.7100271002710026e-06,
+      "loss": 2.0627,
+      "step": 89280
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 2.707841594544978e-06,
+      "loss": 2.0592,
+      "step": 89290
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 2.705656088818953e-06,
+      "loss": 2.0668,
+      "step": 89300
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 2.703470583092928e-06,
+      "loss": 2.0348,
+      "step": 89310
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 2.701285077366903e-06,
+      "loss": 2.0651,
+      "step": 89320
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 2.6990995716408778e-06,
+      "loss": 2.1095,
+      "step": 89330
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 2.696914065914853e-06,
+      "loss": 2.0868,
+      "step": 89340
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 2.694728560188828e-06,
+      "loss": 2.1036,
+      "step": 89350
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 2.692543054462803e-06,
+      "loss": 2.0882,
+      "step": 89360
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 2.690357548736778e-06,
+      "loss": 2.0803,
+      "step": 89370
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 2.688172043010753e-06,
+      "loss": 2.0149,
+      "step": 89380
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 2.6859865372847275e-06,
+      "loss": 2.1028,
+      "step": 89390
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 2.6838010315587026e-06,
+      "loss": 2.0463,
+      "step": 89400
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 2.681615525832678e-06,
+      "loss": 2.0768,
+      "step": 89410
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 2.679430020106653e-06,
+      "loss": 2.0916,
+      "step": 89420
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 2.677244514380628e-06,
+      "loss": 2.0721,
+      "step": 89430
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 2.675059008654603e-06,
+      "loss": 2.0849,
+      "step": 89440
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 2.6728735029285777e-06,
+      "loss": 2.0602,
+      "step": 89450
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 2.6706879972025528e-06,
+      "loss": 2.0891,
+      "step": 89460
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 2.668502491476528e-06,
+      "loss": 2.0532,
+      "step": 89470
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 2.666316985750503e-06,
+      "loss": 2.0871,
+      "step": 89480
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 2.664131480024478e-06,
+      "loss": 2.0804,
+      "step": 89490
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 2.661945974298453e-06,
+      "loss": 2.0761,
+      "step": 89500
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 2.6597604685724275e-06,
+      "loss": 2.074,
+      "step": 89510
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 2.657574962846403e-06,
+      "loss": 2.051,
+      "step": 89520
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 2.655389457120378e-06,
+      "loss": 2.0793,
+      "step": 89530
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 2.653203951394353e-06,
+      "loss": 2.0795,
+      "step": 89540
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 2.651018445668328e-06,
+      "loss": 2.0625,
+      "step": 89550
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 2.648832939942303e-06,
+      "loss": 2.0787,
+      "step": 89560
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 2.6466474342162777e-06,
+      "loss": 2.1181,
+      "step": 89570
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 2.6444619284902527e-06,
+      "loss": 2.0617,
+      "step": 89580
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 2.6422764227642278e-06,
+      "loss": 2.0732,
+      "step": 89590
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 2.640090917038203e-06,
+      "loss": 2.0421,
+      "step": 89600
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 2.637905411312178e-06,
+      "loss": 2.0651,
+      "step": 89610
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 2.635719905586153e-06,
+      "loss": 2.0757,
+      "step": 89620
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 2.6335343998601275e-06,
+      "loss": 2.0988,
+      "step": 89630
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 2.631348894134103e-06,
+      "loss": 2.0749,
+      "step": 89640
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 2.629163388408078e-06,
+      "loss": 2.1053,
+      "step": 89650
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 2.626977882682053e-06,
+      "loss": 2.0607,
+      "step": 89660
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 2.624792376956028e-06,
+      "loss": 2.0764,
+      "step": 89670
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 2.622606871230003e-06,
+      "loss": 2.0674,
+      "step": 89680
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 2.6204213655039777e-06,
+      "loss": 2.0452,
+      "step": 89690
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 2.6182358597779527e-06,
+      "loss": 2.0646,
+      "step": 89700
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 2.6160503540519277e-06,
+      "loss": 2.0583,
+      "step": 89710
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 2.6138648483259027e-06,
+      "loss": 2.0839,
+      "step": 89720
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 2.6116793425998778e-06,
+      "loss": 2.0923,
