diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15769 @@
+{
+  "best_metric": 0.7662602112929114,
+  "best_model_checkpoint": "wav2vec2-base-mirst500/checkpoint-26090",
+  "epoch": 10.0,
+  "global_step": 26090,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1498658489842852e-07,
+      "loss": 2.0809,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2997316979685704e-07,
+      "loss": 2.0826,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4495975469528556e-07,
+      "loss": 2.0793,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.599463395937141e-07,
+      "loss": 2.0825,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.749329244921425e-07,
+      "loss": 2.0785,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.899195093905711e-07,
+      "loss": 2.0799,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.049060942889996e-07,
+      "loss": 2.0789,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.198926791874282e-07,
+      "loss": 2.0798,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0348792640858565e-06,
+      "loss": 2.0765,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.149865848984285e-06,
+      "loss": 2.0825,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2648524338827137e-06,
+      "loss": 2.0783,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3798390187811422e-06,
+      "loss": 2.0757,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4948256036795708e-06,
+      "loss": 2.08,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6098121885779992e-06,
+      "loss": 2.0776,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7247987734764277e-06,
+      "loss": 2.0746,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8397853583748563e-06,
+      "loss": 2.0694,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9547719432732845e-06,
+      "loss": 2.0752,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.069758528171713e-06,
+      "loss": 2.0715,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1847451130701416e-06,
+      "loss": 2.0654,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.29973169796857e-06,
+      "loss": 2.0681,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4147182828669988e-06,
+      "loss": 2.0669,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5297048677654273e-06,
+      "loss": 2.0559,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.644691452663856e-06,
+      "loss": 2.0493,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7596780375622845e-06,
+      "loss": 2.0587,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.874664622460713e-06,
+      "loss": 2.0452,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9896512073591416e-06,
+      "loss": 2.0411,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1046377922575698e-06,
+      "loss": 2.0333,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.2196243771559983e-06,
+      "loss": 2.0246,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.334610962054427e-06,
+      "loss": 2.0095,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.4495975469528555e-06,
+      "loss": 2.0108,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.564584131851284e-06,
+      "loss": 1.9914,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.6795707167497126e-06,
+      "loss": 1.9905,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.7945573016481408e-06,
+      "loss": 2.0021,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.909543886546569e-06,
+      "loss": 2.0083,
+      "step": 340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.0245304714449975e-06,
+      "loss": 1.9917,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.139517056343426e-06,
+      "loss": 1.9845,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.254503641241855e-06,
+      "loss": 1.9991,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.369490226140283e-06,
+      "loss": 1.97,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.484476811038712e-06,
+      "loss": 1.9776,
+      "step": 390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.59946339593714e-06,
+      "loss": 1.9711,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.714449980835569e-06,
+      "loss": 1.9273,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8294365657339975e-06,
+      "loss": 1.9671,
+      "step": 420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.944423150632426e-06,
+      "loss": 1.9618,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.059409735530855e-06,
+      "loss": 1.9592,
+      "step": 440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.174396320429283e-06,
+      "loss": 1.9387,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.289382905327712e-06,
+      "loss": 1.9285,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.392870831736297e-06,
+      "loss": 1.9356,
+      "step": 470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.507857416634726e-06,
+      "loss": 1.8985,
+      "step": 480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.622844001533154e-06,
+      "loss": 1.9485,
+      "step": 490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.737830586431583e-06,
+      "loss": 1.9328,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.8528171713300115e-06,
+      "loss": 1.903,
+      "step": 510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96780375622844e-06,
+      "loss": 1.9217,
+      "step": 520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.082790341126869e-06,
+      "loss": 1.9428,
+      "step": 530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.197776926025297e-06,
+      "loss": 1.9103,
+      "step": 540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.312763510923726e-06,
+      "loss": 1.9054,
+      "step": 550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.427750095822154e-06,
+      "loss": 1.9398,
+      "step": 560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.542736680720583e-06,
+      "loss": 1.9347,
+      "step": 570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.6577232656190115e-06,
+      "loss": 1.8898,
+      "step": 580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.77270985051744e-06,
+      "loss": 1.9083,
+      "step": 590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.887696435415869e-06,
+      "loss": 1.9163,
+      "step": 600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.002683020314297e-06,
+      "loss": 1.8905,
+      "step": 610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.117669605212726e-06,
+      "loss": 1.9142,
+      "step": 620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.2326561901111535e-06,
+      "loss": 1.8944,
+      "step": 630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.347642775009582e-06,
+      "loss": 1.8967,
+      "step": 640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.462629359908011e-06,
+      "loss": 1.8727,
+      "step": 650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.577615944806439e-06,
+      "loss": 1.8877,
+      "step": 660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.692602529704868e-06,
+      "loss": 1.9041,
+      "step": 670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.807589114603296e-06,
+      "loss": 1.8957,
+      "step": 680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.922575699501725e-06,
+      "loss": 1.8808,
+      "step": 690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.037562284400154e-06,
+      "loss": 1.8558,
+      "step": 700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.152548869298582e-06,
+      "loss": 1.8852,
+      "step": 710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.26753545419701e-06,
+      "loss": 1.8812,
+      "step": 720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.38252203909544e-06,
+      "loss": 1.9017,
+      "step": 730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.497508623993868e-06,
+      "loss": 1.8998,
+      "step": 740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.612495208892295e-06,
+      "loss": 1.8669,
+      "step": 750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.727481793790725e-06,
+      "loss": 1.8892,
+      "step": 760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.842468378689152e-06,
+      "loss": 1.8814,
+      "step": 770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.957454963587582e-06,
+      "loss": 1.854,
+      "step": 780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.072441548486009e-06,
+      "loss": 1.8932,
+      "step": 790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.18742813338444e-06,
+      "loss": 1.8865,
+      "step": 800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.302414718282866e-06,
+      "loss": 1.8937,
+      "step": 810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.417401303181296e-06,
+      "loss": 1.8775,
+      "step": 820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.532387888079723e-06,
+      "loss": 1.8759,
+      "step": 830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.647374472978154e-06,
+      "loss": 1.7925,
+      "step": 840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.76236105787658e-06,
+      "loss": 1.8262,
+      "step": 850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.87734764277501e-06,
+      "loss": 1.8571,
+      "step": 860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.992334227673438e-06,
+      "loss": 1.8321,
+      "step": 870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0107320812571868e-05,
+      "loss": 1.8251,
+      "step": 880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0222307397470295e-05,
+      "loss": 1.8128,
+      "step": 890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.0337293982368725e-05,
+      "loss": 1.8149,
+      "step": 900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0452280567267152e-05,
+      "loss": 1.8517,
+      "step": 910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.0567267152165582e-05,
+      "loss": 1.8439,
+      "step": 920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0682253737064009e-05,
+      "loss": 1.83,
+      "step": 930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.079724032196244e-05,
+      "loss": 1.8438,
+      "step": 940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.0912226906860866e-05,
+      "loss": 1.8024,
+      "step": 950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.1027213491759295e-05,
+      "loss": 1.8328,
+      "step": 960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.1142200076657723e-05,
+      "loss": 1.799,
+      "step": 970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.1257186661556152e-05,
+      "loss": 1.8646,
+      "step": 980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.137217324645458e-05,
+      "loss": 1.8245,
+      "step": 990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.1487159831353009e-05,
+      "loss": 1.7966,
+      "step": 1000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1602146416251438e-05,
+      "loss": 1.7852,
+      "step": 1010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1717133001149866e-05,
+      "loss": 1.7661,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1832119586048295e-05,
+      "loss": 1.7789,
+      "step": 1030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1947106170946723e-05,
+      "loss": 1.7793,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2062092755845152e-05,
+      "loss": 1.8356,
+      "step": 1050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.217707934074358e-05,
+      "loss": 1.7949,
+      "step": 1060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2292065925642009e-05,
+      "loss": 1.7382,
+      "step": 1070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2407052510540438e-05,
+      "loss": 1.7778,
+      "step": 1080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2522039095438866e-05,
+      "loss": 1.7256,
+      "step": 1090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2637025680337295e-05,
+      "loss": 1.7275,
+      "step": 1100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2752012265235723e-05,
+      "loss": 1.8011,
+      "step": 1110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2866998850134152e-05,
+      "loss": 1.7839,
+      "step": 1120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.298198543503258e-05,
+      "loss": 1.7867,
+      "step": 1130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3096972019931009e-05,
+      "loss": 1.7877,
+      "step": 1140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3211958604829436e-05,
+      "loss": 1.7485,
+      "step": 1150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3326945189727866e-05,
+      "loss": 1.7283,
+      "step": 1160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3441931774626293e-05,
+      "loss": 1.7347,
+      "step": 1170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3556918359524723e-05,
+      "loss": 1.7604,
+      "step": 1180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.367190494442315e-05,
+      "loss": 1.6977,
+      "step": 1190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.378689152932158e-05,
+      "loss": 1.7332,
+      "step": 1200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3901878114220007e-05,
+      "loss": 1.7263,
+      "step": 1210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.4016864699118438e-05,
+      "loss": 1.7047,
+      "step": 1220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.4131851284016864e-05,
+      "loss": 1.7593,
+      "step": 1230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.4246837868915295e-05,
+      "loss": 1.7368,
+      "step": 1240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.4361824453813722e-05,
+      "loss": 1.7325,
+      "step": 1250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.4476811038712152e-05,
+      "loss": 1.6725,
+      "step": 1260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.4591797623610579e-05,
+      "loss": 1.7199,
+      "step": 1270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.4706784208509007e-05,
+      "loss": 1.6786,
+      "step": 1280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.4821770793407436e-05,
+      "loss": 1.6769,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.4936757378305864e-05,
+      "loss": 1.669,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5051743963204293e-05,
+      "loss": 1.6373,
+      "step": 1310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.516673054810272e-05,
+      "loss": 1.6432,
+      "step": 1320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.528171713300115e-05,
+      "loss": 1.6666,
+      "step": 1330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.539670371789958e-05,
+      "loss": 1.7061,
+      "step": 1340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5511690302798007e-05,
+      "loss": 1.652,
+      "step": 1350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5626676887696436e-05,
+      "loss": 1.6681,
+      "step": 1360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.574166347259486e-05,
+      "loss": 1.5956,
+      "step": 1370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5856650057493293e-05,
+      "loss": 1.6581,
+      "step": 1380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.597163664239172e-05,
+      "loss": 1.6153,
+      "step": 1390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.608662322729015e-05,
+      "loss": 1.6522,
+      "step": 1400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6201609812188575e-05,
+      "loss": 1.5935,
+      "step": 1410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6316596397087007e-05,
+      "loss": 1.6758,
+      "step": 1420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6431582981985436e-05,
+      "loss": 1.5831,
+      "step": 1430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6546569566883865e-05,
+      "loss": 1.6165,
+      "step": 1440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.666155615178229e-05,
+      "loss": 1.6222,
+      "step": 1450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.677654273668072e-05,
+      "loss": 1.606,
+      "step": 1460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.689152932157915e-05,
+      "loss": 1.6208,
+      "step": 1470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.700651590647758e-05,
+      "loss": 1.5936,
+      "step": 1480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7121502491376004e-05,
+      "loss": 1.5332,
+      "step": 1490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7236489076274436e-05,
+      "loss": 1.5827,
+      "step": 1500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7351475661172865e-05,
+      "loss": 1.5474,
+      "step": 1510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7466462246071293e-05,
+      "loss": 1.5729,
+      "step": 1520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7581448830969718e-05,
+      "loss": 1.5742,
+      "step": 1530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.769643541586815e-05,
+      "loss": 1.4756,
+      "step": 1540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.781142200076658e-05,
+      "loss": 1.4994,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7926408585665007e-05,
+      "loss": 1.496,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8041395170563433e-05,
+      "loss": 1.5161,
+      "step": 1570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.815638175546186e-05,
+      "loss": 1.4848,
+      "step": 1580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8271368340360293e-05,
+      "loss": 1.5293,
+      "step": 1590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.838635492525872e-05,
+      "loss": 1.5262,
+      "step": 1600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8501341510157147e-05,
+      "loss": 1.4921,
+      "step": 1610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8616328095055575e-05,
+      "loss": 1.4894,
+      "step": 1620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8731314679954007e-05,
+      "loss": 1.5104,
+      "step": 1630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8846301264852436e-05,
+      "loss": 1.4651,
+      "step": 1640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.896128784975086e-05,
+      "loss": 1.4319,
+      "step": 1650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.907627443464929e-05,
+      "loss": 1.4929,
+      "step": 1660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.919126101954772e-05,
+      "loss": 1.4852,
+      "step": 1670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.930624760444615e-05,
+      "loss": 1.4824,
+      "step": 1680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9421234189344575e-05,
+      "loss": 1.4236,
+      "step": 1690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9536220774243004e-05,
+      "loss": 1.4967,
+      "step": 1700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9651207359141436e-05,
+      "loss": 1.481,
+      "step": 1710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9766193944039865e-05,
+      "loss": 1.4236,
+      "step": 1720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.988118052893829e-05,
+      "loss": 1.4669,
+      "step": 1730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999616711383672e-05,
+      "loss": 1.4596,
+      "step": 1740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.011115369873515e-05,
+      "loss": 1.4675,
+      "step": 1750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.022614028363358e-05,
+      "loss": 1.4355,
+      "step": 1760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.0341126868532004e-05,
+      "loss": 1.4587,
+      "step": 1770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.0456113453430433e-05,
+      "loss": 1.4677,
+      "step": 1780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.057110003832886e-05,
+      "loss": 1.4513,
+      "step": 1790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.0686086623227293e-05,
+      "loss": 1.4613,
+      "step": 1800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.080107320812572e-05,
+      "loss": 1.43,
+      "step": 1810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0916059793024147e-05,
+      "loss": 1.4312,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1031046377922575e-05,
+      "loss": 1.476,
+      "step": 1830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1146032962821007e-05,
+      "loss": 1.4366,
+      "step": 1840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1261019547719433e-05,
+      "loss": 1.4365,
+      "step": 1850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.137600613261786e-05,
+      "loss": 1.4557,
+      "step": 1860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.149099271751629e-05,
+      "loss": 1.4191,
+      "step": 1870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1605979302414722e-05,
+      "loss": 1.4422,
+      "step": 1880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1720965887313147e-05,
+      "loss": 1.472,
+      "step": 1890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1835952472211575e-05,
+      "loss": 1.391,
+      "step": 1900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1950939057110004e-05,
+      "loss": 1.3327,
+      "step": 1910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.2065925642008436e-05,
+      "loss": 1.3913,
+      "step": 1920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.218091222690686e-05,
+      "loss": 1.4128,
+      "step": 1930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.229589881180529e-05,
+      "loss": 1.447,
+      "step": 1940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.241088539670372e-05,
+      "loss": 1.3915,
+      "step": 1950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2525871981602147e-05,
+      "loss": 1.3157,
+      "step": 1960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2640858566500575e-05,
+      "loss": 1.4321,
+      "step": 1970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2755845151399004e-05,
+      "loss": 1.3482,
+      "step": 1980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2870831736297433e-05,
+      "loss": 1.4327,
+      "step": 1990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2985818321195858e-05,
+      "loss": 1.389,
+      "step": 2000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.310080490609429e-05,
+      "loss": 1.3312,
+      "step": 2010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.321579149099272e-05,
+      "loss": 1.3191,
+      "step": 2020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3330778075891147e-05,
+      "loss": 1.3441,
+      "step": 2030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3445764660789572e-05,
+      "loss": 1.4519,
+      "step": 2040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3560751245688004e-05,
+      "loss": 1.4266,
+      "step": 2050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3675737830586433e-05,
+      "loss": 1.3635,
+      "step": 2060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.379072441548486e-05,
+      "loss": 1.4031,
+      "step": 2070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3905711000383286e-05,
+      "loss": 1.3573,
+      "step": 2080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.402069758528172e-05,
+      "loss": 1.3383,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4135684170180147e-05,
+      "loss": 1.3363,
+      "step": 2100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4250670755078576e-05,
+      "loss": 1.3777,
+      "step": 2110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4365657339977e-05,
+      "loss": 1.3103,
+      "step": 2120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4480643924875433e-05,
+      "loss": 1.3438,
+      "step": 2130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.459563050977386e-05,
+      "loss": 1.3614,
+      "step": 2140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.471061709467229e-05,
+      "loss": 1.3448,
+      "step": 2150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4825603679570715e-05,
+      "loss": 1.2961,
+      "step": 2160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4940590264469147e-05,
+      "loss": 1.4005,
+      "step": 2170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5055576849367576e-05,
+      "loss": 1.328,
+      "step": 2180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5170563434266004e-05,
+      "loss": 1.2661,
+      "step": 2190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.528555001916443e-05,
+      "loss": 1.285,
+      "step": 2200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5400536604062858e-05,
+      "loss": 1.3364,
+      "step": 2210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.551552318896129e-05,
+      "loss": 1.3519,
+      "step": 2220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.563050977385972e-05,
+      "loss": 1.3864,
+      "step": 2230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5745496358758144e-05,
+      "loss": 1.2857,
+      "step": 2240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5860482943656572e-05,
+      "loss": 1.3287,
+      "step": 2250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5975469528555004e-05,
+      "loss": 1.3407,
+      "step": 2260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6090456113453433e-05,
+      "loss": 1.3457,
+      "step": 2270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6205442698351858e-05,
+      "loss": 1.2857,
+      "step": 2280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6320429283250286e-05,
+      "loss": 1.3198,
+      "step": 2290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.643541586814872e-05,
+      "loss": 1.338,
+      "step": 2300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6550402453047147e-05,
+      "loss": 1.3066,
+      "step": 2310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6665389037945572e-05,
+      "loss": 1.3071,
+      "step": 2320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6780375622844e-05,
+      "loss": 1.2945,
+      "step": 2330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6895362207742433e-05,
+      "loss": 1.3444,
+      "step": 2340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.701034879264086e-05,
+      "loss": 1.3123,
+      "step": 2350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7125335377539286e-05,
+      "loss": 1.3321,
+      "step": 2360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7240321962437715e-05,
+      "loss": 1.2505,
+      "step": 2370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7355308547336147e-05,
+      "loss": 1.3014,
+      "step": 2380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7470295132234576e-05,
+      "loss": 1.3321,
+      "step": 2390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7585281717133e-05,
+      "loss": 1.3096,
+      "step": 2400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.770026830203143e-05,
+      "loss": 1.31,
+      "step": 2410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7815254886929858e-05,
+      "loss": 1.335,
+      "step": 2420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.793024147182829e-05,
+      "loss": 1.2492,
+      "step": 2430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8045228056726715e-05,
+      "loss": 1.2654,
+      "step": 2440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8160214641625144e-05,
+      "loss": 1.2852,
+      "step": 2450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8275201226523572e-05,
+      "loss": 1.2642,
+      "step": 2460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8390187811422004e-05,
+      "loss": 1.2931,
+      "step": 2470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.850517439632043e-05,
+      "loss": 1.3108,
+      "step": 2480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8620160981218858e-05,
+      "loss": 1.347,
+      "step": 2490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8735147566117286e-05,
+      "loss": 1.3857,
+      "step": 2500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.885013415101572e-05,
+      "loss": 1.3367,
+      "step": 2510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8965120735914144e-05,
+      "loss": 1.3258,
+      "step": 2520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9080107320812572e-05,
+      "loss": 1.3037,
+      "step": 2530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9195093905711e-05,
+      "loss": 1.2947,
+      "step": 2540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9310080490609433e-05,
+      "loss": 1.3097,
+      "step": 2550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9425067075507858e-05,
+      "loss": 1.3048,
+      "step": 2560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9528555001916443e-05,
+      "loss": 1.2779,
+      "step": 2570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9643541586814872e-05,
+      "loss": 1.2939,
+      "step": 2580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.97585281717133e-05,
+      "loss": 1.2575,
+      "step": 2590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.987351475661173e-05,
+      "loss": 1.272,
+      "step": 2600
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.5617947919411638,
+      "eval_loss": 1.1589456796646118,
+      "eval_runtime": 297.7176,
+      "eval_samples_per_second": 140.21,
+      "eval_steps_per_second": 70.107,
+      "step": 2609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9988501341510158e-05,
+      "loss": 1.4261,
+      "step": 2610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9988501341510158e-05,
+      "loss": 1.2925,
+      "step": 2620
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.997572505429922e-05,
+      "loss": 1.3176,
+      "step": 2630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9962948767088286e-05,
+      "loss": 1.2765,
+      "step": 2640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9950172479877348e-05,
+      "loss": 1.2774,
+      "step": 2650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.993739619266641e-05,
+      "loss": 1.2715,
+      "step": 2660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9924619905455476e-05,
+      "loss": 1.2701,
+      "step": 2670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.991184361824454e-05,
+      "loss": 1.241,
+      "step": 2680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.9899067331033604e-05,
+      "loss": 1.2613,
+      "step": 2690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.9886291043822667e-05,
+      "loss": 1.2404,
+      "step": 2700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.987351475661173e-05,
+      "loss": 1.2018,
+      "step": 2710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.986073846940079e-05,
+      "loss": 1.2422,
+      "step": 2720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.9847962182189857e-05,
+      "loss": 1.2276,
+      "step": 2730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.983518589497892e-05,
+      "loss": 1.3064,
+      "step": 2740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.9822409607767985e-05,
+      "loss": 1.2486,
+      "step": 2750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.9809633320557048e-05,
+      "loss": 1.3621,
+      "step": 2760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.979685703334611e-05,
+      "loss": 1.2817,
+      "step": 2770
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.9784080746135172e-05,
+      "loss": 1.2246,
+      "step": 2780
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.977258208764533e-05,
+      "loss": 1.2385,
+      "step": 2790
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.9759805800434395e-05,
+      "loss": 1.2707,
+      "step": 2800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.9747029513223457e-05,
+      "loss": 1.3105,
+      "step": 2810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.9734253226012523e-05,
+      "loss": 1.2547,
+      "step": 2820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.9721476938801585e-05,
+      "loss": 1.2105,
+      "step": 2830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.9708700651590648e-05,
+      "loss": 1.2743,
+      "step": 2840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.969592436437971e-05,
+      "loss": 1.2738,
+      "step": 2850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.9683148077168776e-05,
+      "loss": 1.2489,
+      "step": 2860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.9670371789957838e-05,
+      "loss": 1.2824,
+      "step": 2870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.9657595502746904e-05,
+      "loss": 1.2434,
+      "step": 2880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.9644819215535966e-05,
+      "loss": 1.203,
+      "step": 2890
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.963204292832503e-05,
+      "loss": 1.2387,
+      "step": 2900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.9619266641114094e-05,
+      "loss": 1.2014,
+      "step": 2910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.9606490353903157e-05,
+      "loss": 1.2252,
+      "step": 2920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.9593714066692223e-05,
+      "loss": 1.2449,
+      "step": 2930
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.958093777948128e-05,
+      "loss": 1.2564,
+      "step": 2940
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.9568161492270347e-05,
+      "loss": 1.2215,
+      "step": 2950
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.955538520505941e-05,
+      "loss": 1.2514,
+      "step": 2960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.9542608917848475e-05,
+      "loss": 1.172,
+      "step": 2970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.9529832630637538e-05,
+      "loss": 1.2287,
+      "step": 2980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.95170563434266e-05,
+      "loss": 1.2363,
+      "step": 2990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.9504280056215662e-05,
+      "loss": 1.262,
+      "step": 3000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.9491503769004728e-05,
+      "loss": 1.2625,
+      "step": 3010
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.947872748179379e-05,
+      "loss": 1.1952,
+      "step": 3020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.9465951194582856e-05,
+      "loss": 1.1656,
+      "step": 3030
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.945317490737192e-05,
+      "loss": 1.2057,
+      "step": 3040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.944039862016098e-05,
+      "loss": 1.2663,
+      "step": 3050
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.9427622332950047e-05,
+      "loss": 1.265,
+      "step": 3060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.941484604573911e-05,
+      "loss": 1.2173,
+      "step": 3070
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9402069758528175e-05,
+      "loss": 1.2404,
+      "step": 3080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9389293471317237e-05,
+      "loss": 1.2088,
+      "step": 3090
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.93765171841063e-05,
+      "loss": 1.2703,
+      "step": 3100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.9363740896895362e-05,
+      "loss": 1.2675,
+      "step": 3110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9350964609684428e-05,
+      "loss": 1.2993,
+      "step": 3120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.933818832247349e-05,
+      "loss": 1.2086,
+      "step": 3130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9325412035262556e-05,
+      "loss": 1.2295,
+      "step": 3140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9312635748051615e-05,
+      "loss": 1.2653,
+      "step": 3150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.929985946084068e-05,
+      "loss": 1.2664,
+      "step": 3160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9287083173629743e-05,
+      "loss": 1.1978,
+      "step": 3170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.927430688641881e-05,
+      "loss": 1.247,
+      "step": 3180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.926153059920787e-05,
+      "loss": 1.2309,
+      "step": 3190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9248754311996933e-05,
+      "loss": 1.1825,
+      "step": 3200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9235978024785996e-05,
+      "loss": 1.2008,
+      "step": 3210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.922320173757506e-05,
+      "loss": 1.2321,
+      "step": 3220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9210425450364124e-05,
+      "loss": 1.2492,
+      "step": 3230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.919764916315319e-05,
+      "loss": 1.2045,
+      "step": 3240
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9184872875942252e-05,
+      "loss": 1.2977,
+      "step": 3250
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9172096588731314e-05,
+      "loss": 1.1696,
+      "step": 3260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.915932030152038e-05,
+      "loss": 1.1654,
+      "step": 3270
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9146544014309442e-05,
+      "loss": 1.2436,
+      "step": 3280
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9133767727098508e-05,
+      "loss": 1.2246,
+      "step": 3290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9120991439887567e-05,
+      "loss": 1.1985,
+      "step": 3300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9108215152676633e-05,
+      "loss": 1.2358,
+      "step": 3310
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9095438865465695e-05,
+      "loss": 1.1543,
+      "step": 3320
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.908266257825476e-05,
+      "loss": 1.2323,
+      "step": 3330
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.9069886291043823e-05,
+      "loss": 1.1982,
+      "step": 3340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.9057110003832886e-05,
+      "loss": 1.2133,
+      "step": 3350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.9044333716621948e-05,
+      "loss": 1.2035,
+      "step": 3360
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.9031557429411014e-05,
+      "loss": 1.1577,
+      "step": 3370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.9018781142200076e-05,
