diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,34067 @@
+{
+  "best_metric": 2.486668109893799,
+  "best_model_checkpoint": "checkpoints-mistral-300M/checkpoint-56000",
+  "epoch": 7.5630581204949445,
+  "eval_steps": 1000,
+  "global_step": 56000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00029999999783976374,
+      "loss": 9.6883,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999999846383203,
+      "loss": 8.3424,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999999594355658,
+      "loss": 7.6595,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999999222315021,
+      "loss": 7.402,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002999998730261323,
+      "loss": 7.3015,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999981181946035,
+      "loss": 7.2386,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999973861149105,
+      "loss": 7.182,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999965340223034,
+      "loss": 7.0721,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299999556191685,
+      "loss": 7.0263,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999994469798628,
+      "loss": 6.8929,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999993257667725,
+      "loss": 6.8313,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999991925524238,
+      "loss": 6.7008,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999990473368273,
+      "loss": 6.6717,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002999988901199947,
+      "loss": 6.6225,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999872090193856,
+      "loss": 6.4815,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029999853968267235,
+      "loss": 6.454,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999983464622107,
+      "loss": 6.3762,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299998141240569,
+      "loss": 6.3065,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999979240177636,
+      "loss": 6.2371,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029999769479381193,
+      "loss": 6.2123,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999974535687324,
+      "loss": 6.2118,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999972003425442,
+      "loss": 6.0799,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999969351152677,
+      "loss": 6.0385,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000299996657886924,
+      "loss": 6.0548,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999963686575354,
+      "loss": 6.029,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029999606742712494,
+      "loss": 5.9354,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002999957541957168,
+      "loss": 5.8861,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299995428963336,
+      "loss": 5.8419,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002999950917300085,
+      "loss": 5.8366,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002999947424957614,
+      "loss": 5.7756,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002999943812606226,
+      "loss": 5.7612,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029999400802462105,
+      "loss": 5.7243,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002999936227877865,
+      "loss": 5.6925,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002999932255501498,
+      "loss": 5.6749,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029999281631174277,
+      "loss": 5.6479,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029999239507259824,
+      "loss": 5.5956,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029999196183274967,
+      "loss": 5.6214,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000299991516592232,
+      "loss": 5.5289,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029999105935108063,
+      "loss": 5.5442,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029999059010933225,
+      "loss": 5.4976,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002999901088670244,
+      "loss": 5.4568,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002999896156241956,
+      "loss": 5.4337,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002999891103808853,
+      "loss": 5.4593,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002999885931371338,
+      "loss": 5.4165,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029998806389298265,
+      "loss": 5.3953,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002999875226484742,
+      "loss": 5.3275,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002999869694036516,
+      "loss": 5.3869,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002999864041585593,
+      "loss": 5.2817,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029998582691324237,
+      "loss": 5.3198,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999852376677471,
+      "loss": 5.2841,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029998463642212056,
+      "loss": 5.2466,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029998402317641087,
+      "loss": 5.2742,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029998339793066713,
+      "loss": 5.2403,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999827606849393,
+      "loss": 5.2287,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999821114392785,
+      "loss": 5.2258,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029998145019373656,
+      "loss": 5.1613,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029998077694836636,
+      "loss": 5.1557,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999800917032219,
+      "loss": 5.183,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999793944583578,
+      "loss": 5.1233,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999786852138301,
+      "loss": 5.0988,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029997796396969534,
+      "loss": 5.0862,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999772307260113,
+      "loss": 5.1313,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999764854828367,
+      "loss": 5.0287,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029997572824023104,
+      "loss": 4.9919,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000299974958998255,
+      "loss": 5.0137,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029997417775697017,
+      "loss": 4.9898,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999733845164389,
+      "loss": 4.9773,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999725792767248,
+      "loss": 4.9459,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999717620378922,
+      "loss": 4.9434,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999709328000066,
+      "loss": 4.9227,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029997009156313423,
+      "loss": 4.9197,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999692383273425,
+      "loss": 4.933,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999683730926995,
+      "loss": 4.916,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029996749585927464,
+      "loss": 4.9109,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999666066271381,
+      "loss": 4.8736,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029996570539636086,
+      "loss": 4.8532,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999647921670152,
+      "loss": 4.8317,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029996386693917414,
+      "loss": 4.803,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029996292971291155,
+      "loss": 4.7949,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999619804883027,
+      "loss": 4.7724,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029996101926542336,
+      "loss": 4.7779,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029996004604435046,
+      "loss": 4.7834,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029995906082516183,
+      "loss": 4.7456,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029995806360793637,
+      "loss": 4.7195,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999570543927538,
+      "loss": 4.7132,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999560331796949,
+      "loss": 4.7051,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029995499996884146,
+      "loss": 4.6911,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000299953954760276,
+      "loss": 4.6524,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999528975540822,
+      "loss": 4.6565,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029995182835034466,
+      "loss": 4.6357,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999507471491489,
+      "loss": 4.6317,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999496539505815,
+      "loss": 4.5489,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029994854875472983,
+      "loss": 4.5908,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029994743156168234,
+      "loss": 4.5434,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994630237152847,
+      "loss": 4.5747,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994516118435853,
+      "loss": 4.552,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994400800026377,
+      "loss": 4.5238,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994284281933655,
+      "loss": 4.5142,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994166564167006,
+      "loss": 4.4698,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999404764673584,
+      "loss": 4.5141,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 4.515992641448975,
+      "eval_runtime": 1262.2765,
+      "eval_samples_per_second": 60.316,
+      "eval_steps_per_second": 7.54,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029993927529649685,
+      "loss": 4.4932,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029993806212918144,
+      "loss": 4.5063,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029993683696550926,
+      "loss": 4.4059,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999355998055783,
+      "loss": 4.4535,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029993435064948755,
+      "loss": 4.4165,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000299933089497337,
+      "loss": 4.4083,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993181634922745,
+      "loss": 4.4295,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993053120526094,
+      "loss": 4.3997,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999292340655401,
+      "loss": 4.3731,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999279249301689,
+      "loss": 4.3212,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992660379925186,
+      "loss": 4.3279,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999252706728949,
+      "loss": 4.341,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992392555120453,
+      "loss": 4.3371,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992256843428846,
+      "loss": 4.2964,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992119932225524,
+      "loss": 4.3179,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991981821521433,
+      "loss": 4.3036,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029991842511327643,
+      "loss": 4.3108,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999170200165528,
+      "loss": 4.2924,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029991560292515597,
+      "loss": 4.2827,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999141738391993,
+      "loss": 4.2726,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029991273275879705,
+      "loss": 4.2584,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999112796840647,
+      "loss": 4.18,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990981461511826,
+      "loss": 4.2057,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999083375520752,
+      "loss": 4.2783,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999068484950535,
+      "loss": 4.2394,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990534744417244,
+      "loss": 4.2069,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999038343995521,
+      "loss": 4.19,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002999023093613134,
+      "loss": 4.1566,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029990077232957846,
+      "loss": 4.2053,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998992233044703,
+      "loss": 4.1807,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029989766228611276,
+      "loss": 4.1221,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998960892746308,
+      "loss": 4.1975,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029989450427015025,
+      "loss": 4.1027,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998929072727979,
+      "loss": 4.1417,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998912982827016,
+      "loss": 4.1774,
+      "step": 1350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029988967729999,
+      "loss": 4.1386,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029988804432479286,
+      "loss": 4.1332,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029988639935724077,
+      "loss": 4.1044,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998847423974653,
+      "loss": 4.1211,
+      "step": 1390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998830734455992,
+      "loss": 4.1028,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029988139250177587,
+      "loss": 4.114,
+      "step": 1410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998796995661298,
+      "loss": 4.1213,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998779946387965,
+      "loss": 4.1061,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998762777199123,
+      "loss": 4.0378,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998745488096146,
+      "loss": 4.0598,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998728079080418,
+      "loss": 4.0914,
+      "step": 1460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029987105501533306,
+      "loss": 4.0454,
+      "step": 1470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998692901316287,
+      "loss": 4.0619,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998675132570699,
+      "loss": 4.0344,
+      "step": 1490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029986572439179895,
+      "loss": 4.0503,
+      "step": 1500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002998639235359588,
+      "loss": 4.038,
+      "step": 1510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002998621106896936,
+      "loss": 4.0131,
+      "step": 1520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002998602858531484,
+      "loss": 4.035,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002998584490264692,
+      "loss": 4.029,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029985660020980296,
+      "loss": 4.0702,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002998547394032976,
+      "loss": 3.9678,
+      "step": 1560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000299852866607102,
+      "loss": 3.9994,
+      "step": 1570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000299850981821366,
+      "loss": 4.0349,
+      "step": 1580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029984908504624043,
+      "loss": 3.9855,
+      "step": 1590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000299847176281877,
+      "loss": 3.995,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029984525552842847,
+      "loss": 3.9899,
+      "step": 1610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002998433227860485,
+      "loss": 3.9661,
+      "step": 1620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029984137805489166,
+      "loss": 4.009,
+      "step": 1630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002998394213351137,
+      "loss": 3.9198,
+      "step": 1640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000299837452626871,
+      "loss": 3.9563,
+      "step": 1650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002998354719303212,
+      "loss": 3.9432,
+      "step": 1660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029983347924562273,
+      "loss": 3.9984,
+      "step": 1670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000299831474572935,
+      "loss": 3.991,
+      "step": 1680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002998294579124185,
+      "loss": 3.9759,
+      "step": 1690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002998274292642344,
+      "loss": 3.9539,
+      "step": 1700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002998253886285452,
+      "loss": 3.9262,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000299823336005514,
+      "loss": 3.9282,
+      "step": 1720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002998212713953052,
+      "loss": 3.9466,
+      "step": 1730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029981919479808386,
+      "loss": 3.9303,
+      "step": 1740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002998171062140161,
+      "loss": 3.9155,
+      "step": 1750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029981500564326915,
+      "loss": 3.9133,
+      "step": 1760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000299812893086011,
+      "loss": 3.8839,
+      "step": 1770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002998107685424107,
+      "loss": 3.8817,
+      "step": 1780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029980863201263823,
+      "loss": 3.9205,
+      "step": 1790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029980648349686446,
+      "loss": 3.9162,
+      "step": 1800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029980432299526143,
+      "loss": 3.8682,
+      "step": 1810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002998021505080019,
+      "loss": 3.8901,
+      "step": 1820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029979996603525966,
+      "loss": 3.8445,
+      "step": 1830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029979776957720955,
+      "loss": 3.8295,
+      "step": 1840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029979556113402726,
+      "loss": 3.8621,
+      "step": 1850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002997933407058895,
+      "loss": 3.8556,
+      "step": 1860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000299791108292974,
+      "loss": 3.8938,
+      "step": 1870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002997888638954593,
+      "loss": 3.8572,
+      "step": 1880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029978660751352497,
+      "loss": 3.8279,
+      "step": 1890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002997843391473516,
+      "loss": 3.8352,
+      "step": 1900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002997820587971205,
+      "loss": 3.8915,
+      "step": 1910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002997797664630144,
+      "loss": 3.8182,
+      "step": 1920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029977746214521645,
+      "loss": 3.8138,
+      "step": 1930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002997751458439111,
+      "loss": 3.8405,
+      "step": 1940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002997728175592838,
+      "loss": 3.865,
+      "step": 1950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002997704772915207,
+      "loss": 3.8355,
+      "step": 1960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029976812504080904,
+      "loss": 3.8393,
+      "step": 1970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000299765760807337,
+      "loss": 3.8284,
+      "step": 1980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002997633845912939,
+      "loss": 3.836,
+      "step": 1990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002997609963928697,
+      "loss": 3.7879,
+      "step": 2000
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 3.8531267642974854,
+      "eval_runtime": 1263.3782,
+      "eval_samples_per_second": 60.263,
+      "eval_steps_per_second": 7.533,
+      "step": 2000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029975859621225547,
+      "loss": 3.7696,
+      "step": 2010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029975618404964335,
+      "loss": 3.7891,
+      "step": 2020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029975375990522635,
+      "loss": 3.785,
+      "step": 2030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002997513237791983,
+      "loss": 3.7908,
+      "step": 2040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029974887567175415,
+      "loss": 3.7842,
+      "step": 2050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002997464155830898,
+      "loss": 3.8015,
+      "step": 2060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000299743943513402,
+      "loss": 3.8231,
+      "step": 2070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029974145946288876,
+      "loss": 3.7842,
+      "step": 2080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002997389634317485,
+      "loss": 3.7705,
+      "step": 2090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002997364554201812,
+      "loss": 3.7933,
+      "step": 2100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002997339354283874,
+      "loss": 3.7892,
+      "step": 2110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029973140345656874,
+      "loss": 3.7815,
+      "step": 2120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002997288595049278,
+      "loss": 3.7782,
+      "step": 2130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002997263035736681,
+      "loss": 3.7549,
+      "step": 2140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002997237356629941,
+      "loss": 3.7453,
+      "step": 2150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002997211557731114,
+      "loss": 3.7476,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029971856390422635,
+      "loss": 3.7369,
+      "step": 2170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029971596005654624,
+      "loss": 3.7839,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029971334423027943,
+      "loss": 3.7326,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002997107164256353,
+      "loss": 3.7491,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029970807664282396,
+      "loss": 3.7561,
+      "step": 2210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002997054248820567,
+      "loss": 3.7277,
+      "step": 2220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002997027611435457,
+      "loss": 3.7377,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029970008542750407,
+      "loss": 3.7188,
+      "step": 2240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029969739773414583,
+      "loss": 3.749,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002996946980636861,
+      "loss": 3.7478,
+      "step": 2260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029969198641634085,
+      "loss": 3.7145,
+      "step": 2270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029968926279232697,
+      "loss": 3.718,
+      "step": 2280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002996865271918625,
+      "loss": 3.7107,
+      "step": 2290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002996837796151662,
+      "loss": 3.7283,
+      "step": 2300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002996810200624579,
+      "loss": 3.7105,
+      "step": 2310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002996782485339585,
+      "loss": 3.7127,
+      "step": 2320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029967546502988966,
+      "loss": 3.7119,
+      "step": 2330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002996726695504741,
+      "loss": 3.7244,
+      "step": 2340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002996698620959355,
+      "loss": 3.7023,
+      "step": 2350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029966704266649847,
+      "loss": 3.7158,
+      "step": 2360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029966421126238855,
+      "loss": 3.7225,
+      "step": 2370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029966136788383234,
+      "loss": 3.7294,
+      "step": 2380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029965851253105726,
+      "loss": 3.6898,
+      "step": 2390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002996556452042918,
+      "loss": 3.7663,
+      "step": 2400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002996527659037655,
+      "loss": 3.6646,
+      "step": 2410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029964987462970853,
+      "loss": 3.7028,
+      "step": 2420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002996469713823523,
+      "loss": 3.6833,
+      "step": 2430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029964405616192906,
+      "loss": 3.7037,
+      "step": 2440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002996411289686722,
+      "loss": 3.6912,
+      "step": 2450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002996381898028157,
+      "loss": 3.6827,
+      "step": 2460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029963523866459484,
+      "loss": 3.6791,
+      "step": 2470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002996322755542458,
+      "loss": 3.6742,
+      "step": 2480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002996293004720055,
+      "loss": 3.6954,
+      "step": 2490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029962631341811204,
+      "loss": 3.6884,
+      "step": 2500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002996233143928045,
+      "loss": 3.6578,
+      "step": 2510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029962030339632267,
+      "loss": 3.6415,
+      "step": 2520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029961728042890754,
+      "loss": 3.6443,
+      "step": 2530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029961424549080105,
+      "loss": 3.657,
+      "step": 2540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002996111985822459,
+      "loss": 3.6314,
+      "step": 2550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002996081397034859,
+      "loss": 3.6197,
+      "step": 2560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002996050688547658,
+      "loss": 3.6611,
+      "step": 2570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002996019860363313,
+      "loss": 3.6285,
+      "step": 2580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000299598891248429,
+      "loss": 3.65,
+      "step": 2590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029959578449130654,
+      "loss": 3.6528,
+      "step": 2600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002995926657652126,
+      "loss": 3.6368,
+      "step": 2610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002995895350703965,
+      "loss": 3.6812,
+      "step": 2620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029958639240710894,
+      "loss": 3.6576,
+      "step": 2630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002995832377756012,
+      "loss": 3.6397,
+      "step": 2640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029958007117612573,
+      "loss": 3.625,
+      "step": 2650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002995768926089359,
+      "loss": 3.6114,
+      "step": 2660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000299573702074286,
+      "loss": 3.6461,
+      "step": 2670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029957049957243125,
+      "loss": 3.5942,
+      "step": 2680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029956728510362803,
+      "loss": 3.607,
+      "step": 2690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002995640586681334,
+      "loss": 3.615,
+      "step": 2700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029956082026620553,
+      "loss": 3.6021,
+      "step": 2710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002995575698981035,
+      "loss": 3.6282,
+      "step": 2720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002995543075640874,
+      "loss": 3.598,
+      "step": 2730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002995510332644183,
+      "loss": 3.6252,
+      "step": 2740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029954774699935804,
+      "loss": 3.5939,
+      "step": 2750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029954444876916967,
+      "loss": 3.5992,
+      "step": 2760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029954113857411704,
+      "loss": 3.5652,
+      "step": 2770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002995378164144649,
+      "loss": 3.6183,
+      "step": 2780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002995344822904792,
+      "loss": 3.5739,
+      "step": 2790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029953113620242663,
+      "loss": 3.608,
+      "step": 2800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002995277781505749,
+      "loss": 3.5576,
+      "step": 2810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002995244081351927,
+      "loss": 3.599,
+      "step": 2820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002995210261565497,
+      "loss": 3.5775,
+      "step": 2830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029951763221491634,
+      "loss": 3.6377,
+      "step": 2840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002995142263105644,
+      "loss": 3.5587,
+      "step": 2850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029951080844376615,
+      "loss": 3.5918,
+      "step": 2860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002995073786147952,
+      "loss": 3.5853,
+      "step": 2870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002995039368239259,
+      "loss": 3.5265,
+      "step": 2880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002995004830714336,
+      "loss": 3.5684,
+      "step": 2890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002994970173575947,
+      "loss": 3.5716,
+      "step": 2900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029949353968268647,
+      "loss": 3.558,
+      "step": 2910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029949005004698714,
+      "loss": 3.5642,
+      "step": 2920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002994865484507759,
+      "loss": 3.5517,
+      "step": 2930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029948303489433294,
+      "loss": 3.5486,
+      "step": 2940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002994795093779393,
+      "loss": 3.5693,
+      "step": 2950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002994759719018772,
+      "loss": 3.5195,
+      "step": 2960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002994724224664295,
+      "loss": 3.5802,
+      "step": 2970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002994688610718803,
+      "loss": 3.5658,
+      "step": 2980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002994652877185145,
+      "loss": 3.5215,
+      "step": 2990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029946170240661806,
+      "loss": 3.5484,
+      "step": 3000
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 3.5881383419036865,
+      "eval_runtime": 1262.7068,
+      "eval_samples_per_second": 60.295,
+      "eval_steps_per_second": 7.537,
+      "step": 3000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029945810513647765,
+      "loss": 3.5605,
+      "step": 3010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029945449590838136,
+      "loss": 3.5764,
+      "step": 3020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029945087472261773,
+      "loss": 3.5234,
+      "step": 3030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002994472415794766,
+      "loss": 3.5126,
+      "step": 3040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029944359647924863,
+      "loss": 3.5594,
+      "step": 3050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002994399394222254,
+      "loss": 3.5437,
+      "step": 3060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029943627040869965,
+      "loss": 3.5236,
+      "step": 3070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029943258943896474,
+      "loss": 3.5405,
+      "step": 3080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002994288965133154,
+      "loss": 3.53,
+      "step": 3090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002994251916320469,
+      "loss": 3.5783,
+      "step": 3100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029942147479545574,
+      "loss": 3.5185,
+      "step": 3110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029941774600383933,
+      "loss": 3.5051,
+      "step": 3120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000299414005257496,
+      "loss": 3.5344,
+      "step": 3130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000299410252556725,
+      "loss": 3.5496,
+      "step": 3140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029940648790182663,
+      "loss": 3.5323,
+      "step": 3150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000299402711293102,
+      "loss": 3.5514,
+      "step": 3160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002993989227308534,
+      "loss": 3.5139,
+      "step": 3170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029939512221538385,
+      "loss": 3.5107,
+      "step": 3180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029939130974699745,
+      "loss": 3.5208,
+      "step": 3190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002993874853259993,
+      "loss": 3.4999,
+      "step": 3200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029938364895269525,
+      "loss": 3.5322,
+      "step": 3210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002993798006273924,
+      "loss": 3.4833,
+      "step": 3220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002993759403503985,
+      "loss": 3.5276,
+      "step": 3230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002993720681220225,
+      "loss": 3.5106,
+      "step": 3240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002993681839425742,
+      "loss": 3.5059,
+      "step": 3250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029936428781236437,
+      "loss": 3.4978,
+      "step": 3260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002993603797317047,
+      "loss": 3.5091,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002993564597009079,
+      "loss": 3.4773,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029935252772028757,
+      "loss": 3.4778,
+      "step": 3290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002993485837901584,
+      "loss": 3.5006,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029934462791083585,
+      "loss": 3.5202,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002993406600826364,
+      "loss": 3.513,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002993366803058776,
+      "loss": 3.4845,
+      "step": 3330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029933268858087784,
+      "loss": 3.461,
+      "step": 3340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029932868490795643,
+      "loss": 3.4556,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029932466928743375,
+      "loss": 3.4998,
+      "step": 3360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002993206417196311,
+      "loss": 3.4474,
+      "step": 3370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002993166022048707,
+      "loss": 3.4596,
+      "step": 3380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029931255074347576,
+      "loss": 3.4692,
+      "step": 3390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002993084873357704,
+      "loss": 3.4388,
+      "step": 3400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002993044119820797,
+      "loss": 3.4542,
+      "step": 3410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002993003246827299,
+      "loss": 3.4794,
+      "step": 3420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002992962254380478,
+      "loss": 3.4603,
+      "step": 3430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029929211424836147,
+      "loss": 3.5052,
+      "step": 3440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002992879911139998,
+      "loss": 3.4925,
+      "step": 3450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029928385603529285,
+      "loss": 3.5074,
+      "step": 3460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002992797090125712,
+      "loss": 3.4603,
+      "step": 3470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002992755500461668,
+      "loss": 3.481,
+      "step": 3480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029927137913641247,
+      "loss": 3.4604,
+      "step": 3490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002992671962836417,
+      "loss": 3.4553,
+      "step": 3500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029926300148818937,
+      "loss": 3.4687,
+      "step": 3510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000299258794750391,
+      "loss": 3.4339,
+      "step": 3520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002992545760705832,
+      "loss": 3.449,
+      "step": 3530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029925034544910344,
+      "loss": 3.4514,
+      "step": 3540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029924610288629023,
+      "loss": 3.4281,
+      "step": 3550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002992418483824831,
+      "loss": 3.44,
+      "step": 3560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029923758193802234,
+      "loss": 3.436,
+      "step": 3570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029923330355324936,
+      "loss": 3.4545,
+      "step": 3580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029922901322850643,
+      "loss": 3.4504,
+      "step": 3590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002992247109641368,
+      "loss": 3.4446,
+      "step": 3600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002992203967604847,
+      "loss": 3.4326,
+      "step": 3610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002992160706178954,
+      "loss": 3.4506,
+      "step": 3620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029921173253671495,
+      "loss": 3.4613,
+      "step": 3630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002992073825172904,
+      "loss": 3.4504,
+      "step": 3640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029920302055996985,
+      "loss": 3.4392,
+      "step": 3650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002991986466651023,
+      "loss": 3.4374,
+      "step": 3660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029919426083303765,
+      "loss": 3.4257,
+      "step": 3670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029918986306412684,
+      "loss": 3.4183,
+      "step": 3680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029918545335872165,
+      "loss": 3.4406,
+      "step": 3690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029918103171717504,
+      "loss": 3.4079,
+      "step": 3700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002991765981398407,
+      "loss": 3.4628,
+      "step": 3710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002991721526270733,
+      "loss": 3.4464,
+      "step": 3720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002991676951792286,
+      "loss": 3.4343,
+      "step": 3730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029916322579666324,
+      "loss": 3.4202,
+      "step": 3740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002991587444797348,
+      "loss": 3.4158,
+      "step": 3750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029915425122880175,
+      "loss": 3.4387,
+      "step": 3760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029914974604422365,
+      "loss": 3.4614,
+      "step": 3770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029914522892636095,
+      "loss": 3.4164,
+      "step": 3780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002991406998755751,
+      "loss": 3.4754,
+      "step": 3790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002991361588922284,
+      "loss": 3.412,
+      "step": 3800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029913160597668426,
+      "loss": 3.4467,
+      "step": 3810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029912704112930677,
+      "loss": 3.3988,
+      "step": 3820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002991224643504613,
+      "loss": 3.4257,
+      "step": 3830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002991178756405141,
+      "loss": 3.401,
+      "step": 3840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002991132749998321,
+      "loss": 3.4203,
+      "step": 3850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029910866242878356,
+      "loss": 3.4192,
+      "step": 3860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029910403792773746,
+      "loss": 3.4426,
+      "step": 3870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002990994014970638,
+      "loss": 3.418,
+      "step": 3880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002990947531371335,
+      "loss": 3.389,
+      "step": 3890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029909009284831856,
+      "loss": 3.4054,
+      "step": 3900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029908542063099183,
+      "loss": 3.4464,
+      "step": 3910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029908073648552704,
+      "loss": 3.422,
+      "step": 3920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002990760404122991,
+      "loss": 3.4006,
+      "step": 3930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029907133241168356,
+      "loss": 3.3868,
+      "step": 3940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029906661248405725,
+      "loss": 3.3897,
+      "step": 3950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029906188062979775,
+      "loss": 3.3704,
+      "step": 3960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002990571368492836,
+      "loss": 3.4079,
+      "step": 3970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002990523811428945,
+      "loss": 3.3834,
+      "step": 3980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029904761351101075,
+      "loss": 3.4196,
+      "step": 3990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029904283395401393,
+      "loss": 3.3734,
+      "step": 4000
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 3.4287383556365967,
+      "eval_runtime": 1258.0117,
+      "eval_samples_per_second": 60.52,
+      "eval_steps_per_second": 7.565,
+      "step": 4000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029903804247228644,
+      "loss": 3.3908,
+      "step": 4010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002990332390662116,
+      "loss": 3.4163,
+      "step": 4020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002990284237361737,
+      "loss": 3.4081,
+      "step": 4030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029902359648255805,
+      "loss": 3.3577,
+      "step": 4040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002990187573057509,
+      "loss": 3.3885,
+      "step": 4050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002990139062061394,
+      "loss": 3.4066,
+      "step": 4060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029900904318411165,
+      "loss": 3.3853,
+      "step": 4070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029900416824005677,
+      "loss": 3.3886,
+      "step": 4080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002989992813743648,
+      "loss": 3.3803,
+      "step": 4090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002989943825874267,
+      "loss": 3.4059,
+      "step": 4100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002989894718796345,
+      "loss": 3.3762,
+      "step": 4110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029898454925138094,
+      "loss": 3.4167,
+      "step": 4120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029897961470306004,
+      "loss": 3.3495,
+      "step": 4130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029897466823506654,
+      "loss": 3.3737,
+      "step": 4140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029896970984779616,
+      "loss": 3.3841,
+      "step": 4150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002989647395416456,
+      "loss": 3.3975,
+      "step": 4160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029895975731701266,
+      "loss": 3.3866,
+      "step": 4170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002989547631742959,
+      "loss": 3.3105,
+      "step": 4180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002989497571138948,
+      "loss": 3.3763,
+      "step": 4190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029894473913621,
+      "loss": 3.3882,
+      "step": 4200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000298939709241643,
+      "loss": 3.3671,
+      "step": 4210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002989346674305961,
+      "loss": 3.4105,
+      "step": 4220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002989296137034728,
+      "loss": 3.3696,
+      "step": 4230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002989245480606774,
+      "loss": 3.3327,
+      "step": 4240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029891947050261525,
+      "loss": 3.3353,
+      "step": 4250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002989143810296925,
+      "loss": 3.3668,
+      "step": 4260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029890927964231644,
+      "loss": 3.3709,
+      "step": 4270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002989041663408952,
+      "loss": 3.3784,
+      "step": 4280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002988990411258379,
+      "loss": 3.388,
+      "step": 4290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002988939039975546,
+      "loss": 3.3456,
+      "step": 4300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002988887549564563,
+      "loss": 3.3101,
+      "step": 4310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029888359400295496,
+      "loss": 3.374,
+      "step": 4320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002988784211374635,
+      "loss": 3.3092,
+      "step": 4330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002988732363603958,
+      "loss": 3.3741,
+      "step": 4340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002988680396721667,
+      "loss": 3.3237,
+      "step": 4350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000298862831073192,
+      "loss": 3.3449,
+      "step": 4360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002988576105638883,
+      "loss": 3.3151,
+      "step": 4370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002988523781446735,
+      "loss": 3.3896,
+      "step": 4380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002988471338159661,
+      "loss": 3.3084,
+      "step": 4390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002988418775781858,
+      "loss": 3.3505,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000298836609431753,
+      "loss": 3.3135,
+      "step": 4410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002988313293770893,
+      "loss": 3.3518,
+      "step": 4420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029882603741461706,
+      "loss": 3.3693,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002988207335447598,
+      "loss": 3.3501,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002988154177679418,
+      "loss": 3.3211,
+      "step": 4450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029881009008458844,
+      "loss": 3.3052,
+      "step": 4460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029880475049512585,
+      "loss": 3.3157,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029879939899998137,
+      "loss": 3.3058,
+      "step": 4480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002987940355995831,
+      "loss": 3.32,
+      "step": 4490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002987886602943601,
+      "loss": 3.3461,
+      "step": 4500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002987832730847427,
+      "loss": 3.3216,
+      "step": 4510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029877787397116157,
+      "loss": 3.3265,
+      "step": 4520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000298772462954049,
+      "loss": 3.2859,
+      "step": 4530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002987670400338377,
+      "loss": 3.3683,
+      "step": 4540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002987616052109617,
+      "loss": 3.3287,
+      "step": 4550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029875615848585563,
+      "loss": 3.3218,
+      "step": 4560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002987506998589555,
+      "loss": 3.3386,
+      "step": 4570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029874522933069793,
+      "loss": 3.341,
+      "step": 4580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002987397469015207,
+      "loss": 3.2819,
+      "step": 4590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002987342525718623,
+      "loss": 3.302,
+      "step": 4600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029872874634216245,
+      "loss": 3.3155,
+      "step": 4610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002987232282128617,
+      "loss": 3.3039,
+      "step": 4620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029871769818440145,
+      "loss": 3.2963,
+      "step": 4630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002987121562572242,
+      "loss": 3.2748,
+      "step": 4640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029870660243177345,
+      "loss": 3.3123,
+      "step": 4650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029870103670849335,
+      "loss": 3.3812,
+      "step": 4660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002986954590878294,
+      "loss": 3.3077,
+      "step": 4670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029868986957022777,
+      "loss": 3.3504,
+      "step": 4680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029868426815613566,
+      "loss": 3.3157,
+      "step": 4690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029867865484600124,
+      "loss": 3.2964,
+      "step": 4700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002986730296402737,
+      "loss": 3.2778,
+      "step": 4710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000298667392539403,
+      "loss": 3.2901,
+      "step": 4720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029866174354384025,
+      "loss": 3.2901,
+      "step": 4730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002986560826540373,
+      "loss": 3.2924,
+      "step": 4740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002986504098704472,
+      "loss": 3.3023,
+      "step": 4750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002986447251935237,
+      "loss": 3.3008,
+      "step": 4760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002986390286237218,
+      "loss": 3.2718,
+      "step": 4770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029863332016149706,
+      "loss": 3.3165,
+      "step": 4780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029862759980730635,
+      "loss": 3.3116,
+      "step": 4790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002986218675616073,
+      "loss": 3.283,
+      "step": 4800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002986161234248585,
+      "loss": 3.3114,
+      "step": 4810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002986103673975196,
+      "loss": 3.3064,
+      "step": 4820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029860459948005116,
+      "loss": 3.2989,
+      "step": 4830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002985988196729146,
+      "loss": 3.2692,
+      "step": 4840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029859302797657233,
+      "loss": 3.2622,
+      "step": 4850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002985872243914878,
+      "loss": 3.2947,
+      "step": 4860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029858140891812535,
+      "loss": 3.329,
+      "step": 4870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029857558155695025,
+      "loss": 3.278,
+      "step": 4880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029856974230842864,
+      "loss": 3.2424,
+      "step": 4890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002985638911730279,
+      "loss": 3.3043,
+      "step": 4900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029855802815121604,
+      "loss": 3.3066,
+      "step": 4910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002985521532434622,
+      "loss": 3.293,
+      "step": 4920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002985462664502364,
+      "loss": 3.2835,
+      "step": 4930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029854036777200965,
+      "loss": 3.2571,
+      "step": 4940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002985344572092539,
+      "loss": 3.3262,
+      "step": 4950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029852853476244206,
+      "loss": 3.2634,
+      "step": 4960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029852260043204795,
+      "loss": 3.2552,
+      "step": 4970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029851665421854633,
+      "loss": 3.2921,
+      "step": 4980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029851069612241306,
+      "loss": 3.2653,
+      "step": 4990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002985047261441247,
+      "loss": 3.2722,
+      "step": 5000
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 3.314415693283081,
+      "eval_runtime": 1262.2553,
+      "eval_samples_per_second": 60.317,
+      "eval_steps_per_second": 7.54,
+      "step": 5000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000298498744284159,
+      "loss": 3.299,
+      "step": 5010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002984927505429946,
+      "loss": 3.2565,
+      "step": 5020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000298486744921111,
+      "loss": 3.264,
+      "step": 5030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002984807274189886,
+      "loss": 3.3006,
+      "step": 5040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002984746980371089,
+      "loss": 3.2533,
+      "step": 5050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002984686567759545,
+      "loss": 3.2751,
+      "step": 5060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002984626036360085,
+      "loss": 3.262,
+      "step": 5070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002984565386177554,
+      "loss": 3.2465,
+      "step": 5080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002984504617216803,
+      "loss": 3.2722,
+      "step": 5090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002984443729482694,
+      "loss": 3.2976,
+      "step": 5100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029843827229801003,
+      "loss": 3.2764,
+      "step": 5110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002984321597713901,
+      "loss": 3.2729,
+      "step": 5120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002984260353688988,
+      "loss": 3.2722,
+      "step": 5130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029841989909102604,
+      "loss": 3.284,
+      "step": 5140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002984137509382629,
+      "loss": 3.2443,
+      "step": 5150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029840759091110113,
+      "loss": 3.2197,
+      "step": 5160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002984014190100337,
+      "loss": 3.2432,
+      "step": 5170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029839523523555436,
+      "loss": 3.2188,
+      "step": 5180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029838903958815787,
+      "loss": 3.2312,
+      "step": 5190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029838283206834004,
+      "loss": 3.2435,
+      "step": 5200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002983766126765974,
+      "loss": 3.2374,
+      "step": 5210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029837038141342754,
+      "loss": 3.2422,
+      "step": 5220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002983641382793291,
+      "loss": 3.2507,
+      "step": 5230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029835788327480157,
+      "loss": 3.282,
+      "step": 5240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002983516164003454,
+      "loss": 3.2515,
+      "step": 5250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029834533765646197,
+      "loss": 3.2589,
+      "step": 5260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029833904704365364,
+      "loss": 3.2304,
+      "step": 5270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002983327445624237,
+      "loss": 3.2411,
+      "step": 5280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002983264302132765,
+      "loss": 3.2118,
+      "step": 5290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029832010399671714,
+      "loss": 3.2418,
+      "step": 5300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029831376591325174,
+      "loss": 3.2358,
+      "step": 5310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029830741596338753,
+      "loss": 3.2352,
+      "step": 5320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029830105414763245,
+      "loss": 3.2664,
+      "step": 5330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002982946804664956,
+      "loss": 3.2443,
+      "step": 5340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029828829492048686,
+      "loss": 3.2589,
+      "step": 5350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002982818975101172,
+      "loss": 3.2397,
+      "step": 5360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002982754882358983,
+      "loss": 3.2304,
+      "step": 5370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002982690670983432,
+      "loss": 3.2259,
+      "step": 5380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029826263409796547,
+      "loss": 3.2366,
+      "step": 5390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002982561892352798,
+      "loss": 3.2004,
+      "step": 5400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029824973251080196,
+      "loss": 3.2218,
+      "step": 5410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029824326392504844,
+      "loss": 3.2355,
+      "step": 5420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029823678347853683,
+      "loss": 3.2163,
+      "step": 5430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002982302911717856,
+      "loss": 3.2359,
+      "step": 5440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029822378700531424,
+      "loss": 3.2817,
+      "step": 5450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029821727097964306,
+      "loss": 3.2251,
+      "step": 5460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002982107430952935,
+      "loss": 3.2196,
+      "step": 5470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002982042033527878,
+      "loss": 3.2125,
+      "step": 5480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029819765175264913,
+      "loss": 3.2196,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029819108829540176,
+      "loss": 3.249,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002981845129815708,
+      "loss": 3.2223,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029817792581168236,
+      "loss": 3.2181,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002981713267862634,
+      "loss": 3.2358,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002981647159058419,
+      "loss": 3.2242,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002981580931709469,
+      "loss": 3.1984,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002981514585821082,
+      "loss": 3.2033,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002981448121398566,
+      "loss": 3.2383,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029813815384472394,
+      "loss": 3.1933,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029813148369724283,
+      "loss": 3.2225,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002981248016979471,
+      "loss": 3.2276,
+      "step": 5600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029811810784737127,
+      "loss": 3.2419,
+      "step": 5610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029811140214605094,
+      "loss": 3.2197,
+      "step": 5620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029810468459452255,
+      "loss": 3.1913,
+      "step": 5630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002980979551933237,
+      "loss": 3.2199,
+      "step": 5640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029809121394299266,
+      "loss": 3.248,
+      "step": 5650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029808446084406885,
+      "loss": 3.2327,
+      "step": 5660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002980776958970926,
+      "loss": 3.2153,
+      "step": 5670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002980709191026051,
+      "loss": 3.1731,
+      "step": 5680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002980641304611486,
+      "loss": 3.2494,
+      "step": 5690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002980573299732663,
+      "loss": 3.2158,
+      "step": 5700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029805051763950223,
+      "loss": 3.2046,
+      "step": 5710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029804369346040147,
+      "loss": 3.2824,
+      "step": 5720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029803685743650987,
+      "loss": 3.1937,
+      "step": 5730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002980300095683746,
+      "loss": 3.1755,
+      "step": 5740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002980231498565435,
+      "loss": 3.2206,
+      "step": 5750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029801627830156524,
+      "loss": 3.2374,
+      "step": 5760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002980093949039898,
+      "loss": 3.2115,
+      "step": 5770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002980024996643678,
+      "loss": 3.2323,
+      "step": 5780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002979955925832509,
+      "loss": 3.2223,
+      "step": 5790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029798867366119183,
+      "loss": 3.1721,
+      "step": 5800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029798174289874405,
+      "loss": 3.2005,
+      "step": 5810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002979748002964622,
+      "loss": 3.2027,
+      "step": 5820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029796784585490164,
+      "loss": 3.1967,
+      "step": 5830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029796087957461887,
+      "loss": 3.1974,
+      "step": 5840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002979539014561712,
+      "loss": 3.1973,
+      "step": 5850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000297946911500117,
+      "loss": 3.2019,
+      "step": 5860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002979399097070155,
+      "loss": 3.2292,
+      "step": 5870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002979328960774268,
+      "loss": 3.1824,
+      "step": 5880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029792587061191225,
+      "loss": 3.1908,
+      "step": 5890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029791883331103377,
+      "loss": 3.1768,
+      "step": 5900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002979117841753545,
+      "loss": 3.1947,
+      "step": 5910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029790472320543845,
+      "loss": 3.176,
+      "step": 5920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029789765040185047,
+      "loss": 3.2213,
+      "step": 5930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002978905657651565,
+      "loss": 3.2036,
+      "step": 5940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978834692959234,
+      "loss": 3.1913,
+      "step": 5950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978763609947189,
+      "loss": 3.1837,
+      "step": 5960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978692408621117,
+      "loss": 3.21,
+      "step": 5970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978621088986716,
+      "loss": 3.1888,
+      "step": 5980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978549651049691,
+      "loss": 3.1807,
+      "step": 5990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978478094815758,
+      "loss": 3.2276,
+      "step": 6000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 3.229881525039673,
+      "eval_runtime": 1262.638,
+      "eval_samples_per_second": 60.298,
+      "eval_steps_per_second": 7.537,
+      "step": 6000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978406420290642,
+      "loss": 3.1827,
+      "step": 6010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029783346274800783,
+      "loss": 3.1995,
+      "step": 6020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000297826271638981,
+      "loss": 3.1944,
+      "step": 6030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978190687025591,
+      "loss": 3.1717,
+      "step": 6040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002978118539393185,
+      "loss": 3.1701,
+      "step": 6050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029780462734983623,
+      "loss": 3.184,
+      "step": 6060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029779738893469075,
+      "loss": 3.1846,
+      "step": 6070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029779013869446106,
+      "loss": 3.1516,
+      "step": 6080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029778287662972716,
+      "loss": 3.1818,
+      "step": 6090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002977756027410703,
+      "loss": 3.1829,
+      "step": 6100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002977683170290722,
+      "loss": 3.2006,
+      "step": 6110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000297761019494316,
+      "loss": 3.2065,
+      "step": 6120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029775371013738547,
+      "loss": 3.166,
+      "step": 6130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002977463889588654,
+      "loss": 3.1668,
+      "step": 6140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002977390559593416,
+      "loss": 3.186,
+      "step": 6150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002977317111394007,
+      "loss": 3.2082,
+      "step": 6160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002977243544996304,
+      "loss": 3.1899,
+      "step": 6170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029771698604061933,
+      "loss": 3.1811,
+      "step": 6180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000297709605762957,
+      "loss": 3.1397,
+      "step": 6190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029770221366723387,
+      "loss": 3.1599,
+      "step": 6200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002976948097540414,
+      "loss": 3.158,
+      "step": 6210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000297687394023972,
+      "loss": 3.1851,
+      "step": 6220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029767996647761893,
+      "loss": 3.163,
+      "step": 6230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002976725271155765,
+      "loss": 3.1806,
+      "step": 6240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029766507593843985,
+      "loss": 3.2,
+      "step": 6250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029765761294680524,
+      "loss": 3.1744,
+      "step": 6260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029765013814126974,
+      "loss": 3.1667,
+      "step": 6270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029764265152243136,
+      "loss": 3.2051,
+      "step": 6280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002976351530908892,
+      "loss": 3.183,
+      "step": 6290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029762764284724306,
+      "loss": 3.154,
+      "step": 6300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029762012079209395,
+      "loss": 3.1427,
+      "step": 6310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029761258692604357,
+      "loss": 3.1906,
+      "step": 6320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002976050412496948,
+      "loss": 3.1848,
+      "step": 6330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029759748376365135,
+      "loss": 3.1661,
+      "step": 6340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029758991446851785,
+      "loss": 3.1544,
+      "step": 6350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029758233336489993,
+      "loss": 3.1566,
+      "step": 6360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029757474045340415,
+      "loss": 3.1497,
+      "step": 6370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029756713573463793,
+      "loss": 3.1226,
+      "step": 6380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002975595192092099,
+      "loss": 3.1781,
+      "step": 6390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029755189087772924,
+      "loss": 3.1726,
+      "step": 6400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002975442507408064,
+      "loss": 3.1625,
+      "step": 6410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029753659879905265,
+      "loss": 3.1671,
+      "step": 6420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029752893505308013,
+      "loss": 3.1294,
+      "step": 6430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002975212595035021,
+      "loss": 3.1798,
+      "step": 6440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029751357215093265,
+      "loss": 3.1901,
+      "step": 6450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029750587299598685,
+      "loss": 3.1859,
+      "step": 6460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029749816203928064,
+      "loss": 3.1316,
+      "step": 6470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000297490439281431,
+      "loss": 3.1507,
+      "step": 6480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029748270472305585,
+      "loss": 3.172,
+      "step": 6490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000297474958364774,
+      "loss": 3.1487,
+      "step": 6500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002974672002072052,
+      "loss": 3.109,
+      "step": 6510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002974594302509702,
+      "loss": 3.1347,
+      "step": 6520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002974516484966906,
+      "loss": 3.1345,
+      "step": 6530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002974438549449891,
+      "loss": 3.1585,
+      "step": 6540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029743604959648923,
+      "loss": 3.1138,
+      "step": 6550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002974282324518154,
+      "loss": 3.1808,
+      "step": 6560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002974204035115932,
+      "loss": 3.1491,
+      "step": 6570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002974125627764489,
+      "loss": 3.1422,
+      "step": 6580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002974047102470099,
+      "loss": 3.1711,
+      "step": 6590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029739684592390434,
+      "loss": 3.184,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002973889698077616,
+      "loss": 3.142,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029738108189921174,
+      "loss": 3.1151,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029737318219888594,
+      "loss": 3.1337,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029736527070741613,
+      "loss": 3.1797,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002973573474254353,
+      "loss": 3.1147,
+      "step": 6650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002973494123535776,
+      "loss": 3.1531,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029734146549247764,
+      "loss": 3.1583,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029733350684277136,
+      "loss": 3.1491,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002973255364050955,
+      "loss": 3.1364,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002973175541800878,
+      "loss": 3.1253,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002973095601683869,
+      "loss": 3.1379,
+      "step": 6710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029730155437063233,
+      "loss": 3.1563,
+      "step": 6720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029729353678746464,
+      "loss": 3.1433,
+      "step": 6730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002972855074195254,
+      "loss": 3.1424,
+      "step": 6740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002972774662674569,
+      "loss": 3.1461,
+      "step": 6750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002972694133319026,
+      "loss": 3.1398,
+      "step": 6760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029726134861350677,
+      "loss": 3.1105,
+      "step": 6770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029725327211291465,
+      "loss": 3.0861,
+      "step": 6780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002972451838307724,
+      "loss": 3.1124,
+      "step": 6790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029723708376772724,
+      "loss": 3.1455,
+      "step": 6800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029722897192442714,
+      "loss": 3.18,
+      "step": 6810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029722084830152126,
+      "loss": 3.1196,
+      "step": 6820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002972127128996594,
+      "loss": 3.1463,
+      "step": 6830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029720456571949254,
+      "loss": 3.1278,
+      "step": 6840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002971964067616726,
+      "loss": 3.1672,
+      "step": 6850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002971882360268522,
+      "loss": 3.145,
+      "step": 6860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029718005351568523,
+      "loss": 3.1416,
+      "step": 6870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029717185922882626,
+      "loss": 3.1329,
+      "step": 6880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000297163653166931,
+      "loss": 3.1422,
+      "step": 6890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029715543533065587,
+      "loss": 3.1844,
+      "step": 6900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002971472057206585,
+      "loss": 3.1245,
+      "step": 6910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029713896433759726,
+      "loss": 3.1464,
+      "step": 6920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002971307111821315,
+      "loss": 3.1336,
+      "step": 6930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029712244625492167,
+      "loss": 3.1161,
+      "step": 6940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002971141695566289,
+      "loss": 3.1404,
+      "step": 6950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002971058810879155,
+      "loss": 3.1068,
+      "step": 6960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029709758084944457,
+      "loss": 3.1482,
+      "step": 6970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029708926884188024,
+      "loss": 3.1316,
+      "step": 6980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029708094506588744,
+      "loss": 3.1598,
+      "step": 6990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029707260952213226,
+      "loss": 3.1809,
+      "step": 7000
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 3.1597089767456055,
+      "eval_runtime": 1262.6258,
+      "eval_samples_per_second": 60.299,
+      "eval_steps_per_second": 7.537,
+      "step": 7000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002970642622112816,
+      "loss": 3.1457,
+      "step": 7010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029705590313400325,
+      "loss": 3.1302,
+      "step": 7020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029704753229096607,
+      "loss": 3.1369,
+      "step": 7030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002970391496828398,
+      "loss": 3.0741,
+      "step": 7040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002970307553102951,
+      "loss": 3.1429,
+      "step": 7050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029702234917400354,
+      "loss": 3.0726,
+      "step": 7060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002970139312746378,
+      "loss": 3.1103,
+      "step": 7070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002970055016128714,
+      "loss": 3.1151,
+      "step": 7080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969970601893786,
+      "loss": 3.146,
+      "step": 7090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029698860700483494,
+      "loss": 3.12,
+      "step": 7100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969801420599167,
+      "loss": 3.1103,
+      "step": 7110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029697166535530117,
+      "loss": 3.0877,
+      "step": 7120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969631768916666,
+      "loss": 3.1072,
+      "step": 7130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000296954676669692,
+      "loss": 3.1124,
+      "step": 7140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969461646900576,
+      "loss": 3.1139,
+      "step": 7150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969376409534444,
+      "loss": 3.1035,
+      "step": 7160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002969291054605343,
+      "loss": 3.1059,
+      "step": 7170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002969205582120104,
+      "loss": 3.1141,
+      "step": 7180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029691199920855635,
+      "loss": 3.1194,
+      "step": 7190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000296903428450857,
+      "loss": 3.127,
+      "step": 7200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002968948459395982,
+      "loss": 3.1059,
+      "step": 7210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029688625167546644,
+      "loss": 3.0892,
+      "step": 7220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002968776456591495,
+      "loss": 3.103,
+      "step": 7230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029686902789133584,
+      "loss": 3.0996,
+      "step": 7240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029686039837271494,
+      "loss": 3.0967,
+      "step": 7250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002968517571039774,
+      "loss": 3.0888,
+      "step": 7260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002968431040858144,
+      "loss": 3.0901,
+      "step": 7270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002968344393189183,
+      "loss": 3.1288,
+      "step": 7280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002968257628039825,
+      "loss": 3.0919,
+      "step": 7290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029681707454170105,
+      "loss": 3.1561,
+      "step": 7300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029680837453276913,
+      "loss": 3.117,
+      "step": 7310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029679966277788275,
+      "loss": 3.0694,
+      "step": 7320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002967909392777391,
+      "loss": 3.1108,
+      "step": 7330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029678220403303596,
+      "loss": 3.1357,
+      "step": 7340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002967734570444723,
+      "loss": 3.097,
+      "step": 7350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000296764698312748,
+      "loss": 3.1069,
+      "step": 7360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029675592783856374,
+      "loss": 3.136,
+      "step": 7370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029674714562262126,
+      "loss": 3.1069,
+      "step": 7380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00029673835166562333,
+      "loss": 3.1196,
+      "step": 7390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002967295459682734,
+      "loss": 3.1102,
+      "step": 7400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000296720728531276,
+      "loss": 3.1162,
+      "step": 7410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002967118993553367,
+      "loss": 3.1198,
+      "step": 7420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00029670305844116184,
+      "loss": 3.0783,
