diff --git "a/checkpoint-5724/trainer_state.json" "b/checkpoint-5724/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-5724/trainer_state.json"
@@ -0,0 +1,4038 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 5724,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0069881201956673656,
+      "grad_norm": 14.528067588806152,
+      "learning_rate": 1.99748427672956e-05,
+      "loss": 4.6827,
+      "step": 10
+    },
+    {
+      "epoch": 0.013976240391334731,
+      "grad_norm": 13.050820350646973,
+      "learning_rate": 1.994689028651293e-05,
+      "loss": 4.529,
+      "step": 20
+    },
+    {
+      "epoch": 0.020964360587002098,
+      "grad_norm": 2.765507698059082,
+      "learning_rate": 1.991893780573026e-05,
+      "loss": 3.4294,
+      "step": 30
+    },
+    {
+      "epoch": 0.027952480782669462,
+      "grad_norm": 16.95716667175293,
+      "learning_rate": 1.989098532494759e-05,
+      "loss": 3.352,
+      "step": 40
+    },
+    {
+      "epoch": 0.03494060097833683,
+      "grad_norm": 5.339346408843994,
+      "learning_rate": 1.986303284416492e-05,
+      "loss": 2.2654,
+      "step": 50
+    },
+    {
+      "epoch": 0.041928721174004195,
+      "grad_norm": 7.168309688568115,
+      "learning_rate": 1.983508036338225e-05,
+      "loss": 2.0076,
+      "step": 60
+    },
+    {
+      "epoch": 0.04891684136967156,
+      "grad_norm": 5.793034076690674,
+      "learning_rate": 1.9807127882599582e-05,
+      "loss": 1.6779,
+      "step": 70
+    },
+    {
+      "epoch": 0.055904961565338925,
+      "grad_norm": 6.093694686889648,
+      "learning_rate": 1.9779175401816913e-05,
+      "loss": 1.2564,
+      "step": 80
+    },
+    {
+      "epoch": 0.06289308176100629,
+      "grad_norm": 3.321807622909546,
+      "learning_rate": 1.9751222921034243e-05,
+      "loss": 1.4143,
+      "step": 90
+    },
+    {
+      "epoch": 0.06988120195667366,
+      "grad_norm": 2.9472098350524902,
+      "learning_rate": 1.9723270440251574e-05,
+      "loss": 1.0533,
+      "step": 100
+    },
+    {
+      "epoch": 0.07686932215234102,
+      "grad_norm": 8.515447616577148,
+      "learning_rate": 1.9695317959468904e-05,
+      "loss": 1.2895,
+      "step": 110
+    },
+    {
+      "epoch": 0.08385744234800839,
+      "grad_norm": 3.560924530029297,
+      "learning_rate": 1.9667365478686235e-05,
+      "loss": 1.097,
+      "step": 120
+    },
+    {
+      "epoch": 0.09084556254367575,
+      "grad_norm": 4.986349582672119,
+      "learning_rate": 1.9639412997903566e-05,
+      "loss": 0.9556,
+      "step": 130
+    },
+    {
+      "epoch": 0.09783368273934312,
+      "grad_norm": 1.27532160282135,
+      "learning_rate": 1.9611460517120896e-05,
+      "loss": 0.8306,
+      "step": 140
+    },
+    {
+      "epoch": 0.10482180293501048,
+      "grad_norm": 2.6796844005584717,
+      "learning_rate": 1.9583508036338227e-05,
+      "loss": 0.7747,
+      "step": 150
+    },
+    {
+      "epoch": 0.11180992313067785,
+      "grad_norm": 3.4592413902282715,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.6187,
+      "step": 160
+    },
+    {
+      "epoch": 0.1187980433263452,
+      "grad_norm": 4.833241939544678,
+      "learning_rate": 1.9527603074772888e-05,
+      "loss": 0.6831,
+      "step": 170
+    },
+    {
+      "epoch": 0.12578616352201258,
+      "grad_norm": 1.7203434705734253,
+      "learning_rate": 1.949965059399022e-05,
+      "loss": 0.586,
+      "step": 180
+    },
+    {
+      "epoch": 0.13277428371767994,
+      "grad_norm": 1.333950161933899,
+      "learning_rate": 1.947169811320755e-05,
+      "loss": 0.6214,
+      "step": 190
+    },
+    {
+      "epoch": 0.13976240391334732,
+      "grad_norm": 1.5078257322311401,
+      "learning_rate": 1.944374563242488e-05,
+      "loss": 0.5652,
+      "step": 200
+    },
+    {
+      "epoch": 0.14675052410901468,
+      "grad_norm": 3.7645132541656494,
+      "learning_rate": 1.941579315164221e-05,
+      "loss": 0.5229,
+      "step": 210
+    },
+    {
+      "epoch": 0.15373864430468204,
+      "grad_norm": 1.2441000938415527,
+      "learning_rate": 1.938784067085954e-05,
+      "loss": 0.3669,
+      "step": 220
+    },
+    {
+      "epoch": 0.1607267645003494,
+      "grad_norm": 2.2875142097473145,