+      "loss": 1.2358,
+      "step": 3380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.9006004854989142e-05,
+      "loss": 1.1823,
+      "step": 3390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8993228567778204e-05,
+      "loss": 1.1481,
+      "step": 3400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8980452280567267e-05,
+      "loss": 1.2236,
+      "step": 3410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8967675993356332e-05,
+      "loss": 1.176,
+      "step": 3420
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8954899706145395e-05,
+      "loss": 1.2002,
+      "step": 3430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.894212341893446e-05,
+      "loss": 1.1854,
+      "step": 3440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8929347131723523e-05,
+      "loss": 1.2288,
+      "step": 3450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8916570844512585e-05,
+      "loss": 1.2526,
+      "step": 3460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8903794557301648e-05,
+      "loss": 1.23,
+      "step": 3470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8891018270090713e-05,
+      "loss": 1.2538,
+      "step": 3480
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8878241982879776e-05,
+      "loss": 1.1736,
+      "step": 3490
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.886546569566884e-05,
+      "loss": 1.2168,
+      "step": 3500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.88526894084579e-05,
+      "loss": 1.2636,
+      "step": 3510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.8839913121246966e-05,
+      "loss": 1.1677,
+      "step": 3520
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.882713683403603e-05,
+      "loss": 1.2067,
+      "step": 3530
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8814360546825094e-05,
+      "loss": 1.2135,
+      "step": 3540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8801584259614157e-05,
+      "loss": 1.22,
+      "step": 3550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.878880797240322e-05,
+      "loss": 1.1231,
+      "step": 3560
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.877603168519228e-05,
+      "loss": 1.2263,
+      "step": 3570
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.8763255397981347e-05,
+      "loss": 1.1942,
+      "step": 3580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8750479110770413e-05,
+      "loss": 1.2248,
+      "step": 3590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8737702823559475e-05,
+      "loss": 1.2274,
+      "step": 3600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8724926536348538e-05,
+      "loss": 1.1893,
+      "step": 3610
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.87121502491376e-05,
+      "loss": 1.1591,
+      "step": 3620
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.8699373961926666e-05,
+      "loss": 1.2181,
+      "step": 3630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.8686597674715728e-05,
+      "loss": 1.2559,
+      "step": 3640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.8673821387504794e-05,
+      "loss": 1.2094,
+      "step": 3650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.8661045100293853e-05,
+      "loss": 1.1637,
+      "step": 3660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.864826881308292e-05,
+      "loss": 1.2916,
+      "step": 3670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.863549252587198e-05,
+      "loss": 1.2407,
+      "step": 3680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.8622716238661047e-05,
+      "loss": 1.2177,
+      "step": 3690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.860993995145011e-05,
+      "loss": 1.1724,
+      "step": 3700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.859716366423917e-05,
+      "loss": 1.1939,
+      "step": 3710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.8584387377028234e-05,
+      "loss": 1.1435,
+      "step": 3720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.85716110898173e-05,
+      "loss": 1.1551,
+      "step": 3730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.8558834802606362e-05,
+      "loss": 1.1566,
+      "step": 3740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8546058515395428e-05,
+      "loss": 1.1552,
+      "step": 3750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8533282228184493e-05,
+      "loss": 1.2611,
+      "step": 3760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8520505940973552e-05,
+      "loss": 1.2098,
+      "step": 3770
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.8507729653762618e-05,
+      "loss": 1.2531,
+      "step": 3780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.849495336655168e-05,
+      "loss": 1.244,
+      "step": 3790
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8482177079340746e-05,
+      "loss": 1.1791,
+      "step": 3800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.846940079212981e-05,
+      "loss": 1.1851,
+      "step": 3810
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.845662450491887e-05,
+      "loss": 1.2081,
+      "step": 3820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8443848217707933e-05,
+      "loss": 1.1826,
+      "step": 3830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8431071930497e-05,
+      "loss": 1.1684,
+      "step": 3840
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.841829564328606e-05,
+      "loss": 1.1964,
+      "step": 3850
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.8405519356075127e-05,
+      "loss": 1.2004,
+      "step": 3860
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.8392743068864186e-05,
+      "loss": 1.197,
+      "step": 3870
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.8379966781653252e-05,
+      "loss": 1.1414,
+      "step": 3880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.8367190494442314e-05,
+      "loss": 1.2399,
+      "step": 3890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.835441420723138e-05,
+      "loss": 1.1402,
+      "step": 3900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.8341637920020442e-05,
+      "loss": 1.0983,
+      "step": 3910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.8328861632809505e-05,
+      "loss": 1.1021,
+      "step": 3920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.831608534559857e-05,
+      "loss": 1.0821,
+      "step": 3930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8303309058387633e-05,
+      "loss": 1.2289,
+      "step": 3940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.82905327711767e-05,
+      "loss": 1.2545,
+      "step": 3950
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.827775648396576e-05,
+      "loss": 1.2209,
+      "step": 3960
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8264980196754823e-05,
+      "loss": 1.1316,
+      "step": 3970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8252203909543886e-05,
+      "loss": 1.141,
+      "step": 3980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.823942762233295e-05,
+      "loss": 1.2099,
+      "step": 3990
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8226651335122014e-05,
+      "loss": 1.2144,
+      "step": 4000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.821387504791108e-05,
+      "loss": 1.0854,
+      "step": 4010
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.820109876070014e-05,
+      "loss": 1.1082,
+      "step": 4020
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.8188322473489204e-05,
+      "loss": 1.1284,
+      "step": 4030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.8175546186278267e-05,
+      "loss": 1.1415,
+      "step": 4040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.8162769899067332e-05,
+      "loss": 1.0973,
+      "step": 4050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.8149993611856395e-05,
+      "loss": 1.154,
+      "step": 4060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.813721732464546e-05,
+      "loss": 1.2014,
+      "step": 4070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.812444103743452e-05,
+      "loss": 1.1277,
+      "step": 4080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.8111664750223585e-05,
+      "loss": 1.15,
+      "step": 4090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.809888846301265e-05,
+      "loss": 1.1828,
+      "step": 4100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.8086112175801713e-05,
+      "loss": 1.1453,
+      "step": 4110
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.807333588859078e-05,
+      "loss": 1.2141,
+      "step": 4120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.8060559601379838e-05,
+      "loss": 1.1202,
+      "step": 4130
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8047783314168904e-05,
+      "loss": 1.2477,
+      "step": 4140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8035007026957966e-05,
+      "loss": 1.17,
+      "step": 4150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8022230739747032e-05,
+      "loss": 1.1806,
+      "step": 4160
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8009454452536094e-05,
+      "loss": 1.1241,
+      "step": 4170
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.7996678165325157e-05,
+      "loss": 1.1069,
+      "step": 4180
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.798390187811422e-05,
+      "loss": 1.1165,
+      "step": 4190
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7971125590903285e-05,
+      "loss": 1.149,
+      "step": 4200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7958349303692347e-05,
+      "loss": 1.1921,
+      "step": 4210
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7945573016481413e-05,
+      "loss": 1.1559,
+      "step": 4220
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7932796729270472e-05,
+      "loss": 1.1451,
+      "step": 4230
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7920020442059538e-05,
+      "loss": 1.1556,
+      "step": 4240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.79072441548486e-05,
+      "loss": 1.1168,
+      "step": 4250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7894467867637666e-05,
+      "loss": 1.1272,
+      "step": 4260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.788169158042673e-05,
+      "loss": 1.1107,
+      "step": 4270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.786891529321579e-05,
+      "loss": 1.0901,
+      "step": 4280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.7856139006004856e-05,
+      "loss": 1.073,
+      "step": 4290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.784336271879392e-05,
+      "loss": 1.15,
+      "step": 4300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.7830586431582984e-05,
+      "loss": 1.1264,
+      "step": 4310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.7817810144372047e-05,
+      "loss": 1.2153,
+      "step": 4320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.780503385716111e-05,
+      "loss": 1.147,
+      "step": 4330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.779225756995017e-05,
+      "loss": 1.132,
+      "step": 4340
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7779481282739237e-05,
+      "loss": 1.1926,
+      "step": 4350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.77667049955283e-05,
+      "loss": 1.147,
+      "step": 4360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7753928708317365e-05,
+      "loss": 1.2182,
+      "step": 4370
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.7741152421106428e-05,
+      "loss": 1.2095,
+      "step": 4380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.772837613389549e-05,
+      "loss": 1.1776,
+      "step": 4390
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.7715599846684552e-05,
+      "loss": 1.0947,
+      "step": 4400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.7702823559473618e-05,
+      "loss": 1.0506,
+      "step": 4410
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.769004727226268e-05,
+      "loss": 1.1986,
+      "step": 4420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.7677270985051746e-05,
+      "loss": 1.1531,
+      "step": 4430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.766449469784081e-05,
+      "loss": 1.224,
+      "step": 4440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.765171841062987e-05,
+      "loss": 1.1256,
+      "step": 4450
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.7638942123418937e-05,
+      "loss": 1.2251,
+      "step": 4460
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.7626165836208e-05,
+      "loss": 1.196,
+      "step": 4470
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7613389548997065e-05,
+      "loss": 1.1201,
+      "step": 4480
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7600613261786124e-05,
+      "loss": 1.1389,
+      "step": 4490
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.758783697457519e-05,
+      "loss": 1.1334,
+      "step": 4500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7575060687364252e-05,
+      "loss": 1.1685,
+      "step": 4510
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7562284400153318e-05,
+      "loss": 1.1864,
+      "step": 4520
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.754950811294238e-05,
+      "loss": 1.177,
+      "step": 4530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7536731825731442e-05,
+      "loss": 1.1563,
+      "step": 4540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7523955538520505e-05,
+      "loss": 1.187,
+      "step": 4550
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.751117925130957e-05,
+      "loss": 1.1834,
+      "step": 4560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.7498402964098633e-05,
+      "loss": 1.1543,
+      "step": 4570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.74856266768877e-05,
+      "loss": 1.1232,
+      "step": 4580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7472850389676758e-05,
+      "loss": 1.115,
+      "step": 4590
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7460074102465823e-05,
+      "loss": 1.1729,
+      "step": 4600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.744729781525489e-05,
+      "loss": 1.1257,
+      "step": 4610
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.743452152804395e-05,
+      "loss": 1.1555,
+      "step": 4620
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7421745240833017e-05,
+      "loss": 1.1492,
+      "step": 4630
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7408968953622076e-05,
+      "loss": 1.1116,
+      "step": 4640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7396192666411142e-05,
+      "loss": 1.0887,
+      "step": 4650
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7383416379200204e-05,
+      "loss": 1.1007,
+      "step": 4660
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.737064009198927e-05,
+      "loss": 1.0951,
+      "step": 4670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7357863804778332e-05,
+      "loss": 1.0771,
+      "step": 4680
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7345087517567395e-05,
+      "loss": 1.1075,
+      "step": 4690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7333588859077555e-05,
+      "loss": 1.2025,
+      "step": 4700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7320812571866617e-05,
+      "loss": 1.1542,
+      "step": 4710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.730803628465568e-05,
+      "loss": 1.1391,
+      "step": 4720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7295259997444742e-05,
+      "loss": 1.168,
+      "step": 4730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7282483710233808e-05,
+      "loss": 1.0547,
+      "step": 4740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.726970742302287e-05,
+      "loss": 1.0872,
+      "step": 4750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7256931135811936e-05,
+      "loss": 1.1366,
+      "step": 4760
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7244154848600995e-05,
+      "loss": 1.1685,
+      "step": 4770
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.723137856139006e-05,
+      "loss": 1.1861,
+      "step": 4780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7218602274179123e-05,
+      "loss": 1.1452,
+      "step": 4790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.720582598696819e-05,
+      "loss": 1.1779,
+      "step": 4800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.719304969975725e-05,
+      "loss": 1.0635,
+      "step": 4810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7180273412546314e-05,
+      "loss": 1.045,
+      "step": 4820
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.716749712533538e-05,
+      "loss": 1.2208,
+      "step": 4830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.715472083812444e-05,
+      "loss": 1.1713,
+      "step": 4840
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7141944550913507e-05,
+      "loss": 1.1346,
+      "step": 4850
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.712916826370257e-05,
+      "loss": 1.1371,
+      "step": 4860
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.7116391976491632e-05,
+      "loss": 1.0715,
+      "step": 4870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.7103615689280694e-05,
+      "loss": 1.1294,
+      "step": 4880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.709083940206976e-05,
+      "loss": 1.0605,
+      "step": 4890
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7078063114858823e-05,
+      "loss": 1.131,
+      "step": 4900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.706528682764789e-05,
+      "loss": 1.1283,
+      "step": 4910
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.7052510540436947e-05,
+      "loss": 1.1328,
+      "step": 4920
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.7039734253226013e-05,
+      "loss": 1.0773,
+      "step": 4930
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.7026957966015075e-05,
+      "loss": 1.1057,
+      "step": 4940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.701418167880414e-05,
+      "loss": 1.1657,
+      "step": 4950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.7001405391593204e-05,
+      "loss": 1.1232,
+      "step": 4960
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6988629104382266e-05,
+      "loss": 1.1655,
+      "step": 4970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6975852817171328e-05,
+      "loss": 1.0693,
+      "step": 4980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6963076529960394e-05,
+      "loss": 1.0329,
+      "step": 4990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6950300242749456e-05,
+      "loss": 1.1258,
+      "step": 5000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6937523955538522e-05,
+      "loss": 1.1272,
+      "step": 5010
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6924747668327585e-05,
+      "loss": 1.1643,
+      "step": 5020
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6911971381116647e-05,
+      "loss": 1.1449,
+      "step": 5030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6899195093905713e-05,
+      "loss": 1.1833,
+      "step": 5040
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6886418806694775e-05,
+      "loss": 1.0866,
+      "step": 5050
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.687364251948384e-05,
+      "loss": 1.0456,
+      "step": 5060
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6860866232272903e-05,
+      "loss": 1.1877,
+      "step": 5070
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6848089945061965e-05,
+      "loss": 1.0884,
+      "step": 5080
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6835313657851028e-05,
+      "loss": 1.1218,
+      "step": 5090
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6822537370640094e-05,
+      "loss": 1.1494,
+      "step": 5100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6809761083429156e-05,
+      "loss": 1.1362,
+      "step": 5110
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.679698479621822e-05,
+      "loss": 1.1797,
+      "step": 5120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.678420850900728e-05,
+      "loss": 1.096,
+      "step": 5130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.6771432221796346e-05,
+      "loss": 1.0352,
+      "step": 5140
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.675865593458541e-05,
+      "loss": 1.0766,
+      "step": 5150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6745879647374475e-05,
+      "loss": 1.0958,
+      "step": 5160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6733103360163537e-05,
+      "loss": 1.0997,
+      "step": 5170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.67203270729526e-05,
+      "loss": 1.2081,
+      "step": 5180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6707550785741665e-05,
+      "loss": 1.0706,
+      "step": 5190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6694774498530727e-05,
+      "loss": 1.1818,
+      "step": 5200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6681998211319793e-05,
+      "loss": 1.1918,
+      "step": 5210
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.6628177179407326,
+      "eval_loss": 0.9315594434738159,
+      "eval_runtime": 299.8356,
+      "eval_samples_per_second": 139.22,
+      "eval_steps_per_second": 69.611,
+      "step": 5218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6669221924108855e-05,
+      "loss": 1.1672,
+      "step": 5220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6656445636897918e-05,
+      "loss": 1.1412,
+      "step": 5230
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.664366934968698e-05,
+      "loss": 1.0237,
+      "step": 5240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6630893062476046e-05,
+      "loss": 1.1565,
+      "step": 5250
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.661811677526511e-05,
+      "loss": 1.1093,
+      "step": 5260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.6605340488054174e-05,
+      "loss": 1.0703,
+      "step": 5270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.6592564200843233e-05,
+      "loss": 1.122,
+      "step": 5280
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.65797879136323e-05,
+      "loss": 1.0192,
+      "step": 5290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.656701162642136e-05,
+      "loss": 1.1086,
+      "step": 5300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.6554235339210427e-05,
+      "loss": 1.0331,
+      "step": 5310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.654145905199949e-05,
+      "loss": 1.122,
+      "step": 5320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.652868276478855e-05,
+      "loss": 1.1772,
+      "step": 5330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.6515906477577614e-05,
+      "loss": 1.2321,
+      "step": 5340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.650313019036668e-05,
+      "loss": 1.0778,
+      "step": 5350
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.6490353903155745e-05,
+      "loss": 1.111,
+      "step": 5360
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6477577615944808e-05,
+      "loss": 1.206,
+      "step": 5370
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6464801328733874e-05,
+      "loss": 1.0628,
+      "step": 5380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6452025041522933e-05,
+      "loss": 1.0674,
+      "step": 5390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6439248754312e-05,
+      "loss": 1.1364,
+      "step": 5400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.642647246710106e-05,
+      "loss": 1.0581,
+      "step": 5410
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6413696179890126e-05,
+      "loss": 1.0979,
+      "step": 5420
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.640091989267919e-05,
+      "loss": 1.0712,
+      "step": 5430
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.638814360546825e-05,
+      "loss": 1.1597,
+      "step": 5440
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6375367318257314e-05,
+      "loss": 1.1,
+      "step": 5450
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.636259103104638e-05,
+      "loss": 1.1021,
+      "step": 5460
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.634981474383544e-05,
+      "loss": 1.199,
+      "step": 5470
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6337038456624507e-05,
+      "loss": 1.1968,
+      "step": 5480
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6324262169413566e-05,
+      "loss": 1.1201,
+      "step": 5490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6311485882202632e-05,
+      "loss": 1.0864,
+      "step": 5500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6298709594991694e-05,
+      "loss": 1.1718,
+      "step": 5510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.628593330778076e-05,
+      "loss": 1.1743,
+      "step": 5520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6273157020569826e-05,
+      "loss": 1.0378,
+      "step": 5530
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6260380733358885e-05,
+      "loss": 1.0786,
+      "step": 5540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.624760444614795e-05,
+      "loss": 1.1166,
+      "step": 5550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6234828158937013e-05,
+      "loss": 1.0256,
+      "step": 5560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.622205187172608e-05,
+      "loss": 1.0558,
+      "step": 5570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.620927558451514e-05,
+      "loss": 1.1041,
+      "step": 5580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.6196499297304204e-05,
+      "loss": 1.1053,
+      "step": 5590
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.6183723010093266e-05,
+      "loss": 1.18,
+      "step": 5600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.617094672288233e-05,
+      "loss": 1.1121,
+      "step": 5610
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.6158170435671394e-05,
+      "loss": 1.1014,
+      "step": 5620
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.614539414846046e-05,
+      "loss": 1.0429,
+      "step": 5630
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.613261786124952e-05,
+      "loss": 1.0736,
+      "step": 5640
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.6119841574038584e-05,
+      "loss": 1.0831,
+      "step": 5650
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.6107065286827647e-05,
+      "loss": 1.1652,
+      "step": 5660
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.6094288999616713e-05,
+      "loss": 1.0742,
+      "step": 5670
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6081512712405775e-05,
+      "loss": 1.1007,
+      "step": 5680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6068736425194837e-05,
+      "loss": 1.1308,
+      "step": 5690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6055960137983903e-05,
+      "loss": 1.0248,
+      "step": 5700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.6043183850772965e-05,
+      "loss": 1.0279,
+      "step": 5710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.603040756356203e-05,
+      "loss": 1.1586,
+      "step": 5720
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.6017631276351094e-05,
+      "loss": 1.0694,
+      "step": 5730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.600485498914016e-05,
+      "loss": 1.1462,
+      "step": 5740
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5992078701929218e-05,
+      "loss": 1.1396,
+      "step": 5750
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.5979302414718284e-05,
+      "loss": 1.1173,
+      "step": 5760
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.5966526127507346e-05,
+      "loss": 1.1395,
+      "step": 5770
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.5953749840296412e-05,
+      "loss": 1.0256,
+      "step": 5780
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.5940973553085475e-05,
+      "loss": 1.1937,
+      "step": 5790
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.5928197265874537e-05,
+      "loss": 1.1789,
+      "step": 5800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.59154209786636e-05,
+      "loss": 1.1477,
+      "step": 5810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5902644691452665e-05,
+      "loss": 1.1234,
+      "step": 5820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5889868404241727e-05,
+      "loss": 1.1487,
+      "step": 5830
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5877092117030793e-05,
+      "loss": 1.0304,
+      "step": 5840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5864315829819852e-05,
+      "loss": 1.1853,
+      "step": 5850
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.5851539542608918e-05,
+      "loss": 1.1154,
+      "step": 5860
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.5838763255397984e-05,
+      "loss": 1.0777,
+      "step": 5870
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.5825986968187046e-05,
+      "loss": 1.0639,
+      "step": 5880
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.581321068097611e-05,
+      "loss": 1.0546,
+      "step": 5890
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.580043439376517e-05,
+      "loss": 1.0165,
+      "step": 5900
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5787658106554236e-05,
+      "loss": 1.1897,
+      "step": 5910
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.57748818193433e-05,
+      "loss": 1.1065,
+      "step": 5920
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5762105532132365e-05,
+      "loss": 1.0956,
+      "step": 5930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.5749329244921427e-05,
+      "loss": 1.0996,
+      "step": 5940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.573655295771049e-05,
+      "loss": 1.0707,
+      "step": 5950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.572377667049955e-05,
+      "loss": 1.1295,
+      "step": 5960
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.5711000383288617e-05,
+      "loss": 1.0654,
+      "step": 5970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.569822409607768e-05,
+      "loss": 1.1163,
+      "step": 5980
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5685447808866745e-05,
+      "loss": 1.1043,
+      "step": 5990
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5672671521655804e-05,
+      "loss": 1.1324,
+      "step": 6000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.565989523444487e-05,
+      "loss": 1.082,
+      "step": 6010
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5647118947233933e-05,
+      "loss": 1.1027,
+      "step": 6020
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5634342660023e-05,
+      "loss": 1.0954,
+      "step": 6030
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5621566372812064e-05,
+      "loss": 1.0809,
+      "step": 6040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5608790085601126e-05,
+      "loss": 1.0714,
+      "step": 6050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.559601379839019e-05,
+      "loss": 1.0627,
+      "step": 6060
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.558323751117925e-05,
+      "loss": 1.0128,
+      "step": 6070
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5570461223968317e-05,
+      "loss": 1.0972,
+      "step": 6080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.555768493675738e-05,
+      "loss": 1.0176,
+      "step": 6090
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5544908649546445e-05,
+      "loss": 1.0691,
+      "step": 6100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5532132362335504e-05,
+      "loss": 1.0893,
+      "step": 6110
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.551935607512457e-05,
+      "loss": 1.1509,
+      "step": 6120
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.5506579787913632e-05,
+      "loss": 1.1002,
+      "step": 6130
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.5493803500702698e-05,
+      "loss": 1.1349,
+      "step": 6140
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.548102721349176e-05,
+      "loss": 1.0991,
+      "step": 6150
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.5468250926280823e-05,
+      "loss": 1.1392,
+      "step": 6160
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.5455474639069885e-05,
+      "loss": 1.0253,
+      "step": 6170
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.544269835185895e-05,
+      "loss": 1.0333,
+      "step": 6180
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5429922064648013e-05,
+      "loss": 1.0522,
+      "step": 6190
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.541714577743708e-05,
+      "loss": 1.0827,
+      "step": 6200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.540436949022614e-05,
+      "loss": 1.0338,
+      "step": 6210
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.5391593203015204e-05,
+      "loss": 1.0513,
+      "step": 6220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.537881691580427e-05,
+      "loss": 1.0698,
+      "step": 6230
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.536604062859333e-05,
+      "loss": 1.107,
+      "step": 6240
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5353264341382397e-05,
+      "loss": 1.0761,
+      "step": 6250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5340488054171456e-05,
+      "loss": 1.1251,
+      "step": 6260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5327711766960522e-05,
+      "loss": 1.0736,
+      "step": 6270
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5314935479749584e-05,
+      "loss": 1.0833,
+      "step": 6280
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.530215919253865e-05,
+      "loss": 1.0835,
+      "step": 6290
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5289382905327713e-05,
+      "loss": 0.9721,
+      "step": 6300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5276606618116775e-05,
+      "loss": 1.047,
+      "step": 6310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5263830330905837e-05,
+      "loss": 1.1411,
+      "step": 6320
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5251054043694903e-05,
+      "loss": 1.0874,
+      "step": 6330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5238277756483965e-05,
+      "loss": 1.14,
+      "step": 6340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.522550146927303e-05,
+      "loss": 1.1002,
+      "step": 6350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5212725182062094e-05,
+      "loss": 1.0454,
+      "step": 6360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5199948894851156e-05,
+      "loss": 1.1183,
+      "step": 6370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.518717260764022e-05,
+      "loss": 1.1313,
+      "step": 6380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5174396320429284e-05,
+      "loss": 1.2169,
+      "step": 6390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.516162003321835e-05,
+      "loss": 1.0862,
+      "step": 6400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5148843746007412e-05,
+      "loss": 1.0721,
+      "step": 6410
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5136067458796475e-05,
+      "loss": 1.0286,
+      "step": 6420
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5123291171585537e-05,
+      "loss": 1.073,
+      "step": 6430
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5110514884374603e-05,
+      "loss": 1.0554,
+      "step": 6440
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5097738597163665e-05,
+      "loss": 1.004,
+      "step": 6450
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.508496230995273e-05,
+      "loss": 1.1401,
+      "step": 6460
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.507218602274179e-05,
+      "loss": 1.0613,
+      "step": 6470
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5059409735530855e-05,
+      "loss": 1.0989,
+      "step": 6480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5046633448319918e-05,
+      "loss": 1.1485,
+      "step": 6490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5033857161108984e-05,
+      "loss": 1.0263,
+      "step": 6500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5021080873898046e-05,
+      "loss": 1.0634,
+      "step": 6510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5008304586687108e-05,
+      "loss": 1.1238,
+      "step": 6520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.499552829947617e-05,
+      "loss": 1.176,
+      "step": 6530
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4982752012265236e-05,
+      "loss": 1.1438,
+      "step": 6540
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4969975725054302e-05,
+      "loss": 1.0184,
+      "step": 6550
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4957199437843365e-05,
+      "loss": 1.1115,
+      "step": 6560
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4944423150632427e-05,
+      "loss": 1.0265,
+      "step": 6570
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.493164686342149e-05,
+      "loss": 1.1034,
+      "step": 6580
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4918870576210555e-05,
+      "loss": 1.0778,
+      "step": 6590
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4906094288999617e-05,
+      "loss": 1.1707,
+      "step": 6600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4893318001788683e-05,
+      "loss": 1.0163,
+      "step": 6610
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4880541714577742e-05,
+      "loss": 1.1181,
+      "step": 6620
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4867765427366808e-05,
+      "loss": 1.0864,
+      "step": 6630
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.485498914015587e-05,
+      "loss": 1.0583,
+      "step": 6640
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4842212852944936e-05,
+      "loss": 1.0578,
+      "step": 6650
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4829436565734e-05,
+      "loss": 1.173,
+      "step": 6660
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.481666027852306e-05,
+      "loss": 1.0768,