+      "step": 7430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00029669420578945884,
+      "loss": 3.0867,
+      "step": 7440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002966853414009359,
+      "loss": 3.0681,
+      "step": 7450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002966764652763023,
+      "loss": 3.1045,
+      "step": 7460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002966675774162682,
+      "loss": 3.1053,
+      "step": 7470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002966586778215447,
+      "loss": 3.0944,
+      "step": 7480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002966497664928439,
+      "loss": 3.0843,
+      "step": 7490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00029664084343087867,
+      "loss": 3.1086,
+      "step": 7500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002966319086363631,
+      "loss": 3.0846,
+      "step": 7510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00029662296211001185,
+      "loss": 3.0417,
+      "step": 7520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00029661400385254085,
+      "loss": 3.0938,
+      "step": 7530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00029660503386466685,
+      "loss": 3.1287,
+      "step": 7540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002965960521471075,
+      "loss": 3.0479,
+      "step": 7550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002965870587005813,
+      "loss": 3.1001,
+      "step": 7560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000296578053525808,
+      "loss": 3.0934,
+      "step": 7570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000296569036623508,
+      "loss": 3.1251,
+      "step": 7580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002965600079944027,
+      "loss": 3.0948,
+      "step": 7590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002965509676392145,
+      "loss": 3.0783,
+      "step": 7600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002965419155586667,
+      "loss": 3.0542,
+      "step": 7610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029653285175348357,
+      "loss": 3.0946,
+      "step": 7620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002965237762243903,
+      "loss": 3.074,
+      "step": 7630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000296514688972113,
+      "loss": 3.0811,
+      "step": 7640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029650558999737867,
+      "loss": 3.08,
+      "step": 7650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002964964793009154,
+      "loss": 3.073,
+      "step": 7660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029648735688345205,
+      "loss": 3.0804,
+      "step": 7670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029647822274571855,
+      "loss": 3.0701,
+      "step": 7680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002964690768884457,
+      "loss": 3.0487,
+      "step": 7690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029645991931236517,
+      "loss": 3.0867,
+      "step": 7700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002964507500182098,
+      "loss": 3.0916,
+      "step": 7710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002964415690067131,
+      "loss": 3.0782,
+      "step": 7720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002964323762786096,
+      "loss": 3.089,
+      "step": 7730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00029642317183463493,
+      "loss": 3.0416,
+      "step": 7740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00029641395567552546,
+      "loss": 3.0844,
+      "step": 7750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002964047278020185,
+      "loss": 3.1009,
+      "step": 7760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002963954882148524,
+      "loss": 3.0975,
+      "step": 7770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002963862369147665,
+      "loss": 3.0932,
+      "step": 7780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00029637697390250085,
+      "loss": 3.0968,
+      "step": 7790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002963676991787967,
+      "loss": 3.0679,
+      "step": 7800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00029635841274439594,
+      "loss": 3.0705,
+      "step": 7810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002963491146000417,
+      "loss": 3.0923,
+      "step": 7820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002963398047464779,
+      "loss": 3.0742,
+      "step": 7830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002963304831844494,
+      "loss": 3.0896,
+      "step": 7840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029632114991470196,
+      "loss": 3.0747,
+      "step": 7850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002963118049379824,
+      "loss": 3.1024,
+      "step": 7860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029630244825503836,
+      "loss": 3.0963,
+      "step": 7870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002962930798666184,
+      "loss": 3.0835,
+      "step": 7880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002962836997734722,
+      "loss": 3.0745,
+      "step": 7890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002962743079763501,
+      "loss": 3.0786,
+      "step": 7900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029626490447600366,
+      "loss": 3.0697,
+      "step": 7910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029625548927318516,
+      "loss": 3.084,
+      "step": 7920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00029624606236864796,
+      "loss": 3.059,
+      "step": 7930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002962366237631462,
+      "loss": 3.0793,
+      "step": 7940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029622717345743514,
+      "loss": 3.0691,
+      "step": 7950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002962177114522709,
+      "loss": 3.1112,
+      "step": 7960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029620823774841043,
+      "loss": 3.1004,
+      "step": 7970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002961987523466117,
+      "loss": 3.0406,
+      "step": 7980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029618925524763377,
+      "loss": 3.0851,
+      "step": 7990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029617974645223643,
+      "loss": 3.0706,
+      "step": 8000
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 3.1043083667755127,
+      "eval_runtime": 1262.8212,
+      "eval_samples_per_second": 60.29,
+      "eval_steps_per_second": 7.536,
+      "step": 8000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002961702259611804,
+      "loss": 3.0704,
+      "step": 8010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002961606937752274,
+      "loss": 3.051,
+      "step": 8020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029615114989514017,
+      "loss": 3.0859,
+      "step": 8030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002961415943216823,
+      "loss": 3.0962,
+      "step": 8040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002961320270556182,
+      "loss": 3.0713,
+      "step": 8050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029612244809771357,
+      "loss": 3.0823,
+      "step": 8060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002961128574487346,
+      "loss": 3.061,
+      "step": 8070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002961032551094486,
+      "loss": 3.074,
+      "step": 8080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000296093641080624,
+      "loss": 3.0551,
+      "step": 8090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029608401536303,
+      "loss": 3.0811,
+      "step": 8100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002960743779574365,
+      "loss": 3.0746,
+      "step": 8110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002960647288646149,
+      "loss": 3.0514,
+      "step": 8120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000296055068085337,
+      "loss": 3.1027,
+      "step": 8130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029604539562037586,
+      "loss": 3.0493,
+      "step": 8140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002960357114705053,
+      "loss": 3.0647,
+      "step": 8150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029602601563650014,
+      "loss": 3.0594,
+      "step": 8160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029601630811913613,
+      "loss": 3.058,
+      "step": 8170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029600658891919,
+      "loss": 3.0755,
+      "step": 8180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002959968580374393,
+      "loss": 3.0111,
+      "step": 8190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002959871154746627,
+      "loss": 3.0019,
+      "step": 8200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029597736123163955,
+      "loss": 3.0586,
+      "step": 8210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002959675953091504,
+      "loss": 3.0724,
+      "step": 8220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002959578177079765,
+      "loss": 3.0824,
+      "step": 8230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029594802842890016,
+      "loss": 3.0672,
+      "step": 8240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029593822747270467,
+      "loss": 3.0526,
+      "step": 8250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029592841484017425,
+      "loss": 3.0724,
+      "step": 8260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002959185905320938,
+      "loss": 3.0718,
+      "step": 8270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029590875454924947,
+      "loss": 3.0034,
+      "step": 8280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002958989068924283,
+      "loss": 3.0563,
+      "step": 8290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000295889047562418,
+      "loss": 3.0657,
+      "step": 8300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002958791765600076,
+      "loss": 3.0595,
+      "step": 8310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002958692938859867,
+      "loss": 3.0517,
+      "step": 8320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029585939954114614,
+      "loss": 3.0472,
+      "step": 8330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029584949352627746,
+      "loss": 3.0924,
+      "step": 8340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029583957584217326,
+      "loss": 3.0252,
+      "step": 8350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029582964648962697,
+      "loss": 3.0424,
+      "step": 8360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029581970546943315,
+      "loss": 3.0643,
+      "step": 8370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002958097527823871,
+      "loss": 3.0959,
+      "step": 8380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002957997884292852,
+      "loss": 3.0493,
+      "step": 8390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002957898124109246,
+      "loss": 3.0442,
+      "step": 8400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002957798247281034,
+      "loss": 3.068,
+      "step": 8410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029576982538162083,
+      "loss": 3.0389,
+      "step": 8420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002957598143722769,
+      "loss": 3.0408,
+      "step": 8430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029574979170087255,
+      "loss": 3.0643,
+      "step": 8440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002957397573682097,
+      "loss": 3.0681,
+      "step": 8450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002957297113750912,
+      "loss": 3.0757,
+      "step": 8460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002957196537223208,
+      "loss": 3.0495,
+      "step": 8470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029570958441070316,
+      "loss": 3.019,
+      "step": 8480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029569950344104397,
+      "loss": 3.0433,
+      "step": 8490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029568941081414976,
+      "loss": 3.0481,
+      "step": 8500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029567930653082804,
+      "loss": 3.0305,
+      "step": 8510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029566919059188725,
+      "loss": 3.0699,
+      "step": 8520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029565906299813675,
+      "loss": 3.0219,
+      "step": 8530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029564892375038686,
+      "loss": 3.0687,
+      "step": 8540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029563877284944875,
+      "loss": 3.0251,
+      "step": 8550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029562861029613463,
+      "loss": 3.0146,
+      "step": 8560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002956184360912576,
+      "loss": 3.0418,
+      "step": 8570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029560825023563163,
+      "loss": 3.0322,
+      "step": 8580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029559805273007166,
+      "loss": 3.0739,
+      "step": 8590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002955878435753937,
+      "loss": 3.0625,
+      "step": 8600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002955776227724145,
+      "loss": 3.0672,
+      "step": 8610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002955673903219518,
+      "loss": 3.0304,
+      "step": 8620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029555714622482435,
+      "loss": 3.001,
+      "step": 8630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002955468904818516,
+      "loss": 3.0411,
+      "step": 8640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029553662309385434,
+      "loss": 3.0182,
+      "step": 8650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002955263440616539,
+      "loss": 3.0461,
+      "step": 8660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002955160533860727,
+      "loss": 3.0539,
+      "step": 8670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000295505751067934,
+      "loss": 2.9968,
+      "step": 8680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002954954371080623,
+      "loss": 3.0386,
+      "step": 8690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002954851115072827,
+      "loss": 3.0238,
+      "step": 8700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029547477426642127,
+      "loss": 3.0155,
+      "step": 8710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002954644253863052,
+      "loss": 3.0014,
+      "step": 8720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002954540648677624,
+      "loss": 3.0408,
+      "step": 8730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002954436927116218,
+      "loss": 3.0541,
+      "step": 8740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002954333089187133,
+      "loss": 2.9756,
+      "step": 8750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029542291348986767,
+      "loss": 3.0619,
+      "step": 8760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002954125064259167,
+      "loss": 3.0218,
+      "step": 8770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029540208772769295,
+      "loss": 3.0225,
+      "step": 8780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002953916573960301,
+      "loss": 3.0219,
+      "step": 8790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002953812154317626,
+      "loss": 3.0366,
+      "step": 8800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029537076183572584,
+      "loss": 3.0551,
+      "step": 8810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002953602966087564,
+      "loss": 3.0189,
+      "step": 8820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002953498197516914,
+      "loss": 3.0355,
+      "step": 8830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029533933126536923,
+      "loss": 3.0119,
+      "step": 8840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002953288311506289,
+      "loss": 3.0208,
+      "step": 8850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002953183194083106,
+      "loss": 3.0384,
+      "step": 8860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029530779603925537,
+      "loss": 3.0288,
+      "step": 8870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002952972610443052,
+      "loss": 3.0126,
+      "step": 8880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029528671442430283,
+      "loss": 3.026,
+      "step": 8890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029527615618009227,
+      "loss": 3.0238,
+      "step": 8900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002952655863125182,
+      "loss": 3.0282,
+      "step": 8910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002952550048224262,
+      "loss": 2.9681,
+      "step": 8920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002952444117106631,
+      "loss": 3.0175,
+      "step": 8930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002952338069780762,
+      "loss": 3.0304,
+      "step": 8940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002952231906255142,
+      "loss": 3.05,
+      "step": 8950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029521256265382625,
+      "loss": 3.0625,
+      "step": 8960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002952019230638629,
+      "loss": 3.0095,
+      "step": 8970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002951912718564753,
+      "loss": 3.0204,
+      "step": 8980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029518060903251575,
+      "loss": 3.0424,
+      "step": 8990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029516993459283717,
+      "loss": 3.0185,
+      "step": 9000
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 3.0578036308288574,
+      "eval_runtime": 1263.0467,
+      "eval_samples_per_second": 60.279,
+      "eval_steps_per_second": 7.535,
+      "step": 9000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029515924853829376,
+      "loss": 3.0597,
+      "step": 9010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002951485508697405,
+      "loss": 3.0298,
+      "step": 9020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029513784158803326,
+      "loss": 3.0113,
+      "step": 9030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029512712069402884,
+      "loss": 3.0256,
+      "step": 9040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002951163881885851,
+      "loss": 3.0211,
+      "step": 9050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002951056440725606,
+      "loss": 2.9975,
+      "step": 9060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029509488834681514,
+      "loss": 2.9914,
+      "step": 9070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002950841210122091,
+      "loss": 3.0043,
+      "step": 9080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029507334206960406,
+      "loss": 3.0386,
+      "step": 9090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002950625515198624,
+      "loss": 3.0045,
+      "step": 9100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029505174936384745,
+      "loss": 3.0167,
+      "step": 9110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029504093560242344,
+      "loss": 2.9882,
+      "step": 9120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002950301102364557,
+      "loss": 3.0267,
+      "step": 9130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002950192732668102,
+      "loss": 3.0081,
+      "step": 9140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029500842469435415,
+      "loss": 3.0016,
+      "step": 9150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002949975645199554,
+      "loss": 3.0318,
+      "step": 9160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002949866927444829,
+      "loss": 3.0074,
+      "step": 9170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029497580936880646,
+      "loss": 3.0361,
+      "step": 9180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002949649143937969,
+      "loss": 2.9849,
+      "step": 9190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029495400782032587,
+      "loss": 3.0168,
+      "step": 9200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029494308964926597,
+      "loss": 2.9908,
+      "step": 9210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029493215988149083,
+      "loss": 3.0127,
+      "step": 9220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002949212185178749,
+      "loss": 3.0461,
+      "step": 9230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029491026555929353,
+      "loss": 3.0364,
+      "step": 9240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002948993010066231,
+      "loss": 3.0262,
+      "step": 9250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002948883248607409,
+      "loss": 3.0245,
+      "step": 9260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000294877337122525,
+      "loss": 3.0012,
+      "step": 9270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002948663377928547,
+      "loss": 2.996,
+      "step": 9280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002948553268726099,
+      "loss": 3.0212,
+      "step": 9290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002948443043626715,
+      "loss": 2.9864,
+      "step": 9300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002948332702639216,
+      "loss": 3.0105,
+      "step": 9310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029482222457724296,
+      "loss": 3.0253,
+      "step": 9320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029481116730351926,
+      "loss": 3.0308,
+      "step": 9330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002948000984436352,
+      "loss": 3.0255,
+      "step": 9340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002947890179984764,
+      "loss": 3.0345,
+      "step": 9350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002947779259689294,
+      "loss": 3.0168,
+      "step": 9360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002947668223558817,
+      "loss": 3.0096,
+      "step": 9370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002947557071602216,
+      "loss": 2.9982,
+      "step": 9380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029474458038283846,
+      "loss": 2.9963,
+      "step": 9390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029473344202462256,
+      "loss": 3.063,
+      "step": 9400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029472229208646496,
+      "loss": 2.9875,
+      "step": 9410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029471113056925784,
+      "loss": 2.9658,
+      "step": 9420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029469995747389416,
+      "loss": 3.0146,
+      "step": 9430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029468877280126795,
+      "loss": 2.9802,
+      "step": 9440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000294677576552274,
+      "loss": 2.9839,
+      "step": 9450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002946663687278082,
+      "loss": 2.9994,
+      "step": 9460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002946551493287671,
+      "loss": 3.0042,
+      "step": 9470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002946439183560485,
+      "loss": 2.993,
+      "step": 9480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000294632675810551,
+      "loss": 3.0364,
+      "step": 9490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029462142169317393,
+      "loss": 2.9943,
+      "step": 9500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002946101560048179,
+      "loss": 3.0059,
+      "step": 9510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029459887874638417,
+      "loss": 3.0195,
+      "step": 9520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029458758991877503,
+      "loss": 3.001,
+      "step": 9530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029457628952289373,
+      "loss": 2.972,
+      "step": 9540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002945649775596444,
+      "loss": 3.0164,
+      "step": 9550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029455365402993196,
+      "loss": 2.9772,
+      "step": 9560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029454231893466255,
+      "loss": 2.9813,
+      "step": 9570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000294530972274743,
+      "loss": 2.9855,
+      "step": 9580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029451961405108117,
+      "loss": 2.9526,
+      "step": 9590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029450824426458585,
+      "loss": 2.9906,
+      "step": 9600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029449686291616663,
+      "loss": 3.0095,
+      "step": 9610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002944854700067342,
+      "loss": 3.0197,
+      "step": 9620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029447406553720004,
+      "loss": 2.9829,
+      "step": 9630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002944626495084766,
+      "loss": 2.9755,
+      "step": 9640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029445122192147733,
+      "loss": 3.0022,
+      "step": 9650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002944397827771165,
+      "loss": 2.9964,
+      "step": 9660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002944283320763094,
+      "loss": 2.9992,
+      "step": 9670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000294416869819972,
+      "loss": 3.0028,
+      "step": 9680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002944053960090216,
+      "loss": 2.9762,
+      "step": 9690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029439391064437604,
+      "loss": 3.0013,
+      "step": 9700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002943824137269544,
+      "loss": 2.9737,
+      "step": 9710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002943709052576764,
+      "loss": 3.0341,
+      "step": 9720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002943593852374629,
+      "loss": 2.9832,
+      "step": 9730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029434785366723557,
+      "loss": 2.9471,
+      "step": 9740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000294336310547917,
+      "loss": 2.9905,
+      "step": 9750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002943247558804309,
+      "loss": 3.015,
+      "step": 9760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002943131896657015,
+      "loss": 2.9692,
+      "step": 9770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029430161190465435,
+      "loss": 3.0242,
+      "step": 9780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002942900225982158,
+      "loss": 2.9922,
+      "step": 9790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000294278421747313,
+      "loss": 2.9768,
+      "step": 9800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002942668093528742,
+      "loss": 2.9744,
+      "step": 9810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002942551854158284,
+      "loss": 2.9578,
+      "step": 9820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029424354993710575,
+      "loss": 2.9956,
+      "step": 9830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002942319029176371,
+      "loss": 3.0201,
+      "step": 9840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002942202443583543,
+      "loss": 3.0025,
+      "step": 9850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029420857426019014,
+      "loss": 3.0371,
+      "step": 9860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029419689262407837,
+      "loss": 2.9824,
+      "step": 9870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002941851994509536,
+      "loss": 2.9894,
+      "step": 9880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002941734947417514,
+      "loss": 3.0096,
+      "step": 9890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002941617784974083,
+      "loss": 2.9696,
+      "step": 9900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002941500507188616,
+      "loss": 2.996,
+      "step": 9910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029413831140704966,
+      "loss": 2.9802,
+      "step": 9920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029412656056291176,
+      "loss": 2.9473,
+      "step": 9930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000294114798187388,
+      "loss": 2.9769,
+      "step": 9940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002941030242814196,
+      "loss": 2.97,
+      "step": 9950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002940912388459484,
+      "loss": 2.9625,
+      "step": 9960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002940794418819175,
+      "loss": 3.0007,
+      "step": 9970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029406763339027064,
+      "loss": 2.9699,
+      "step": 9980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029405581337195273,
+      "loss": 3.0,
+      "step": 9990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029404398182790936,
+      "loss": 2.9496,
+      "step": 10000
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 3.015676736831665,
+      "eval_runtime": 1263.3403,
+      "eval_samples_per_second": 60.265,
+      "eval_steps_per_second": 7.533,
+      "step": 10000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002940321387590872,
+      "loss": 2.967,
+      "step": 10010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029402028416643375,
+      "loss": 2.9832,
+      "step": 10020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029400841805089756,
+      "loss": 2.9475,
+      "step": 10030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002939965404134281,
+      "loss": 2.9791,
+      "step": 10040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002939846512549754,
+      "loss": 2.9918,
+      "step": 10050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000293972750576491,
+      "loss": 2.9811,
+      "step": 10060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029396083837892684,
+      "loss": 2.9826,
+      "step": 10070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002939489146632362,
+      "loss": 2.9575,
+      "step": 10080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029393697943037286,
+      "loss": 2.947,
+      "step": 10090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000293925032681292,
+      "loss": 2.9215,
+      "step": 10100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029391307441694914,
+      "loss": 2.9739,
+      "step": 10110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029390110463830134,
+      "loss": 2.9678,
+      "step": 10120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029388912334630613,
+      "loss": 2.9656,
+      "step": 10130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002938771305419222,
+      "loss": 2.9903,
+      "step": 10140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000293865126226109,
+      "loss": 2.9693,
+      "step": 10150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002938531103998271,
+      "loss": 2.9837,
+      "step": 10160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002938410830640377,
+      "loss": 2.9656,
+      "step": 10170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029382904421970327,
+      "loss": 3.0083,
+      "step": 10180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029381699386778686,
+      "loss": 2.9379,
+      "step": 10190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002938049320092527,
+      "loss": 2.9609,
+      "step": 10200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002937928586450659,
+      "loss": 2.9488,
+      "step": 10210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029378077377619226,
+      "loss": 2.9555,
+      "step": 10220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002937686774035988,
+      "loss": 2.9645,
+      "step": 10230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029375656952825333,
+      "loss": 2.9751,
+      "step": 10240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002937444501511245,
+      "loss": 2.9976,
+      "step": 10250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029373231927318216,
+      "loss": 2.9421,
+      "step": 10260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002937201768953967,
+      "loss": 2.9879,
+      "step": 10270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029370802301873964,
+      "loss": 2.9481,
+      "step": 10280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002936958576441835,
+      "loss": 2.9474,
+      "step": 10290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002936836807727015,
+      "loss": 2.9652,
+      "step": 10300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029367149240526794,
+      "loss": 2.9473,
+      "step": 10310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000293659292542858,
+      "loss": 3.0229,
+      "step": 10320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002936470811864478,
+      "loss": 2.9465,
+      "step": 10330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029363485833701434,
+      "loss": 2.9975,
+      "step": 10340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029362262399553553,
+      "loss": 2.9536,
+      "step": 10350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002936103781629902,
+      "loss": 2.9521,
+      "step": 10360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029359812084035823,
+      "loss": 2.9676,
+      "step": 10370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029358585202862025,
+      "loss": 2.9556,
+      "step": 10380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029357357172875786,
+      "loss": 2.9422,
+      "step": 10390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029356127994175355,
+      "loss": 2.9717,
+      "step": 10400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029354897666859087,
+      "loss": 2.9954,
+      "step": 10410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029353666191025415,
+      "loss": 2.9506,
+      "step": 10420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029352433566772867,
+      "loss": 2.9587,
+      "step": 10430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002935119979420006,
+      "loss": 2.9852,
+      "step": 10440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002934996487340571,
+      "loss": 2.9638,
+      "step": 10450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002934872880448863,
+      "loss": 2.9688,
+      "step": 10460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029347491587547697,
+      "loss": 2.9501,
+      "step": 10470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002934625322268192,
+      "loss": 2.9686,
+      "step": 10480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029345013709990365,
+      "loss": 2.9953,
+      "step": 10490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029343773049572203,
+      "loss": 2.9843,
+      "step": 10500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029342531241526715,
+      "loss": 2.9598,
+      "step": 10510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029341288285953234,
+      "loss": 2.93,
+      "step": 10520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029340044182951224,
+      "loss": 2.9703,
+      "step": 10530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029338798932620217,
+      "loss": 2.9547,
+      "step": 10540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029337552535059843,
+      "loss": 2.991,
+      "step": 10550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029336304990369824,
+      "loss": 2.9727,
+      "step": 10560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002933505629864998,
+      "loss": 2.9521,
+      "step": 10570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029333806460000214,
+      "loss": 2.936,
+      "step": 10580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029332555474520527,
+      "loss": 2.9629,
+      "step": 10590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029331303342311,
+      "loss": 2.9565,
+      "step": 10600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002933005006347183,
+      "loss": 2.9509,
+      "step": 10610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002932879563810327,
+      "loss": 2.9245,
+      "step": 10620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000293275400663057,
+      "loss": 2.97,
+      "step": 10630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029326283348179576,
+      "loss": 2.9516,
+      "step": 10640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002932502548382544,
+      "loss": 2.9821,
+      "step": 10650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002932376647334393,
+      "loss": 2.9267,
+      "step": 10660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002932250631683579,
+      "loss": 2.9599,
+      "step": 10670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029321245014401836,
+      "loss": 2.9609,
+      "step": 10680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029319982566142986,
+      "loss": 2.9653,
+      "step": 10690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002931871897216024,
+      "loss": 2.9393,
+      "step": 10700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029317454232554704,
+      "loss": 2.9789,
+      "step": 10710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002931618834742756,
+      "loss": 2.9535,
+      "step": 10720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029314921316880106,
+      "loss": 2.94,
+      "step": 10730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002931365314101369,
+      "loss": 2.9243,
+      "step": 10740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029312383819929806,
+      "loss": 2.9311,
+      "step": 10750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002931111335372999,
+      "loss": 2.9383,
+      "step": 10760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029309841742515905,
+      "loss": 2.952,
+      "step": 10770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029308568986389274,
+      "loss": 2.946,
+      "step": 10780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029307295085451943,
+      "loss": 2.9612,
+      "step": 10790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029306020039805827,
+      "loss": 2.9499,
+      "step": 10800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002930474384955295,
+      "loss": 2.9664,
+      "step": 10810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029303466514795404,
+      "loss": 2.9936,
+      "step": 10820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000293021880356354,
+      "loss": 2.9327,
+      "step": 10830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002930090841217522,
+      "loss": 2.9356,
+      "step": 10840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002929962764451725,
+      "loss": 2.9567,
+      "step": 10850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002929834573276396,
+      "loss": 2.9716,
+      "step": 10860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002929706267701792,
+      "loss": 2.9511,
+      "step": 10870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002929577847738177,
+      "loss": 2.968,
+      "step": 10880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029294493133958277,
+      "loss": 2.923,
+      "step": 10890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029293206646850267,
+      "loss": 2.9527,
+      "step": 10900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002929191901616067,
+      "loss": 2.9627,
+      "step": 10910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029290630241992515,
+      "loss": 2.9357,
+      "step": 10920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002928934032444891,
+      "loss": 2.9435,
+      "step": 10930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029288049263633063,
+      "loss": 2.9429,
+      "step": 10940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029286757059648265,
+      "loss": 2.9281,
+      "step": 10950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029285463712597914,
+      "loss": 2.9606,
+      "step": 10960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002928416922258547,
+      "loss": 2.9604,
+      "step": 10970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002928287358971453,
+      "loss": 2.9297,
+      "step": 10980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002928157681408873,
+      "loss": 2.9722,
+      "step": 10990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002928027889581184,
+      "loss": 2.9374,
+      "step": 11000
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 2.98154354095459,
+      "eval_runtime": 1262.7818,
+      "eval_samples_per_second": 60.291,
+      "eval_steps_per_second": 7.537,
+      "step": 11000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029278979834987704,
+      "loss": 2.9293,
+      "step": 11010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029277679631720246,
+      "loss": 2.9367,
+      "step": 11020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029276378286113505,
+      "loss": 2.956,
+      "step": 11030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029275075798271597,
+      "loss": 2.927,
+      "step": 11040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029273772168298735,
+      "loss": 2.9089,
+      "step": 11050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002927246739629921,
+      "loss": 2.9383,
+      "step": 11060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029271161482377427,
+      "loss": 2.9081,
+      "step": 11070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029269854426637864,
+      "loss": 2.9513,
+      "step": 11080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029268546229185106,
+      "loss": 2.9391,
+      "step": 11090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029267236890123807,
+      "loss": 2.9849,
+      "step": 11100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002926592640955874,
+      "loss": 2.9485,
+      "step": 11110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029264614787594736,
+      "loss": 2.9386,
+      "step": 11120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029263302024336757,
+      "loss": 2.9357,
+      "step": 11130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029261988119889823,
+      "loss": 2.9154,
+      "step": 11140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002926067307435906,
+      "loss": 2.9139,
+      "step": 11150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029259356887849684,
+      "loss": 2.9032,
+      "step": 11160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029258039560467003,
+      "loss": 2.9259,
+      "step": 11170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002925672109231641,
+      "loss": 2.9256,
+      "step": 11180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000292554014835034,
+      "loss": 2.9335,
+      "step": 11190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029254080734133554,
+      "loss": 2.9027,
+      "step": 11200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002925275884431253,
+      "loss": 2.9042,
+      "step": 11210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029251435814146105,
+      "loss": 2.9122,
+      "step": 11220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002925011164374013,
+      "loss": 2.9645,
+      "step": 11230
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002924878633320055,
+      "loss": 2.9235,
+      "step": 11240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000292474598826334,
+      "loss": 2.9188,
+      "step": 11250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029246132292144804,
+      "loss": 2.9332,
+      "step": 11260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029244803561840986,
+      "loss": 2.895,
+      "step": 11270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002924347369182826,
+      "loss": 2.9131,
+      "step": 11280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029242142682213017,
+      "loss": 2.9401,
+      "step": 11290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002924081053310176,
+      "loss": 2.9622,
+      "step": 11300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002923947724460106,
+      "loss": 2.9234,
+      "step": 11310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000292381428168176,
+      "loss": 2.8935,
+      "step": 11320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029236807249858146,
+      "loss": 2.8994,
+      "step": 11330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029235470543829555,
+      "loss": 2.8563,
+      "step": 11340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002923413269883877,
+      "loss": 2.9086,
+      "step": 11350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002923279371499284,
+      "loss": 2.9191,
+      "step": 11360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029231453592398884,
+      "loss": 2.9275,
+      "step": 11370
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029230112331164135,
+      "loss": 2.903,
+      "step": 11380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029228769931395897,
+      "loss": 2.9273,
+      "step": 11390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029227426393201576,
+      "loss": 2.907,
+      "step": 11400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029226081716688664,
+      "loss": 2.8869,
+      "step": 11410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029224735901964754,
+      "loss": 2.8847,
+      "step": 11420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029223388949137523,
+      "loss": 2.9373,
+      "step": 11430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002922204085831473,
+      "loss": 2.9338,
+      "step": 11440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002922069162960424,
+      "loss": 2.893,
+      "step": 11450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029219341263114003,
+      "loss": 2.9534,
+      "step": 11460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029217989758952055,
+      "loss": 2.907,
+      "step": 11470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002921663711722654,
+      "loss": 2.8934,
+      "step": 11480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002921528333804567,
+      "loss": 2.9087,
+      "step": 11490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002921392842151776,
+      "loss": 2.8718,
+      "step": 11500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029212572367751224,
+      "loss": 2.9288,
+      "step": 11510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002921121517685455,
+      "loss": 2.9068,
+      "step": 11520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002920985684893633,
+      "loss": 2.8809,
+      "step": 11530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029208497384105236,
+      "loss": 2.9073,
+      "step": 11540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002920713678247004,
+      "loss": 2.9239,
+      "step": 11550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029205775044139606,
+      "loss": 2.9061,
+      "step": 11560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002920441216922288,
+      "loss": 2.8997,
+      "step": 11570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029203048157828905,
+      "loss": 2.9225,
+      "step": 11580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029201683010066814,
+      "loss": 2.9108,
+      "step": 11590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029200316726045835,
+      "loss": 2.88,
+      "step": 11600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002919894930587528,
+      "loss": 2.9091,
+      "step": 11610
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002919758074966455,
+      "loss": 2.883,
+      "step": 11620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002919621105752314,
+      "loss": 2.9072,
+      "step": 11630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002919484022956065,
+      "loss": 2.8887,
+      "step": 11640
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029193468265886746,
+      "loss": 2.8988,
+      "step": 11650
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029192095166611205,
+      "loss": 2.8958,
+      "step": 11660
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002919072093184388,
+      "loss": 2.8826,
+      "step": 11670
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029189345561694727,
+      "loss": 2.8953,
+      "step": 11680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002918796905627379,
+      "loss": 2.8785,
+      "step": 11690
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029186591415691194,
+      "loss": 2.9246,
+      "step": 11700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029185212640057165,
+      "loss": 2.8806,
+      "step": 11710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002918383272948202,
+      "loss": 2.8333,
+      "step": 11720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002918245168407616,
+      "loss": 2.8866,
+      "step": 11730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002918106950395008,
+      "loss": 2.8747,
+      "step": 11740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002917968618921437,
+      "loss": 2.897,
+      "step": 11750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002917830173997971,
+      "loss": 2.8977,
+      "step": 11760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029176916156356867,
+      "loss": 2.8972,
+      "step": 11770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002917552943845669,
+      "loss": 2.9113,
+      "step": 11780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002917414158639014,
+      "loss": 2.8794,
+      "step": 11790
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002917275260026825,
+      "loss": 2.8887,
+      "step": 11800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029171362480202157,
+      "loss": 2.9034,
+      "step": 11810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029169971226303076,
+      "loss": 2.94,
+      "step": 11820
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002916857883868233,
+      "loss": 2.909,
+      "step": 11830
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002916718531745131,
+      "loss": 2.9229,
+      "step": 11840
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029165790662721516,
+      "loss": 2.9078,
+      "step": 11850
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029164394874604533,
+      "loss": 2.9164,
+      "step": 11860
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029162997953212036,
+      "loss": 2.9093,
+      "step": 11870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002916159989865579,
+      "loss": 2.8978,
+      "step": 11880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002916020071104765,
+      "loss": 2.8562,
+      "step": 11890
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002915880039049957,
+      "loss": 2.9251,
+      "step": 11900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002915739893712358,
+      "loss": 2.8971,
+      "step": 11910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029155996351031807,
+      "loss": 2.8574,
+      "step": 11920
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029154592632336484,
+      "loss": 2.892,
+      "step": 11930
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029153187781149904,
+      "loss": 2.8924,
+      "step": 11940
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029151781797584475,
+      "loss": 2.8857,
+      "step": 11950
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002915037468175269,
+      "loss": 2.9205,
+      "step": 11960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029148966433767133,
+      "loss": 2.8966,
+      "step": 11970
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002914755705374046,
+      "loss": 2.916,
+      "step": 11980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029146146541785456,
+      "loss": 2.894,
+      "step": 11990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002914473489801496,
+      "loss": 2.8794,
+      "step": 12000
+    },
+    {
+      "epoch": 1.08,
+      "eval_loss": 2.9487366676330566,
+      "eval_runtime": 1263.1596,
+      "eval_samples_per_second": 60.273,
+      "eval_steps_per_second": 7.534,
+      "step": 12000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002914332212254192,
+      "loss": 2.8792,
+      "step": 12010
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029141908215479367,
+      "loss": 2.9167,
+      "step": 12020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002914049317694043,
+      "loss": 2.882,
+      "step": 12030
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029139077007038326,
+      "loss": 2.8444,
+      "step": 12040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029137659705886354,
+      "loss": 2.8763,
+      "step": 12050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029136241273597916,
+      "loss": 2.9046,
+      "step": 12060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000291348217102865,
+      "loss": 2.9113,
+      "step": 12070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002913340101606568,
+      "loss": 2.8746,
+      "step": 12080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029131979191049124,
+      "loss": 2.8993,
+      "step": 12090
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029130556235350594,
+      "loss": 2.9011,
+      "step": 12100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002912913214908393,
+      "loss": 2.9036,
+      "step": 12110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029127706932363087,
+      "loss": 2.9042,
+      "step": 12120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002912628058530208,
+      "loss": 2.9092,
+      "step": 12130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029124853108015037,
+      "loss": 2.875,
+      "step": 12140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029123424500616165,
+      "loss": 2.8744,
+      "step": 12150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00029121994763219765,
+      "loss": 2.8978,
+      "step": 12160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029120563895940237,
+      "loss": 2.9034,
+      "step": 12170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002911913189889205,
+      "loss": 2.8915,
+      "step": 12180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029117698772189785,
+      "loss": 2.9131,
+      "step": 12190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000291162645159481,
+      "loss": 2.8716,
+      "step": 12200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002911482913028175,
+      "loss": 2.9156,
+      "step": 12210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029113392615305584,
+      "loss": 2.8888,
+      "step": 12220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029111954971134524,
+      "loss": 2.8679,
+      "step": 12230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000291105161978836,
+      "loss": 2.9288,
+      "step": 12240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002910907629566793,
+      "loss": 2.8852,
+      "step": 12250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029107635264602716,
+      "loss": 2.8976,
+      "step": 12260
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002910619310480325,
+      "loss": 2.9128,
+      "step": 12270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002910474981638492,
+      "loss": 2.91,
+      "step": 12280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000291033053994632,
+      "loss": 2.8897,
+      "step": 12290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029101859854153663,
+      "loss": 2.9202,
+      "step": 12300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002910041318057196,
+      "loss": 2.862,
+      "step": 12310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029098965378833837,
+      "loss": 2.8729,
+      "step": 12320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002909751644905513,
+      "loss": 2.8668,
+      "step": 12330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002909606639135177,
+      "loss": 2.9044,
+      "step": 12340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029094615205839765,
+      "loss": 2.879,
+      "step": 12350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002909316289263523,
+      "loss": 2.8554,
+      "step": 12360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002909170945185437,
+      "loss": 2.8854,
+      "step": 12370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002909025488361346,
+      "loss": 2.8889,
+      "step": 12380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002908879918802889,
+      "loss": 2.8689,
+      "step": 12390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002908734236521711,
+      "loss": 2.8778,
+      "step": 12400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00029085884415294694,
+      "loss": 2.8776,
+      "step": 12410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002908442533837829,
+      "loss": 2.8688,
+      "step": 12420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002908296513458463,
+      "loss": 2.868,
+      "step": 12430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00029081503804030554,
+      "loss": 2.8812,
+      "step": 12440
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00029080041346832964,
+      "loss": 2.9104,
+      "step": 12450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00029078577763108886,
+      "loss": 2.9102,
+      "step": 12460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00029077113052975416,
+      "loss": 2.9185,
+      "step": 12470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002907564721654973,
+      "loss": 2.8828,
+      "step": 12480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00029074327000883717,
+      "loss": 2.9025,
+      "step": 12490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002907285902482603,
+      "loss": 2.8892,
+      "step": 12500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00029071389922816497,
+      "loss": 2.8846,
+      "step": 12510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002906991969497265,
+      "loss": 2.87,
+      "step": 12520
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002906844834141214,
+      "loss": 2.9197,
+      "step": 12530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00029066975862252674,
+      "loss": 2.896,
+      "step": 12540
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002906550225761206,
+      "loss": 2.8778,
+      "step": 12550
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00029064027527608214,
+      "loss": 2.8632,
+      "step": 12560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002906255167235912,
+      "loss": 2.8683,
+      "step": 12570
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00029061074691982857,
+      "loss": 2.8946,
+      "step": 12580
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00029059596586597595,
+      "loss": 2.8625,
+      "step": 12590
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00029058117356321603,
+      "loss": 2.8757,
+      "step": 12600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00029056637001273223,
+      "loss": 2.8585,
+      "step": 12610
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000290551555215709,
+      "loss": 2.8617,
+      "step": 12620
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002905367291733317,
+      "loss": 2.9061,
+      "step": 12630
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002905218918867865,
+      "loss": 2.8895,
+      "step": 12640
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002905070433572605,
+      "loss": 2.8989,
+      "step": 12650
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00029049218358594175,
+      "loss": 2.8633,
+      "step": 12660
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00029047731257401916,
+      "loss": 2.8429,
+      "step": 12670
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00029046243032268245,
+      "loss": 2.8718,
+      "step": 12680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002904475368331225,
+      "loss": 2.8743,
+      "step": 12690
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002904326321065307,
+      "loss": 2.8724,
+      "step": 12700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00029041771614409976,
+      "loss": 2.8991,
+      "step": 12710
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029040278894702304,
+      "loss": 2.8635,
+      "step": 12720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029038785051649477,
+      "loss": 2.8774,
+      "step": 12730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002903729008537102,
+      "loss": 2.8722,
+      "step": 12740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029035793995986546,
+      "loss": 2.8878,
+      "step": 12750
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029034296783615754,
+      "loss": 2.8739,
+      "step": 12760
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029032798448378426,
+      "loss": 2.8993,
+      "step": 12770
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029031298990394455,
+      "loss": 2.9003,
+      "step": 12780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029029798409783807,
+      "loss": 2.8649,
+      "step": 12790
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002902829670666653,
+      "loss": 2.8764,
+      "step": 12800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029026793881162786,
+      "loss": 2.8859,
+      "step": 12810
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029025289933392814,
+      "loss": 2.8774,
+      "step": 12820
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00029023784863476933,
+      "loss": 2.8984,
+      "step": 12830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00029022278671535576,
+      "loss": 2.913,
+      "step": 12840
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002902077135768924,
+      "loss": 2.8776,
+      "step": 12850
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002901926292205853,
+      "loss": 2.9011,
+      "step": 12860
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00029017753364764124,
+      "loss": 2.8829,
+      "step": 12870
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00029016242685926815,
+      "loss": 2.8622,
+      "step": 12880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002901473088566746,
+      "loss": 2.8948,
+      "step": 12890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00029013217964107016,
+      "loss": 2.8574,
+      "step": 12900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00029011703921366536,
+      "loss": 2.8581,
+      "step": 12910
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002901018875756715,
+      "loss": 2.8854,
+      "step": 12920
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00029008672472830097,
+      "loss": 2.8544,
+      "step": 12930
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00029007155067276675,
+      "loss": 2.8412,
+      "step": 12940
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000290056365410283,
+      "loss": 2.811,
+      "step": 12950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002900411689420647,
+      "loss": 2.8333,
+      "step": 12960
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00029002596126932755,
+      "loss": 2.8668,
+      "step": 12970
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002900107423932885,
+      "loss": 2.8999,
+      "step": 12980
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028999551231516515,
+      "loss": 2.8599,
+      "step": 12990
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028998027103617594,
+      "loss": 2.8407,
+      "step": 13000
+    },
+    {
+      "epoch": 1.17,
+      "eval_loss": 2.9229493141174316,
+      "eval_runtime": 1263.3336,
+      "eval_samples_per_second": 60.265,
+      "eval_steps_per_second": 7.533,
+      "step": 13000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028996501855754026,
+      "loss": 2.9025,
+      "step": 13010
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028994975488047867,
+      "loss": 2.8272,
+      "step": 13020
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028993448000621223,
+      "loss": 2.8932,
+      "step": 13030
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000289919193935963,
+      "loss": 2.8812,
+      "step": 13040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002899038966709542,
+      "loss": 2.8603,
+      "step": 13050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028988858821240963,
+      "loss": 2.8881,
+      "step": 13060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002898732685615541,
+      "loss": 2.8568,
+      "step": 13070
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028985793771961336,
+      "loss": 2.8705,
+      "step": 13080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028984259568781397,
+      "loss": 2.8014,
+      "step": 13090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002898272424673834,
+      "loss": 2.8487,
+      "step": 13100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002898118780595501,
+      "loss": 2.8529,
+      "step": 13110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028979650246554334,
+      "loss": 2.8686,
+      "step": 13120
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002897811156865933,
+      "loss": 2.8453,
+      "step": 13130
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028976571772393106,
+      "loss": 2.8334,
+      "step": 13140
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002897503085787886,
+      "loss": 2.8772,
+      "step": 13150
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028973488825239876,
+      "loss": 2.849,
+      "step": 13160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028971945674599527,
+      "loss": 2.867,
+      "step": 13170
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002897040140608129,
+      "loss": 2.8626,
+      "step": 13180
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002896885601980871,
+      "loss": 2.8564,
+      "step": 13190
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002896730951590543,
+      "loss": 2.9008,
+      "step": 13200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028965761894495195,
+      "loss": 2.8803,
+      "step": 13210
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002896421315570182,
+      "loss": 2.8754,
+      "step": 13220
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028962663299649216,
+      "loss": 2.8841,
+      "step": 13230
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002896111232646139,
+      "loss": 2.8824,
+      "step": 13240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002895956023626243,
+      "loss": 2.8689,
+      "step": 13250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028958007029176525,
+      "loss": 2.8624,
+      "step": 13260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028956452705327935,
+      "loss": 2.8362,
+      "step": 13270
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002895489726484102,
+      "loss": 2.8418,
+      "step": 13280
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028953340707840237,
+      "loss": 2.8637,
+      "step": 13290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028951783034450115,
+      "loss": 2.8377,
+      "step": 13300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002895022424479529,
+      "loss": 2.8498,
+      "step": 13310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002894866433900047,
+      "loss": 2.8335,
+      "step": 13320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002894710331719047,
+      "loss": 2.8665,
+      "step": 13330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028945541179490177,
+      "loss": 2.88,
+      "step": 13340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002894397792602458,
+      "loss": 2.8361,
+      "step": 13350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028942413556918753,
+      "loss": 2.8658,
+      "step": 13360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028940848072297863,
+      "loss": 2.8588,
+      "step": 13370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028939281472287155,
+      "loss": 2.8231,
+      "step": 13380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028937713757011974,
+      "loss": 2.8813,
+      "step": 13390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002893614492659775,
+      "loss": 2.8376,
+      "step": 13400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002893457498117,
+      "loss": 2.8628,
+      "step": 13410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028933003920854344,
+      "loss": 2.8485,
+      "step": 13420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002893143174577647,
+      "loss": 2.8758,
+      "step": 13430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028929858456062176,
+      "loss": 2.8467,
+      "step": 13440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028928284051837324,
+      "loss": 2.8291,
+      "step": 13450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028926708533227896,
+      "loss": 2.8567,
+      "step": 13460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002892513190035994,
+      "loss": 2.8715,
+      "step": 13470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000289235541533596,
+      "loss": 2.8416,
+      "step": 13480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002892197529235311,
+      "loss": 2.8513,
+      "step": 13490
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028920395317466786,
+      "loss": 2.8656,
+      "step": 13500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028918814228827053,
+      "loss": 2.8684,
+      "step": 13510
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002891723202656041,
+      "loss": 2.8399,
+      "step": 13520
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028915648710793433,
+      "loss": 2.8528,
+      "step": 13530
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028914064281652816,
+      "loss": 2.8794,
+      "step": 13540
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002891247873926532,
+      "loss": 2.8402,
+      "step": 13550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028910892083757806,
+      "loss": 2.8503,
+      "step": 13560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002890930431525722,
+      "loss": 2.8808,
+      "step": 13570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028907715433890586,
+      "loss": 2.8806,
+      "step": 13580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028906125439785046,
+      "loss": 2.8515,
+      "step": 13590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028904534333067805,
+      "loss": 2.8634,
+      "step": 13600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002890294211386617,
+      "loss": 2.8522,
+      "step": 13610
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002890134878230752,
+      "loss": 2.8498,
+      "step": 13620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002889975433851935,
+      "loss": 2.8567,
+      "step": 13630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028898158782629226,
+      "loss": 2.8494,
+      "step": 13640
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000288965621147648,
+      "loss": 2.8521,
+      "step": 13650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028894964335053825,
+      "loss": 2.824,
+      "step": 13660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002889336544362413,
+      "loss": 2.8537,
+      "step": 13670
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028891765440603655,
+      "loss": 2.8738,
+      "step": 13680
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028890164326120395,
+      "loss": 2.8488,
+      "step": 13690
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028888562100302473,
+      "loss": 2.8285,
+      "step": 13700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002888695876327807,
+      "loss": 2.8555,
+      "step": 13710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028885354315175463,
+      "loss": 2.8921,
+      "step": 13720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028883748756123033,
+      "loss": 2.8779,
+      "step": 13730
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002888214208624923,
+      "loss": 2.8778,
+      "step": 13740
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000288805343056826,
+      "loss": 2.8604,
+      "step": 13750
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002887892541455179,
+      "loss": 2.8824,
+      "step": 13760
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028877315412985524,
+      "loss": 2.8633,
+      "step": 13770
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002887570430111261,
+      "loss": 2.838,
+      "step": 13780
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028874092079061943,
+      "loss": 2.8596,
+      "step": 13790
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028872478746962534,
+      "loss": 2.8819,
+      "step": 13800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002887086430494345,
+      "loss": 2.8203,
+      "step": 13810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002886924875313387,
+      "loss": 2.8448,
+      "step": 13820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002886763209166305,
+      "loss": 2.8693,
+      "step": 13830
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028866014320660325,
+      "loss": 2.8599,
+      "step": 13840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028864395440255145,
+      "loss": 2.8393,
+      "step": 13850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028862775450577027,
+      "loss": 2.8356,
+      "step": 13860
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002886115435175559,
+      "loss": 2.8746,
+      "step": 13870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028859532143920533,
+      "loss": 2.8577,
+      "step": 13880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028857908827201645,
+      "loss": 2.8642,
+      "step": 13890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002885628440172881,
+      "loss": 2.8502,
+      "step": 13900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002885465886763199,
+      "loss": 2.9011,
+      "step": 13910
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002885303222504125,
+      "loss": 2.8102,
+      "step": 13920
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002885140447408673,
+      "loss": 2.8282,
+      "step": 13930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028849775614898666,
+      "loss": 2.8375,
+      "step": 13940
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002884814564760738,
+      "loss": 2.8522,
+      "step": 13950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028846514572343284,
+      "loss": 2.8408,
+      "step": 13960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002884488238923688,
+      "loss": 2.8531,
+      "step": 13970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028843249098418753,
+      "loss": 2.8616,
+      "step": 13980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002884161470001958,
+      "loss": 2.8589,
+      "step": 13990
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002883997919417014,
+      "loss": 2.8818,
+      "step": 14000
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 2.8972880840301514,
+      "eval_runtime": 1263.2852,
+      "eval_samples_per_second": 60.267,
+      "eval_steps_per_second": 7.534,
+      "step": 14000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002883834258100127,
+      "loss": 2.8734,
+      "step": 14010
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002883670486064392,
+      "loss": 2.8292,
+      "step": 14020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028835066033229133,
+      "loss": 2.8097,
+      "step": 14030
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028833426098888015,
+      "loss": 2.8275,
+      "step": 14040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028831785057751777,
+      "loss": 2.8601,
+      "step": 14050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028830142909951723,
+      "loss": 2.8597,
+      "step": 14060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002882849965561924,
+      "loss": 2.8811,
+      "step": 14070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028826855294885787,
+      "loss": 2.8467,
+      "step": 14080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002882520982788295,
+      "loss": 2.834,
+      "step": 14090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002882356325474237,
+      "loss": 2.869,
+      "step": 14100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028821915575595776,
+      "loss": 2.8522,
+      "step": 14110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002882026679057501,
+      "loss": 2.8671,
+      "step": 14120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002881861689981199,
+      "loss": 2.856,
+      "step": 14130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028816965903438724,
+      "loss": 2.8499,
+      "step": 14140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028815313801587296,
+      "loss": 2.8608,
+      "step": 14150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028813660594389893,
+      "loss": 2.8426,
+      "step": 14160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028812006281978784,
+      "loss": 2.8084,
+      "step": 14170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002881035086448633,
+      "loss": 2.8582,
+      "step": 14180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028808694342044984,
+      "loss": 2.8273,
+      "step": 14190
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002880703671478728,
+      "loss": 2.8396,
+      "step": 14200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028805377982845835,
+      "loss": 2.8398,
+      "step": 14210
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002880371814635337,
+      "loss": 2.817,
+      "step": 14220
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002880205720544269,
+      "loss": 2.8301,
+      "step": 14230
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002880039516024667,
+      "loss": 2.8222,
+      "step": 14240
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028798732010898296,
+      "loss": 2.8912,
+      "step": 14250
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002879706775753064,
+      "loss": 2.8675,
+      "step": 14260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002879540240027685,
+      "loss": 2.8088,
+      "step": 14270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028793735939270173,
+      "loss": 2.8414,
+      "step": 14280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002879206837464394,
+      "loss": 2.8669,
+      "step": 14290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002879039970653157,
+      "loss": 2.8797,
+      "step": 14300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002878872993506657,
+      "loss": 2.8476,
+      "step": 14310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028787059060382536,
+      "loss": 2.8455,
+      "step": 14320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028785387082613155,
+      "loss": 2.8212,
+      "step": 14330
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000287837140018922,
+      "loss": 2.8515,
+      "step": 14340
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002878203981835353,
+      "loss": 2.8689,
+      "step": 14350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002878036453213109,
+      "loss": 2.8221,
+      "step": 14360
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002877868814335893,
+      "loss": 2.8424,
+      "step": 14370
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028777010652171166,
+      "loss": 2.8326,
+      "step": 14380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028775332058702017,
+      "loss": 2.8744,
+      "step": 14390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002877365236308578,
+      "loss": 2.8526,
+      "step": 14400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002877197156545684,
+      "loss": 2.8356,
+      "step": 14410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028770289665949695,
+      "loss": 2.8536,
+      "step": 14420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002876860666469889,
+      "loss": 2.8638,
+      "step": 14430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028766922561839094,
+      "loss": 2.818,
+      "step": 14440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002876523735750505,
+      "loss": 2.8244,
+      "step": 14450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028763551051831583,
+      "loss": 2.7943,
+      "step": 14460
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028761863644953615,
+      "loss": 2.8016,
+      "step": 14470
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028760175137006145,
+      "loss": 2.8272,