+      "learning_rate": 1.935988819007687e-05,
+      "loss": 0.4162,
+      "step": 230
+    },
+    {
+      "epoch": 0.16771488469601678,
+      "grad_norm": 2.1816585063934326,
+      "learning_rate": 1.9331935709294202e-05,
+      "loss": 0.3836,
+      "step": 240
+    },
+    {
+      "epoch": 0.17470300489168414,
+      "grad_norm": 3.4903249740600586,
+      "learning_rate": 1.9303983228511532e-05,
+      "loss": 0.4131,
+      "step": 250
+    },
+    {
+      "epoch": 0.1816911250873515,
+      "grad_norm": 2.1905159950256348,
+      "learning_rate": 1.9276030747728863e-05,
+      "loss": 0.3725,
+      "step": 260
+    },
+    {
+      "epoch": 0.18867924528301888,
+      "grad_norm": 3.6003665924072266,
+      "learning_rate": 1.9248078266946194e-05,
+      "loss": 0.4131,
+      "step": 270
+    },
+    {
+      "epoch": 0.19566736547868624,
+      "grad_norm": 3.012568235397339,
+      "learning_rate": 1.9220125786163524e-05,
+      "loss": 0.3991,
+      "step": 280
+    },
+    {
+      "epoch": 0.2026554856743536,
+      "grad_norm": 1.2049503326416016,
+      "learning_rate": 1.9192173305380855e-05,
+      "loss": 0.2732,
+      "step": 290
+    },
+    {
+      "epoch": 0.20964360587002095,
+      "grad_norm": 2.0943682193756104,
+      "learning_rate": 1.9164220824598185e-05,
+      "loss": 0.3052,
+      "step": 300
+    },
+    {
+      "epoch": 0.21663172606568834,
+      "grad_norm": 4.1970086097717285,
+      "learning_rate": 1.9136268343815516e-05,
+      "loss": 0.2544,
+      "step": 310
+    },
+    {
+      "epoch": 0.2236198462613557,
+      "grad_norm": 2.6826064586639404,
+      "learning_rate": 1.9108315863032846e-05,
+      "loss": 0.1921,
+      "step": 320
+    },
+    {
+      "epoch": 0.23060796645702306,
+      "grad_norm": 6.013642311096191,
+      "learning_rate": 1.9080363382250177e-05,
+      "loss": 0.3014,
+      "step": 330
+    },
+    {
+      "epoch": 0.2375960866526904,
+      "grad_norm": 4.2075629234313965,
+      "learning_rate": 1.9052410901467508e-05,
+      "loss": 0.2569,
+      "step": 340
+    },
+    {
+      "epoch": 0.2445842068483578,
+      "grad_norm": 0.5800023078918457,
+      "learning_rate": 1.9024458420684838e-05,
+      "loss": 0.1486,
+      "step": 350
+    },
+    {
+      "epoch": 0.25157232704402516,
+      "grad_norm": 8.123499870300293,
+      "learning_rate": 1.899650593990217e-05,
+      "loss": 0.2215,
+      "step": 360
+    },
+    {
+      "epoch": 0.2585604472396925,
+      "grad_norm": 4.439255237579346,
+      "learning_rate": 1.89685534591195e-05,
+      "loss": 0.1692,
+      "step": 370
+    },
+    {
+      "epoch": 0.2655485674353599,
+      "grad_norm": 4.177631855010986,
+      "learning_rate": 1.894060097833683e-05,
+      "loss": 0.1176,
+      "step": 380
+    },
+    {
+      "epoch": 0.27253668763102723,
+      "grad_norm": 0.5754140019416809,
+      "learning_rate": 1.891264849755416e-05,
+      "loss": 0.1538,
+      "step": 390
+    },
+    {
+      "epoch": 0.27952480782669464,
+      "grad_norm": 0.5612320303916931,
+      "learning_rate": 1.888469601677149e-05,
+      "loss": 0.1054,
+      "step": 400
+    },
+    {
+      "epoch": 0.286512928022362,
+      "grad_norm": 1.949729561805725,
+      "learning_rate": 1.8856743535988818e-05,
+      "loss": 0.0824,
+      "step": 410
+    },
+    {
+      "epoch": 0.29350104821802936,
+      "grad_norm": 3.153822898864746,
+      "learning_rate": 1.8828791055206152e-05,
+      "loss": 0.1174,
+      "step": 420
+    },
+    {
+      "epoch": 0.3004891684136967,
+      "grad_norm": 0.6545250415802002,
+      "learning_rate": 1.8800838574423483e-05,
+      "loss": 0.1305,
+      "step": 430
+    },
+    {
+      "epoch": 0.3074772886093641,
+      "grad_norm": 1.2621614933013916,
+      "learning_rate": 1.8772886093640813e-05,
+      "loss": 0.0684,
+      "step": 440
+    },
+    {
+      "epoch": 0.31446540880503143,
+      "grad_norm": 0.3028416335582733,
+      "learning_rate": 1.8744933612858144e-05,
+      "loss": 0.0863,
+      "step": 450
+    },
+    {
+      "epoch": 0.3214535290006988,
+      "grad_norm": 0.4452091455459595,
+      "learning_rate": 1.8716981132075474e-05,
+      "loss": 0.1016,
+      "step": 460
+    },
+    {