+      "step": 6670
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4803883991312123e-05,
+      "loss": 1.0241,
+      "step": 6680
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.479110770410119e-05,
+      "loss": 1.1056,
+      "step": 6690
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.477833141689025e-05,
+      "loss": 1.2032,
+      "step": 6700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4765555129679317e-05,
+      "loss": 1.0829,
+      "step": 6710
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4752778842468383e-05,
+      "loss": 1.1422,
+      "step": 6720
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.474000255525744e-05,
+      "loss": 1.0895,
+      "step": 6730
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4727226268046507e-05,
+      "loss": 0.9601,
+      "step": 6740
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.471444998083557e-05,
+      "loss": 1.0642,
+      "step": 6750
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4701673693624635e-05,
+      "loss": 1.1043,
+      "step": 6760
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4688897406413698e-05,
+      "loss": 1.0912,
+      "step": 6770
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.467612111920276e-05,
+      "loss": 1.0568,
+      "step": 6780
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4663344831991823e-05,
+      "loss": 0.9949,
+      "step": 6790
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.465056854478089e-05,
+      "loss": 1.1519,
+      "step": 6800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.463779225756995e-05,
+      "loss": 1.0177,
+      "step": 6810
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4625015970359016e-05,
+      "loss": 0.9917,
+      "step": 6820
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4612239683148075e-05,
+      "loss": 1.0757,
+      "step": 6830
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.459946339593714e-05,
+      "loss": 1.0863,
+      "step": 6840
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4586687108726204e-05,
+      "loss": 1.0913,
+      "step": 6850
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.457391082151527e-05,
+      "loss": 1.1372,
+      "step": 6860
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.456113453430433e-05,
+      "loss": 0.9983,
+      "step": 6870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4548358247093394e-05,
+      "loss": 1.0834,
+      "step": 6880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4535581959882456e-05,
+      "loss": 1.0411,
+      "step": 6890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4522805672671522e-05,
+      "loss": 1.0315,
+      "step": 6900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4510029385460588e-05,
+      "loss": 1.0517,
+      "step": 6910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.449725309824965e-05,
+      "loss": 1.1103,
+      "step": 6920
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.4484476811038713e-05,
+      "loss": 1.0671,
+      "step": 6930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.4471700523827775e-05,
+      "loss": 0.9529,
+      "step": 6940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.445892423661684e-05,
+      "loss": 1.017,
+      "step": 6950
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.4446147949405903e-05,
+      "loss": 1.1169,
+      "step": 6960
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.443337166219497e-05,
+      "loss": 0.9866,
+      "step": 6970
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4420595374984028e-05,
+      "loss": 1.0956,
+      "step": 6980
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4407819087773094e-05,
+      "loss": 1.0235,
+      "step": 6990
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4395042800562156e-05,
+      "loss": 1.0232,
+      "step": 7000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.438226651335122e-05,
+      "loss": 1.0942,
+      "step": 7010
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.4369490226140284e-05,
+      "loss": 1.0338,
+      "step": 7020
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.4356713938929346e-05,
+      "loss": 1.1155,
+      "step": 7030
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.434393765171841e-05,
+      "loss": 1.0933,
+      "step": 7040
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.4331161364507474e-05,
+      "loss": 1.0331,
+      "step": 7050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.4318385077296537e-05,
+      "loss": 1.0647,
+      "step": 7060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.4305608790085603e-05,
+      "loss": 1.0587,
+      "step": 7070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.429283250287467e-05,
+      "loss": 1.0124,
+      "step": 7080
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.4280056215663727e-05,
+      "loss": 1.0816,
+      "step": 7090
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.4267279928452793e-05,
+      "loss": 1.0868,
+      "step": 7100
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4254503641241855e-05,
+      "loss": 1.1505,
+      "step": 7110
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.424172735403092e-05,
+      "loss": 1.077,
+      "step": 7120
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4228951066819984e-05,
+      "loss": 1.1635,
+      "step": 7130
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.4216174779609046e-05,
+      "loss": 1.1038,
+      "step": 7140
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.4203398492398108e-05,
+      "loss": 1.0356,
+      "step": 7150
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.4190622205187174e-05,
+      "loss": 1.1225,
+      "step": 7160
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4177845917976236e-05,
+      "loss": 1.067,
+      "step": 7170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4165069630765302e-05,
+      "loss": 1.0761,
+      "step": 7180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.415229334355436e-05,
+      "loss": 1.0828,
+      "step": 7190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4139517056343427e-05,
+      "loss": 1.0658,
+      "step": 7200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.412674076913249e-05,
+      "loss": 1.0643,
+      "step": 7210
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4113964481921555e-05,
+      "loss": 1.0683,
+      "step": 7220
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.4101188194710617e-05,
+      "loss": 1.0343,
+      "step": 7230
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.408841190749968e-05,
+      "loss": 1.0698,
+      "step": 7240
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4075635620288745e-05,
+      "loss": 1.1194,
+      "step": 7250
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4062859333077808e-05,
+      "loss": 1.0786,
+      "step": 7260
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4050083045866874e-05,
+      "loss": 1.0877,
+      "step": 7270
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4037306758655936e-05,
+      "loss": 1.0789,
+      "step": 7280
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4024530471444998e-05,
+      "loss": 1.0631,
+      "step": 7290
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.401175418423406e-05,
+      "loss": 1.0943,
+      "step": 7300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3998977897023126e-05,
+      "loss": 1.082,
+      "step": 7310
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.398620160981219e-05,
+      "loss": 1.1354,
+      "step": 7320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.3973425322601255e-05,
+      "loss": 1.0272,
+      "step": 7330
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.3960649035390313e-05,
+      "loss": 1.0843,
+      "step": 7340
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.394787274817938e-05,
+      "loss": 1.0707,
+      "step": 7350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.393509646096844e-05,
+      "loss": 1.1286,
+      "step": 7360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.3922320173757507e-05,
+      "loss": 1.0692,
+      "step": 7370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.390954388654657e-05,
+      "loss": 1.1927,
+      "step": 7380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.3896767599335635e-05,
+      "loss": 1.0788,
+      "step": 7390
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.3883991312124694e-05,
+      "loss": 1.0983,
+      "step": 7400
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.387121502491376e-05,
+      "loss": 1.135,
+      "step": 7410
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.3858438737702826e-05,
+      "loss": 1.149,
+      "step": 7420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.384566245049189e-05,
+      "loss": 1.1294,
+      "step": 7430
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.3832886163280954e-05,
+      "loss": 1.041,
+      "step": 7440
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3820109876070013e-05,
+      "loss": 1.0936,
+      "step": 7450
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.380733358885908e-05,
+      "loss": 1.041,
+      "step": 7460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.379455730164814e-05,
+      "loss": 1.0955,
+      "step": 7470
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3781781014437207e-05,
+      "loss": 1.0294,
+      "step": 7480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.376900472722627e-05,
+      "loss": 1.1155,
+      "step": 7490
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.375622844001533e-05,
+      "loss": 1.0319,
+      "step": 7500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.3743452152804394e-05,
+      "loss": 1.0478,
+      "step": 7510
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.373067586559346e-05,
+      "loss": 1.0321,
+      "step": 7520
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.3717899578382522e-05,
+      "loss": 1.1729,
+      "step": 7530
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.3705123291171588e-05,
+      "loss": 1.0052,
+      "step": 7540
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.3692347003960647e-05,
+      "loss": 1.0182,
+      "step": 7550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.3679570716749713e-05,
+      "loss": 0.9899,
+      "step": 7560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.3666794429538775e-05,
+      "loss": 1.042,
+      "step": 7570
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.365401814232784e-05,
+      "loss": 1.017,
+      "step": 7580
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.3641241855116906e-05,
+      "loss": 1.0711,
+      "step": 7590
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.3628465567905965e-05,
+      "loss": 1.0517,
+      "step": 7600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.361568928069503e-05,
+      "loss": 1.1192,
+      "step": 7610
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.3602912993484094e-05,
+      "loss": 0.9748,
+      "step": 7620
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.359013670627316e-05,
+      "loss": 1.1557,
+      "step": 7630
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.357736041906222e-05,
+      "loss": 1.0267,
+      "step": 7640
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3564584131851284e-05,
+      "loss": 1.023,
+      "step": 7650
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3551807844640346e-05,
+      "loss": 1.0621,
+      "step": 7660
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3539031557429412e-05,
+      "loss": 1.1223,
+      "step": 7670
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3526255270218474e-05,
+      "loss": 1.0851,
+      "step": 7680
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.351347898300754e-05,
+      "loss": 1.002,
+      "step": 7690
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.3500702695796603e-05,
+      "loss": 0.983,
+      "step": 7700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.3487926408585665e-05,
+      "loss": 1.1106,
+      "step": 7710
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.3475150121374727e-05,
+      "loss": 1.0228,
+      "step": 7720
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.3462373834163793e-05,
+      "loss": 1.0445,
+      "step": 7730
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3449597546952855e-05,
+      "loss": 1.0605,
+      "step": 7740
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.343682125974192e-05,
+      "loss": 1.1248,
+      "step": 7750
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3424044972530984e-05,
+      "loss": 1.0091,
+      "step": 7760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3411268685320046e-05,
+      "loss": 1.0451,
+      "step": 7770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.339849239810911e-05,
+      "loss": 1.0389,
+      "step": 7780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3385716110898174e-05,
+      "loss": 0.9988,
+      "step": 7790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.337293982368724e-05,
+      "loss": 1.0669,
+      "step": 7800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.33601635364763e-05,
+      "loss": 1.0262,
+      "step": 7810
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.3347387249265364e-05,
+      "loss": 1.0755,
+      "step": 7820
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.6825096423352418,
+      "eval_loss": 0.8937900066375732,
+      "eval_runtime": 301.1324,
+      "eval_samples_per_second": 138.62,
+      "eval_steps_per_second": 69.312,
+      "step": 7827
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.3334610962054427e-05,
+      "loss": 1.0878,
+      "step": 7830
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.3321834674843493e-05,
+      "loss": 1.025,
+      "step": 7840
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.3309058387632555e-05,
+      "loss": 1.0359,
+      "step": 7850
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.3296282100421617e-05,
+      "loss": 1.0187,
+      "step": 7860
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.328350581321068e-05,
+      "loss": 1.0652,
+      "step": 7870
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.3270729525999745e-05,
+      "loss": 1.018,
+      "step": 7880
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.3257953238788808e-05,
+      "loss": 1.0593,
+      "step": 7890
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.3245176951577874e-05,
+      "loss": 0.9571,
+      "step": 7900
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.3232400664366933e-05,
+      "loss": 0.9714,
+      "step": 7910
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.3219624377155998e-05,
+      "loss": 1.0267,
+      "step": 7920
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.3206848089945064e-05,
+      "loss": 1.0624,
+      "step": 7930
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.3194071802734126e-05,
+      "loss": 1.0104,
+      "step": 7940
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.3181295515523192e-05,
+      "loss": 1.0656,
+      "step": 7950
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.316851922831225e-05,
+      "loss": 0.9538,
+      "step": 7960
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.3155742941101317e-05,
+      "loss": 1.0332,
+      "step": 7970
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.314296665389038e-05,
+      "loss": 0.9867,
+      "step": 7980
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.3130190366679445e-05,
+      "loss": 1.0243,
+      "step": 7990
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3117414079468507e-05,
+      "loss": 1.0271,
+      "step": 8000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.310463779225757e-05,
+      "loss": 1.013,
+      "step": 8010
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3091861505046632e-05,
+      "loss": 1.0212,
+      "step": 8020
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3079085217835698e-05,
+      "loss": 1.0156,
+      "step": 8030
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.306630893062476e-05,
+      "loss": 1.0226,
+      "step": 8040
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.3053532643413826e-05,
+      "loss": 0.954,
+      "step": 8050
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.3040756356202888e-05,
+      "loss": 1.031,
+      "step": 8060
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.302798006899195e-05,
+      "loss": 1.0905,
+      "step": 8070
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.3015203781781013e-05,
+      "loss": 1.0685,
+      "step": 8080
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.300242749457008e-05,
+      "loss": 1.0371,
+      "step": 8090
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2989651207359145e-05,
+      "loss": 1.1171,
+      "step": 8100
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.2976874920148207e-05,
+      "loss": 1.0627,
+      "step": 8110
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.296409863293727e-05,
+      "loss": 1.0545,
+      "step": 8120
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.295132234572633e-05,
+      "loss": 1.0362,
+      "step": 8130
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.2938546058515397e-05,
+      "loss": 1.0701,
+      "step": 8140
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.292576977130446e-05,
+      "loss": 0.961,
+      "step": 8150
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.2912993484093525e-05,
+      "loss": 1.0286,
+      "step": 8160
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.2900217196882584e-05,
+      "loss": 1.0491,
+      "step": 8170
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.288744090967165e-05,
+      "loss": 1.1052,
+      "step": 8180
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.2874664622460713e-05,
+      "loss": 1.0902,
+      "step": 8190
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.286188833524978e-05,
+      "loss": 1.0374,
+      "step": 8200
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.284911204803884e-05,
+      "loss": 0.9469,
+      "step": 8210
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2836335760827903e-05,
+      "loss": 1.0683,
+      "step": 8220
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2823559473616965e-05,
+      "loss": 1.0952,
+      "step": 8230
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.281078318640603e-05,
+      "loss": 1.0811,
+      "step": 8240
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2798006899195094e-05,
+      "loss": 1.0438,
+      "step": 8250
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.278523061198416e-05,
+      "loss": 0.983,
+      "step": 8260
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.277245432477322e-05,
+      "loss": 1.0498,
+      "step": 8270
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2759678037562284e-05,
+      "loss": 0.9876,
+      "step": 8280
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.274690175035135e-05,
+      "loss": 1.0483,
+      "step": 8290
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.2734125463140412e-05,
+      "loss": 1.0201,
+      "step": 8300
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.2721349175929478e-05,
+      "loss": 1.0603,
+      "step": 8310
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.2708572888718537e-05,
+      "loss": 0.98,
+      "step": 8320
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.2695796601507603e-05,
+      "loss": 1.0186,
+      "step": 8330
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.2683020314296665e-05,
+      "loss": 1.0364,
+      "step": 8340
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.267024402708573e-05,
+      "loss": 0.9915,
+      "step": 8350
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.2657467739874793e-05,
+      "loss": 0.9961,
+      "step": 8360
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.2644691452663855e-05,
+      "loss": 1.0486,
+      "step": 8370
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.2631915165452918e-05,
+      "loss": 1.1112,
+      "step": 8380
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.2619138878241984e-05,
+      "loss": 1.0118,
+      "step": 8390
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.2606362591031046e-05,
+      "loss": 1.0305,
+      "step": 8400
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.259358630382011e-05,
+      "loss": 1.0391,
+      "step": 8410
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.2580810016609174e-05,
+      "loss": 1.0419,
+      "step": 8420
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.2568033729398236e-05,
+      "loss": 1.0441,
+      "step": 8430
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.2555257442187302e-05,
+      "loss": 1.014,
+      "step": 8440
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2542481154976364e-05,
+      "loss": 0.9706,
+      "step": 8450
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.252970486776543e-05,
+      "loss": 1.125,
+      "step": 8460
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.2516928580554493e-05,
+      "loss": 1.0316,
+      "step": 8470
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.2504152293343555e-05,
+      "loss": 1.0275,
+      "step": 8480
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.2491376006132617e-05,
+      "loss": 1.1037,
+      "step": 8490
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.2478599718921683e-05,
+      "loss": 1.0732,
+      "step": 8500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.2465823431710745e-05,
+      "loss": 1.1062,
+      "step": 8510
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.245304714449981e-05,
+      "loss": 1.0378,
+      "step": 8520
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.244027085728887e-05,
+      "loss": 1.1277,
+      "step": 8530
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.2427494570077936e-05,
+      "loss": 0.9765,
+      "step": 8540
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.2414718282866998e-05,
+      "loss": 1.0515,
+      "step": 8550
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.2401941995656064e-05,
+      "loss": 1.0798,
+      "step": 8560
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.2389165708445126e-05,
+      "loss": 0.9839,
+      "step": 8570
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.237638942123419e-05,
+      "loss": 1.0819,
+      "step": 8580
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.236361313402325e-05,
+      "loss": 0.9844,
+      "step": 8590
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.2350836846812317e-05,
+      "loss": 1.0126,
+      "step": 8600
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.2338060559601383e-05,
+      "loss": 1.0232,
+      "step": 8610
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.2325284272390445e-05,
+      "loss": 1.0272,
+      "step": 8620
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.2312507985179507e-05,
+      "loss": 1.0461,
+      "step": 8630
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.229973169796857e-05,
+      "loss": 1.064,
+      "step": 8640
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.2286955410757635e-05,
+      "loss": 1.0858,
+      "step": 8650
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.2274179123546698e-05,
+      "loss": 1.0288,
+      "step": 8660
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.2261402836335764e-05,
+      "loss": 0.9986,
+      "step": 8670
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.2248626549124823e-05,
+      "loss": 1.0781,
+      "step": 8680
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.2235850261913888e-05,
+      "loss": 1.0184,
+      "step": 8690
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.222307397470295e-05,
+      "loss": 0.9895,
+      "step": 8700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.2210297687492016e-05,
+      "loss": 1.0041,
+      "step": 8710
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.219752140028108e-05,
+      "loss": 1.0071,
+      "step": 8720
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.2184745113070145e-05,
+      "loss": 1.0364,
+      "step": 8730
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.2171968825859203e-05,
+      "loss": 1.0183,
+      "step": 8740
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.215919253864827e-05,
+      "loss": 0.9923,
+      "step": 8750
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.214641625143733e-05,
+      "loss": 1.0698,
+      "step": 8760
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.2133639964226397e-05,
+      "loss": 1.0158,
+      "step": 8770
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.2120863677015463e-05,
+      "loss": 0.9953,
+      "step": 8780
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.2108087389804522e-05,
+      "loss": 1.0502,
+      "step": 8790
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.2095311102593588e-05,
+      "loss": 0.9623,
+      "step": 8800
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.208253481538265e-05,
+      "loss": 0.924,
+      "step": 8810
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.2069758528171716e-05,
+      "loss": 1.0329,
+      "step": 8820
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.205698224096078e-05,
+      "loss": 1.0636,
+      "step": 8830
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.204420595374984e-05,
+      "loss": 1.025,
+      "step": 8840
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.2031429666538903e-05,
+      "loss": 1.0375,
+      "step": 8850
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.201865337932797e-05,
+      "loss": 1.0329,
+      "step": 8860
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.200587709211703e-05,
+      "loss": 1.0646,
+      "step": 8870
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.1993100804906097e-05,
+      "loss": 1.0791,
+      "step": 8880
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.1980324517695156e-05,
+      "loss": 1.0558,
+      "step": 8890
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.196754823048422e-05,
+      "loss": 0.9135,
+      "step": 8900
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.1954771943273284e-05,
+      "loss": 1.0741,
+      "step": 8910
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.194199565606235e-05,
+      "loss": 1.0717,
+      "step": 8920
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.1929219368851412e-05,
+      "loss": 1.0419,
+      "step": 8930
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.1916443081640474e-05,
+      "loss": 1.0776,
+      "step": 8940
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.1903666794429537e-05,
+      "loss": 1.0317,
+      "step": 8950
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.1890890507218603e-05,
+      "loss": 1.097,
+      "step": 8960
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.187811422000767e-05,
+      "loss": 1.0727,
+      "step": 8970
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.186533793279673e-05,
+      "loss": 0.9845,
+      "step": 8980
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.1852561645585793e-05,
+      "loss": 0.9579,
+      "step": 8990
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.1839785358374855e-05,
+      "loss": 0.9434,
+      "step": 9000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.182700907116392e-05,
+      "loss": 1.0293,
+      "step": 9010
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.1814232783952984e-05,
+      "loss": 1.0813,
+      "step": 9020
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.180145649674205e-05,
+      "loss": 1.0201,
+      "step": 9030
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.178868020953111e-05,
+      "loss": 1.0174,
+      "step": 9040
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.1775903922320174e-05,
+      "loss": 1.0147,
+      "step": 9050
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.1763127635109236e-05,
+      "loss": 1.078,
+      "step": 9060
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.1750351347898302e-05,
+      "loss": 1.0205,
+      "step": 9070
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.1737575060687364e-05,
+      "loss": 1.1099,
+      "step": 9080
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.172479877347643e-05,
+      "loss": 1.0465,
+      "step": 9090
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.171202248626549e-05,
+      "loss": 1.0107,
+      "step": 9100
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.1699246199054555e-05,
+      "loss": 0.99,
+      "step": 9110
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.1686469911843617e-05,
+      "loss": 1.0071,
+      "step": 9120
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.1673693624632683e-05,
+      "loss": 0.997,
+      "step": 9130
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.166091733742175e-05,
+      "loss": 1.0344,
+      "step": 9140
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.1648141050210808e-05,
+      "loss": 1.0395,
+      "step": 9150
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.1635364762999874e-05,
+      "loss": 0.9957,
+      "step": 9160
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.1622588475788936e-05,
+      "loss": 0.9928,
+      "step": 9170
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.1609812188578e-05,
+      "loss": 1.0512,
+      "step": 9180
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.1597035901367064e-05,
+      "loss": 1.0085,
+      "step": 9190
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.1584259614156126e-05,
+      "loss": 1.021,
+      "step": 9200
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.157148332694519e-05,
+      "loss": 1.1114,
+      "step": 9210
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.1558707039734254e-05,
+      "loss": 1.053,
+      "step": 9220
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.1545930752523317e-05,
+      "loss": 0.959,
+      "step": 9230
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.1533154465312383e-05,
+      "loss": 1.0537,
+      "step": 9240
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.152037817810144e-05,
+      "loss": 1.1059,
+      "step": 9250
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.1507601890890507e-05,
+      "loss": 1.0334,
+      "step": 9260
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.149482560367957e-05,
+      "loss": 0.945,
+      "step": 9270
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.1482049316468635e-05,
+      "loss": 1.0205,
+      "step": 9280
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.1469273029257698e-05,
+      "loss": 0.8946,
+      "step": 9290
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.145649674204676e-05,
+      "loss": 1.1158,
+      "step": 9300
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.1443720454835826e-05,
+      "loss": 1.0037,
+      "step": 9310
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.1430944167624888e-05,
+      "loss": 1.0036,
+      "step": 9320
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.1418167880413954e-05,
+      "loss": 1.0294,
+      "step": 9330
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.1405391593203016e-05,
+      "loss": 0.9538,
+      "step": 9340
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.139261530599208e-05,
+      "loss": 1.0533,
+      "step": 9350
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.137983901878114e-05,
+      "loss": 1.0851,
+      "step": 9360
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.1367062731570207e-05,
+      "loss": 1.0317,
+      "step": 9370
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.135428644435927e-05,
+      "loss": 1.0627,
+      "step": 9380
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.1341510157148335e-05,
+      "loss": 1.0275,
+      "step": 9390
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.1328733869937397e-05,
+      "loss": 1.034,
+      "step": 9400
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.131595758272646e-05,
+      "loss": 1.0159,
+      "step": 9410
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.1303181295515522e-05,
+      "loss": 0.9994,
+      "step": 9420
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.1290405008304588e-05,
+      "loss": 1.0029,
+      "step": 9430
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.127762872109365e-05,
+      "loss": 1.054,
+      "step": 9440
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.1264852433882716e-05,
+      "loss": 1.0247,
+      "step": 9450
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.1252076146671775e-05,
+      "loss": 1.101,
+      "step": 9460
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.123929985946084e-05,
+      "loss": 1.0384,
+      "step": 9470
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.1226523572249906e-05,
+      "loss": 1.0056,
+      "step": 9480
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.121374728503897e-05,
+      "loss": 1.0253,
+      "step": 9490
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.1200970997828035e-05,
+      "loss": 1.0022,
+      "step": 9500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.1188194710617094e-05,
+      "loss": 1.0184,
+      "step": 9510
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.117541842340616e-05,
+      "loss": 1.084,
+      "step": 9520
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.116264213619522e-05,
+      "loss": 1.0293,
+      "step": 9530
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.1149865848984287e-05,
+      "loss": 1.0071,
+      "step": 9540
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.113708956177335e-05,
+      "loss": 1.0419,
+      "step": 9550
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.1124313274562412e-05,
+      "loss": 1.025,
+      "step": 9560
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.1111536987351474e-05,
+      "loss": 1.064,
+      "step": 9570
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.109876070014054e-05,
+      "loss": 1.077,
+      "step": 9580
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.1085984412929603e-05,
+      "loss": 1.0143,
+      "step": 9590
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.107320812571867e-05,
+      "loss": 1.0703,
+      "step": 9600
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.1060431838507727e-05,
+      "loss": 1.0376,
+      "step": 9610
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.1047655551296793e-05,
+      "loss": 1.0564,
+      "step": 9620
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.1034879264085855e-05,
+      "loss": 1.0056,
+      "step": 9630
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.102210297687492e-05,
+      "loss": 1.0423,
+      "step": 9640
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.1010604318385078e-05,
+      "loss": 0.9948,
+      "step": 9650
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.099782803117414e-05,
+      "loss": 0.9852,
+      "step": 9660
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0985051743963206e-05,
+      "loss": 0.9679,
+      "step": 9670
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.097227545675227e-05,
+      "loss": 1.0069,
+      "step": 9680
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.095949916954133e-05,
+      "loss": 1.0064,
+      "step": 9690
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.0946722882330397e-05,
+      "loss": 1.0177,
+      "step": 9700
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.093394659511946e-05,
+      "loss": 1.0382,
+      "step": 9710
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.0921170307908525e-05,
+      "loss": 1.0895,
+      "step": 9720
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.0908394020697587e-05,
+      "loss": 0.946,
+      "step": 9730
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.089561773348665e-05,
+      "loss": 1.0327,
+      "step": 9740
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.0882841446275712e-05,
+      "loss": 1.0571,
+      "step": 9750
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.0870065159064777e-05,
+      "loss": 0.9581,
+      "step": 9760
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.085728887185384e-05,
+      "loss": 0.9532,
+      "step": 9770
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0844512584642906e-05,
+      "loss": 1.0014,
+      "step": 9780
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0831736297431965e-05,
+      "loss": 0.9949,
+      "step": 9790
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.081896001022103e-05,
+      "loss": 1.0254,
+      "step": 9800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.0806183723010093e-05,
+      "loss": 1.0585,
+      "step": 9810
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.079340743579916e-05,
+      "loss": 1.0318,
+      "step": 9820
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.078063114858822e-05,
+      "loss": 0.9836,
+      "step": 9830
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.0767854861377283e-05,
+      "loss": 0.9845,
+      "step": 9840
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.0755078574166346e-05,
+      "loss": 1.0269,
+      "step": 9850
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.074230228695541e-05,
+      "loss": 0.9764,
+      "step": 9860
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.0729525999744477e-05,
+      "loss": 1.004,
+      "step": 9870
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.071674971253354e-05,
+      "loss": 1.0853,
+      "step": 9880
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.0703973425322602e-05,
+      "loss": 1.0041,
+      "step": 9890
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.0691197138111664e-05,
+      "loss": 1.076,
+      "step": 9900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.067842085090073e-05,
+      "loss": 1.0294,
+      "step": 9910
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.0665644563689792e-05,
+      "loss": 1.0616,
+      "step": 9920
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.0652868276478858e-05,
+      "loss": 1.0558,
+      "step": 9930
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.0640091989267917e-05,
+      "loss": 1.0142,
+      "step": 9940