+      "step": 14480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002875848552812428,
+      "loss": 2.8609,
+      "step": 14490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028756794818443203,
+      "loss": 2.8304,
+      "step": 14500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002875510300809818,
+      "loss": 2.8735,
+      "step": 14510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028753410097224574,
+      "loss": 2.8034,
+      "step": 14520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028751716085957825,
+      "loss": 2.8156,
+      "step": 14530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002875002097443348,
+      "loss": 2.8469,
+      "step": 14540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002874832476278716,
+      "loss": 2.84,
+      "step": 14550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002874662745115457,
+      "loss": 2.9009,
+      "step": 14560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002874509893030922,
+      "loss": 2.8205,
+      "step": 14570
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028743399529076926,
+      "loss": 2.8091,
+      "step": 14580
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002874169902825243,
+      "loss": 2.8475,
+      "step": 14590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028739997427971786,
+      "loss": 2.8635,
+      "step": 14600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028738294728371136,
+      "loss": 2.8233,
+      "step": 14610
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028736590929586705,
+      "loss": 2.8227,
+      "step": 14620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028734886031754825,
+      "loss": 2.8168,
+      "step": 14630
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028733180035011896,
+      "loss": 2.8192,
+      "step": 14640
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028731472939494407,
+      "loss": 2.8113,
+      "step": 14650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002872976474533895,
+      "loss": 2.8201,
+      "step": 14660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002872805545268219,
+      "loss": 2.8451,
+      "step": 14670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028726345061660886,
+      "loss": 2.8116,
+      "step": 14680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028724633572411887,
+      "loss": 2.8684,
+      "step": 14690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028722920985072123,
+      "loss": 2.8398,
+      "step": 14700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028721207299778616,
+      "loss": 2.8349,
+      "step": 14710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002871966404437733,
+      "loss": 2.8537,
+      "step": 14720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002871794827334953,
+      "loss": 2.8556,
+      "step": 14730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028716231404765847,
+      "loss": 2.8086,
+      "step": 14740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002871451343876364,
+      "loss": 2.8458,
+      "step": 14750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028712794375480383,
+      "loss": 2.8608,
+      "step": 14760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028711074215053594,
+      "loss": 2.8203,
+      "step": 14770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002870935295762091,
+      "loss": 2.8529,
+      "step": 14780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002870763060332005,
+      "loss": 2.8379,
+      "step": 14790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002870590715228881,
+      "loss": 2.8465,
+      "step": 14800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002870418260466508,
+      "loss": 2.8215,
+      "step": 14810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028702456960586853,
+      "loss": 2.8507,
+      "step": 14820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002870073022019218,
+      "loss": 2.8312,
+      "step": 14830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002869900238361922,
+      "loss": 2.8634,
+      "step": 14840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002869727345100622,
+      "loss": 2.8278,
+      "step": 14850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000286955434224915,
+      "loss": 2.831,
+      "step": 14860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002869381229821349,
+      "loss": 2.8304,
+      "step": 14870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028692080078310684,
+      "loss": 2.829,
+      "step": 14880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028690346762921684,
+      "loss": 2.8449,
+      "step": 14890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028688612352185163,
+      "loss": 2.809,
+      "step": 14900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028686876846239893,
+      "loss": 2.8028,
+      "step": 14910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002868514024522472,
+      "loss": 2.7941,
+      "step": 14920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002868340254927861,
+      "loss": 2.8791,
+      "step": 14930
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028681663758540565,
+      "loss": 2.8196,
+      "step": 14940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002867992387314973,
+      "loss": 2.8129,
+      "step": 14950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028678182893245296,
+      "loss": 2.8541,
+      "step": 14960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028676440818966556,
+      "loss": 2.8517,
+      "step": 14970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000286746976504529,
+      "loss": 2.8156,
+      "step": 14980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002867295338784379,
+      "loss": 2.8261,
+      "step": 14990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002867120803127878,
+      "loss": 2.8167,
+      "step": 15000
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 2.872957468032837,
+      "eval_runtime": 1262.9899,
+      "eval_samples_per_second": 60.282,
+      "eval_steps_per_second": 7.535,
+      "step": 15000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002866946158089752,
+      "loss": 2.8172,
+      "step": 15010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002866771403683974,
+      "loss": 2.7923,
+      "step": 15020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028665965399245257,
+      "loss": 2.8121,
+      "step": 15030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028664215668253976,
+      "loss": 2.8356,
+      "step": 15040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028662464844005894,
+      "loss": 2.7803,
+      "step": 15050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002866071292664109,
+      "loss": 2.8083,
+      "step": 15060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002865895991629974,
+      "loss": 2.8096,
+      "step": 15070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002865720581312209,
+      "loss": 2.8289,
+      "step": 15080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002865545061724849,
+      "loss": 2.8068,
+      "step": 15090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028653694328819364,
+      "loss": 2.8527,
+      "step": 15100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002865193694797524,
+      "loss": 2.8359,
+      "step": 15110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002865017847485671,
+      "loss": 2.8384,
+      "step": 15120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028648418909604486,
+      "loss": 2.8627,
+      "step": 15130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028646658252359336,
+      "loss": 2.7973,
+      "step": 15140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028644896503262127,
+      "loss": 2.8094,
+      "step": 15150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002864313366245382,
+      "loss": 2.8173,
+      "step": 15160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002864136973007546,
+      "loss": 2.8325,
+      "step": 15170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002863960470626817,
+      "loss": 2.8285,
+      "step": 15180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002863783859117317,
+      "loss": 2.8359,
+      "step": 15190
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002863607138493176,
+      "loss": 2.8359,
+      "step": 15200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002863430308768534,
+      "loss": 2.8361,
+      "step": 15210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028632533699575393,
+      "loss": 2.8295,
+      "step": 15220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028630763220743475,
+      "loss": 2.8242,
+      "step": 15230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002862899165133124,
+      "loss": 2.8268,
+      "step": 15240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028627218991480433,
+      "loss": 2.7912,
+      "step": 15250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002862544524133288,
+      "loss": 2.833,
+      "step": 15260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000286236704010305,
+      "loss": 2.8341,
+      "step": 15270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000286218944707153,
+      "loss": 2.8229,
+      "step": 15280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028620117450529355,
+      "loss": 2.8157,
+      "step": 15290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028618339340614855,
+      "loss": 2.8182,
+      "step": 15300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002861656014111406,
+      "loss": 2.8378,
+      "step": 15310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028614779852169326,
+      "loss": 2.8277,
+      "step": 15320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002861299847392308,
+      "loss": 2.8464,
+      "step": 15330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028611216006517866,
+      "loss": 2.8219,
+      "step": 15340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028609432450096275,
+      "loss": 2.811,
+      "step": 15350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028607647804801027,
+      "loss": 2.7976,
+      "step": 15360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000286058620707749,
+      "loss": 2.8224,
+      "step": 15370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002860407524816076,
+      "loss": 2.8299,
+      "step": 15380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028602287337101584,
+      "loss": 2.8172,
+      "step": 15390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028600498337740416,
+      "loss": 2.8165,
+      "step": 15400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002859870825022039,
+      "loss": 2.7969,
+      "step": 15410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028596917074684723,
+      "loss": 2.852,
+      "step": 15420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028595124811276737,
+      "loss": 2.8559,
+      "step": 15430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002859333146013982,
+      "loss": 2.822,
+      "step": 15440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028591537021417456,
+      "loss": 2.8138,
+      "step": 15450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028589741495253215,
+      "loss": 2.8003,
+      "step": 15460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028587944881790765,
+      "loss": 2.8698,
+      "step": 15470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002858614718117383,
+      "loss": 2.8223,
+      "step": 15480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002858434839354627,
+      "loss": 2.787,
+      "step": 15490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002858254851905198,
+      "loss": 2.804,
+      "step": 15500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00028580747557834977,
+      "loss": 2.8136,
+      "step": 15510
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002857894551003935,
+      "loss": 2.7749,
+      "step": 15520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00028577142375809284,
+      "loss": 2.8065,
+      "step": 15530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00028575338155289033,
+      "loss": 2.7651,
+      "step": 15540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002857353284862296,
+      "loss": 2.8154,
+      "step": 15550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002857172645595551,
+      "loss": 2.8187,
+      "step": 15560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00028569918977431203,
+      "loss": 2.82,
+      "step": 15570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002856811041319465,
+      "loss": 2.8147,
+      "step": 15580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00028566300763390557,
+      "loss": 2.8222,
+      "step": 15590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002856449002816371,
+      "loss": 2.8247,
+      "step": 15600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002856267820765899,
+      "loss": 2.8323,
+      "step": 15610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00028560865302021345,
+      "loss": 2.828,
+      "step": 15620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00028559051311395834,
+      "loss": 2.8422,
+      "step": 15630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000285572362359276,
+      "loss": 2.8088,
+      "step": 15640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002855542007576184,
+      "loss": 2.779,
+      "step": 15650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002855360283104388,
+      "loss": 2.7575,
+      "step": 15660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002855178450191912,
+      "loss": 2.8205,
+      "step": 15670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00028549965088533026,
+      "loss": 2.8411,
+      "step": 15680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00028548144591031185,
+      "loss": 2.8028,
+      "step": 15690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002854632300955924,
+      "loss": 2.8024,
+      "step": 15700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002854450034426293,
+      "loss": 2.8021,
+      "step": 15710
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000285426765952881,
+      "loss": 2.823,
+      "step": 15720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002854085176278065,
+      "loss": 2.7644,
+      "step": 15730
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00028539025846886594,
+      "loss": 2.8003,
+      "step": 15740
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002853719884775202,
+      "loss": 2.7852,
+      "step": 15750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000285353707655231,
+      "loss": 2.772,
+      "step": 15760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002853354160034609,
+      "loss": 2.8258,
+      "step": 15770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002853171135236735,
+      "loss": 2.7767,
+      "step": 15780
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00028529880021733317,
+      "loss": 2.7892,
+      "step": 15790
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000285280476085905,
+      "loss": 2.8171,
+      "step": 15800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00028526214113085526,
+      "loss": 2.8164,
+      "step": 15810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002852437953536508,
+      "loss": 2.8091,
+      "step": 15820
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00028522543875575943,
+      "loss": 2.8332,
+      "step": 15830
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002852070713386499,
+      "loss": 2.8134,
+      "step": 15840
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00028518869310379167,
+      "loss": 2.8324,
+      "step": 15850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002851703040526553,
+      "loss": 2.7974,
+      "step": 15860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000285151904186712,
+      "loss": 2.8245,
+      "step": 15870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00028513349350743383,
+      "loss": 2.8271,
+      "step": 15880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002851150720162939,
+      "loss": 2.7935,
+      "step": 15890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00028509663971476617,
+      "loss": 2.8118,
+      "step": 15900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00028507819660432524,
+      "loss": 2.8601,
+      "step": 15910
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002850597426864468,
+      "loss": 2.7988,
+      "step": 15920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002850412779626073,
+      "loss": 2.7915,
+      "step": 15930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00028502280243428407,
+      "loss": 2.7844,
+      "step": 15940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00028500431610295533,
+      "loss": 2.8153,
+      "step": 15950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00028498581897010015,
+      "loss": 2.8269,
+      "step": 15960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00028496731103719845,
+      "loss": 2.8431,
+      "step": 15970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000284948792305731,
+      "loss": 2.8081,
+      "step": 15980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00028493026277717954,
+      "loss": 2.8047,
+      "step": 15990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00028491172245302654,
+      "loss": 2.7941,
+      "step": 16000
+    },
+    {
+      "epoch": 1.44,
+      "eval_loss": 2.8514840602874756,
+      "eval_runtime": 1268.5535,
+      "eval_samples_per_second": 60.017,
+      "eval_steps_per_second": 7.502,
+      "step": 16000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002848931713347554,
+      "loss": 2.839,
+      "step": 16010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00028487460942385037,
+      "loss": 2.7998,
+      "step": 16020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00028485603672179655,
+      "loss": 2.774,
+      "step": 16030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00028483745323007996,
+      "loss": 2.7941,
+      "step": 16040
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002848188589501874,
+      "loss": 2.8237,
+      "step": 16050
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002848002538836066,
+      "loss": 2.8063,
+      "step": 16060
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00028478163803182606,
+      "loss": 2.7756,
+      "step": 16070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002847630113963353,
+      "loss": 2.78,
+      "step": 16080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002847443739786246,
+      "loss": 2.8029,
+      "step": 16090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00028472572578018513,
+      "loss": 2.8227,
+      "step": 16100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002847070668025088,
+      "loss": 2.7702,
+      "step": 16110
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00028468839704708863,
+      "loss": 2.8218,
+      "step": 16120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00028466971651541827,
+      "loss": 2.7949,
+      "step": 16130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00028465102520899236,
+      "loss": 2.7995,
+      "step": 16140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002846323231293064,
+      "loss": 2.7902,
+      "step": 16150
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002846136102778567,
+      "loss": 2.8055,
+      "step": 16160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002845948866561404,
+      "loss": 2.823,
+      "step": 16170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002845761522656556,
+      "loss": 2.8113,
+      "step": 16180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00028455740710790116,
+      "loss": 2.7878,
+      "step": 16190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00028453865118437696,
+      "loss": 2.7761,
+      "step": 16200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00028451988449658356,
+      "loss": 2.7934,
+      "step": 16210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002845011070460225,
+      "loss": 2.7917,
+      "step": 16220
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002844823188341961,
+      "loss": 2.7993,
+      "step": 16230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00028446351986260765,
+      "loss": 2.7683,
+      "step": 16240
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00028444471013276115,
+      "loss": 2.7853,
+      "step": 16250
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002844258896461616,
+      "loss": 2.7708,
+      "step": 16260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002844070584043147,
+      "loss": 2.8363,
+      "step": 16270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002843882164087273,
+      "loss": 2.7858,
+      "step": 16280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00028436936366090675,
+      "loss": 2.8234,
+      "step": 16290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00028435050016236156,
+      "loss": 2.8003,
+      "step": 16300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00028433162591460094,
+      "loss": 2.7967,
+      "step": 16310
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002843127409191349,
+      "loss": 2.8137,
+      "step": 16320
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002842938451774745,
+      "loss": 2.7782,
+      "step": 16330
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00028427493869113155,
+      "loss": 2.7808,
+      "step": 16340
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00028425602146161876,
+      "loss": 2.7854,
+      "step": 16350
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002842370934904496,
+      "loss": 2.773,
+      "step": 16360
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00028421815477913854,
+      "loss": 2.8009,
+      "step": 16370
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002841992053292008,
+      "loss": 2.8074,
+      "step": 16380
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00028418024514215254,
+      "loss": 2.8121,
+      "step": 16390
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002841612742195107,
+      "loss": 2.8207,
+      "step": 16400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002841422925627932,
+      "loss": 2.7889,
+      "step": 16410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00028412330017351864,
+      "loss": 2.8007,
+      "step": 16420
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002841042970532066,
+      "loss": 2.7748,
+      "step": 16430
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002840852832033775,
+      "loss": 2.7747,
+      "step": 16440
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002840662586255527,
+      "loss": 2.7457,
+      "step": 16450
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00028404722332125423,
+      "loss": 2.8344,
+      "step": 16460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00028402817729200505,
+      "loss": 2.8059,
+      "step": 16470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002840091205393291,
+      "loss": 2.7955,
+      "step": 16480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002839900530647511,
+      "loss": 2.7599,
+      "step": 16490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002839709748697965,
+      "loss": 2.8051,
+      "step": 16500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00028395188595599184,
+      "loss": 2.7782,
+      "step": 16510
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00028393278632486436,
+      "loss": 2.7968,
+      "step": 16520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002839136759779422,
+      "loss": 2.8001,
+      "step": 16530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00028389455491675434,
+      "loss": 2.7848,
+      "step": 16540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002838754231428306,
+      "loss": 2.8013,
+      "step": 16550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00028385628065770175,
+      "loss": 2.7676,
+      "step": 16560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002838371274628994,
+      "loss": 2.7984,
+      "step": 16570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00028381796355995585,
+      "loss": 2.801,
+      "step": 16580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00028379878895040445,
+      "loss": 2.7779,
+      "step": 16590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00028377960363577923,
+      "loss": 2.7952,
+      "step": 16600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00028376040761761537,
+      "loss": 2.8156,
+      "step": 16610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00028374120089744864,
+      "loss": 2.8107,
+      "step": 16620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002837219834768157,
+      "loss": 2.8015,
+      "step": 16630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002837027553572541,
+      "loss": 2.7967,
+      "step": 16640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002836835165403024,
+      "loss": 2.7936,
+      "step": 16650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002836642670274997,
+      "loss": 2.7867,
+      "step": 16660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002836450068203863,
+      "loss": 2.7614,
+      "step": 16670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000283625735920503,
+      "loss": 2.7914,
+      "step": 16680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00028360645432939176,
+      "loss": 2.7763,
+      "step": 16690
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00028358716204859527,
+      "loss": 2.773,
+      "step": 16700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000283567859079657,
+      "loss": 2.7908,
+      "step": 16710
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00028354854542412145,
+      "loss": 2.7907,
+      "step": 16720
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002835292210835338,
+      "loss": 2.7697,
+      "step": 16730
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00028350988605944027,
+      "loss": 2.7846,
+      "step": 16740
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00028349054035338775,
+      "loss": 2.7728,
+      "step": 16750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002834711839669241,
+      "loss": 2.7676,
+      "step": 16760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00028345181690159796,
+      "loss": 2.7971,
+      "step": 16770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002834324391589589,
+      "loss": 2.785,
+      "step": 16780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002834130507405573,
+      "loss": 2.7731,
+      "step": 16790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002833936516479445,
+      "loss": 2.7943,
+      "step": 16800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002833742418826724,
+      "loss": 2.7685,
+      "step": 16810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00028335482144629405,
+      "loss": 2.7462,
+      "step": 16820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002833353903403633,
+      "loss": 2.8114,
+      "step": 16830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00028331594856643473,
+      "loss": 2.8018,
+      "step": 16840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00028329649612606394,
+      "loss": 2.8054,
+      "step": 16850
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00028327703302080724,
+      "loss": 2.7772,
+      "step": 16860
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002832575592522218,
+      "loss": 2.7714,
+      "step": 16870
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00028323807482186577,
+      "loss": 2.794,
+      "step": 16880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002832185797312981,
+      "loss": 2.771,
+      "step": 16890
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002831990739820785,
+      "loss": 2.7964,
+      "step": 16900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00028317955757576756,
+      "loss": 2.8071,
+      "step": 16910
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00028316003051392685,
+      "loss": 2.7844,
+      "step": 16920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002831404927981187,
+      "loss": 2.7721,
+      "step": 16930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00028312094442990623,
+      "loss": 2.818,
+      "step": 16940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002831013854108535,
+      "loss": 2.773,
+      "step": 16950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002830818157425255,
+      "loss": 2.8021,
+      "step": 16960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002830622354264879,
+      "loss": 2.7836,
+      "step": 16970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00028304264446430734,
+      "loss": 2.788,
+      "step": 16980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002830230428575512,
+      "loss": 2.7696,
+      "step": 16990
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002830053923116547,
+      "loss": 2.7878,
+      "step": 17000
+    },
+    {
+      "epoch": 1.53,
+      "eval_loss": 2.831134080886841,
+      "eval_runtime": 1263.6147,
+      "eval_samples_per_second": 60.252,
+      "eval_steps_per_second": 7.532,
+      "step": 17000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002829857704845263,
+      "loss": 2.7969,
+      "step": 17010
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00028296613801737287,
+      "loss": 2.7805,
+      "step": 17020
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000282946494911765,
+      "loss": 2.7996,
+      "step": 17030
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00028292684116927445,
+      "loss": 2.7949,
+      "step": 17040
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00028290717679147354,
+      "loss": 2.7809,
+      "step": 17050
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002828875017799358,
+      "loss": 2.8041,
+      "step": 17060
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00028286781613623516,
+      "loss": 2.7772,
+      "step": 17070
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002828481198619468,
+      "loss": 2.832,
+      "step": 17080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002828284129586466,
+      "loss": 2.7831,
+      "step": 17090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002828086954279112,
+      "loss": 2.7743,
+      "step": 17100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00028278896727131823,
+      "loss": 2.7768,
+      "step": 17110
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002827692284904461,
+      "loss": 2.766,
+      "step": 17120
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002827494790868741,
+      "loss": 2.7618,
+      "step": 17130
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002827297190621823,
+      "loss": 2.7931,
+      "step": 17140
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002827099484179518,
+      "loss": 2.7817,
+      "step": 17150
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00028269016715576427,
+      "loss": 2.7673,
+      "step": 17160
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002826703752772025,
+      "loss": 2.7645,
+      "step": 17170
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00028265057278385,
+      "loss": 2.8033,
+      "step": 17180
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00028263075967729104,
+      "loss": 2.7809,
+      "step": 17190
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002826109359591109,
+      "loss": 2.7653,
+      "step": 17200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002825911016308957,
+      "loss": 2.7781,
+      "step": 17210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00028257125669423235,
+      "loss": 2.7948,
+      "step": 17220
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002825514011507086,
+      "loss": 2.7718,
+      "step": 17230
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00028253153500191303,
+      "loss": 2.7748,
+      "step": 17240
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00028251165824943515,
+      "loss": 2.7825,
+      "step": 17250
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00028249177089486525,
+      "loss": 2.7963,
+      "step": 17260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002824718729397945,
+      "loss": 2.7781,
+      "step": 17270
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028245196438581494,
+      "loss": 2.7542,
+      "step": 17280
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002824320452345194,
+      "loss": 2.7798,
+      "step": 17290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028241211548750157,
+      "loss": 2.8216,
+      "step": 17300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028239217514635604,
+      "loss": 2.7877,
+      "step": 17310
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002823722242126782,
+      "loss": 2.7644,
+      "step": 17320
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028235226268806426,
+      "loss": 2.7605,
+      "step": 17330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028233229057411137,
+      "loss": 2.7921,
+      "step": 17340
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028231230787241745,
+      "loss": 2.815,
+      "step": 17350
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002822923145845813,
+      "loss": 2.7722,
+      "step": 17360
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002822723107122026,
+      "loss": 2.7834,
+      "step": 17370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002822522962568817,
+      "loss": 2.7587,
+      "step": 17380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002822322712202201,
+      "loss": 2.7863,
+      "step": 17390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002822122356038198,
+      "loss": 2.7879,
+      "step": 17400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000282192189409284,
+      "loss": 2.7886,
+      "step": 17410
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002821721326382165,
+      "loss": 2.7432,
+      "step": 17420
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00028215206529222197,
+      "loss": 2.7841,
+      "step": 17430
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00028213198737290606,
+      "loss": 2.7909,
+      "step": 17440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002821118988818751,
+      "loss": 2.7897,
+      "step": 17450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002820917998207364,
+      "loss": 2.7748,
+      "step": 17460
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000282071690191098,
+      "loss": 2.795,
+      "step": 17470
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002820515699945689,
+      "loss": 2.8268,
+      "step": 17480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002820314392327589,
+      "loss": 2.7267,
+      "step": 17490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00028201129790727863,
+      "loss": 2.7389,
+      "step": 17500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002819911460197395,
+      "loss": 2.7982,
+      "step": 17510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002819709835717539,
+      "loss": 2.7856,
+      "step": 17520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00028195081056493505,
+      "loss": 2.7392,
+      "step": 17530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00028193062700089685,
+      "loss": 2.7506,
+      "step": 17540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002819104328812543,
+      "loss": 2.7787,
+      "step": 17550
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00028189022820762296,
+      "loss": 2.7892,
+      "step": 17560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002818700129816195,
+      "loss": 2.7726,
+      "step": 17570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002818497872048612,
+      "loss": 2.7934,
+      "step": 17580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002818295508789664,
+      "loss": 2.791,
+      "step": 17590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002818093040055542,
+      "loss": 2.7656,
+      "step": 17600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002817890465862444,
+      "loss": 2.7514,
+      "step": 17610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002817687786226579,
+      "loss": 2.7655,
+      "step": 17620
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00028174850011641623,
+      "loss": 2.7691,
+      "step": 17630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002817282110691419,
+      "loss": 2.7638,
+      "step": 17640
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00028170791148245816,
+      "loss": 2.7956,
+      "step": 17650
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002816876013579892,
+      "loss": 2.8035,
+      "step": 17660
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00028166728069736,
+      "loss": 2.777,
+      "step": 17670
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00028164694950219636,
+      "loss": 2.7354,
+      "step": 17680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000281626607774125,
+      "loss": 2.7733,
+      "step": 17690
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00028160625551477346,
+      "loss": 2.7782,
+      "step": 17700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00028158589272577,
+      "loss": 2.741,
+      "step": 17710
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00028156551940874383,
+      "loss": 2.7702,
+      "step": 17720
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002815451355653251,
+      "loss": 2.7685,
+      "step": 17730
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002815247411971446,
+      "loss": 2.8023,
+      "step": 17740
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002815043363058341,
+      "loss": 2.7486,
+      "step": 17750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002814839208930262,
+      "loss": 2.6952,
+      "step": 17760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002814634949603542,
+      "loss": 2.7706,
+      "step": 17770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002814430585094525,
+      "loss": 2.725,
+      "step": 17780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00028142261154195605,
+      "loss": 2.7801,
+      "step": 17790
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002814021540595009,
+      "loss": 2.7242,
+      "step": 17800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002813816860637237,
+      "loss": 2.8025,
+      "step": 17810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00028136120755626226,
+      "loss": 2.7772,
+      "step": 17820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002813407185387549,
+      "loss": 2.7322,
+      "step": 17830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00028132021901284095,
+      "loss": 2.75,
+      "step": 17840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00028129970898016054,
+      "loss": 2.7896,
+      "step": 17850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002812791884423547,
+      "loss": 2.8227,
+      "step": 17860
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00028125865740106517,
+      "loss": 2.7679,
+      "step": 17870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002812381158579347,
+      "loss": 2.7763,
+      "step": 17880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002812175638146067,
+      "loss": 2.7294,
+      "step": 17890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00028119700127272564,
+      "loss": 2.7738,
+      "step": 17900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002811764282339366,
+      "loss": 2.7751,
+      "step": 17910
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002811558446998856,
+      "loss": 2.8096,
+      "step": 17920
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002811352506722196,
+      "loss": 2.7508,
+      "step": 17930
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002811146461525862,
+      "loss": 2.7451,
+      "step": 17940
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000281094031142634,
+      "loss": 2.7668,
+      "step": 17950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00028107340564401237,
+      "loss": 2.7547,
+      "step": 17960
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002810527696583715,
+      "loss": 2.7785,
+      "step": 17970
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002810321231873625,
+      "loss": 2.7814,
+      "step": 17980
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002810114662326372,
+      "loss": 2.7674,
+      "step": 17990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00028099079879584843,
+      "loss": 2.7894,
+      "step": 18000
+    },
+    {
+      "epoch": 1.62,
+      "eval_loss": 2.811331272125244,
+      "eval_runtime": 1263.416,
+      "eval_samples_per_second": 60.261,
+      "eval_steps_per_second": 7.533,
+      "step": 18000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00028097012087864966,
+      "loss": 2.7772,
+      "step": 18010
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002809494324826954,
+      "loss": 2.7323,
+      "step": 18020
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002809287336096408,
+      "loss": 2.7622,
+      "step": 18030
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002809080242611421,
+      "loss": 2.7919,
+      "step": 18040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002808873044388561,
+      "loss": 2.7508,
+      "step": 18050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002808665741444406,
+      "loss": 2.7731,
+      "step": 18060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002808458333795542,
+      "loss": 2.7796,
+      "step": 18070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00028082508214585637,
+      "loss": 2.7653,
+      "step": 18080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002808043204450074,
+      "loss": 2.7775,
+      "step": 18090
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002807835482786683,
+      "loss": 2.7689,
+      "step": 18100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002807627656485012,
+      "loss": 2.7282,
+      "step": 18110
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002807419725561688,
+      "loss": 2.7501,
+      "step": 18120
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002807211690033347,
+      "loss": 2.7759,
+      "step": 18130
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002807003549916634,
+      "loss": 2.7365,
+      "step": 18140
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00028067953052282016,
+      "loss": 2.7815,
+      "step": 18150
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00028065869559847117,
+      "loss": 2.7814,
+      "step": 18160
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002806378502202834,
+      "loss": 2.7076,
+      "step": 18170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002806169943899247,
+      "loss": 2.7783,
+      "step": 18180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002805961281090637,
+      "loss": 2.7978,
+      "step": 18190
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002805752513793698,
+      "loss": 2.7997,
+      "step": 18200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00028055436420251333,
+      "loss": 2.7909,
+      "step": 18210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00028053346658016564,
+      "loss": 2.7604,
+      "step": 18220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00028051255851399846,
+      "loss": 2.754,
+      "step": 18230
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002804916400056848,
+      "loss": 2.7349,
+      "step": 18240
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00028047071105689826,
+      "loss": 2.7578,
+      "step": 18250
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00028044977166931335,
+      "loss": 2.7981,
+      "step": 18260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00028042882184460533,
+      "loss": 2.7726,
+      "step": 18270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002804078615844505,
+      "loss": 2.7484,
+      "step": 18280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002803868908905258,
+      "loss": 2.7701,
+      "step": 18290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002803659097645091,
+      "loss": 2.7735,
+      "step": 18300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000280344918208079,
+      "loss": 2.7464,
+      "step": 18310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00028032391622291505,
+      "loss": 2.7827,
+      "step": 18320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002803029038106976,
+      "loss": 2.7866,
+      "step": 18330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00028028188097310784,
+      "loss": 2.729,
+      "step": 18340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002802608477118277,
+      "loss": 2.7397,
+      "step": 18350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00028023980402854014,
+      "loss": 2.7617,
+      "step": 18360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002802187499249288,
+      "loss": 2.7758,
+      "step": 18370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002801976854026781,
+      "loss": 2.7497,
+      "step": 18380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002801766104634735,
+      "loss": 2.7171,
+      "step": 18390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00028015552510900123,
+      "loss": 2.748,
+      "step": 18400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00028013442934094813,
+      "loss": 2.7523,
+      "step": 18410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00028011332316100216,
+      "loss": 2.7973,
+      "step": 18420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00028009220657085196,
+      "loss": 2.7552,
+      "step": 18430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000280071079572187,
+      "loss": 2.8005,
+      "step": 18440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002800499421666978,
+      "loss": 2.7603,
+      "step": 18450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00028002879435607526,
+      "loss": 2.7503,
+      "step": 18460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00028000763614201166,
+      "loss": 2.7959,
+      "step": 18470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002799864675261997,
+      "loss": 2.7276,
+      "step": 18480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002799652885103331,
+      "loss": 2.7529,
+      "step": 18490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027994409909610634,
+      "loss": 2.7713,
+      "step": 18500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027992289928521476,
+      "loss": 2.7678,
+      "step": 18510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002799016890793545,
+      "loss": 2.7314,
+      "step": 18520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002798804684802227,
+      "loss": 2.7776,
+      "step": 18530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027985923748951696,
+      "loss": 2.7932,
+      "step": 18540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002798379961089362,
+      "loss": 2.7762,
+      "step": 18550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002798167443401797,
+      "loss": 2.7393,
+      "step": 18560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027979548218494795,
+      "loss": 2.7434,
+      "step": 18570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000279774209644942,
+      "loss": 2.7569,
+      "step": 18580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002797529267218639,
+      "loss": 2.7583,
+      "step": 18590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002797316334174164,
+      "loss": 2.7476,
+      "step": 18600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002797103297333033,
+      "loss": 2.7871,
+      "step": 18610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002796890156712289,
+      "loss": 2.755,
+      "step": 18620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027966769123289866,
+      "loss": 2.7261,
+      "step": 18630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002796463564200186,
+      "loss": 2.7433,
+      "step": 18640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027962501123429575,
+      "loss": 2.7307,
+      "step": 18650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002796036556774379,
+      "loss": 2.7416,
+      "step": 18660
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002795822897511537,
+      "loss": 2.7374,
+      "step": 18670
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002795609134571526,
+      "loss": 2.7156,
+      "step": 18680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027953952679714495,
+      "loss": 2.7674,
+      "step": 18690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002795181297728417,
+      "loss": 2.7347,
+      "step": 18700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000279496722385955,
+      "loss": 2.7402,
+      "step": 18710
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002794753046381975,
+      "loss": 2.7561,
+      "step": 18720
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027945387653128287,
+      "loss": 2.7698,
+      "step": 18730
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002794324380669255,
+      "loss": 2.7907,
+      "step": 18740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002794109892468407,
+      "loss": 2.7777,
+      "step": 18750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002793895300727445,
+      "loss": 2.7502,
+      "step": 18760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027936806054635385,
+      "loss": 2.7644,
+      "step": 18770
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002793465806693866,
+      "loss": 2.7495,
+      "step": 18780
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027932509044356114,
+      "loss": 2.7571,
+      "step": 18790
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027930358987059703,
+      "loss": 2.7378,
+      "step": 18800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027928207895221445,
+      "loss": 2.7546,
+      "step": 18810
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002792605576901344,
+      "loss": 2.7447,
+      "step": 18820
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002792390260860789,
+      "loss": 2.7413,
+      "step": 18830
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00027921748414177057,
+      "loss": 2.7417,
+      "step": 18840
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000279195931858933,
+      "loss": 2.746,
+      "step": 18850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002791743692392905,
+      "loss": 2.7415,
+      "step": 18860
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00027915279628456835,
+      "loss": 2.7538,
+      "step": 18870
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002791312129964925,
+      "loss": 2.7677,
+      "step": 18880
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00027910961937678986,
+      "loss": 2.7829,
+      "step": 18890
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002790880154271881,
+      "loss": 2.748,
+      "step": 18900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002790664011494157,
+      "loss": 2.7782,
+      "step": 18910
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000279044776545202,
+      "loss": 2.7211,
+      "step": 18920
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002790231416162771,
+      "loss": 2.7796,
+      "step": 18930
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00027900149636437215,
+      "loss": 2.7279,
+      "step": 18940
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002789798407912188,
+      "loss": 2.7511,
+      "step": 18950
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00027895817489854974,
+      "loss": 2.7582,
+      "step": 18960
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00027893649868809844,
+      "loss": 2.754,
+      "step": 18970
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00027891481216159916,
+      "loss": 2.7683,
+      "step": 18980
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00027889311532078705,
+      "loss": 2.7168,
+      "step": 18990
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00027887140816739797,
+      "loss": 2.7158,
+      "step": 19000
+    },
+    {
+      "epoch": 1.71,
+      "eval_loss": 2.7934746742248535,
+      "eval_runtime": 1263.8082,
+      "eval_samples_per_second": 60.243,
+      "eval_steps_per_second": 7.53,
+      "step": 19000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00027884969070316877,
+      "loss": 2.756,
+      "step": 19010
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00027882796292983704,
+      "loss": 2.7506,
+      "step": 19020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00027880622484914114,
+      "loss": 2.7568,
+      "step": 19030
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002787844764628203,
+      "loss": 2.7515,
+      "step": 19040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00027876271777261456,
+      "loss": 2.7746,
+      "step": 19050
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002787409487802649,
+      "loss": 2.7478,
+      "step": 19060
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000278719169487513,
+      "loss": 2.758,
+      "step": 19070
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00027869737989610134,
+      "loss": 2.7395,
+      "step": 19080
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002786755800077733,
+      "loss": 2.7505,
+      "step": 19090
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002786537698242731,
+      "loss": 2.6873,
+      "step": 19100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00027863194934734574,
+      "loss": 2.7713,
+      "step": 19110
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000278610118578737,
+      "loss": 2.7564,
+      "step": 19120
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00027858827752019354,
+      "loss": 2.7131,
+      "step": 19130
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00027856642617346284,
+      "loss": 2.7643,
+      "step": 19140
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002785445645402933,
+      "loss": 2.766,
+      "step": 19150
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000278524880276981,
+      "loss": 2.7423,
+      "step": 19160
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00027850299910439695,
+      "loss": 2.7561,
+      "step": 19170
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002784811076504487,
+      "loss": 2.769,
+      "step": 19180
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00027845920591688776,
+      "loss": 2.7806,
+      "step": 19190
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00027843729390546646,
+      "loss": 2.7789,
+      "step": 19200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00027841537161793793,
+      "loss": 2.763,
+      "step": 19210
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00027839343905605615,
+      "loss": 2.7475,
+      "step": 19220
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00027837149622157595,
+      "loss": 2.7232,
+      "step": 19230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002783495431162529,
+      "loss": 2.7625,
+      "step": 19240
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002783275797418435,
+      "loss": 2.7186,
+      "step": 19250
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00027830560610010496,
+      "loss": 2.7641,
+      "step": 19260
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00027828362219279543,
+      "loss": 2.7396,
+      "step": 19270
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002782616280216737,
+      "loss": 2.7302,
+      "step": 19280
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002782396235884996,
+      "loss": 2.7295,
+      "step": 19290
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002782176088950336,
+      "loss": 2.7549,
+      "step": 19300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002781955839430371,
+      "loss": 2.7284,
+      "step": 19310
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00027817354873427237,
+      "loss": 2.7926,
+      "step": 19320
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00027815150327050234,
+      "loss": 2.7759,
+      "step": 19330
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002781294475534908,
+      "loss": 2.7291,
+      "step": 19340
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002781073815850024,
+      "loss": 2.7289,
+      "step": 19350
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002780853053668028,
+      "loss": 2.72,
+      "step": 19360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00027806321890065807,
+      "loss": 2.7248,
+      "step": 19370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002780411221883354,
+      "loss": 2.7517,
+      "step": 19380
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002780190152316027,
+      "loss": 2.6992,
+      "step": 19390
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00027799689803222874,
+      "loss": 2.7111,
+      "step": 19400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00027797477059198313,
+      "loss": 2.7219,
+      "step": 19410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002779526329126362,
+      "loss": 2.7601,
+      "step": 19420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00027793048499595916,
+      "loss": 2.7242,
+      "step": 19430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00027790832684372407,
+      "loss": 2.73,
+      "step": 19440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00027788615845770375,
+      "loss": 2.7367,
+      "step": 19450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00027786397983967185,
+      "loss": 2.7546,
+      "step": 19460
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002778417909914029,
+      "loss": 2.7341,
+      "step": 19470
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002778195919146722,
+      "loss": 2.7497,
+      "step": 19480
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002777973826112558,
+      "loss": 2.7105,
+      "step": 19490
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002777751630829307,
+      "loss": 2.7486,
+      "step": 19500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002777529333314747,
+      "loss": 2.7033,
+      "step": 19510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00027773069335866623,
+      "loss": 2.7358,
+      "step": 19520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002777084431662848,
+      "loss": 2.7388,
+      "step": 19530
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00027768618275611067,
+      "loss": 2.7162,
+      "step": 19540
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002776639121299247,
+      "loss": 2.7521,
+      "step": 19550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00027764163128950886,
+      "loss": 2.7108,
+      "step": 19560
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00027761934023664577,
+      "loss": 2.7396,
+      "step": 19570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002775970389731189,
+      "loss": 2.7477,
+      "step": 19580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002775747275007126,
+      "loss": 2.7326,
+      "step": 19590
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002775524058212119,
+      "loss": 2.7043,
+      "step": 19600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002775300739364028,
+      "loss": 2.7459,
+      "step": 19610
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.000277507731848072,
+      "loss": 2.7518,
+      "step": 19620
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002774853795580071,
+      "loss": 2.7684,
+      "step": 19630
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002774630170679965,
+      "loss": 2.7568,
+      "step": 19640
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002774406443798293,
+      "loss": 2.7393,
+      "step": 19650
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002774182614952956,
+      "loss": 2.7363,
+      "step": 19660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002773958684161862,
+      "loss": 2.713,
+      "step": 19670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002773734651442927,
+      "loss": 2.7079,
+      "step": 19680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002773510516814076,
+      "loss": 2.7513,
+      "step": 19690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00027732862802932415,
+      "loss": 2.7311,
+      "step": 19700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002773061941898365,
+      "loss": 2.7397,
+      "step": 19710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002772837501647395,
+      "loss": 2.7411,
+      "step": 19720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00027726129595582885,
+      "loss": 2.7292,
+      "step": 19730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00027723883156490116,
+      "loss": 2.7469,
+      "step": 19740
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002772163569937537,
+      "loss": 2.7298,
+      "step": 19750
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00027719387224418466,
+      "loss": 2.769,
+      "step": 19760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00027717137731799303,
+      "loss": 2.7438,
+      "step": 19770
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002771488722169785,
+      "loss": 2.7379,
+      "step": 19780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002771263569429419,
+      "loss": 2.7574,
+      "step": 19790
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002771038314976844,
+      "loss": 2.7202,
+      "step": 19800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002770812958830084,
+      "loss": 2.7544,
+      "step": 19810
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002770587501007169,
+      "loss": 2.726,
+      "step": 19820
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002770361941526137,
+      "loss": 2.7102,
+      "step": 19830
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002770136280405036,
+      "loss": 2.7194,
+      "step": 19840
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000276991051766192,
+      "loss": 2.7154,
+      "step": 19850
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002769684653314851,
+      "loss": 2.7355,
+      "step": 19860
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00027694586873819023,
+      "loss": 2.7237,
+      "step": 19870
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00027692326198811517,
+      "loss": 2.7766,
+      "step": 19880
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00027690064508306873,
+      "loss": 2.7352,
+      "step": 19890
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002768780180248604,
+      "loss": 2.751,
+      "step": 19900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002768553808153006,
+      "loss": 2.7255,
+      "step": 19910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00027683273345620036,
+      "loss": 2.7258,
+      "step": 19920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002768100759493719,
+      "loss": 2.7494,
+      "step": 19930
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00027678740829662786,
+      "loss": 2.7546,
+      "step": 19940
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002767647304997819,
+      "loss": 2.7291,
+      "step": 19950
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00027674204256064843,
+      "loss": 2.6885,
+      "step": 19960
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002767193444810427,
+      "loss": 2.7994,
+      "step": 19970
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002766966362627807,
+      "loss": 2.7725,
+      "step": 19980
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002766739179076793,
+      "loss": 2.7636,
+      "step": 19990
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002766511894175563,
+      "loss": 2.7409,
+      "step": 20000
+    },
+    {
+      "epoch": 1.8,
+      "eval_loss": 2.7765214443206787,
+      "eval_runtime": 1263.162,
+      "eval_samples_per_second": 60.273,
+      "eval_steps_per_second": 7.534,
+      "step": 20000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00027662845079422997,
+      "loss": 2.7273,
+      "step": 20010
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002766057020395197,
+      "loss": 2.754,
+      "step": 20020
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002765829431552456,
+      "loss": 2.732,
+      "step": 20030
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00027656017414322867,
+      "loss": 2.7365,
+      "step": 20040
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002765373950052904,
+      "loss": 2.7809,
+      "step": 20050
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002765146057432536,
+      "loss": 2.7123,
+      "step": 20060
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002764918063589413,
+      "loss": 2.7172,
+      "step": 20070
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00027646899685417785,
+      "loss": 2.7514,
+      "step": 20080
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00027644617723078817,
+      "loss": 2.7528,
+      "step": 20090
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00027642334749059807,
+      "loss": 2.6872,
+      "step": 20100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002764005076354341,
+      "loss": 2.7482,
+      "step": 20110
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00027637765766712363,
+      "loss": 2.7519,
+      "step": 20120
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00027635479758749483,
+      "loss": 2.7476,
+      "step": 20130
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00027633192739837673,
+      "loss": 2.7617,
+      "step": 20140
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002763090471015992,
+      "loss": 2.738,
+      "step": 20150
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002762861566989928,
+      "loss": 2.7408,
+      "step": 20160
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000276263256192389,
+      "loss": 2.7842,
+      "step": 20170
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00027624034558361996,
+      "loss": 2.7567,
+      "step": 20180
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00027621742487451886,
+      "loss": 2.716,
+      "step": 20190
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002761944940669195,
+      "loss": 2.7112,
+      "step": 20200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002761715531626565,
+      "loss": 2.7244,
+      "step": 20210
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00027614860216356536,
+      "loss": 2.6776,
+      "step": 20220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002761256410714824,
+      "loss": 2.7224,
+      "step": 20230
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00027610266988824467,
+      "loss": 2.7531,
+      "step": 20240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00027607968861569006,
+      "loss": 2.7224,
+      "step": 20250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002760566972556573,
+      "loss": 2.7044,
+      "step": 20260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002760336958099859,
+      "loss": 2.7505,
+      "step": 20270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002760106842805161,
+      "loss": 2.736,
+      "step": 20280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00027598766266908914,
+      "loss": 2.7717,
+      "step": 20290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00027596463097754687,
+      "loss": 2.7652,
+      "step": 20300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00027594158920773206,
+      "loss": 2.7459,
+      "step": 20310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002759185373614882,
+      "loss": 2.737,
+      "step": 20320
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002758954754406598,
+      "loss": 2.733,
+      "step": 20330
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002758724034470918,
+      "loss": 2.7306,
+      "step": 20340
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002758493213826303,
+      "loss": 2.7219,
+      "step": 20350
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00027582622924912203,
+      "loss": 2.7295,
+      "step": 20360
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00027580312704841457,
+      "loss": 2.7542,
+      "step": 20370
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00027578001478235627,
+      "loss": 2.7308,
+      "step": 20380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00027575689245279634,
+      "loss": 2.725,
+      "step": 20390
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00027573376006158475,
+      "loss": 2.7234,
+      "step": 20400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00027571061761057236,
+      "loss": 2.6868,
+      "step": 20410
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00027568746510161065,
+      "loss": 2.7467,
+      "step": 20420
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00027566430253655213,
+      "loss": 2.7495,
+      "step": 20430
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00027564112991724994,
+      "loss": 2.7332,
+      "step": 20440
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002756179472455581,
+      "loss": 2.7263,
+      "step": 20450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002755947545233315,
+      "loss": 2.7243,
+      "step": 20460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002755715517524257,
+      "loss": 2.7154,
+      "step": 20470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002755483389346971,
+      "loss": 2.7447,
+      "step": 20480
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00027552511607200296,
+      "loss": 2.7077,
+      "step": 20490
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00027550188316620135,
+      "loss": 2.7525,
+      "step": 20500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00027547864021915103,
+      "loss": 2.6968,
+      "step": 20510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002754553872327117,
+      "loss": 2.739,
+      "step": 20520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00027543212420874376,
+      "loss": 2.7044,
+      "step": 20530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002754088511491085,
+      "loss": 2.7162,
+      "step": 20540
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002753855680556679,
+      "loss": 2.7355,
+      "step": 20550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00027536227493028483,
+      "loss": 2.7267,
+      "step": 20560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000275338971774823,
+      "loss": 2.7338,
+      "step": 20570
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002753156585911469,
+      "loss": 2.6969,
+      "step": 20580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00027529233538112164,
+      "loss": 2.7265,
+      "step": 20590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00027526900214661337,
+      "loss": 2.7139,
+      "step": 20600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00027524565888948895,
+      "loss": 2.6866,
+      "step": 20610
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002752223056116161,
+      "loss": 2.7548,
+      "step": 20620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002751989423148632,
+      "loss": 2.7161,
+      "step": 20630
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002751755690010995,
+      "loss": 2.7375,
+      "step": 20640
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00027515218567219515,
+      "loss": 2.7078,
+      "step": 20650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00027512879233002096,
+      "loss": 2.7212,
+      "step": 20660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002751053889764487,
+      "loss": 2.7112,
+      "step": 20670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002750819756133507,
+      "loss": 2.7365,
+      "step": 20680
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002750585522426003,
+      "loss": 2.7332,
+      "step": 20690
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002750351188660716,
+      "loss": 2.7404,
+      "step": 20700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002750116754856394,
+      "loss": 2.7635,
+      "step": 20710
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00027498822210317953,
+      "loss": 2.7143,
+      "step": 20720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00027496475872056835,
+      "loss": 2.7464,
+      "step": 20730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002749412853396831,
+      "loss": 2.7322,
+      "step": 20740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00027491780196240193,
+      "loss": 2.7462,
+      "step": 20750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002748943085906037,
+      "loss": 2.7016,
+      "step": 20760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000274870805226168,
+      "loss": 2.7718,
+      "step": 20770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002748472918709755,
+      "loss": 2.7385,
+      "step": 20780
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002748237685269073,
+      "loss": 2.685,
+      "step": 20790
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00027480023519584556,
+      "loss": 2.7409,
+      "step": 20800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002747766918796731,
+      "loss": 2.7232,
+      "step": 20810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00027475313858027364,
+      "loss": 2.7167,
+      "step": 20820
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002747295752995316,
+      "loss": 2.7333,
+      "step": 20830
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002747060020393323,
+      "loss": 2.7035,
+      "step": 20840
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00027468241880156173,
+      "loss": 2.7258,
+      "step": 20850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00027465882558810683,
+      "loss": 2.7216,
+      "step": 20860
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002746352224008553,
+      "loss": 2.7116,
+      "step": 20870
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002746116092416954,
+      "loss": 2.7273,
+      "step": 20880
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00027458798611251664,
+      "loss": 2.7009,
+      "step": 20890
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00027456435301520895,
+      "loss": 2.7148,
+      "step": 20900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002745407099516632,
+      "loss": 2.6928,
+      "step": 20910
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000274517056923771,
+      "loss": 2.733,
+      "step": 20920
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002744933939334249,
+      "loss": 2.7477,
+      "step": 20930
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00027446972098251805,
+      "loss": 2.7072,
+      "step": 20940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00027444603807294456,
+      "loss": 2.7042,
+      "step": 20950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002744223452065992,
+      "loss": 2.6988,
+      "step": 20960
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00027439864238537766,
+      "loss": 2.7384,
+      "step": 20970
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002743749296111764,
+      "loss": 2.696,
+      "step": 20980
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002743512068858926,
+      "loss": 2.7048,
+      "step": 20990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002743274742114243,
+      "loss": 2.7349,
+      "step": 21000
+    },
+    {
+      "epoch": 1.89,
+      "eval_loss": 2.7612814903259277,
+      "eval_runtime": 1263.568,
+      "eval_samples_per_second": 60.254,
+      "eval_steps_per_second": 7.532,
+      "step": 21000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002743037315896703,
+      "loss": 2.7178,
+      "step": 21010
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00027427997902253025,
+      "loss": 2.776,
+      "step": 21020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002742562165119045,
+      "loss": 2.6859,
+      "step": 21030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00027423244405969436,
+      "loss": 2.7197,
+      "step": 21040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002742086616678018,
+      "loss": 2.7091,
+      "step": 21050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00027418486933812955,
+      "loss": 2.7233,
+      "step": 21060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00027416106707258125,
+      "loss": 2.7227,
+      "step": 21070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00027413725487306133,
+      "loss": 2.7429,
+      "step": 21080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00027411343274147496,
+      "loss": 2.7234,
+      "step": 21090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00027408960067972806,
+      "loss": 2.7257,
+      "step": 21100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00027406575868972746,
+      "loss": 2.722,
+      "step": 21110
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00027404190677338065,
+      "loss": 2.7203,
+      "step": 21120
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002740180449325961,
+      "loss": 2.7463,
+      "step": 21130
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002739941731692829,
+      "loss": 2.6861,
+      "step": 21140
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00027397029148535105,
+      "loss": 2.734,
+      "step": 21150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00027394639988271126,
+      "loss": 2.7138,
+      "step": 21160
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00027392249836327503,
+      "loss": 2.7113,
+      "step": 21170
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002738985869289548,
+      "loss": 2.6762,
+      "step": 21180
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00027387466558166353,
+      "loss": 2.6911,
+      "step": 21190
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002738507343233153,
+      "loss": 2.733,
+      "step": 21200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00027382679315582473,
+      "loss": 2.7385,
+      "step": 21210
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00027380284208110734,
+      "loss": 2.7411,
+      "step": 21220
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00027377888110107943,
+      "loss": 2.7224,
+      "step": 21230
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00027375491021765805,
+      "loss": 2.7173,
+      "step": 21240
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00027373092943276114,
+      "loss": 2.7432,
+      "step": 21250
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002737069387483074,
+      "loss": 2.713,
+      "step": 21260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002736829381662162,
+      "loss": 2.6877,
+      "step": 21270
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00027365892768840784,
+      "loss": 2.7221,
+      "step": 21280
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002736349073168034,
+      "loss": 2.6928,
+      "step": 21290
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00027361087705332464,
+      "loss": 2.699,
+      "step": 21300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00027358683689989425,
+      "loss": 2.7239,
+      "step": 21310
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002735627868584357,
+      "loss": 2.7304,
+      "step": 21320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00027353872693087306,
+      "loss": 2.6927,
+      "step": 21330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002735146571191315,
+      "loss": 2.7285,
+      "step": 21340
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00027349057742513666,
+      "loss": 2.7263,
+      "step": 21350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002734664878508153,
+      "loss": 2.6903,
+      "step": 21360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002734423883980946,
+      "loss": 2.7265,
+      "step": 21370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002734182790689029,
+      "loss": 2.7076,
+      "step": 21380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00027339415986516905,
+      "loss": 2.6826,
+      "step": 21390
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002733700307888228,
+      "loss": 2.7201,
+      "step": 21400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002733458918417948,
+      "loss": 2.7005,
+      "step": 21410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00027332174302601624,
+      "loss": 2.7034,
+      "step": 21420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002732975843434193,
+      "loss": 2.7143,
+      "step": 21430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002732734157959369,
+      "loss": 2.749,
+      "step": 21440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002732492373855027,
+      "loss": 2.7557,