+      "step": 89730
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 2.609493836873853e-06,
+      "loss": 2.1329,
+      "step": 89740
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 2.6073083311478274e-06,
+      "loss": 2.0612,
+      "step": 89750
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 2.605122825421803e-06,
+      "loss": 2.0779,
+      "step": 89760
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 2.602937319695778e-06,
+      "loss": 2.0612,
+      "step": 89770
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 2.600751813969753e-06,
+      "loss": 2.0477,
+      "step": 89780
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 2.598566308243728e-06,
+      "loss": 2.0745,
+      "step": 89790
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 2.596380802517703e-06,
+      "loss": 2.1039,
+      "step": 89800
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 2.5941952967916776e-06,
+      "loss": 2.0503,
+      "step": 89810
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 2.5920097910656526e-06,
+      "loss": 2.1511,
+      "step": 89820
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 2.5898242853396277e-06,
+      "loss": 2.0361,
+      "step": 89830
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 2.5876387796136027e-06,
+      "loss": 2.0663,
+      "step": 89840
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 2.5854532738875777e-06,
+      "loss": 2.1112,
+      "step": 89850
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 2.5832677681615528e-06,
+      "loss": 2.0745,
+      "step": 89860
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 2.5810822624355282e-06,
+      "loss": 2.0636,
+      "step": 89870
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 2.578896756709503e-06,
+      "loss": 2.0722,
+      "step": 89880
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 2.576711250983478e-06,
+      "loss": 2.063,
+      "step": 89890
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 2.574525745257453e-06,
+      "loss": 2.0415,
+      "step": 89900
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 2.572340239531428e-06,
+      "loss": 2.0791,
+      "step": 89910
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 2.570154733805403e-06,
+      "loss": 2.053,
+      "step": 89920
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 2.5679692280793776e-06,
+      "loss": 2.1067,
+      "step": 89930
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 2.5657837223533526e-06,
+      "loss": 2.0833,
+      "step": 89940
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 2.5635982166273276e-06,
+      "loss": 2.0624,
+      "step": 89950
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 2.5614127109013027e-06,
+      "loss": 2.1011,
+      "step": 89960
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 2.5592272051752777e-06,
+      "loss": 2.0829,
+      "step": 89970
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 2.5570416994492523e-06,
+      "loss": 2.09,
+      "step": 89980
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 2.554856193723228e-06,
+      "loss": 2.1204,
+      "step": 89990
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 2.5526706879972028e-06,
+      "loss": 2.0776,
+      "step": 90000
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 2.550485182271178e-06,
+      "loss": 2.0872,
+      "step": 90010
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 2.548299676545153e-06,
+      "loss": 2.1256,
+      "step": 90020
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 2.546114170819128e-06,
+      "loss": 2.0983,
+      "step": 90030
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 2.543928665093103e-06,
+      "loss": 2.0869,
+      "step": 90040
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 2.5417431593670775e-06,
+      "loss": 2.0534,
+      "step": 90050
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 2.5395576536410526e-06,
+      "loss": 2.0589,
+      "step": 90060
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 2.5373721479150276e-06,
+      "loss": 2.07,
+      "step": 90070
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 2.5351866421890026e-06,
+      "loss": 2.0551,
+      "step": 90080
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 2.5330011364629776e-06,
+      "loss": 2.0632,
+      "step": 90090
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 2.5308156307369523e-06,
+      "loss": 2.0856,
+      "step": 90100
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 2.528630125010928e-06,
+      "loss": 2.0825,
+      "step": 90110
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 2.5264446192849027e-06,
+      "loss": 2.0954,
+      "step": 90120
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 2.5242591135588778e-06,
+      "loss": 2.0943,
+      "step": 90130
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 2.522073607832853e-06,
+      "loss": 2.0684,
+      "step": 90140
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 2.519888102106828e-06,
+      "loss": 2.1011,
+      "step": 90150
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 2.517702596380803e-06,
+      "loss": 2.1,
+      "step": 90160
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 2.5155170906547775e-06,
+      "loss": 2.0621,
+      "step": 90170
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 2.5133315849287525e-06,
+      "loss": 2.0808,
+      "step": 90180