+      "epoch": 0.3284416491963662,
+      "grad_norm": 0.4046468436717987,
+      "learning_rate": 1.8689028651292805e-05,
+      "loss": 0.2317,
+      "step": 470
+    },
+    {
+      "epoch": 0.33542976939203356,
+      "grad_norm": 0.5121694803237915,
+      "learning_rate": 1.8661076170510135e-05,
+      "loss": 0.1965,
+      "step": 480
+    },
+    {
+      "epoch": 0.3424178895877009,
+      "grad_norm": 2.7892794609069824,
+      "learning_rate": 1.8633123689727466e-05,
+      "loss": 0.2082,
+      "step": 490
+    },
+    {
+      "epoch": 0.3494060097833683,
+      "grad_norm": 0.29012227058410645,
+      "learning_rate": 1.8605171208944793e-05,
+      "loss": 0.1393,
+      "step": 500
+    },
+    {
+      "epoch": 0.35639412997903563,
+      "grad_norm": 0.14969919621944427,
+      "learning_rate": 1.8577218728162124e-05,
+      "loss": 0.1917,
+      "step": 510
+    },
+    {
+      "epoch": 0.363382250174703,
+      "grad_norm": 2.5301618576049805,
+      "learning_rate": 1.8549266247379458e-05,
+      "loss": 0.0939,
+      "step": 520
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 0.5171930193901062,
+      "learning_rate": 1.8521313766596788e-05,
+      "loss": 0.0643,
+      "step": 530
+    },
+    {
+      "epoch": 0.37735849056603776,
+      "grad_norm": 0.23746764659881592,
+      "learning_rate": 1.849336128581412e-05,
+      "loss": 0.2094,
+      "step": 540
+    },
+    {
+      "epoch": 0.3843466107617051,
+      "grad_norm": 1.2422730922698975,
+      "learning_rate": 1.846540880503145e-05,
+      "loss": 0.109,
+      "step": 550
+    },
+    {
+      "epoch": 0.3913347309573725,
+      "grad_norm": 0.8466539978981018,
+      "learning_rate": 1.843745632424878e-05,
+      "loss": 0.1514,
+      "step": 560
+    },
+    {
+      "epoch": 0.39832285115303984,
+      "grad_norm": 0.201154425740242,
+      "learning_rate": 1.840950384346611e-05,
+      "loss": 0.1597,
+      "step": 570
+    },
+    {
+      "epoch": 0.4053109713487072,
+      "grad_norm": 0.19054199755191803,
+      "learning_rate": 1.838155136268344e-05,
+      "loss": 0.0955,
+      "step": 580
+    },
+    {
+      "epoch": 0.41229909154437455,
+      "grad_norm": 1.8036576509475708,
+      "learning_rate": 1.8353598881900768e-05,
+      "loss": 0.1369,
+      "step": 590
+    },
+    {
+      "epoch": 0.4192872117400419,
+      "grad_norm": 0.3122437596321106,
+      "learning_rate": 1.83256464011181e-05,
+      "loss": 0.1439,
+      "step": 600
+    },
+    {
+      "epoch": 0.42627533193570927,
+      "grad_norm": 2.6353561878204346,
+      "learning_rate": 1.829769392033543e-05,
+      "loss": 0.1566,
+      "step": 610
+    },
+    {
+      "epoch": 0.4332634521313767,
+      "grad_norm": 3.645707368850708,
+      "learning_rate": 1.8269741439552763e-05,
+      "loss": 0.2505,
+      "step": 620
+    },
+    {
+      "epoch": 0.44025157232704404,
+      "grad_norm": 0.20553793013095856,
+      "learning_rate": 1.8241788958770094e-05,
+      "loss": 0.0509,
+      "step": 630
+    },
+    {
+      "epoch": 0.4472396925227114,
+      "grad_norm": 3.8460299968719482,
+      "learning_rate": 1.8213836477987425e-05,
+      "loss": 0.118,
+      "step": 640
+    },
+    {
+      "epoch": 0.45422781271837875,
+      "grad_norm": 0.4825759828090668,
+      "learning_rate": 1.8185883997204755e-05,
+      "loss": 0.0671,
+      "step": 650
+    },
+    {
+      "epoch": 0.4612159329140461,
+      "grad_norm": 2.5147883892059326,
+      "learning_rate": 1.8157931516422086e-05,
+      "loss": 0.1381,
+      "step": 660
+    },
+    {
+      "epoch": 0.46820405310971347,
+      "grad_norm": 4.72428035736084,
+      "learning_rate": 1.8129979035639413e-05,
+      "loss": 0.1093,
+      "step": 670
+    },
+    {
+      "epoch": 0.4751921733053808,
+      "grad_norm": 3.091625928878784,
+      "learning_rate": 1.8102026554856743e-05,
+      "loss": 0.2363,
+      "step": 680
+    },
+    {
+      "epoch": 0.48218029350104824,
+      "grad_norm": 0.10304142534732819,
+      "learning_rate": 1.8074074074074074e-05,
+      "loss": 0.1347,
+      "step": 690
+    },
+    {
+      "epoch": 0.4891684136967156,
+      "grad_norm": 0.4455307424068451,
+      "learning_rate": 1.8046121593291405e-05,
+      "loss": 0.1064,