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.0627315702056983e-05,
+      "loss": 0.965,
+      "step": 9950
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.0614539414846045e-05,
+      "loss": 0.9732,
+      "step": 9960
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.060176312763511e-05,
+      "loss": 1.0222,
+      "step": 9970
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.0588986840424173e-05,
+      "loss": 1.0496,
+      "step": 9980
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.0576210553213236e-05,
+      "loss": 0.9679,
+      "step": 9990
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.0563434266002298e-05,
+      "loss": 1.0011,
+      "step": 10000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.0550657978791364e-05,
+      "loss": 1.0534,
+      "step": 10010
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.0537881691580426e-05,
+      "loss": 0.972,
+      "step": 10020
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.0525105404369492e-05,
+      "loss": 1.0555,
+      "step": 10030
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0512329117158558e-05,
+      "loss": 1.0325,
+      "step": 10040
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0499552829947617e-05,
+      "loss": 0.9442,
+      "step": 10050
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.0486776542736682e-05,
+      "loss": 0.9743,
+      "step": 10060
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.0474000255525745e-05,
+      "loss": 1.0768,
+      "step": 10070
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.046122396831481e-05,
+      "loss": 0.943,
+      "step": 10080
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.0448447681103873e-05,
+      "loss": 1.055,
+      "step": 10090
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.0435671393892935e-05,
+      "loss": 0.968,
+      "step": 10100
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.0422895106681997e-05,
+      "loss": 1.0523,
+      "step": 10110
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.0410118819471063e-05,
+      "loss": 1.0725,
+      "step": 10120
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.0397342532260126e-05,
+      "loss": 1.0593,
+      "step": 10130
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.038456624504919e-05,
+      "loss": 0.8884,
+      "step": 10140
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.037178995783825e-05,
+      "loss": 0.9884,
+      "step": 10150
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.0359013670627316e-05,
+      "loss": 1.0388,
+      "step": 10160
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.034623738341638e-05,
+      "loss": 1.0385,
+      "step": 10170
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.0333461096205444e-05,
+      "loss": 0.9961,
+      "step": 10180
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.0320684808994507e-05,
+      "loss": 0.986,
+      "step": 10190
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.030790852178357e-05,
+      "loss": 1.0116,
+      "step": 10200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.0295132234572635e-05,
+      "loss": 0.9868,
+      "step": 10210
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.0282355947361697e-05,
+      "loss": 1.0432,
+      "step": 10220
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.0269579660150763e-05,
+      "loss": 0.9453,
+      "step": 10230
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.0256803372939825e-05,
+      "loss": 1.0286,
+      "step": 10240
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.0244027085728887e-05,
+      "loss": 1.0106,
+      "step": 10250
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.023125079851795e-05,
+      "loss": 1.0025,
+      "step": 10260
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.0218474511307016e-05,
+      "loss": 1.02,
+      "step": 10270
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.0205698224096078e-05,
+      "loss": 0.9797,
+      "step": 10280
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.0192921936885144e-05,
+      "loss": 1.0164,
+      "step": 10290
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.0180145649674203e-05,
+      "loss": 1.0135,
+      "step": 10300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.016736936246327e-05,
+      "loss": 0.9658,
+      "step": 10310
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.015459307525233e-05,
+      "loss": 0.9884,
+      "step": 10320
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.0141816788041397e-05,
+      "loss": 0.9869,
+      "step": 10330
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.012904050083046e-05,
+      "loss": 1.0391,
+      "step": 10340
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.011626421361952e-05,
+      "loss": 0.9192,
+      "step": 10350
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.0103487926408584e-05,
+      "loss": 1.0853,
+      "step": 10360
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.009071163919765e-05,
+      "loss": 0.9712,
+      "step": 10370
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0077935351986715e-05,
+      "loss": 1.0299,
+      "step": 10380
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0065159064775777e-05,
+      "loss": 1.0221,
+      "step": 10390
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0052382777564843e-05,
+      "loss": 1.0338,
+      "step": 10400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0039606490353902e-05,
+      "loss": 0.9901,
+      "step": 10410
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0026830203142968e-05,
+      "loss": 0.9988,
+      "step": 10420
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.001405391593203e-05,
+      "loss": 1.0048,
+      "step": 10430
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.7040461873847113,
+      "eval_loss": 0.8600494265556335,
+      "eval_runtime": 306.4466,
+      "eval_samples_per_second": 136.216,
+      "eval_steps_per_second": 68.11,
+      "step": 10436
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0001277628721096e-05,
+      "loss": 0.9459,
+      "step": 10440
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.998850134151016e-05,
+      "loss": 0.9764,
+      "step": 10450
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.997572505429922e-05,
+      "loss": 1.0754,
+      "step": 10460
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9962948767088283e-05,
+      "loss": 1.0515,
+      "step": 10470
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.995017247987735e-05,
+      "loss": 1.0065,
+      "step": 10480
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.993739619266641e-05,
+      "loss": 1.0253,
+      "step": 10490
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9924619905455477e-05,
+      "loss": 0.9243,
+      "step": 10500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9911843618244536e-05,
+      "loss": 0.9527,
+      "step": 10510
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9899067331033602e-05,
+      "loss": 1.0599,
+      "step": 10520
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9886291043822664e-05,
+      "loss": 0.9384,
+      "step": 10530
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.987351475661173e-05,
+      "loss": 1.0183,
+      "step": 10540
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9860738469400796e-05,
+      "loss": 1.0682,
+      "step": 10550
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.9847962182189855e-05,
+      "loss": 1.0485,
+      "step": 10560
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.983518589497892e-05,
+      "loss": 1.0225,
+      "step": 10570
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.9822409607767983e-05,
+      "loss": 0.9636,
+      "step": 10580
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.980963332055705e-05,
+      "loss": 0.9476,
+      "step": 10590
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.979685703334611e-05,
+      "loss": 0.9997,
+      "step": 10600
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.9784080746135173e-05,
+      "loss": 1.0073,
+      "step": 10610
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.9771304458924236e-05,
+      "loss": 0.9891,
+      "step": 10620
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.97585281717133e-05,
+      "loss": 1.0127,
+      "step": 10630
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.9745751884502364e-05,
+      "loss": 0.9135,
+      "step": 10640
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.973297559729143e-05,
+      "loss": 0.9873,
+      "step": 10650
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.972019931008049e-05,
+      "loss": 1.0475,
+      "step": 10660
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.9707423022869554e-05,
+      "loss": 1.0128,
+      "step": 10670
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.9694646735658616e-05,
+      "loss": 0.9845,
+      "step": 10680
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.9681870448447682e-05,
+      "loss": 0.9993,
+      "step": 10690
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.9669094161236745e-05,
+      "loss": 0.9962,
+      "step": 10700
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.965631787402581e-05,
+      "loss": 1.017,
+      "step": 10710
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.964354158681487e-05,
+      "loss": 0.9382,
+      "step": 10720
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.9630765299603935e-05,
+      "loss": 1.0317,
+      "step": 10730
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.9617989012393e-05,
+      "loss": 1.0399,
+      "step": 10740
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.9605212725182063e-05,
+      "loss": 1.0362,
+      "step": 10750
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.959243643797113e-05,
+      "loss": 0.9808,
+      "step": 10760
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.9579660150760188e-05,
+      "loss": 0.9382,
+      "step": 10770
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.9566883863549254e-05,
+      "loss": 0.983,
+      "step": 10780
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.9554107576338316e-05,
+      "loss": 0.9385,
+      "step": 10790
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.9541331289127382e-05,
+      "loss": 0.9472,
+      "step": 10800
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.9528555001916444e-05,
+      "loss": 1.0646,
+      "step": 10810
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.9515778714705507e-05,
+      "loss": 0.9611,
+      "step": 10820
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.950300242749457e-05,
+      "loss": 1.0051,
+      "step": 10830
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.9490226140283635e-05,
+      "loss": 0.9689,
+      "step": 10840
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.9477449853072697e-05,
+      "loss": 1.0728,
+      "step": 10850
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.9464673565861763e-05,
+      "loss": 0.9823,
+      "step": 10860
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.9451897278650822e-05,
+      "loss": 0.9576,
+      "step": 10870
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.9439120991439887e-05,
+      "loss": 1.0259,
+      "step": 10880
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.942634470422895e-05,
+      "loss": 1.0078,
+      "step": 10890
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9413568417018016e-05,
+      "loss": 0.9973,
+      "step": 10900
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.940079212980708e-05,
+      "loss": 1.0395,
+      "step": 10910
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.938801584259614e-05,
+      "loss": 1.0008,
+      "step": 10920
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9375239555385206e-05,
+      "loss": 1.0114,
+      "step": 10930
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.936246326817427e-05,
+      "loss": 0.9611,
+      "step": 10940
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9349686980963334e-05,
+      "loss": 0.9867,
+      "step": 10950
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9336910693752397e-05,
+      "loss": 0.9812,
+      "step": 10960
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.932413440654146e-05,
+      "loss": 1.0514,
+      "step": 10970
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.931135811933052e-05,
+      "loss": 1.0008,
+      "step": 10980
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.9298581832119587e-05,
+      "loss": 1.1416,
+      "step": 10990
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.928580554490865e-05,
+      "loss": 0.9546,
+      "step": 11000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.9273029257697715e-05,
+      "loss": 0.9935,
+      "step": 11010
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.9260252970486777e-05,
+      "loss": 1.0199,
+      "step": 11020
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.924747668327584e-05,
+      "loss": 0.982,
+      "step": 11030
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.9234700396064902e-05,
+      "loss": 0.9565,
+      "step": 11040
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.9221924108853968e-05,
+      "loss": 1.0392,
+      "step": 11050
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.920914782164303e-05,
+      "loss": 1.0471,
+      "step": 11060
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.9196371534432096e-05,
+      "loss": 0.994,
+      "step": 11070
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.918359524722116e-05,
+      "loss": 1.0374,
+      "step": 11080
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.917081896001022e-05,
+      "loss": 0.9268,
+      "step": 11090
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.9158042672799287e-05,
+      "loss": 0.9702,
+      "step": 11100
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.914526638558835e-05,
+      "loss": 0.9774,
+      "step": 11110
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.9132490098377415e-05,
+      "loss": 0.9614,
+      "step": 11120
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.9119713811166474e-05,
+      "loss": 1.0095,
+      "step": 11130
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.910693752395554e-05,
+      "loss": 1.0305,
+      "step": 11140
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.9094161236744602e-05,
+      "loss": 0.9512,
+      "step": 11150
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.9081384949533667e-05,
+      "loss": 1.0214,
+      "step": 11160
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.906860866232273e-05,
+      "loss": 0.9975,
+      "step": 11170
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.9055832375111792e-05,
+      "loss": 0.9814,
+      "step": 11180
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.9043056087900855e-05,
+      "loss": 0.9298,
+      "step": 11190
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.903027980068992e-05,
+      "loss": 0.9918,
+      "step": 11200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.9017503513478983e-05,
+      "loss": 1.0058,
+      "step": 11210
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.900472722626805e-05,
+      "loss": 0.9774,
+      "step": 11220
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.8991950939057107e-05,
+      "loss": 0.9492,
+      "step": 11230
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.8979174651846173e-05,
+      "loss": 1.0274,
+      "step": 11240
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.896639836463524e-05,
+      "loss": 0.8601,
+      "step": 11250
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.89536220774243e-05,
+      "loss": 1.1159,
+      "step": 11260
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.8940845790213367e-05,
+      "loss": 1.0081,
+      "step": 11270
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.8928069503002426e-05,
+      "loss": 1.0421,
+      "step": 11280
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.8915293215791492e-05,
+      "loss": 1.0195,
+      "step": 11290
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.8902516928580554e-05,
+      "loss": 0.9817,
+      "step": 11300
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.888974064136962e-05,
+      "loss": 0.9258,
+      "step": 11310
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.8876964354158682e-05,
+      "loss": 1.0313,
+      "step": 11320
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.8864188066947745e-05,
+      "loss": 1.0741,
+      "step": 11330
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.8851411779736807e-05,
+      "loss": 0.9507,
+      "step": 11340
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.8838635492525873e-05,
+      "loss": 1.004,
+      "step": 11350
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.8825859205314935e-05,
+      "loss": 0.9851,
+      "step": 11360
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.8813082918104e-05,
+      "loss": 0.9581,
+      "step": 11370
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.8800306630893063e-05,
+      "loss": 1.0172,
+      "step": 11380
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.8787530343682126e-05,
+      "loss": 1.0631,
+      "step": 11390
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.8774754056471188e-05,
+      "loss": 0.9673,
+      "step": 11400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.8761977769260254e-05,
+      "loss": 0.9853,
+      "step": 11410
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.874920148204932e-05,
+      "loss": 0.9911,
+      "step": 11420
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.8736425194838382e-05,
+      "loss": 1.0252,
+      "step": 11430
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.8723648907627444e-05,
+      "loss": 0.9863,
+      "step": 11440
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.8710872620416507e-05,
+      "loss": 0.9982,
+      "step": 11450
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.8698096333205572e-05,
+      "loss": 0.9994,
+      "step": 11460
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8685320045994635e-05,
+      "loss": 0.9207,
+      "step": 11470
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.86725437587837e-05,
+      "loss": 1.0277,
+      "step": 11480
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.865976747157276e-05,
+      "loss": 1.0076,
+      "step": 11490
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8646991184361825e-05,
+      "loss": 0.9431,
+      "step": 11500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8634214897150887e-05,
+      "loss": 1.0096,
+      "step": 11510
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.8621438609939953e-05,
+      "loss": 0.9848,
+      "step": 11520
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.8608662322729016e-05,
+      "loss": 1.0283,
+      "step": 11530
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.8595886035518078e-05,
+      "loss": 0.9438,
+      "step": 11540
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.858310974830714e-05,
+      "loss": 0.954,
+      "step": 11550
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.8570333461096206e-05,
+      "loss": 0.9268,
+      "step": 11560
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.855755717388527e-05,
+      "loss": 0.9719,
+      "step": 11570
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.8544780886674334e-05,
+      "loss": 1.0252,
+      "step": 11580
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.8532004599463397e-05,
+      "loss": 0.9273,
+      "step": 11590
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.851922831225246e-05,
+      "loss": 1.0104,
+      "step": 11600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.8506452025041525e-05,
+      "loss": 1.0399,
+      "step": 11610
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.8493675737830587e-05,
+      "loss": 1.0342,
+      "step": 11620
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.8480899450619653e-05,
+      "loss": 0.9441,
+      "step": 11630
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.8468123163408712e-05,
+      "loss": 0.949,
+      "step": 11640
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.8455346876197777e-05,
+      "loss": 1.0071,
+      "step": 11650
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.844257058898684e-05,
+      "loss": 1.0028,
+      "step": 11660
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.8429794301775906e-05,
+      "loss": 1.013,
+      "step": 11670
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.8417018014564968e-05,
+      "loss": 0.983,
+      "step": 11680
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.840424172735403e-05,
+      "loss": 1.0092,
+      "step": 11690
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.8391465440143093e-05,
+      "loss": 0.9959,
+      "step": 11700
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.837868915293216e-05,
+      "loss": 0.981,
+      "step": 11710
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.836591286572122e-05,
+      "loss": 1.0389,
+      "step": 11720
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.8353136578510287e-05,
+      "loss": 0.972,
+      "step": 11730
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.834036029129935e-05,
+      "loss": 0.975,
+      "step": 11740
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.832758400408841e-05,
+      "loss": 0.9072,
+      "step": 11750
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.8314807716877477e-05,
+      "loss": 1.013,
+      "step": 11760
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.830203142966654e-05,
+      "loss": 0.9246,
+      "step": 11770
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.8289255142455605e-05,
+      "loss": 1.0591,
+      "step": 11780
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.827775648396576e-05,
+      "loss": 0.9849,
+      "step": 11790
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.8264980196754824e-05,
+      "loss": 1.026,
+      "step": 11800
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.825220390954389e-05,
+      "loss": 1.061,
+      "step": 11810
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.823942762233295e-05,
+      "loss": 1.0452,
+      "step": 11820
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8226651335122015e-05,
+      "loss": 0.9118,
+      "step": 11830
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8213875047911077e-05,
+      "loss": 0.9805,
+      "step": 11840
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8201098760700143e-05,
+      "loss": 0.9626,
+      "step": 11850
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8188322473489205e-05,
+      "loss": 1.053,
+      "step": 11860
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8175546186278268e-05,
+      "loss": 1.0135,
+      "step": 11870
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.816276989906733e-05,
+      "loss": 0.9814,
+      "step": 11880
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.8149993611856396e-05,
+      "loss": 0.9771,
+      "step": 11890
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.8137217324645458e-05,
+      "loss": 1.0723,
+      "step": 11900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.8124441037434524e-05,
+      "loss": 1.0822,
+      "step": 11910
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.8111664750223583e-05,
+      "loss": 0.9721,
+      "step": 11920
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.809888846301265e-05,
+      "loss": 1.0491,
+      "step": 11930
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.808611217580171e-05,
+      "loss": 1.035,
+      "step": 11940
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8073335888590777e-05,
+      "loss": 0.9814,
+      "step": 11950
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.806055960137984e-05,
+      "loss": 1.0619,
+      "step": 11960
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.80477833141689e-05,
+      "loss": 1.0584,
+      "step": 11970
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.8035007026957967e-05,
+      "loss": 0.8946,
+      "step": 11980
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.802223073974703e-05,
+      "loss": 0.9485,
+      "step": 11990
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.8009454452536095e-05,
+      "loss": 0.9647,
+      "step": 12000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.7996678165325158e-05,
+      "loss": 0.9895,
+      "step": 12010
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.7983901878114223e-05,
+      "loss": 0.982,
+      "step": 12020
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.7971125590903282e-05,
+      "loss": 0.9681,
+      "step": 12030
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.7958349303692348e-05,
+      "loss": 1.0043,
+      "step": 12040
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.794557301648141e-05,
+      "loss": 1.0314,
+      "step": 12050
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.7932796729270476e-05,
+      "loss": 1.0258,
+      "step": 12060
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.792002044205954e-05,
+      "loss": 1.0732,
+      "step": 12070
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.79072441548486e-05,
+      "loss": 1.0242,
+      "step": 12080
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.7894467867637663e-05,
+      "loss": 0.9567,
+      "step": 12090
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.788169158042673e-05,
+      "loss": 0.9396,
+      "step": 12100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.786891529321579e-05,
+      "loss": 0.9696,
+      "step": 12110
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.7856139006004857e-05,
+      "loss": 1.0943,
+      "step": 12120
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.7843362718793916e-05,
+      "loss": 0.9693,
+      "step": 12130
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.7830586431582982e-05,
+      "loss": 0.9157,
+      "step": 12140
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.7817810144372048e-05,
+      "loss": 0.9401,
+      "step": 12150
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.780503385716111e-05,
+      "loss": 0.9495,
+      "step": 12160
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.7792257569950176e-05,
+      "loss": 0.936,
+      "step": 12170
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.7779481282739235e-05,
+      "loss": 0.9385,
+      "step": 12180
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.77667049955283e-05,
+      "loss": 0.9262,
+      "step": 12190
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.7753928708317363e-05,
+      "loss": 1.0073,
+      "step": 12200
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.774115242110643e-05,
+      "loss": 1.0286,
+      "step": 12210
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.772837613389549e-05,
+      "loss": 0.9921,
+      "step": 12220
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.7715599846684553e-05,
+      "loss": 0.9697,
+      "step": 12230
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.7702823559473616e-05,
+      "loss": 1.0053,
+      "step": 12240
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.769004727226268e-05,
+      "loss": 1.0966,
+      "step": 12250
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.7677270985051744e-05,
+      "loss": 0.9925,
+      "step": 12260
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.766449469784081e-05,
+      "loss": 1.0647,
+      "step": 12270
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.765171841062987e-05,
+      "loss": 1.0401,
+      "step": 12280
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7638942123418934e-05,
+      "loss": 0.9786,
+      "step": 12290
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7626165836207997e-05,
+      "loss": 1.0068,
+      "step": 12300
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.7613389548997062e-05,
+      "loss": 0.999,
+      "step": 12310
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.7600613261786125e-05,
+      "loss": 1.0131,
+      "step": 12320
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.758783697457519e-05,
+      "loss": 0.9801,
+      "step": 12330
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.7575060687364253e-05,
+      "loss": 0.998,
+      "step": 12340
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.7562284400153315e-05,
+      "loss": 0.9733,
+      "step": 12350
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.754950811294238e-05,
+      "loss": 1.005,
+      "step": 12360
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.7536731825731443e-05,
+      "loss": 1.0983,
+      "step": 12370
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.752395553852051e-05,
+      "loss": 0.9449,
+      "step": 12380
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.7511179251309568e-05,
+      "loss": 0.8714,
+      "step": 12390
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.7498402964098634e-05,
+      "loss": 1.0294,
+      "step": 12400
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7485626676887696e-05,
+      "loss": 0.9287,
+      "step": 12410
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7472850389676762e-05,
+      "loss": 0.9364,
+      "step": 12420
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7460074102465824e-05,
+      "loss": 0.9208,
+      "step": 12430
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.7447297815254887e-05,
+      "loss": 0.9774,
+      "step": 12440
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.743452152804395e-05,
+      "loss": 1.0821,
+      "step": 12450
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.7421745240833015e-05,
+      "loss": 0.9437,
+      "step": 12460
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.7408968953622077e-05,
+      "loss": 0.9584,
+      "step": 12470
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.7396192666411143e-05,
+      "loss": 1.0863,
+      "step": 12480
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.7383416379200202e-05,
+      "loss": 0.9668,
+      "step": 12490
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.7370640091989268e-05,
+      "loss": 0.9899,
+      "step": 12500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.7357863804778333e-05,
+      "loss": 1.0286,
+      "step": 12510
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.7345087517567396e-05,
+      "loss": 1.0362,
+      "step": 12520
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.733231123035646e-05,
+      "loss": 0.9126,
+      "step": 12530
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.731953494314552e-05,
+      "loss": 0.9714,
+      "step": 12540
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.7306758655934586e-05,
+      "loss": 1.002,
+      "step": 12550
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.729398236872365e-05,
+      "loss": 0.9519,
+      "step": 12560
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7281206081512714e-05,
+      "loss": 0.9881,
+      "step": 12570
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7268429794301777e-05,
+      "loss": 1.0048,
+      "step": 12580
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.725565350709084e-05,
+      "loss": 0.991,
+      "step": 12590
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.72428772198799e-05,
+      "loss": 1.0719,
+      "step": 12600
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.7230100932668967e-05,
+      "loss": 0.9645,
+      "step": 12610
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.721732464545803e-05,
+      "loss": 0.9867,
+      "step": 12620
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.7204548358247095e-05,
+      "loss": 0.9421,
+      "step": 12630
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.7191772071036154e-05,
+      "loss": 0.8979,
+      "step": 12640
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.717899578382522e-05,
+      "loss": 1.0321,
+      "step": 12650
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.7166219496614282e-05,
+      "loss": 0.993,
+      "step": 12660
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.7153443209403348e-05,
+      "loss": 0.989,
+      "step": 12670
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.7140666922192414e-05,
+      "loss": 0.9692,
+      "step": 12680
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.7127890634981476e-05,
+      "loss": 1.0294,
+      "step": 12690
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.711511434777054e-05,
+      "loss": 0.8949,
+      "step": 12700
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.71023380605596e-05,
+      "loss": 0.9204,
+      "step": 12710
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7089561773348667e-05,
+      "loss": 0.9863,
+      "step": 12720
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.707678548613773e-05,
+      "loss": 0.9261,
+      "step": 12730
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7064009198926795e-05,
+      "loss": 1.0644,
+      "step": 12740
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7051232911715854e-05,
+      "loss": 0.9793,
+      "step": 12750
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.703845662450492e-05,
+      "loss": 0.9588,
+      "step": 12760
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7025680337293982e-05,
+      "loss": 1.019,
+      "step": 12770
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7012904050083048e-05,
+      "loss": 1.0621,
+      "step": 12780
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.700012776287211e-05,
+      "loss": 0.9755,
+      "step": 12790
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6987351475661172e-05,
+      "loss": 0.9292,
+      "step": 12800
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6974575188450235e-05,
+      "loss": 1.0005,
+      "step": 12810
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.69617989012393e-05,
+      "loss": 1.0226,
+      "step": 12820
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6949022614028363e-05,
+      "loss": 1.0131,
+      "step": 12830
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.693624632681743e-05,
+      "loss": 0.963,
+      "step": 12840
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.692347003960649e-05,
+      "loss": 1.0655,
+      "step": 12850
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6910693752395553e-05,
+      "loss": 0.996,
+      "step": 12860
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.689791746518462e-05,
+      "loss": 0.996,
+      "step": 12870
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.688514117797368e-05,
+      "loss": 1.0607,
+      "step": 12880
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6872364890762747e-05,
+      "loss": 0.9656,
+      "step": 12890
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6859588603551806e-05,
+      "loss": 0.9606,
+      "step": 12900
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6846812316340872e-05,
+      "loss": 0.9681,
+      "step": 12910
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6834036029129934e-05,
+      "loss": 0.9653,
+      "step": 12920
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6821259741919e-05,
+      "loss": 1.0569,
+      "step": 12930
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6808483454708062e-05,
+      "loss": 0.9666,
+      "step": 12940
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6795707167497125e-05,
+      "loss": 1.0277,
+      "step": 12950
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6782930880286187e-05,
+      "loss": 0.9886,
+      "step": 12960
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6770154593075253e-05,
+      "loss": 0.9212,
+      "step": 12970
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6757378305864315e-05,
+      "loss": 0.9798,
+      "step": 12980
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.674460201865338e-05,
+      "loss": 1.0217,
+      "step": 12990
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6731825731442443e-05,
+      "loss": 1.0463,
+      "step": 13000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6719049444231506e-05,
+      "loss": 1.0125,
+      "step": 13010
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.670627315702057e-05,
+      "loss": 0.9281,
+      "step": 13020
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6693496869809634e-05,
+      "loss": 1.02,
+      "step": 13030
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.66807205825987e-05,
+      "loss": 0.9933,
+      "step": 13040
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.7321706633447524,
+      "eval_loss": 0.7696112990379333,
+      "eval_runtime": 295.2461,
+      "eval_samples_per_second": 141.384,
+      "eval_steps_per_second": 70.694,
+      "step": 13045
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6667944295387762e-05,
+      "loss": 0.9511,
+      "step": 13050
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6655168008176824e-05,
+      "loss": 0.8864,
+      "step": 13060
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6642391720965887e-05,
+      "loss": 1.041,
+      "step": 13070
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.6629615433754952e-05,
+      "loss": 0.9998,
+      "step": 13080
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.6616839146544015e-05,
+      "loss": 0.9828,
+      "step": 13090
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.660406285933308e-05,
+      "loss": 0.985,
+      "step": 13100
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.659128657212214e-05,
+      "loss": 1.005,
+      "step": 13110
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6578510284911205e-05,
+      "loss": 0.8982,
+      "step": 13120
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.6565733997700268e-05,
+      "loss": 0.9311,
+      "step": 13130
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6552957710489333e-05,
+      "loss": 0.9662,
+      "step": 13140
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6540181423278396e-05,
+      "loss": 0.8836,
+      "step": 13150
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.6527405136067458e-05,
+      "loss": 0.9816,
+      "step": 13160