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 2.5111460792027275e-06,
+      "loss": 2.0685,
+      "step": 90190
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 2.5089605734767026e-06,
+      "loss": 2.0975,
+      "step": 90200
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 2.5067750677506776e-06,
+      "loss": 2.1214,
+      "step": 90210
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 2.504589562024653e-06,
+      "loss": 2.0444,
+      "step": 90220
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 2.502404056298628e-06,
+      "loss": 2.0681,
+      "step": 90230
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 2.5002185505726027e-06,
+      "loss": 2.0925,
+      "step": 90240
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 2.4980330448465777e-06,
+      "loss": 2.0748,
+      "step": 90250
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 2.4958475391205528e-06,
+      "loss": 2.0671,
+      "step": 90260
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 2.493662033394528e-06,
+      "loss": 2.0779,
+      "step": 90270
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 2.491476527668503e-06,
+      "loss": 2.0967,
+      "step": 90280
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 2.4892910219424774e-06,
+      "loss": 2.0902,
+      "step": 90290
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 2.4871055162164525e-06,
+      "loss": 2.0755,
+      "step": 90300
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 2.484920010490428e-06,
+      "loss": 2.112,
+      "step": 90310
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 2.482734504764403e-06,
+      "loss": 2.0727,
+      "step": 90320
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 2.4805489990383776e-06,
+      "loss": 2.0789,
+      "step": 90330
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 2.4783634933123526e-06,
+      "loss": 2.0831,
+      "step": 90340
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 2.4761779875863276e-06,
+      "loss": 2.0973,
+      "step": 90350
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 2.4739924818603027e-06,
+      "loss": 2.0737,
+      "step": 90360
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 2.4718069761342777e-06,
+      "loss": 2.0598,
+      "step": 90370
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 2.4696214704082527e-06,
+      "loss": 2.07,
+      "step": 90380
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 2.4674359646822278e-06,
+      "loss": 2.1152,
+      "step": 90390
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 2.4652504589562028e-06,
+      "loss": 2.0588,
+      "step": 90400
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 2.4630649532301774e-06,
+      "loss": 2.0721,
+      "step": 90410
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 2.4608794475041524e-06,
+      "loss": 2.0525,
+      "step": 90420
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 2.458693941778128e-06,
+      "loss": 2.094,
+      "step": 90430
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 2.456508436052103e-06,
+      "loss": 2.0897,
+      "step": 90440
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 2.4543229303260775e-06,
+      "loss": 2.0903,
+      "step": 90450
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 2.4521374246000526e-06,
+      "loss": 2.0956,
+      "step": 90460
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 2.4499519188740276e-06,
+      "loss": 2.062,
+      "step": 90470
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 2.4477664131480026e-06,
+      "loss": 2.1055,
+      "step": 90480
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 2.4455809074219776e-06,
+      "loss": 2.0473,
+      "step": 90490
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 2.4433954016959527e-06,
+      "loss": 2.064,
+      "step": 90500
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 2.4412098959699277e-06,
+      "loss": 2.0637,
+      "step": 90510
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 2.0996,
+      "step": 90520
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 2.4368388845178774e-06,
+      "loss": 2.0807,
+      "step": 90530
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 2.434653378791853e-06,
+      "loss": 2.0889,
+      "step": 90540
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 2.432467873065828e-06,
+      "loss": 2.0811,
+      "step": 90550
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 2.430282367339803e-06,
+      "loss": 2.0477,
+      "step": 90560
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 2.4280968616137775e-06,
+      "loss": 2.0859,
+      "step": 90570
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 2.4259113558877525e-06,
+      "loss": 2.0582,
+      "step": 90580
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 2.4237258501617275e-06,
+      "loss": 2.0939,
+      "step": 90590
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 2.4215403444357026e-06,
+      "loss": 2.0854,
+      "step": 90600
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 2.1055,
+      "step": 90610
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 2.4171693329836526e-06,
+      "loss": 2.0899,
+      "step": 90620
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 2.4149838272576277e-06,
+      "loss": 2.0653,
+      "step": 90630
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 2.4127983215316027e-06,
+      "loss": 2.1071,
+      "step": 90640