+      "step": 700
+    },
+    {
+      "epoch": 0.49615653389238296,
+      "grad_norm": 0.11594414710998535,
+      "learning_rate": 1.8018169112508735e-05,
+      "loss": 0.123,
+      "step": 710
+    },
+    {
+      "epoch": 0.5031446540880503,
+      "grad_norm": 0.1224653497338295,
+      "learning_rate": 1.799021663172607e-05,
+      "loss": 0.1263,
+      "step": 720
+    },
+    {
+      "epoch": 0.5101327742837177,
+      "grad_norm": 4.951889991760254,
+      "learning_rate": 1.79622641509434e-05,
+      "loss": 0.2111,
+      "step": 730
+    },
+    {
+      "epoch": 0.517120894479385,
+      "grad_norm": 0.16316205263137817,
+      "learning_rate": 1.793431167016073e-05,
+      "loss": 0.0366,
+      "step": 740
+    },
+    {
+      "epoch": 0.5241090146750524,
+      "grad_norm": 0.1896398663520813,
+      "learning_rate": 1.790635918937806e-05,
+      "loss": 0.1656,
+      "step": 750
+    },
+    {
+      "epoch": 0.5310971348707197,
+      "grad_norm": 0.3920069932937622,
+      "learning_rate": 1.7878406708595388e-05,
+      "loss": 0.2804,
+      "step": 760
+    },
+    {
+      "epoch": 0.5380852550663872,
+      "grad_norm": 0.12546010315418243,
+      "learning_rate": 1.785045422781272e-05,
+      "loss": 0.1404,
+      "step": 770
+    },
+    {
+      "epoch": 0.5450733752620545,
+      "grad_norm": 5.382510185241699,
+      "learning_rate": 1.782250174703005e-05,
+      "loss": 0.1194,
+      "step": 780
+    },
+    {
+      "epoch": 0.5520614954577219,
+      "grad_norm": 0.3077394366264343,
+      "learning_rate": 1.779454926624738e-05,
+      "loss": 0.1744,
+      "step": 790
+    },
+    {
+      "epoch": 0.5590496156533893,
+      "grad_norm": 0.41757112741470337,
+      "learning_rate": 1.776659678546471e-05,
+      "loss": 0.0729,
+      "step": 800
+    },
+    {
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.17300930619239807,
+      "learning_rate": 1.773864430468204e-05,
+      "loss": 0.1443,
+      "step": 810
+    },
+    {
+      "epoch": 0.573025856044724,
+      "grad_norm": 0.09419082850217819,
+      "learning_rate": 1.7710691823899375e-05,
+      "loss": 0.1587,
+      "step": 820
+    },
+    {
+      "epoch": 0.5800139762403913,
+      "grad_norm": 0.14113612473011017,
+      "learning_rate": 1.7682739343116705e-05,
+      "loss": 0.1196,
+      "step": 830
+    },
+    {
+      "epoch": 0.5870020964360587,
+      "grad_norm": 0.6690120697021484,
+      "learning_rate": 1.7654786862334036e-05,
+      "loss": 0.0619,
+      "step": 840
+    },
+    {
+      "epoch": 0.593990216631726,
+      "grad_norm": 0.5076020956039429,
+      "learning_rate": 1.7626834381551363e-05,
+      "loss": 0.2922,
+      "step": 850
+    },
+    {
+      "epoch": 0.6009783368273934,
+      "grad_norm": 0.12461690604686737,
+      "learning_rate": 1.7598881900768694e-05,
+      "loss": 0.0351,
+      "step": 860
+    },
+    {
+      "epoch": 0.6079664570230608,
+      "grad_norm": 5.236762523651123,
+      "learning_rate": 1.7570929419986024e-05,
+      "loss": 0.1571,
+      "step": 870
+    },
+    {
+      "epoch": 0.6149545772187281,
+      "grad_norm": 0.12950463593006134,
+      "learning_rate": 1.7542976939203355e-05,
+      "loss": 0.0204,
+      "step": 880
+    },
+    {
+      "epoch": 0.6219426974143956,
+      "grad_norm": 0.08026222884654999,
+      "learning_rate": 1.7515024458420685e-05,
+      "loss": 0.1185,
+      "step": 890
+    },
+    {
+      "epoch": 0.6289308176100629,
+      "grad_norm": 4.237085342407227,
+      "learning_rate": 1.7487071977638016e-05,
+      "loss": 0.0876,
+      "step": 900
+    },
+    {
+      "epoch": 0.6359189378057303,
+      "grad_norm": 0.916748046875,
+      "learning_rate": 1.7459119496855346e-05,
+      "loss": 0.0619,
+      "step": 910
+    },
+    {
+      "epoch": 0.6429070580013976,
+      "grad_norm": 4.51674747467041,
+      "learning_rate": 1.743116701607268e-05,
+      "loss": 0.1116,
+      "step": 920
+    },
+    {
+      "epoch": 0.649895178197065,
+      "grad_norm": 1.1469999551773071,
+      "learning_rate": 1.740321453529001e-05,
+      "loss": 0.0147,
+      "step": 930
+    },
+    {
+      "epoch": 0.6568832983927324,
+      "grad_norm": 2.3627333641052246,
+      "learning_rate": 1.7375262054507338e-05,