+      "step": 21450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002732250491140512,
+      "loss": 2.7415,
+      "step": 21460
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002732008509835177,
+      "loss": 2.7474,
+      "step": 21470
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00027317664299583813,
+      "loss": 2.7196,
+      "step": 21480
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00027315242515294955,
+      "loss": 2.7365,
+      "step": 21490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002731281974567894,
+      "loss": 2.7302,
+      "step": 21500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002731039599092963,
+      "loss": 2.6979,
+      "step": 21510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002730797125124093,
+      "loss": 2.6821,
+      "step": 21520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002730554552680685,
+      "loss": 2.6877,
+      "step": 21530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002730311881782146,
+      "loss": 2.7016,
+      "step": 21540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002730093393810372,
+      "loss": 2.7009,
+      "step": 21550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00027298505359005834,
+      "loss": 2.7403,
+      "step": 21560
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002729607579591991,
+      "loss": 2.734,
+      "step": 21570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002729364524904035,
+      "loss": 2.731,
+      "step": 21580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00027291213718561605,
+      "loss": 2.7112,
+      "step": 21590
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002728878120467824,
+      "loss": 2.7173,
+      "step": 21600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002728634770758485,
+      "loss": 2.7242,
+      "step": 21610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00027283913227476155,
+      "loss": 2.6897,
+      "step": 21620
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002728147776454693,
+      "loss": 2.7383,
+      "step": 21630
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002727904131899203,
+      "loss": 2.7007,
+      "step": 21640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000272766038910064,
+      "loss": 2.6835,
+      "step": 21650
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002727416548078504,
+      "loss": 2.7231,
+      "step": 21660
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00027271726088523064,
+      "loss": 2.6974,
+      "step": 21670
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002726928571441563,
+      "loss": 2.6922,
+      "step": 21680
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00027266844358658,
+      "loss": 2.726,
+      "step": 21690
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00027264402021445495,
+      "loss": 2.7183,
+      "step": 21700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002726195870297353,
+      "loss": 2.7248,
+      "step": 21710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00027259514403437583,
+      "loss": 2.7048,
+      "step": 21720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002725706912303323,
+      "loss": 2.7228,
+      "step": 21730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002725462286195611,
+      "loss": 2.7347,
+      "step": 21740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002725217562040194,
+      "loss": 2.6957,
+      "step": 21750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002724972739856653,
+      "loss": 2.7226,
+      "step": 21760
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00027247278196645746,
+      "loss": 2.7064,
+      "step": 21770
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00027244828014835564,
+      "loss": 2.7008,
+      "step": 21780
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002724237685333201,
+      "loss": 2.7013,
+      "step": 21790
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00027239924712331193,
+      "loss": 2.7459,
+      "step": 21800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002723747159202932,
+      "loss": 2.7047,
+      "step": 21810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002723501749262264,
+      "loss": 2.6991,
+      "step": 21820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002723256241430753,
+      "loss": 2.7051,
+      "step": 21830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00027230106357280394,
+      "loss": 2.7176,
+      "step": 21840
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00027227649321737754,
+      "loss": 2.7063,
+      "step": 21850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00027225191307876184,
+      "loss": 2.6705,
+      "step": 21860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00027222732315892355,
+      "loss": 2.7088,
+      "step": 21870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00027220272345983005,
+      "loss": 2.7096,
+      "step": 21880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00027217811398344945,
+      "loss": 2.7226,
+      "step": 21890
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002721534947317508,
+      "loss": 2.7118,
+      "step": 21900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00027212886570670387,
+      "loss": 2.7064,
+      "step": 21910
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002721042269102792,
+      "loss": 2.7088,
+      "step": 21920
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00027207957834444807,
+      "loss": 2.686,
+      "step": 21930
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00027205492001118256,
+      "loss": 2.7142,
+      "step": 21940
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002720302519124556,
+      "loss": 2.7236,
+      "step": 21950
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002720055740502409,
+      "loss": 2.6873,
+      "step": 21960
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00027198088642651277,
+      "loss": 2.708,
+      "step": 21970
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00027195618904324656,
+      "loss": 2.7035,
+      "step": 21980
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002719314819024182,
+      "loss": 2.7356,
+      "step": 21990
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00027190676500600454,
+      "loss": 2.6631,
+      "step": 22000
+    },
+    {
+      "epoch": 1.98,
+      "eval_loss": 2.7451064586639404,
+      "eval_runtime": 1263.4248,
+      "eval_samples_per_second": 60.261,
+      "eval_steps_per_second": 7.533,
+      "step": 22000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002718820383559831,
+      "loss": 2.7473,
+      "step": 22010
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002718573019543322,
+      "loss": 2.7095,
+      "step": 22020
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00027183255580303105,
+      "loss": 2.6788,
+      "step": 22030
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00027180779990405943,
+      "loss": 2.697,
+      "step": 22040
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002717830342593982,
+      "loss": 2.682,
+      "step": 22050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00027175825887102876,
+      "loss": 2.6548,
+      "step": 22060
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00027173347374093333,
+      "loss": 2.7592,
+      "step": 22070
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002717086788710949,
+      "loss": 2.6695,
+      "step": 22080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002716838742634974,
+      "loss": 2.7335,
+      "step": 22090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00027165905992012533,
+      "loss": 2.685,
+      "step": 22100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00027163423584296406,
+      "loss": 2.6582,
+      "step": 22110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002716094020339997,
+      "loss": 2.7059,
+      "step": 22120
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00027158455849521926,
+      "loss": 2.679,
+      "step": 22130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002715597052286104,
+      "loss": 2.6812,
+      "step": 22140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00027153484223616154,
+      "loss": 2.6677,
+      "step": 22150
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00027150996951986204,
+      "loss": 2.6568,
+      "step": 22160
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00027148508708170184,
+      "loss": 2.6909,
+      "step": 22170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00027146019492367183,
+      "loss": 2.6955,
+      "step": 22180
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00027143529304776357,
+      "loss": 2.7042,
+      "step": 22190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002714103814559694,
+      "loss": 2.6916,
+      "step": 22200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00027138546015028253,
+      "loss": 2.7031,
+      "step": 22210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002713605291326968,
+      "loss": 2.6827,
+      "step": 22220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002713355884052069,
+      "loss": 2.6184,
+      "step": 22230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00027131063796980847,
+      "loss": 2.6497,
+      "step": 22240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002712856778284976,
+      "loss": 2.6396,
+      "step": 22250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002712607079832714,
+      "loss": 2.6511,
+      "step": 22260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002712357284361276,
+      "loss": 2.6349,
+      "step": 22270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00027121073918906486,
+      "loss": 2.6652,
+      "step": 22280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002711857402440825,
+      "loss": 2.639,
+      "step": 22290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00027116073160318065,
+      "loss": 2.6505,
+      "step": 22300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002711357132683603,
+      "loss": 2.6544,
+      "step": 22310
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000271110685241623,
+      "loss": 2.6352,
+      "step": 22320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002710856475249713,
+      "loss": 2.6813,
+      "step": 22330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002710606001204084,
+      "loss": 2.6627,
+      "step": 22340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002710355430299383,
+      "loss": 2.6397,
+      "step": 22350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002710104762555659,
+      "loss": 2.6427,
+      "step": 22360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002709853997992966,
+      "loss": 2.6182,
+      "step": 22370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00027096031366313685,
+      "loss": 2.6559,
+      "step": 22380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002709352178490937,
+      "loss": 2.6584,
+      "step": 22390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002709101123591751,
+      "loss": 2.698,
+      "step": 22400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002708849971953896,
+      "loss": 2.6569,
+      "step": 22410
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00027085987235974675,
+      "loss": 2.6377,
+      "step": 22420
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00027083473785425666,
+      "loss": 2.6539,
+      "step": 22430
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00027080959368093043,
+      "loss": 2.6811,
+      "step": 22440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00027078443984177963,
+      "loss": 2.6569,
+      "step": 22450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00027075927633881693,
+      "loss": 2.6943,
+      "step": 22460
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002707341031740557,
+      "loss": 2.6722,
+      "step": 22470
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002707089203495098,
+      "loss": 2.6723,
+      "step": 22480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002706837278671942,
+      "loss": 2.6343,
+      "step": 22490
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002706585257291245,
+      "loss": 2.6524,
+      "step": 22500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00027063331393731716,
+      "loss": 2.6812,
+      "step": 22510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00027060809249378925,
+      "loss": 2.6771,
+      "step": 22520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002705828614005587,
+      "loss": 2.6543,
+      "step": 22530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002705576206596443,
+      "loss": 2.6807,
+      "step": 22540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00027053237027306544,
+      "loss": 2.6563,
+      "step": 22550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00027050711024284243,
+      "loss": 2.6954,
+      "step": 22560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002704818405709963,
+      "loss": 2.6544,
+      "step": 22570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002704565612595488,
+      "loss": 2.641,
+      "step": 22580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00027043127231052256,
+      "loss": 2.6696,
+      "step": 22590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002704059737259408,
+      "loss": 2.6565,
+      "step": 22600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002703806655078278,
+      "loss": 2.6674,
+      "step": 22610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002703553476582083,
+      "loss": 2.6669,
+      "step": 22620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00027033002017910795,
+      "loss": 2.6534,
+      "step": 22630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00027030468307255335,
+      "loss": 2.6641,
+      "step": 22640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00027027933634057145,
+      "loss": 2.6797,
+      "step": 22650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002702539799851903,
+      "loss": 2.6546,
+      "step": 22660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00027022861400843867,
+      "loss": 2.6584,
+      "step": 22670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002702032384123461,
+      "loss": 2.6698,
+      "step": 22680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002701778531989427,
+      "loss": 2.6591,
+      "step": 22690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002701524583702596,
+      "loss": 2.625,
+      "step": 22700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002701270539283287,
+      "loss": 2.6204,
+      "step": 22710
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00027010163987518243,
+      "loss": 2.6625,
+      "step": 22720
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002700762162128542,
+      "loss": 2.663,
+      "step": 22730
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00027005078294337813,
+      "loss": 2.6264,
+      "step": 22740
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002700253400687891,
+      "loss": 2.6479,
+      "step": 22750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002699998875911228,
+      "loss": 2.6593,
+      "step": 22760
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00026997442551241553,
+      "loss": 2.6553,
+      "step": 22770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00026994895383470463,
+      "loss": 2.6761,
+      "step": 22780
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00026992347256002793,
+      "loss": 2.699,
+      "step": 22790
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00026989798169042427,
+      "loss": 2.6369,
+      "step": 22800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00026987248122793304,
+      "loss": 2.6706,
+      "step": 22810
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00026984697117459454,
+      "loss": 2.6345,
+      "step": 22820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002698214515324498,
+      "loss": 2.6609,
+      "step": 22830
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00026979592230354064,
+      "loss": 2.6377,
+      "step": 22840
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002697703834899096,
+      "loss": 2.6514,
+      "step": 22850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002697448350936,
+      "loss": 2.6877,
+      "step": 22860
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002697192771166559,
+      "loss": 2.6672,
+      "step": 22870
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00026969370956112226,
+      "loss": 2.6712,
+      "step": 22880
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00026966813242904465,
+      "loss": 2.6537,
+      "step": 22890
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00026964254572246944,
+      "loss": 2.663,
+      "step": 22900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002696169494434438,
+      "loss": 2.6845,
+      "step": 22910
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002695913435940157,
+      "loss": 2.6592,
+      "step": 22920
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002695657281762338,
+      "loss": 2.6618,
+      "step": 22930
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00026954010319214756,
+      "loss": 2.6882,
+      "step": 22940
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002695144686438072,
+      "loss": 2.6556,
+      "step": 22950
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00026948882453326373,
+      "loss": 2.6427,
+      "step": 22960
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00026946317086256884,
+      "loss": 2.6675,
+      "step": 22970
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00026943750763377516,
+      "loss": 2.6623,
+      "step": 22980
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00026941183484893586,
+      "loss": 2.6479,
+      "step": 22990
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00026938615251010506,
+      "loss": 2.6766,
+      "step": 23000
+    },
+    {
+      "epoch": 2.07,
+      "eval_loss": 2.735337018966675,
+      "eval_runtime": 1264.0312,
+      "eval_samples_per_second": 60.232,
+      "eval_steps_per_second": 7.529,
+      "step": 23000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002693604606193375,
+      "loss": 2.6436,
+      "step": 23010
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002693347591786888,
+      "loss": 2.632,
+      "step": 23020
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00026930904819021535,
+      "loss": 2.609,
+      "step": 23030
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00026928332765597423,
+      "loss": 2.6647,
+      "step": 23040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002692575975780232,
+      "loss": 2.6611,
+      "step": 23050
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00026923185795842103,
+      "loss": 2.6555,
+      "step": 23060
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002692061087992271,
+      "loss": 2.6643,
+      "step": 23070
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00026918035010250145,
+      "loss": 2.6424,
+      "step": 23080
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002691545818703051,
+      "loss": 2.6641,
+      "step": 23090
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002691288041046997,
+      "loss": 2.6972,
+      "step": 23100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00026910301680774777,
+      "loss": 2.6618,
+      "step": 23110
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002690772199815124,
+      "loss": 2.6505,
+      "step": 23120
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00026905141362805766,
+      "loss": 2.6847,
+      "step": 23130
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00026902559774944824,
+      "loss": 2.6569,
+      "step": 23140
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00026899977234774963,
+      "loss": 2.6587,
+      "step": 23150
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002689739374250281,
+      "loss": 2.6543,
+      "step": 23160
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00026894809298335065,
+      "loss": 2.6618,
+      "step": 23170
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002689222390247851,
+      "loss": 2.6704,
+      "step": 23180
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00026889637555139995,
+      "loss": 2.6634,
+      "step": 23190
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00026887050256526457,
+      "loss": 2.6572,
+      "step": 23200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00026884462006844897,
+      "loss": 2.6775,
+      "step": 23210
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002688187280630239,
+      "loss": 2.6957,
+      "step": 23220
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00026879282655106117,
+      "loss": 2.6749,
+      "step": 23230
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00026876691553463295,
+      "loss": 2.6725,
+      "step": 23240
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00026874099501581237,
+      "loss": 2.6512,
+      "step": 23250
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00026871506499667334,
+      "loss": 2.6248,
+      "step": 23260
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00026868912547929044,
+      "loss": 2.6705,
+      "step": 23270
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00026866317646573914,
+      "loss": 2.6961,
+      "step": 23280
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002686372179580955,
+      "loss": 2.6844,
+      "step": 23290
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00026861124995843647,
+      "loss": 2.6602,
+      "step": 23300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002685852724688397,
+      "loss": 2.7021,
+      "step": 23310
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00026855928549138365,
+      "loss": 2.6364,
+      "step": 23320
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002685332890281474,
+      "loss": 2.6544,
+      "step": 23330
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00026850728308121106,
+      "loss": 2.7008,
+      "step": 23340
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00026848126765265524,
+      "loss": 2.6673,
+      "step": 23350
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002684552427445614,
+      "loss": 2.6673,
+      "step": 23360
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002684292083590118,
+      "loss": 2.6379,
+      "step": 23370
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002684031644980893,
+      "loss": 2.6795,
+      "step": 23380
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00026837711116387784,
+      "loss": 2.6405,
+      "step": 23390
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002683510483584617,
+      "loss": 2.6447,
+      "step": 23400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002683249760839263,
+      "loss": 2.6421,
+      "step": 23410
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002682988943423576,
+      "loss": 2.6447,
+      "step": 23420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002682728031358423,
+      "loss": 2.6437,
+      "step": 23430
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000268246702466468,
+      "loss": 2.6512,
+      "step": 23440
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00026822059233632295,
+      "loss": 2.6891,
+      "step": 23450
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002681944727474962,
+      "loss": 2.6796,
+      "step": 23460
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00026816834370207754,
+      "loss": 2.6511,
+      "step": 23470
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002681422052021576,
+      "loss": 2.6522,
+      "step": 23480
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002681160572498275,
+      "loss": 2.6648,
+      "step": 23490
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00026808989984717946,
+      "loss": 2.6608,
+      "step": 23500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00026806373299630626,
+      "loss": 2.6534,
+      "step": 23510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00026803755669930145,
+      "loss": 2.6407,
+      "step": 23520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00026801137095825946,
+      "loss": 2.642,
+      "step": 23530
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002679851757752753,
+      "loss": 2.6585,
+      "step": 23540
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00026795897115244476,
+      "loss": 2.6501,
+      "step": 23550
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00026793275709186454,
+      "loss": 2.6267,
+      "step": 23560
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002679065335956319,
+      "loss": 2.633,
+      "step": 23570
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00026788030066584505,
+      "loss": 2.6647,
+      "step": 23580
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00026785405830460283,
+      "loss": 2.6537,
+      "step": 23590
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00026782780651400483,
+      "loss": 2.6714,
+      "step": 23600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002678015452961514,
+      "loss": 2.6709,
+      "step": 23610
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002677752746531437,
+      "loss": 2.6604,
+      "step": 23620
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002677489945870837,
+      "loss": 2.6742,
+      "step": 23630
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002677227051000738,
+      "loss": 2.6575,
+      "step": 23640
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00026769640619421763,
+      "loss": 2.6517,
+      "step": 23650
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002676700978716192,
+      "loss": 2.6133,
+      "step": 23660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00026764378013438344,
+      "loss": 2.6595,
+      "step": 23670
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000267617452984616,
+      "loss": 2.6393,
+      "step": 23680
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002675911164244233,
+      "loss": 2.689,
+      "step": 23690
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002675647704559124,
+      "loss": 2.6703,
+      "step": 23700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00026753841508119137,
+      "loss": 2.6525,
+      "step": 23710
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026751468720337553,
+      "loss": 2.6388,
+      "step": 23720
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026748831396266503,
+      "loss": 2.6462,
+      "step": 23730
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002674619313218615,
+      "loss": 2.6396,
+      "step": 23740
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002674355392830759,
+      "loss": 2.6466,
+      "step": 23750
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026740913784841956,
+      "loss": 2.634,
+      "step": 23760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026738272702000505,
+      "loss": 2.6976,
+      "step": 23770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026735630679994537,
+      "loss": 2.6415,
+      "step": 23780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002673298771903544,
+      "loss": 2.6388,
+      "step": 23790
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002673034381933467,
+      "loss": 2.6735,
+      "step": 23800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00026727698981103776,
+      "loss": 2.6671,
+      "step": 23810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002672505320455435,
+      "loss": 2.648,
+      "step": 23820
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002672240648989808,
+      "loss": 2.6732,
+      "step": 23830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002671975883734674,
+      "loss": 2.6442,
+      "step": 23840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002671711024711215,
+      "loss": 2.6535,
+      "step": 23850
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002671446071940623,
+      "loss": 2.6452,
+      "step": 23860
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002671181025444096,
+      "loss": 2.6649,
+      "step": 23870
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00026709158852428406,
+      "loss": 2.6806,
+      "step": 23880
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000267065065135807,
+      "loss": 2.6555,
+      "step": 23890
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002670385323811005,
+      "loss": 2.6702,
+      "step": 23900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002670119902622875,
+      "loss": 2.6833,
+      "step": 23910
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00026698543878149147,
+      "loss": 2.6406,
+      "step": 23920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002669588779408368,
+      "loss": 2.6551,
+      "step": 23930
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002669323077424487,
+      "loss": 2.6525,
+      "step": 23940
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00026690572818845285,
+      "loss": 2.6318,
+      "step": 23950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.000266879139280976,
+      "loss": 2.6674,
+      "step": 23960
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002668525410221454,
+      "loss": 2.6447,
+      "step": 23970
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002668259334140892,
+      "loss": 2.6536,
+      "step": 23980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00026679931645893617,
+      "loss": 2.641,
+      "step": 23990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.000266772690158816,
+      "loss": 2.6405,
+      "step": 24000
+    },
+    {
+      "epoch": 2.16,
+      "eval_loss": 2.7231175899505615,
+      "eval_runtime": 1264.5004,
+      "eval_samples_per_second": 60.21,
+      "eval_steps_per_second": 7.526,
+      "step": 24000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002667460545158589,
+      "loss": 2.6594,
+      "step": 24010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00026671940953219603,
+      "loss": 2.6442,
+      "step": 24020
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00026669275520995925,
+      "loss": 2.6616,
+      "step": 24030
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00026666609155128113,
+      "loss": 2.6735,
+      "step": 24040
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002666394185582949,
+      "loss": 2.6428,
+      "step": 24050
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00026661273623313473,
+      "loss": 2.6556,
+      "step": 24060
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00026658604457793544,
+      "loss": 2.6151,
+      "step": 24070
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002665593435948325,
+      "loss": 2.6974,
+      "step": 24080
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002665326332859624,
+      "loss": 2.6932,
+      "step": 24090
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00026650591365346197,
+      "loss": 2.6586,
+      "step": 24100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00026647918469946923,
+      "loss": 2.6823,
+      "step": 24110
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002664524464261226,
+      "loss": 2.6524,
+      "step": 24120
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00026642569883556145,
+      "loss": 2.6691,
+      "step": 24130
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002663989419299257,
+      "loss": 2.638,
+      "step": 24140
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00026637217571135624,
+      "loss": 2.6834,
+      "step": 24150
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002663454001819946,
+      "loss": 2.6746,
+      "step": 24160
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00026631861534398305,
+      "loss": 2.6679,
+      "step": 24170
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002662918211994646,
+      "loss": 2.6834,
+      "step": 24180
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00026626501775058297,
+      "loss": 2.6508,
+      "step": 24190
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00026623820499948274,
+      "loss": 2.6317,
+      "step": 24200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00026621138294830916,
+      "loss": 2.683,
+      "step": 24210
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002661845515992082,
+      "loss": 2.6172,
+      "step": 24220
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00026615771095432654,
+      "loss": 2.6415,
+      "step": 24230
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002661308610158118,
+      "loss": 2.6631,
+      "step": 24240
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00026610400178581214,
+      "loss": 2.6737,
+      "step": 24250
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00026607713326647656,
+      "loss": 2.6554,
+      "step": 24260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002660502554599547,
+      "loss": 2.6621,
+      "step": 24270
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00026602336836839714,
+      "loss": 2.6333,
+      "step": 24280
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00026599647199395494,
+      "loss": 2.6368,
+      "step": 24290
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002659695663387802,
+      "loss": 2.6607,
+      "step": 24300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002659426514050255,
+      "loss": 2.6462,
+      "step": 24310
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00026591572719484425,
+      "loss": 2.6604,
+      "step": 24320
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00026588879371039075,
+      "loss": 2.6415,
+      "step": 24330
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00026586185095381983,
+      "loss": 2.6347,
+      "step": 24340
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002658348989272871,
+      "loss": 2.6678,
+      "step": 24350
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00026580793763294905,
+      "loss": 2.6935,
+      "step": 24360
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002657809670729628,
+      "loss": 2.643,
+      "step": 24370
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002657539872494862,
+      "loss": 2.6501,
+      "step": 24380
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002657269981646779,
+      "loss": 2.6712,
+      "step": 24390
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002656999998206972,
+      "loss": 2.6427,
+      "step": 24400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002656729922197043,
+      "loss": 2.6432,
+      "step": 24410
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00026564597536385995,
+      "loss": 2.6935,
+      "step": 24420
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00026561894925532584,
+      "loss": 2.6481,
+      "step": 24430
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002655919138962642,
+      "loss": 2.6387,
+      "step": 24440
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002655648692888381,
+      "loss": 2.6703,
+      "step": 24450
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002655378154352114,
+      "loss": 2.6562,
+      "step": 24460
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00026551075233754865,
+      "loss": 2.6688,
+      "step": 24470
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002654836799980151,
+      "loss": 2.666,
+      "step": 24480
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00026545659841877673,
+      "loss": 2.6336,
+      "step": 24490
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002654295076020004,
+      "loss": 2.6735,
+      "step": 24500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00026540240754985354,
+      "loss": 2.6842,
+      "step": 24510
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00026537529826450444,
+      "loss": 2.6403,
+      "step": 24520
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002653481797481221,
+      "loss": 2.6763,
+      "step": 24530
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002653210520028761,
+      "loss": 2.6646,
+      "step": 24540
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00026529391503093703,
+      "loss": 2.6481,
+      "step": 24550
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00026526676883447607,
+      "loss": 2.6687,
+      "step": 24560
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002652396134156651,
+      "loss": 2.6625,
+      "step": 24570
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00026521244877667687,
+      "loss": 2.6279,
+      "step": 24580
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00026518527491968465,
+      "loss": 2.6805,
+      "step": 24590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002651580918468628,
+      "loss": 2.6323,
+      "step": 24600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00026513089956038596,
+      "loss": 2.6475,
+      "step": 24610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00026510369806242995,
+      "loss": 2.6528,
+      "step": 24620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.000265076487355171,
+      "loss": 2.6519,
+      "step": 24630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00026504926744078626,
+      "loss": 2.6169,
+      "step": 24640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002650220383214536,
+      "loss": 2.6677,
+      "step": 24650
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002649947999993515,
+      "loss": 2.6405,
+      "step": 24660
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002649675524766593,
+      "loss": 2.6453,
+      "step": 24670
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002649402957555571,
+      "loss": 2.6488,
+      "step": 24680
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00026491302983822556,
+      "loss": 2.6533,
+      "step": 24690
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002648857547268463,
+      "loss": 2.6636,
+      "step": 24700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002648584704236015,
+      "loss": 2.6273,
+      "step": 24710
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002648311769306742,
+      "loss": 2.6273,
+      "step": 24720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002648038742502481,
+      "loss": 2.6358,
+      "step": 24730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00026477656238450755,
+      "loss": 2.6492,
+      "step": 24740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002647492413356379,
+      "loss": 2.6656,
+      "step": 24750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00026472191110582494,
+      "loss": 2.6408,
+      "step": 24760
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00026469457169725545,
+      "loss": 2.6482,
+      "step": 24770
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002646672231121167,
+      "loss": 2.6603,
+      "step": 24780
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002646398653525969,
+      "loss": 2.6544,
+      "step": 24790
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00026461249842088487,
+      "loss": 2.6632,
+      "step": 24800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002645851223191702,
+      "loss": 2.6463,
+      "step": 24810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00026455773704964333,
+      "loss": 2.6192,
+      "step": 24820
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002645303426144951,
+      "loss": 2.6263,
+      "step": 24830
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00026450293901591755,
+      "loss": 2.6222,
+      "step": 24840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.000264475526256103,
+      "loss": 2.6275,
+      "step": 24850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00026444810433724487,
+      "loss": 2.6384,
+      "step": 24860
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00026442067326153704,
+      "loss": 2.6589,
+      "step": 24870
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002643932330311743,
+      "loss": 2.6406,
+      "step": 24880
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.000264365783648352,
+      "loss": 2.6616,
+      "step": 24890
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002643383251152665,
+      "loss": 2.6627,
+      "step": 24900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00026431085743411467,
+      "loss": 2.6272,
+      "step": 24910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00026428338060709405,
+      "loss": 2.6525,
+      "step": 24920
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002642558946364032,
+      "loss": 2.6264,
+      "step": 24930
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002642283995242411,
+      "loss": 2.6472,
+      "step": 24940
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00026420089527280763,
+      "loss": 2.6635,
+      "step": 24950
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00026417338188430346,
+      "loss": 2.6461,
+      "step": 24960
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002641458593609298,
+      "loss": 2.64,
+      "step": 24970
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00026411832770488865,
+      "loss": 2.6025,
+      "step": 24980
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00026409078691838297,
+      "loss": 2.633,
+      "step": 24990
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002640632370036161,
+      "loss": 2.6707,
+      "step": 25000
+    },
+    {
+      "epoch": 2.25,
+      "eval_loss": 2.7120893001556396,
+      "eval_runtime": 1263.6911,
+      "eval_samples_per_second": 60.248,
+      "eval_steps_per_second": 7.531,
+      "step": 25000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002640356779627923,
+      "loss": 2.6599,
+      "step": 25010
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00026400810979811665,
+      "loss": 2.6417,
+      "step": 25020
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002639805325117947,
+      "loss": 2.5979,
+      "step": 25030
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00026395294610603295,
+      "loss": 2.6324,
+      "step": 25040
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00026392535058303854,
+      "loss": 2.6434,
+      "step": 25050
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00026389774594501927,
+      "loss": 2.6619,
+      "step": 25060
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002638701321941839,
+      "loss": 2.6561,
+      "step": 25070
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002638425093327417,
+      "loss": 2.66,
+      "step": 25080
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002638148773629027,
+      "loss": 2.6716,
+      "step": 25090
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002637872362868778,
+      "loss": 2.6328,
+      "step": 25100
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002637595861068784,
+      "loss": 2.667,
+      "step": 25110
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002637319268251169,
+      "loss": 2.6682,
+      "step": 25120
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00026370425844380615,
+      "loss": 2.6636,
+      "step": 25130
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00026367658096515995,
+      "loss": 2.6585,
+      "step": 25140
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00026364889439139266,
+      "loss": 2.6497,
+      "step": 25150
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002636211987247196,
+      "loss": 2.6509,
+      "step": 25160
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002635934939673565,
+      "loss": 2.6308,
+      "step": 25170
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026356578012152006,
+      "loss": 2.6371,
+      "step": 25180
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026353805718942763,
+      "loss": 2.6526,
+      "step": 25190
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026351032517329725,
+      "loss": 2.656,
+      "step": 25200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026348258407534776,
+      "loss": 2.6576,
+      "step": 25210
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026345483389779873,
+      "loss": 2.6572,
+      "step": 25220
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026342707464287033,
+      "loss": 2.6561,
+      "step": 25230
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026339930631278356,
+      "loss": 2.6528,
+      "step": 25240
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026337152890976016,
+      "loss": 2.6396,
+      "step": 25250
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002633437424360226,
+      "loss": 2.6319,
+      "step": 25260
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000263315946893794,
+      "loss": 2.6284,
+      "step": 25270
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026328814228529816,
+      "loss": 2.6697,
+      "step": 25280
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002632603286127598,
+      "loss": 2.6753,
+      "step": 25290
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002632325058784043,
+      "loss": 2.6157,
+      "step": 25300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002632046740844576,
+      "loss": 2.664,
+      "step": 25310
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026317683323314654,
+      "loss": 2.654,
+      "step": 25320
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002631489833266986,
+      "loss": 2.6402,
+      "step": 25330
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026312112436734217,
+      "loss": 2.6435,
+      "step": 25340
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000263093256357306,
+      "loss": 2.6467,
+      "step": 25350
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002630653792988199,
+      "loss": 2.6169,
+      "step": 25360
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026303749319411417,
+      "loss": 2.6185,
+      "step": 25370
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026300959804542013,
+      "loss": 2.6452,
+      "step": 25380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026298169385496943,
+      "loss": 2.6563,
+      "step": 25390
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002629537806249948,
+      "loss": 2.6449,
+      "step": 25400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026292585835772947,
+      "loss": 2.6277,
+      "step": 25410
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026289792705540743,
+      "loss": 2.6532,
+      "step": 25420
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026286998672026354,
+      "loss": 2.6446,
+      "step": 25430
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026284203735453315,
+      "loss": 2.6502,
+      "step": 25440
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002628140789604526,
+      "loss": 2.6515,
+      "step": 25450
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002627861115402586,
+      "loss": 2.655,
+      "step": 25460
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026275813509618904,
+      "loss": 2.6368,
+      "step": 25470
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002627301496304821,
+      "loss": 2.6532,
+      "step": 25480
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026270215514537687,
+      "loss": 2.6739,
+      "step": 25490
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00026267415164311317,
+      "loss": 2.6647,
+      "step": 25500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00026264613912593165,
+      "loss": 2.6469,
+      "step": 25510
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00026261811759607334,
+      "loss": 2.6421,
+      "step": 25520
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00026259008705578044,
+      "loss": 2.669,
+      "step": 25530
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00026256204750729545,
+      "loss": 2.6902,
+      "step": 25540
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00026253399895286185,
+      "loss": 2.6523,
+      "step": 25550
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002625059413947238,
+      "loss": 2.6445,
+      "step": 25560
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002624778748351261,
+      "loss": 2.6397,
+      "step": 25570
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00026244979927631434,
+      "loss": 2.6374,
+      "step": 25580
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002624217147205348,
+      "loss": 2.6135,
+      "step": 25590
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00026239362117003455,
+      "loss": 2.665,
+      "step": 25600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00026236551862706125,
+      "loss": 2.6618,
+      "step": 25610
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00026233740709386336,
+      "loss": 2.6739,
+      "step": 25620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00026230928657269,
+      "loss": 2.6384,
+      "step": 25630
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00026228115706579126,
+      "loss": 2.6706,
+      "step": 25640
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002622530185754175,
+      "loss": 2.6408,
+      "step": 25650
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00026222487110382013,
+      "loss": 2.6556,
+      "step": 25660
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002621967146532513,
+      "loss": 2.6576,
+      "step": 25670
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002621685492259636,
+      "loss": 2.671,
+      "step": 25680
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00026214037482421063,
+      "loss": 2.6594,
+      "step": 25690
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00026211219145024655,
+      "loss": 2.6432,
+      "step": 25700
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002620839991063263,
+      "loss": 2.6352,
+      "step": 25710
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002620557977947055,
+      "loss": 2.6456,
+      "step": 25720
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00026202758751764046,
+      "loss": 2.638,
+      "step": 25730
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002619993682773883,
+      "loss": 2.6695,
+      "step": 25740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002619711400762068,
+      "loss": 2.6253,
+      "step": 25750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00026194290291635446,
+      "loss": 2.6662,
+      "step": 25760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00026191465680009047,
+      "loss": 2.6558,
+      "step": 25770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00026188640172967483,
+      "loss": 2.6533,
+      "step": 25780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002618581377073681,
+      "loss": 2.6301,
+      "step": 25790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002618298647354318,
+      "loss": 2.655,
+      "step": 25800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002618015828161278,
+      "loss": 2.6082,
+      "step": 25810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00026177329195171907,
+      "loss": 2.5971,
+      "step": 25820
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00026174499214446907,
+      "loss": 2.6402,
+      "step": 25830
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00026171668339664206,
+      "loss": 2.6247,
+      "step": 25840
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00026168836571050294,
+      "loss": 2.6361,
+      "step": 25850
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002616600390883174,
+      "loss": 2.6592,
+      "step": 25860
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00026163170353235186,
+      "loss": 2.602,
+      "step": 25870
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002616033590448733,
+      "loss": 2.6626,
+      "step": 25880
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00026157500562814964,
+      "loss": 2.6424,
+      "step": 25890
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00026154664328444934,
+      "loss": 2.6731,
+      "step": 25900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00026151827201604166,
+      "loss": 2.6157,
+      "step": 25910
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00026148989182519654,
+      "loss": 2.6278,
+      "step": 25920
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002614615027141846,
+      "loss": 2.626,
+      "step": 25930
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00026143310468527736,
+      "loss": 2.6222,
+      "step": 25940
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002614075388363381,
+      "loss": 2.6222,
+      "step": 25950
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00026137912386968974,
+      "loss": 2.6442,
+      "step": 25960
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00026135069999173703,
+      "loss": 2.6336,
+      "step": 25970
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002613222672047541,
+      "loss": 2.6313,
+      "step": 25980
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00026129382551101583,
+      "loss": 2.6083,
+      "step": 25990
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002612653749127978,
+      "loss": 2.6362,
+      "step": 26000
+    },
+    {
+      "epoch": 2.34,
+      "eval_loss": 2.700486898422241,
+      "eval_runtime": 1263.1131,
+      "eval_samples_per_second": 60.276,
+      "eval_steps_per_second": 7.535,
+      "step": 26000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002612369154123763,
+      "loss": 2.6266,
+      "step": 26010
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002612084470120284,
+      "loss": 2.6674,
+      "step": 26020
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00026117996971403173,
+      "loss": 2.6532,
+      "step": 26030
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002611514835206648,
+      "loss": 2.6543,
+      "step": 26040
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002611229884342067,
+      "loss": 2.6371,
+      "step": 26050
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002610944844569373,
+      "loss": 2.6642,
+      "step": 26060
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002610659715911372,
+      "loss": 2.644,
+      "step": 26070
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002610374498390876,
+      "loss": 2.6528,
+      "step": 26080
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002610089192030706,
+      "loss": 2.6684,
+      "step": 26090
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002609803796853688,
+      "loss": 2.6266,
+      "step": 26100
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002609518312882657,
+      "loss": 2.6644,
+      "step": 26110
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002609232740140454,
+      "loss": 2.6601,
+      "step": 26120
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002608947078649926,
+      "loss": 2.6545,
+      "step": 26130
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00026086613284339303,
+      "loss": 2.6205,
+      "step": 26140
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002608375489515328,
+      "loss": 2.6598,
+      "step": 26150
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000260808956191699,
+      "loss": 2.6384,
+      "step": 26160
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002607803545661792,
+      "loss": 2.6571,
+      "step": 26170
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00026075174407726187,
+      "loss": 2.6544,
+      "step": 26180
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00026072312472723595,
+      "loss": 2.6134,
+      "step": 26190
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002606944965183914,
+      "loss": 2.6493,
+      "step": 26200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002606658594530186,
+      "loss": 2.6418,
+      "step": 26210
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00026063721353340887,
+      "loss": 2.6523,
+      "step": 26220
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00026060855876185407,
+      "loss": 2.655,
+      "step": 26230
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00026057989514064687,
+      "loss": 2.6378,
+      "step": 26240
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00026055122267208057,
+      "loss": 2.6239,
+      "step": 26250
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002605225413584493,
+      "loss": 2.629,
+      "step": 26260
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002604938512020477,
+      "loss": 2.617,
+      "step": 26270
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002604651522051713,
+      "loss": 2.641,
+      "step": 26280
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00026043644437011624,
+      "loss": 2.6384,
+      "step": 26290
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00026040772769917944,
+      "loss": 2.675,
+      "step": 26300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00026037900219465844,
+      "loss": 2.6581,
+      "step": 26310
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002603502678588516,
+      "loss": 2.6602,
+      "step": 26320
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00026032152469405783,
+      "loss": 2.5972,
+      "step": 26330
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002602927727025769,
+      "loss": 2.6499,
+      "step": 26340
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002602640118867092,
+      "loss": 2.642,
+      "step": 26350
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002602352422487558,
+      "loss": 2.6306,
+      "step": 26360
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00026020646379101863,
+      "loss": 2.6261,
+      "step": 26370
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00026017767651580014,
+      "loss": 2.6032,
+      "step": 26380
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002601488804254036,
+      "loss": 2.6372,
+      "step": 26390
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002601200755221329,
+      "loss": 2.6395,
+      "step": 26400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002600912618082927,
+      "loss": 2.6781,
+      "step": 26410
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002600624392861884,
+      "loss": 2.659,
+      "step": 26420
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00026003360795812596,
+      "loss": 2.6367,
+      "step": 26430
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00026000476782641225,
+      "loss": 2.6508,
+      "step": 26440
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002599759188933547,
+      "loss": 2.6424,
+      "step": 26450
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00025994706116126145,
+      "loss": 2.6267,
+      "step": 26460
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00025991819463244134,
+      "loss": 2.6397,
+      "step": 26470
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000259889319309204,
+      "loss": 2.6268,
+      "step": 26480
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002598604351938596,
+      "loss": 2.6214,
+      "step": 26490
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00025983154228871933,
+      "loss": 2.6438,
+      "step": 26500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00025980264059609473,
+      "loss": 2.643,
+      "step": 26510
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00025977373011829816,
+      "loss": 2.656,
+      "step": 26520
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002597448108576428,
+      "loss": 2.6274,
+      "step": 26530
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00025971588281644244,
+      "loss": 2.682,
+      "step": 26540
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00025968694599701154,
+      "loss": 2.6556,
+      "step": 26550
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002596580004016653,
+      "loss": 2.6255,
+      "step": 26560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002596290460327197,
+      "loss": 2.6284,
+      "step": 26570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002596000828924911,
+      "loss": 2.6611,
+      "step": 26580
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002595711109832971,
+      "loss": 2.6291,
+      "step": 26590
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00025954213030745556,
+      "loss": 2.6435,
+      "step": 26600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002595131408672852,
+      "loss": 2.6114,
+      "step": 26610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002594841426651054,
+      "loss": 2.6122,
+      "step": 26620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002594551357032363,
+      "loss": 2.5895,
+      "step": 26630
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002594261199839988,
+      "loss": 2.6345,
+      "step": 26640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00025939709550971426,
+      "loss": 2.6345,
+      "step": 26650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000259368062282705,
+      "loss": 2.6356,
+      "step": 26660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002593390203052939,
+      "loss": 2.6262,
+      "step": 26670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002593099695798045,
+      "loss": 2.6497,
+      "step": 26680
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00025928091010856124,
+      "loss": 2.6472,
+      "step": 26690
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000259251841893889,
+      "loss": 2.6421,
+      "step": 26700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00025922276493811364,
+      "loss": 2.6373,
+      "step": 26710
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00025919367924356147,
+      "loss": 2.6517,
+      "step": 26720
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00025916458481255964,
+      "loss": 2.6348,
+      "step": 26730
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002591354816474359,
+      "loss": 2.6152,
+      "step": 26740
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00025910636975051877,
+      "loss": 2.6211,
+      "step": 26750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002590772491241375,
+      "loss": 2.6199,
+      "step": 26760
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000259048119770622,
+      "loss": 2.6577,
+      "step": 26770
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002590189816923029,
+      "loss": 2.6229,
+      "step": 26780
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002589898348915114,
+      "loss": 2.6401,
+      "step": 26790
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00025896067937057957,
+      "loss": 2.6162,
+      "step": 26800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002589315151318401,
+      "loss": 2.6195,
+      "step": 26810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00025890234217762633,
+      "loss": 2.6487,
+      "step": 26820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00025887316051027246,
+      "loss": 2.6564,
+      "step": 26830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002588439701321131,
+      "loss": 2.6278,
+      "step": 26840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00025881477104548403,
+      "loss": 2.647,
+      "step": 26850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002587855632527211,
+      "loss": 2.619,
+      "step": 26860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002587563467561614,
+      "loss": 2.6712,
+      "step": 26870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00025872712155814245,
+      "loss": 2.6339,
+      "step": 26880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002586978876610025,
+      "loss": 2.5873,
+      "step": 26890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002586686450670805,
+      "loss": 2.6321,
+      "step": 26900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00025863939377871625,
+      "loss": 2.6067,
+      "step": 26910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002586101337982499,
+      "loss": 2.6486,
+      "step": 26920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002585808651280227,
+      "loss": 2.6362,
+      "step": 26930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00025855158777037625,
+      "loss": 2.6282,
+      "step": 26940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002585223017276531,
+      "loss": 2.6589,
+      "step": 26950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00025849300700219634,
+      "loss": 2.6195,
+      "step": 26960
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002584637035963498,
+      "loss": 2.616,
+      "step": 26970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000258434391512458,
+      "loss": 2.6291,
+      "step": 26980
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002584050707528662,
+      "loss": 2.6551,
+      "step": 26990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002583757413199203,
+      "loss": 2.5997,
+      "step": 27000
+    },
+    {
+      "epoch": 2.43,
+      "eval_loss": 2.690363645553589,
+      "eval_runtime": 1263.6402,
+      "eval_samples_per_second": 60.251,
+      "eval_steps_per_second": 7.531,
+      "step": 27000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00025834640321596697,
+      "loss": 2.5928,
+      "step": 27010
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00025831705644335334,
+      "loss": 2.6011,
+      "step": 27020
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002582877010044276,
+      "loss": 2.6337,
+      "step": 27030
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002582583369015384,
+      "loss": 2.662,
+      "step": 27040
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.000258228964137035,
+      "loss": 2.6354,
+      "step": 27050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00025819958271326764,
+      "loss": 2.6132,
+      "step": 27060
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.000258170192632587,
+      "loss": 2.6435,
+      "step": 27070
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00025814079389734453,
+      "loss": 2.6094,
+      "step": 27080
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002581113865098925,
+      "loss": 2.6655,
+      "step": 27090
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00025808197047258365,
+      "loss": 2.6705,
+      "step": 27100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00025805254578777147,
+      "loss": 2.6679,
+      "step": 27110
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00025802311245781034,
+      "loss": 2.6348,
+      "step": 27120
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002579936704850551,
+      "loss": 2.6463,
+      "step": 27130
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002579642198718614,
+      "loss": 2.6241,
+      "step": 27140
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00025793476062058553,
+      "loss": 2.5907,
+      "step": 27150
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002579052927335845,
+      "loss": 2.6458,
+      "step": 27160
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00025787581621321596,
+      "loss": 2.6393,
+      "step": 27170
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002578463310618383,
+      "loss": 2.6293,
+      "step": 27180
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00025781683728181063,
+      "loss": 2.6347,
+      "step": 27190
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00025778733487549267,
+      "loss": 2.647,
+      "step": 27200
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00025775782384524485,
+      "loss": 2.6172,
+      "step": 27210
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002577283041934284,
+      "loss": 2.6305,
+      "step": 27220
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002576987759224051,
+      "loss": 2.6292,
+      "step": 27230
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002576692390345374,
+      "loss": 2.6381,
+      "step": 27240
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00025763969353218865,
+      "loss": 2.6481,
+      "step": 27250
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00025761013941772267,
+      "loss": 2.6291,
+      "step": 27260
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000257580576693504,
+      "loss": 2.6277,
+      "step": 27270
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000257551005361898,
+      "loss": 2.6295,
+      "step": 27280
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002575214254252706,
+      "loss": 2.6482,
+      "step": 27290
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00025749183688598845,
+      "loss": 2.6418,
+      "step": 27300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002574622397464189,
+      "loss": 2.6324,
+      "step": 27310
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00025743263400893003,
+      "loss": 2.6168,
+      "step": 27320
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00025740301967589044,
+      "loss": 2.6435,
+      "step": 27330
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00025737339674966966,
+      "loss": 2.5989,
+      "step": 27340
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00025734376523263767,
+      "loss": 2.6593,
+      "step": 27350
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002573141251271653,
+      "loss": 2.6309,
+      "step": 27360
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002572844764356241,
+      "loss": 2.6454,
+      "step": 27370
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002572548191603861,
+      "loss": 2.6072,
+      "step": 27380
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002572251533038242,
+      "loss": 2.5836,
+      "step": 27390
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002571954788683119,
+      "loss": 2.6087,
+      "step": 27400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00025716579585622343,
+      "loss": 2.6292,
+      "step": 27410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00025713610426993376,
+      "loss": 2.6093,
+      "step": 27420
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00025710640411181834,
+      "loss": 2.6241,
+      "step": 27430
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00025707669538425356,
+      "loss": 2.6353,
+      "step": 27440
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00025704697808961633,
+      "loss": 2.6386,
+      "step": 27450
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002570172522302843,
+      "loss": 2.6318,
+      "step": 27460
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00025698751780863573,
+      "loss": 2.6452,
+      "step": 27470
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00025695777482704973,
+      "loss": 2.6157,
+      "step": 27480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.000256928023287906,
+      "loss": 2.6014,
+      "step": 27490
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00025689826319358485,
+      "loss": 2.6325,
+      "step": 27500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002568684945464674,
+      "loss": 2.6113,
+      "step": 27510
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002568387173489354,
+      "loss": 2.6292,
+      "step": 27520
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002568089316033713,
+      "loss": 2.6288,
+      "step": 27530
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002567821171257658,
+      "loss": 2.6382,
+      "step": 27540
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00025675231514550665,
+      "loss": 2.6373,
+      "step": 27550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00025672250462412837,
+      "loss": 2.6106,
+      "step": 27560
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00025669268556401597,
+      "loss": 2.6103,
+      "step": 27570
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002566628579675553,
+      "loss": 2.5983,
+      "step": 27580
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00025663302183713283,
+      "loss": 2.6468,
+      "step": 27590
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002566031771751356,
+      "loss": 2.6013,
+      "step": 27600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002565733239839516,
+      "loss": 2.6713,
+      "step": 27610
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002565434622659692,
+      "loss": 2.6496,
+      "step": 27620
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00025651359202357766,
+      "loss": 2.646,
+      "step": 27630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002564837132591669,
+      "loss": 2.6163,
+      "step": 27640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002564538259751274,
+      "loss": 2.6337,
+      "step": 27650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002564239301738505,
+      "loss": 2.6128,
+      "step": 27660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00025639402585772813,
+      "loss": 2.6349,
+      "step": 27670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002563641130291528,
+      "loss": 2.6409,
+      "step": 27680
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00025633419169051783,
+      "loss": 2.6071,
+      "step": 27690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002563042618442172,
+      "loss": 2.607,
+      "step": 27700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002562743234926456,
+      "loss": 2.6273,
+      "step": 27710
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002562443766381983,
+      "loss": 2.6505,
+      "step": 27720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00025621442128327134,
+      "loss": 2.5932,
+      "step": 27730
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002561844574302614,
+      "loss": 2.6234,
+      "step": 27740
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00025615448508156587,
+      "loss": 2.6188,
+      "step": 27750
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00025612450423958275,
+      "loss": 2.6151,
+      "step": 27760
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002560945149067108,
+      "loss": 2.6607,
+      "step": 27770
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00025606451708534944,
+      "loss": 2.6177,
+      "step": 27780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002560345107778988,
+      "loss": 2.6467,
+      "step": 27790
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002560044959867596,
+      "loss": 2.6371,
+      "step": 27800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002559744727143332,
+      "loss": 2.6277,
+      "step": 27810
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002559444409630219,
+      "loss": 2.6369,
+      "step": 27820
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00025591440073522836,
+      "loss": 2.6027,
+      "step": 27830
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002558843520333561,
+      "loss": 2.6138,
+      "step": 27840
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00025585429485980933,
+      "loss": 2.6297,
+      "step": 27850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002558242292169928,
+      "loss": 2.6153,
+      "step": 27860
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00025579415510731214,
+      "loss": 2.6005,
+      "step": 27870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00025576407253317344,
+      "loss": 2.663,
+      "step": 27880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002557339814969836,
+      "loss": 2.6354,
+      "step": 27890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00025570388200115015,
+      "loss": 2.6487,
+      "step": 27900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00025567377404808137,
+      "loss": 2.5764,
+      "step": 27910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00025564365764018613,
+      "loss": 2.6429,
+      "step": 27920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00025561353277987395,
+      "loss": 2.6119,
+      "step": 27930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002555833994695552,
+      "loss": 2.6163,
+      "step": 27940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00025555325771164065,
+      "loss": 2.6485,
+      "step": 27950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.000255523107508542,
+      "loss": 2.6139,
+      "step": 27960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002554929488626716,
+      "loss": 2.5986,
+      "step": 27970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00025546278177644233,
+      "loss": 2.6251,
+      "step": 27980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002554326062522678,
+      "loss": 2.653,
+      "step": 27990
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002554024222925623,
+      "loss": 2.6549,
+      "step": 28000
+    },
+    {
+      "epoch": 2.52,
+      "eval_loss": 2.6797590255737305,
+      "eval_runtime": 1263.3374,
+      "eval_samples_per_second": 60.265,
+      "eval_steps_per_second": 7.533,
+      "step": 28000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00025537222989974087,
+      "loss": 2.6035,
+      "step": 28010
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00025534202907621917,
+      "loss": 2.6274,
+      "step": 28020
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00025531181982441347,
+      "loss": 2.5943,
+      "step": 28030
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00025528160214674083,
+      "loss": 2.6128,
+      "step": 28040
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002552513760456189,
+      "loss": 2.5925,
+      "step": 28050
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00025522114152346604,
+      "loss": 2.6624,
+      "step": 28060
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002551908985827012,
+      "loss": 2.6373,
+      "step": 28070
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002551606472257443,
+      "loss": 2.6027,
+      "step": 28080
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00025513038745501545,
+      "loss": 2.6004,
+      "step": 28090
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00025510011927293584,
+      "loss": 2.6269,
+      "step": 28100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00025506984268192724,
+      "loss": 2.6214,
+      "step": 28110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00025503955768441187,
+      "loss": 2.6009,
+      "step": 28120
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00025500926428281297,
+      "loss": 2.6579,
+      "step": 28130
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002549789624795541,
+      "loss": 2.6204,
+      "step": 28140
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00025494865227705985,
+      "loss": 2.6166,
+      "step": 28150
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002549183336777552,
+      "loss": 2.6151,
+      "step": 28160
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002548880066840659,
+      "loss": 2.6082,
+      "step": 28170
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00025485767129841837,
+      "loss": 2.6386,
+      "step": 28180
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002548273275232398,
+      "loss": 2.6071,
+      "step": 28190
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00025479697536095786,
+      "loss": 2.602,
+      "step": 28200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00025476661481400105,
+      "loss": 2.6316,
+      "step": 28210
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002547362458847984,
+      "loss": 2.6369,
+      "step": 28220
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00025470586857577975,
+      "loss": 2.6032,
+      "step": 28230
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00025467548288937556,
+      "loss": 2.5833,
+      "step": 28240
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002546450888280169,
+      "loss": 2.5961,
+      "step": 28250
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00025461468639413566,
+      "loss": 2.6248,
+      "step": 28260
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00025458427559016416,
+      "loss": 2.6269,
+      "step": 28270
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002545538564185356,
+      "loss": 2.6072,
+      "step": 28280
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00025452342888168385,
+      "loss": 2.6306,
+      "step": 28290
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002544929929820432,
+      "loss": 2.6375,
+      "step": 28300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00025446254872204903,
+      "loss": 2.6667,
+      "step": 28310
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00025443209610413693,
+      "loss": 2.6314,
+      "step": 28320
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002544016351307435,
+      "loss": 2.6453,
+      "step": 28330
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00025437116580430576,
+      "loss": 2.6056,
+      "step": 28340
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002543406881272617,
+      "loss": 2.6088,
+      "step": 28350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002543102021020497,
+      "loss": 2.6031,
+      "step": 28360
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002542797077311089,
+      "loss": 2.6131,
+      "step": 28370
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00025424920501687914,
+      "loss": 2.5997,
+      "step": 28380
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002542186939618009,
+      "loss": 2.6531,
+      "step": 28390
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00025418817456831534,