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 2.4106128158055773e-06,
+      "loss": 2.0521,
+      "step": 90650
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 2.4084273100795528e-06,
+      "loss": 2.0937,
+      "step": 90660
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 2.406241804353528e-06,
+      "loss": 2.097,
+      "step": 90670
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 2.404056298627503e-06,
+      "loss": 2.0859,
+      "step": 90680
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 2.4018707929014774e-06,
+      "loss": 2.0679,
+      "step": 90690
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 2.3996852871754525e-06,
+      "loss": 2.118,
+      "step": 90700
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 2.3974997814494275e-06,
+      "loss": 2.0955,
+      "step": 90710
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 2.3953142757234025e-06,
+      "loss": 2.0938,
+      "step": 90720
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 2.3931287699973776e-06,
+      "loss": 2.1075,
+      "step": 90730
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 2.3909432642713526e-06,
+      "loss": 2.0787,
+      "step": 90740
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 2.3887577585453276e-06,
+      "loss": 2.0699,
+      "step": 90750
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 2.3865722528193027e-06,
+      "loss": 2.0726,
+      "step": 90760
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 2.3843867470932773e-06,
+      "loss": 2.0805,
+      "step": 90770
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 2.3822012413672527e-06,
+      "loss": 2.1235,
+      "step": 90780
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 2.3800157356412278e-06,
+      "loss": 2.0846,
+      "step": 90790
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 2.3778302299152024e-06,
+      "loss": 2.0822,
+      "step": 90800
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 2.3756447241891774e-06,
+      "loss": 2.0483,
+      "step": 90810
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 2.3734592184631524e-06,
+      "loss": 2.0449,
+      "step": 90820
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 2.371273712737128e-06,
+      "loss": 2.0769,
+      "step": 90830
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 2.3690882070111025e-06,
+      "loss": 2.0572,
+      "step": 90840
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 2.3669027012850775e-06,
+      "loss": 2.0297,
+      "step": 90850
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 2.3647171955590525e-06,
+      "loss": 2.0761,
+      "step": 90860
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 2.3625316898330276e-06,
+      "loss": 2.0431,
+      "step": 90870
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 2.3603461841070026e-06,
+      "loss": 2.0578,
+      "step": 90880
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 2.3581606783809776e-06,
+      "loss": 2.0385,
+      "step": 90890
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 2.3559751726549527e-06,
+      "loss": 2.1019,
+      "step": 90900
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 2.3537896669289277e-06,
+      "loss": 2.0684,
+      "step": 90910
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 2.3516041612029023e-06,
+      "loss": 2.0976,
+      "step": 90920
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 2.3494186554768773e-06,
+      "loss": 2.0473,
+      "step": 90930
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 2.3472331497508524e-06,
+      "loss": 2.0575,
+      "step": 90940
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 2.345047644024828e-06,
+      "loss": 2.0493,
+      "step": 90950
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 2.3428621382988024e-06,
+      "loss": 2.1107,
+      "step": 90960
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 2.3406766325727775e-06,
+      "loss": 2.053,
+      "step": 90970
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 2.3384911268467525e-06,
+      "loss": 2.0793,
+      "step": 90980
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 2.3363056211207275e-06,
+      "loss": 2.0636,
+      "step": 90990
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 2.3341201153947026e-06,
+      "loss": 2.089,
+      "step": 91000
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 2.3319346096686776e-06,
+      "loss": 2.079,
+      "step": 91010
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 2.3297491039426526e-06,
+      "loss": 2.1021,
+      "step": 91020
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 2.3275635982166277e-06,
+      "loss": 2.0777,
+      "step": 91030
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 2.3253780924906023e-06,
+      "loss": 2.1171,
+      "step": 91040
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 2.3231925867645773e-06,
+      "loss": 2.0726,
+      "step": 91050
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 2.3210070810385523e-06,
+      "loss": 2.0509,
+      "step": 91060
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 2.318821575312528e-06,
+      "loss": 2.063,
+      "step": 91070
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 2.3166360695865024e-06,
+      "loss": 2.064,
+      "step": 91080
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 2.3144505638604774e-06,
+      "loss": 2.0809,
+      "step": 91090
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 2.3122650581344525e-06,