+      "loss": 0.1485,
+      "step": 940
+    },
+    {
+      "epoch": 0.6638714185883997,
+      "grad_norm": 8.216973304748535,
+      "learning_rate": 1.734730957372467e-05,
+      "loss": 0.0472,
+      "step": 950
+    },
+    {
+      "epoch": 0.6708595387840671,
+      "grad_norm": 3.5901684761047363,
+      "learning_rate": 1.7319357092942e-05,
+      "loss": 0.0979,
+      "step": 960
+    },
+    {
+      "epoch": 0.6778476589797344,
+      "grad_norm": 7.015125274658203,
+      "learning_rate": 1.729140461215933e-05,
+      "loss": 0.0537,
+      "step": 970
+    },
+    {
+      "epoch": 0.6848357791754018,
+      "grad_norm": 4.372157096862793,
+      "learning_rate": 1.726345213137666e-05,
+      "loss": 0.1152,
+      "step": 980
+    },
+    {
+      "epoch": 0.6918238993710691,
+      "grad_norm": 4.862729072570801,
+      "learning_rate": 1.723549965059399e-05,
+      "loss": 0.1131,
+      "step": 990
+    },
+    {
+      "epoch": 0.6988120195667366,
+      "grad_norm": 0.07732150703668594,
+      "learning_rate": 1.720754716981132e-05,
+      "loss": 0.0707,
+      "step": 1000
+    },
+    {
+      "epoch": 0.705800139762404,
+      "grad_norm": 0.19970574975013733,
+      "learning_rate": 1.7179594689028652e-05,
+      "loss": 0.0738,
+      "step": 1010
+    },
+    {
+      "epoch": 0.7127882599580713,
+      "grad_norm": 0.21677608788013458,
+      "learning_rate": 1.7151642208245983e-05,
+      "loss": 0.0992,
+      "step": 1020
+    },
+    {
+      "epoch": 0.7197763801537387,
+      "grad_norm": 0.10738188028335571,
+      "learning_rate": 1.7123689727463313e-05,
+      "loss": 0.0103,
+      "step": 1030
+    },
+    {
+      "epoch": 0.726764500349406,
+      "grad_norm": 3.7014143466949463,
+      "learning_rate": 1.7095737246680644e-05,
+      "loss": 0.0803,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7337526205450734,
+      "grad_norm": 0.04619055986404419,
+      "learning_rate": 1.7067784765897974e-05,
+      "loss": 0.049,
+      "step": 1050
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.2753339409828186,
+      "learning_rate": 1.7039832285115305e-05,
+      "loss": 0.144,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7477288609364081,
+      "grad_norm": 2.8835108280181885,
+      "learning_rate": 1.7011879804332635e-05,
+      "loss": 0.0885,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.22654394805431366,
+      "learning_rate": 1.6983927323549966e-05,
+      "loss": 0.0512,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7617051013277428,
+      "grad_norm": 2.421400547027588,
+      "learning_rate": 1.6955974842767297e-05,
+      "loss": 0.1544,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7686932215234102,
+      "grad_norm": 0.03806561604142189,
+      "learning_rate": 1.6928022361984627e-05,
+      "loss": 0.0225,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7756813417190775,
+      "grad_norm": 0.08347579091787338,
+      "learning_rate": 1.6900069881201958e-05,
+      "loss": 0.1386,
+      "step": 1110
+    },
+    {
+      "epoch": 0.782669461914745,
+      "grad_norm": 0.2059401273727417,
+      "learning_rate": 1.687211740041929e-05,
+      "loss": 0.2406,
+      "step": 1120
+    },
+    {
+      "epoch": 0.7896575821104123,
+      "grad_norm": 0.060870036482810974,
+      "learning_rate": 1.684416491963662e-05,
+      "loss": 0.1158,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7966457023060797,
+      "grad_norm": 0.14614495635032654,
+      "learning_rate": 1.681621243885395e-05,
+      "loss": 0.1938,
+      "step": 1140
+    },
+    {
+      "epoch": 0.803633822501747,
+      "grad_norm": 0.09593985229730606,
+      "learning_rate": 1.678825995807128e-05,
+      "loss": 0.0627,
+      "step": 1150
+    },
+    {
+      "epoch": 0.8106219426974144,
+      "grad_norm": 9.543107032775879,
+      "learning_rate": 1.676030747728861e-05,
+      "loss": 0.1365,
+      "step": 1160
+    },
+    {
+      "epoch": 0.8176100628930818,
+      "grad_norm": 0.09678485244512558,
+      "learning_rate": 1.673235499650594e-05,
+      "loss": 0.1438,
+      "step": 1170
+    },
+    {
+      "epoch": 0.8245981830887491,
+      "grad_norm": 0.24556872248649597,
+      "learning_rate": 1.6704402515723272e-05,