+      "loss": 2.612,
+      "step": 28400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00025415764683886424,
+      "loss": 2.619,
+      "step": 28410
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00025412711077589013,
+      "loss": 2.6117,
+      "step": 28420
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00025409656638183617,
+      "loss": 2.6031,
+      "step": 28430
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002540660136591461,
+      "loss": 2.5895,
+      "step": 28440
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002540354526102645,
+      "loss": 2.6347,
+      "step": 28450
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00025400488323763646,
+      "loss": 2.6542,
+      "step": 28460
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00025397430554370787,
+      "loss": 2.6041,
+      "step": 28470
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002539437195309251,
+      "loss": 2.6317,
+      "step": 28480
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002539131252017354,
+      "loss": 2.6455,
+      "step": 28490
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00025388252255858655,
+      "loss": 2.5973,
+      "step": 28500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000253851911603927,
+      "loss": 2.5907,
+      "step": 28510
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002538212923402059,
+      "loss": 2.636,
+      "step": 28520
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002537906647698731,
+      "loss": 2.6462,
+      "step": 28530
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00025376002889537905,
+      "loss": 2.6225,
+      "step": 28540
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00025372938471917493,
+      "loss": 2.6518,
+      "step": 28550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00025369873224371245,
+      "loss": 2.613,
+      "step": 28560
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00025366807147144413,
+      "loss": 2.617,
+      "step": 28570
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002536374024048231,
+      "loss": 2.6451,
+      "step": 28580
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00025360672504630313,
+      "loss": 2.6099,
+      "step": 28590
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00025357603939833874,
+      "loss": 2.6077,
+      "step": 28600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000253545345463385,
+      "loss": 2.6044,
+      "step": 28610
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002535146432438977,
+      "loss": 2.5952,
+      "step": 28620
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002534839327423333,
+      "loss": 2.6289,
+      "step": 28630
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002534532139611489,
+      "loss": 2.6629,
+      "step": 28640
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002534224869028022,
+      "loss": 2.6134,
+      "step": 28650
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00025339175156975175,
+      "loss": 2.5999,
+      "step": 28660
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002533610079644566,
+      "loss": 2.6361,
+      "step": 28670
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002533302560893765,
+      "loss": 2.6299,
+      "step": 28680
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002532994959469718,
+      "loss": 2.6365,
+      "step": 28690
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002532687275397038,
+      "loss": 2.5844,
+      "step": 28700
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00025323795087003393,
+      "loss": 2.6086,
+      "step": 28710
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002532071659404249,
+      "loss": 2.6036,
+      "step": 28720
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00025317637275333956,
+      "loss": 2.6364,
+      "step": 28730
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002531455713112418,
+      "loss": 2.6287,
+      "step": 28740
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002531147616165958,
+      "loss": 2.6353,
+      "step": 28750
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002530839436718667,
+      "loss": 2.6406,
+      "step": 28760
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002530531174795203,
+      "loss": 2.629,
+      "step": 28770
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00025302228304202286,
+      "loss": 2.6374,
+      "step": 28780
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00025299144036184144,
+      "loss": 2.6322,
+      "step": 28790
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002529605894414437,
+      "loss": 2.6251,
+      "step": 28800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000252929730283298,
+      "loss": 2.6198,
+      "step": 28810
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00025289886288987337,
+      "loss": 2.6192,
+      "step": 28820
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00025286798726363943,
+      "loss": 2.62,
+      "step": 28830
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00025283710340706644,
+      "loss": 2.616,
+      "step": 28840
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00025280621132262553,
+      "loss": 2.6026,
+      "step": 28850
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002527753110127883,
+      "loss": 2.6121,
+      "step": 28860
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00025274440248002696,
+      "loss": 2.5817,
+      "step": 28870
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002527134857268145,
+      "loss": 2.5615,
+      "step": 28880
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00025268256075562456,
+      "loss": 2.646,
+      "step": 28890
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00025265162756893134,
+      "loss": 2.6246,
+      "step": 28900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00025262068616920983,
+      "loss": 2.6183,
+      "step": 28910
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002525897365589356,
+      "loss": 2.6126,
+      "step": 28920
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002525587787405849,
+      "loss": 2.6263,
+      "step": 28930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002525278127166346,
+      "loss": 2.6323,
+      "step": 28940
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00025249683848956224,
+      "loss": 2.6392,
+      "step": 28950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002524658560618461,
+      "loss": 2.5911,
+      "step": 28960
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002524348654359649,
+      "loss": 2.5907,
+      "step": 28970
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002524038666143984,
+      "loss": 2.6144,
+      "step": 28980
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00025237285959962656,
+      "loss": 2.6206,
+      "step": 28990
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002523418443941303,
+      "loss": 2.6056,
+      "step": 29000
+    },
+    {
+      "epoch": 2.61,
+      "eval_loss": 2.668823719024658,
+      "eval_runtime": 1263.5959,
+      "eval_samples_per_second": 60.253,
+      "eval_steps_per_second": 7.532,
+      "step": 29000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002523108210003911,
+      "loss": 2.5908,
+      "step": 29010
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00025227978942089107,
+      "loss": 2.6331,
+      "step": 29020
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002522487496581131,
+      "loss": 2.6106,
+      "step": 29030
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002522177017145405,
+      "loss": 2.5996,
+      "step": 29040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002521866455926575,
+      "loss": 2.6401,
+      "step": 29050
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002521555812949488,
+      "loss": 2.6334,
+      "step": 29060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00025212450882389984,
+      "loss": 2.6186,
+      "step": 29070
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002520934281819967,
+      "loss": 2.6238,
+      "step": 29080
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002520623393717261,
+      "loss": 2.5777,
+      "step": 29090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00025203124239557535,
+      "loss": 2.6314,
+      "step": 29100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002520001372560326,
+      "loss": 2.6349,
+      "step": 29110
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00025196902395558643,
+      "loss": 2.6492,
+      "step": 29120
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002519379024967262,
+      "loss": 2.6058,
+      "step": 29130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002519067728819419,
+      "loss": 2.6009,
+      "step": 29140
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00025187563511372424,
+      "loss": 2.6125,
+      "step": 29150
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00025184448919456445,
+      "loss": 2.5516,
+      "step": 29160
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002518133351269545,
+      "loss": 2.5986,
+      "step": 29170
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002517821729133869,
+      "loss": 2.6104,
+      "step": 29180
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000251751002556355,
+      "loss": 2.6504,
+      "step": 29190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00025171982405835266,
+      "loss": 2.6157,
+      "step": 29200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00025168863742187447,
+      "loss": 2.6126,
+      "step": 29210
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002516574426494156,
+      "loss": 2.6597,
+      "step": 29220
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00025162623974347195,
+      "loss": 2.6181,
+      "step": 29230
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00025159502870653993,
+      "loss": 2.61,
+      "step": 29240
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002515638095411168,
+      "loss": 2.6395,
+      "step": 29250
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002515325822497003,
+      "loss": 2.6067,
+      "step": 29260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00025150134683478887,
+      "loss": 2.6332,
+      "step": 29270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002514701032988817,
+      "loss": 2.59,
+      "step": 29280
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002514388516444785,
+      "loss": 2.6282,
+      "step": 29290
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00025140759187407974,
+      "loss": 2.6176,
+      "step": 29300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00025137632399018634,
+      "loss": 2.577,
+      "step": 29310
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00025134504799530005,
+      "loss": 2.593,
+      "step": 29320
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00025131376389192327,
+      "loss": 2.6118,
+      "step": 29330
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00025128247168255903,
+      "loss": 2.5751,
+      "step": 29340
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00025125117136971087,
+      "loss": 2.5833,
+      "step": 29350
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002512198629558832,
+      "loss": 2.6211,
+      "step": 29360
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002511885464435809,
+      "loss": 2.6347,
+      "step": 29370
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002511572218353095,
+      "loss": 2.5715,
+      "step": 29380
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002511258891335754,
+      "loss": 2.6336,
+      "step": 29390
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002510945483408855,
+      "loss": 2.6206,
+      "step": 29400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002510631994597471,
+      "loss": 2.574,
+      "step": 29410
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00025103184249266865,
+      "loss": 2.6126,
+      "step": 29420
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00025100047744215886,
+      "loss": 2.5847,
+      "step": 29430
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002509691043107272,
+      "loss": 2.5932,
+      "step": 29440
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00025093772310088376,
+      "loss": 2.5843,
+      "step": 29450
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002509063338151394,
+      "loss": 2.6233,
+      "step": 29460
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002508749364560055,
+      "loss": 2.619,
+      "step": 29470
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002508435310259941,
+      "loss": 2.5996,
+      "step": 29480
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000250812117527618,
+      "loss": 2.6227,
+      "step": 29490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00025078069596339047,
+      "loss": 2.6098,
+      "step": 29500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00025074926633582555,
+      "loss": 2.608,
+      "step": 29510
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002507178286474378,
+      "loss": 2.6171,
+      "step": 29520
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002506863829007426,
+      "loss": 2.6127,
+      "step": 29530
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00025065492909825585,
+      "loss": 2.5868,
+      "step": 29540
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00025062346724249416,
+      "loss": 2.614,
+      "step": 29550
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00025059199733597475,
+      "loss": 2.6235,
+      "step": 29560
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00025056051938121543,
+      "loss": 2.58,
+      "step": 29570
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00025052903338073474,
+      "loss": 2.623,
+      "step": 29580
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002504975393370519,
+      "loss": 2.5991,
+      "step": 29590
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00025046603725268664,
+      "loss": 2.6041,
+      "step": 29600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002504345271301594,
+      "loss": 2.6143,
+      "step": 29610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002504030089719913,
+      "loss": 2.5939,
+      "step": 29620
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002503714827807041,
+      "loss": 2.6093,
+      "step": 29630
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00025033994855882,
+      "loss": 2.6272,
+      "step": 29640
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002503084063088623,
+      "loss": 2.6128,
+      "step": 29650
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00025027685603335435,
+      "loss": 2.5937,
+      "step": 29660
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00025024529773482064,
+      "loss": 2.6295,
+      "step": 29670
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002502137314157861,
+      "loss": 2.6284,
+      "step": 29680
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002501821570787762,
+      "loss": 2.5958,
+      "step": 29690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002501505747263173,
+      "loss": 2.6098,
+      "step": 29700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002501189843609362,
+      "loss": 2.6052,
+      "step": 29710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00025008738598516036,
+      "loss": 2.6097,
+      "step": 29720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000250055779601518,
+      "loss": 2.5909,
+      "step": 29730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002500241652125379,
+      "loss": 2.6408,
+      "step": 29740
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00024999254282074944,
+      "loss": 2.6035,
+      "step": 29750
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00024996091242868277,
+      "loss": 2.588,
+      "step": 29760
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002499292740388685,
+      "loss": 2.5836,
+      "step": 29770
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00024989762765383804,
+      "loss": 2.6139,
+      "step": 29780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002498659732761233,
+      "loss": 2.5912,
+      "step": 29790
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00024983431090825703,
+      "loss": 2.5806,
+      "step": 29800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002498026405527724,
+      "loss": 2.6205,
+      "step": 29810
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002497709622122034,
+      "loss": 2.6426,
+      "step": 29820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002497392758890845,
+      "loss": 2.6206,
+      "step": 29830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002497075815859509,
+      "loss": 2.588,
+      "step": 29840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002496758793053384,
+      "loss": 2.6072,
+      "step": 29850
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00024964416904978354,
+      "loss": 2.616,
+      "step": 29860
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00024961245082182336,
+      "loss": 2.6155,
+      "step": 29870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002495807246239955,
+      "loss": 2.6199,
+      "step": 29880
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002495489904588385,
+      "loss": 2.6225,
+      "step": 29890
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00024951724832889135,
+      "loss": 2.5982,
+      "step": 29900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002494854982366937,
+      "loss": 2.6457,
+      "step": 29910
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002494537401847857,
+      "loss": 2.6137,
+      "step": 29920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002494219741757083,
+      "loss": 2.5866,
+      "step": 29930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002493902002120033,
+      "loss": 2.6166,
+      "step": 29940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000249361596845563,
+      "loss": 2.6329,
+      "step": 29950
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002493298077750694,
+      "loss": 2.6191,
+      "step": 29960
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00024929801075732213,
+      "loss": 2.5873,
+      "step": 29970
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00024926620579486525,
+      "loss": 2.6268,
+      "step": 29980
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00024923439289024343,
+      "loss": 2.6124,
+      "step": 29990
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00024920257204600195,
+      "loss": 2.5722,
+      "step": 30000
+    },
+    {
+      "epoch": 2.7,
+      "eval_loss": 2.659433603286743,
+      "eval_runtime": 1263.6858,
+      "eval_samples_per_second": 60.248,
+      "eval_steps_per_second": 7.531,
+      "step": 30000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002491707432646867,
+      "loss": 2.603,
+      "step": 30010
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00024913890654884444,
+      "loss": 2.6335,
+      "step": 30020
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00024910706190102227,
+      "loss": 2.6283,
+      "step": 30030
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002490752093237681,
+      "loss": 2.6351,
+      "step": 30040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002490433488196303,
+      "loss": 2.6096,
+      "step": 30050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002490114803911581,
+      "loss": 2.5792,
+      "step": 30060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002489796040409012,
+      "loss": 2.633,
+      "step": 30070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00024894771977141,
+      "loss": 2.5738,
+      "step": 30080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00024891582758523553,
+      "loss": 2.6021,
+      "step": 30090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00024888392748492944,
+      "loss": 2.6271,
+      "step": 30100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000248852019473044,
+      "loss": 2.5913,
+      "step": 30110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00024882010355213217,
+      "loss": 2.6452,
+      "step": 30120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002487881797247474,
+      "loss": 2.5934,
+      "step": 30130
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002487562479934441,
+      "loss": 2.6216,
+      "step": 30140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00024872430836077677,
+      "loss": 2.6278,
+      "step": 30150
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002486923608293011,
+      "loss": 2.6123,
+      "step": 30160
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002486604054015731,
+      "loss": 2.6379,
+      "step": 30170
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00024862844208014954,
+      "loss": 2.6395,
+      "step": 30180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002485964708675876,
+      "loss": 2.606,
+      "step": 30190
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00024856449176644544,
+      "loss": 2.6165,
+      "step": 30200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00024853250477928156,
+      "loss": 2.6123,
+      "step": 30210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002485005099086553,
+      "loss": 2.5991,
+      "step": 30220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002484685071571264,
+      "loss": 2.5601,
+      "step": 30230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002484364965272554,
+      "loss": 2.5935,
+      "step": 30240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002484044780216035,
+      "loss": 2.6041,
+      "step": 30250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00024837245164273245,
+      "loss": 2.5765,
+      "step": 30260
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002483404173932045,
+      "loss": 2.6068,
+      "step": 30270
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002483083752755829,
+      "loss": 2.5957,
+      "step": 30280
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002482763252924311,
+      "loss": 2.5972,
+      "step": 30290
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00024824426744631347,
+      "loss": 2.6408,
+      "step": 30300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002482122017397949,
+      "loss": 2.6023,
+      "step": 30310
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002481801281754409,
+      "loss": 2.5811,
+      "step": 30320
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002481480467558177,
+      "loss": 2.6158,
+      "step": 30330
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00024811595748349205,
+      "loss": 2.5907,
+      "step": 30340
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002480838603610314,
+      "loss": 2.624,
+      "step": 30350
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002480517553910038,
+      "loss": 2.5972,
+      "step": 30360
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002480196425759778,
+      "loss": 2.624,
+      "step": 30370
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002479875219185229,
+      "loss": 2.6174,
+      "step": 30380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000247955393421209,
+      "loss": 2.6014,
+      "step": 30390
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00024792325708660655,
+      "loss": 2.5828,
+      "step": 30400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00024789111291728677,
+      "loss": 2.6217,
+      "step": 30410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002478589609158216,
+      "loss": 2.617,
+      "step": 30420
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002478268010847833,
+      "loss": 2.6064,
+      "step": 30430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000247794633426745,
+      "loss": 2.6197,
+      "step": 30440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00024776245794428054,
+      "loss": 2.6093,
+      "step": 30450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00024773027463996397,
+      "loss": 2.5543,
+      "step": 30460
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002476980835163705,
+      "loss": 2.6024,
+      "step": 30470
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002476658845760755,
+      "loss": 2.5681,
+      "step": 30480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002476336778216553,
+      "loss": 2.6116,
+      "step": 30490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002476014632556867,
+      "loss": 2.5989,
+      "step": 30500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002475692408807471,
+      "loss": 2.6319,
+      "step": 30510
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00024753701069941454,
+      "loss": 2.5912,
+      "step": 30520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00024750477271426785,
+      "loss": 2.6453,
+      "step": 30530
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00024747252692788624,
+      "loss": 2.6095,
+      "step": 30540
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002474402733428497,
+      "loss": 2.6319,
+      "step": 30550
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00024740801196173874,
+      "loss": 2.6164,
+      "step": 30560
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002473757427871346,
+      "loss": 2.6021,
+      "step": 30570
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002473434658216191,
+      "loss": 2.6311,
+      "step": 30580
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002473111810677747,
+      "loss": 2.5883,
+      "step": 30590
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002472788885281844,
+      "loss": 2.6085,
+      "step": 30600
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00024724658820543196,
+      "loss": 2.605,
+      "step": 30610
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00024721428010210166,
+      "loss": 2.6134,
+      "step": 30620
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002471819642207784,
+      "loss": 2.6335,
+      "step": 30630
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002471496405640477,
+      "loss": 2.6022,
+      "step": 30640
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002471173091344959,
+      "loss": 2.5853,
+      "step": 30650
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002470849699347096,
+      "loss": 2.5845,
+      "step": 30660
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00024705262296727634,
+      "loss": 2.6061,
+      "step": 30670
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002470202682347841,
+      "loss": 2.6195,
+      "step": 30680
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002469879057398216,
+      "loss": 2.6249,
+      "step": 30690
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00024695553548497806,
+      "loss": 2.5917,
+      "step": 30700
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00024692315747284343,
+      "loss": 2.6263,
+      "step": 30710
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002468907717060082,
+      "loss": 2.6051,
+      "step": 30720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00024685837818706356,
+      "loss": 2.5841,
+      "step": 30730
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00024682597691860124,
+      "loss": 2.6086,
+      "step": 30740
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00024679356790321364,
+      "loss": 2.5541,
+      "step": 30750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002467611511434938,
+      "loss": 2.5885,
+      "step": 30760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002467287266420352,
+      "loss": 2.6312,
+      "step": 30770
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00024669629440143224,
+      "loss": 2.5981,
+      "step": 30780
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00024666385442427975,
+      "loss": 2.6009,
+      "step": 30790
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00024663140671317315,
+      "loss": 2.6335,
+      "step": 30800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002465989512707086,
+      "loss": 2.6223,
+      "step": 30810
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002465664880994828,
+      "loss": 2.6129,
+      "step": 30820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002465340172020931,
+      "loss": 2.6056,
+      "step": 30830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002465015385811374,
+      "loss": 2.6362,
+      "step": 30840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002464690522392144,
+      "loss": 2.6048,
+      "step": 30850
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00024643655817892316,
+      "loss": 2.5962,
+      "step": 30860
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00024640405640286353,
+      "loss": 2.6003,
+      "step": 30870
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00024637154691363596,
+      "loss": 2.59,
+      "step": 30880
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00024633902971384145,
+      "loss": 2.5869,
+      "step": 30890
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00024630650480608177,
+      "loss": 2.6185,
+      "step": 30900
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00024627397219295904,
+      "loss": 2.6151,
+      "step": 30910
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002462414318770762,
+      "loss": 2.5897,
+      "step": 30920
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002462088838610369,
+      "loss": 2.6064,
+      "step": 30930
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002461763281474451,
+      "loss": 2.6172,
+      "step": 30940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002461437647389056,
+      "loss": 2.5775,
+      "step": 30950
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002461111936380237,
+      "loss": 2.6073,
+      "step": 30960
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00024607861484740557,
+      "loss": 2.6036,
+      "step": 30970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00024604602836965757,
+      "loss": 2.6054,
+      "step": 30980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00024601343420738696,
+      "loss": 2.5912,
+      "step": 30990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002459808323632016,
+      "loss": 2.6179,
+      "step": 31000
+    },
+    {
+      "epoch": 2.79,
+      "eval_loss": 2.650871515274048,
+      "eval_runtime": 1263.5804,
+      "eval_samples_per_second": 60.253,
+      "eval_steps_per_second": 7.532,
+      "step": 31000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00024594822283970997,
+      "loss": 2.5965,
+      "step": 31010
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00024591560563952104,
+      "loss": 2.6082,
+      "step": 31020
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00024588298076524446,
+      "loss": 2.614,
+      "step": 31030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002458503482194905,
+      "loss": 2.6199,
+      "step": 31040
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002458177080048702,
+      "loss": 2.6159,
+      "step": 31050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00024578506012399486,
+      "loss": 2.6167,
+      "step": 31060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002457524045794767,
+      "loss": 2.5906,
+      "step": 31070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00024571974137392843,
+      "loss": 2.6055,
+      "step": 31080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002456870705099634,
+      "loss": 2.6112,
+      "step": 31090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00024565439199019556,
+      "loss": 2.6065,
+      "step": 31100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002456217058172395,
+      "loss": 2.5929,
+      "step": 31110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00024558901199371033,
+      "loss": 2.6149,
+      "step": 31120
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00024555631052222395,
+      "loss": 2.5816,
+      "step": 31130
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00024552360140539666,
+      "loss": 2.585,
+      "step": 31140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002454908846458455,
+      "loss": 2.5664,
+      "step": 31150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00024545816024618817,
+      "loss": 2.5768,
+      "step": 31160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00024542542820904284,
+      "loss": 2.6151,
+      "step": 31170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002453926885370283,
+      "loss": 2.6222,
+      "step": 31180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002453599412327642,
+      "loss": 2.5875,
+      "step": 31190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002453271862988704,
+      "loss": 2.6053,
+      "step": 31200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00024529442373796774,
+      "loss": 2.6162,
+      "step": 31210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002452616535526774,
+      "loss": 2.5869,
+      "step": 31220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00024522887574562137,
+      "loss": 2.5561,
+      "step": 31230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002451960903194221,
+      "loss": 2.5713,
+      "step": 31240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002451632972767027,
+      "loss": 2.596,
+      "step": 31250
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000245130496620087,
+      "loss": 2.627,
+      "step": 31260
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002450976883521992,
+      "loss": 2.6159,
+      "step": 31270
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002450648724756644,
+      "loss": 2.608,
+      "step": 31280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00024503204899310797,
+      "loss": 2.6203,
+      "step": 31290
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00024499921790715627,
+      "loss": 2.6115,
+      "step": 31300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002449663792204359,
+      "loss": 2.573,
+      "step": 31310
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00024493353293557444,
+      "loss": 2.5986,
+      "step": 31320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002449006790551997,
+      "loss": 2.6017,
+      "step": 31330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00024486781758194037,
+      "loss": 2.6273,
+      "step": 31340
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00024483494851842557,
+      "loss": 2.6078,
+      "step": 31350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00024480207186728525,
+      "loss": 2.5889,
+      "step": 31360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00024476918763114975,
+      "loss": 2.582,
+      "step": 31370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002447362958126501,
+      "loss": 2.6068,
+      "step": 31380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00024470339641441787,
+      "loss": 2.579,
+      "step": 31390
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002446704894390854,
+      "loss": 2.63,
+      "step": 31400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00024463757488928556,
+      "loss": 2.587,
+      "step": 31410
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00024460465276765166,
+      "loss": 2.5655,
+      "step": 31420
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00024457172307681784,
+      "loss": 2.578,
+      "step": 31430
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002445387858194188,
+      "loss": 2.6026,
+      "step": 31440
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002445058409980898,
+      "loss": 2.6131,
+      "step": 31450
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00024447288861546666,
+      "loss": 2.605,
+      "step": 31460
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00024443992867418583,
+      "loss": 2.6185,
+      "step": 31470
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002444069611768845,
+      "loss": 2.5842,
+      "step": 31480
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00024437398612620027,
+      "loss": 2.617,
+      "step": 31490
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00024434100352477154,
+      "loss": 2.5817,
+      "step": 31500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002443080133752371,
+      "loss": 2.6219,
+      "step": 31510
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002442750156802365,
+      "loss": 2.6447,
+      "step": 31520
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002442420104424098,
+      "loss": 2.6091,
+      "step": 31530
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00024420899766439775,
+      "loss": 2.5724,
+      "step": 31540
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00024417597734884163,
+      "loss": 2.6027,
+      "step": 31550
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002441429494983834,
+      "loss": 2.5772,
+      "step": 31560
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00024410991411566552,
+      "loss": 2.6058,
+      "step": 31570
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00024407687120333116,
+      "loss": 2.6152,
+      "step": 31580
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00024404382076402404,
+      "loss": 2.6136,
+      "step": 31590
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00024401076280038844,
+      "loss": 2.61,
+      "step": 31600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00024397769731506933,
+      "loss": 2.6085,
+      "step": 31610
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002439446243107122,
+      "loss": 2.5876,
+      "step": 31620
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00024391154378996323,
+      "loss": 2.6214,
+      "step": 31630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002438784557554691,
+      "loss": 2.586,
+      "step": 31640
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00024384536020987717,
+      "loss": 2.5865,
+      "step": 31650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00024381225715583535,
+      "loss": 2.5955,
+      "step": 31660
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00024377914659599225,
+      "loss": 2.6072,
+      "step": 31670
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00024374602853299688,
+      "loss": 2.5898,
+      "step": 31680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00024371290296949905,
+      "loss": 2.6172,
+      "step": 31690
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00024367976990814912,
+      "loss": 2.6052,
+      "step": 31700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00024364662935159794,
+      "loss": 2.6043,
+      "step": 31710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00024361348130249712,
+      "loss": 2.5674,
+      "step": 31720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002435803257634988,
+      "loss": 2.5895,
+      "step": 31730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00024354716273725566,
+      "loss": 2.5874,
+      "step": 31740
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00024351399222642106,
+      "loss": 2.6131,
+      "step": 31750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00024348081423364892,
+      "loss": 2.6347,
+      "step": 31760
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00024344762876159372,
+      "loss": 2.5781,
+      "step": 31770
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00024341443581291073,
+      "loss": 2.5955,
+      "step": 31780
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00024338123539025553,
+      "loss": 2.5775,
+      "step": 31790
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00024334802749628452,
+      "loss": 2.595,
+      "step": 31800
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002433148121336546,
+      "loss": 2.5966,
+      "step": 31810
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00024328158930502332,
+      "loss": 2.5557,
+      "step": 31820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00024324835901304878,
+      "loss": 2.5976,
+      "step": 31830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002432151212603896,
+      "loss": 2.5887,
+      "step": 31840
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002431818760497053,
+      "loss": 2.5976,
+      "step": 31850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002431486233836556,
+      "loss": 2.5942,
+      "step": 31860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00024311536326490104,
+      "loss": 2.5816,
+      "step": 31870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00024308209569610278,
+      "loss": 2.5946,
+      "step": 31880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002430488206799225,
+      "loss": 2.596,
+      "step": 31890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00024301553821902248,
+      "loss": 2.6123,
+      "step": 31900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00024298224831606558,
+      "loss": 2.5744,
+      "step": 31910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002429489509737153,
+      "loss": 2.5838,
+      "step": 31920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00024291564619463577,
+      "loss": 2.5786,
+      "step": 31930
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00024288233398149162,
+      "loss": 2.55,
+      "step": 31940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002428490143369481,
+      "loss": 2.5815,
+      "step": 31950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002428156872636711,
+      "loss": 2.608,
+      "step": 31960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00024278235276432706,
+      "loss": 2.6063,
+      "step": 31970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00024274901084158306,
+      "loss": 2.6064,
+      "step": 31980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00024271566149810676,
+      "loss": 2.5783,
+      "step": 31990
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00024268230473656628,
+      "loss": 2.6064,
+      "step": 32000
+    },
+    {
+      "epoch": 2.88,
+      "eval_loss": 2.642347574234009,
+      "eval_runtime": 1263.5641,
+      "eval_samples_per_second": 60.254,
+      "eval_steps_per_second": 7.532,
+      "step": 32000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002426489405596306,
+      "loss": 2.6091,
+      "step": 32010
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00024261556896996907,
+      "loss": 2.5474,
+      "step": 32020
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00024258218997025174,
+      "loss": 2.5809,
+      "step": 32030
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00024254880356314922,
+      "loss": 2.6079,
+      "step": 32040
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00024251540975133267,
+      "loss": 2.5749,
+      "step": 32050
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00024248200853747392,
+      "loss": 2.5906,
+      "step": 32060
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00024244859992424538,
+      "loss": 2.5891,
+      "step": 32070
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00024241518391432,
+      "loss": 2.6299,
+      "step": 32080
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0002423817605103714,
+      "loss": 2.6047,
+      "step": 32090
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00024234832971507363,
+      "loss": 2.5702,
+      "step": 32100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0002423148915311016,
+      "loss": 2.6169,
+      "step": 32110
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00024228144596113058,
+      "loss": 2.5708,
+      "step": 32120
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00024224799300783645,
+      "loss": 2.5769,
+      "step": 32130
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00024221453267389583,
+      "loss": 2.6089,
+      "step": 32140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00024218106496198583,
+      "loss": 2.5697,
+      "step": 32150
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002421475898747841,
+      "loss": 2.5958,
+      "step": 32160
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00024211410741496902,
+      "loss": 2.5991,
+      "step": 32170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002420806175852194,
+      "loss": 2.5993,
+      "step": 32180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00024204712038821473,
+      "loss": 2.5959,
+      "step": 32190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00024201361582663514,
+      "loss": 2.5605,
+      "step": 32200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00024198010390316129,
+      "loss": 2.5905,
+      "step": 32210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002419465846204744,
+      "loss": 2.5889,
+      "step": 32220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002419130579812562,
+      "loss": 2.5931,
+      "step": 32230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00024187952398818928,
+      "loss": 2.5787,
+      "step": 32240
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00024184598264395655,
+      "loss": 2.5762,
+      "step": 32250
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002418124339512417,
+      "loss": 2.5672,
+      "step": 32260
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00024177887791272884,
+      "loss": 2.5687,
+      "step": 32270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00024174531453110274,
+      "loss": 2.6034,
+      "step": 32280
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00024171174380904884,
+      "loss": 2.622,
+      "step": 32290
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00024167816574925297,
+      "loss": 2.5955,
+      "step": 32300
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00024164458035440183,
+      "loss": 2.5779,
+      "step": 32310
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00024161098762718241,
+      "loss": 2.5157,
+      "step": 32320
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0002415773875702825,
+      "loss": 2.5865,
+      "step": 32330
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00024154378018639036,
+      "loss": 2.5942,
+      "step": 32340
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0002415101654781949,
+      "loss": 2.559,
+      "step": 32350
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00024147654344838557,
+      "loss": 2.5977,
+      "step": 32360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0002414429140996524,
+      "loss": 2.5687,
+      "step": 32370
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002414092774346861,
+      "loss": 2.5951,
+      "step": 32380
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024137563345617786,
+      "loss": 2.5944,
+      "step": 32390
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024134198216681945,
+      "loss": 2.5854,
+      "step": 32400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024130832356930338,
+      "loss": 2.5963,
+      "step": 32410
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024127465766632256,
+      "loss": 2.5891,
+      "step": 32420
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024124098446057053,
+      "loss": 2.5676,
+      "step": 32430
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024120730395474149,
+      "loss": 2.5837,
+      "step": 32440
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024117361615153015,
+      "loss": 2.5601,
+      "step": 32450
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002411399210536318,
+      "loss": 2.5968,
+      "step": 32460
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024110621866374243,
+      "loss": 2.6093,
+      "step": 32470
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00024107250898455846,
+      "loss": 2.5893,
+      "step": 32480
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024103879201877696,
+      "loss": 2.6069,
+      "step": 32490
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024100506776909565,
+      "loss": 2.5655,
+      "step": 32500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024097133623821263,
+      "loss": 2.5783,
+      "step": 32510
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024093759742882688,
+      "loss": 2.6044,
+      "step": 32520
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002409038513436377,
+      "loss": 2.5776,
+      "step": 32530
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024087009798534503,
+      "loss": 2.5765,
+      "step": 32540
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024083633735664956,
+      "loss": 2.6053,
+      "step": 32550
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024080256946025234,
+      "loss": 2.5465,
+      "step": 32560
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002407687942988552,
+      "loss": 2.5908,
+      "step": 32570
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00024073501187516026,
+      "loss": 2.6163,
+      "step": 32580
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002407012221918706,
+      "loss": 2.5756,
+      "step": 32590
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024066742525168962,
+      "loss": 2.5476,
+      "step": 32600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024063362105732135,
+      "loss": 2.5997,
+      "step": 32610
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024059980961147043,
+      "loss": 2.6063,
+      "step": 32620
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024056599091684208,
+      "loss": 2.5984,
+      "step": 32630
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0002405321649761421,
+      "loss": 2.544,
+      "step": 32640
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024049833179207687,
+      "loss": 2.5911,
+      "step": 32650
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0002404644913673533,
+      "loss": 2.5848,
+      "step": 32660
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024043064370467897,
+      "loss": 2.5677,
+      "step": 32670
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024039678880676194,
+      "loss": 2.6191,
+      "step": 32680
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024036292667631092,
+      "loss": 2.5772,
+      "step": 32690
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0002403290573160352,
+      "loss": 2.5997,
+      "step": 32700
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00024029518072864458,
+      "loss": 2.6296,
+      "step": 32710
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002402612969168495,
+      "loss": 2.6108,
+      "step": 32720
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.000240227405883361,
+      "loss": 2.547,
+      "step": 32730
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00024019350763089062,
+      "loss": 2.5928,
+      "step": 32740
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002401596021621505,
+      "loss": 2.5985,
+      "step": 32750
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00024012568947985341,
+      "loss": 2.5797,
+      "step": 32760
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00024009176958671264,
+      "loss": 2.5643,
+      "step": 32770
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002400578424854421,
+      "loss": 2.592,
+      "step": 32780
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00024002390817875625,
+      "loss": 2.5841,
+      "step": 32790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00023998996666937012,
+      "loss": 2.5868,
+      "step": 32800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00023995601795999934,
+      "loss": 2.5606,
+      "step": 32810
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00023992206205336003,
+      "loss": 2.5898,
+      "step": 32820
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00023988809895216908,
+      "loss": 2.6087,
+      "step": 32830
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00023985412865914376,
+      "loss": 2.5999,
+      "step": 32840
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00023982015117700206,
+      "loss": 2.5626,
+      "step": 32850
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002397861665084624,
+      "loss": 2.616,
+      "step": 32860
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00023975217465624387,
+      "loss": 2.5496,
+      "step": 32870
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002397181756230661,
+      "loss": 2.6027,
+      "step": 32880
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00023968416941164936,
+      "loss": 2.5701,
+      "step": 32890
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00023965015602471439,
+      "loss": 2.594,
+      "step": 32900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00023961613546498263,
+      "loss": 2.5971,
+      "step": 32910
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00023958210773517596,
+      "loss": 2.6047,
+      "step": 32920
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002395480728380169,
+      "loss": 2.5663,
+      "step": 32930
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00023951403077622857,
+      "loss": 2.5634,
+      "step": 32940
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00023947998155253463,
+      "loss": 2.5848,
+      "step": 32950
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00023944592516965928,
+      "loss": 2.5627,
+      "step": 32960
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00023941186163032734,
+      "loss": 2.5954,
+      "step": 32970
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00023937779093726418,
+      "loss": 2.5967,
+      "step": 32980
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00023934371309319584,
+      "loss": 2.6168,
+      "step": 32990
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002393096281008488,
+      "loss": 2.5836,
+      "step": 33000
+    },
+    {
+      "epoch": 2.97,
+      "eval_loss": 2.6340036392211914,
+      "eval_runtime": 1263.5015,
+      "eval_samples_per_second": 60.257,
+      "eval_steps_per_second": 7.532,
+      "step": 33000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00023927553596295005,
+      "loss": 2.6018,
+      "step": 33010
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002392414366822274,
+      "loss": 2.5834,
+      "step": 33020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000239207330261409,
+      "loss": 2.5973,
+      "step": 33030
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00023917321670322377,
+      "loss": 2.552,
+      "step": 33040
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00023913909601040097,
+      "loss": 2.5937,
+      "step": 33050
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002391083812890017,
+      "loss": 2.5922,
+      "step": 33060
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00023907424704788913,
+      "loss": 2.5641,
+      "step": 33070
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002390401056800575,
+      "loss": 2.5588,
+      "step": 33080
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002390059571882384,
+      "loss": 2.5351,
+      "step": 33090
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00023897180157516398,
+      "loss": 2.5733,
+      "step": 33100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00023893763884356702,
+      "loss": 2.5834,
+      "step": 33110
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00023890346899618087,
+      "loss": 2.578,
+      "step": 33120
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00023886929203573937,
+      "loss": 2.5722,
+      "step": 33130
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00023883510796497695,
+      "loss": 2.5668,
+      "step": 33140
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00023880091678662866,
+      "loss": 2.5644,
+      "step": 33150
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023876671850343018,
+      "loss": 2.5687,
+      "step": 33160
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023873251311811754,
+      "loss": 2.563,
+      "step": 33170
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023869830063342757,
+      "loss": 2.581,
+      "step": 33180
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023866408105209746,
+      "loss": 2.5884,
+      "step": 33190
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002386298543768652,
+      "loss": 2.6068,
+      "step": 33200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023859562061046916,
+      "loss": 2.6162,
+      "step": 33210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023856137975564835,
+      "loss": 2.5807,
+      "step": 33220
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023852713181514226,
+      "loss": 2.6039,
+      "step": 33230
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023849287679169118,
+      "loss": 2.5615,
+      "step": 33240
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023845861468803573,
+      "loss": 2.5629,
+      "step": 33250
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00023842434550691712,
+      "loss": 2.586,
+      "step": 33260
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00023839006925107728,
+      "loss": 2.5552,
+      "step": 33270
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00023835578592325854,
+      "loss": 2.5873,
+      "step": 33280
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00023832149552620394,
+      "loss": 2.574,
+      "step": 33290
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00023828719806265696,
+      "loss": 2.5856,
+      "step": 33300
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002382528935353617,
+      "loss": 2.59,
+      "step": 33310
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00023821858194706286,
+      "loss": 2.5742,
+      "step": 33320
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002381842633005056,
+      "loss": 2.5426,
+      "step": 33330
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00023814993759843574,
+      "loss": 2.5364,
+      "step": 33340
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00023811560484359964,
+      "loss": 2.5586,
+      "step": 33350
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002380812650387443,
+      "loss": 2.5815,
+      "step": 33360
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00023804691818661705,
+      "loss": 2.5356,
+      "step": 33370
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00023801256428996606,
+      "loss": 2.5734,
+      "step": 33380
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002379782033515399,
+      "loss": 2.5226,
+      "step": 33390
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002379438353740877,
+      "loss": 2.5352,
+      "step": 33400
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002379094603603593,
+      "loss": 2.5526,
+      "step": 33410
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00023787507831310494,
+      "loss": 2.5234,
+      "step": 33420
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00023784068923507547,
+      "loss": 2.5437,
+      "step": 33430
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002378062931290224,
+      "loss": 2.56,
+      "step": 33440
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002377718899976976,
+      "loss": 2.558,
+      "step": 33450
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00023773747984385372,
+      "loss": 2.5618,
+      "step": 33460
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002377030626702438,
+      "loss": 2.5509,
+      "step": 33470
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00023766863847962154,
+      "loss": 2.5826,
+      "step": 33480
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00023763420727474123,
+      "loss": 2.5489,
+      "step": 33490
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002375997690583576,
+      "loss": 2.5215,
+      "step": 33500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000237565323833226,
+      "loss": 2.5525,
+      "step": 33510
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00023753087160210243,
+      "loss": 2.5317,
+      "step": 33520
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00023749641236774332,
+      "loss": 2.5262,
+      "step": 33530
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002374619461329057,
+      "loss": 2.5435,
+      "step": 33540
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00023742747290034718,
+      "loss": 2.5591,
+      "step": 33550
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002373929926728259,
+      "loss": 2.5648,
+      "step": 33560
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00023735850545310064,
+      "loss": 2.5089,
+      "step": 33570
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00023732401124393064,
+      "loss": 2.5383,
+      "step": 33580
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00023728951004807573,
+      "loss": 2.5348,
+      "step": 33590
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002372550018682963,
+      "loss": 2.5509,
+      "step": 33600
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00023722048670735333,
+      "loss": 2.547,
+      "step": 33610
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00023718596456800834,
+      "loss": 2.5545,
+      "step": 33620
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002371514354530234,
+      "loss": 2.5499,
+      "step": 33630
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002371168993651611,
+      "loss": 2.5126,
+      "step": 33640
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002370823563071847,
+      "loss": 2.568,
+      "step": 33650
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00023704780628185787,
+      "loss": 2.5201,
+      "step": 33660
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000237013249291945,
+      "loss": 2.5413,
+      "step": 33670
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002369786853402109,
+      "loss": 2.5448,
+      "step": 33680
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00023694411442942096,
+      "loss": 2.5302,
+      "step": 33690
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002369095365623412,
+      "loss": 2.5534,
+      "step": 33700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002368749517417381,
+      "loss": 2.5015,
+      "step": 33710
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00023684035997037888,
+      "loss": 2.5512,
+      "step": 33720
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00023680576125103105,
+      "loss": 2.5209,
+      "step": 33730
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00023677115558646284,
+      "loss": 2.5557,
+      "step": 33740
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00023673654297944304,
+      "loss": 2.5462,
+      "step": 33750
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00023670192343274091,
+      "loss": 2.5728,
+      "step": 33760
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00023666729694912637,
+      "loss": 2.5405,
+      "step": 33770
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00023663266353136982,
+      "loss": 2.5235,
+      "step": 33780
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002365980231822422,
+      "loss": 2.5483,
+      "step": 33790
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00023656337590451516,
+      "loss": 2.536,
+      "step": 33800
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002365287217009606,
+      "loss": 2.5296,
+      "step": 33810
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00023649406057435135,
+      "loss": 2.5247,
+      "step": 33820
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023645939252746047,
+      "loss": 2.5626,
+      "step": 33830
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023642471756306178,
+      "loss": 2.4923,
+      "step": 33840
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023639003568392953,
+      "loss": 2.5507,
+      "step": 33850
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002363553468928386,
+      "loss": 2.5887,
+      "step": 33860
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023632065119256438,
+      "loss": 2.5568,
+      "step": 33870
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002362859485858829,
+      "loss": 2.5593,
+      "step": 33880
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023625123907557052,
+      "loss": 2.5635,
+      "step": 33890
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023621652266440446,
+      "loss": 2.5114,
+      "step": 33900
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023618179935516222,
+      "loss": 2.5288,
+      "step": 33910
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002361470691506221,
+      "loss": 2.5388,
+      "step": 33920
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00023611233205356272,
+      "loss": 2.5427,
+      "step": 33930
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023607758806676335,
+      "loss": 2.5431,
+      "step": 33940
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023604283719300386,
+      "loss": 2.5608,
+      "step": 33950
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0002360080794350646,
+      "loss": 2.5511,
+      "step": 33960
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023597331479572649,
+      "loss": 2.5647,
+      "step": 33970
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023593854327777104,
+      "loss": 2.5247,
+      "step": 33980
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023590376488398022,
+      "loss": 2.5677,
+      "step": 33990
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023586897961713665,
+      "loss": 2.5502,
+      "step": 34000
+    },
+    {
+      "epoch": 3.06,
+      "eval_loss": 2.6285486221313477,
+      "eval_runtime": 1264.0404,
+      "eval_samples_per_second": 60.231,
+      "eval_steps_per_second": 7.529,
+      "step": 34000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023583418748002345,
+      "loss": 2.5216,
+      "step": 34010
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023579938847542422,
+      "loss": 2.5229,
+      "step": 34020
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023576458260612333,
+      "loss": 2.5441,
+      "step": 34030
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00023572976987490542,
+      "loss": 2.5722,
+      "step": 34040
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023569495028455593,
+      "loss": 2.512,
+      "step": 34050
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023566012383786062,
+      "loss": 2.5504,
+      "step": 34060
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023562529053760596,
+      "loss": 2.5454,
+      "step": 34070
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023559045038657896,
+      "loss": 2.5383,
+      "step": 34080
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023555560338756702,
+      "loss": 2.5168,
+      "step": 34090
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023552074954335834,
+      "loss": 2.5625,
+      "step": 34100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023548588885674148,
+      "loss": 2.535,
+      "step": 34110
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023545102133050552,
+      "loss": 2.5493,
+      "step": 34120
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023541614696744032,
+      "loss": 2.5287,
+      "step": 34130
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023538126577033605,
+      "loss": 2.5455,
+      "step": 34140
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00023534637774198346,
+      "loss": 2.5233,
+      "step": 34150
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000235311482885174,
+      "loss": 2.5308,
+      "step": 34160
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00023527658120269947,
+      "loss": 2.5467,
+      "step": 34170
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002352416726973524,
+      "loss": 2.5395,
+      "step": 34180
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00023520675737192574,
+      "loss": 2.5124,
+      "step": 34190
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000235171835229213,
+      "loss": 2.5279,
+      "step": 34200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00023513690627200825,
+      "loss": 2.5435,
+      "step": 34210
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002351019705031061,
+      "loss": 2.5583,
+      "step": 34220
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00023506702792530182,
+      "loss": 2.5438,
+      "step": 34230
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000235032078541391,
+      "loss": 2.5558,
+      "step": 34240
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00023499712235416998,
+      "loss": 2.5611,
+      "step": 34250
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00023496215936643554,
+      "loss": 2.5523,
+      "step": 34260
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023492718958098498,
+      "loss": 2.4899,
+      "step": 34270
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023489221300061625,
+      "loss": 2.5226,
+      "step": 34280
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023485722962812778,
+      "loss": 2.5181,
+      "step": 34290
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023482223946631845,
+      "loss": 2.5117,
+      "step": 34300
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0002347872425179879,
+      "loss": 2.5721,
+      "step": 34310
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023475223878593614,
+      "loss": 2.5356,
+      "step": 34320
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023471722827296375,
+      "loss": 2.5278,
+      "step": 34330
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023468221098187192,
+      "loss": 2.5411,
+      "step": 34340
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023464718691546235,
+      "loss": 2.5635,
+      "step": 34350
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023461215607653723,
+      "loss": 2.5282,
+      "step": 34360
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00023457711846789934,
+      "loss": 2.5358,
+      "step": 34370
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023454207409235198,
+      "loss": 2.5755,
+      "step": 34380
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023450702295269905,
+      "loss": 2.534,
+      "step": 34390
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023447196505174494,
+      "loss": 2.5137,
+      "step": 34400
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023443690039229458,
+      "loss": 2.5622,
+      "step": 34410
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023440182897715346,
+      "loss": 2.5126,
+      "step": 34420
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023436675080912753,
+      "loss": 2.5401,
+      "step": 34430
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023433166589102348,
+      "loss": 2.5192,
+      "step": 34440
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023429657422564828,
+      "loss": 2.5395,
+      "step": 34450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023426147581580964,
+      "loss": 2.527,
+      "step": 34460
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023422637066431579,
+      "loss": 2.5319,
+      "step": 34470
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00023419125877397528,
+      "loss": 2.5467,
+      "step": 34480
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023415614014759756,
+      "loss": 2.5265,
+      "step": 34490
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023412101478799234,
+      "loss": 2.5488,
+      "step": 34500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023408588269796993,
+      "loss": 2.5501,
+      "step": 34510
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023405074388034125,
+      "loss": 2.5369,
+      "step": 34520
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023401559833791767,
+      "loss": 2.56,
+      "step": 34530
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002339804460735112,
+      "loss": 2.5707,
+      "step": 34540
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002339452870899343,
+      "loss": 2.5314,
+      "step": 34550
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023391012138999996,
+      "loss": 2.5128,
+      "step": 34560
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023387494897652178,
+      "loss": 2.5311,
+      "step": 34570
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023383976985231385,
+      "loss": 2.5542,
+      "step": 34580
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00023380458402019085,
+      "loss": 2.5494,
+      "step": 34590
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023376939148296791,
+      "loss": 2.516,
+      "step": 34600
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023373419224346072,
+      "loss": 2.5212,
+      "step": 34610
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023369898630448563,
+      "loss": 2.5228,
+      "step": 34620
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023366377366885929,
+      "loss": 2.5595,
+      "step": 34630
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023362855433939906,
+      "loss": 2.545,
+      "step": 34640
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0002335933283189229,
+      "loss": 2.4774,
+      "step": 34650
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.000233558095610249,
+      "loss": 2.5454,
+      "step": 34660
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0002335228562161965,
+      "loss": 2.5405,
+      "step": 34670
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0002334876101395847,
+      "loss": 2.5199,
+      "step": 34680
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023345235738323367,
+      "loss": 2.5155,
+      "step": 34690
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023341709794996393,
+      "loss": 2.5275,
+      "step": 34700
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00023338183184259652,
+      "loss": 2.5411,
+      "step": 34710
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002333465590639531,
+      "loss": 2.5147,
+      "step": 34720
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002333112796168557,
+      "loss": 2.5397,
+      "step": 34730
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00023327599350412702,
+      "loss": 2.5133,
+      "step": 34740
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00023324070072859032,
+      "loss": 2.526,
+      "step": 34750
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00023320540129306928,
+      "loss": 2.5372,
+      "step": 34760
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00023317009520038815,
+      "loss": 2.5378,
+      "step": 34770
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00023313478245337178,
+      "loss": 2.5412,
+      "step": 34780
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002330994630548454,
+      "loss": 2.5624,
+      "step": 34790
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00023306413700763503,
+      "loss": 2.5344,
+      "step": 34800
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00023302880431456689,
+      "loss": 2.537,
+      "step": 34810
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00023299346497846802,
+      "loss": 2.5211,
+      "step": 34820
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002329581190021658,
+      "loss": 2.5372,
+      "step": 34830
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023292276638848833,
+      "loss": 2.5255,
+      "step": 34840
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.000232887407140264,
+      "loss": 2.5642,
+      "step": 34850
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023285204126032193,
+      "loss": 2.542,
+      "step": 34860
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023281666875149169,
+      "loss": 2.518,
+      "step": 34870
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023278128961660337,
+      "loss": 2.5308,
+      "step": 34880
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023274590385848762,
+      "loss": 2.5425,
+      "step": 34890
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023271051147997563,
+      "loss": 2.525,
+      "step": 34900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002326751124838991,
+      "loss": 2.5184,
+      "step": 34910
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002326397068730902,
+      "loss": 2.5247,
+      "step": 34920
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00023260429465038176,
+      "loss": 2.5398,
+      "step": 34930
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00023256887581860704,
+      "loss": 2.5301,
+      "step": 34940
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00023253345038059986,
+      "loss": 2.5186,
+      "step": 34950
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00023249801833919454,
+      "loss": 2.5567,
+      "step": 34960
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.000232462579697226,
+      "loss": 2.5246,
+      "step": 34970
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00023242713445752959,
+      "loss": 2.5448,
+      "step": 34980
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00023239168262294128,
+      "loss": 2.5183,
+      "step": 34990
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00023235622419629754,
+      "loss": 2.5428,
+      "step": 35000
+    },
+    {
+      "epoch": 3.15,
+      "eval_loss": 2.6218130588531494,
+      "eval_runtime": 1264.586,
+      "eval_samples_per_second": 60.205,
+      "eval_steps_per_second": 7.526,
+      "step": 35000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00023232075918043528,
+      "loss": 2.5275,
+      "step": 35010
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00023228528757819207,
+      "loss": 2.5427,
+      "step": 35020
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00023224980939240593,
+      "loss": 2.5521,
+      "step": 35030
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0002322143246259155,
+      "loss": 2.5646,
+      "step": 35040
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00023217883328155973,
+      "loss": 2.5704,
+      "step": 35050
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002321433353621783,
+      "loss": 2.538,
+      "step": 35060
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002321078308706114,
+      "loss": 2.5549,
+      "step": 35070
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00023207231980969972,
+      "loss": 2.5607,
+      "step": 35080
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00023203680218228436,
+      "loss": 2.5849,
+      "step": 35090
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00023200127799120703,
+      "loss": 2.5454,
+      "step": 35100
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00023196574723931008,
+      "loss": 2.5439,
+      "step": 35110
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00023193020992943626,
+      "loss": 2.5474,
+      "step": 35120
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00023189466606442875,
+      "loss": 2.5501,
+      "step": 35130
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00023185911564713152,
+      "loss": 2.5448,
+      "step": 35140
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00023182355868038882,
+      "loss": 2.5578,
+      "step": 35150
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002317879951670456,
+      "loss": 2.5471,
+      "step": 35160
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00023175242510994716,
+      "loss": 2.5597,
+      "step": 35170
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00023171684851193945,
+      "loss": 2.5531,
+      "step": 35180
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00023168126537586892,
+      "loss": 2.5087,
+      "step": 35190
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002316456757045825,
+      "loss": 2.5265,
+      "step": 35200
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00023161007950092775,
+      "loss": 2.5042,
+      "step": 35210
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00023157447676775263,
+      "loss": 2.5731,
+      "step": 35220
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002315388675079057,
+      "loss": 2.5273,
+      "step": 35230
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00023150325172423595,
+      "loss": 2.5614,
+      "step": 35240
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00023146762941959295,
+      "loss": 2.5716,
+      "step": 35250
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00023143200059682687,
+      "loss": 2.5572,
+      "step": 35260
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00023139636525878825,
+      "loss": 2.5358,
+      "step": 35270
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002313607234083283,
+      "loss": 2.5795,
+      "step": 35280
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00023132507504829868,
+      "loss": 2.5475,
+      "step": 35290
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002312894201815515,
+      "loss": 2.5487,
+      "step": 35300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00023125375881093951,
+      "loss": 2.567,
+      "step": 35310
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00023121809093931597,
+      "loss": 2.5532,
+      "step": 35320
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002311824165695345,
+      "loss": 2.5272,
+      "step": 35330
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002311467357044495,