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.651462884885652e-05,
+      "loss": 0.9692,
+      "step": 13170
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.6501852561645586e-05,
+      "loss": 0.9834,
+      "step": 13180
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6489076274434652e-05,
+      "loss": 0.962,
+      "step": 13190
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6476299987223714e-05,
+      "loss": 0.8939,
+      "step": 13200
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.6463523700012777e-05,
+      "loss": 1.0154,
+      "step": 13210
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.645074741280184e-05,
+      "loss": 1.0056,
+      "step": 13220
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6437971125590905e-05,
+      "loss": 0.9138,
+      "step": 13230
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.6425194838379967e-05,
+      "loss": 0.8636,
+      "step": 13240
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6412418551169033e-05,
+      "loss": 0.9425,
+      "step": 13250
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.6399642263958092e-05,
+      "loss": 0.9244,
+      "step": 13260
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6386865976747158e-05,
+      "loss": 1.0511,
+      "step": 13270
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.637408968953622e-05,
+      "loss": 0.9437,
+      "step": 13280
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.6361313402325286e-05,
+      "loss": 0.9603,
+      "step": 13290
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.6348537115114348e-05,
+      "loss": 0.9806,
+      "step": 13300
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.633576082790341e-05,
+      "loss": 1.0418,
+      "step": 13310
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.6322984540692473e-05,
+      "loss": 0.9977,
+      "step": 13320
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.631020825348154e-05,
+      "loss": 0.9562,
+      "step": 13330
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.62974319662706e-05,
+      "loss": 0.9789,
+      "step": 13340
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6284655679059667e-05,
+      "loss": 0.981,
+      "step": 13350
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.6271879391848732e-05,
+      "loss": 0.9924,
+      "step": 13360
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.625910310463779e-05,
+      "loss": 0.9103,
+      "step": 13370
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.6246326817426857e-05,
+      "loss": 0.9105,
+      "step": 13380
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.623355053021592e-05,
+      "loss": 0.9018,
+      "step": 13390
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6220774243004985e-05,
+      "loss": 1.008,
+      "step": 13400
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.6207997955794048e-05,
+      "loss": 0.9163,
+      "step": 13410
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.619522166858311e-05,
+      "loss": 0.9285,
+      "step": 13420
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6182445381372172e-05,
+      "loss": 1.0056,
+      "step": 13430
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.6169669094161238e-05,
+      "loss": 1.0056,
+      "step": 13440
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.61568928069503e-05,
+      "loss": 1.0101,
+      "step": 13450
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6144116519739366e-05,
+      "loss": 0.916,
+      "step": 13460
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.6131340232528425e-05,
+      "loss": 0.8751,
+      "step": 13470
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.611856394531749e-05,
+      "loss": 0.9678,
+      "step": 13480
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.6105787658106553e-05,
+      "loss": 0.9592,
+      "step": 13490
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.609301137089562e-05,
+      "loss": 0.9084,
+      "step": 13500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.608023508368468e-05,
+      "loss": 1.0476,
+      "step": 13510
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.6067458796473744e-05,
+      "loss": 0.9773,
+      "step": 13520
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.605468250926281e-05,
+      "loss": 0.9905,
+      "step": 13530
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6041906222051872e-05,
+      "loss": 1.0131,
+      "step": 13540
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.6029129934840938e-05,
+      "loss": 0.9017,
+      "step": 13550
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.601635364763e-05,
+      "loss": 0.9386,
+      "step": 13560
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.6003577360419062e-05,
+      "loss": 0.8702,
+      "step": 13570
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5990801073208125e-05,
+      "loss": 0.9276,
+      "step": 13580
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.597802478599719e-05,
+      "loss": 0.9717,
+      "step": 13590
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5965248498786253e-05,
+      "loss": 0.9509,
+      "step": 13600
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.595247221157532e-05,
+      "loss": 0.9955,
+      "step": 13610
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5939695924364378e-05,
+      "loss": 0.9675,
+      "step": 13620
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5926919637153443e-05,
+      "loss": 0.9306,
+      "step": 13630
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5914143349942506e-05,
+      "loss": 1.0027,
+      "step": 13640
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.590136706273157e-05,
+      "loss": 0.9495,
+      "step": 13650
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5888590775520634e-05,
+      "loss": 0.9397,
+      "step": 13660
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5875814488309696e-05,
+      "loss": 0.9794,
+      "step": 13670
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.586303820109876e-05,
+      "loss": 0.9686,
+      "step": 13680
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5850261913887824e-05,
+      "loss": 1.0132,
+      "step": 13690
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.583748562667689e-05,
+      "loss": 0.9909,
+      "step": 13700
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5824709339465952e-05,
+      "loss": 0.9189,
+      "step": 13710
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5811933052255018e-05,
+      "loss": 0.9516,
+      "step": 13720
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5799156765044077e-05,
+      "loss": 0.8799,
+      "step": 13730
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5786380477833143e-05,
+      "loss": 0.9896,
+      "step": 13740
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5773604190622205e-05,
+      "loss": 0.9109,
+      "step": 13750
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.576082790341127e-05,
+      "loss": 1.0067,
+      "step": 13760
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.5748051616200333e-05,
+      "loss": 0.9225,
+      "step": 13770
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.5735275328989396e-05,
+      "loss": 0.9435,
+      "step": 13780
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.5722499041778458e-05,
+      "loss": 0.9736,
+      "step": 13790
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.5709722754567524e-05,
+      "loss": 0.8851,
+      "step": 13800
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.5696946467356586e-05,
+      "loss": 0.9621,
+      "step": 13810
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.5684170180145652e-05,
+      "loss": 0.9986,
+      "step": 13820
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.567139389293471e-05,
+      "loss": 1.0393,
+      "step": 13830
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.5658617605723777e-05,
+      "loss": 0.9755,
+      "step": 13840
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.564584131851284e-05,
+      "loss": 0.9263,
+      "step": 13850
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.5633065031301905e-05,
+      "loss": 0.9422,
+      "step": 13860
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.562028874409097e-05,
+      "loss": 0.9224,
+      "step": 13870
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.560751245688003e-05,
+      "loss": 0.9424,
+      "step": 13880
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.5594736169669095e-05,
+      "loss": 1.0011,
+      "step": 13890
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.5581959882458158e-05,
+      "loss": 0.9618,
+      "step": 13900
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.5569183595247223e-05,
+      "loss": 0.9749,
+      "step": 13910
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.5556407308036286e-05,
+      "loss": 0.9849,
+      "step": 13920
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.5543631020825348e-05,
+      "loss": 0.9788,
+      "step": 13930
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.553085473361441e-05,
+      "loss": 1.0272,
+      "step": 13940
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.5518078446403476e-05,
+      "loss": 0.9311,
+      "step": 13950
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.550530215919254e-05,
+      "loss": 0.9522,
+      "step": 13960
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.5492525871981604e-05,
+      "loss": 0.9477,
+      "step": 13970
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.5479749584770663e-05,
+      "loss": 1.0389,
+      "step": 13980
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.546697329755973e-05,
+      "loss": 1.0002,
+      "step": 13990
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.545419701034879e-05,
+      "loss": 0.9685,
+      "step": 14000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.5441420723137857e-05,
+      "loss": 1.0261,
+      "step": 14010
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.542864443592692e-05,
+      "loss": 0.9437,
+      "step": 14020
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.5415868148715985e-05,
+      "loss": 0.9554,
+      "step": 14030
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.5403091861505048e-05,
+      "loss": 0.9672,
+      "step": 14040
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.539031557429411e-05,
+      "loss": 1.0075,
+      "step": 14050
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.5377539287083176e-05,
+      "loss": 0.9677,
+      "step": 14060
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.5364762999872238e-05,
+      "loss": 0.9479,
+      "step": 14070
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.5351986712661304e-05,
+      "loss": 0.8863,
+      "step": 14080
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.5339210425450363e-05,
+      "loss": 1.02,
+      "step": 14090
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.532643413823943e-05,
+      "loss": 1.0428,
+      "step": 14100
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.531365785102849e-05,
+      "loss": 1.0251,
+      "step": 14110
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.5300881563817557e-05,
+      "loss": 0.9373,
+      "step": 14120
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.528810527660662e-05,
+      "loss": 0.9771,
+      "step": 14130
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.527532898939568e-05,
+      "loss": 0.8552,
+      "step": 14140
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.5262552702184744e-05,
+      "loss": 1.0306,
+      "step": 14150
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.524977641497381e-05,
+      "loss": 1.0745,
+      "step": 14160
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.5237000127762872e-05,
+      "loss": 0.9924,
+      "step": 14170
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.5224223840551936e-05,
+      "loss": 1.0457,
+      "step": 14180
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.5211447553340998e-05,
+      "loss": 0.9553,
+      "step": 14190
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.5198671266130062e-05,
+      "loss": 0.9672,
+      "step": 14200
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.5185894978919128e-05,
+      "loss": 0.9946,
+      "step": 14210
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.517311869170819e-05,
+      "loss": 0.9928,
+      "step": 14220
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.5160342404497255e-05,
+      "loss": 0.9947,
+      "step": 14230
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.5147566117286317e-05,
+      "loss": 1.0306,
+      "step": 14240
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.5134789830075381e-05,
+      "loss": 1.0141,
+      "step": 14250
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.5122013542864443e-05,
+      "loss": 0.9512,
+      "step": 14260
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.5109237255653509e-05,
+      "loss": 0.9386,
+      "step": 14270
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.509646096844257e-05,
+      "loss": 0.9933,
+      "step": 14280
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.5083684681231635e-05,
+      "loss": 0.9377,
+      "step": 14290
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.5070908394020698e-05,
+      "loss": 0.9007,
+      "step": 14300
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.5058132106809762e-05,
+      "loss": 0.9382,
+      "step": 14310
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.5045355819598824e-05,
+      "loss": 0.8972,
+      "step": 14320
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.5032579532387888e-05,
+      "loss": 1.0583,
+      "step": 14330
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.501980324517695e-05,
+      "loss": 0.992,
+      "step": 14340
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.5007026957966016e-05,
+      "loss": 0.9516,
+      "step": 14350
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.4994250670755079e-05,
+      "loss": 0.9898,
+      "step": 14360
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4981474383544143e-05,
+      "loss": 0.9197,
+      "step": 14370
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4968698096333205e-05,
+      "loss": 0.9178,
+      "step": 14380
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.495592180912227e-05,
+      "loss": 0.9039,
+      "step": 14390
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4943145521911333e-05,
+      "loss": 0.9795,
+      "step": 14400
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4930369234700396e-05,
+      "loss": 1.0887,
+      "step": 14410
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.491759294748946e-05,
+      "loss": 0.9682,
+      "step": 14420
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4904816660278524e-05,
+      "loss": 1.0227,
+      "step": 14430
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4892040373067586e-05,
+      "loss": 0.9106,
+      "step": 14440
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.487926408585665e-05,
+      "loss": 0.9762,
+      "step": 14450
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.4866487798645713e-05,
+      "loss": 0.8884,
+      "step": 14460
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4853711511434778e-05,
+      "loss": 0.9572,
+      "step": 14470
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4840935224223842e-05,
+      "loss": 0.965,
+      "step": 14480
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4828158937012905e-05,
+      "loss": 0.9821,
+      "step": 14490
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4815382649801969e-05,
+      "loss": 1.0218,
+      "step": 14500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4802606362591031e-05,
+      "loss": 0.9829,
+      "step": 14510
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4789830075380095e-05,
+      "loss": 0.9546,
+      "step": 14520
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.477705378816916e-05,
+      "loss": 0.9789,
+      "step": 14530
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4764277500958222e-05,
+      "loss": 0.9226,
+      "step": 14540
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4751501213747286e-05,
+      "loss": 0.9627,
+      "step": 14550
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4738724926536348e-05,
+      "loss": 0.9271,
+      "step": 14560
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4725948639325412e-05,
+      "loss": 0.9789,
+      "step": 14570
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4713172352114476e-05,
+      "loss": 0.962,
+      "step": 14580
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4700396064903539e-05,
+      "loss": 0.9724,
+      "step": 14590
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.4687619777692603e-05,
+      "loss": 0.9596,
+      "step": 14600
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.4674843490481667e-05,
+      "loss": 0.9355,
+      "step": 14610
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.4662067203270729e-05,
+      "loss": 0.9744,
+      "step": 14620
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4649290916059793e-05,
+      "loss": 0.9599,
+      "step": 14630
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4636514628848857e-05,
+      "loss": 0.9202,
+      "step": 14640
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4623738341637921e-05,
+      "loss": 0.9353,
+      "step": 14650
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4610962054426985e-05,
+      "loss": 1.0016,
+      "step": 14660
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4598185767216048e-05,
+      "loss": 0.9177,
+      "step": 14670
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4585409480005112e-05,
+      "loss": 0.8621,
+      "step": 14680
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4572633192794174e-05,
+      "loss": 0.9553,
+      "step": 14690
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4559856905583238e-05,
+      "loss": 0.9857,
+      "step": 14700
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4547080618372302e-05,
+      "loss": 0.9332,
+      "step": 14710
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4534304331161364e-05,
+      "loss": 0.9719,
+      "step": 14720
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4521528043950429e-05,
+      "loss": 0.9366,
+      "step": 14730
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4508751756739491e-05,
+      "loss": 0.8713,
+      "step": 14740
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4495975469528555e-05,
+      "loss": 0.925,
+      "step": 14750
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4483199182317619e-05,
+      "loss": 0.9503,
+      "step": 14760
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4470422895106681e-05,
+      "loss": 0.9484,
+      "step": 14770
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4457646607895745e-05,
+      "loss": 0.9774,
+      "step": 14780
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.444487032068481e-05,
+      "loss": 0.9409,
+      "step": 14790
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4432094033473872e-05,
+      "loss": 0.9403,
+      "step": 14800
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4419317746262936e-05,
+      "loss": 0.8935,
+      "step": 14810
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4406541459052e-05,
+      "loss": 0.9192,
+      "step": 14820
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4393765171841064e-05,
+      "loss": 0.9793,
+      "step": 14830
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4380988884630128e-05,
+      "loss": 0.9864,
+      "step": 14840
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.436821259741919e-05,
+      "loss": 0.947,
+      "step": 14850
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4355436310208255e-05,
+      "loss": 0.8894,
+      "step": 14860
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4342660022997317e-05,
+      "loss": 0.9636,
+      "step": 14870
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4329883735786381e-05,
+      "loss": 1.0282,
+      "step": 14880
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4317107448575445e-05,
+      "loss": 0.8897,
+      "step": 14890
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4304331161364507e-05,
+      "loss": 0.9213,
+      "step": 14900
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4291554874153571e-05,
+      "loss": 0.984,
+      "step": 14910
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4278778586942634e-05,
+      "loss": 0.9425,
+      "step": 14920
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4266002299731698e-05,
+      "loss": 0.8759,
+      "step": 14930
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4253226012520762e-05,
+      "loss": 1.0491,
+      "step": 14940
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4240449725309824e-05,
+      "loss": 0.9523,
+      "step": 14950
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4227673438098888e-05,
+      "loss": 0.9773,
+      "step": 14960
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4214897150887952e-05,
+      "loss": 1.0238,
+      "step": 14970
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4202120863677015e-05,
+      "loss": 0.8875,
+      "step": 14980
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.418934457646608e-05,
+      "loss": 0.9249,
+      "step": 14990
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4176568289255143e-05,
+      "loss": 0.931,
+      "step": 15000
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4163792002044207e-05,
+      "loss": 0.9782,
+      "step": 15010
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4151015714833271e-05,
+      "loss": 0.9454,
+      "step": 15020
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4138239427622333e-05,
+      "loss": 0.985,
+      "step": 15030
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4125463140411397e-05,
+      "loss": 0.9814,
+      "step": 15040
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.411268685320046e-05,
+      "loss": 0.994,
+      "step": 15050
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4099910565989524e-05,
+      "loss": 0.981,
+      "step": 15060
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.4087134278778588e-05,
+      "loss": 0.8869,
+      "step": 15070
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.407435799156765e-05,
+      "loss": 0.8948,
+      "step": 15080
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.4061581704356714e-05,
+      "loss": 1.0393,
+      "step": 15090
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4048805417145778e-05,
+      "loss": 0.9359,
+      "step": 15100
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.403602912993484e-05,
+      "loss": 0.9872,
+      "step": 15110
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.4023252842723905e-05,
+      "loss": 0.9779,
+      "step": 15120
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.4010476555512967e-05,
+      "loss": 1.022,
+      "step": 15130
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.3997700268302031e-05,
+      "loss": 0.9735,
+      "step": 15140
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.3984923981091095e-05,
+      "loss": 0.9278,
+      "step": 15150
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.397214769388016e-05,
+      "loss": 0.9582,
+      "step": 15160
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.3959371406669223e-05,
+      "loss": 0.9715,
+      "step": 15170
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.3946595119458286e-05,
+      "loss": 0.9954,
+      "step": 15180
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.393381883224735e-05,
+      "loss": 0.9101,
+      "step": 15190
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3921042545036414e-05,
+      "loss": 0.8907,
+      "step": 15200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3908266257825476e-05,
+      "loss": 0.896,
+      "step": 15210
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.389548997061454e-05,
+      "loss": 1.0151,
+      "step": 15220
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.3882713683403603e-05,
+      "loss": 0.9799,
+      "step": 15230
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.3869937396192667e-05,
+      "loss": 0.9436,
+      "step": 15240
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.385716110898173e-05,
+      "loss": 0.9756,
+      "step": 15250
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3844384821770793e-05,
+      "loss": 0.9419,
+      "step": 15260
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3831608534559857e-05,
+      "loss": 0.9821,
+      "step": 15270
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3818832247348921e-05,
+      "loss": 0.9894,
+      "step": 15280
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3806055960137984e-05,
+      "loss": 0.9639,
+      "step": 15290
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3793279672927048e-05,
+      "loss": 0.9347,
+      "step": 15300
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.378050338571611e-05,
+      "loss": 0.9394,
+      "step": 15310
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.3767727098505174e-05,
+      "loss": 0.9341,
+      "step": 15320
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.375495081129424e-05,
+      "loss": 0.9435,
+      "step": 15330
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.3742174524083302e-05,
+      "loss": 0.9809,
+      "step": 15340
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.3729398236872366e-05,
+      "loss": 0.9137,
+      "step": 15350
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3716621949661429e-05,
+      "loss": 0.9425,
+      "step": 15360
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3703845662450493e-05,
+      "loss": 0.9033,
+      "step": 15370
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3691069375239557e-05,
+      "loss": 0.9462,
+      "step": 15380
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3678293088028619e-05,
+      "loss": 0.9076,
+      "step": 15390
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3665516800817683e-05,
+      "loss": 0.9702,
+      "step": 15400
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3652740513606745e-05,
+      "loss": 0.9613,
+      "step": 15410
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.363996422639581e-05,
+      "loss": 0.9325,
+      "step": 15420
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3627187939184874e-05,
+      "loss": 0.9128,
+      "step": 15430
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3614411651973936e-05,
+      "loss": 0.9716,
+      "step": 15440
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3601635364763e-05,
+      "loss": 0.912,
+      "step": 15450
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3588859077552064e-05,
+      "loss": 0.8953,
+      "step": 15460
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3576082790341126e-05,
+      "loss": 0.9225,
+      "step": 15470
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.356330650313019e-05,
+      "loss": 1.0659,
+      "step": 15480
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3550530215919253e-05,
+      "loss": 0.9815,
+      "step": 15490
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3537753928708319e-05,
+      "loss": 0.9231,
+      "step": 15500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3524977641497383e-05,
+      "loss": 0.9542,
+      "step": 15510
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.3512201354286445e-05,
+      "loss": 0.8962,
+      "step": 15520
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.3499425067075509e-05,
+      "loss": 0.976,
+      "step": 15530
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3486648779864571e-05,
+      "loss": 0.9271,
+      "step": 15540
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3473872492653635e-05,
+      "loss": 0.9489,
+      "step": 15550
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.34610962054427e-05,
+      "loss": 0.9426,
+      "step": 15560
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.3448319918231762e-05,
+      "loss": 0.983,
+      "step": 15570
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.3435543631020826e-05,
+      "loss": 0.9935,
+      "step": 15580
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3422767343809888e-05,
+      "loss": 0.9129,
+      "step": 15590
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3409991056598952e-05,
+      "loss": 0.949,
+      "step": 15600
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3397214769388016e-05,
+      "loss": 1.095,
+      "step": 15610
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3384438482177079e-05,
+      "loss": 0.8744,
+      "step": 15620
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3371662194966143e-05,
+      "loss": 0.9047,
+      "step": 15630
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3358885907755207e-05,
+      "loss": 0.9946,
+      "step": 15640
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.334610962054427e-05,
+      "loss": 0.9818,
+      "step": 15650
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.7412021177203364,
+      "eval_loss": 0.7523965239524841,
+      "eval_runtime": 305.2278,
+      "eval_samples_per_second": 136.76,
+      "eval_steps_per_second": 68.382,
+      "step": 15654
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.9586,
+      "step": 15660
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3320557046122397e-05,
+      "loss": 0.8459,
+      "step": 15670
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3307780758911461e-05,
+      "loss": 0.8638,
+      "step": 15680
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3295004471700525e-05,
+      "loss": 0.8783,
+      "step": 15690
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3282228184489588e-05,
+      "loss": 0.9379,
+      "step": 15700
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3269451897278652e-05,
+      "loss": 0.8332,
+      "step": 15710
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3256675610067714e-05,
+      "loss": 1.028,
+      "step": 15720
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3243899322856778e-05,
+      "loss": 0.8896,
+      "step": 15730
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3231123035645842e-05,
+      "loss": 0.9274,
+      "step": 15740
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3218346748434905e-05,
+      "loss": 0.9472,
+      "step": 15750
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3205570461223969e-05,
+      "loss": 0.9594,
+      "step": 15760
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3192794174013033e-05,
+      "loss": 0.9271,
+      "step": 15770
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.3180017886802095e-05,
+      "loss": 0.943,
+      "step": 15780
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.316724159959116e-05,
+      "loss": 0.9039,
+      "step": 15790
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3154465312380222e-05,
+      "loss": 1.0425,
+      "step": 15800
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3141689025169286e-05,
+      "loss": 1.0,
+      "step": 15810
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.312891273795835e-05,
+      "loss": 0.9602,
+      "step": 15820
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3116136450747412e-05,
+      "loss": 0.955,
+      "step": 15830
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3103360163536476e-05,
+      "loss": 0.9675,
+      "step": 15840
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.309058387632554e-05,
+      "loss": 0.882,
+      "step": 15850
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.3077807589114604e-05,
+      "loss": 0.9206,
+      "step": 15860
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.306630893062476e-05,
+      "loss": 0.9227,
+      "step": 15870
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3053532643413823e-05,
+      "loss": 0.9688,
+      "step": 15880
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3040756356202887e-05,
+      "loss": 0.914,
+      "step": 15890
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3027980068991952e-05,
+      "loss": 0.9487,
+      "step": 15900
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.3015203781781016e-05,
+      "loss": 0.928,
+      "step": 15910
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.300242749457008e-05,
+      "loss": 1.0184,
+      "step": 15920
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.2989651207359142e-05,
+      "loss": 0.9388,
+      "step": 15930
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.2976874920148206e-05,
+      "loss": 0.9807,
+      "step": 15940
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.2964098632937268e-05,
+      "loss": 0.9869,
+      "step": 15950
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.2951322345726332e-05,
+      "loss": 0.9385,
+      "step": 15960
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.2938546058515397e-05,
+      "loss": 0.9272,
+      "step": 15970
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.2925769771304459e-05,
+      "loss": 0.915,
+      "step": 15980
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.2912993484093523e-05,
+      "loss": 0.9132,
+      "step": 15990
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.2900217196882585e-05,
+      "loss": 0.8885,
+      "step": 16000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.288744090967165e-05,
+      "loss": 0.9054,
+      "step": 16010
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.2874664622460713e-05,
+      "loss": 0.8998,
+      "step": 16020
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.2861888335249776e-05,
+      "loss": 0.8836,
+      "step": 16030
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.284911204803884e-05,
+      "loss": 0.9405,
+      "step": 16040
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.2836335760827902e-05,
+      "loss": 0.9278,
+      "step": 16050
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.2823559473616966e-05,
+      "loss": 0.9781,
+      "step": 16060
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.2810783186406032e-05,
+      "loss": 0.976,
+      "step": 16070
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.2798006899195094e-05,
+      "loss": 0.9477,
+      "step": 16080
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.2785230611984158e-05,
+      "loss": 0.9721,
+      "step": 16090
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.2772454324773223e-05,
+      "loss": 0.9329,
+      "step": 16100
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.2759678037562285e-05,
+      "loss": 0.9914,
+      "step": 16110
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.2746901750351349e-05,
+      "loss": 1.0077,
+      "step": 16120
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.2734125463140411e-05,
+      "loss": 0.9494,
+      "step": 16130
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.2721349175929475e-05,
+      "loss": 0.8976,
+      "step": 16140
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.270857288871854e-05,
+      "loss": 0.9463,
+      "step": 16150
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.2695796601507602e-05,
+      "loss": 0.9427,
+      "step": 16160
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.2683020314296666e-05,
+      "loss": 0.9242,
+      "step": 16170
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.2670244027085728e-05,
+      "loss": 0.9318,
+      "step": 16180
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.2657467739874792e-05,
+      "loss": 0.8956,
+      "step": 16190
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.2644691452663856e-05,
+      "loss": 0.9115,
+      "step": 16200
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.2631915165452919e-05,
+      "loss": 0.9006,
+      "step": 16210
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.2619138878241983e-05,
+      "loss": 0.9181,
+      "step": 16220
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.2606362591031047e-05,
+      "loss": 0.9392,
+      "step": 16230
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.259358630382011e-05,
+      "loss": 0.9432,
+      "step": 16240
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.2580810016609175e-05,
+      "loss": 0.9381,
+      "step": 16250
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.2568033729398237e-05,
+      "loss": 0.8903,
+      "step": 16260
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.2555257442187301e-05,
+      "loss": 0.8658,
+      "step": 16270
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.2542481154976365e-05,
+      "loss": 0.9013,
+      "step": 16280
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.2529704867765428e-05,
+      "loss": 0.8659,
+      "step": 16290
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.2516928580554492e-05,
+      "loss": 0.8926,
+      "step": 16300
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.2504152293343554e-05,
+      "loss": 0.9089,
+      "step": 16310
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.2491376006132618e-05,
+      "loss": 0.9359,
+      "step": 16320
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.2478599718921682e-05,
+      "loss": 0.8845,
+      "step": 16330
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.2465823431710745e-05,
+      "loss": 0.9725,
+      "step": 16340
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.2453047144499809e-05,
+      "loss": 0.9351,
+      "step": 16350
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.2440270857288871e-05,
+      "loss": 0.8831,
+      "step": 16360
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.2427494570077935e-05,
+      "loss": 0.9873,
+      "step": 16370
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.2414718282867e-05,
+      "loss": 0.9707,
+      "step": 16380
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.2401941995656062e-05,
+      "loss": 0.8314,
+      "step": 16390
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.2389165708445126e-05,
+      "loss": 0.9669,
+      "step": 16400
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.2376389421234191e-05,
+      "loss": 0.8932,