+      "loss": 0.2233,
+      "step": 1180
+    },
+    {
+      "epoch": 0.8315863032844165,
+      "grad_norm": 0.06892254948616028,
+      "learning_rate": 1.6676450034940602e-05,
+      "loss": 0.0601,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8385744234800838,
+      "grad_norm": 0.6429911255836487,
+      "learning_rate": 1.6648497554157933e-05,
+      "loss": 0.132,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8455625436757512,
+      "grad_norm": 0.03480253368616104,
+      "learning_rate": 1.6620545073375263e-05,
+      "loss": 0.0472,
+      "step": 1210
+    },
+    {
+      "epoch": 0.8525506638714185,
+      "grad_norm": 0.25101321935653687,
+      "learning_rate": 1.6592592592592594e-05,
+      "loss": 0.0485,
+      "step": 1220
+    },
+    {
+      "epoch": 0.859538784067086,
+      "grad_norm": 0.0525231771171093,
+      "learning_rate": 1.6564640111809925e-05,
+      "loss": 0.1824,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8665269042627534,
+      "grad_norm": 4.694910526275635,
+      "learning_rate": 1.6536687631027255e-05,
+      "loss": 0.1666,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8735150244584207,
+      "grad_norm": 0.09604217112064362,
+      "learning_rate": 1.6508735150244586e-05,
+      "loss": 0.0638,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8805031446540881,
+      "grad_norm": 9.125068664550781,
+      "learning_rate": 1.6480782669461916e-05,
+      "loss": 0.0744,
+      "step": 1260
+    },
+    {
+      "epoch": 0.8874912648497554,
+      "grad_norm": 0.5258679986000061,
+      "learning_rate": 1.6452830188679247e-05,
+      "loss": 0.019,
+      "step": 1270
+    },
+    {
+      "epoch": 0.8944793850454228,
+      "grad_norm": 3.6146976947784424,
+      "learning_rate": 1.6424877707896577e-05,
+      "loss": 0.0141,
+      "step": 1280
+    },
+    {
+      "epoch": 0.9014675052410901,
+      "grad_norm": 0.05512338504195213,
+      "learning_rate": 1.6396925227113908e-05,
+      "loss": 0.0138,
+      "step": 1290
+    },
+    {
+      "epoch": 0.9084556254367575,
+      "grad_norm": 0.028468603268265724,
+      "learning_rate": 1.636897274633124e-05,
+      "loss": 0.0815,
+      "step": 1300
+    },
+    {
+      "epoch": 0.9154437456324249,
+      "grad_norm": 0.7173348665237427,
+      "learning_rate": 1.634102026554857e-05,
+      "loss": 0.0919,
+      "step": 1310
+    },
+    {
+      "epoch": 0.9224318658280922,
+      "grad_norm": 0.06138516217470169,
+      "learning_rate": 1.63130677847659e-05,
+      "loss": 0.1383,
+      "step": 1320
+    },
+    {
+      "epoch": 0.9294199860237596,
+      "grad_norm": 10.367135047912598,
+      "learning_rate": 1.628511530398323e-05,
+      "loss": 0.1265,
+      "step": 1330
+    },
+    {
+      "epoch": 0.9364081062194269,
+      "grad_norm": 0.3321629762649536,
+      "learning_rate": 1.625716282320056e-05,
+      "loss": 0.1149,
+      "step": 1340
+    },
+    {
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.21935391426086426,
+      "learning_rate": 1.622921034241789e-05,
+      "loss": 0.1147,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9503843466107617,
+      "grad_norm": 4.630390644073486,
+      "learning_rate": 1.6201257861635222e-05,
+      "loss": 0.1159,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9573724668064291,
+      "grad_norm": 0.03748277202248573,
+      "learning_rate": 1.6173305380852552e-05,
+      "loss": 0.0029,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9643605870020965,
+      "grad_norm": 0.09677435457706451,
+      "learning_rate": 1.6145352900069883e-05,
+      "loss": 0.179,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9713487071977638,
+      "grad_norm": 0.022979794070124626,
+      "learning_rate": 1.6117400419287214e-05,
+      "loss": 0.0034,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9783368273934312,
+      "grad_norm": 3.3972229957580566,
+      "learning_rate": 1.6089447938504544e-05,
+      "loss": 0.1633,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9853249475890985,
+      "grad_norm": 0.0410487987101078,
+      "learning_rate": 1.6061495457721875e-05,
+      "loss": 0.0758,
+      "step": 1410
+    },
+    {
+      "epoch": 0.9923130677847659,