+      "loss": 2.5713,
+      "step": 35340
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00023111104834691565,
+      "loss": 2.5536,
+      "step": 35350
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00023107535449978827,
+      "loss": 2.5198,
+      "step": 35360
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00023103965416592328,
+      "loss": 2.5366,
+      "step": 35370
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00023100394734817687,
+      "loss": 2.5221,
+      "step": 35380
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00023096823404940596,
+      "loss": 2.5396,
+      "step": 35390
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00023093251427246794,
+      "loss": 2.548,
+      "step": 35400
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002308967880202207,
+      "loss": 2.565,
+      "step": 35410
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002308610552955227,
+      "loss": 2.5749,
+      "step": 35420
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002308253161012328,
+      "loss": 2.5329,
+      "step": 35430
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002307895704402104,
+      "loss": 2.5557,
+      "step": 35440
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00023075381831531562,
+      "loss": 2.5543,
+      "step": 35450
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002307180597294088,
+      "loss": 2.5304,
+      "step": 35460
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00023068229468535103,
+      "loss": 2.515,
+      "step": 35470
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00023064652318600378,
+      "loss": 2.561,
+      "step": 35480
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023061074523422908,
+      "loss": 2.5396,
+      "step": 35490
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002305749608328896,
+      "loss": 2.5518,
+      "step": 35500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023053916998484815,
+      "loss": 2.5428,
+      "step": 35510
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023050337269296854,
+      "loss": 2.5334,
+      "step": 35520
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023046756896011477,
+      "loss": 2.5552,
+      "step": 35530
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023043175878915146,
+      "loss": 2.5364,
+      "step": 35540
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023039594218294374,
+      "loss": 2.5359,
+      "step": 35550
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023036011914435731,
+      "loss": 2.5275,
+      "step": 35560
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023032428967625823,
+      "loss": 2.5395,
+      "step": 35570
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023028845378151325,
+      "loss": 2.5493,
+      "step": 35580
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00023025261146298944,
+      "loss": 2.5814,
+      "step": 35590
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002302167627235547,
+      "loss": 2.5242,
+      "step": 35600
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00023018090756607705,
+      "loss": 2.5616,
+      "step": 35610
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002301450459934253,
+      "loss": 2.5647,
+      "step": 35620
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002301091780084687,
+      "loss": 2.5564,
+      "step": 35630
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00023007330361407696,
+      "loss": 2.5165,
+      "step": 35640
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002300374228131204,
+      "loss": 2.5354,
+      "step": 35650
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00023000153560846974,
+      "loss": 2.5221,
+      "step": 35660
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022996564200299625,
+      "loss": 2.5183,
+      "step": 35670
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002299297419995718,
+      "loss": 2.581,
+      "step": 35680
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022989383560106873,
+      "loss": 2.5515,
+      "step": 35690
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002298579228103598,
+      "loss": 2.5513,
+      "step": 35700
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022982200363031837,
+      "loss": 2.5525,
+      "step": 35710
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022978607806381824,
+      "loss": 2.5537,
+      "step": 35720
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022975014611373382,
+      "loss": 2.5372,
+      "step": 35730
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022971420778294005,
+      "loss": 2.5351,
+      "step": 35740
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022967826307431215,
+      "loss": 2.5358,
+      "step": 35750
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022964231199072612,
+      "loss": 2.5402,
+      "step": 35760
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022960635453505836,
+      "loss": 2.5422,
+      "step": 35770
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0002295703907101858,
+      "loss": 2.5527,
+      "step": 35780
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022953442051898575,
+      "loss": 2.5417,
+      "step": 35790
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022949844396433623,
+      "loss": 2.5558,
+      "step": 35800
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022946246104911565,
+      "loss": 2.5573,
+      "step": 35810
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022942647177620292,
+      "loss": 2.5627,
+      "step": 35820
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002293904761484776,
+      "loss": 2.5196,
+      "step": 35830
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002293544741688196,
+      "loss": 2.5398,
+      "step": 35840
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022931846584010938,
+      "loss": 2.5565,
+      "step": 35850
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022928245116522796,
+      "loss": 2.5483,
+      "step": 35860
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002292464301470568,
+      "loss": 2.5339,
+      "step": 35870
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022921040278847782,
+      "loss": 2.5439,
+      "step": 35880
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022917436909237367,
+      "loss": 2.5675,
+      "step": 35890
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002291383290616273,
+      "loss": 2.5568,
+      "step": 35900
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022910228269912218,
+      "loss": 2.5317,
+      "step": 35910
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022906623000774242,
+      "loss": 2.5478,
+      "step": 35920
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002290301709903725,
+      "loss": 2.5251,
+      "step": 35930
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00022899410564989746,
+      "loss": 2.5906,
+      "step": 35940
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002289580339892028,
+      "loss": 2.5211,
+      "step": 35950
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00022892195601117467,
+      "loss": 2.5447,
+      "step": 35960
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00022888587171869957,
+      "loss": 2.5596,
+      "step": 35970
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002288497811146645,
+      "loss": 2.5121,
+      "step": 35980
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00022881368420195713,
+      "loss": 2.5419,
+      "step": 35990
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002287775809834655,
+      "loss": 2.5342,
+      "step": 36000
+    },
+    {
+      "epoch": 3.24,
+      "eval_loss": 2.616030216217041,
+      "eval_runtime": 1263.6218,
+      "eval_samples_per_second": 60.251,
+      "eval_steps_per_second": 7.532,
+      "step": 36000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002287414714620781,
+      "loss": 2.5282,
+      "step": 36010
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002287053556406841,
+      "loss": 2.5521,
+      "step": 36020
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00022866923352217304,
+      "loss": 2.5577,
+      "step": 36030
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00022863310510943505,
+      "loss": 2.5327,
+      "step": 36040
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002285969704053607,
+      "loss": 2.5428,
+      "step": 36050
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.000228560829412841,
+      "loss": 2.5665,
+      "step": 36060
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002285246821347677,
+      "loss": 2.5757,
+      "step": 36070
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002284885285740328,
+      "loss": 2.5669,
+      "step": 36080
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00022845236873352885,
+      "loss": 2.5326,
+      "step": 36090
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002284162026161491,
+      "loss": 2.5317,
+      "step": 36100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.000228380030224787,
+      "loss": 2.5303,
+      "step": 36110
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00022834385156233678,
+      "loss": 2.5386,
+      "step": 36120
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00022830766663169298,
+      "loss": 2.5501,
+      "step": 36130
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00022827147543575072,
+      "loss": 2.538,
+      "step": 36140
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002282352779774057,
+      "loss": 2.5529,
+      "step": 36150
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022819907425955386,
+      "loss": 2.5418,
+      "step": 36160
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022816286428509194,
+      "loss": 2.5329,
+      "step": 36170
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022812664805691707,
+      "loss": 2.5744,
+      "step": 36180
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022809042557792681,
+      "loss": 2.5469,
+      "step": 36190
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022805419685101926,
+      "loss": 2.4985,
+      "step": 36200
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022801796187909307,
+      "loss": 2.5499,
+      "step": 36210
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022798172066504737,
+      "loss": 2.5391,
+      "step": 36220
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022794547321178176,
+      "loss": 2.5202,
+      "step": 36230
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022790921952219634,
+      "loss": 2.5254,
+      "step": 36240
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00022787295959919172,
+      "loss": 2.5388,
+      "step": 36250
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.000227836693445669,
+      "loss": 2.5534,
+      "step": 36260
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00022780042106452982,
+      "loss": 2.5483,
+      "step": 36270
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0002277641424586763,
+      "loss": 2.5301,
+      "step": 36280
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00022772785763101094,
+      "loss": 2.5116,
+      "step": 36290
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.000227691566584437,
+      "loss": 2.5393,
+      "step": 36300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.000227655269321858,
+      "loss": 2.5193,
+      "step": 36310
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00022761896584617803,
+      "loss": 2.5059,
+      "step": 36320
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00022758265616030173,
+      "loss": 2.5739,
+      "step": 36330
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00022754634026713415,
+      "loss": 2.4966,
+      "step": 36340
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00022751001816958088,
+      "loss": 2.5305,
+      "step": 36350
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.000227473689870548,
+      "loss": 2.5277,
+      "step": 36360
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00022743735537294216,
+      "loss": 2.539,
+      "step": 36370
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002274010146796704,
+      "loss": 2.5431,
+      "step": 36380
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00022736466779364024,
+      "loss": 2.5321,
+      "step": 36390
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002273283147177598,
+      "loss": 2.5451,
+      "step": 36400
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00022729195545493762,
+      "loss": 2.532,
+      "step": 36410
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002272555900080828,
+      "loss": 2.5405,
+      "step": 36420
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00022721921838010487,
+      "loss": 2.5142,
+      "step": 36430
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00022718284057391384,
+      "loss": 2.5479,
+      "step": 36440
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00022714645659242032,
+      "loss": 2.528,
+      "step": 36450
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002271100664385353,
+      "loss": 2.5443,
+      "step": 36460
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00022707367011517033,
+      "loss": 2.5606,
+      "step": 36470
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00022703726762523743,
+      "loss": 2.5323,
+      "step": 36480
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00022700085897164911,
+      "loss": 2.5208,
+      "step": 36490
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002269644441573184,
+      "loss": 2.5404,
+      "step": 36500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002269280231851588,
+      "loss": 2.5107,
+      "step": 36510
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002268915960580843,
+      "loss": 2.547,
+      "step": 36520
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002268551627790094,
+      "loss": 2.5482,
+      "step": 36530
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00022681872335084905,
+      "loss": 2.5515,
+      "step": 36540
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00022678227777651872,
+      "loss": 2.5579,
+      "step": 36550
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00022674582605893446,
+      "loss": 2.5494,
+      "step": 36560
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002267093682010126,
+      "loss": 2.5287,
+      "step": 36570
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00022667290420567014,
+      "loss": 2.5273,
+      "step": 36580
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002266364340758246,
+      "loss": 2.5543,
+      "step": 36590
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002265999578143938,
+      "loss": 2.5365,
+      "step": 36600
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00022656347542429622,
+      "loss": 2.4881,
+      "step": 36610
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00022652698690845075,
+      "loss": 2.4872,
+      "step": 36620
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00022649049226977676,
+      "loss": 2.554,
+      "step": 36630
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002264539915111942,
+      "loss": 2.5258,
+      "step": 36640
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00022641748463562338,
+      "loss": 2.5765,
+      "step": 36650
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00022638097164598523,
+      "loss": 2.5476,
+      "step": 36660
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00022634445254520111,
+      "loss": 2.5109,
+      "step": 36670
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00022630792733619286,
+      "loss": 2.5464,
+      "step": 36680
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00022627139602188274,
+      "loss": 2.5503,
+      "step": 36690
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002262348586051937,
+      "loss": 2.5309,
+      "step": 36700
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022619831508904895,
+      "loss": 2.539,
+      "step": 36710
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022616176547637234,
+      "loss": 2.554,
+      "step": 36720
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022612520977008813,
+      "loss": 2.5272,
+      "step": 36730
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022608864797312114,
+      "loss": 2.5368,
+      "step": 36740
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0002260520800883966,
+      "loss": 2.5399,
+      "step": 36750
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022601550611884026,
+      "loss": 2.5721,
+      "step": 36760
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022597892606737843,
+      "loss": 2.5599,
+      "step": 36770
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022594233993693773,
+      "loss": 2.5384,
+      "step": 36780
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022590574773044542,
+      "loss": 2.5423,
+      "step": 36790
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0002258691494508292,
+      "loss": 2.5126,
+      "step": 36800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022583254510101722,
+      "loss": 2.5431,
+      "step": 36810
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00022579593468393817,
+      "loss": 2.5154,
+      "step": 36820
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00022575931820252124,
+      "loss": 2.5326,
+      "step": 36830
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00022572269565969595,
+      "loss": 2.5461,
+      "step": 36840
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0002256860670583926,
+      "loss": 2.5112,
+      "step": 36850
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00022564943240154168,
+      "loss": 2.5306,
+      "step": 36860
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0002256127916920743,
+      "loss": 2.5709,
+      "step": 36870
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000225576144932922,
+      "loss": 2.5095,
+      "step": 36880
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00022553949212701692,
+      "loss": 2.5453,
+      "step": 36890
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00022550283327729163,
+      "loss": 2.5364,
+      "step": 36900
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00022546616838667904,
+      "loss": 2.5365,
+      "step": 36910
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00022542949745811273,
+      "loss": 2.5196,
+      "step": 36920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00022539282049452672,
+      "loss": 2.5284,
+      "step": 36930
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002253561374988554,
+      "loss": 2.5452,
+      "step": 36940
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00022531944847403382,
+      "loss": 2.5518,
+      "step": 36950
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00022528275342299737,
+      "loss": 2.5096,
+      "step": 36960
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00022524605234868205,
+      "loss": 2.5242,
+      "step": 36970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002252093452540242,
+      "loss": 2.519,
+      "step": 36980
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00022517263214196065,
+      "loss": 2.4987,
+      "step": 36990
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002251359130154289,
+      "loss": 2.5152,
+      "step": 37000
+    },
+    {
+      "epoch": 3.33,
+      "eval_loss": 2.6090149879455566,
+      "eval_runtime": 1262.9896,
+      "eval_samples_per_second": 60.282,
+      "eval_steps_per_second": 7.535,
+      "step": 37000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00022508695265669482,
+      "loss": 2.5202,
+      "step": 37010
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00022505021620121524,
+      "loss": 2.497,
+      "step": 37020
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0002250134737399803,
+      "loss": 2.5073,
+      "step": 37030
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00022497672527593035,
+      "loss": 2.4808,
+      "step": 37040
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00022493997081200596,
+      "loss": 2.5143,
+      "step": 37050
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00022490321035114843,
+      "loss": 2.5039,
+      "step": 37060
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002248664438962995,
+      "loss": 2.5001,
+      "step": 37070
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00022482967145040123,
+      "loss": 2.4968,
+      "step": 37080
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00022479289301639632,
+      "loss": 2.5072,
+      "step": 37090
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00022475610859722786,
+      "loss": 2.4914,
+      "step": 37100
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00022471931819583948,
+      "loss": 2.4979,
+      "step": 37110
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00022468252181517527,
+      "loss": 2.4905,
+      "step": 37120
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002246457194581798,
+      "loss": 2.5321,
+      "step": 37130
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00022460891112779807,
+      "loss": 2.4953,
+      "step": 37140
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00022457209682697564,
+      "loss": 2.5037,
+      "step": 37150
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002245352765586585,
+      "loss": 2.4955,
+      "step": 37160
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002244984503257931,
+      "loss": 2.4816,
+      "step": 37170
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00022446161813132644,
+      "loss": 2.4784,
+      "step": 37180
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002244247799782059,
+      "loss": 2.5061,
+      "step": 37190
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002243879358693794,
+      "loss": 2.494,
+      "step": 37200
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00022435108580779535,
+      "loss": 2.5103,
+      "step": 37210
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00022431422979640256,
+      "loss": 2.4721,
+      "step": 37220
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00022427736783815037,
+      "loss": 2.4896,
+      "step": 37230
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00022424049993598867,
+      "loss": 2.4757,
+      "step": 37240
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00022420362609286765,
+      "loss": 2.4791,
+      "step": 37250
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00022416674631173806,
+      "loss": 2.4785,
+      "step": 37260
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00022412986059555122,
+      "loss": 2.4892,
+      "step": 37270
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00022409296894725884,
+      "loss": 2.4816,
+      "step": 37280
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00022405607136981298,
+      "loss": 2.4703,
+      "step": 37290
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002240191678661665,
+      "loss": 2.489,
+      "step": 37300
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00022398225843927235,
+      "loss": 2.4818,
+      "step": 37310
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002239453430920842,
+      "loss": 2.4642,
+      "step": 37320
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00022390842182755616,
+      "loss": 2.4782,
+      "step": 37330
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00022387149464864282,
+      "loss": 2.4777,
+      "step": 37340
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00022383456155829914,
+      "loss": 2.4977,
+      "step": 37350
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00022379762255948064,
+      "loss": 2.4931,
+      "step": 37360
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00022376067765514325,
+      "loss": 2.4828,
+      "step": 37370
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00022372372684824357,
+      "loss": 2.4579,
+      "step": 37380
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00022368677014173835,
+      "loss": 2.4708,
+      "step": 37390
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00022364980753858505,
+      "loss": 2.4915,
+      "step": 37400
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00022361283904174156,
+      "loss": 2.4878,
+      "step": 37410
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00022357586465416616,
+      "loss": 2.4729,
+      "step": 37420
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002235388843788177,
+      "loss": 2.4556,
+      "step": 37430
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00022350189821865551,
+      "loss": 2.4587,
+      "step": 37440
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002234649061766392,
+      "loss": 2.513,
+      "step": 37450
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002234279082557291,
+      "loss": 2.467,
+      "step": 37460
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002233909044588858,
+      "loss": 2.4558,
+      "step": 37470
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002233538947890706,
+      "loss": 2.4901,
+      "step": 37480
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00022331687924924504,
+      "loss": 2.4741,
+      "step": 37490
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00022327985784237128,
+      "loss": 2.4623,
+      "step": 37500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002232428305714118,
+      "loss": 2.4857,
+      "step": 37510
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00022320579743932969,
+      "loss": 2.498,
+      "step": 37520
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00022316875844908843,
+      "loss": 2.4922,
+      "step": 37530
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00022313171360365208,
+      "loss": 2.4896,
+      "step": 37540
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00022309466290598498,
+      "loss": 2.491,
+      "step": 37550
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002230576063590521,
+      "loss": 2.4634,
+      "step": 37560
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00022302054396581883,
+      "loss": 2.5019,
+      "step": 37570
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00022298347572925103,
+      "loss": 2.5329,
+      "step": 37580
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00022294640165231497,
+      "loss": 2.4792,
+      "step": 37590
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00022290932173797744,
+      "loss": 2.4849,
+      "step": 37600
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002228722359892057,
+      "loss": 2.4949,
+      "step": 37610
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002228351444089675,
+      "loss": 2.4966,
+      "step": 37620
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00022279804700023098,
+      "loss": 2.4837,
+      "step": 37630
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00022276094376596483,
+      "loss": 2.4818,
+      "step": 37640
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00022272383470913813,
+      "loss": 2.4963,
+      "step": 37650
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00022268671983272052,
+      "loss": 2.4965,
+      "step": 37660
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00022264959913968196,
+      "loss": 2.4721,
+      "step": 37670
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002226124726329931,
+      "loss": 2.4883,
+      "step": 37680
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002225753403156248,
+      "loss": 2.4927,
+      "step": 37690
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00022253820219054855,
+      "loss": 2.4988,
+      "step": 37700
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00022250105826073622,
+      "loss": 2.4758,
+      "step": 37710
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00022246390852916024,
+      "loss": 2.4772,
+      "step": 37720
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00022242675299879346,
+      "loss": 2.4828,
+      "step": 37730
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00022238959167260913,
+      "loss": 2.4912,
+      "step": 37740
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00022235242455358105,
+      "loss": 2.4839,
+      "step": 37750
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002223152516446834,
+      "loss": 2.4917,
+      "step": 37760
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00022227807294889095,
+      "loss": 2.4803,
+      "step": 37770
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00022224088846917886,
+      "loss": 2.4768,
+      "step": 37780
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00022220369820852263,
+      "loss": 2.4834,
+      "step": 37790
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00022216650216989845,
+      "loss": 2.5038,
+      "step": 37800
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00022212930035628287,
+      "loss": 2.4667,
+      "step": 37810
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002220920927706528,
+      "loss": 2.4742,
+      "step": 37820
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002220548794159858,
+      "loss": 2.5051,
+      "step": 37830
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002220176602952598,
+      "loss": 2.4736,
+      "step": 37840
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002219804354114531,
+      "loss": 2.4624,
+      "step": 37850
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00022194320476754466,
+      "loss": 2.4914,
+      "step": 37860
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00022190596836651374,
+      "loss": 2.4565,
+      "step": 37870
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002218687262113401,
+      "loss": 2.481,
+      "step": 37880
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00022183147830500403,
+      "loss": 2.4866,
+      "step": 37890
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002217942246504862,
+      "loss": 2.4902,
+      "step": 37900
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00022175696525076772,
+      "loss": 2.5325,
+      "step": 37910
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00022171970010883028,
+      "loss": 2.4894,
+      "step": 37920
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00022168242922765587,
+      "loss": 2.4837,
+      "step": 37930
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00022164515261022713,
+      "loss": 2.4668,
+      "step": 37940
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00022160787025952698,
+      "loss": 2.4487,
+      "step": 37950
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00022157058217853886,
+      "loss": 2.5002,
+      "step": 37960
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002215332883702467,
+      "loss": 2.4839,
+      "step": 37970
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002214959888376349,
+      "loss": 2.5165,
+      "step": 37980
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00022145868358368828,
+      "loss": 2.4817,
+      "step": 37990
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002214213726113921,
+      "loss": 2.5138,
+      "step": 38000
+    },
+    {
+      "epoch": 5.13,
+      "eval_loss": 2.5765559673309326,
+      "eval_runtime": 1269.7704,
+      "eval_samples_per_second": 59.96,
+      "eval_steps_per_second": 4.997,
+      "step": 38000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00022138405592373204,
+      "loss": 2.4842,
+      "step": 38010
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002213467335236944,
+      "loss": 2.499,
+      "step": 38020
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00022130940541426582,
+      "loss": 2.4801,
+      "step": 38030
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00022127207159843338,
+      "loss": 2.4645,
+      "step": 38040
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00022123473207918468,
+      "loss": 2.4837,
+      "step": 38050
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00022119738685950772,
+      "loss": 2.4809,
+      "step": 38060
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.000221160035942391,
+      "loss": 2.4883,
+      "step": 38070
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00022112267933082345,
+      "loss": 2.5176,
+      "step": 38080
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00022108531702779447,
+      "loss": 2.5117,
+      "step": 38090
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00022104794903629393,
+      "loss": 2.4763,
+      "step": 38100
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00022101057535931206,
+      "loss": 2.5114,
+      "step": 38110
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00022097319599983966,
+      "loss": 2.4798,
+      "step": 38120
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.000220935810960868,
+      "loss": 2.4528,
+      "step": 38130
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00022089842024538867,
+      "loss": 2.4756,
+      "step": 38140
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00022086102385639384,
+      "loss": 2.4838,
+      "step": 38150
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00022082362179687612,
+      "loss": 2.4891,
+      "step": 38160
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00022078621406982838,
+      "loss": 2.494,
+      "step": 38170
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002207488006782443,
+      "loss": 2.487,
+      "step": 38180
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00022071138162511775,
+      "loss": 2.4666,
+      "step": 38190
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002206739569134431,
+      "loss": 2.4851,
+      "step": 38200
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00022063652654621515,
+      "loss": 2.4879,
+      "step": 38210
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00022059909052642927,
+      "loss": 2.466,
+      "step": 38220
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00022056164885708121,
+      "loss": 2.4934,
+      "step": 38230
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00022052420154116716,
+      "loss": 2.5006,
+      "step": 38240
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002204867485816837,
+      "loss": 2.4816,
+      "step": 38250
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00022044928998162804,
+      "loss": 2.5133,
+      "step": 38260
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00022041182574399765,
+      "loss": 2.4763,
+      "step": 38270
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00022037435587179063,
+      "loss": 2.4742,
+      "step": 38280
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00022033688036800534,
+      "loss": 2.5107,
+      "step": 38290
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00022029939923564078,
+      "loss": 2.499,
+      "step": 38300
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00022026191247769618,
+      "loss": 2.4868,
+      "step": 38310
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00022022442009717144,
+      "loss": 2.5135,
+      "step": 38320
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00022018692209706684,
+      "loss": 2.4816,
+      "step": 38330
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002201494184803831,
+      "loss": 2.4928,
+      "step": 38340
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00022011190925012126,
+      "loss": 2.499,
+      "step": 38350
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00022007439440928305,
+      "loss": 2.4898,
+      "step": 38360
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00022003687396087042,
+      "loss": 2.487,
+      "step": 38370
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00021999934790788595,
+      "loss": 2.4823,
+      "step": 38380
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002199618162533326,
+      "loss": 2.486,
+      "step": 38390
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00021992427900021378,
+      "loss": 2.457,
+      "step": 38400
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002198867361515332,
+      "loss": 2.5269,
+      "step": 38410
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00021984918771029532,
+      "loss": 2.4705,
+      "step": 38420
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00021981163367950482,
+      "loss": 2.4646,
+      "step": 38430
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00021977407406216696,
+      "loss": 2.4876,
+      "step": 38440
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00021973650886128725,
+      "loss": 2.4861,
+      "step": 38450
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00021969893807987186,
+      "loss": 2.4821,
+      "step": 38460
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00021966136172092728,
+      "loss": 2.5098,
+      "step": 38470
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00021962377978746058,
+      "loss": 2.4809,
+      "step": 38480
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00021958619228247905,
+      "loss": 2.5004,
+      "step": 38490
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00021954859920899074,
+      "loss": 2.4562,
+      "step": 38500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00021951100057000377,
+      "loss": 2.492,
+      "step": 38510
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00021947339636852698,
+      "loss": 2.5046,
+      "step": 38520
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002194357866075696,
+      "loss": 2.4737,
+      "step": 38530
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00021939817129014132,
+      "loss": 2.5025,
+      "step": 38540
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00021936055041925216,
+      "loss": 2.4868,
+      "step": 38550
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00021932292399791267,
+      "loss": 2.531,
+      "step": 38560
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00021928529202913386,
+      "loss": 2.4976,
+      "step": 38570
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00021924765451592716,
+      "loss": 2.4981,
+      "step": 38580
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00021921001146130438,
+      "loss": 2.5074,
+      "step": 38590
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.000219172362868278,
+      "loss": 2.4963,
+      "step": 38600
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002191347087398606,
+      "loss": 2.496,
+      "step": 38610
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00021909704907906543,
+      "loss": 2.4693,
+      "step": 38620
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00021905938388890617,
+      "loss": 2.481,
+      "step": 38630
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002190217131723969,
+      "loss": 2.4872,
+      "step": 38640
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00021898403693255215,
+      "loss": 2.4845,
+      "step": 38650
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00021894635517238692,
+      "loss": 2.4909,
+      "step": 38660
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00021890866789491654,
+      "loss": 2.4818,
+      "step": 38670
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00021887097510315693,
+      "loss": 2.4625,
+      "step": 38680
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00021883327680012436,
+      "loss": 2.4324,
+      "step": 38690
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002187955729888356,
+      "loss": 2.4801,
+      "step": 38700
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00021875786367230778,
+      "loss": 2.4891,
+      "step": 38710
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00021872014885355854,
+      "loss": 2.4768,
+      "step": 38720
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00021868242853560594,
+      "loss": 2.5122,
+      "step": 38730
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00021864470272146857,
+      "loss": 2.4826,
+      "step": 38740
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00021860697141416516,
+      "loss": 2.4975,
+      "step": 38750
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002185692346167153,
+      "loss": 2.4608,
+      "step": 38760
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00021853149233213865,
+      "loss": 2.4986,
+      "step": 38770
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00021849374456345556,
+      "loss": 2.4931,
+      "step": 38780
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002184559913136867,
+      "loss": 2.4647,
+      "step": 38790
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00021841823258585323,
+      "loss": 2.4797,
+      "step": 38800
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00021838046838297664,
+      "loss": 2.4908,
+      "step": 38810
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00021834269870807906,
+      "loss": 2.487,
+      "step": 38820
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00021830492356418282,
+      "loss": 2.477,
+      "step": 38830
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002182671429543109,
+      "loss": 2.498,
+      "step": 38840
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00021822935688148656,
+      "loss": 2.481,
+      "step": 38850
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00021819156534873363,
+      "loss": 2.4582,
+      "step": 38860
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002181537683590762,
+      "loss": 2.4622,
+      "step": 38870
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.000218115965915539,
+      "loss": 2.4994,
+      "step": 38880
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00021807815802114705,
+      "loss": 2.4852,
+      "step": 38890
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00021804034467892594,
+      "loss": 2.4635,
+      "step": 38900
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002180025258919015,
+      "loss": 2.4845,
+      "step": 38910
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00021796470166310014,
+      "loss": 2.4978,
+      "step": 38920
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002179268719955487,
+      "loss": 2.5048,
+      "step": 38930
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00021788903689227447,
+      "loss": 2.492,
+      "step": 38940
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00021785119635630505,
+      "loss": 2.4945,
+      "step": 38950
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00021781335039066865,
+      "loss": 2.5089,
+      "step": 38960
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00021777549899839368,
+      "loss": 2.4868,
+      "step": 38970
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00021773764218250926,
+      "loss": 2.4667,
+      "step": 38980
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00021769977994604475,
+      "loss": 2.481,
+      "step": 38990
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00021766191229203008,
+      "loss": 2.5032,
+      "step": 39000
+    },
+    {
+      "epoch": 5.27,
+      "eval_loss": 2.568270683288574,
+      "eval_runtime": 1269.1498,
+      "eval_samples_per_second": 59.989,
+      "eval_steps_per_second": 4.999,
+      "step": 39000
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00021762403922349547,
+      "loss": 2.4998,
+      "step": 39010
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002175861607434716,
+      "loss": 2.5132,
+      "step": 39020
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00021754827685498972,
+      "loss": 2.4757,
+      "step": 39030
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00021751038756108142,
+      "loss": 2.4997,
+      "step": 39040
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002174724928647786,
+      "loss": 2.4666,
+      "step": 39050
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002174345927691139,
+      "loss": 2.4702,
+      "step": 39060
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00021739668727712,
+      "loss": 2.5078,
+      "step": 39070
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00021735877639183037,
+      "loss": 2.5101,
+      "step": 39080
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002173208601162787,
+      "loss": 2.5053,
+      "step": 39090
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00021728293845349922,
+      "loss": 2.4808,
+      "step": 39100
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00021724501140652644,
+      "loss": 2.4872,
+      "step": 39110
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00021720707897839543,
+      "loss": 2.489,
+      "step": 39120
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00021716914117214176,
+      "loss": 2.5087,
+      "step": 39130
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00021713119799080126,
+      "loss": 2.4739,
+      "step": 39140
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00021709324943741022,
+      "loss": 2.5089,
+      "step": 39150
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00021705529551500552,
+      "loss": 2.4704,
+      "step": 39160
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00021701733622662425,
+      "loss": 2.4836,
+      "step": 39170
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00021697937157530404,
+      "loss": 2.481,
+      "step": 39180
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.000216941401564083,
+      "loss": 2.4477,
+      "step": 39190
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00021690342619599961,
+      "loss": 2.4396,
+      "step": 39200
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00021686544547409273,
+      "loss": 2.4799,
+      "step": 39210
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00021682745940140166,
+      "loss": 2.4848,
+      "step": 39220
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00021678946798096626,
+      "loss": 2.4674,
+      "step": 39230
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00021675147121582672,
+      "loss": 2.4852,
+      "step": 39240
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00021671346910902357,
+      "loss": 2.4766,
+      "step": 39250
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00021667546166359794,
+      "loss": 2.489,
+      "step": 39260
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00021663744888259127,
+      "loss": 2.4977,
+      "step": 39270
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00021659943076904548,
+      "loss": 2.4933,
+      "step": 39280
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00021656140732600289,
+      "loss": 2.5051,
+      "step": 39290
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00021652337855650628,
+      "loss": 2.4856,
+      "step": 39300
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00021648534446359877,
+      "loss": 2.4729,
+      "step": 39310
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00021644730505032401,
+      "loss": 2.4799,
+      "step": 39320
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00021640926031972604,
+      "loss": 2.4914,
+      "step": 39330
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00021637121027484935,
+      "loss": 2.5298,
+      "step": 39340
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00021633315491873873,
+      "loss": 2.4918,
+      "step": 39350
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002162950942544396,
+      "loss": 2.4563,
+      "step": 39360
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002162570282849976,
+      "loss": 2.4751,
+      "step": 39370
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00021621895701345894,
+      "loss": 2.4494,
+      "step": 39380
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002161808804428702,
+      "loss": 2.4831,
+      "step": 39390
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002161427985762784,
+      "loss": 2.4517,
+      "step": 39400
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00021610471141673092,
+      "loss": 2.4993,
+      "step": 39410
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00021606661896727563,
+      "loss": 2.4844,
+      "step": 39420
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00021602852123096087,
+      "loss": 2.4581,
+      "step": 39430
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002159904182108353,
+      "loss": 2.4611,
+      "step": 39440
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00021595230990994804,
+      "loss": 2.4935,
+      "step": 39450
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00021591419633134863,
+      "loss": 2.5092,
+      "step": 39460
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.000215876077478087,
+      "loss": 2.5045,
+      "step": 39470
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00021583795335321362,
+      "loss": 2.4956,
+      "step": 39480
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00021579982395977926,
+      "loss": 2.4638,
+      "step": 39490
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002157616893008352,
+      "loss": 2.4856,
+      "step": 39500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00021572354937943303,
+      "loss": 2.5163,
+      "step": 39510
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00021568540419862487,
+      "loss": 2.5191,
+      "step": 39520
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00021564725376146317,
+      "loss": 2.4523,
+      "step": 39530
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00021560909807100093,
+      "loss": 2.5048,
+      "step": 39540
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00021557093713029144,
+      "loss": 2.5037,
+      "step": 39550
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00021553277094238846,
+      "loss": 2.4992,
+      "step": 39560
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00021549459951034615,
+      "loss": 2.4644,
+      "step": 39570
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00021545642283721911,
+      "loss": 2.4748,
+      "step": 39580
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00021541824092606243,
+      "loss": 2.4916,
+      "step": 39590
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00021538005377993148,
+      "loss": 2.4974,
+      "step": 39600
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002153418614018821,
+      "loss": 2.4745,
+      "step": 39610
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002153036637949706,
+      "loss": 2.4754,
+      "step": 39620
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00021526546096225366,
+      "loss": 2.4771,
+      "step": 39630
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00021522725290678844,
+      "loss": 2.4807,
+      "step": 39640
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00021518903963163244,
+      "loss": 2.4823,
+      "step": 39650
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00021515082113984357,
+      "loss": 2.5006,
+      "step": 39660
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002151125974344802,
+      "loss": 2.4761,
+      "step": 39670
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00021507436851860113,
+      "loss": 2.4651,
+      "step": 39680
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00021503613439526558,
+      "loss": 2.4811,
+      "step": 39690
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00021499789506753317,
+      "loss": 2.485,
+      "step": 39700
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00021495965053846384,
+      "loss": 2.4871,
+      "step": 39710
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00021492140081111813,
+      "loss": 2.4603,
+      "step": 39720
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00021488314588855687,
+      "loss": 2.4935,
+      "step": 39730
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002148448857738414,
+      "loss": 2.5072,
+      "step": 39740
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00021480662047003332,
+      "loss": 2.4818,
+      "step": 39750
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00021476834998019486,
+      "loss": 2.4773,
+      "step": 39760
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002147300743073884,
+      "loss": 2.4497,
+      "step": 39770
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00021469179345467694,
+      "loss": 2.4704,
+      "step": 39780
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00021465350742512387,
+      "loss": 2.499,
+      "step": 39790
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00021461521622179292,
+      "loss": 2.4724,
+      "step": 39800
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002145769198477483,
+      "loss": 2.4815,
+      "step": 39810
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002145386183060546,
+      "loss": 2.4503,
+      "step": 39820
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00021450031159977678,
+      "loss": 2.4864,
+      "step": 39830
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00021446199973198035,
+      "loss": 2.469,
+      "step": 39840
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002144236827057311,
+      "loss": 2.4699,
+      "step": 39850
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002143853605240953,
+      "loss": 2.4664,
+      "step": 39860
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00021434703319013956,
+      "loss": 2.4798,
+      "step": 39870
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00021430870070693103,
+      "loss": 2.4822,
+      "step": 39880
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002142703630775371,
+      "loss": 2.475,
+      "step": 39890
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00021423202030502577,
+      "loss": 2.4813,
+      "step": 39900
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002141936723924653,
+      "loss": 2.4792,
+      "step": 39910
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00021415531934292442,
+      "loss": 2.4467,
+      "step": 39920
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00021411696115947222,
+      "loss": 2.5251,
+      "step": 39930
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00021407859784517834,
+      "loss": 2.4428,
+      "step": 39940
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00021404022940311264,
+      "loss": 2.4636,
+      "step": 39950
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00021400185583634552,
+      "loss": 2.4907,
+      "step": 39960
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00021396347714794776,
+      "loss": 2.4867,
+      "step": 39970
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00021392509334099052,
+      "loss": 2.4745,
+      "step": 39980
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002138867044185454,
+      "loss": 2.4653,
+      "step": 39990
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00021384831038368446,
+      "loss": 2.4783,
+      "step": 40000
+    },
+    {
+      "epoch": 5.4,
+      "eval_loss": 2.560861587524414,
+      "eval_runtime": 1269.0909,
+      "eval_samples_per_second": 59.992,
+      "eval_steps_per_second": 5.0,
+      "step": 40000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00021380991123948,
+      "loss": 2.457,
+      "step": 40010
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00021377150698900497,
+      "loss": 2.4846,
+      "step": 40020
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00021373309763533249,
+      "loss": 2.5044,
+      "step": 40030
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00021369468318153628,
+      "loss": 2.4751,
+      "step": 40040
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002136562636306903,
+      "loss": 2.4568,
+      "step": 40050
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00021361783898586913,
+      "loss": 2.5042,
+      "step": 40060
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00021357940925014748,
+      "loss": 2.4718,
+      "step": 40070
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002135409744266007,
+      "loss": 2.4524,
+      "step": 40080
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00021350253451830447,
+      "loss": 2.4863,
+      "step": 40090
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002134640895283349,
+      "loss": 2.4508,
+      "step": 40100
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002134256394597684,
+      "loss": 2.4936,
+      "step": 40110
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00021338718431568188,
+      "loss": 2.4764,
+      "step": 40120
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00021334872409915269,
+      "loss": 2.4547,
+      "step": 40130
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00021331025881325856,
+      "loss": 2.4998,
+      "step": 40140
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002132717884610775,
+      "loss": 2.4971,
+      "step": 40150
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00021323331304568818,
+      "loss": 2.4779,
+      "step": 40160
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00021319483257016935,
+      "loss": 2.4987,
+      "step": 40170
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00021315634703760044,
+      "loss": 2.471,
+      "step": 40180
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00021311785645106114,
+      "loss": 2.5032,
+      "step": 40190
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00021307936081363168,
+      "loss": 2.4883,
+      "step": 40200
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002130408601283925,
+      "loss": 2.471,
+      "step": 40210
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002130023543984246,
+      "loss": 2.4743,
+      "step": 40220
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00021296384362680927,
+      "loss": 2.5117,
+      "step": 40230
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002129253278166284,
+      "loss": 2.5009,
+      "step": 40240
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00021288680697096395,
+      "loss": 2.4941,
+      "step": 40250
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00021284828109289865,
+      "loss": 2.4847,
+      "step": 40260
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00021280975018551538,
+      "loss": 2.4897,
+      "step": 40270
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002127712142518975,
+      "loss": 2.4767,
+      "step": 40280
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00021273267329512883,
+      "loss": 2.5074,
+      "step": 40290
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002126941273182935,
+      "loss": 2.4967,
+      "step": 40300
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002126555763244761,
+      "loss": 2.4694,
+      "step": 40310
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002126170203167616,
+      "loss": 2.4723,
+      "step": 40320
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002125784592982353,
+      "loss": 2.4825,
+      "step": 40330
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002125398932719831,
+      "loss": 2.5009,
+      "step": 40340
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00021250132224109107,
+      "loss": 2.462,
+      "step": 40350
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00021246274620864589,
+      "loss": 2.4844,
+      "step": 40360
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00021242416517773443,
+      "loss": 2.4681,
+      "step": 40370
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002123855791514441,
+      "loss": 2.4734,
+      "step": 40380
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00021234698813286267,
+      "loss": 2.4891,
+      "step": 40390
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00021230839212507845,
+      "loss": 2.4771,
+      "step": 40400
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00021226979113117978,
+      "loss": 2.4799,
+      "step": 40410
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00021223118515425581,
+      "loss": 2.4828,
+      "step": 40420
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002121925741973958,
+      "loss": 2.4678,
+      "step": 40430
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00021215395826368962,
+      "loss": 2.4834,
+      "step": 40440
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002121153373562274,
+      "loss": 2.5042,
+      "step": 40450
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00021207671147809967,
+      "loss": 2.4904,
+      "step": 40460
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002120380806323974,
+      "loss": 2.4585,
+      "step": 40470
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00021199944482221204,
+      "loss": 2.4818,
+      "step": 40480
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00021196080405063523,
+      "loss": 2.4875,
+      "step": 40490
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00021192215832075925,
+      "loss": 2.4678,
+      "step": 40500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00021188350763567655,
+      "loss": 2.4804,
+      "step": 40510
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00021184485199848008,
+      "loss": 2.4457,
+      "step": 40520
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00021180619141226325,
+      "loss": 2.4887,
+      "step": 40530
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002117675258801198,
+      "loss": 2.4995,
+      "step": 40540
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002117288554051438,
+      "loss": 2.4506,
+      "step": 40550
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00021169017999042987,
+      "loss": 2.4915,
+      "step": 40560
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00021165149963907283,
+      "loss": 2.4901,
+      "step": 40570
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002116128143541681,
+      "loss": 2.4646,
+      "step": 40580
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002115741241388114,
+      "loss": 2.446,
+      "step": 40590
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00021153542899609882,
+      "loss": 2.4857,
+      "step": 40600
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002114967289291268,
+      "loss": 2.4821,
+      "step": 40610
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00021145802394099234,
+      "loss": 2.4845,
+      "step": 40620
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00021141931403479265,
+      "loss": 2.4813,
+      "step": 40630
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002113805992136255,
+      "loss": 2.469,
+      "step": 40640
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00021134187948058892,
+      "loss": 2.4598,
+      "step": 40650
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00021130315483878145,
+      "loss": 2.4646,
+      "step": 40660
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00021126442529130187,
+      "loss": 2.4652,
+      "step": 40670
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00021122569084124947,
+      "loss": 2.4993,
+      "step": 40680
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00021118695149172388,
+      "loss": 2.4883,
+      "step": 40690
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00021114820724582527,
+      "loss": 2.4603,
+      "step": 40700
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00021110945810665392,
+      "loss": 2.4551,
+      "step": 40710
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00021107070407731077,
+      "loss": 2.4616,
+      "step": 40720
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00021103194516089696,
+      "loss": 2.4797,
+      "step": 40730
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00021099318136051415,
+      "loss": 2.4461,
+      "step": 40740
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00021095441267926428,
+      "loss": 2.4594,
+      "step": 40750
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00021091563912024986,
+      "loss": 2.4628,
+      "step": 40760
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00021087686068657356,
+      "loss": 2.4577,
+      "step": 40770
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002108380773813386,
+      "loss": 2.4635,
+      "step": 40780
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00021079928920764854,
+      "loss": 2.4793,
+      "step": 40790
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00021076049616860736,
+      "loss": 2.4339,
+      "step": 40800
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00021072169826731932,
+      "loss": 2.5019,
+      "step": 40810
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00021068289550688925,
+      "loss": 2.4837,
+      "step": 40820
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00021064408789042214,
+      "loss": 2.456,
+      "step": 40830
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00021060527542102365,
+      "loss": 2.4725,
+      "step": 40840
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00021056645810179958,
+      "loss": 2.4755,
+      "step": 40850
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00021052763593585626,
+      "loss": 2.4978,
+      "step": 40860
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00021048880892630031,
+      "loss": 2.4651,
+      "step": 40870
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00021044997707623882,
+      "loss": 2.4659,
+      "step": 40880
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00021041114038877927,
+      "loss": 2.4653,
+      "step": 40890
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00021037229886702947,
+      "loss": 2.4749,
+      "step": 40900
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002103334525140976,
+      "loss": 2.4841,
+      "step": 40910
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00021029460133309235,
+      "loss": 2.4595,
+      "step": 40920
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002102557453271226,
+      "loss": 2.4742,
+      "step": 40930
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00021021688449929785,
+      "loss": 2.4934,
+      "step": 40940
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00021017801885272778,
+      "loss": 2.4774,
+      "step": 40950
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002101391483905226,
+      "loss": 2.4606,
+      "step": 40960
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002101002731157928,
+      "loss": 2.4823,
+      "step": 40970
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00021006139303164937,
+      "loss": 2.4842,
+      "step": 40980
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002100225081412035,
+      "loss": 2.4572,
+      "step": 40990
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00020998361844756701,
+      "loss": 2.4519,
+      "step": 41000
+    },
+    {
+      "epoch": 5.54,
+      "eval_loss": 2.554502248764038,
+      "eval_runtime": 1269.2475,
+      "eval_samples_per_second": 59.984,
+      "eval_steps_per_second": 4.999,
+      "step": 41000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002099447239538519,
+      "loss": 2.4557,
+      "step": 41010
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002099058246631707,
+      "loss": 2.4632,
+      "step": 41020
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00020986692057863607,
+      "loss": 2.4847,
+      "step": 41030
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00020982801170336152,
+      "loss": 2.4799,
+      "step": 41040
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002097890980404604,
+      "loss": 2.4771,
+      "step": 41050
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002097501795930469,
+      "loss": 2.4819,
+      "step": 41060
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00020971125636423527,
+      "loss": 2.5033,
+      "step": 41070
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00020967232835714033,
+      "loss": 2.4828,
+      "step": 41080
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002096333955748772,
+      "loss": 2.4652,
+      "step": 41090
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002095944580205614,
+      "loss": 2.4774,
+      "step": 41100
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00020955551569730885,
+      "loss": 2.4715,
+      "step": 41110
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00020951656860823586,
+      "loss": 2.4742,
+      "step": 41120
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00020947761675645901,
+      "loss": 2.4598,
+      "step": 41130
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00020943866014509546,
+      "loss": 2.474,
+      "step": 41140
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00020939969877726262,
+      "loss": 2.4592,
+      "step": 41150
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00020936073265607824,
+      "loss": 2.4653,
+      "step": 41160
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00020932176178466052,
+      "loss": 2.4862,
+      "step": 41170
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00020928278616612813,
+      "loss": 2.4957,
+      "step": 41180
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002092438058035999,
+      "loss": 2.5022,
+      "step": 41190
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00020920482070019522,
+      "loss": 2.4821,
+      "step": 41200
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002091658308590338,
+      "loss": 2.4559,
+      "step": 41210
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00020912683628323577,
+      "loss": 2.4591,
+      "step": 41220
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00020908783697592146,
+      "loss": 2.5039,
+      "step": 41230
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002090488329402119,
+      "loss": 2.4594,
+      "step": 41240
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00020900982417922816,
+      "loss": 2.4705,
+      "step": 41250
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00020897081069609196,
+      "loss": 2.4791,
+      "step": 41260
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00020893179249392526,
+      "loss": 2.4889,
+      "step": 41270
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002088927695758503,
+      "loss": 2.4817,
+      "step": 41280
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00020885374194498996,
+      "loss": 2.5149,
+      "step": 41290
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00020881470960446735,
+      "loss": 2.4629,
+      "step": 41300
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00020877567255740586,
+      "loss": 2.5145,
+      "step": 41310
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00020873663080692945,
+      "loss": 2.4725,
+      "step": 41320
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00020869758435616232,
+      "loss": 2.4778,
+      "step": 41330
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00020865853320822908,
+      "loss": 2.4793,
+      "step": 41340
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00020861947736625474,
+      "loss": 2.4559,
+      "step": 41350
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002085804168333647,
+      "loss": 2.4807,
+      "step": 41360
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00020854135161268468,
+      "loss": 2.4569,
+      "step": 41370
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002085022817073408,
+      "loss": 2.5051,
+      "step": 41380
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00020846320712045955,
+      "loss": 2.4575,
+      "step": 41390
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00020842412785516785,
+      "loss": 2.4685,
+      "step": 41400
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00020838504391459291,
+      "loss": 2.4737,
+      "step": 41410
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00020834595530186233,
+      "loss": 2.4929,
+      "step": 41420
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00020830686202010415,
+      "loss": 2.4838,
+      "step": 41430
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002082677640724467,
+      "loss": 2.4692,
+      "step": 41440
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002082286614620187,
+      "loss": 2.4722,
+      "step": 41450
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00020818955419194937,
+      "loss": 2.4508,
+      "step": 41460
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002081504422653681,
+      "loss": 2.4584,
+      "step": 41470
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00020811132568540475,
+      "loss": 2.4972,
+      "step": 41480
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002080722044551896,
+      "loss": 2.492,
+      "step": 41490
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00020803307857785325,
+      "loss": 2.4687,
+      "step": 41500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00020799394805652663,
+      "loss": 2.4831,
+      "step": 41510
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00020795481289434116,
+      "loss": 2.4702,
+      "step": 41520
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0002079156730944285,
+      "loss": 2.4691,
+      "step": 41530
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00020787652865992076,
+      "loss": 2.4848,
+      "step": 41540
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00020783737959395037,
+      "loss": 2.4782,
+      "step": 41550
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00020779822589965023,
+      "loss": 2.4715,
+      "step": 41560
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00020775906758015348,
+      "loss": 2.4653,
+      "step": 41570
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00020771990463859373,
+      "loss": 2.4667,
+      "step": 41580
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00020768073707810484,
+      "loss": 2.497,
+      "step": 41590
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00020764156490182128,
+      "loss": 2.4884,
+      "step": 41600
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00020760238811287762,
+      "loss": 2.4863,
+      "step": 41610
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002075632067144089,
+      "loss": 2.4548,
+      "step": 41620
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002075240207095506,
+      "loss": 2.473,
+      "step": 41630
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00020748483010143845,
+      "loss": 2.5016,
+      "step": 41640
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002074456348932087,
+      "loss": 2.4647,
+      "step": 41650
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00020740643508799776,
+      "loss": 2.5118,
+      "step": 41660
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00020736723068894258,
+      "loss": 2.468,
+      "step": 41670
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00020732802169918038,
+      "loss": 2.4878,
+      "step": 41680
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00020728880812184882,
+      "loss": 2.4699,
+      "step": 41690
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00020724958996008597,
+      "loss": 2.4741,
+      "step": 41700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00020721036721703005,
+      "loss": 2.4482,
+      "step": 41710
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002071711398958199,
+      "loss": 2.4694,
+      "step": 41720
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00020713190799959453,
+      "loss": 2.4633,
+      "step": 41730
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00020709267153149342,
+      "loss": 2.4812,
+      "step": 41740
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00020705343049465643,
+      "loss": 2.4957,
+      "step": 41750
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00020701418489222376,
+      "loss": 2.4698,
+      "step": 41760
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00020697493472733588,
+      "loss": 2.4607,
+      "step": 41770
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002069356800031338,
+      "loss": 2.4739,
+      "step": 41780
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00020689642072275878,
+      "loss": 2.46,
+      "step": 41790
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00020685715688935254,
+      "loss": 2.4599,
+      "step": 41800
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00020681788850605696,
+      "loss": 2.5052,
+      "step": 41810
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002067786155760145,
+      "loss": 2.4855,
+      "step": 41820
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00020673933810236788,
+      "loss": 2.4535,
+      "step": 41830
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00020670005608826024,
+      "loss": 2.4798,
+      "step": 41840
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.000206660769536835,
+      "loss": 2.4492,
+      "step": 41850
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00020662147845123608,
+      "loss": 2.4571,
+      "step": 41860
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00020658218283460756,
+      "loss": 2.4481,
+      "step": 41870
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00020654288269009407,
+      "loss": 2.467,
+      "step": 41880
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00020650357802084053,
+      "loss": 2.4843,
+      "step": 41890
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00020646426882999226,
+      "loss": 2.5013,
+      "step": 41900
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00020642495512069483,
+      "loss": 2.4709,
+      "step": 41910
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00020638563689609425,
+      "loss": 2.4852,
+      "step": 41920
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00020634631415933697,
+      "loss": 2.4607,
+      "step": 41930
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00020630698691356963,
+      "loss": 2.4832,
+      "step": 41940
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00020626765516193932,
+      "loss": 2.4762,
+      "step": 41950
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00020622831890759364,
+      "loss": 2.4693,
+      "step": 41960
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002061889781536802,
+      "loss": 2.4691,
+      "step": 41970
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00020614963290334725,
+      "loss": 2.4866,
+      "step": 41980
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002061102831597434,
+      "loss": 2.495,
+      "step": 41990
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00020607092892601744,
+      "loss": 2.4918,
+      "step": 42000
+    },
+    {
+      "epoch": 5.67,
+      "eval_loss": 2.5471692085266113,
+      "eval_runtime": 1269.1259,
+      "eval_samples_per_second": 59.99,
+      "eval_steps_per_second": 5.0,
+      "step": 42000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00020603157020531864,
+      "loss": 2.4637,
+      "step": 42010
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00020599220700079666,
+      "loss": 2.5047,
+      "step": 42020
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002059528393156014,
+      "loss": 2.4426,
+      "step": 42030
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002059134671528832,
+      "loss": 2.4907,
+      "step": 42040
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00020587409051579277,
+      "loss": 2.4824,
+      "step": 42050
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00020583470940748114,
+      "loss": 2.4361,
+      "step": 42060
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00020579532383109973,
+      "loss": 2.4727,
+      "step": 42070
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002057559337898003,
+      "loss": 2.473,
+      "step": 42080
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00020571653928673492,
+      "loss": 2.4746,
+      "step": 42090
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002056771403250561,
+      "loss": 2.4848,
+      "step": 42100
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00020563773690791666,
+      "loss": 2.4888,
+      "step": 42110
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00020559832903846982,
+      "loss": 2.4469,
+      "step": 42120
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00020555891671986901,
+      "loss": 2.4442,
+      "step": 42130
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00020551949995526827,
+      "loss": 2.4599,
+      "step": 42140
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00020548007874782176,
+      "loss": 2.4513,
+      "step": 42150
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00020544065310068417,
+      "loss": 2.4665,
+      "step": 42160
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00020540122301701034,
+      "loss": 2.431,
+      "step": 42170
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00020536178849995572,
+      "loss": 2.4766,
+      "step": 42180
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002053223495526759,
+      "loss": 2.4579,
+      "step": 42190
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00020528290617832697,
+      "loss": 2.4797,
+      "step": 42200
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00020524345838006524,
+      "loss": 2.4796,
+      "step": 42210
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00020520400616104757,
+      "loss": 2.4668,
+      "step": 42220
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002051645495244309,
+      "loss": 2.4713,
+      "step": 42230
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00020512508847337274,