+      "step": 16410
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.2363613134023254e-05,
+      "loss": 0.9126,
+      "step": 16420
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.2350836846812318e-05,
+      "loss": 0.9265,
+      "step": 16430
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.233806055960138e-05,
+      "loss": 1.0193,
+      "step": 16440
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.2325284272390444e-05,
+      "loss": 0.9645,
+      "step": 16450
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.2312507985179508e-05,
+      "loss": 0.9286,
+      "step": 16460
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.229973169796857e-05,
+      "loss": 0.9656,
+      "step": 16470
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2286955410757635e-05,
+      "loss": 0.9616,
+      "step": 16480
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2274179123546697e-05,
+      "loss": 0.9177,
+      "step": 16490
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2261402836335761e-05,
+      "loss": 0.9946,
+      "step": 16500
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2248626549124825e-05,
+      "loss": 0.9088,
+      "step": 16510
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2235850261913887e-05,
+      "loss": 1.0018,
+      "step": 16520
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.2223073974702952e-05,
+      "loss": 0.9384,
+      "step": 16530
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.2210297687492014e-05,
+      "loss": 0.9727,
+      "step": 16540
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.2197521400281078e-05,
+      "loss": 1.0048,
+      "step": 16550
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2184745113070142e-05,
+      "loss": 0.9819,
+      "step": 16560
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2171968825859204e-05,
+      "loss": 0.8762,
+      "step": 16570
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2159192538648268e-05,
+      "loss": 0.9546,
+      "step": 16580
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2146416251437334e-05,
+      "loss": 0.9259,
+      "step": 16590
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2133639964226397e-05,
+      "loss": 0.9361,
+      "step": 16600
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.212086367701546e-05,
+      "loss": 0.9629,
+      "step": 16610
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2108087389804523e-05,
+      "loss": 0.9916,
+      "step": 16620
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2095311102593587e-05,
+      "loss": 0.9547,
+      "step": 16630
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.2082534815382651e-05,
+      "loss": 0.9804,
+      "step": 16640
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.2069758528171713e-05,
+      "loss": 0.9321,
+      "step": 16650
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2056982240960777e-05,
+      "loss": 0.9202,
+      "step": 16660
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.204420595374984e-05,
+      "loss": 0.9242,
+      "step": 16670
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2031429666538904e-05,
+      "loss": 0.8881,
+      "step": 16680
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.2018653379327968e-05,
+      "loss": 0.9281,
+      "step": 16690
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.200587709211703e-05,
+      "loss": 0.9828,
+      "step": 16700
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.1993100804906094e-05,
+      "loss": 0.9089,
+      "step": 16710
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.1980324517695157e-05,
+      "loss": 0.9634,
+      "step": 16720
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.196754823048422e-05,
+      "loss": 0.9364,
+      "step": 16730
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.1954771943273285e-05,
+      "loss": 0.929,
+      "step": 16740
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.1941995656062347e-05,
+      "loss": 0.8833,
+      "step": 16750
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.1929219368851413e-05,
+      "loss": 0.9461,
+      "step": 16760
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.1916443081640477e-05,
+      "loss": 0.9085,
+      "step": 16770
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.190366679442954e-05,
+      "loss": 0.9154,
+      "step": 16780
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.1890890507218603e-05,
+      "loss": 0.9294,
+      "step": 16790
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.1878114220007666e-05,
+      "loss": 0.9265,
+      "step": 16800
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.186533793279673e-05,
+      "loss": 0.9582,
+      "step": 16810
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.1852561645585794e-05,
+      "loss": 0.9213,
+      "step": 16820
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.1839785358374856e-05,
+      "loss": 0.9956,
+      "step": 16830
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.182700907116392e-05,
+      "loss": 0.8833,
+      "step": 16840
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.1814232783952983e-05,
+      "loss": 0.8304,
+      "step": 16850
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.1801456496742047e-05,
+      "loss": 0.9825,
+      "step": 16860
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.178868020953111e-05,
+      "loss": 0.8428,
+      "step": 16870
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.1775903922320173e-05,
+      "loss": 0.8903,
+      "step": 16880
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.1763127635109237e-05,
+      "loss": 1.0443,
+      "step": 16890
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.1750351347898301e-05,
+      "loss": 0.8667,
+      "step": 16900
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.1737575060687364e-05,
+      "loss": 0.919,
+      "step": 16910
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.1724798773476428e-05,
+      "loss": 0.9725,
+      "step": 16920
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.1712022486265492e-05,
+      "loss": 1.0142,
+      "step": 16930
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.1699246199054556e-05,
+      "loss": 0.8658,
+      "step": 16940
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.168646991184362e-05,
+      "loss": 0.9889,
+      "step": 16950
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.1673693624632682e-05,
+      "loss": 0.9083,
+      "step": 16960
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.1662194966142839e-05,
+      "loss": 1.0192,
+      "step": 16970
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.1649418678931903e-05,
+      "loss": 0.9716,
+      "step": 16980
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.1636642391720967e-05,
+      "loss": 0.9514,
+      "step": 16990
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.162386610451003e-05,
+      "loss": 0.9483,
+      "step": 17000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.1611089817299094e-05,
+      "loss": 0.9177,
+      "step": 17010
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.1598313530088158e-05,
+      "loss": 0.9618,
+      "step": 17020
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.158553724287722e-05,
+      "loss": 0.925,
+      "step": 17030
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.1572760955666284e-05,
+      "loss": 0.9186,
+      "step": 17040
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.1559984668455348e-05,
+      "loss": 0.9548,
+      "step": 17050
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.154720838124441e-05,
+      "loss": 0.8696,
+      "step": 17060
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.1534432094033475e-05,
+      "loss": 0.9062,
+      "step": 17070
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.1521655806822537e-05,
+      "loss": 0.911,
+      "step": 17080
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.1508879519611601e-05,
+      "loss": 0.901,
+      "step": 17090
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.1496103232400665e-05,
+      "loss": 0.8725,
+      "step": 17100
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.1483326945189727e-05,
+      "loss": 0.9769,
+      "step": 17110
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.1470550657978791e-05,
+      "loss": 0.9614,
+      "step": 17120
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.1457774370767854e-05,
+      "loss": 0.8504,
+      "step": 17130
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.1444998083556918e-05,
+      "loss": 0.8765,
+      "step": 17140
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.1432221796345984e-05,
+      "loss": 1.0312,
+      "step": 17150
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.1419445509135046e-05,
+      "loss": 0.94,
+      "step": 17160
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.140666922192411e-05,
+      "loss": 0.9744,
+      "step": 17170
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.1393892934713172e-05,
+      "loss": 0.9472,
+      "step": 17180
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.1381116647502236e-05,
+      "loss": 0.9806,
+      "step": 17190
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.13683403602913e-05,
+      "loss": 0.9444,
+      "step": 17200
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.1355564073080363e-05,
+      "loss": 0.9197,
+      "step": 17210
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.1342787785869427e-05,
+      "loss": 1.0164,
+      "step": 17220
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.1330011498658491e-05,
+      "loss": 0.886,
+      "step": 17230
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.1317235211447553e-05,
+      "loss": 0.9585,
+      "step": 17240
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.1304458924236617e-05,
+      "loss": 0.9107,
+      "step": 17250
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.129168263702568e-05,
+      "loss": 0.9869,
+      "step": 17260
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.1278906349814744e-05,
+      "loss": 0.9038,
+      "step": 17270
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.1266130062603808e-05,
+      "loss": 0.9483,
+      "step": 17280
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.125335377539287e-05,
+      "loss": 0.9691,
+      "step": 17290
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.1240577488181934e-05,
+      "loss": 0.8956,
+      "step": 17300
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.1227801200970997e-05,
+      "loss": 0.9302,
+      "step": 17310
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.121502491376006e-05,
+      "loss": 0.9601,
+      "step": 17320
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.1202248626549126e-05,
+      "loss": 1.024,
+      "step": 17330
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.1189472339338189e-05,
+      "loss": 0.8333,
+      "step": 17340
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.1176696052127253e-05,
+      "loss": 0.9275,
+      "step": 17350
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.1163919764916315e-05,
+      "loss": 0.9635,
+      "step": 17360
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.115114347770538e-05,
+      "loss": 0.9281,
+      "step": 17370
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.1138367190494443e-05,
+      "loss": 0.8926,
+      "step": 17380
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.1125590903283506e-05,
+      "loss": 0.9328,
+      "step": 17390
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.111281461607257e-05,
+      "loss": 0.9769,
+      "step": 17400
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.1100038328861634e-05,
+      "loss": 0.9333,
+      "step": 17410
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.1087262041650696e-05,
+      "loss": 0.9036,
+      "step": 17420
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.107448575443976e-05,
+      "loss": 0.9757,
+      "step": 17430
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.1061709467228823e-05,
+      "loss": 0.9172,
+      "step": 17440
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.1048933180017887e-05,
+      "loss": 0.9398,
+      "step": 17450
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.103615689280695e-05,
+      "loss": 0.9642,
+      "step": 17460
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.1023380605596013e-05,
+      "loss": 0.9532,
+      "step": 17470
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.1010604318385077e-05,
+      "loss": 1.044,
+      "step": 17480
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.099782803117414e-05,
+      "loss": 0.8964,
+      "step": 17490
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.0985051743963205e-05,
+      "loss": 0.9378,
+      "step": 17500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.097227545675227e-05,
+      "loss": 0.9711,
+      "step": 17510
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.0959499169541332e-05,
+      "loss": 0.9112,
+      "step": 17520
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.0946722882330396e-05,
+      "loss": 0.9576,
+      "step": 17530
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.0933946595119458e-05,
+      "loss": 0.9392,
+      "step": 17540
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.0921170307908522e-05,
+      "loss": 0.897,
+      "step": 17550
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.0908394020697586e-05,
+      "loss": 0.9241,
+      "step": 17560
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.0895617733486649e-05,
+      "loss": 0.9095,
+      "step": 17570
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.0882841446275713e-05,
+      "loss": 0.8671,
+      "step": 17580
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.0870065159064777e-05,
+      "loss": 0.97,
+      "step": 17590
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.0857288871853839e-05,
+      "loss": 1.0,
+      "step": 17600
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.0844512584642903e-05,
+      "loss": 1.016,
+      "step": 17610
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.0831736297431965e-05,
+      "loss": 0.9341,
+      "step": 17620
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.081896001022103e-05,
+      "loss": 0.9678,
+      "step": 17630
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.0806183723010094e-05,
+      "loss": 0.9584,
+      "step": 17640
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.0793407435799156e-05,
+      "loss": 0.9703,
+      "step": 17650
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.078063114858822e-05,
+      "loss": 0.9487,
+      "step": 17660
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.0767854861377284e-05,
+      "loss": 0.8318,
+      "step": 17670
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.0755078574166348e-05,
+      "loss": 0.9905,
+      "step": 17680
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.0742302286955412e-05,
+      "loss": 0.9442,
+      "step": 17690
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.0729525999744475e-05,
+      "loss": 0.9329,
+      "step": 17700
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.0716749712533539e-05,
+      "loss": 0.868,
+      "step": 17710
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.0703973425322603e-05,
+      "loss": 0.9682,
+      "step": 17720
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.0691197138111665e-05,
+      "loss": 0.9541,
+      "step": 17730
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.0678420850900729e-05,
+      "loss": 0.9863,
+      "step": 17740
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.0665644563689791e-05,
+      "loss": 0.9433,
+      "step": 17750
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.0652868276478855e-05,
+      "loss": 0.9281,
+      "step": 17760
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.064009198926792e-05,
+      "loss": 0.9478,
+      "step": 17770
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.0627315702056982e-05,
+      "loss": 0.9242,
+      "step": 17780
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.0614539414846046e-05,
+      "loss": 0.8586,
+      "step": 17790
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.0601763127635108e-05,
+      "loss": 0.876,
+      "step": 17800
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.0588986840424172e-05,
+      "loss": 0.9752,
+      "step": 17810
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.0576210553213236e-05,
+      "loss": 0.9239,
+      "step": 17820
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.0563434266002299e-05,
+      "loss": 0.8582,
+      "step": 17830
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.0550657978791365e-05,
+      "loss": 1.0195,
+      "step": 17840
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.0537881691580427e-05,
+      "loss": 0.864,
+      "step": 17850
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.0525105404369491e-05,
+      "loss": 0.8776,
+      "step": 17860
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.0512329117158555e-05,
+      "loss": 0.9705,
+      "step": 17870
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.0499552829947617e-05,
+      "loss": 0.98,
+      "step": 17880
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.0486776542736681e-05,
+      "loss": 0.9534,
+      "step": 17890
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.0474000255525745e-05,
+      "loss": 0.939,
+      "step": 17900
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.0461223968314808e-05,
+      "loss": 0.9252,
+      "step": 17910
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.0448447681103872e-05,
+      "loss": 0.9684,
+      "step": 17920
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.0435671393892934e-05,
+      "loss": 0.8427,
+      "step": 17930
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.0422895106681998e-05,
+      "loss": 0.9144,
+      "step": 17940
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.0410118819471062e-05,
+      "loss": 0.9737,
+      "step": 17950
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.0397342532260125e-05,
+      "loss": 0.8941,
+      "step": 17960
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.0384566245049189e-05,
+      "loss": 0.9312,
+      "step": 17970
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.0371789957838251e-05,
+      "loss": 0.8658,
+      "step": 17980
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.0359013670627315e-05,
+      "loss": 1.0223,
+      "step": 17990
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.034623738341638e-05,
+      "loss": 0.9213,
+      "step": 18000
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.0333461096205443e-05,
+      "loss": 0.973,
+      "step": 18010
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.0320684808994507e-05,
+      "loss": 0.872,
+      "step": 18020
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.030790852178357e-05,
+      "loss": 0.9214,
+      "step": 18030
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.0295132234572634e-05,
+      "loss": 0.9037,
+      "step": 18040
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.0282355947361698e-05,
+      "loss": 0.9465,
+      "step": 18050
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.026957966015076e-05,
+      "loss": 1.0212,
+      "step": 18060
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.0256803372939824e-05,
+      "loss": 0.9216,
+      "step": 18070
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.0244027085728888e-05,
+      "loss": 0.8502,
+      "step": 18080
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.023125079851795e-05,
+      "loss": 1.0534,
+      "step": 18090
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.0218474511307015e-05,
+      "loss": 0.9301,
+      "step": 18100
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.0205698224096077e-05,
+      "loss": 0.9158,
+      "step": 18110
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.0192921936885141e-05,
+      "loss": 0.921,
+      "step": 18120
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.0180145649674205e-05,
+      "loss": 0.961,
+      "step": 18130
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.0167369362463268e-05,
+      "loss": 0.9089,
+      "step": 18140
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.0154593075252332e-05,
+      "loss": 0.9501,
+      "step": 18150
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.0141816788041394e-05,
+      "loss": 0.9391,
+      "step": 18160
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.0129040500830458e-05,
+      "loss": 0.8947,
+      "step": 18170
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.0116264213619524e-05,
+      "loss": 0.9362,
+      "step": 18180
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.0103487926408586e-05,
+      "loss": 0.957,
+      "step": 18190
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.009071163919765e-05,
+      "loss": 0.8636,
+      "step": 18200
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.0077935351986713e-05,
+      "loss": 0.8835,
+      "step": 18210
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.0065159064775777e-05,
+      "loss": 0.967,
+      "step": 18220
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.005238277756484e-05,
+      "loss": 0.8585,
+      "step": 18230
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.0039606490353903e-05,
+      "loss": 0.8728,
+      "step": 18240
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.0026830203142967e-05,
+      "loss": 0.8082,
+      "step": 18250
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.0014053915932031e-05,
+      "loss": 0.948,
+      "step": 18260
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.7396928826390053,
+      "eval_loss": 0.7853737473487854,
+      "eval_runtime": 306.6153,
+      "eval_samples_per_second": 136.141,
+      "eval_steps_per_second": 68.072,
+      "step": 18263
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.0001277628721094e-05,
+      "loss": 0.917,
+      "step": 18270
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.988501341510158e-06,
+      "loss": 0.9677,
+      "step": 18280
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.97572505429922e-06,
+      "loss": 0.9087,
+      "step": 18290
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.962948767088284e-06,
+      "loss": 0.9864,
+      "step": 18300
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.950172479877348e-06,
+      "loss": 0.9212,
+      "step": 18310
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.93739619266641e-06,
+      "loss": 0.9485,
+      "step": 18320
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.924619905455475e-06,
+      "loss": 0.9506,
+      "step": 18330
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.911843618244537e-06,
+      "loss": 0.9752,
+      "step": 18340
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.899067331033601e-06,
+      "loss": 0.9582,
+      "step": 18350
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.886291043822667e-06,
+      "loss": 0.9211,
+      "step": 18360
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.873514756611729e-06,
+      "loss": 0.8882,
+      "step": 18370
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.860738469400793e-06,
+      "loss": 0.9447,
+      "step": 18380
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.847962182189857e-06,
+      "loss": 0.9005,
+      "step": 18390
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.83518589497892e-06,
+      "loss": 0.8327,
+      "step": 18400
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.822409607767984e-06,
+      "loss": 0.8961,
+      "step": 18410
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.809633320557046e-06,
+      "loss": 0.8904,
+      "step": 18420
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.79685703334611e-06,
+      "loss": 0.9239,
+      "step": 18430
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.784080746135174e-06,
+      "loss": 0.9298,
+      "step": 18440
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.771304458924236e-06,
+      "loss": 0.9099,
+      "step": 18450
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.7585281717133e-06,
+      "loss": 0.9328,
+      "step": 18460
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.745751884502363e-06,
+      "loss": 0.9255,
+      "step": 18470
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.732975597291427e-06,
+      "loss": 0.8669,
+      "step": 18480
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.720199310080491e-06,
+      "loss": 0.8689,
+      "step": 18490
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.707423022869553e-06,
+      "loss": 0.9246,
+      "step": 18500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.694646735658617e-06,
+      "loss": 0.8907,
+      "step": 18510
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.68187044844768e-06,
+      "loss": 0.9067,
+      "step": 18520
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.669094161236745e-06,
+      "loss": 0.9925,
+      "step": 18530
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.65631787402581e-06,
+      "loss": 0.9197,
+      "step": 18540
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.643541586814872e-06,
+      "loss": 0.9109,
+      "step": 18550
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.630765299603936e-06,
+      "loss": 0.9298,
+      "step": 18560
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.617989012393e-06,
+      "loss": 0.966,
+      "step": 18570
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.605212725182062e-06,
+      "loss": 0.965,
+      "step": 18580
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.592436437971126e-06,
+      "loss": 0.8526,
+      "step": 18590
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.579660150760189e-06,
+      "loss": 0.883,
+      "step": 18600
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.566883863549253e-06,
+      "loss": 0.8712,
+      "step": 18610
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.554107576338317e-06,
+      "loss": 0.9506,
+      "step": 18620
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.54133128912738e-06,
+      "loss": 0.8746,
+      "step": 18630
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.528555001916443e-06,
+      "loss": 0.938,
+      "step": 18640
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.515778714705506e-06,
+      "loss": 0.9669,
+      "step": 18650
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.50300242749457e-06,
+      "loss": 0.9333,
+      "step": 18660
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.490226140283634e-06,
+      "loss": 0.8384,
+      "step": 18670
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.477449853072696e-06,
+      "loss": 0.8996,
+      "step": 18680
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.46467356586176e-06,
+      "loss": 0.8694,
+      "step": 18690
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.451897278650824e-06,
+      "loss": 0.8928,
+      "step": 18700
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.439120991439888e-06,
+      "loss": 0.9085,
+      "step": 18710
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.426344704228952e-06,
+      "loss": 0.9642,
+      "step": 18720
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.413568417018015e-06,
+      "loss": 0.9233,
+      "step": 18730
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.400792129807079e-06,
+      "loss": 0.9378,
+      "step": 18740
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.388015842596143e-06,
+      "loss": 0.929,
+      "step": 18750
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.375239555385205e-06,
+      "loss": 0.8891,
+      "step": 18760
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.36246326817427e-06,
+      "loss": 0.8708,
+      "step": 18770
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.349686980963332e-06,
+      "loss": 0.9237,
+      "step": 18780
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.336910693752396e-06,
+      "loss": 0.8177,
+      "step": 18790
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.32413440654146e-06,
+      "loss": 0.8765,
+      "step": 18800
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.311358119330522e-06,
+      "loss": 0.8823,
+      "step": 18810
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.298581832119586e-06,
+      "loss": 0.936,
+      "step": 18820
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.285805544908649e-06,
+      "loss": 0.9233,
+      "step": 18830
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.273029257697713e-06,
+      "loss": 0.9215,
+      "step": 18840
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.260252970486777e-06,
+      "loss": 0.955,
+      "step": 18850
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.247476683275839e-06,
+      "loss": 0.9017,
+      "step": 18860
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.234700396064905e-06,
+      "loss": 0.909,
+      "step": 18870
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.221924108853967e-06,
+      "loss": 0.9449,
+      "step": 18880
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.209147821643031e-06,
+      "loss": 0.8397,
+      "step": 18890
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.196371534432095e-06,
+      "loss": 0.8534,
+      "step": 18900
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.183595247221158e-06,
+      "loss": 0.8681,
+      "step": 18910
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.170818960010222e-06,
+      "loss": 0.9161,
+      "step": 18920
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.158042672799286e-06,
+      "loss": 0.9247,
+      "step": 18930
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.145266385588348e-06,
+      "loss": 0.9136,
+      "step": 18940
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.132490098377412e-06,
+      "loss": 0.8871,
+      "step": 18950
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.119713811166475e-06,
+      "loss": 1.0471,
+      "step": 18960
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.106937523955539e-06,
+      "loss": 0.8436,
+      "step": 18970
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.094161236744603e-06,
+      "loss": 0.9205,
+      "step": 18980
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.081384949533665e-06,
+      "loss": 0.9451,
+      "step": 18990
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.068608662322729e-06,
+      "loss": 0.9911,
+      "step": 19000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 9.055832375111791e-06,
+      "loss": 0.9336,
+      "step": 19010
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 9.043056087900855e-06,
+      "loss": 0.9144,
+      "step": 19020
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 9.03027980068992e-06,
+      "loss": 0.9352,
+      "step": 19030
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 9.017503513478984e-06,
+      "loss": 0.9402,
+      "step": 19040
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 9.004727226268048e-06,
+      "loss": 0.942,
+      "step": 19050
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.991950939057112e-06,
+      "loss": 0.8221,
+      "step": 19060
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.979174651846174e-06,
+      "loss": 0.9079,
+      "step": 19070
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.966398364635238e-06,
+      "loss": 0.9532,
+      "step": 19080
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.9536220774243e-06,
+      "loss": 0.8565,
+      "step": 19090
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.940845790213365e-06,
+      "loss": 0.9071,
+      "step": 19100
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.928069503002429e-06,
+      "loss": 0.8813,
+      "step": 19110
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.915293215791491e-06,
+      "loss": 0.9561,
+      "step": 19120
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.902516928580555e-06,
+      "loss": 0.9081,
+      "step": 19130
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.889740641369617e-06,
+      "loss": 0.9111,
+      "step": 19140
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.876964354158681e-06,
+      "loss": 0.9259,
+      "step": 19150
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.864188066947745e-06,
+      "loss": 0.924,
+      "step": 19160
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.851411779736808e-06,
+      "loss": 0.881,
+      "step": 19170
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.838635492525872e-06,
+      "loss": 0.9141,
+      "step": 19180
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.825859205314934e-06,
+      "loss": 0.9058,
+      "step": 19190
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.813082918103998e-06,
+      "loss": 0.8971,
+      "step": 19200
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.800306630893062e-06,
+      "loss": 0.957,
+      "step": 19210
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.787530343682126e-06,
+      "loss": 0.9251,
+      "step": 19220
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.77475405647119e-06,
+      "loss": 0.9212,
+      "step": 19230
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.761977769260255e-06,
+      "loss": 0.8916,
+      "step": 19240
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.749201482049317e-06,
+      "loss": 0.9184,
+      "step": 19250
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.736425194838381e-06,
+      "loss": 0.9677,
+      "step": 19260
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.723648907627443e-06,
+      "loss": 0.8744,
+      "step": 19270
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.710872620416507e-06,
+      "loss": 0.9614,
+      "step": 19280
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.698096333205571e-06,
+      "loss": 0.8792,
+      "step": 19290
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.685320045994634e-06,
+      "loss": 0.9592,
+      "step": 19300
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.672543758783698e-06,
+      "loss": 0.8911,
+      "step": 19310
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.65976747157276e-06,
+      "loss": 0.891,
+      "step": 19320
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.646991184361824e-06,
+      "loss": 0.8937,
+      "step": 19330
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.634214897150888e-06,
+      "loss": 0.9239,
+      "step": 19340
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.62143860993995e-06,
+      "loss": 0.8465,
+      "step": 19350
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.608662322729015e-06,
+      "loss": 0.9427,
+      "step": 19360
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.595886035518077e-06,
+      "loss": 0.8904,
+      "step": 19370
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.583109748307141e-06,
+      "loss": 1.0235,
+      "step": 19380
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.570333461096207e-06,
+      "loss": 0.895,
+      "step": 19390
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.55755717388527e-06,
+      "loss": 0.8767,
+      "step": 19400
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.544780886674333e-06,
+      "loss": 0.9346,
+      "step": 19410
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.532004599463397e-06,
+      "loss": 0.9869,
+      "step": 19420
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.51922831225246e-06,
+      "loss": 0.9067,
+      "step": 19430
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.506452025041524e-06,
+      "loss": 0.8218,
+      "step": 19440
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.493675737830586e-06,
+      "loss": 0.8634,
+      "step": 19450
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.48089945061965e-06,
+      "loss": 0.9779,
+      "step": 19460
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.468123163408714e-06,
+      "loss": 0.9544,
+      "step": 19470
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.455346876197777e-06,
+      "loss": 0.9239,
+      "step": 19480
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.44257058898684e-06,
+      "loss": 0.921,
+      "step": 19490
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.429794301775903e-06,
+      "loss": 0.894,
+      "step": 19500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.417018014564967e-06,
+      "loss": 0.9174,
+      "step": 19510
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.404241727354031e-06,
+      "loss": 0.8581,
+      "step": 19520
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.391465440143094e-06,
+      "loss": 0.9115,
+      "step": 19530
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.378689152932158e-06,
+      "loss": 0.9198,
+      "step": 19540
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.365912865721222e-06,
+      "loss": 0.9195,
+      "step": 19550
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.353136578510286e-06,
+      "loss": 0.9593,
+      "step": 19560
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.34036029129935e-06,
+      "loss": 0.9765,
+      "step": 19570
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.327584004088412e-06,
+      "loss": 0.8636,
+      "step": 19580
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.314807716877476e-06,
+      "loss": 0.8528,
+      "step": 19590
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.30203142966654e-06,
+      "loss": 0.8767,
+      "step": 19600
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.289255142455603e-06,
+      "loss": 0.8928,
+      "step": 19610
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.276478855244667e-06,
+      "loss": 0.8565,
+      "step": 19620
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.263702568033729e-06,
+      "loss": 0.996,
+      "step": 19630
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.250926280822793e-06,
+      "loss": 0.9569,
+      "step": 19640
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.238149993611857e-06,
+      "loss": 0.916,
+      "step": 19650