+      "grad_norm": 0.1519143134355545,
+      "learning_rate": 1.6033542976939205e-05,
+      "loss": 0.0967,
+      "step": 1420
+    },
+    {
+      "epoch": 0.9993011879804332,
+      "grad_norm": 0.11000066250562668,
+      "learning_rate": 1.6005590496156536e-05,
+      "loss": 0.1054,
+      "step": 1430
+    },
+    {
+      "epoch": 1.0062893081761006,
+      "grad_norm": 9.595348358154297,
+      "learning_rate": 1.5977638015373866e-05,
+      "loss": 0.0367,
+      "step": 1440
+    },
+    {
+      "epoch": 1.013277428371768,
+      "grad_norm": 0.044866789132356644,
+      "learning_rate": 1.5949685534591197e-05,
+      "loss": 0.145,
+      "step": 1450
+    },
+    {
+      "epoch": 1.0202655485674355,
+      "grad_norm": 6.3100361824035645,
+      "learning_rate": 1.5921733053808524e-05,
+      "loss": 0.2702,
+      "step": 1460
+    },
+    {
+      "epoch": 1.0272536687631026,
+      "grad_norm": 1.0666788816452026,
+      "learning_rate": 1.5893780573025858e-05,
+      "loss": 0.1257,
+      "step": 1470
+    },
+    {
+      "epoch": 1.03424178895877,
+      "grad_norm": 15.676182746887207,
+      "learning_rate": 1.586582809224319e-05,
+      "loss": 0.0373,
+      "step": 1480
+    },
+    {
+      "epoch": 1.0412299091544375,
+      "grad_norm": 13.464042663574219,
+      "learning_rate": 1.583787561146052e-05,
+      "loss": 0.1107,
+      "step": 1490
+    },
+    {
+      "epoch": 1.0482180293501049,
+      "grad_norm": 0.029935991391539574,
+      "learning_rate": 1.580992313067785e-05,
+      "loss": 0.0778,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0552061495457723,
+      "grad_norm": 1.0842170715332031,
+      "learning_rate": 1.578197064989518e-05,
+      "loss": 0.2003,
+      "step": 1510
+    },
+    {
+      "epoch": 1.0621942697414395,
+      "grad_norm": 0.8753694891929626,
+      "learning_rate": 1.575401816911251e-05,
+      "loss": 0.0869,
+      "step": 1520
+    },
+    {
+      "epoch": 1.069182389937107,
+      "grad_norm": 0.03639127314090729,
+      "learning_rate": 1.572606568832984e-05,
+      "loss": 0.0035,
+      "step": 1530
+    },
+    {
+      "epoch": 1.0761705101327743,
+      "grad_norm": 0.10639658570289612,
+      "learning_rate": 1.5698113207547172e-05,
+      "loss": 0.0709,
+      "step": 1540
+    },
+    {
+      "epoch": 1.0831586303284417,
+      "grad_norm": 0.09013110399246216,
+      "learning_rate": 1.56701607267645e-05,
+      "loss": 0.0041,
+      "step": 1550
+    },
+    {
+      "epoch": 1.090146750524109,
+      "grad_norm": 0.1398763209581375,
+      "learning_rate": 1.564220824598183e-05,
+      "loss": 0.0039,
+      "step": 1560
+    },
+    {
+      "epoch": 1.0971348707197763,
+      "grad_norm": 0.13840065896511078,
+      "learning_rate": 1.5614255765199164e-05,
+      "loss": 0.0288,
+      "step": 1570
+    },
+    {
+      "epoch": 1.1041229909154437,
+      "grad_norm": 0.03229377046227455,
+      "learning_rate": 1.5586303284416494e-05,
+      "loss": 0.023,
+      "step": 1580
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.09332744032144547,
+      "learning_rate": 1.5558350803633825e-05,
+      "loss": 0.148,
+      "step": 1590
+    },
+    {
+      "epoch": 1.1180992313067786,
+      "grad_norm": 0.24537114799022675,
+      "learning_rate": 1.5530398322851156e-05,
+      "loss": 0.0143,
+      "step": 1600
+    },
+    {
+      "epoch": 1.1250873515024458,
+      "grad_norm": 0.02455071546137333,
+      "learning_rate": 1.5502445842068486e-05,
+      "loss": 0.0378,
+      "step": 1610
+    },
+    {
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.027023492380976677,
+      "learning_rate": 1.5474493361285817e-05,
+      "loss": 0.0558,
+      "step": 1620
+    },
+    {
+      "epoch": 1.1390635918937806,
+      "grad_norm": 0.048673246055841446,
+      "learning_rate": 1.5446540880503147e-05,
+      "loss": 0.0519,
+      "step": 1630
+    },
+    {
+      "epoch": 1.146051712089448,
+      "grad_norm": 8.062767028808594,
+      "learning_rate": 1.5418588399720474e-05,
+      "loss": 0.0404,
+      "step": 1640
+    },
+    {
+      "epoch": 1.1530398322851152,
+      "grad_norm": 8.358656883239746,
+      "learning_rate": 1.5390635918937805e-05,
+      "loss": 0.1054,
+      "step": 1650
+    },