+      "loss": 2.4502,
+      "step": 42240
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0002050856230110309,
+      "loss": 2.4616,
+      "step": 42250
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00020504615314056356,
+      "loss": 2.4784,
+      "step": 42260
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00020500667886512913,
+      "loss": 2.4766,
+      "step": 42270
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00020496720018788658,
+      "loss": 2.4532,
+      "step": 42280
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00020492771711199501,
+      "loss": 2.4763,
+      "step": 42290
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0002048882296406141,
+      "loss": 2.4548,
+      "step": 42300
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00020484873777690366,
+      "loss": 2.4774,
+      "step": 42310
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00020480924152402398,
+      "loss": 2.4793,
+      "step": 42320
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002047697408851357,
+      "loss": 2.4512,
+      "step": 42330
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00020473023586339975,
+      "loss": 2.4743,
+      "step": 42340
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00020469072646197745,
+      "loss": 2.4844,
+      "step": 42350
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00020465121268403055,
+      "loss": 2.4568,
+      "step": 42360
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00020461169453272096,
+      "loss": 2.4584,
+      "step": 42370
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00020457217201121105,
+      "loss": 2.4511,
+      "step": 42380
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002045326451226636,
+      "loss": 2.4974,
+      "step": 42390
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002044931138702417,
+      "loss": 2.5048,
+      "step": 42400
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00020445357825710868,
+      "loss": 2.4747,
+      "step": 42410
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00020441403828642837,
+      "loss": 2.4695,
+      "step": 42420
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00020437449396136482,
+      "loss": 2.4713,
+      "step": 42430
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00020433494528508252,
+      "loss": 2.4735,
+      "step": 42440
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00020429539226074627,
+      "loss": 2.4628,
+      "step": 42450
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002042558348915213,
+      "loss": 2.4582,
+      "step": 42460
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00020421627318057304,
+      "loss": 2.4751,
+      "step": 42470
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00020417670713106736,
+      "loss": 2.4714,
+      "step": 42480
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00020413713674617045,
+      "loss": 2.4793,
+      "step": 42490
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00020409756202904888,
+      "loss": 2.498,
+      "step": 42500
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00020405798298286958,
+      "loss": 2.4465,
+      "step": 42510
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00020401839961079968,
+      "loss": 2.4549,
+      "step": 42520
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00020397881191600684,
+      "loss": 2.4307,
+      "step": 42530
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00020393921990165902,
+      "loss": 2.4334,
+      "step": 42540
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00020389962357092442,
+      "loss": 2.4896,
+      "step": 42550
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00020386002292697173,
+      "loss": 2.472,
+      "step": 42560
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00020382041797296988,
+      "loss": 2.4767,
+      "step": 42570
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002037808087120882,
+      "loss": 2.464,
+      "step": 42580
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00020374119514749637,
+      "loss": 2.4777,
+      "step": 42590
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002037015772823643,
+      "loss": 2.4514,
+      "step": 42600
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00020366195511986251,
+      "loss": 2.4833,
+      "step": 42610
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0002036223286631615,
+      "loss": 2.4539,
+      "step": 42620
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00020358269791543248,
+      "loss": 2.4681,
+      "step": 42630
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00020354306287984663,
+      "loss": 2.4599,
+      "step": 42640
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00020350342355957585,
+      "loss": 2.4594,
+      "step": 42650
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00020346377995779206,
+      "loss": 2.4673,
+      "step": 42660
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00020342413207766782,
+      "loss": 2.4665,
+      "step": 42670
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00020338447992237577,
+      "loss": 2.4719,
+      "step": 42680
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00020334482349508898,
+      "loss": 2.4851,
+      "step": 42690
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00020330516279898098,
+      "loss": 2.4755,
+      "step": 42700
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002032654978372255,
+      "loss": 2.4573,
+      "step": 42710
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002032258286129966,
+      "loss": 2.4707,
+      "step": 42720
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00020318615512946884,
+      "loss": 2.4884,
+      "step": 42730
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00020314647738981694,
+      "loss": 2.4536,
+      "step": 42740
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00020310679539721607,
+      "loss": 2.469,
+      "step": 42750
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00020306710915484166,
+      "loss": 2.479,
+      "step": 42760
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00020302741866586962,
+      "loss": 2.4842,
+      "step": 42770
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00020298772393347603,
+      "loss": 2.4715,
+      "step": 42780
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002029480249608374,
+      "loss": 2.4758,
+      "step": 42790
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00020290832175113058,
+      "loss": 2.4645,
+      "step": 42800
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00020286861430753278,
+      "loss": 2.4602,
+      "step": 42810
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00020282890263322148,
+      "loss": 2.4406,
+      "step": 42820
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00020278918673137454,
+      "loss": 2.4553,
+      "step": 42830
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00020274946660517012,
+      "loss": 2.4841,
+      "step": 42840
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002027097422577868,
+      "loss": 2.4822,
+      "step": 42850
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002026700136924034,
+      "loss": 2.4672,
+      "step": 42860
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00020263028091219919,
+      "loss": 2.463,
+      "step": 42870
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00020259054392035363,
+      "loss": 2.4921,
+      "step": 42880
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002025508027200467,
+      "loss": 2.4394,
+      "step": 42890
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00020251105731445852,
+      "loss": 2.4929,
+      "step": 42900
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00020247130770676967,
+      "loss": 2.4455,
+      "step": 42910
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00020243155390016108,
+      "loss": 2.4725,
+      "step": 42920
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00020239179589781397,
+      "loss": 2.4708,
+      "step": 42930
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002023520337029098,
+      "loss": 2.4638,
+      "step": 42940
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002023122673186306,
+      "loss": 2.4401,
+      "step": 42950
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00020227249674815853,
+      "loss": 2.4907,
+      "step": 42960
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00020223272199467623,
+      "loss": 2.4784,
+      "step": 42970
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00020219294306136648,
+      "loss": 2.4602,
+      "step": 42980
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00020215315995141258,
+      "loss": 2.4695,
+      "step": 42990
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002021133726679981,
+      "loss": 2.4591,
+      "step": 43000
+    },
+    {
+      "epoch": 5.81,
+      "eval_loss": 2.5411388874053955,
+      "eval_runtime": 1269.129,
+      "eval_samples_per_second": 59.99,
+      "eval_steps_per_second": 4.999,
+      "step": 43000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00020207358121430698,
+      "loss": 2.4675,
+      "step": 43010
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00020203378559352337,
+      "loss": 2.4821,
+      "step": 43020
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00020199398580883197,
+      "loss": 2.4569,
+      "step": 43030
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00020195418186341754,
+      "loss": 2.4905,
+      "step": 43040
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002019143737604654,
+      "loss": 2.4404,
+      "step": 43050
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002018745615031611,
+      "loss": 2.467,
+      "step": 43060
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00020183474509469058,
+      "loss": 2.4447,
+      "step": 43070
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00020179492453824,
+      "loss": 2.4805,
+      "step": 43080
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00020175509983699598,
+      "loss": 2.4832,
+      "step": 43090
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002017152709941454,
+      "loss": 2.4672,
+      "step": 43100
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00020167543801287555,
+      "loss": 2.456,
+      "step": 43110
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00020163560089637392,
+      "loss": 2.4429,
+      "step": 43120
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00020159575964782842,
+      "loss": 2.4287,
+      "step": 43130
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00020155591427042723,
+      "loss": 2.4802,
+      "step": 43140
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00020151606476735896,
+      "loss": 2.4723,
+      "step": 43150
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00020148019668978778,
+      "loss": 2.4427,
+      "step": 43160
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00020144033935673772,
+      "loss": 2.4765,
+      "step": 43170
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00020140047790726922,
+      "loss": 2.4483,
+      "step": 43180
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00020136061234457206,
+      "loss": 2.4537,
+      "step": 43190
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00020132074267183655,
+      "loss": 2.4451,
+      "step": 43200
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00020128086889225312,
+      "loss": 2.4245,
+      "step": 43210
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00020124099100901258,
+      "loss": 2.4773,
+      "step": 43220
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00020120110902530612,
+      "loss": 2.4838,
+      "step": 43230
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00020116122294432519,
+      "loss": 2.4396,
+      "step": 43240
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00020112133276926157,
+      "loss": 2.4741,
+      "step": 43250
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002010814385033075,
+      "loss": 2.4388,
+      "step": 43260
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00020104154014965537,
+      "loss": 2.4827,
+      "step": 43270
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.000201001637711498,
+      "loss": 2.4781,
+      "step": 43280
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002009617311920285,
+      "loss": 2.4622,
+      "step": 43290
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00020092182059444028,
+      "loss": 2.4894,
+      "step": 43300
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00020088190592192713,
+      "loss": 2.4448,
+      "step": 43310
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002008419871776832,
+      "loss": 2.4658,
+      "step": 43320
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00020080206436490287,
+      "loss": 2.4876,
+      "step": 43330
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00020076213748678088,
+      "loss": 2.4874,
+      "step": 43340
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00020072220654651234,
+      "loss": 2.4711,
+      "step": 43350
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00020068227154729267,
+      "loss": 2.4688,
+      "step": 43360
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00020064233249231749,
+      "loss": 2.4501,
+      "step": 43370
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00020060238938478296,
+      "loss": 2.4949,
+      "step": 43380
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00020056244222788536,
+      "loss": 2.4755,
+      "step": 43390
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00020052249102482146,
+      "loss": 2.4613,
+      "step": 43400
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00020048253577878823,
+      "loss": 2.4779,
+      "step": 43410
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00020044257649298313,
+      "loss": 2.4617,
+      "step": 43420
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020040261317060367,
+      "loss": 2.4851,
+      "step": 43430
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0002003626458148479,
+      "loss": 2.4453,
+      "step": 43440
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020032267442891418,
+      "loss": 2.4717,
+      "step": 43450
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020028269901600115,
+      "loss": 2.4684,
+      "step": 43460
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020024271957930768,
+      "loss": 2.474,
+      "step": 43470
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020020273612203309,
+      "loss": 2.4314,
+      "step": 43480
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020016274864737701,
+      "loss": 2.4656,
+      "step": 43490
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00020012275715853935,
+      "loss": 2.4719,
+      "step": 43500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00020008276165872033,
+      "loss": 2.4889,
+      "step": 43510
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00020004276215112064,
+      "loss": 2.4206,
+      "step": 43520
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00020000275863894096,
+      "loss": 2.4577,
+      "step": 43530
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00019996275112538266,
+      "loss": 2.4732,
+      "step": 43540
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00019992273961364716,
+      "loss": 2.4665,
+      "step": 43550
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0001998827241069364,
+      "loss": 2.4989,
+      "step": 43560
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0001998427046084525,
+      "loss": 2.4787,
+      "step": 43570
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00019980268112139798,
+      "loss": 2.49,
+      "step": 43580
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00019976265364897558,
+      "loss": 2.4584,
+      "step": 43590
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00019972262219438846,
+      "loss": 2.4544,
+      "step": 43600
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0001996825867608401,
+      "loss": 2.4424,
+      "step": 43610
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00019964254735153422,
+      "loss": 2.4751,
+      "step": 43620
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00019960250396967492,
+      "loss": 2.4537,
+      "step": 43630
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00019956245661846656,
+      "loss": 2.4604,
+      "step": 43640
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00019952240530111387,
+      "loss": 2.4782,
+      "step": 43650
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00019948235002082196,
+      "loss": 2.4703,
+      "step": 43660
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00019944229078079607,
+      "loss": 2.4771,
+      "step": 43670
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00019940222758424192,
+      "loss": 2.4628,
+      "step": 43680
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00019936216043436552,
+      "loss": 2.4711,
+      "step": 43690
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00019932208933437308,
+      "loss": 2.4672,
+      "step": 43700
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0001992820142874713,
+      "loss": 2.5034,
+      "step": 43710
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00019924193529686712,
+      "loss": 2.4551,
+      "step": 43720
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00019920185236576765,
+      "loss": 2.4717,
+      "step": 43730
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00019916176549738065,
+      "loss": 2.4742,
+      "step": 43740
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00019912167469491384,
+      "loss": 2.4512,
+      "step": 43750
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00019908157996157555,
+      "loss": 2.4714,
+      "step": 43760
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00019904148130057414,
+      "loss": 2.4909,
+      "step": 43770
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0001990013787151186,
+      "loss": 2.466,
+      "step": 43780
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00019896127220841786,
+      "loss": 2.4609,
+      "step": 43790
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00019892116178368152,
+      "loss": 2.4771,
+      "step": 43800
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001988810474441193,
+      "loss": 2.4726,
+      "step": 43810
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001988409291929413,
+      "loss": 2.4756,
+      "step": 43820
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00019880080703335785,
+      "loss": 2.483,
+      "step": 43830
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001987606809685797,
+      "loss": 2.4737,
+      "step": 43840
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00019872055100181784,
+      "loss": 2.4694,
+      "step": 43850
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00019868041713628367,
+      "loss": 2.463,
+      "step": 43860
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001986402793751887,
+      "loss": 2.4546,
+      "step": 43870
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.000198600137721745,
+      "loss": 2.4435,
+      "step": 43880
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001985599921791647,
+      "loss": 2.4781,
+      "step": 43890
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00019851984275066054,
+      "loss": 2.4482,
+      "step": 43900
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00019847968943944525,
+      "loss": 2.4401,
+      "step": 43910
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001984395322487322,
+      "loss": 2.4637,
+      "step": 43920
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001983993711817347,
+      "loss": 2.4782,
+      "step": 43930
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00019835920624166665,
+      "loss": 2.4491,
+      "step": 43940
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001983190374317422,
+      "loss": 2.5098,
+      "step": 43950
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00019827886475517578,
+      "loss": 2.4507,
+      "step": 43960
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001982386882151821,
+      "loss": 2.47,
+      "step": 43970
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00019819850781497626,
+      "loss": 2.4464,
+      "step": 43980
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00019815832355777355,
+      "loss": 2.4305,
+      "step": 43990
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001981181354467897,
+      "loss": 2.4756,
+      "step": 44000
+    },
+    {
+      "epoch": 5.94,
+      "eval_loss": 2.535437822341919,
+      "eval_runtime": 1269.0292,
+      "eval_samples_per_second": 59.995,
+      "eval_steps_per_second": 5.0,
+      "step": 44000
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00019807794348524065,
+      "loss": 2.4382,
+      "step": 44010
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00019803774767634277,
+      "loss": 2.4432,
+      "step": 44020
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00019799754802331257,
+      "loss": 2.4465,
+      "step": 44030
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00019795734452936697,
+      "loss": 2.4528,
+      "step": 44040
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0001979171371977232,
+      "loss": 2.4641,
+      "step": 44050
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00019787692603159877,
+      "loss": 2.4507,
+      "step": 44060
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00019783671103421152,
+      "loss": 2.4466,
+      "step": 44070
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00019779649220877957,
+      "loss": 2.4871,
+      "step": 44080
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00019775626955852133,
+      "loss": 2.4524,
+      "step": 44090
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00019771604308665554,
+      "loss": 2.4853,
+      "step": 44100
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001976758127964013,
+      "loss": 2.4546,
+      "step": 44110
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00019763557869097796,
+      "loss": 2.4785,
+      "step": 44120
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00019759534077360517,
+      "loss": 2.4803,
+      "step": 44130
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00019755509904750284,
+      "loss": 2.4603,
+      "step": 44140
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00019751485351589133,
+      "loss": 2.4352,
+      "step": 44150
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00019747460418199114,
+      "loss": 2.4724,
+      "step": 44160
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0001974343510490232,
+      "loss": 2.4509,
+      "step": 44170
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00019739409412020868,
+      "loss": 2.4819,
+      "step": 44180
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00019735383339876904,
+      "loss": 2.4686,
+      "step": 44190
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0001973135688879261,
+      "loss": 2.492,
+      "step": 44200
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00019727330059090192,
+      "loss": 2.4858,
+      "step": 44210
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00019723302851091897,
+      "loss": 2.4667,
+      "step": 44220
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00019719275265119983,
+      "loss": 2.4498,
+      "step": 44230
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0001971524730149676,
+      "loss": 2.4472,
+      "step": 44240
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00019711218960544554,
+      "loss": 2.4825,
+      "step": 44250
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00019707190242585732,
+      "loss": 2.48,
+      "step": 44260
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00019703161147942674,
+      "loss": 2.4534,
+      "step": 44270
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0001969913167693781,
+      "loss": 2.443,
+      "step": 44280
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00019695101829893586,
+      "loss": 2.489,
+      "step": 44290
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0001969107160713248,
+      "loss": 2.4746,
+      "step": 44300
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00019687041008977012,
+      "loss": 2.4829,
+      "step": 44310
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00019683010035749724,
+      "loss": 2.4478,
+      "step": 44320
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00019678978687773173,
+      "loss": 2.4747,
+      "step": 44330
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00019674946965369974,
+      "loss": 2.4661,
+      "step": 44340
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001967091486886275,
+      "loss": 2.466,
+      "step": 44350
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00019666882398574172,
+      "loss": 2.4518,
+      "step": 44360
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00019662849554826918,
+      "loss": 2.4558,
+      "step": 44370
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001965881633794372,
+      "loss": 2.448,
+      "step": 44380
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00019654782748247324,
+      "loss": 2.4876,
+      "step": 44390
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00019650748786060508,
+      "loss": 2.4389,
+      "step": 44400
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00019646714451706086,
+      "loss": 2.4567,
+      "step": 44410
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00019642679745506903,
+      "loss": 2.5036,
+      "step": 44420
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00019638644667785814,
+      "loss": 2.4461,
+      "step": 44430
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00019634609218865736,
+      "loss": 2.4168,
+      "step": 44440
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00019630573399069583,
+      "loss": 2.4105,
+      "step": 44450
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00019626537208720327,
+      "loss": 2.4163,
+      "step": 44460
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019622500648140947,
+      "loss": 2.417,
+      "step": 44470
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019618463717654473,
+      "loss": 2.4254,
+      "step": 44480
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019614426417583936,
+      "loss": 2.3788,
+      "step": 44490
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019610388748252425,
+      "loss": 2.4101,
+      "step": 44500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019606350709983048,
+      "loss": 2.4378,
+      "step": 44510
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001960231230309894,
+      "loss": 2.4292,
+      "step": 44520
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019598273527923258,
+      "loss": 2.3993,
+      "step": 44530
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019594234384779215,
+      "loss": 2.4199,
+      "step": 44540
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019590194873990018,
+      "loss": 2.3988,
+      "step": 44550
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001958615499587893,
+      "loss": 2.4226,
+      "step": 44560
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001958211475076924,
+      "loss": 2.4346,
+      "step": 44570
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019578074138984253,
+      "loss": 2.4329,
+      "step": 44580
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019574033160847312,
+      "loss": 2.4202,
+      "step": 44590
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001956999181668179,
+      "loss": 2.4431,
+      "step": 44600
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001956595010681109,
+      "loss": 2.4107,
+      "step": 44610
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019561908031558647,
+      "loss": 2.392,
+      "step": 44620
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019557865591247908,
+      "loss": 2.3989,
+      "step": 44630
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019553822786202372,
+      "loss": 2.4116,
+      "step": 44640
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019549779616745552,
+      "loss": 2.4127,
+      "step": 44650
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019545736083200998,
+      "loss": 2.4391,
+      "step": 44660
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019541692185892284,
+      "loss": 2.4043,
+      "step": 44670
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019537647925143024,
+      "loss": 2.4239,
+      "step": 44680
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019533603301276836,
+      "loss": 2.4172,
+      "step": 44690
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019529558314617396,
+      "loss": 2.4153,
+      "step": 44700
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019525512965488392,
+      "loss": 2.4234,
+      "step": 44710
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001952146725421355,
+      "loss": 2.4404,
+      "step": 44720
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019517421181116618,
+      "loss": 2.4031,
+      "step": 44730
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019513374746521375,
+      "loss": 2.4353,
+      "step": 44740
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019509327950751628,
+      "loss": 2.4175,
+      "step": 44750
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019505280794131216,
+      "loss": 2.414,
+      "step": 44760
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019501233276984005,
+      "loss": 2.4094,
+      "step": 44770
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019497185399633895,
+      "loss": 2.4099,
+      "step": 44780
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019493137162404802,
+      "loss": 2.4047,
+      "step": 44790
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019489088565620682,
+      "loss": 2.4114,
+      "step": 44800
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019485039609605512,
+      "loss": 2.4199,
+      "step": 44810
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019480990294683313,
+      "loss": 2.4246,
+      "step": 44820
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001947694062117812,
+      "loss": 2.4037,
+      "step": 44830
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019472890589413994,
+      "loss": 2.4312,
+      "step": 44840
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019468840199715035,
+      "loss": 2.4394,
+      "step": 44850
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019464789452405373,
+      "loss": 2.4416,
+      "step": 44860
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019460738347809153,
+      "loss": 2.4383,
+      "step": 44870
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019456686886250565,
+      "loss": 2.4326,
+      "step": 44880
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019452635068053814,
+      "loss": 2.426,
+      "step": 44890
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019448582893543144,
+      "loss": 2.389,
+      "step": 44900
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019444530363042819,
+      "loss": 2.4545,
+      "step": 44910
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001944047747687714,
+      "loss": 2.4349,
+      "step": 44920
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019436424235370423,
+      "loss": 2.4316,
+      "step": 44930
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001943237063884703,
+      "loss": 2.4241,
+      "step": 44940
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019428316687631337,
+      "loss": 2.4244,
+      "step": 44950
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001942426238204776,
+      "loss": 2.4234,
+      "step": 44960
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001942020772242073,
+      "loss": 2.4132,
+      "step": 44970
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019416152709074717,
+      "loss": 2.4465,
+      "step": 44980
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001941209734233422,
+      "loss": 2.4269,
+      "step": 44990
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019408041622523753,
+      "loss": 2.4434,
+      "step": 45000
+    },
+    {
+      "epoch": 6.08,
+      "eval_loss": 2.5345442295074463,
+      "eval_runtime": 1269.0646,
+      "eval_samples_per_second": 59.993,
+      "eval_steps_per_second": 5.0,
+      "step": 45000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019403985549967876,
+      "loss": 2.4377,
+      "step": 45010
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019399929124991168,
+      "loss": 2.4184,
+      "step": 45020
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001939587234791823,
+      "loss": 2.4172,
+      "step": 45030
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019391815219073707,
+      "loss": 2.4167,
+      "step": 45040
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019387757738782254,
+      "loss": 2.3992,
+      "step": 45050
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019383699907368578,
+      "loss": 2.4135,
+      "step": 45060
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019379641725157383,
+      "loss": 2.4192,
+      "step": 45070
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019375583192473425,
+      "loss": 2.4079,
+      "step": 45080
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001937152430964148,
+      "loss": 2.4317,
+      "step": 45090
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019367465076986352,
+      "loss": 2.3917,
+      "step": 45100
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019363405494832875,
+      "loss": 2.4152,
+      "step": 45110
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001935934556350591,
+      "loss": 2.4341,
+      "step": 45120
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019355285283330343,
+      "loss": 2.4299,
+      "step": 45130
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019351224654631097,
+      "loss": 2.422,
+      "step": 45140
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.000193471636777331,
+      "loss": 2.436,
+      "step": 45150
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019343102352961346,
+      "loss": 2.4114,
+      "step": 45160
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019339040680640818,
+      "loss": 2.432,
+      "step": 45170
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019334978661096557,
+      "loss": 2.4085,
+      "step": 45180
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019330916294653605,
+      "loss": 2.4398,
+      "step": 45190
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019326853581637057,
+      "loss": 2.3995,
+      "step": 45200
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019322790522372017,
+      "loss": 2.4231,
+      "step": 45210
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019318727117183628,
+      "loss": 2.4098,
+      "step": 45220
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019314663366397052,
+      "loss": 2.4187,
+      "step": 45230
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001931059927033749,
+      "loss": 2.4202,
+      "step": 45240
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019306534829330153,
+      "loss": 2.4195,
+      "step": 45250
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019302470043700308,
+      "loss": 2.4241,
+      "step": 45260
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019298404913773213,
+      "loss": 2.4368,
+      "step": 45270
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019294339439874183,
+      "loss": 2.4241,
+      "step": 45280
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019290273622328546,
+      "loss": 2.4309,
+      "step": 45290
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019286207461461665,
+      "loss": 2.4105,
+      "step": 45300
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019282140957598926,
+      "loss": 2.4132,
+      "step": 45310
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019278074111065744,
+      "loss": 2.4111,
+      "step": 45320
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019274006922187557,
+      "loss": 2.4489,
+      "step": 45330
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001926993939128984,
+      "loss": 2.413,
+      "step": 45340
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00019265871518698083,
+      "loss": 2.4419,
+      "step": 45350
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00019261803304737815,
+      "loss": 2.408,
+      "step": 45360
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00019257734749734582,
+      "loss": 2.4048,
+      "step": 45370
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00019253665854013975,
+      "loss": 2.3964,
+      "step": 45380
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00019249596617901585,
+      "loss": 2.4331,
+      "step": 45390
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00019245527041723053,
+      "loss": 2.4055,
+      "step": 45400
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001924145712580404,
+      "loss": 2.422,
+      "step": 45410
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00019237386870470233,
+      "loss": 2.4549,
+      "step": 45420
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00019233316276047344,
+      "loss": 2.4309,
+      "step": 45430
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001922924534286112,
+      "loss": 2.4269,
+      "step": 45440
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00019225174071237316,
+      "loss": 2.4132,
+      "step": 45450
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00019221102461501743,
+      "loss": 2.4163,
+      "step": 45460
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001921703051398022,
+      "loss": 2.4228,
+      "step": 45470
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00019212958228998595,
+      "loss": 2.444,
+      "step": 45480
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001920888560688275,
+      "loss": 2.4368,
+      "step": 45490
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001920481264795858,
+      "loss": 2.4401,
+      "step": 45500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00019200739352552023,
+      "loss": 2.4042,
+      "step": 45510
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00019196665720989036,
+      "loss": 2.426,
+      "step": 45520
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00019192591753595603,
+      "loss": 2.4393,
+      "step": 45530
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001918851745069774,
+      "loss": 2.4186,
+      "step": 45540
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00019184442812621478,
+      "loss": 2.4334,
+      "step": 45550
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00019180367839692882,
+      "loss": 2.4069,
+      "step": 45560
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001917629253223805,
+      "loss": 2.4232,
+      "step": 45570
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00019172216890583105,
+      "loss": 2.434,
+      "step": 45580
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00019168140915054184,
+      "loss": 2.4273,
+      "step": 45590
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001916406460597746,
+      "loss": 2.4285,
+      "step": 45600
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00019159987963679134,
+      "loss": 2.4381,
+      "step": 45610
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001915591098848544,
+      "loss": 2.4134,
+      "step": 45620
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00019151833680722613,
+      "loss": 2.4266,
+      "step": 45630
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00019147756040716946,
+      "loss": 2.419,
+      "step": 45640
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001914367806879474,
+      "loss": 2.4442,
+      "step": 45650
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00019139599765282328,
+      "loss": 2.422,
+      "step": 45660
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00019135521130506065,
+      "loss": 2.4315,
+      "step": 45670
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00019131442164792344,
+      "loss": 2.4441,
+      "step": 45680
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001912736286846757,
+      "loss": 2.4079,
+      "step": 45690
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00019123283241858185,
+      "loss": 2.42,
+      "step": 45700
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00019119203285290648,
+      "loss": 2.4429,
+      "step": 45710
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00019115122999091458,
+      "loss": 2.4481,
+      "step": 45720
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00019111042383587124,
+      "loss": 2.4482,
+      "step": 45730
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.000191069614391042,
+      "loss": 2.4234,
+      "step": 45740
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00019102880165969247,
+      "loss": 2.4502,
+      "step": 45750
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00019098798564508865,
+      "loss": 2.4317,
+      "step": 45760
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00019094716635049676,
+      "loss": 2.4328,
+      "step": 45770
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001909063437791833,
+      "loss": 2.4391,
+      "step": 45780
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.000190865517934415,
+      "loss": 2.4513,
+      "step": 45790
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001908246888194589,
+      "loss": 2.4145,
+      "step": 45800
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00019078385643758224,
+      "loss": 2.3955,
+      "step": 45810
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001907430207920526,
+      "loss": 2.4497,
+      "step": 45820
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00019070218188613772,
+      "loss": 2.4198,
+      "step": 45830
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00019066133972310575,
+      "loss": 2.4139,
+      "step": 45840
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00019062049430622493,
+      "loss": 2.4346,
+      "step": 45850
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00019057964563876383,
+      "loss": 2.4108,
+      "step": 45860
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001905387937239913,
+      "loss": 2.4519,
+      "step": 45870
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001904979385651765,
+      "loss": 2.4307,
+      "step": 45880
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00019045708016558874,
+      "loss": 2.4433,
+      "step": 45890
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00019041621852849763,
+      "loss": 2.4233,
+      "step": 45900
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00019037535365717308,
+      "loss": 2.4408,
+      "step": 45910
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001903344855548852,
+      "loss": 2.4105,
+      "step": 45920
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001902936142249044,
+      "loss": 2.4338,
+      "step": 45930
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00019025273967050133,
+      "loss": 2.4343,
+      "step": 45940
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001902118618949469,
+      "loss": 2.4111,
+      "step": 45950
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00019017098090151223,
+      "loss": 2.4483,
+      "step": 45960
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00019013009669346885,
+      "loss": 2.4401,
+      "step": 45970
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00019008920927408838,
+      "loss": 2.4046,
+      "step": 45980
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00019004831864664277,
+      "loss": 2.4233,
+      "step": 45990
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00019000742481440415,
+      "loss": 2.4312,
+      "step": 46000
+    },
+    {
+      "epoch": 6.21,
+      "eval_loss": 2.5300614833831787,
+      "eval_runtime": 1269.1082,
+      "eval_samples_per_second": 59.991,
+      "eval_steps_per_second": 5.0,
+      "step": 46000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00018996652778064508,
+      "loss": 2.4175,
+      "step": 46010
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00018992562754863827,
+      "loss": 2.4362,
+      "step": 46020
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00018988472412165657,
+      "loss": 2.4372,
+      "step": 46030
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00018984381750297333,
+      "loss": 2.4333,
+      "step": 46040
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00018980290769586195,
+      "loss": 2.4231,
+      "step": 46050
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00018976199470359617,
+      "loss": 2.4398,
+      "step": 46060
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00018972107852944996,
+      "loss": 2.4525,
+      "step": 46070
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00018968015917669766,
+      "loss": 2.4293,
+      "step": 46080
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00018963923664861365,
+      "loss": 2.4222,
+      "step": 46090
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00018959831094847275,
+      "loss": 2.4402,
+      "step": 46100
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001895573820795499,
+      "loss": 2.4206,
+      "step": 46110
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00018951645004512043,
+      "loss": 2.4311,
+      "step": 46120
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00018947551484845983,
+      "loss": 2.4245,
+      "step": 46130
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00018943457649284385,
+      "loss": 2.4217,
+      "step": 46140
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001893936349815485,
+      "loss": 2.4314,
+      "step": 46150
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00018935269031785006,
+      "loss": 2.4365,
+      "step": 46160
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00018931174250502502,
+      "loss": 2.438,
+      "step": 46170
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00018927079154635026,
+      "loss": 2.4398,
+      "step": 46180
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00018922983744510264,
+      "loss": 2.4273,
+      "step": 46190
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00018918888020455957,
+      "loss": 2.4405,
+      "step": 46200
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001891479198279985,
+      "loss": 2.4504,
+      "step": 46210
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00018910695631869727,
+      "loss": 2.4403,
+      "step": 46220
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00018906598967993388,
+      "loss": 2.4238,
+      "step": 46230
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001890250199149866,
+      "loss": 2.4039,
+      "step": 46240
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00018898404702713396,
+      "loss": 2.4152,
+      "step": 46250
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00018894307101965477,
+      "loss": 2.4168,
+      "step": 46260
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.000188902091895828,
+      "loss": 2.438,
+      "step": 46270
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00018886110965893305,
+      "loss": 2.4251,
+      "step": 46280
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001888201243122493,
+      "loss": 2.4177,
+      "step": 46290
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00018877913585905657,
+      "loss": 2.4587,
+      "step": 46300
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00018873814430263495,
+      "loss": 2.4234,
+      "step": 46310
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001886971496462647,
+      "loss": 2.4196,
+      "step": 46320
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00018865615189322629,
+      "loss": 2.4472,
+      "step": 46330
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001886151510468005,
+      "loss": 2.4428,
+      "step": 46340
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00018857414711026836,
+      "loss": 2.4073,
+      "step": 46350
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00018853314008691115,
+      "loss": 2.4382,
+      "step": 46360
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00018849212998001035,
+      "loss": 2.4395,
+      "step": 46370
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00018845111679284773,
+      "loss": 2.4194,
+      "step": 46380
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00018841010052870534,
+      "loss": 2.4157,
+      "step": 46390
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00018836908119086534,
+      "loss": 2.4282,
+      "step": 46400
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00018832805878261032,
+      "loss": 2.4224,
+      "step": 46410
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00018828703330722298,
+      "loss": 2.4221,
+      "step": 46420
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00018824600476798629,
+      "loss": 2.4223,
+      "step": 46430
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001882049731681835,
+      "loss": 2.3979,
+      "step": 46440
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001881639385110981,
+      "loss": 2.4142,
+      "step": 46450
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00018812290080001378,
+      "loss": 2.4226,
+      "step": 46460
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00018808186003821457,
+      "loss": 2.4453,
+      "step": 46470
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00018804081622898462,
+      "loss": 2.4472,
+      "step": 46480
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00018799976937560842,
+      "loss": 2.4361,
+      "step": 46490
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00018795871948137065,
+      "loss": 2.4218,
+      "step": 46500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00018791766654955623,
+      "loss": 2.4328,
+      "step": 46510
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001878766105834504,
+      "loss": 2.4336,
+      "step": 46520
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00018783555158633858,
+      "loss": 2.4407,
+      "step": 46530
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001877944895615064,
+      "loss": 2.426,
+      "step": 46540
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001877534245122398,
+      "loss": 2.4379,
+      "step": 46550
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001877123564418249,
+      "loss": 2.4158,
+      "step": 46560
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00018767128535354815,
+      "loss": 2.4592,
+      "step": 46570
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001876302112506962,
+      "loss": 2.4206,
+      "step": 46580
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00018758913413655584,
+      "loss": 2.4175,
+      "step": 46590
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001875480540144143,
+      "loss": 2.4591,
+      "step": 46600
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00018750697088755882,
+      "loss": 2.4428,
+      "step": 46610
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00018746588475927712,
+      "loss": 2.4381,
+      "step": 46620
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00018742479563285696,
+      "loss": 2.4166,
+      "step": 46630
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00018738370351158647,
+      "loss": 2.4311,
+      "step": 46640
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0001873426083987539,
+      "loss": 2.4352,
+      "step": 46650
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00018730151029764796,
+      "loss": 2.4308,
+      "step": 46660
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00018726040921155724,
+      "loss": 2.4427,
+      "step": 46670
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00018721930514377098,
+      "loss": 2.4214,
+      "step": 46680
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00018717819809757836,
+      "loss": 2.4347,
+      "step": 46690
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001871370880762689,
+      "loss": 2.4208,
+      "step": 46700
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001870959750831323,
+      "loss": 2.4392,
+      "step": 46710
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00018705485912145865,
+      "loss": 2.4181,
+      "step": 46720
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00018701374019453816,
+      "loss": 2.4328,
+      "step": 46730
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00018697261830566127,
+      "loss": 2.4409,
+      "step": 46740
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00018693149345811865,
+      "loss": 2.4304,
+      "step": 46750
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001868903656552013,
+      "loss": 2.4364,
+      "step": 46760
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00018684923490020036,
+      "loss": 2.4417,
+      "step": 46770
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00018680810119640732,
+      "loss": 2.4155,
+      "step": 46780
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00018676696454711372,
+      "loss": 2.3956,
+      "step": 46790
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00018672582495561153,
+      "loss": 2.4175,
+      "step": 46800
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00018668468242519275,
+      "loss": 2.4588,
+      "step": 46810
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00018664353695914987,
+      "loss": 2.4244,
+      "step": 46820
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00018660238856077545,
+      "loss": 2.4032,
+      "step": 46830
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00018656123723336229,
+      "loss": 2.3936,
+      "step": 46840
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001865200829802034,
+      "loss": 2.4191,
+      "step": 46850
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00018647892580459213,
+      "loss": 2.4187,
+      "step": 46860
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00018643776570982204,
+      "loss": 2.4356,
+      "step": 46870
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00018639660269918683,
+      "loss": 2.4544,
+      "step": 46880
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00018635543677598054,
+      "loss": 2.4445,
+      "step": 46890
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001863142679434974,
+      "loss": 2.4267,
+      "step": 46900
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00018627309620503174,
+      "loss": 2.3972,
+      "step": 46910
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00018623192156387843,
+      "loss": 2.4023,
+      "step": 46920
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00018619074402333232,
+      "loss": 2.4268,
+      "step": 46930
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00018614956358668858,
+      "loss": 2.4165,
+      "step": 46940
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00018610838025724254,
+      "loss": 2.4278,
+      "step": 46950
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0001860671940382899,
+      "loss": 2.4191,
+      "step": 46960
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00018602600493312647,
+      "loss": 2.4168,
+      "step": 46970
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00018598481294504837,
+      "loss": 2.4298,
+      "step": 46980
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00018594361807735185,
+      "loss": 2.4154,
+      "step": 46990
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001859024203333335,
+      "loss": 2.4576,
+      "step": 47000
+    },
+    {
+      "epoch": 6.35,
+      "eval_loss": 2.524200916290283,
+      "eval_runtime": 1269.0927,
+      "eval_samples_per_second": 59.992,
+      "eval_steps_per_second": 5.0,
+      "step": 47000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00018586121971629003,
+      "loss": 2.4382,
+      "step": 47010
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00018582001622951858,
+      "loss": 2.4301,
+      "step": 47020
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00018577880987631623,
+      "loss": 2.4483,
+      "step": 47030
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00018573760065998058,
+      "loss": 2.415,
+      "step": 47040
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00018569638858380918,
+      "loss": 2.4593,
+      "step": 47050
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00018565517365110005,
+      "loss": 2.4253,
+      "step": 47060
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00018561395586515127,
+      "loss": 2.4342,
+      "step": 47070
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00018557273522926135,
+      "loss": 2.4278,
+      "step": 47080
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00018553151174672874,
+      "loss": 2.432,
+      "step": 47090
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00018549028542085236,
+      "loss": 2.4109,
+      "step": 47100
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001854490562549312,
+      "loss": 2.4158,
+      "step": 47110
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00018540782425226468,
+      "loss": 2.4136,
+      "step": 47120
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00018536658941615216,
+      "loss": 2.4238,
+      "step": 47130
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0001853253517498935,
+      "loss": 2.3896,
+      "step": 47140
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00018528411125678858,
+      "loss": 2.3975,
+      "step": 47150
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00018524286794013763,
+      "loss": 2.4139,
+      "step": 47160
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0001852016218032411,
+      "loss": 2.4319,
+      "step": 47170
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00018516037284939965,
+      "loss": 2.4363,
+      "step": 47180
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00018511912108191408,
+      "loss": 2.4373,
+      "step": 47190
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00018507786650408551,
+      "loss": 2.4372,
+      "step": 47200
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00018503660911921527,
+      "loss": 2.4199,
+      "step": 47210
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00018499534893060492,
+      "loss": 2.4104,
+      "step": 47220
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00018495408594155623,
+      "loss": 2.3844,
+      "step": 47230
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001849128201553712,
+      "loss": 2.4263,
+      "step": 47240
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00018487155157535196,
+      "loss": 2.4327,
+      "step": 47250
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00018483028020480107,
+      "loss": 2.4286,
+      "step": 47260
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00018478900604702116,
+      "loss": 2.4475,
+      "step": 47270
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00018474772910531513,
+      "loss": 2.4277,
+      "step": 47280
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.000184706449382986,
+      "loss": 2.4154,
+      "step": 47290
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00018466516688333724,
+      "loss": 2.4218,
+      "step": 47300
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00018462388160967228,
+      "loss": 2.4185,
+      "step": 47310
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.000184582593565295,
+      "loss": 2.4215,
+      "step": 47320
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00018454130275350934,
+      "loss": 2.4162,
+      "step": 47330
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001845000091776196,
+      "loss": 2.4189,
+      "step": 47340
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00018445871284093006,
+      "loss": 2.3936,
+      "step": 47350
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00018441741374674553,
+      "loss": 2.4113,
+      "step": 47360
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00018437611189837084,
+      "loss": 2.3972,
+      "step": 47370
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00018433480729911113,
+      "loss": 2.4552,
+      "step": 47380
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00018429349995227167,
+      "loss": 2.432,
+      "step": 47390
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00018425218986115806,
+      "loss": 2.4329,
+      "step": 47400
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00018421087702907596,
+      "loss": 2.4379,
+      "step": 47410
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001841695614593315,
+      "loss": 2.4367,
+      "step": 47420
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001841282431552308,
+      "loss": 2.447,
+      "step": 47430
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001840869221200803,
+      "loss": 2.4534,
+      "step": 47440
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00018404559835718658,
+      "loss": 2.4294,
+      "step": 47450
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00018400427186985658,
+      "loss": 2.4579,
+      "step": 47460
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00018396294266139737,
+      "loss": 2.4136,
+      "step": 47470
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001839216107351162,
+      "loss": 2.422,
+      "step": 47480
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001838802760943206,
+      "loss": 2.4242,
+      "step": 47490
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00018383893874231833,
+      "loss": 2.4364,
+      "step": 47500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001837975986824172,
+      "loss": 2.4399,
+      "step": 47510
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00018375625591792558,
+      "loss": 2.4325,
+      "step": 47520
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00018371491045215174,
+      "loss": 2.4137,
+      "step": 47530
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00018367356228840426,
+      "loss": 2.4195,
+      "step": 47540
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00018363221142999198,
+      "loss": 2.398,
+      "step": 47550
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00018359085788022389,
+      "loss": 2.4459,
+      "step": 47560
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00018354950164240925,
+      "loss": 2.4363,
+      "step": 47570
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00018350814271985758,
+      "loss": 2.418,
+      "step": 47580
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001834667811158785,
+      "loss": 2.4388,
+      "step": 47590
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00018342541683378185,
+      "loss": 2.4287,
+      "step": 47600
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00018338404987687778,
+      "loss": 2.4602,
+      "step": 47610
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001833426802484766,
+      "loss": 2.4225,
+      "step": 47620
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00018330130795188884,
+      "loss": 2.4282,
+      "step": 47630
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00018325993299042528,
+      "loss": 2.4193,
+      "step": 47640
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00018321855536739676,
+      "loss": 2.3947,
+      "step": 47650
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00018317717508611455,
+      "loss": 2.4353,
+      "step": 47660
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00018313579214988997,
+      "loss": 2.4067,
+      "step": 47670
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00018309440656203474,
+      "loss": 2.4447,
+      "step": 47680
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001830530183258605,
+      "loss": 2.4432,
+      "step": 47690
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00018301162744467937,
+      "loss": 2.4356,
+      "step": 47700
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00018297023392180355,
+      "loss": 2.4176,
+      "step": 47710
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00018292883776054543,
+      "loss": 2.4125,
+      "step": 47720
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00018288743896421774,
+      "loss": 2.4272,
+      "step": 47730
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00018284603753613334,
+      "loss": 2.4124,
+      "step": 47740
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00018280463347960523,
+      "loss": 2.4418,
+      "step": 47750
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00018276322679794675,
+      "loss": 2.4383,
+      "step": 47760
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00018272181749447137,
+      "loss": 2.4287,
+      "step": 47770
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00018268040557249286,
+      "loss": 2.4473,
+      "step": 47780
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00018263899103532503,
+      "loss": 2.4011,
+      "step": 47790
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001825975738862821,
+      "loss": 2.4261,
+      "step": 47800
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00018255615412867828,
+      "loss": 2.4361,
+      "step": 47810
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00018251473176582824,
+      "loss": 2.4021,
+      "step": 47820
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00018247330680104662,
+      "loss": 2.4058,
+      "step": 47830
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00018243187923764847,
+      "loss": 2.4461,
+      "step": 47840
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001823904490789489,
+      "loss": 2.4156,
+      "step": 47850
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00018234901632826328,
+      "loss": 2.4222,
+      "step": 47860
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001823075809889072,
+      "loss": 2.4379,
+      "step": 47870
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00018226614306419647,
+      "loss": 2.4106,
+      "step": 47880
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00018222470255744704,
+      "loss": 2.4263,
+      "step": 47890
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00018218325947197517,
+      "loss": 2.4426,
+      "step": 47900
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00018214181381109718,
+      "loss": 2.4142,
+      "step": 47910
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00018210036557812976,
+      "loss": 2.4439,
+      "step": 47920
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00018205891477638967,
+      "loss": 2.4484,
+      "step": 47930
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00018201746140919403,
+      "loss": 2.4397,
+      "step": 47940
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00018197600547985993,
+      "loss": 2.4241,
+      "step": 47950
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001819345469917049,
+      "loss": 2.4068,
+      "step": 47960
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00018189723216731552,
+      "loss": 2.4217,
+      "step": 47970
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00018185576882654095,
+      "loss": 2.4494,
+      "step": 47980
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00018181430293656717,
+      "loss": 2.4293,
+      "step": 47990
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00018177283450071237,
+      "loss": 2.4343,
+      "step": 48000
+    },
+    {
+      "epoch": 6.48,
+      "eval_loss": 2.51920485496521,
+      "eval_runtime": 1269.1416,
+      "eval_samples_per_second": 59.989,
+      "eval_steps_per_second": 4.999,
+      "step": 48000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.000181731363522295,
+      "loss": 2.439,
+      "step": 48010
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00018168989000463367,
+      "loss": 2.4372,
+      "step": 48020
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00018164841395104732,
+      "loss": 2.4401,
+      "step": 48030
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001816069353648549,
+      "loss": 2.409,
+      "step": 48040
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00018156545424937568,
+      "loss": 2.4308,
+      "step": 48050
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00018152397060792905,
+      "loss": 2.4192,
+      "step": 48060
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00018148248444383485,
+      "loss": 2.4217,
+      "step": 48070
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001814409957604127,
+      "loss": 2.4151,
+      "step": 48080
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00018139950456098284,
+      "loss": 2.4377,
+      "step": 48090
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00018135801084886543,
+      "loss": 2.4306,
+      "step": 48100
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00018131651462738094,
+      "loss": 2.4153,
+      "step": 48110
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00018127501589985004,
+      "loss": 2.4268,
+      "step": 48120
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00018123351466959364,
+      "loss": 2.4053,
+      "step": 48130
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00018119201093993268,
+      "loss": 2.4353,
+      "step": 48140
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00018115050471418853,
+      "loss": 2.408,
+      "step": 48150
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00018110899599568255,
+      "loss": 2.4664,
+      "step": 48160
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00018106748478773652,
+      "loss": 2.4006,
+      "step": 48170
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001810259710936722,
+      "loss": 2.4056,
+      "step": 48180
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00018098445491681171,
+      "loss": 2.4207,
+      "step": 48190
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00018094293626047723,
+      "loss": 2.4536,
+      "step": 48200
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001809014151279913,
+      "loss": 2.4044,
+      "step": 48210
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001808598915226765,
+      "loss": 2.4349,
+      "step": 48220
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00018081836544785575,
+      "loss": 2.4451,
+      "step": 48230
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.000180776836906852,
+      "loss": 2.4533,
+      "step": 48240
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00018073530590298863,
+      "loss": 2.4258,
+      "step": 48250
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00018069377243958894,
+      "loss": 2.4356,
+      "step": 48260
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00018065223651997668,
+      "loss": 2.4422,
+      "step": 48270
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00018061069814747564,
+      "loss": 2.4243,
+      "step": 48280
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0001805691573254099,
+      "loss": 2.465,
+      "step": 48290
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00018052761405710352,
+      "loss": 2.4028,
+      "step": 48300
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0001804860683458811,
+      "loss": 2.4408,
+      "step": 48310
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001804445201950672,
+      "loss": 2.4346,
+      "step": 48320
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00018040296960798667,
+      "loss": 2.423,
+      "step": 48330
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001803614165879645,
+      "loss": 2.427,
+      "step": 48340
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00018031986113832584,
+      "loss": 2.4177,
+      "step": 48350
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00018027830326239615,
+      "loss": 2.4327,
+      "step": 48360
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00018023674296350098,
+      "loss": 2.4291,
+      "step": 48370
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00018019518024496613,
+      "loss": 2.4331,
+      "step": 48380
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00018015361511011763,
+      "loss": 2.445,
+      "step": 48390
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001801120475622816,
+      "loss": 2.4391,
+      "step": 48400
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001800704776047844,
+      "loss": 2.4424,
+      "step": 48410
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00018002890524095258,
+      "loss": 2.4094,
+      "step": 48420
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.000179987330474113,
+      "loss": 2.4389,
+      "step": 48430
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00017994575330759245,
+      "loss": 2.4044,
+      "step": 48440
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001799041737447182,
+      "loss": 2.4346,
+      "step": 48450
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00017986259178881745,
+      "loss": 2.4335,
+      "step": 48460
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00017982100744321782,
+      "loss": 2.4541,
+      "step": 48470
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.000179779420711247,
+      "loss": 2.4235,
+      "step": 48480
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00017973783159623286,
+      "loss": 2.4437,
+      "step": 48490
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00017969624010150353,
+      "loss": 2.4389,
+      "step": 48500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00017965464623038725,
+      "loss": 2.4278,
+      "step": 48510
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001796130499862125,
+      "loss": 2.4313,
+      "step": 48520
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00017957145137230803,
+      "loss": 2.4308,
+      "step": 48530
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00017952985039200258,
+      "loss": 2.4458,
+      "step": 48540
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001794882470486253,
+      "loss": 2.4339,
+      "step": 48550
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00017944664134550526,
+      "loss": 2.4169,
+      "step": 48560
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00017940503328597198,
+      "loss": 2.4321,
+      "step": 48570
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001793634228733551,
+      "loss": 2.4548,
+      "step": 48580
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001793218101109844,
+      "loss": 2.4301,
+      "step": 48590
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00017928019500218977,
+      "loss": 2.4239,
+      "step": 48600
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017923857755030152,
+      "loss": 2.4441,
+      "step": 48610
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017919695775864993,
+      "loss": 2.4252,
+      "step": 48620
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017915533563056556,
+      "loss": 2.414,
+      "step": 48630
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017911371116937917,
+      "loss": 2.4485,
+      "step": 48640
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017907208437842169,
+      "loss": 2.4132,
+      "step": 48650
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017903045526102414,
+      "loss": 2.4464,
+      "step": 48660
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017898882382051784,
+      "loss": 2.4277,
+      "step": 48670
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00017894719006023436,
+      "loss": 2.44,
+      "step": 48680
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001789055539835053,
+      "loss": 2.4523,
+      "step": 48690
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001788639155936625,
+      "loss": 2.4156,
+      "step": 48700
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00017882227489403798,
+      "loss": 2.4047,
+      "step": 48710
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00017878063188796404,
+      "loss": 2.4436,
+      "step": 48720
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.000178738986578773,
+      "loss": 2.4305,
+      "step": 48730
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001786973389697975,
+      "loss": 2.4352,
+      "step": 48740
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00017865568906437032,
+      "loss": 2.4437,
+      "step": 48750
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001786140368658244,
+      "loss": 2.4439,
+      "step": 48760
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001785723823774928,
+      "loss": 2.4526,
+      "step": 48770
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00017853072560270898,
+      "loss": 2.4612,
+      "step": 48780
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001784890665448064,
+      "loss": 2.4298,
+      "step": 48790
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00017844740520711868,
+      "loss": 2.4224,
+      "step": 48800
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00017840574159297978,
+      "loss": 2.4458,
+      "step": 48810
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001783640757057237,
+      "loss": 2.438,
+      "step": 48820
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00017832240754868473,
+      "loss": 2.4198,
+      "step": 48830
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00017828073712519726,
+      "loss": 2.4561,
+      "step": 48840
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00017823906443859588,
+      "loss": 2.4146,
+      "step": 48850
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00017819738949221534,
+      "loss": 2.4198,
+      "step": 48860
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00017815571228939066,
+      "loss": 2.4207,
+      "step": 48870
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00017811403283345692,
+      "loss": 2.4211,
+      "step": 48880
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001780723511277495,
+      "loss": 2.4122,
+      "step": 48890
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00017803066717560388,
+      "loss": 2.3985,
+      "step": 48900
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00017798898098035578,
+      "loss": 2.4029,
+      "step": 48910
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00017794729254534095,
+      "loss": 2.4214,
+      "step": 48920
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00017790560187389557,
+      "loss": 2.3944,
+      "step": 48930
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0001778639089693557,
+      "loss": 2.4512,
+      "step": 48940
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00017782221383505793,
+      "loss": 2.3924,
+      "step": 48950
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0001777805164743387,
+      "loss": 2.45,
+      "step": 48960
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00017773881689053477,
+      "loss": 2.3966,
+      "step": 48970
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00017769711508698312,
+      "loss": 2.4289,
+      "step": 48980
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00017765541106702087,
+      "loss": 2.4291,
+      "step": 48990
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00017761370483398523,
+      "loss": 2.426,
+      "step": 49000
+    },
+    {
+      "epoch": 6.62,
+      "eval_loss": 2.5139381885528564,
+      "eval_runtime": 1269.1203,
+      "eval_samples_per_second": 59.99,
+      "eval_steps_per_second": 5.0,
+      "step": 49000
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00017757199639121374,
+      "loss": 2.418,
+      "step": 49010
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.000177530285742044,
+      "loss": 2.423,
+      "step": 49020
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001774885728898139,
+      "loss": 2.446,
+      "step": 49030
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00017744685783786131,
+      "loss": 2.4315,
+      "step": 49040
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00017740514058952453,
+      "loss": 2.4257,
+      "step": 49050
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001773634211481418,
+      "loss": 2.4195,
+      "step": 49060
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00017732169951705166,
+      "loss": 2.4299,
+      "step": 49070
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00017727997569959286,
+      "loss": 2.4125,
+      "step": 49080
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00017723824969910427,
+      "loss": 2.4359,
+      "step": 49090
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00017719652151892488,
+      "loss": 2.4263,
+      "step": 49100
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001771547911623939,
+      "loss": 2.4185,
+      "step": 49110
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00017711305863285084,
+      "loss": 2.4269,
+      "step": 49120
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00017707132393363518,
+      "loss": 2.4268,
+      "step": 49130
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00017702958706808665,
+      "loss": 2.4388,