+      "loss": 2.0817,
+      "step": 91100
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 2.3100795524084275e-06,
+      "loss": 2.0452,
+      "step": 91110
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 2.3078940466824025e-06,
+      "loss": 2.0618,
+      "step": 91120
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 2.3057085409563776e-06,
+      "loss": 2.0629,
+      "step": 91130
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 2.3035230352303526e-06,
+      "loss": 2.1089,
+      "step": 91140
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 2.3013375295043276e-06,
+      "loss": 2.0768,
+      "step": 91150
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 2.2991520237783022e-06,
+      "loss": 2.0893,
+      "step": 91160
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 2.2969665180522773e-06,
+      "loss": 2.0565,
+      "step": 91170
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 2.2947810123262527e-06,
+      "loss": 2.0759,
+      "step": 91180
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 2.2925955066002277e-06,
+      "loss": 2.0768,
+      "step": 91190
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 2.2904100008742024e-06,
+      "loss": 2.0965,
+      "step": 91200
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 2.2882244951481774e-06,
+      "loss": 2.0992,
+      "step": 91210
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 2.2860389894221524e-06,
+      "loss": 2.1085,
+      "step": 91220
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 2.2838534836961275e-06,
+      "loss": 2.0494,
+      "step": 91230
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 2.2816679779701025e-06,
+      "loss": 2.0803,
+      "step": 91240
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 2.2794824722440775e-06,
+      "loss": 2.1278,
+      "step": 91250
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 2.2772969665180525e-06,
+      "loss": 2.1157,
+      "step": 91260
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 2.2751114607920276e-06,
+      "loss": 2.0706,
+      "step": 91270
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 2.272925955066002e-06,
+      "loss": 2.0886,
+      "step": 91280
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 2.2707404493399772e-06,
+      "loss": 2.0852,
+      "step": 91290
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 2.2685549436139527e-06,
+      "loss": 2.0434,
+      "step": 91300
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 2.2663694378879277e-06,
+      "loss": 2.0911,
+      "step": 91310
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 2.2641839321619023e-06,
+      "loss": 2.0617,
+      "step": 91320
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 2.2619984264358773e-06,
+      "loss": 2.0718,
+      "step": 91330
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 2.2598129207098524e-06,
+      "loss": 2.0885,
+      "step": 91340
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 2.2576274149838274e-06,
+      "loss": 2.0731,
+      "step": 91350
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 2.2554419092578024e-06,
+      "loss": 2.0854,
+      "step": 91360
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 2.2532564035317775e-06,
+      "loss": 2.115,
+      "step": 91370
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 2.2510708978057525e-06,
+      "loss": 2.1156,
+      "step": 91380
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 2.2488853920797275e-06,
+      "loss": 2.0955,
+      "step": 91390
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 2.246699886353702e-06,
+      "loss": 2.0452,
+      "step": 91400
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 2.244514380627677e-06,
+      "loss": 2.0606,
+      "step": 91410
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 2.2423288749016526e-06,
+      "loss": 2.1259,
+      "step": 91420
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 2.2401433691756277e-06,
+      "loss": 2.0351,
+      "step": 91430
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 2.2379578634496023e-06,
+      "loss": 2.1185,
+      "step": 91440
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 2.2357723577235773e-06,
+      "loss": 2.0954,
+      "step": 91450
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 2.2335868519975523e-06,
+      "loss": 2.0836,
+      "step": 91460
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 2.2314013462715274e-06,
+      "loss": 2.0784,
+      "step": 91470
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 2.2292158405455024e-06,
+      "loss": 2.0594,
+      "step": 91480
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 2.2270303348194774e-06,
+      "loss": 2.1081,
+      "step": 91490
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 2.2248448290934525e-06,
+      "loss": 2.0317,
+      "step": 91500
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 2.2226593233674275e-06,
+      "loss": 2.1398,
+      "step": 91510
+    },
+    {
+      "epoch": 18.0,
+      "eval_BLEU": 10.4597,
+      "eval_gen_len": 12.1899,
+      "eval_loss": 3.369558095932007,
+      "eval_runtime": 136.7056,
+      "eval_samples_per_second": 78.731,
+      "eval_steps_per_second": 0.315,
+      "step": 91512
+    }
+  ],
+  "max_steps": 101680,
+  "num_train_epochs": 20,
+  "total_flos": 5.233235502030651e+18,
+  "trial_name": null,
+  "trial_params": null
+}