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.22537370640092e-06,
+      "loss": 0.8914,
+      "step": 19660
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.212597419189984e-06,
+      "loss": 0.8659,
+      "step": 19670
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.199821131979046e-06,
+      "loss": 0.892,
+      "step": 19680
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 8.18704484476811e-06,
+      "loss": 0.9377,
+      "step": 19690
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 8.174268557557174e-06,
+      "loss": 0.9182,
+      "step": 19700
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 8.161492270346236e-06,
+      "loss": 0.9041,
+      "step": 19710
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 8.1487159831353e-06,
+      "loss": 0.8693,
+      "step": 19720
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 8.135939695924366e-06,
+      "loss": 0.8469,
+      "step": 19730
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 8.123163408713429e-06,
+      "loss": 0.9394,
+      "step": 19740
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 8.110387121502493e-06,
+      "loss": 0.9496,
+      "step": 19750
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 8.097610834291555e-06,
+      "loss": 0.8135,
+      "step": 19760
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 8.084834547080619e-06,
+      "loss": 0.9006,
+      "step": 19770
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 8.072058259869683e-06,
+      "loss": 0.8794,
+      "step": 19780
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 8.059281972658745e-06,
+      "loss": 0.8807,
+      "step": 19790
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 8.04650568544781e-06,
+      "loss": 0.9595,
+      "step": 19800
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 8.033729398236872e-06,
+      "loss": 0.9101,
+      "step": 19810
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 8.020953111025936e-06,
+      "loss": 0.8634,
+      "step": 19820
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 8.008176823815e-06,
+      "loss": 1.003,
+      "step": 19830
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.995400536604062e-06,
+      "loss": 0.9065,
+      "step": 19840
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.982624249393126e-06,
+      "loss": 0.9402,
+      "step": 19850
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.969847962182189e-06,
+      "loss": 0.9089,
+      "step": 19860
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.957071674971253e-06,
+      "loss": 0.9288,
+      "step": 19870
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.944295387760317e-06,
+      "loss": 0.9206,
+      "step": 19880
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.93151910054938e-06,
+      "loss": 0.8987,
+      "step": 19890
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.918742813338445e-06,
+      "loss": 0.8514,
+      "step": 19900
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.905966526127509e-06,
+      "loss": 0.9646,
+      "step": 19910
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.893190238916571e-06,
+      "loss": 0.8446,
+      "step": 19920
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.880413951705635e-06,
+      "loss": 0.9699,
+      "step": 19930
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.867637664494698e-06,
+      "loss": 0.9842,
+      "step": 19940
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.854861377283762e-06,
+      "loss": 0.9289,
+      "step": 19950
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.842085090072826e-06,
+      "loss": 0.8707,
+      "step": 19960
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.829308802861888e-06,
+      "loss": 0.9045,
+      "step": 19970
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.816532515650952e-06,
+      "loss": 0.8601,
+      "step": 19980
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.803756228440015e-06,
+      "loss": 0.8528,
+      "step": 19990
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.790979941229079e-06,
+      "loss": 0.896,
+      "step": 20000
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.778203654018143e-06,
+      "loss": 0.9011,
+      "step": 20010
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.765427366807205e-06,
+      "loss": 0.8921,
+      "step": 20020
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.75265107959627e-06,
+      "loss": 0.8781,
+      "step": 20030
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.739874792385332e-06,
+      "loss": 0.9161,
+      "step": 20040
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.727098505174396e-06,
+      "loss": 0.951,
+      "step": 20050
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.71432221796346e-06,
+      "loss": 0.8912,
+      "step": 20060
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.701545930752524e-06,
+      "loss": 0.8727,
+      "step": 20070
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.688769643541588e-06,
+      "loss": 0.8656,
+      "step": 20080
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.675993356330652e-06,
+      "loss": 0.878,
+      "step": 20090
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.663217069119714e-06,
+      "loss": 0.9238,
+      "step": 20100
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.650440781908778e-06,
+      "loss": 0.8911,
+      "step": 20110
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.63766449469784e-06,
+      "loss": 0.9215,
+      "step": 20120
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.624888207486905e-06,
+      "loss": 0.8791,
+      "step": 20130
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.612111920275968e-06,
+      "loss": 0.929,
+      "step": 20140
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.599335633065031e-06,
+      "loss": 0.8388,
+      "step": 20150
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.586559345854095e-06,
+      "loss": 0.9093,
+      "step": 20160
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.5737830586431584e-06,
+      "loss": 0.8885,
+      "step": 20170
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.561006771432222e-06,
+      "loss": 0.943,
+      "step": 20180
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.548230484221285e-06,
+      "loss": 0.8925,
+      "step": 20190
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.535454197010349e-06,
+      "loss": 0.9925,
+      "step": 20200
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.522677909799412e-06,
+      "loss": 0.9013,
+      "step": 20210
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.509901622588475e-06,
+      "loss": 0.9073,
+      "step": 20220
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.497125335377539e-06,
+      "loss": 0.9233,
+      "step": 20230
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.484349048166603e-06,
+      "loss": 0.9192,
+      "step": 20240
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.471572760955667e-06,
+      "loss": 0.9543,
+      "step": 20250
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.45879647374473e-06,
+      "loss": 0.9246,
+      "step": 20260
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.446020186533793e-06,
+      "loss": 0.8648,
+      "step": 20270
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.433243899322856e-06,
+      "loss": 0.9527,
+      "step": 20280
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.420467612111921e-06,
+      "loss": 0.9113,
+      "step": 20290
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.407691324900984e-06,
+      "loss": 0.9147,
+      "step": 20300
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.394915037690048e-06,
+      "loss": 0.9232,
+      "step": 20310
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.382138750479111e-06,
+      "loss": 0.8593,
+      "step": 20320
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.369362463268174e-06,
+      "loss": 0.8959,
+      "step": 20330
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.356586176057238e-06,
+      "loss": 0.9735,
+      "step": 20340
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.343809888846301e-06,
+      "loss": 0.957,
+      "step": 20350
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.3310336016353645e-06,
+      "loss": 0.8511,
+      "step": 20360
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.3182573144244286e-06,
+      "loss": 0.9581,
+      "step": 20370
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.305481027213493e-06,
+      "loss": 0.8906,
+      "step": 20380
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.292704740002556e-06,
+      "loss": 0.8391,
+      "step": 20390
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.279928452791619e-06,
+      "loss": 0.9566,
+      "step": 20400
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.267152165580682e-06,
+      "loss": 0.9526,
+      "step": 20410
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.2543758783697455e-06,
+      "loss": 0.9003,
+      "step": 20420
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.2415995911588095e-06,
+      "loss": 0.882,
+      "step": 20430
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.228823303947873e-06,
+      "loss": 0.9138,
+      "step": 20440
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.216047016736936e-06,
+      "loss": 0.9135,
+      "step": 20450
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.203270729526e-06,
+      "loss": 0.9206,
+      "step": 20460
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.190494442315064e-06,
+      "loss": 0.8777,
+      "step": 20470
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.177718155104127e-06,
+      "loss": 0.992,
+      "step": 20480
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.1649418678931905e-06,
+      "loss": 0.8647,
+      "step": 20490
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 7.152165580682254e-06,
+      "loss": 0.8841,
+      "step": 20500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 7.139389293471317e-06,
+      "loss": 0.8797,
+      "step": 20510
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 7.126613006260381e-06,
+      "loss": 0.9484,
+      "step": 20520
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 7.113836719049444e-06,
+      "loss": 0.857,
+      "step": 20530
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 7.101060431838507e-06,
+      "loss": 0.8707,
+      "step": 20540
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 7.088284144627571e-06,
+      "loss": 0.9466,
+      "step": 20550
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 7.0755078574166355e-06,
+      "loss": 0.9228,
+      "step": 20560
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 7.062731570205699e-06,
+      "loss": 0.8967,
+      "step": 20570
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 7.049955282994762e-06,
+      "loss": 0.9361,
+      "step": 20580
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 7.037178995783825e-06,
+      "loss": 0.9126,
+      "step": 20590
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 7.024402708572889e-06,
+      "loss": 0.9717,
+      "step": 20600
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 7.011626421361952e-06,
+      "loss": 0.9346,
+      "step": 20610
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.9988501341510156e-06,
+      "loss": 0.9664,
+      "step": 20620
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.98607384694008e-06,
+      "loss": 0.8919,
+      "step": 20630
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.9745751884502364e-06,
+      "loss": 0.8816,
+      "step": 20640
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.9617989012393e-06,
+      "loss": 1.0376,
+      "step": 20650
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.949022614028364e-06,
+      "loss": 0.9102,
+      "step": 20660
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.936246326817427e-06,
+      "loss": 0.9019,
+      "step": 20670
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.923470039606491e-06,
+      "loss": 0.8672,
+      "step": 20680
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.910693752395554e-06,
+      "loss": 0.846,
+      "step": 20690
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.897917465184617e-06,
+      "loss": 0.9165,
+      "step": 20700
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.8851411779736814e-06,
+      "loss": 0.927,
+      "step": 20710
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.872364890762745e-06,
+      "loss": 0.831,
+      "step": 20720
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.859588603551808e-06,
+      "loss": 0.8768,
+      "step": 20730
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.846812316340871e-06,
+      "loss": 0.9396,
+      "step": 20740
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.834036029129935e-06,
+      "loss": 0.9237,
+      "step": 20750
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.821259741918998e-06,
+      "loss": 0.9281,
+      "step": 20760
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.808483454708062e-06,
+      "loss": 0.9473,
+      "step": 20770
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.795707167497126e-06,
+      "loss": 0.8429,
+      "step": 20780
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.782930880286189e-06,
+      "loss": 0.9076,
+      "step": 20790
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.770154593075253e-06,
+      "loss": 0.8291,
+      "step": 20800
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.757378305864316e-06,
+      "loss": 0.9032,
+      "step": 20810
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.744602018653379e-06,
+      "loss": 0.9355,
+      "step": 20820
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.7318257314424425e-06,
+      "loss": 0.9126,
+      "step": 20830
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.7190494442315066e-06,
+      "loss": 0.8687,
+      "step": 20840
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.70627315702057e-06,
+      "loss": 0.8658,
+      "step": 20850
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.693496869809634e-06,
+      "loss": 0.8725,
+      "step": 20860
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.680720582598697e-06,
+      "loss": 0.9855,
+      "step": 20870
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.7598878853939582,
+      "eval_loss": 0.6982513070106506,
+      "eval_runtime": 298.5542,
+      "eval_samples_per_second": 139.817,
+      "eval_steps_per_second": 69.91,
+      "step": 20872
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.66794429538776e-06,
+      "loss": 0.9312,
+      "step": 20880
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.655168008176824e-06,
+      "loss": 0.8315,
+      "step": 20890
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.6423917209658875e-06,
+      "loss": 0.904,
+      "step": 20900
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.629615433754951e-06,
+      "loss": 0.9193,
+      "step": 20910
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.616839146544014e-06,
+      "loss": 0.8572,
+      "step": 20920
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.604062859333078e-06,
+      "loss": 0.9464,
+      "step": 20930
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.591286572122142e-06,
+      "loss": 0.8234,
+      "step": 20940
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.578510284911205e-06,
+      "loss": 0.8723,
+      "step": 20950
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.5657339977002685e-06,
+      "loss": 0.9415,
+      "step": 20960
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.552957710489332e-06,
+      "loss": 0.9802,
+      "step": 20970
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.540181423278396e-06,
+      "loss": 0.8766,
+      "step": 20980
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.527405136067459e-06,
+      "loss": 0.8674,
+      "step": 20990
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.514628848856522e-06,
+      "loss": 0.9455,
+      "step": 21000
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.501852561645585e-06,
+      "loss": 0.9649,
+      "step": 21010
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.489076274434649e-06,
+      "loss": 0.9409,
+      "step": 21020
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.4762999872237135e-06,
+      "loss": 0.9398,
+      "step": 21030
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.463523700012777e-06,
+      "loss": 0.855,
+      "step": 21040
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.45074741280184e-06,
+      "loss": 0.897,
+      "step": 21050
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.437971125590903e-06,
+      "loss": 0.8957,
+      "step": 21060
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.425194838379967e-06,
+      "loss": 0.9142,
+      "step": 21070
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.41241855116903e-06,
+      "loss": 0.8697,
+      "step": 21080
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.3996422639580936e-06,
+      "loss": 0.8454,
+      "step": 21090
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.386865976747157e-06,
+      "loss": 0.906,
+      "step": 21100
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.374089689536222e-06,
+      "loss": 0.9329,
+      "step": 21110
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.361313402325285e-06,
+      "loss": 0.8734,
+      "step": 21120
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.348537115114348e-06,
+      "loss": 0.9396,
+      "step": 21130
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.335760827903411e-06,
+      "loss": 0.9262,
+      "step": 21140
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.3229845406924745e-06,
+      "loss": 0.9217,
+      "step": 21150
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.310208253481539e-06,
+      "loss": 0.8764,
+      "step": 21160
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.297431966270602e-06,
+      "loss": 0.8846,
+      "step": 21170
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.284655679059665e-06,
+      "loss": 0.8609,
+      "step": 21180
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.271879391848728e-06,
+      "loss": 0.8395,
+      "step": 21190
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.259103104637793e-06,
+      "loss": 0.8809,
+      "step": 21200
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.246326817426856e-06,
+      "loss": 0.8261,
+      "step": 21210
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.2335505302159195e-06,
+      "loss": 0.9381,
+      "step": 21220
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.220774243004983e-06,
+      "loss": 0.9072,
+      "step": 21230
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.207997955794047e-06,
+      "loss": 0.9348,
+      "step": 21240
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.19522166858311e-06,
+      "loss": 0.8415,
+      "step": 21250
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.182445381372173e-06,
+      "loss": 0.9064,
+      "step": 21260
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.1696690941612364e-06,
+      "loss": 0.9454,
+      "step": 21270
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.1568928069503e-06,
+      "loss": 0.9834,
+      "step": 21280
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.1441165197393645e-06,
+      "loss": 0.8969,
+      "step": 21290
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.131340232528428e-06,
+      "loss": 0.9067,
+      "step": 21300
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.118563945317491e-06,
+      "loss": 0.8998,
+      "step": 21310
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.105787658106554e-06,
+      "loss": 0.8821,
+      "step": 21320
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.093011370895618e-06,
+      "loss": 0.877,
+      "step": 21330
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.0802350836846814e-06,
+      "loss": 0.9425,
+      "step": 21340
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.067458796473745e-06,
+      "loss": 0.9162,
+      "step": 21350
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.054682509262808e-06,
+      "loss": 0.8861,
+      "step": 21360
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.041906222051872e-06,
+      "loss": 0.9061,
+      "step": 21370
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.029129934840936e-06,
+      "loss": 0.8659,
+      "step": 21380
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.016353647629999e-06,
+      "loss": 0.9983,
+      "step": 21390
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.003577360419062e-06,
+      "loss": 0.9033,
+      "step": 21400
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.990801073208126e-06,
+      "loss": 0.8876,
+      "step": 21410
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.97802478599719e-06,
+      "loss": 0.86,
+      "step": 21420
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.965248498786253e-06,
+      "loss": 0.9695,
+      "step": 21430
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.952472211575316e-06,
+      "loss": 0.9871,
+      "step": 21440
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.939695924364379e-06,
+      "loss": 0.9573,
+      "step": 21450
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.926919637153443e-06,
+      "loss": 0.8751,
+      "step": 21460
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.914143349942507e-06,
+      "loss": 0.9682,
+      "step": 21470
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.901367062731571e-06,
+      "loss": 0.936,
+      "step": 21480
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.888590775520634e-06,
+      "loss": 0.9071,
+      "step": 21490
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.875814488309697e-06,
+      "loss": 0.8265,
+      "step": 21500
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.863038201098761e-06,
+      "loss": 0.8778,
+      "step": 21510
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.850261913887824e-06,
+      "loss": 0.9853,
+      "step": 21520
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.8374856266768875e-06,
+      "loss": 0.9725,
+      "step": 21530
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.8247093394659516e-06,
+      "loss": 0.9485,
+      "step": 21540
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.811933052255015e-06,
+      "loss": 0.8038,
+      "step": 21550
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.799156765044079e-06,
+      "loss": 0.8746,
+      "step": 21560
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.786380477833142e-06,
+      "loss": 0.9441,
+      "step": 21570
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.773604190622205e-06,
+      "loss": 0.876,
+      "step": 21580
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.7608279034112684e-06,
+      "loss": 0.9058,
+      "step": 21590
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.7480516162003325e-06,
+      "loss": 0.8759,
+      "step": 21600
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.735275328989396e-06,
+      "loss": 0.8642,
+      "step": 21610
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.722499041778459e-06,
+      "loss": 0.8023,
+      "step": 21620
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.709722754567523e-06,
+      "loss": 0.8719,
+      "step": 21630
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.696946467356586e-06,
+      "loss": 0.8601,
+      "step": 21640
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.68417018014565e-06,
+      "loss": 0.9119,
+      "step": 21650
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.6713938929347135e-06,
+      "loss": 0.8946,
+      "step": 21660
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.658617605723777e-06,
+      "loss": 0.8619,
+      "step": 21670
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.64584131851284e-06,
+      "loss": 0.8055,
+      "step": 21680
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.633065031301904e-06,
+      "loss": 0.9409,
+      "step": 21690
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.620288744090967e-06,
+      "loss": 0.8886,
+      "step": 21700
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.60751245688003e-06,
+      "loss": 0.9549,
+      "step": 21710
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.594736169669094e-06,
+      "loss": 0.8889,
+      "step": 21720
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.581959882458158e-06,
+      "loss": 0.8451,
+      "step": 21730
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.569183595247222e-06,
+      "loss": 0.9339,
+      "step": 21740
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.556407308036285e-06,
+      "loss": 0.8483,
+      "step": 21750
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.543631020825348e-06,
+      "loss": 0.8599,
+      "step": 21760
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.530854733614411e-06,
+      "loss": 0.7685,
+      "step": 21770
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.518078446403475e-06,
+      "loss": 0.932,
+      "step": 21780
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.5053021591925386e-06,
+      "loss": 0.9624,
+      "step": 21790
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.492525871981603e-06,
+      "loss": 0.8659,
+      "step": 21800
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.479749584770666e-06,
+      "loss": 0.9084,
+      "step": 21810
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.466973297559729e-06,
+      "loss": 0.912,
+      "step": 21820
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.454197010348793e-06,
+      "loss": 0.8673,
+      "step": 21830
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.441420723137856e-06,
+      "loss": 0.8506,
+      "step": 21840
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.4286444359269195e-06,
+      "loss": 0.7869,
+      "step": 21850
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.415868148715983e-06,
+      "loss": 0.9777,
+      "step": 21860
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.403091861505047e-06,
+      "loss": 0.9148,
+      "step": 21870
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.39031557429411e-06,
+      "loss": 0.8874,
+      "step": 21880
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.377539287083174e-06,
+      "loss": 0.8703,
+      "step": 21890
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.364762999872237e-06,
+      "loss": 0.9067,
+      "step": 21900
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.351986712661301e-06,
+      "loss": 0.8131,
+      "step": 21910
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.3392104254503645e-06,
+      "loss": 0.8348,
+      "step": 21920
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.326434138239428e-06,
+      "loss": 0.8441,
+      "step": 21930
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.313657851028491e-06,
+      "loss": 1.042,
+      "step": 21940
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.300881563817554e-06,
+      "loss": 0.8645,
+      "step": 21950
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.288105276606618e-06,
+      "loss": 0.8494,
+      "step": 21960
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.275328989395682e-06,
+      "loss": 0.9312,
+      "step": 21970
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.2625527021847455e-06,
+      "loss": 0.8624,
+      "step": 21980
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.249776414973809e-06,
+      "loss": 0.9628,
+      "step": 21990
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.237000127762873e-06,
+      "loss": 0.9118,
+      "step": 22000
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.224223840551936e-06,
+      "loss": 0.9334,
+      "step": 22010
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.211447553340999e-06,
+      "loss": 0.8442,
+      "step": 22020
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.198671266130062e-06,
+      "loss": 0.9181,
+      "step": 22030
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.185894978919126e-06,
+      "loss": 0.8719,
+      "step": 22040
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.17311869170819e-06,
+      "loss": 0.8794,
+      "step": 22050
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.160342404497254e-06,
+      "loss": 0.9218,
+      "step": 22060
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.147566117286317e-06,
+      "loss": 1.0097,
+      "step": 22070
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.13478983007538e-06,
+      "loss": 0.9816,
+      "step": 22080
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.122013542864444e-06,
+      "loss": 0.9781,
+      "step": 22090
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.109237255653507e-06,
+      "loss": 0.8809,
+      "step": 22100
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.096460968442571e-06,
+      "loss": 0.9249,
+      "step": 22110
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.083684681231634e-06,
+      "loss": 0.9203,
+      "step": 22120
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.070908394020697e-06,
+      "loss": 0.9798,
+      "step": 22130
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.058132106809762e-06,
+      "loss": 0.8419,
+      "step": 22140
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.045355819598825e-06,
+      "loss": 0.9809,
+      "step": 22150
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.032579532387888e-06,
+      "loss": 0.9398,
+      "step": 22160
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.0198032451769515e-06,
+      "loss": 0.9118,
+      "step": 22170
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.007026957966016e-06,
+      "loss": 0.8401,
+      "step": 22180
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.994250670755079e-06,
+      "loss": 0.9148,
+      "step": 22190
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.981474383544142e-06,
+      "loss": 0.8719,
+      "step": 22200
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.968698096333205e-06,
+      "loss": 0.8894,
+      "step": 22210
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.9559218091222684e-06,
+      "loss": 0.9964,
+      "step": 22220
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.943145521911333e-06,
+      "loss": 0.8524,
+      "step": 22230
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.9303692347003966e-06,
+      "loss": 0.9801,
+      "step": 22240
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.91759294748946e-06,
+      "loss": 0.9724,
+      "step": 22250
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.904816660278523e-06,
+      "loss": 0.8593,
+      "step": 22260
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.892040373067587e-06,
+      "loss": 0.8703,
+      "step": 22270
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.87926408585665e-06,
+      "loss": 0.9036,
+      "step": 22280
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.8664877986457134e-06,
+      "loss": 0.894,
+      "step": 22290
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.853711511434777e-06,
+      "loss": 0.7996,
+      "step": 22300
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.84093522422384e-06,
+      "loss": 0.8349,
+      "step": 22310
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.828158937012905e-06,
+      "loss": 0.8002,
+      "step": 22320
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.815382649801968e-06,
+      "loss": 0.9046,
+      "step": 22330
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.802606362591031e-06,
+      "loss": 0.968,
+      "step": 22340
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.789830075380094e-06,
+      "loss": 0.8509,
+      "step": 22350
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.7770537881691585e-06,
+      "loss": 0.8947,
+      "step": 22360
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.764277500958222e-06,
+      "loss": 0.7755,
+      "step": 22370
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.751501213747285e-06,
+      "loss": 0.8074,
+      "step": 22380
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.738724926536348e-06,
+      "loss": 0.8925,
+      "step": 22390
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.725948639325412e-06,
+      "loss": 0.866,
+      "step": 22400
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.713172352114476e-06,
+      "loss": 0.9228,
+      "step": 22410
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.700396064903539e-06,
+      "loss": 0.9365,
+      "step": 22420
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.687619777692603e-06,
+      "loss": 0.8845,
+      "step": 22430
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.674843490481666e-06,
+      "loss": 0.9089,
+      "step": 22440
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.66206720327073e-06,
+      "loss": 0.899,
+      "step": 22450
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.649290916059793e-06,
+      "loss": 0.8708,
+      "step": 22460
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.636514628848856e-06,
+      "loss": 0.8776,
+      "step": 22470
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.6237383416379195e-06,
+      "loss": 0.8305,
+      "step": 22480
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.6109620544269836e-06,
+      "loss": 0.8431,
+      "step": 22490
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.598185767216048e-06,
+      "loss": 0.9398,
+      "step": 22500
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.585409480005111e-06,
+      "loss": 0.9261,
+      "step": 22510
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.572633192794174e-06,
+      "loss": 0.8081,
+      "step": 22520
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.559856905583237e-06,
+      "loss": 0.9381,
+      "step": 22530
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.547080618372301e-06,
+      "loss": 0.8543,
+      "step": 22540
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.5343043311613645e-06,
+      "loss": 0.8721,
+      "step": 22550
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.521528043950428e-06,
+      "loss": 0.9618,
+      "step": 22560
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.508751756739492e-06,
+      "loss": 0.8853,
+      "step": 22570
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.495975469528556e-06,
+      "loss": 0.9033,
+      "step": 22580
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.483199182317619e-06,
+      "loss": 0.9493,
+      "step": 22590
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.470422895106682e-06,
+      "loss": 0.9213,
+      "step": 22600
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.4576466078957455e-06,
+      "loss": 0.8884,
+      "step": 22610
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.444870320684809e-06,
+      "loss": 0.8742,
+      "step": 22620
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.432094033473873e-06,
+      "loss": 0.9303,
+      "step": 22630
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.419317746262936e-06,
+      "loss": 0.8751,
+      "step": 22640
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.406541459051999e-06,
+      "loss": 0.8977,
+      "step": 22650
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.393765171841063e-06,
+      "loss": 0.916,
+      "step": 22660
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.380988884630127e-06,
+      "loss": 1.0154,
+      "step": 22670
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.3682125974191905e-06,
+      "loss": 0.9187,
+      "step": 22680
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.355436310208254e-06,
+      "loss": 0.9263,
+      "step": 22690
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.342660022997317e-06,
+      "loss": 0.8458,
+      "step": 22700
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.32988373578638e-06,
+      "loss": 0.9166,
+      "step": 22710
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.317107448575444e-06,
+      "loss": 0.8883,
+      "step": 22720
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.304331161364507e-06,
+      "loss": 0.9751,
+      "step": 22730
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.291554874153571e-06,
+      "loss": 0.8521,
+      "step": 22740
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.278778586942635e-06,
+      "loss": 0.9101,
+      "step": 22750
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.266002299731699e-06,
+      "loss": 0.9253,
+      "step": 22760
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.253226012520762e-06,
+      "loss": 0.8784,
+      "step": 22770
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.240449725309825e-06,
+      "loss": 0.928,
+      "step": 22780
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.227673438098888e-06,
+      "loss": 0.8649,
+      "step": 22790
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.2148971508879515e-06,
+      "loss": 0.8783,
+      "step": 22800
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.202120863677016e-06,
+      "loss": 0.9044,
+      "step": 22810
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.189344576466079e-06,
+      "loss": 0.8893,
+      "step": 22820
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.176568289255143e-06,
+      "loss": 0.8716,
+      "step": 22830
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.163792002044206e-06,
+      "loss": 0.9598,
+      "step": 22840
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.15101571483327e-06,
+      "loss": 0.7954,
+      "step": 22850
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.138239427622333e-06,
+      "loss": 0.8604,
+      "step": 22860
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.1254631404113965e-06,
+      "loss": 0.8798,
+      "step": 22870
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.11268685320046e-06,
+      "loss": 0.9116,
+      "step": 22880
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.099910565989523e-06,
+      "loss": 0.9244,
+      "step": 22890
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.087134278778587e-06,
+      "loss": 0.9035,