+      "step": 49140
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00017698784803954522,
+      "loss": 2.4035,
+      "step": 49150
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001769461068513509,
+      "loss": 2.42,
+      "step": 49160
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00017690436350684406,
+      "loss": 2.4491,
+      "step": 49170
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00017686261800936505,
+      "loss": 2.4031,
+      "step": 49180
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00017682087036225452,
+      "loss": 2.4447,
+      "step": 49190
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00017677912056885322,
+      "loss": 2.4088,
+      "step": 49200
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001767373686325021,
+      "loss": 2.4127,
+      "step": 49210
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001766956145565423,
+      "loss": 2.4034,
+      "step": 49220
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00017665385834431517,
+      "loss": 2.412,
+      "step": 49230
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.000176612099999162,
+      "loss": 2.4295,
+      "step": 49240
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00017657033952442466,
+      "loss": 2.435,
+      "step": 49250
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00017652857692344474,
+      "loss": 2.4006,
+      "step": 49260
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00017648681219956434,
+      "loss": 2.4085,
+      "step": 49270
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00017644504535612555,
+      "loss": 2.4228,
+      "step": 49280
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00017640327639647073,
+      "loss": 2.4053,
+      "step": 49290
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00017636150532394235,
+      "loss": 2.4148,
+      "step": 49300
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00017631973214188304,
+      "loss": 2.4347,
+      "step": 49310
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001762779568536356,
+      "loss": 2.4197,
+      "step": 49320
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00017623617946254313,
+      "loss": 2.4396,
+      "step": 49330
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00017619439997194864,
+      "loss": 2.4113,
+      "step": 49340
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001761526183851956,
+      "loss": 2.4368,
+      "step": 49350
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001761108347056274,
+      "loss": 2.4061,
+      "step": 49360
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00017606904893658772,
+      "loss": 2.4189,
+      "step": 49370
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001760272610814204,
+      "loss": 2.4203,
+      "step": 49380
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00017598547114346954,
+      "loss": 2.4375,
+      "step": 49390
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00017594367912607912,
+      "loss": 2.3955,
+      "step": 49400
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00017590188503259355,
+      "loss": 2.417,
+      "step": 49410
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00017586008886635735,
+      "loss": 2.4173,
+      "step": 49420
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001758182906307152,
+      "loss": 2.43,
+      "step": 49430
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00017577649032901184,
+      "loss": 2.4127,
+      "step": 49440
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00017573468796459234,
+      "loss": 2.4116,
+      "step": 49450
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00017569288354080183,
+      "loss": 2.4191,
+      "step": 49460
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00017565107706098558,
+      "loss": 2.4192,
+      "step": 49470
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00017560926852848918,
+      "loss": 2.4052,
+      "step": 49480
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001755674579466582,
+      "loss": 2.4179,
+      "step": 49490
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001755256453188385,
+      "loss": 2.4099,
+      "step": 49500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00017548383064837605,
+      "loss": 2.4264,
+      "step": 49510
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00017544201393861696,
+      "loss": 2.4412,
+      "step": 49520
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00017540019519290755,
+      "loss": 2.4306,
+      "step": 49530
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00017535837441459435,
+      "loss": 2.4211,
+      "step": 49540
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00017531655160702388,
+      "loss": 2.4118,
+      "step": 49550
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00017527472677354304,
+      "loss": 2.4135,
+      "step": 49560
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00017523289991749867,
+      "loss": 2.4277,
+      "step": 49570
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.000175191071042238,
+      "loss": 2.4101,
+      "step": 49580
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00017514924015110827,
+      "loss": 2.4253,
+      "step": 49590
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001751074072474569,
+      "loss": 2.4158,
+      "step": 49600
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001750655723346315,
+      "loss": 2.4012,
+      "step": 49610
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001750237354159798,
+      "loss": 2.4288,
+      "step": 49620
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00017498189649484982,
+      "loss": 2.4089,
+      "step": 49630
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00017494005557458953,
+      "loss": 2.4172,
+      "step": 49640
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00017489821265854726,
+      "loss": 2.4151,
+      "step": 49650
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00017485636775007134,
+      "loss": 2.4395,
+      "step": 49660
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001748145208525104,
+      "loss": 2.4215,
+      "step": 49670
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00017477267196921306,
+      "loss": 2.4444,
+      "step": 49680
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00017473082110352835,
+      "loss": 2.4276,
+      "step": 49690
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00017468896825880518,
+      "loss": 2.3968,
+      "step": 49700
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00017464711343839287,
+      "loss": 2.417,
+      "step": 49710
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00017460525664564057,
+      "loss": 2.4467,
+      "step": 49720
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.000174563397883898,
+      "loss": 2.4561,
+      "step": 49730
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00017452153715651474,
+      "loss": 2.416,
+      "step": 49740
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00017447967446684062,
+      "loss": 2.422,
+      "step": 49750
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0001744378098182256,
+      "loss": 2.4274,
+      "step": 49760
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00017439594321401992,
+      "loss": 2.4574,
+      "step": 49770
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00017435407465757376,
+      "loss": 2.4034,
+      "step": 49780
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00017431220415223766,
+      "loss": 2.4341,
+      "step": 49790
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00017427033170136215,
+      "loss": 2.4562,
+      "step": 49800
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00017422845730829812,
+      "loss": 2.4278,
+      "step": 49810
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001741865809763963,
+      "loss": 2.4246,
+      "step": 49820
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.000174144702709008,
+      "loss": 2.4244,
+      "step": 49830
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00017410282250948428,
+      "loss": 2.4396,
+      "step": 49840
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00017406094038117657,
+      "loss": 2.4521,
+      "step": 49850
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00017401905632743642,
+      "loss": 2.4352,
+      "step": 49860
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00017397717035161548,
+      "loss": 2.4211,
+      "step": 49870
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001739352824570657,
+      "loss": 2.4493,
+      "step": 49880
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.000173893392647139,
+      "loss": 2.4523,
+      "step": 49890
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00017385150092518752,
+      "loss": 2.4139,
+      "step": 49900
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00017380960729456368,
+      "loss": 2.4442,
+      "step": 49910
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00017376771175861978,
+      "loss": 2.4246,
+      "step": 49920
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00017372581432070856,
+      "loss": 2.4098,
+      "step": 49930
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00017368391498418274,
+      "loss": 2.4192,
+      "step": 49940
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00017364201375239527,
+      "loss": 2.4354,
+      "step": 49950
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00017360011062869915,
+      "loss": 2.4358,
+      "step": 49960
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00017355820561644765,
+      "loss": 2.4135,
+      "step": 49970
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00017351629871899412,
+      "loss": 2.4453,
+      "step": 49980
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001734743899396921,
+      "loss": 2.4354,
+      "step": 49990
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00017343247928189526,
+      "loss": 2.4136,
+      "step": 50000
+    },
+    {
+      "epoch": 6.75,
+      "eval_loss": 2.508427143096924,
+      "eval_runtime": 1269.0734,
+      "eval_samples_per_second": 59.993,
+      "eval_steps_per_second": 5.0,
+      "step": 50000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00017339056674895744,
+      "loss": 2.4184,
+      "step": 50010
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001733486523442325,
+      "loss": 2.3999,
+      "step": 50020
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00017330673607107475,
+      "loss": 2.4114,
+      "step": 50030
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00017326481793283834,
+      "loss": 2.4218,
+      "step": 50040
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001732228979328777,
+      "loss": 2.437,
+      "step": 50050
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00017318097607454746,
+      "loss": 2.4249,
+      "step": 50060
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00017313905236120226,
+      "loss": 2.4288,
+      "step": 50070
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00017309712679619704,
+      "loss": 2.4458,
+      "step": 50080
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00017305519938288678,
+      "loss": 2.4387,
+      "step": 50090
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00017301327012462662,
+      "loss": 2.4234,
+      "step": 50100
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00017297133902477194,
+      "loss": 2.4307,
+      "step": 50110
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001729294060866781,
+      "loss": 2.4384,
+      "step": 50120
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001728874713137008,
+      "loss": 2.4242,
+      "step": 50130
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00017284553470919576,
+      "loss": 2.4318,
+      "step": 50140
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00017280359627651893,
+      "loss": 2.4351,
+      "step": 50150
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00017276165601902622,
+      "loss": 2.4444,
+      "step": 50160
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00017271971394007391,
+      "loss": 2.4279,
+      "step": 50170
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00017267777004301835,
+      "loss": 2.4285,
+      "step": 50180
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00017263582433121606,
+      "loss": 2.4452,
+      "step": 50190
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00017259387680802353,
+      "loss": 2.4323,
+      "step": 50200
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00017255192747679766,
+      "loss": 2.4254,
+      "step": 50210
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00017250997634089528,
+      "loss": 2.4068,
+      "step": 50220
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00017246802340367356,
+      "loss": 2.4161,
+      "step": 50230
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00017242606866848961,
+      "loss": 2.4303,
+      "step": 50240
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00017238411213870084,
+      "loss": 2.4369,
+      "step": 50250
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001723421538176647,
+      "loss": 2.4202,
+      "step": 50260
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00017230019370873884,
+      "loss": 2.4115,
+      "step": 50270
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00017225823181528106,
+      "loss": 2.4443,
+      "step": 50280
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00017221626814064927,
+      "loss": 2.4013,
+      "step": 50290
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00017217430268820152,
+      "loss": 2.4149,
+      "step": 50300
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00017213233546129605,
+      "loss": 2.4431,
+      "step": 50310
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00017209036646329115,
+      "loss": 2.4246,
+      "step": 50320
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00017204839569754543,
+      "loss": 2.4201,
+      "step": 50330
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00017200642316741741,
+      "loss": 2.45,
+      "step": 50340
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00017196444887626592,
+      "loss": 2.418,
+      "step": 50350
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00017192247282744985,
+      "loss": 2.4209,
+      "step": 50360
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00017188049502432824,
+      "loss": 2.4415,
+      "step": 50370
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001718385154702603,
+      "loss": 2.4135,
+      "step": 50380
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00017179653416860545,
+      "loss": 2.4064,
+      "step": 50390
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00017175455112272302,
+      "loss": 2.4215,
+      "step": 50400
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00017171256633597275,
+      "loss": 2.4241,
+      "step": 50410
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00017167057981171428,
+      "loss": 2.4384,
+      "step": 50420
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00017162859155330762,
+      "loss": 2.4401,
+      "step": 50430
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00017158660156411273,
+      "loss": 2.4426,
+      "step": 50440
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00017154460984748985,
+      "loss": 2.4265,
+      "step": 50450
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0001715026164067992,
+      "loss": 2.4476,
+      "step": 50460
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00017146062124540127,
+      "loss": 2.4226,
+      "step": 50470
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00017141862436665666,
+      "loss": 2.4354,
+      "step": 50480
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0001713766257739261,
+      "loss": 2.4206,
+      "step": 50490
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0001713346254705704,
+      "loss": 2.4011,
+      "step": 50500
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00017129262345995063,
+      "loss": 2.3865,
+      "step": 50510
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0001712506197454278,
+      "loss": 2.4251,
+      "step": 50520
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0001712086143303633,
+      "loss": 2.4241,
+      "step": 50530
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001711666072181185,
+      "loss": 2.4144,
+      "step": 50540
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00017112459841205497,
+      "loss": 2.4302,
+      "step": 50550
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001710825879155343,
+      "loss": 2.4211,
+      "step": 50560
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00017104057573191837,
+      "loss": 2.4213,
+      "step": 50570
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00017099856186456912,
+      "loss": 2.4287,
+      "step": 50580
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001709565463168487,
+      "loss": 2.4157,
+      "step": 50590
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00017091452909211917,
+      "loss": 2.4209,
+      "step": 50600
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.000170872510193743,
+      "loss": 2.4276,
+      "step": 50610
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0001708304896250826,
+      "loss": 2.4386,
+      "step": 50620
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00017078846738950064,
+      "loss": 2.4267,
+      "step": 50630
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00017074644349035988,
+      "loss": 2.4087,
+      "step": 50640
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0001707044179310232,
+      "loss": 2.4206,
+      "step": 50650
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00017066239071485356,
+      "loss": 2.4272,
+      "step": 50660
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00017062036184521418,
+      "loss": 2.4405,
+      "step": 50670
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0001705783313254683,
+      "loss": 2.4312,
+      "step": 50680
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00017053629915897938,
+      "loss": 2.4402,
+      "step": 50690
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00017049426534911094,
+      "loss": 2.4131,
+      "step": 50700
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00017045222989922666,
+      "loss": 2.4578,
+      "step": 50710
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00017041019281269034,
+      "loss": 2.4254,
+      "step": 50720
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0001703681540928659,
+      "loss": 2.4529,
+      "step": 50730
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00017032611374311743,
+      "loss": 2.4278,
+      "step": 50740
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0001702840717668092,
+      "loss": 2.4524,
+      "step": 50750
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00017024202816730546,
+      "loss": 2.4582,
+      "step": 50760
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00017019998294797066,
+      "loss": 2.4172,
+      "step": 50770
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00017015793611216946,
+      "loss": 2.3983,
+      "step": 50780
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00017011588766326655,
+      "loss": 2.4037,
+      "step": 50790
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001700738376046268,
+      "loss": 2.3946,
+      "step": 50800
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001700317859396151,
+      "loss": 2.4456,
+      "step": 50810
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00016998973267159668,
+      "loss": 2.4177,
+      "step": 50820
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00016994767780393664,
+      "loss": 2.4253,
+      "step": 50830
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00016990562134000048,
+      "loss": 2.4566,
+      "step": 50840
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0001698635632831536,
+      "loss": 2.4181,
+      "step": 50850
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00016982150363676168,
+      "loss": 2.4237,
+      "step": 50860
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00016977944240419038,
+      "loss": 2.4231,
+      "step": 50870
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00016973737958880566,
+      "loss": 2.4131,
+      "step": 50880
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0001696953151939735,
+      "loss": 2.4182,
+      "step": 50890
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00016965324922306002,
+      "loss": 2.4214,
+      "step": 50900
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00016961118167943147,
+      "loss": 2.4113,
+      "step": 50910
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00016956911256645418,
+      "loss": 2.422,
+      "step": 50920
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0001695270418874947,
+      "loss": 2.4352,
+      "step": 50930
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00016948496964591965,
+      "loss": 2.4063,
+      "step": 50940
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00016944289584509584,
+      "loss": 2.4382,
+      "step": 50950
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00016940082048839004,
+      "loss": 2.4218,
+      "step": 50960
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00016935874357916933,
+      "loss": 2.3886,
+      "step": 50970
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001693166651208008,
+      "loss": 2.4259,
+      "step": 50980
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001692745851166518,
+      "loss": 2.4374,
+      "step": 50990
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001692325035700895,
+      "loss": 2.4463,
+      "step": 51000
+    },
+    {
+      "epoch": 6.89,
+      "eval_loss": 2.5036706924438477,
+      "eval_runtime": 1266.3668,
+      "eval_samples_per_second": 60.121,
+      "eval_steps_per_second": 5.01,
+      "step": 51000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001691904204844816,
+      "loss": 2.4009,
+      "step": 51010
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00016914833586319561,
+      "loss": 2.4192,
+      "step": 51020
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00016910624970959933,
+      "loss": 2.422,
+      "step": 51030
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00016906416202706057,
+      "loss": 2.4292,
+      "step": 51040
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001690220728189474,
+      "loss": 2.413,
+      "step": 51050
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00016897998208862785,
+      "loss": 2.4395,
+      "step": 51060
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00016893788983947015,
+      "loss": 2.4453,
+      "step": 51070
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00016889579607484275,
+      "loss": 2.4345,
+      "step": 51080
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00016885370079811407,
+      "loss": 2.4535,
+      "step": 51090
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00016881160401265269,
+      "loss": 2.4217,
+      "step": 51100
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00016876950572182733,
+      "loss": 2.3907,
+      "step": 51110
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00016872740592900685,
+      "loss": 2.4347,
+      "step": 51120
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00016868530463756017,
+      "loss": 2.4032,
+      "step": 51130
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0001686432018508564,
+      "loss": 2.4185,
+      "step": 51140
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0001686010975722648,
+      "loss": 2.4495,
+      "step": 51150
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00016855899180515457,
+      "loss": 2.4396,
+      "step": 51160
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00016851688455289522,
+      "loss": 2.43,
+      "step": 51170
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00016847477581885623,
+      "loss": 2.4327,
+      "step": 51180
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00016843266560640744,
+      "loss": 2.3907,
+      "step": 51190
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00016839055391891842,
+      "loss": 2.453,
+      "step": 51200
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00016834844075975928,
+      "loss": 2.407,
+      "step": 51210
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00016830632613229986,
+      "loss": 2.4216,
+      "step": 51220
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00016826421003991046,
+      "loss": 2.4298,
+      "step": 51230
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001682220924859613,
+      "loss": 2.4323,
+      "step": 51240
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00016817997347382273,
+      "loss": 2.4157,
+      "step": 51250
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00016813785300686524,
+      "loss": 2.4274,
+      "step": 51260
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001680957310884595,
+      "loss": 2.4391,
+      "step": 51270
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0001680536077219762,
+      "loss": 2.4411,
+      "step": 51280
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00016801148291078618,
+      "loss": 2.4135,
+      "step": 51290
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0001679693566582604,
+      "loss": 2.4163,
+      "step": 51300
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00016792722896777,
+      "loss": 2.4289,
+      "step": 51310
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00016788509984268606,
+      "loss": 2.4178,
+      "step": 51320
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00016784296928637992,
+      "loss": 2.4393,
+      "step": 51330
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00016780083730222307,
+      "loss": 2.3959,
+      "step": 51340
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.000167758703893587,
+      "loss": 2.4157,
+      "step": 51350
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00016771656906384335,
+      "loss": 2.3975,
+      "step": 51360
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00016767443281636384,
+      "loss": 2.4207,
+      "step": 51370
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00016763229515452045,
+      "loss": 2.4342,
+      "step": 51380
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00016759015608168513,
+      "loss": 2.4087,
+      "step": 51390
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0001675480156012299,
+      "loss": 2.4041,
+      "step": 51400
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00016750587371652708,
+      "loss": 2.438,
+      "step": 51410
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00016746373043094893,
+      "loss": 2.4218,
+      "step": 51420
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00016742158574786792,
+      "loss": 2.4331,
+      "step": 51430
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00016737943967065657,
+      "loss": 2.4343,
+      "step": 51440
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00016733729220268762,
+      "loss": 2.4106,
+      "step": 51450
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001672951433473338,
+      "loss": 2.4398,
+      "step": 51460
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00016725299310796788,
+      "loss": 2.4304,
+      "step": 51470
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00016721084148796304,
+      "loss": 2.4165,
+      "step": 51480
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001671686884906923,
+      "loss": 2.4086,
+      "step": 51490
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00016712653411952888,
+      "loss": 2.4027,
+      "step": 51500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0001670885940135916,
+      "loss": 2.4215,
+      "step": 51510
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00016704643704132572,
+      "loss": 2.4302,
+      "step": 51520
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00016700427870495004,
+      "loss": 2.424,
+      "step": 51530
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00016696211900783835,
+      "loss": 2.4212,
+      "step": 51540
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00016691995795336425,
+      "loss": 2.4015,
+      "step": 51550
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00016687779554490172,
+      "loss": 2.3958,
+      "step": 51560
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0001668356317858247,
+      "loss": 2.4423,
+      "step": 51570
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0001667934666795072,
+      "loss": 2.4209,
+      "step": 51580
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00016675130022932345,
+      "loss": 2.4277,
+      "step": 51590
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00016670913243864782,
+      "loss": 2.4557,
+      "step": 51600
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00016666696331085458,
+      "loss": 2.4296,
+      "step": 51610
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00016662479284931827,
+      "loss": 2.4251,
+      "step": 51620
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00016658262105741356,
+      "loss": 2.4133,
+      "step": 51630
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00016654044793851518,
+      "loss": 2.4304,
+      "step": 51640
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00016649827349599788,
+      "loss": 2.4311,
+      "step": 51650
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00016645609773323664,
+      "loss": 2.394,
+      "step": 51660
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00016641392065360649,
+      "loss": 2.4141,
+      "step": 51670
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00016637174226048253,
+      "loss": 2.4032,
+      "step": 51680
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00016632956255724005,
+      "loss": 2.4238,
+      "step": 51690
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00016628738154725442,
+      "loss": 2.3856,
+      "step": 51700
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00016624519923390108,
+      "loss": 2.4068,
+      "step": 51710
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00016620301562055557,
+      "loss": 2.4314,
+      "step": 51720
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00016616083071059354,
+      "loss": 2.4041,
+      "step": 51730
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00016611864450739088,
+      "loss": 2.4153,
+      "step": 51740
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00016607645701432333,
+      "loss": 2.4202,
+      "step": 51750
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00016603426823476693,
+      "loss": 2.3951,
+      "step": 51760
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00016599207817209773,
+      "loss": 2.4116,
+      "step": 51770
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001659498868296919,
+      "loss": 2.4161,
+      "step": 51780
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001659076942109258,
+      "loss": 2.4023,
+      "step": 51790
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00016586550031917575,
+      "loss": 2.423,
+      "step": 51800
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00016582330515781823,
+      "loss": 2.4052,
+      "step": 51810
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00016578110873022986,
+      "loss": 2.3956,
+      "step": 51820
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00016573891103978736,
+      "loss": 2.4131,
+      "step": 51830
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00016569671208986747,
+      "loss": 2.351,
+      "step": 51840
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0001656545118838471,
+      "loss": 2.3669,
+      "step": 51850
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00016561231042510328,
+      "loss": 2.3666,
+      "step": 51860
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00016557010771701302,
+      "loss": 2.356,
+      "step": 51870
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00016552790376295353,
+      "loss": 2.3847,
+      "step": 51880
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00016548569856630218,
+      "loss": 2.3918,
+      "step": 51890
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00016544349213043637,
+      "loss": 2.3812,
+      "step": 51900
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0001654012844587335,
+      "loss": 2.3656,
+      "step": 51910
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00016535907555457119,
+      "loss": 2.3414,
+      "step": 51920
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00016531686542132714,
+      "loss": 2.3855,
+      "step": 51930
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00016527465406237918,
+      "loss": 2.3851,
+      "step": 51940
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00016523244148110514,
+      "loss": 2.3781,
+      "step": 51950
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00016519022768088303,
+      "loss": 2.3689,
+      "step": 51960
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00016514801266509093,
+      "loss": 2.3736,
+      "step": 51970
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00016510579643710696,
+      "loss": 2.3578,
+      "step": 51980
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00016506357900030946,
+      "loss": 2.3636,
+      "step": 51990
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00016502136035807688,
+      "loss": 2.345,
+      "step": 52000
+    },
+    {
+      "epoch": 7.02,
+      "eval_loss": 2.501615047454834,
+      "eval_runtime": 1269.0509,
+      "eval_samples_per_second": 59.994,
+      "eval_steps_per_second": 5.0,
+      "step": 52000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00016497914051378753,
+      "loss": 2.3666,
+      "step": 52010
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00016493691947082003,
+      "loss": 2.4002,
+      "step": 52020
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00016489469723255306,
+      "loss": 2.3886,
+      "step": 52030
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00016485247380236542,
+      "loss": 2.3433,
+      "step": 52040
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0001648102491836359,
+      "loss": 2.3796,
+      "step": 52050
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00016476802337974349,
+      "loss": 2.3937,
+      "step": 52060
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00016472579639406715,
+      "loss": 2.3452,
+      "step": 52070
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0001646835682299861,
+      "loss": 2.3601,
+      "step": 52080
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0001646413388908795,
+      "loss": 2.3738,
+      "step": 52090
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0001645991083801268,
+      "loss": 2.3512,
+      "step": 52100
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00016455687670110727,
+      "loss": 2.3835,
+      "step": 52110
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0001645146438572005,
+      "loss": 2.3721,
+      "step": 52120
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00016447240985178612,
+      "loss": 2.3917,
+      "step": 52130
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0001644301746882438,
+      "loss": 2.3671,
+      "step": 52140
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00016438793836995324,
+      "loss": 2.3676,
+      "step": 52150
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0001643457009002945,
+      "loss": 2.3975,
+      "step": 52160
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00016430346228264747,
+      "loss": 2.3852,
+      "step": 52170
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00016426122252039213,
+      "loss": 2.3669,
+      "step": 52180
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00016421898161690876,
+      "loss": 2.3785,
+      "step": 52190
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.0001641767395755776,
+      "loss": 2.3786,
+      "step": 52200
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00016413449639977895,
+      "loss": 2.3495,
+      "step": 52210
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00016409225209289323,
+      "loss": 2.3903,
+      "step": 52220
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00016405000665830098,
+      "loss": 2.3202,
+      "step": 52230
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00016400776009938288,
+      "loss": 2.3615,
+      "step": 52240
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00016396551241951954,
+      "loss": 2.3572,
+      "step": 52250
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0001639232636220918,
+      "loss": 2.3885,
+      "step": 52260
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0001638810137104805,
+      "loss": 2.3325,
+      "step": 52270
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00016383876268806665,
+      "loss": 2.3542,
+      "step": 52280
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0001637965105582313,
+      "loss": 2.3847,
+      "step": 52290
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00016375425732435565,
+      "loss": 2.3982,
+      "step": 52300
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00016371200298982084,
+      "loss": 2.3845,
+      "step": 52310
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.00016366974755800825,
+      "loss": 2.3684,
+      "step": 52320
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.00016362749103229925,
+      "loss": 2.374,
+      "step": 52330
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.00016358523341607545,
+      "loss": 2.3724,
+      "step": 52340
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0001635429747127183,
+      "loss": 2.3868,
+      "step": 52350
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.00016350071492560962,
+      "loss": 2.3679,
+      "step": 52360
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.00016345845405813103,
+      "loss": 2.3927,
+      "step": 52370
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.00016341619211366444,
+      "loss": 2.3767,
+      "step": 52380
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00016337392909559177,
+      "loss": 2.3394,
+      "step": 52390
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.0001633316650072951,
+      "loss": 2.386,
+      "step": 52400
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00016328939985215643,
+      "loss": 2.3692,
+      "step": 52410
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00016324713363355807,
+      "loss": 2.3919,
+      "step": 52420
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.0001632048663548822,
+      "loss": 2.4083,
+      "step": 52430
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00016316259801951122,
+      "loss": 2.3866,
+      "step": 52440
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.0001631203286308276,
+      "loss": 2.3528,
+      "step": 52450
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00016307805819221384,
+      "loss": 2.3646,
+      "step": 52460
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00016303578670705254,
+      "loss": 2.3532,
+      "step": 52470
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.0001629935141787264,
+      "loss": 2.3677,
+      "step": 52480
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00016295124061061823,
+      "loss": 2.3711,
+      "step": 52490
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.0001629089660061109,
+      "loss": 2.3689,
+      "step": 52500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00016286669036858734,
+      "loss": 2.3691,
+      "step": 52510
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00016282441370143053,
+      "loss": 2.3867,
+      "step": 52520
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00016278213600802368,
+      "loss": 2.4009,
+      "step": 52530
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.0001627398572917499,
+      "loss": 2.3961,
+      "step": 52540
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00016269757755599254,
+      "loss": 2.3894,
+      "step": 52550
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.0001626552968041349,
+      "loss": 2.3839,
+      "step": 52560
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00016261301503956048,
+      "loss": 2.4152,
+      "step": 52570
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00016257073226565272,
+      "loss": 2.3554,
+      "step": 52580
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.0001625284484857952,
+      "loss": 2.3944,
+      "step": 52590
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00016248616370337173,
+      "loss": 2.3651,
+      "step": 52600
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00016244387792176606,
+      "loss": 2.3657,
+      "step": 52610
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00016240159114436187,
+      "loss": 2.353,
+      "step": 52620
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00016235930337454323,
+      "loss": 2.4128,
+      "step": 52630
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00016231701461569408,
+      "loss": 2.406,
+      "step": 52640
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00016227472487119859,
+      "loss": 2.3702,
+      "step": 52650
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00016223243414444078,
+      "loss": 2.3783,
+      "step": 52660
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.000162190142438805,
+      "loss": 2.3655,
+      "step": 52670
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00016214784975767546,
+      "loss": 2.3777,
+      "step": 52680
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00016210555610443665,
+      "loss": 2.3936,
+      "step": 52690
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00016206326148247299,
+      "loss": 2.3755,
+      "step": 52700
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00016202096589516907,
+      "loss": 2.3673,
+      "step": 52710
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0001619786693459095,
+      "loss": 2.3877,
+      "step": 52720
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00016193637183807894,
+      "loss": 2.3831,
+      "step": 52730
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0001618940733750622,
+      "loss": 2.3859,
+      "step": 52740
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0001618517739602442,
+      "loss": 2.3247,
+      "step": 52750
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00016180947359700976,
+      "loss": 2.3614,
+      "step": 52760
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00016176717228874398,
+      "loss": 2.3757,
+      "step": 52770
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00016172487003883186,
+      "loss": 2.3934,
+      "step": 52780
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00016168256685065866,
+      "loss": 2.3608,
+      "step": 52790
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0001616402627276095,
+      "loss": 2.3759,
+      "step": 52800
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00016159795767306984,
+      "loss": 2.3819,
+      "step": 52810
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0001615556516904249,
+      "loss": 2.3816,
+      "step": 52820
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00016151334478306024,
+      "loss": 2.3761,
+      "step": 52830
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00016147103695436136,
+      "loss": 2.3789,
+      "step": 52840
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0001614287282077139,
+      "loss": 2.3732,
+      "step": 52850
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0001613864185465035,
+      "loss": 2.3915,
+      "step": 52860
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00016134410797411595,
+      "loss": 2.3755,
+      "step": 52870
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00016130179649393697,
+      "loss": 2.3602,
+      "step": 52880
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00016125948410935262,
+      "loss": 2.3775,
+      "step": 52890
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00016121717082374874,
+      "loss": 2.3959,
+      "step": 52900
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00016117485664051144,
+      "loss": 2.3682,
+      "step": 52910
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00016113254156302682,
+      "loss": 2.3939,
+      "step": 52920
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00016109022559468106,
+      "loss": 2.3582,
+      "step": 52930
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0001610479087388604,
+      "loss": 2.3912,
+      "step": 52940
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00016100559099895126,
+      "loss": 2.3843,
+      "step": 52950
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00016096327237833988,
+      "loss": 2.388,
+      "step": 52960
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0001609209528804129,
+      "loss": 2.3856,
+      "step": 52970
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0001608786325085567,
+      "loss": 2.4109,
+      "step": 52980
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.000160836311266158,
+      "loss": 2.3768,
+      "step": 52990
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00016079398915660344,
+      "loss": 2.3736,
+      "step": 53000
+    },
+    {
+      "epoch": 7.16,
+      "eval_loss": 2.498974323272705,
+      "eval_runtime": 1269.1099,
+      "eval_samples_per_second": 59.991,
+      "eval_steps_per_second": 5.0,
+      "step": 53000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0001607516661832798,
+      "loss": 2.3692,
+      "step": 53010
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00016070934234957386,
+      "loss": 2.3728,
+      "step": 53020
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00016066701765887252,
+      "loss": 2.3631,
+      "step": 53030
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0001606246921145627,
+      "loss": 2.4021,
+      "step": 53040
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00016058236572003153,
+      "loss": 2.38,
+      "step": 53050
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00016054003847866598,
+      "loss": 2.3843,
+      "step": 53060
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0001604977103938533,
+      "loss": 2.3883,
+      "step": 53070
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00016045538146898067,
+      "loss": 2.3977,
+      "step": 53080
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00016041305170743535,
+      "loss": 2.3998,
+      "step": 53090
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00016037072111260478,
+      "loss": 2.3981,
+      "step": 53100
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00016032838968787637,
+      "loss": 2.378,
+      "step": 53110
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00016028605743663755,
+      "loss": 2.3811,
+      "step": 53120
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00016024372436227593,
+      "loss": 2.3739,
+      "step": 53130
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00016020139046817915,
+      "loss": 2.3814,
+      "step": 53140
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00016015905575773493,
+      "loss": 2.3703,
+      "step": 53150
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00016011672023433088,
+      "loss": 2.3645,
+      "step": 53160
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00016007438390135505,
+      "loss": 2.374,
+      "step": 53170
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0001600320467621951,
+      "loss": 2.4049,
+      "step": 53180
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0001599897088202391,
+      "loss": 2.3718,
+      "step": 53190
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00015994737007887503,
+      "loss": 2.3586,
+      "step": 53200
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00015990503054149106,
+      "loss": 2.3959,
+      "step": 53210
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00015986269021147522,
+      "loss": 2.3931,
+      "step": 53220
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00015982034909221577,
+      "loss": 2.3994,
+      "step": 53230
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00015977800718710094,
+      "loss": 2.3769,
+      "step": 53240
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00015973566449951915,
+      "loss": 2.3743,
+      "step": 53250
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00015969332103285867,
+      "loss": 2.4115,
+      "step": 53260
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.0001596509767905081,
+      "loss": 2.3871,
+      "step": 53270
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0001596086317758558,
+      "loss": 2.3739,
+      "step": 53280
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0001595662859922905,
+      "loss": 2.3925,
+      "step": 53290
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00015952393944320076,
+      "loss": 2.3945,
+      "step": 53300
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0001594815921319753,
+      "loss": 2.3841,
+      "step": 53310
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0001594392440620029,
+      "loss": 2.3876,
+      "step": 53320
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00015939689523667238,
+      "loss": 2.3755,
+      "step": 53330
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0001593545456593726,
+      "loss": 2.3854,
+      "step": 53340
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00015931219533349257,
+      "loss": 2.3593,
+      "step": 53350
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00015926984426242123,
+      "loss": 2.3575,
+      "step": 53360
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00015922749244954768,
+      "loss": 2.4062,
+      "step": 53370
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00015918513989826105,
+      "loss": 2.3784,
+      "step": 53380
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0001591427866119505,
+      "loss": 2.4051,
+      "step": 53390
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00015910043259400526,
+      "loss": 2.4096,
+      "step": 53400
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00015905807784781474,
+      "loss": 2.4,
+      "step": 53410
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00015901572237676817,
+      "loss": 2.3893,
+      "step": 53420
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.000158973366184255,
+      "loss": 2.3792,
+      "step": 53430
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00015893100927366475,
+      "loss": 2.3567,
+      "step": 53440
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00015888865164838699,
+      "loss": 2.426,
+      "step": 53450
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00015884629331181115,
+      "loss": 2.3917,
+      "step": 53460
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00015880393426732708,
+      "loss": 2.3577,
+      "step": 53470
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00015876157451832432,
+      "loss": 2.373,
+      "step": 53480
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00015871921406819265,
+      "loss": 2.3895,
+      "step": 53490
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.000158676852920322,
+      "loss": 2.3697,
+      "step": 53500
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00015863872729347323,
+      "loss": 2.3958,
+      "step": 53510
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00015859636482923756,
+      "loss": 2.3718,
+      "step": 53520
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00015855400167709356,
+      "loss": 2.408,
+      "step": 53530
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00015851163784043137,
+      "loss": 2.3926,
+      "step": 53540
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.000158469273322641,
+      "loss": 2.3638,
+      "step": 53550
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00015842690812711268,
+      "loss": 2.3928,
+      "step": 53560
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0001583845422572365,
+      "loss": 2.3419,
+      "step": 53570
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00015834217571640278,
+      "loss": 2.3828,
+      "step": 53580
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00015829980850800182,
+      "loss": 2.3953,
+      "step": 53590
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00015825744063542398,
+      "loss": 2.4096,
+      "step": 53600
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00015821507210205964,
+      "loss": 2.3768,
+      "step": 53610
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00015817270291129933,
+      "loss": 2.4019,
+      "step": 53620
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00015813033306653348,
+      "loss": 2.3957,
+      "step": 53630
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.0001580879625711527,
+      "loss": 2.3887,
+      "step": 53640
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0001580455914285476,
+      "loss": 2.3974,
+      "step": 53650
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0001580032196421089,
+      "loss": 2.386,
+      "step": 53660
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00015796084721522724,
+      "loss": 2.3749,
+      "step": 53670
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00015791847415129345,
+      "loss": 2.3716,
+      "step": 53680
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00015787610045369836,
+      "loss": 2.4072,
+      "step": 53690
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00015783372612583287,
+      "loss": 2.4066,
+      "step": 53700
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00015779135117108785,
+      "loss": 2.4103,
+      "step": 53710
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00015774897559285429,
+      "loss": 2.3711,
+      "step": 53720
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00015770659939452322,
+      "loss": 2.3433,
+      "step": 53730
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00015766422257948575,
+      "loss": 2.4116,
+      "step": 53740
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00015762184515113296,
+      "loss": 2.3872,
+      "step": 53750
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00015757946711285608,
+      "loss": 2.3717,
+      "step": 53760
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0001575370884680463,
+      "loss": 2.3443,
+      "step": 53770
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00015749470922009484,
+      "loss": 2.392,
+      "step": 53780
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0001574523293723931,
+      "loss": 2.3626,
+      "step": 53790
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0001574099489283325,
+      "loss": 2.3829,
+      "step": 53800
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0001573675678913043,
+      "loss": 2.3798,
+      "step": 53810
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00015732518626470008,
+      "loss": 2.3873,
+      "step": 53820
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0001572828040519113,
+      "loss": 2.3962,
+      "step": 53830
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00015724042125632955,
+      "loss": 2.3441,
+      "step": 53840
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0001571980378813464,
+      "loss": 2.3832,
+      "step": 53850
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0001571556539303536,
+      "loss": 2.3811,
+      "step": 53860
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0001571132694067427,
+      "loss": 2.3864,
+      "step": 53870
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0001570708843139055,
+      "loss": 2.391,
+      "step": 53880
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00015702849865523383,
+      "loss": 2.374,
+      "step": 53890
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00015698611243411954,
+      "loss": 2.3615,
+      "step": 53900
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00015694372565395442,
+      "loss": 2.4158,
+      "step": 53910
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00015690133831813052,
+      "loss": 2.3971,
+      "step": 53920
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0001568589504300396,
+      "loss": 2.4231,
+      "step": 53930
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00015681656199307387,
+      "loss": 2.3763,
+      "step": 53940
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00015677417301062532,
+      "loss": 2.3637,
+      "step": 53950
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00015673178348608608,
+      "loss": 2.3753,
+      "step": 53960
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00015668939342284823,
+      "loss": 2.3662,
+      "step": 53970
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.000156647002824304,
+      "loss": 2.3817,
+      "step": 53980
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0001566046116938456,
+      "loss": 2.3897,
+      "step": 53990
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00015656222003486534,
+      "loss": 2.4092,
+      "step": 54000
+    },
+    {
+      "epoch": 7.29,
+      "eval_loss": 2.495549440383911,
+      "eval_runtime": 1269.0814,
+      "eval_samples_per_second": 59.992,
+      "eval_steps_per_second": 5.0,
+      "step": 54000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0001565198278507555,
+      "loss": 2.4011,
+      "step": 54010
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00015647743514490841,
+      "loss": 2.3827,
+      "step": 54020
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0001564350419207165,
+      "loss": 2.3895,
+      "step": 54030
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00015639264818157223,
+      "loss": 2.4156,
+      "step": 54040
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00015635025393086806,
+      "loss": 2.3716,
+      "step": 54050
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00015630785917199655,
+      "loss": 2.3799,
+      "step": 54060
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00015626546390835015,
+      "loss": 2.421,
+      "step": 54070
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00015622306814332162,
+      "loss": 2.4113,
+      "step": 54080
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00015618067188030347,
+      "loss": 2.3873,
+      "step": 54090
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00015613827512268847,
+      "loss": 2.4048,
+      "step": 54100
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00015609587787386928,
+      "loss": 2.3748,
+      "step": 54110
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00015605348013723872,
+      "loss": 2.3931,
+      "step": 54120
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0001560110819161895,
+      "loss": 2.3927,
+      "step": 54130
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00015596868321411458,
+      "loss": 2.3976,
+      "step": 54140
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00015592628403440672,
+      "loss": 2.3699,
+      "step": 54150
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00015588388438045895,
+      "loss": 2.3865,
+      "step": 54160
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00015584148425566415,
+      "loss": 2.3903,
+      "step": 54170
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0001557990836634153,
+      "loss": 2.3865,
+      "step": 54180
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00015575668260710547,
+      "loss": 2.3699,
+      "step": 54190
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00015571428109012776,
+      "loss": 2.3503,
+      "step": 54200
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0001556718791158752,
+      "loss": 2.3842,
+      "step": 54210
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00015562947668774097,
+      "loss": 2.3857,
+      "step": 54220
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00015558707380911824,
+      "loss": 2.3985,
+      "step": 54230
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00015554467048340016,
+      "loss": 2.3772,
+      "step": 54240
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.0001555022667139801,
+      "loss": 2.3845,
+      "step": 54250
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00015545986250425128,
+      "loss": 2.4131,
+      "step": 54260
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00015541745785760695,
+      "loss": 2.3746,
+      "step": 54270
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.0001553750527774406,
+      "loss": 2.3819,
+      "step": 54280
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.0001553326472671455,
+      "loss": 2.3479,
+      "step": 54290
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00015529024133011517,
+      "loss": 2.3855,
+      "step": 54300
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.000155247834969743,
+      "loss": 2.4035,
+      "step": 54310
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.00015520542818942253,
+      "loss": 2.382,
+      "step": 54320
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0001551630209925472,
+      "loss": 2.3877,
+      "step": 54330
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.00015512061338251063,
+      "loss": 2.3987,
+      "step": 54340
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.00015507820536270642,
+      "loss": 2.3585,
+      "step": 54350
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.00015503579693652823,
+      "loss": 2.41,
+      "step": 54360
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0001549933881073696,
+      "loss": 2.4075,
+      "step": 54370
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0001549509788786243,
+      "loss": 2.4046,
+      "step": 54380
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00015490856925368598,
+      "loss": 2.368,
+      "step": 54390
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00015486615923594854,
+      "loss": 2.4147,
+      "step": 54400
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0001548237488288056,
+      "loss": 2.3625,
+      "step": 54410
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00015478133803565111,
+      "loss": 2.3899,
+      "step": 54420
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0001547389268598788,
+      "loss": 2.351,
+      "step": 54430
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0001546965153048826,
+      "loss": 2.4011,
+      "step": 54440
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00015465410337405639,
+      "loss": 2.4328,
+      "step": 54450
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00015461169107079418,
+      "loss": 2.3649,
+      "step": 54460
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00015456927839848983,
+      "loss": 2.3886,
+      "step": 54470
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00015452686536053745,
+      "loss": 2.3691,
+      "step": 54480
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00015448445196033096,
+      "loss": 2.402,
+      "step": 54490
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0001544420382012645,
+      "loss": 2.4017,
+      "step": 54500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00015439962408673205,
+      "loss": 2.3755,
+      "step": 54510
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00015435720962012788,
+      "loss": 2.413,
+      "step": 54520
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00015431479480484595,
+      "loss": 2.4007,
+      "step": 54530
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00015427237964428053,
+      "loss": 2.3774,
+      "step": 54540
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0001542299641418258,
+      "loss": 2.3731,
+      "step": 54550
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.000154187548300876,
+      "loss": 2.352,
+      "step": 54560
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00015414513212482533,
+      "loss": 2.3826,
+      "step": 54570
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00015410271561706816,
+      "loss": 2.3807,
+      "step": 54580
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0001540602987809986,
+      "loss": 2.3928,
+      "step": 54590
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0001540178816200112,
+      "loss": 2.4112,
+      "step": 54600
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0001539754641375002,
+      "loss": 2.3764,
+      "step": 54610
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00015393304633686005,
+      "loss": 2.4027,
+      "step": 54620
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0001538906282214851,
+      "loss": 2.3737,
+      "step": 54630
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00015384820979476974,
+      "loss": 2.391,
+      "step": 54640
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00015380579106010847,
+      "loss": 2.4182,
+      "step": 54650
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00015376337202089584,
+      "loss": 2.3786,
+      "step": 54660
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00015372095268052628,
+      "loss": 2.3708,
+      "step": 54670
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00015367853304239432,
+      "loss": 2.3843,
+      "step": 54680
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00015363611310989452,
+      "loss": 2.3619,
+      "step": 54690
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0001535936928864215,
+      "loss": 2.3856,
+      "step": 54700
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00015355127237536977,
+      "loss": 2.4118,
+      "step": 54710
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00015350885158013405,
+      "loss": 2.4135,
+      "step": 54720
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0001534664305041089,
+      "loss": 2.3996,
+      "step": 54730
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00015342400915068906,
+      "loss": 2.3949,
+      "step": 54740
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0001533815875232692,
+      "loss": 2.4016,
+      "step": 54750
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.00015333916562524402,
+      "loss": 2.3921,
+      "step": 54760
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0001532967434600083,
+      "loss": 2.3949,
+      "step": 54770
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0001532543210309567,
+      "loss": 2.4179,
+      "step": 54780
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.00015321189834148404,
+      "loss": 2.3816,
+      "step": 54790
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0001531694753949852,
+      "loss": 2.4064,
+      "step": 54800
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.00015312705219485486,
+      "loss": 2.3717,
+      "step": 54810
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.00015308462874448798,
+      "loss": 2.4069,
+      "step": 54820
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00015304220504727936,
+      "loss": 2.4026,
+      "step": 54830
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00015299978110662384,
+      "loss": 2.3758,
+      "step": 54840
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00015295735692591637,
+      "loss": 2.3761,
+      "step": 54850
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0001529149325085519,
+      "loss": 2.396,
+      "step": 54860
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0001528725078579253,
+      "loss": 2.3905,
+      "step": 54870
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00015283008297743158,
+      "loss": 2.4021,
+      "step": 54880
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0001527876578704656,
+      "loss": 2.4063,
+      "step": 54890
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0001527452325404225,
+      "loss": 2.3735,
+      "step": 54900
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.0001527028069906972,
+      "loss": 2.3916,
+      "step": 54910
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00015266038122468476,
+      "loss": 2.377,
+      "step": 54920
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.0001526179552457802,
+      "loss": 2.3909,
+      "step": 54930
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00015257552905737856,
+      "loss": 2.3785,
+      "step": 54940
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.000152533102662875,
+      "loss": 2.3817,
+      "step": 54950
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00015249067606566455,
+      "loss": 2.3779,
+      "step": 54960
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00015244824926914234,
+      "loss": 2.3642,
+      "step": 54970
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0001524058222767035,
+      "loss": 2.4137,
+      "step": 54980
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00015236339509174313,
+      "loss": 2.3739,
+      "step": 54990
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00015232096771765645,
+      "loss": 2.3689,
+      "step": 55000
+    },
+    {
+      "epoch": 7.43,
+      "eval_loss": 2.491696357727051,
+      "eval_runtime": 1269.0901,
+      "eval_samples_per_second": 59.992,
+      "eval_steps_per_second": 5.0,
+      "step": 55000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0001522785401578386,
+      "loss": 2.3798,
+      "step": 55010
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00015223611241568476,
+      "loss": 2.3472,
+      "step": 55020
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00015219368449459014,
+      "loss": 2.3876,
+      "step": 55030
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00015215125639794996,
+      "loss": 2.3859,
+      "step": 55040
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00015210882812915946,
+      "loss": 2.3973,
+      "step": 55050
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00015206639969161386,
+      "loss": 2.3579,
+      "step": 55060
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00015202397108870843,
+      "loss": 2.3997,
+      "step": 55070
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.0001519815423238385,
+      "loss": 2.3811,
+      "step": 55080
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00015193911340039917,
+      "loss": 2.4171,
+      "step": 55090
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00015189668432178595,
+      "loss": 2.4079,
+      "step": 55100
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.000151854255091394,
+      "loss": 2.3845,
+      "step": 55110
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00015181182571261882,
+      "loss": 2.382,
+      "step": 55120
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00015176939618885552,
+      "loss": 2.374,
+      "step": 55130
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00015172696652349956,
+      "loss": 2.37,
+      "step": 55140
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00015168453671994628,
+      "loss": 2.3853,
+      "step": 55150
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.0001516421067815911,
+      "loss": 2.4129,
+      "step": 55160
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.0001515996767118293,
+      "loss": 2.3929,
+      "step": 55170
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00015155724651405637,
+      "loss": 2.4043,
+      "step": 55180
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00015151481619166756,
+      "loss": 2.3979,
+      "step": 55190
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00015147238574805846,
+      "loss": 2.383,
+      "step": 55200
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.0001514299551866244,
+      "loss": 2.3746,
+      "step": 55210
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.0001513875245107608,
+      "loss": 2.4096,
+      "step": 55220
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00015134509372386308,
+      "loss": 2.3669,
+      "step": 55230
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.0001513026628293267,
+      "loss": 2.388,
+      "step": 55240
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00015126023183054715,
+      "loss": 2.4047,
+      "step": 55250
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.0001512178007309199,
+      "loss": 2.3876,
+      "step": 55260
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00015117536953384035,
+      "loss": 2.3622,
+      "step": 55270
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00015113293824270405,
+      "loss": 2.3936,
+      "step": 55280
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00015109050686090637,
+      "loss": 2.3752,
+      "step": 55290
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.000151048075391843,
+      "loss": 2.4035,
+      "step": 55300
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00015100564383890922,
+      "loss": 2.4041,
+      "step": 55310
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00015096321220550073,
+      "loss": 2.4009,
+      "step": 55320
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00015092078049501294,
+      "loss": 2.3807,
+      "step": 55330
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00015087834871084136,
+      "loss": 2.3979,
+      "step": 55340
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00015083591685638155,
+      "loss": 2.3754,
+      "step": 55350
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00015079348493502904,
+      "loss": 2.3546,
+      "step": 55360
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00015075105295017936,
+      "loss": 2.399,
+      "step": 55370
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00015070862090522805,
+      "loss": 2.3907,
+      "step": 55380
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00015066618880357066,
+      "loss": 2.3923,
+      "step": 55390
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0001506237566486027,
+      "loss": 2.397,
+      "step": 55400
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00015058132444371977,
+      "loss": 2.3958,
+      "step": 55410
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0001505388921923175,
+      "loss": 2.3976,
+      "step": 55420
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.00015049645989779133,
+      "loss": 2.3802,
+      "step": 55430
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.00015045402756353683,
+      "loss": 2.3918,
+      "step": 55440
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.00015041159519294965,
+      "loss": 2.3794,
+      "step": 55450
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0001503691627894253,
+      "loss": 2.3951,
+      "step": 55460
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.00015032673035635943,
+      "loss": 2.3966,
+      "step": 55470
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.00015028429789714753,
+      "loss": 2.3741,
+      "step": 55480
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0001502418654151852,
+      "loss": 2.4115,
+      "step": 55490
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00015019943291386803,
+      "loss": 2.3954,
+      "step": 55500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0001501570003965916,
+      "loss": 2.3732,
+      "step": 55510
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00015011456786675155,
+      "loss": 2.3809,
+      "step": 55520
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0001500721353277434,
+      "loss": 2.4006,
+      "step": 55530
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00015002970278296275,
+      "loss": 2.4026,
+      "step": 55540
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00014998727023580517,
+      "loss": 2.3866,
+      "step": 55550
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00014994483768966628,
+      "loss": 2.3986,
+      "step": 55560
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00014990240514794171,
+      "loss": 2.3746,
+      "step": 55570
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00014986421586697017,
+      "loss": 2.4081,
+      "step": 55580
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00014982178334298747,
+      "loss": 2.3785,
+      "step": 55590
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00014977935083326625,
+      "loss": 2.3879,
+      "step": 55600
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00014973691834120208,
+      "loss": 2.3909,
+      "step": 55610
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.0001496944858701905,
+      "loss": 2.3808,
+      "step": 55620
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00014965205342362713,
+      "loss": 2.4131,
+      "step": 55630
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.0001496096210049076,
+      "loss": 2.3817,
+      "step": 55640
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.00014956718861742737,
+      "loss": 2.3674,
+      "step": 55650
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0001495247562645821,
+      "loss": 2.4019,
+      "step": 55660
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0001494823239497673,
+      "loss": 2.4173,
+      "step": 55670
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.00014943989167637864,
+      "loss": 2.3774,
+      "step": 55680
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0001493974594478116,
+      "loss": 2.3991,
+      "step": 55690
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0001493550272674617,
+      "loss": 2.3883,
+      "step": 55700
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.00014931259513872457,
+      "loss": 2.3742,
+      "step": 55710
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00014927016306499584,
+      "loss": 2.3661,
+      "step": 55720
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00014922773104967087,
+      "loss": 2.3832,
+      "step": 55730
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00014918529909614531,
+      "loss": 2.402,
+      "step": 55740
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00014914286720781468,
+      "loss": 2.4098,
+      "step": 55750
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00014910043538807463,
+      "loss": 2.3815,
+      "step": 55760
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0001490580036403205,
+      "loss": 2.3613,
+      "step": 55770
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00014901557196794787,
+      "loss": 2.3666,
+      "step": 55780
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0001489731403743523,
+      "loss": 2.3887,
+      "step": 55790
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00014893070886292932,
+      "loss": 2.3725,
+      "step": 55800
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00014888827743707433,
+      "loss": 2.3926,
+      "step": 55810
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0001488458461001829,
+      "loss": 2.3647,
+      "step": 55820
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00014880341485565054,
+      "loss": 2.3527,
+      "step": 55830
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00014876098370687272,
+      "loss": 2.3828,
+      "step": 55840
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00014871855265724486,
+      "loss": 2.4066,
+      "step": 55850
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00014867612171016247,
+      "loss": 2.3771,
+      "step": 55860
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.000148633690869021,
+      "loss": 2.4037,
+      "step": 55870
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00014859126013721594,
+      "loss": 2.4069,
+      "step": 55880
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00014854882951814265,
+      "loss": 2.387,
+      "step": 55890
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.0001485063990151966,
+      "loss": 2.3872,
+      "step": 55900
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00014846396863177322,
+      "loss": 2.4004,
+      "step": 55910
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.000148421538371268,
+      "loss": 2.3886,
+      "step": 55920
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.0001483791082370762,
+      "loss": 2.3934,
+      "step": 55930
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00014833667823259324,
+      "loss": 2.4024,
+      "step": 55940
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00014829424836121456,
+      "loss": 2.4158,
+      "step": 55950
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0001482518186263356,
+      "loss": 2.3883,
+      "step": 55960
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00014820938903135157,
+      "loss": 2.4039,
+      "step": 55970
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00014816695957965787,
+      "loss": 2.3849,
+      "step": 55980
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00014812453027464987,
+      "loss": 2.3864,
+      "step": 55990
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0001480821011197229,
+      "loss": 2.3797,
+      "step": 56000
+    },
+    {
+      "epoch": 7.56,
+      "eval_loss": 2.486668109893799,
+      "eval_runtime": 1269.1258,
+      "eval_samples_per_second": 59.99,
+      "eval_steps_per_second": 5.0,
+      "step": 56000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 111060,
+  "num_train_epochs": 15,
+  "save_steps": 1000,
+  "total_flos": 1.5657271333748736e+19,
+  "trial_name": null,
+  "trial_params": null
+}