+      "step": 22900
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.07435799156765e-06,
+      "loss": 0.8316,
+      "step": 22910
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.061581704356714e-06,
+      "loss": 0.8543,
+      "step": 22920
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.0488054171457775e-06,
+      "loss": 0.8646,
+      "step": 22930
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.0360291299348416e-06,
+      "loss": 0.9101,
+      "step": 22940
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.023252842723905e-06,
+      "loss": 0.9426,
+      "step": 22950
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.010476555512968e-06,
+      "loss": 0.9152,
+      "step": 22960
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.997700268302031e-06,
+      "loss": 0.8991,
+      "step": 22970
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.984923981091094e-06,
+      "loss": 0.9125,
+      "step": 22980
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.9721476938801584e-06,
+      "loss": 0.8832,
+      "step": 22990
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.9593714066692225e-06,
+      "loss": 0.8985,
+      "step": 23000
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.946595119458286e-06,
+      "loss": 0.8441,
+      "step": 23010
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.933818832247349e-06,
+      "loss": 0.8681,
+      "step": 23020
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.921042545036413e-06,
+      "loss": 0.9633,
+      "step": 23030
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.908266257825476e-06,
+      "loss": 0.9412,
+      "step": 23040
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.895489970614539e-06,
+      "loss": 0.8792,
+      "step": 23050
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.882713683403603e-06,
+      "loss": 0.9904,
+      "step": 23060
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.869937396192666e-06,
+      "loss": 0.9428,
+      "step": 23070
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.85716110898173e-06,
+      "loss": 0.9492,
+      "step": 23080
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.844384821770794e-06,
+      "loss": 0.9462,
+      "step": 23090
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.831608534559857e-06,
+      "loss": 0.8152,
+      "step": 23100
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.81883224734892e-06,
+      "loss": 0.8701,
+      "step": 23110
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.806055960137984e-06,
+      "loss": 0.9181,
+      "step": 23120
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.7932796729270476e-06,
+      "loss": 0.7705,
+      "step": 23130
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.780503385716111e-06,
+      "loss": 0.8511,
+      "step": 23140
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.7677270985051745e-06,
+      "loss": 0.9121,
+      "step": 23150
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.7549508112942377e-06,
+      "loss": 0.8985,
+      "step": 23160
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.7421745240833013e-06,
+      "loss": 0.9216,
+      "step": 23170
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.729398236872365e-06,
+      "loss": 0.8353,
+      "step": 23180
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.716621949661428e-06,
+      "loss": 0.8624,
+      "step": 23190
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.703845662450492e-06,
+      "loss": 0.8525,
+      "step": 23200
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.6910693752395554e-06,
+      "loss": 0.8716,
+      "step": 23210
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.678293088028619e-06,
+      "loss": 0.8408,
+      "step": 23220
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.6655168008176823e-06,
+      "loss": 0.9026,
+      "step": 23230
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.6527405136067463e-06,
+      "loss": 0.902,
+      "step": 23240
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.6399642263958095e-06,
+      "loss": 0.9572,
+      "step": 23250
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.6271879391848727e-06,
+      "loss": 0.895,
+      "step": 23260
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.6144116519739364e-06,
+      "loss": 0.9128,
+      "step": 23270
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.601635364763e-06,
+      "loss": 0.9269,
+      "step": 23280
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5888590775520636e-06,
+      "loss": 0.9065,
+      "step": 23290
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.576082790341127e-06,
+      "loss": 0.8654,
+      "step": 23300
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5633065031301905e-06,
+      "loss": 0.8228,
+      "step": 23310
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5505302159192537e-06,
+      "loss": 1.054,
+      "step": 23320
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5377539287083177e-06,
+      "loss": 0.827,
+      "step": 23330
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.524977641497381e-06,
+      "loss": 0.8659,
+      "step": 23340
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5122013542864446e-06,
+      "loss": 0.9061,
+      "step": 23350
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4994250670755078e-06,
+      "loss": 0.9262,
+      "step": 23360
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.4866487798645714e-06,
+      "loss": 0.9465,
+      "step": 23370
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.473872492653635e-06,
+      "loss": 0.8705,
+      "step": 23380
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.4610962054426983e-06,
+      "loss": 0.7709,
+      "step": 23390
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.448319918231762e-06,
+      "loss": 0.9401,
+      "step": 23400
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.4355436310208255e-06,
+      "loss": 0.8963,
+      "step": 23410
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.422767343809889e-06,
+      "loss": 0.9124,
+      "step": 23420
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.4099910565989524e-06,
+      "loss": 0.867,
+      "step": 23430
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.397214769388016e-06,
+      "loss": 0.8558,
+      "step": 23440
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.3844384821770796e-06,
+      "loss": 0.9042,
+      "step": 23450
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.371662194966143e-06,
+      "loss": 0.8928,
+      "step": 23460
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.3588859077552065e-06,
+      "loss": 0.9059,
+      "step": 23470
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.3461096205442697e-06,
+      "loss": 0.9017,
+      "step": 23480
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.7608221737776394,
+      "eval_loss": 0.7128584384918213,
+      "eval_runtime": 301.0875,
+      "eval_samples_per_second": 138.641,
+      "eval_steps_per_second": 69.322,
+      "step": 23481
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.9507,
+      "step": 23490
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.320557046122397e-06,
+      "loss": 0.9181,
+      "step": 23500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.3077807589114606e-06,
+      "loss": 0.9228,
+      "step": 23510
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.295004471700524e-06,
+      "loss": 0.9009,
+      "step": 23520
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.2822281844895874e-06,
+      "loss": 0.8826,
+      "step": 23530
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.269451897278651e-06,
+      "loss": 0.8958,
+      "step": 23540
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.2566756100677143e-06,
+      "loss": 0.9058,
+      "step": 23550
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.243899322856778e-06,
+      "loss": 0.8906,
+      "step": 23560
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.231123035645841e-06,
+      "loss": 0.9041,
+      "step": 23570
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.218346748434905e-06,
+      "loss": 0.886,
+      "step": 23580
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.2055704612239684e-06,
+      "loss": 0.8974,
+      "step": 23590
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.192794174013032e-06,
+      "loss": 0.9301,
+      "step": 23600
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.1800178868020952e-06,
+      "loss": 0.9374,
+      "step": 23610
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.167241599591159e-06,
+      "loss": 0.7817,
+      "step": 23620
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.1544653123802225e-06,
+      "loss": 0.8456,
+      "step": 23630
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.1416890251692857e-06,
+      "loss": 0.8351,
+      "step": 23640
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.1289127379583493e-06,
+      "loss": 0.9484,
+      "step": 23650
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.1161364507474125e-06,
+      "loss": 0.8637,
+      "step": 23660
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.1033601635364766e-06,
+      "loss": 0.8785,
+      "step": 23670
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.09058387632554e-06,
+      "loss": 0.9531,
+      "step": 23680
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0778075891146034e-06,
+      "loss": 0.886,
+      "step": 23690
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0650313019036667e-06,
+      "loss": 0.8663,
+      "step": 23700
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.0522550146927307e-06,
+      "loss": 0.8406,
+      "step": 23710
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.039478727481794e-06,
+      "loss": 0.8344,
+      "step": 23720
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.0267024402708576e-06,
+      "loss": 0.9386,
+      "step": 23730
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.0139261530599208e-06,
+      "loss": 0.8994,
+      "step": 23740
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.001149865848984e-06,
+      "loss": 0.9545,
+      "step": 23750
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.988373578638048e-06,
+      "loss": 0.8869,
+      "step": 23760
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9755972914271112e-06,
+      "loss": 0.9274,
+      "step": 23770
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.962821004216175e-06,
+      "loss": 0.9779,
+      "step": 23780
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.950044717005238e-06,
+      "loss": 0.9058,
+      "step": 23790
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.937268429794302e-06,
+      "loss": 0.8455,
+      "step": 23800
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.9244921425833653e-06,
+      "loss": 0.85,
+      "step": 23810
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.911715855372429e-06,
+      "loss": 0.9097,
+      "step": 23820
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.898939568161492e-06,
+      "loss": 0.8656,
+      "step": 23830
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.886163280950556e-06,
+      "loss": 0.8325,
+      "step": 23840
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8733869937396195e-06,
+      "loss": 0.8249,
+      "step": 23850
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.8606107065286827e-06,
+      "loss": 0.9265,
+      "step": 23860
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.8478344193177463e-06,
+      "loss": 0.902,
+      "step": 23870
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.83505813210681e-06,
+      "loss": 0.9086,
+      "step": 23880
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.8222818448958736e-06,
+      "loss": 0.9146,
+      "step": 23890
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.8095055576849368e-06,
+      "loss": 0.9267,
+      "step": 23900
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7967292704740004e-06,
+      "loss": 0.8712,
+      "step": 23910
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7839529832630636e-06,
+      "loss": 0.8805,
+      "step": 23920
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7711766960521272e-06,
+      "loss": 0.8443,
+      "step": 23930
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.758400408841191e-06,
+      "loss": 0.8683,
+      "step": 23940
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.745624121630254e-06,
+      "loss": 0.9162,
+      "step": 23950
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.7328478344193177e-06,
+      "loss": 0.8848,
+      "step": 23960
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.7200715472083814e-06,
+      "loss": 0.9599,
+      "step": 23970
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.707295259997445e-06,
+      "loss": 0.82,
+      "step": 23980
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.694518972786508e-06,
+      "loss": 0.9204,
+      "step": 23990
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.681742685575572e-06,
+      "loss": 0.8583,
+      "step": 24000
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6689663983646355e-06,
+      "loss": 0.9028,
+      "step": 24010
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.656190111153699e-06,
+      "loss": 0.8569,
+      "step": 24020
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.6434138239427623e-06,
+      "loss": 0.7817,
+      "step": 24030
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.6306375367318255e-06,
+      "loss": 0.8167,
+      "step": 24040
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.617861249520889e-06,
+      "loss": 0.8659,
+      "step": 24050
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.6050849623099528e-06,
+      "loss": 0.8956,
+      "step": 24060
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5923086750990164e-06,
+      "loss": 0.8504,
+      "step": 24070
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5795323878880796e-06,
+      "loss": 0.9213,
+      "step": 24080
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5667561006771433e-06,
+      "loss": 0.9327,
+      "step": 24090
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.553979813466207e-06,
+      "loss": 0.8836,
+      "step": 24100
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.5412035262552705e-06,
+      "loss": 0.917,
+      "step": 24110
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.5284272390443337e-06,
+      "loss": 0.8959,
+      "step": 24120
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.515650951833397e-06,
+      "loss": 0.8548,
+      "step": 24130
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.502874664622461e-06,
+      "loss": 0.9371,
+      "step": 24140
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4900983774115242e-06,
+      "loss": 0.8789,
+      "step": 24150
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.477322090200588e-06,
+      "loss": 0.8334,
+      "step": 24160
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.464545802989651e-06,
+      "loss": 0.9277,
+      "step": 24170
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.451769515778715e-06,
+      "loss": 0.9204,
+      "step": 24180
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.4389932285677783e-06,
+      "loss": 0.8449,
+      "step": 24190
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.426216941356842e-06,
+      "loss": 0.9497,
+      "step": 24200
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.413440654145905e-06,
+      "loss": 0.9056,
+      "step": 24210
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.4006643669349684e-06,
+      "loss": 0.8841,
+      "step": 24220
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3878880797240324e-06,
+      "loss": 0.8616,
+      "step": 24230
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3751117925130956e-06,
+      "loss": 0.8244,
+      "step": 24240
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3623355053021593e-06,
+      "loss": 0.8809,
+      "step": 24250
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.3495592180912225e-06,
+      "loss": 0.9184,
+      "step": 24260
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.3367829308802865e-06,
+      "loss": 0.8938,
+      "step": 24270
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.3240066436693497e-06,
+      "loss": 0.8416,
+      "step": 24280
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.3112303564584134e-06,
+      "loss": 0.8895,
+      "step": 24290
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2984540692474766e-06,
+      "loss": 0.8441,
+      "step": 24300
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2856777820365402e-06,
+      "loss": 0.868,
+      "step": 24310
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.272901494825604e-06,
+      "loss": 0.9158,
+      "step": 24320
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.260125207614667e-06,
+      "loss": 0.8591,
+      "step": 24330
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.2473489204037307e-06,
+      "loss": 0.8936,
+      "step": 24340
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.234572633192794e-06,
+      "loss": 0.8518,
+      "step": 24350
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.221796345981858e-06,
+      "loss": 0.8639,
+      "step": 24360
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.209020058770921e-06,
+      "loss": 0.9183,
+      "step": 24370
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.196243771559985e-06,
+      "loss": 0.8916,
+      "step": 24380
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.183467484349048e-06,
+      "loss": 0.9673,
+      "step": 24390
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.170691197138112e-06,
+      "loss": 0.9182,
+      "step": 24400
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1579149099271753e-06,
+      "loss": 0.917,
+      "step": 24410
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1451386227162385e-06,
+      "loss": 0.9187,
+      "step": 24420
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.132362335505302e-06,
+      "loss": 0.8496,
+      "step": 24430
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.1195860482943658e-06,
+      "loss": 0.8817,
+      "step": 24440
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.1068097610834294e-06,
+      "loss": 0.8311,
+      "step": 24450
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0940334738724926e-06,
+      "loss": 0.8131,
+      "step": 24460
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0812571866615562e-06,
+      "loss": 0.8677,
+      "step": 24470
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0684808994506194e-06,
+      "loss": 0.8458,
+      "step": 24480
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.0557046122396835e-06,
+      "loss": 0.8945,
+      "step": 24490
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.0429283250287467e-06,
+      "loss": 0.9614,
+      "step": 24500
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.03015203781781e-06,
+      "loss": 0.9041,
+      "step": 24510
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.0173757506068736e-06,
+      "loss": 0.874,
+      "step": 24520
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.004599463395937e-06,
+      "loss": 0.7861,
+      "step": 24530
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.991823176185001e-06,
+      "loss": 0.8658,
+      "step": 24540
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.979046888974064e-06,
+      "loss": 0.8822,
+      "step": 24550
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9662706017631277e-06,
+      "loss": 0.9477,
+      "step": 24560
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9534943145521913e-06,
+      "loss": 0.8248,
+      "step": 24570
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.940718027341255e-06,
+      "loss": 0.8301,
+      "step": 24580
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.927941740130318e-06,
+      "loss": 0.8404,
+      "step": 24590
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.9151654529193813e-06,
+      "loss": 0.9288,
+      "step": 24600
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.9023891657084454e-06,
+      "loss": 0.8764,
+      "step": 24610
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8896128784975088e-06,
+      "loss": 0.8312,
+      "step": 24620
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8768365912865722e-06,
+      "loss": 0.9223,
+      "step": 24630
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8640603040756357e-06,
+      "loss": 0.9138,
+      "step": 24640
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.851284016864699e-06,
+      "loss": 0.8945,
+      "step": 24650
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8385077296537625e-06,
+      "loss": 0.8871,
+      "step": 24660
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8257314424428261e-06,
+      "loss": 0.8953,
+      "step": 24670
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8129551552318896e-06,
+      "loss": 0.8712,
+      "step": 24680
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8001788680209532e-06,
+      "loss": 0.7844,
+      "step": 24690
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7874025808100166e-06,
+      "loss": 0.8303,
+      "step": 24700
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7746262935990803e-06,
+      "loss": 0.8612,
+      "step": 24710
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7618500063881437e-06,
+      "loss": 0.8528,
+      "step": 24720
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7490737191772073e-06,
+      "loss": 0.88,
+      "step": 24730
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7362974319662705e-06,
+      "loss": 0.8481,
+      "step": 24740
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.7235211447553342e-06,
+      "loss": 0.9088,
+      "step": 24750
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.7107448575443976e-06,
+      "loss": 0.9231,
+      "step": 24760
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6979685703334612e-06,
+      "loss": 0.9253,
+      "step": 24770
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6851922831225246e-06,
+      "loss": 0.8375,
+      "step": 24780
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.672415995911588e-06,
+      "loss": 0.8787,
+      "step": 24790
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6596397087006517e-06,
+      "loss": 0.8975,
+      "step": 24800
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.646863421489715e-06,
+      "loss": 0.8286,
+      "step": 24810
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6340871342787787e-06,
+      "loss": 0.8539,
+      "step": 24820
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.6213108470678422e-06,
+      "loss": 0.8729,
+      "step": 24830
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.6085345598569056e-06,
+      "loss": 0.8833,
+      "step": 24840
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.595758272645969e-06,
+      "loss": 0.9273,
+      "step": 24850
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5829819854350326e-06,
+      "loss": 0.9133,
+      "step": 24860
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.570205698224096e-06,
+      "loss": 0.9569,
+      "step": 24870
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5574294110131597e-06,
+      "loss": 0.8787,
+      "step": 24880
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5446531238022231e-06,
+      "loss": 0.8777,
+      "step": 24890
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5331544653123803e-06,
+      "loss": 0.9187,
+      "step": 24900
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.5203781781014438e-06,
+      "loss": 0.9011,
+      "step": 24910
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.5076018908905072e-06,
+      "loss": 0.984,
+      "step": 24920
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4948256036795708e-06,
+      "loss": 0.9167,
+      "step": 24930
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4820493164686342e-06,
+      "loss": 0.7452,
+      "step": 24940
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4692730292576977e-06,
+      "loss": 0.9521,
+      "step": 24950
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4564967420467613e-06,
+      "loss": 0.934,
+      "step": 24960
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4437204548358247e-06,
+      "loss": 0.9003,
+      "step": 24970
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4309441676248883e-06,
+      "loss": 0.8777,
+      "step": 24980
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.4181678804139518e-06,
+      "loss": 0.9531,
+      "step": 24990
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.4053915932030152e-06,
+      "loss": 0.9781,
+      "step": 25000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3926153059920786e-06,
+      "loss": 0.8931,
+      "step": 25010
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3798390187811422e-06,
+      "loss": 0.8565,
+      "step": 25020
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3670627315702057e-06,
+      "loss": 0.9222,
+      "step": 25030
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3542864443592693e-06,
+      "loss": 0.8402,
+      "step": 25040
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3415101571483327e-06,
+      "loss": 0.8715,
+      "step": 25050
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3287338699373963e-06,
+      "loss": 0.852,
+      "step": 25060
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3159575827264598e-06,
+      "loss": 0.8933,
+      "step": 25070
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3031812955155232e-06,
+      "loss": 0.8523,
+      "step": 25080
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2904050083045868e-06,
+      "loss": 0.8297,
+      "step": 25090
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.27762872109365e-06,
+      "loss": 0.9061,
+      "step": 25100
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2648524338827137e-06,
+      "loss": 0.858,
+      "step": 25110
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.252076146671777e-06,
+      "loss": 0.8085,
+      "step": 25120
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2392998594608407e-06,
+      "loss": 0.8008,
+      "step": 25130
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.2265235722499041e-06,
+      "loss": 0.9034,
+      "step": 25140
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.2137472850389678e-06,
+      "loss": 0.9067,
+      "step": 25150
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.2009709978280312e-06,
+      "loss": 0.9019,
+      "step": 25160
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1881947106170948e-06,
+      "loss": 0.9576,
+      "step": 25170
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1754184234061583e-06,
+      "loss": 0.9147,
+      "step": 25180
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1626421361952219e-06,
+      "loss": 0.9099,
+      "step": 25190
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.149865848984285e-06,
+      "loss": 0.8634,
+      "step": 25200
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1370895617733487e-06,
+      "loss": 0.9054,
+      "step": 25210
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1243132745624121e-06,
+      "loss": 0.9334,
+      "step": 25220
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1115369873514756e-06,
+      "loss": 0.8752,
+      "step": 25230
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0987607001405392e-06,
+      "loss": 0.9368,
+      "step": 25240
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0859844129296026e-06,
+      "loss": 0.8216,
+      "step": 25250
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0732081257186663e-06,
+      "loss": 0.9001,
+      "step": 25260
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0604318385077297e-06,
+      "loss": 0.9092,
+      "step": 25270
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0476555512967933e-06,
+      "loss": 0.9125,
+      "step": 25280
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0348792640858565e-06,
+      "loss": 0.8771,
+      "step": 25290
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.0221029768749202e-06,
+      "loss": 0.903,
+      "step": 25300
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.0093266896639836e-06,
+      "loss": 0.7834,
+      "step": 25310
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.965504024530472e-07,
+      "loss": 0.8227,
+      "step": 25320
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.837741152421106e-07,
+      "loss": 0.8716,
+      "step": 25330
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.709978280311743e-07,
+      "loss": 0.802,
+      "step": 25340
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.582215408202377e-07,
+      "loss": 0.8235,
+      "step": 25350
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.454452536093012e-07,
+      "loss": 0.9491,
+      "step": 25360
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.326689663983646e-07,
+      "loss": 0.892,
+      "step": 25370
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 9.198926791874282e-07,
+      "loss": 0.9325,
+      "step": 25380
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 9.071163919764917e-07,
+      "loss": 0.9197,
+      "step": 25390
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.943401047655552e-07,
+      "loss": 0.885,
+      "step": 25400
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.815638175546186e-07,
+      "loss": 0.8702,
+      "step": 25410
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.687875303436822e-07,
+      "loss": 0.8566,
+      "step": 25420
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.560112431327457e-07,
+      "loss": 0.8361,
+      "step": 25430
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.432349559218092e-07,
+      "loss": 0.8913,
+      "step": 25440
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.304586687108726e-07,
+      "loss": 0.9078,
+      "step": 25450
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 8.176823814999361e-07,
+      "loss": 0.9227,
+      "step": 25460
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 8.049060942889996e-07,
+      "loss": 0.8385,
+      "step": 25470
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.921298070780631e-07,
+      "loss": 0.8668,
+      "step": 25480
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.793535198671266e-07,
+      "loss": 0.8811,
+      "step": 25490
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.665772326561902e-07,
+      "loss": 0.9254,
+      "step": 25500
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.538009454452536e-07,
+      "loss": 0.8265,
+      "step": 25510
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.410246582343171e-07,
+      "loss": 0.9664,
+      "step": 25520
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.282483710233806e-07,
+      "loss": 0.9129,
+      "step": 25530
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.154720838124442e-07,
+      "loss": 0.923,
+      "step": 25540
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.026957966015076e-07,
+      "loss": 0.8464,
+      "step": 25550
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.899195093905711e-07,
+      "loss": 0.9125,
+      "step": 25560
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.771432221796346e-07,
+      "loss": 0.8628,
+      "step": 25570
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.643669349686982e-07,
+      "loss": 0.9041,
+      "step": 25580
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.515906477577616e-07,
+      "loss": 0.7553,
+      "step": 25590
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.38814360546825e-07,
+      "loss": 0.9134,
+      "step": 25600
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.260380733358885e-07,
+      "loss": 0.8363,
+      "step": 25610
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.132617861249521e-07,
+      "loss": 0.9235,
+      "step": 25620
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.004854989140156e-07,
+      "loss": 0.8643,
+      "step": 25630
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.877092117030791e-07,
+      "loss": 0.8314,
+      "step": 25640
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.749329244921425e-07,
+      "loss": 0.8884,
+      "step": 25650
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.621566372812061e-07,
+      "loss": 0.9008,
+      "step": 25660
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.493803500702696e-07,
+      "loss": 0.836,
+      "step": 25670
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.366040628593331e-07,
+      "loss": 0.9073,
+      "step": 25680
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.238277756483967e-07,
+      "loss": 0.8563,
+      "step": 25690
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.110514884374601e-07,
+      "loss": 0.8679,
+      "step": 25700
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.982752012265236e-07,
+      "loss": 0.8408,
+      "step": 25710
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.854989140155871e-07,
+      "loss": 0.8618,
+      "step": 25720
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.727226268046506e-07,
+      "loss": 0.8408,
+      "step": 25730
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.599463395937141e-07,
+      "loss": 0.9157,
+      "step": 25740
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.471700523827776e-07,
+      "loss": 0.846,
+      "step": 25750
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.343937651718411e-07,
+      "loss": 0.8534,
+      "step": 25760
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.216174779609046e-07,
+      "loss": 0.8676,
+      "step": 25770
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.0884119074996803e-07,
+      "loss": 0.8592,
+      "step": 25780
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.9606490353903156e-07,
+      "loss": 0.8271,
+      "step": 25790
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.832886163280951e-07,
+      "loss": 0.8538,
+      "step": 25800
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.7051232911715856e-07,
+      "loss": 0.8675,
+      "step": 25810
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.577360419062221e-07,
+      "loss": 0.9328,
+      "step": 25820
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.4495975469528556e-07,
+      "loss": 0.8798,
+      "step": 25830
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.321834674843491e-07,
+      "loss": 0.9539,
+      "step": 25840
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.194071802734125e-07,
+      "loss": 0.833,
+      "step": 25850
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.0663089306247604e-07,
+      "loss": 0.8749,
+      "step": 25860
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.9385460585153956e-07,
+      "loss": 0.8933,
+      "step": 25870
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.8107831864060304e-07,
+      "loss": 0.8271,
+      "step": 25880
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.6830203142966656e-07,
+      "loss": 0.8959,
+      "step": 25890
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.5552574421873004e-07,
+      "loss": 0.832,
+      "step": 25900
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.4274945700779357e-07,
+      "loss": 0.8726,
+      "step": 25910
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.2997316979685704e-07,
+      "loss": 0.9485,
+      "step": 25920
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.1719688258592054e-07,
+      "loss": 0.8342,
+      "step": 25930
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.0442059537498401e-07,
+      "loss": 0.9169,
+      "step": 25940
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.9164430816404754e-07,
+      "loss": 0.7973,
+      "step": 25950
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.7886802095311104e-07,
+      "loss": 0.8861,
+      "step": 25960
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6609173374217454e-07,
+      "loss": 0.8022,
+      "step": 25970
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.5331544653123802e-07,
+      "loss": 0.776,
+      "step": 25980
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.4053915932030152e-07,
+      "loss": 0.8827,
+      "step": 25990
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.2776287210936502e-07,
+      "loss": 0.799,
+      "step": 26000
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.1498658489842852e-07,
+      "loss": 0.8771,
+      "step": 26010
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0221029768749201e-07,
+      "loss": 0.848,
+      "step": 26020
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.943401047655552e-08,
+      "loss": 0.9332,
+      "step": 26030
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.665772326561901e-08,
+      "loss": 0.9762,
+      "step": 26040
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.388143605468251e-08,
+      "loss": 0.8634,
+      "step": 26050
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.1105148843746004e-08,
+      "loss": 0.7815,
+      "step": 26060
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.8328861632809504e-08,
+      "loss": 0.9461,
+      "step": 26070
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.5552574421873002e-08,
+      "loss": 0.8885,
+      "step": 26080
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.2776287210936501e-08,
+      "loss": 0.8495,
+      "step": 26090
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.7662602112929114,
+      "eval_loss": 0.7000287771224976,
+      "eval_runtime": 299.7126,
+      "eval_samples_per_second": 139.277,
+      "eval_steps_per_second": 69.64,
+      "step": 26090
+    },
+    {
+      "epoch": 10.0,
+      "step": 26090,
+      "total_flos": 1.6966966119854244e+19,
+      "train_loss": 1.0529476437910437,
+      "train_runtime": 11696.8841,
+      "train_samples_per_second": 142.747,
+      "train_steps_per_second": 2.231
+    }
+  ],
+  "max_steps": 26090,
+  "num_train_epochs": 10,
+  "total_flos": 1.6966966119854244e+19,
+  "trial_name": null,
+  "trial_params": null
+}