+    {
+      "epoch": 1.1600279524807826,
+      "grad_norm": 1.3057444095611572,
+      "learning_rate": 1.5362683438155136e-05,
+      "loss": 0.1049,
+      "step": 1660
+    },
+    {
+      "epoch": 1.16701607267645,
+      "grad_norm": 0.018700115382671356,
+      "learning_rate": 1.533473095737247e-05,
+      "loss": 0.0106,
+      "step": 1670
+    },
+    {
+      "epoch": 1.1740041928721174,
+      "grad_norm": 8.731624603271484,
+      "learning_rate": 1.53067784765898e-05,
+      "loss": 0.0541,
+      "step": 1680
+    },
+    {
+      "epoch": 1.1809923130677848,
+      "grad_norm": 1.358834147453308,
+      "learning_rate": 1.527882599580713e-05,
+      "loss": 0.1515,
+      "step": 1690
+    },
+    {
+      "epoch": 1.187980433263452,
+      "grad_norm": 0.05055036395788193,
+      "learning_rate": 1.525087351502446e-05,
+      "loss": 0.1181,
+      "step": 1700
+    },
+    {
+      "epoch": 1.1949685534591195,
+      "grad_norm": 0.19085977971553802,
+      "learning_rate": 1.5222921034241792e-05,
+      "loss": 0.0624,
+      "step": 1710
+    },
+    {
+      "epoch": 1.2019566736547869,
+      "grad_norm": 1.282493233680725,
+      "learning_rate": 1.5194968553459122e-05,
+      "loss": 0.016,
+      "step": 1720
+    },
+    {
+      "epoch": 1.2089447938504543,
+      "grad_norm": 0.03596067801117897,
+      "learning_rate": 1.5167016072676451e-05,
+      "loss": 0.0211,
+      "step": 1730
+    },
+    {
+      "epoch": 1.2159329140461215,
+      "grad_norm": 0.033778220415115356,
+      "learning_rate": 1.5139063591893782e-05,
+      "loss": 0.0105,
+      "step": 1740
+    },
+    {
+      "epoch": 1.2229210342417889,
+      "grad_norm": 0.022970452904701233,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 0.1634,
+      "step": 1750
+    },
+    {
+      "epoch": 1.2299091544374563,
+      "grad_norm": 0.08840584754943848,
+      "learning_rate": 1.5083158630328443e-05,
+      "loss": 0.0062,
+      "step": 1760
+    },
+    {
+      "epoch": 1.2368972746331237,
+      "grad_norm": 0.0703321248292923,
+      "learning_rate": 1.5055206149545773e-05,
+      "loss": 0.0554,
+      "step": 1770
+    },
+    {
+      "epoch": 1.2438853948287911,
+      "grad_norm": 2.026752471923828,
+      "learning_rate": 1.5027253668763104e-05,
+      "loss": 0.03,
+      "step": 1780
+    },
+    {
+      "epoch": 1.2508735150244585,
+      "grad_norm": 0.08599984645843506,
+      "learning_rate": 1.4999301187980435e-05,
+      "loss": 0.0303,
+      "step": 1790
+    },
+    {
+      "epoch": 1.2578616352201257,
+      "grad_norm": 0.0998782068490982,
+      "learning_rate": 1.4971348707197765e-05,
+      "loss": 0.1389,
+      "step": 1800
+    },
+    {
+      "epoch": 1.2648497554157931,
+      "grad_norm": 1.7642574310302734,
+      "learning_rate": 1.4943396226415094e-05,
+      "loss": 0.0556,
+      "step": 1810
+    },
+    {
+      "epoch": 1.2718378756114606,
+      "grad_norm": 0.012091516517102718,
+      "learning_rate": 1.4915443745632425e-05,
+      "loss": 0.1116,
+      "step": 1820
+    },
+    {
+      "epoch": 1.2788259958071277,
+      "grad_norm": 0.030048305168747902,
+      "learning_rate": 1.4887491264849757e-05,
+      "loss": 0.0433,
+      "step": 1830
+    },
+    {
+      "epoch": 1.2858141160027952,
+      "grad_norm": 0.025840098038315773,
+      "learning_rate": 1.4859538784067087e-05,
+      "loss": 0.038,
+      "step": 1840
+    },
+    {
+      "epoch": 1.2928022361984626,
+      "grad_norm": 0.4607846736907959,
+      "learning_rate": 1.4831586303284418e-05,
+      "loss": 0.0815,
+      "step": 1850
+    },
+    {
+      "epoch": 1.29979035639413,
+      "grad_norm": 3.925401210784912,
+      "learning_rate": 1.4803633822501749e-05,
+      "loss": 0.0708,
+      "step": 1860
+    },
+    {
+      "epoch": 1.3067784765897974,
+      "grad_norm": 5.653714656829834,
+      "learning_rate": 1.4775681341719079e-05,
+      "loss": 0.0166,
+      "step": 1870
+    },
+    {
+      "epoch": 1.3137665967854648,
+      "grad_norm": 0.02135496586561203,
+      "learning_rate": 1.474772886093641e-05,
+      "loss": 0.0782,
+      "step": 1880
+    },
+    {
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.020742664113640785,
+      "learning_rate": 1.471977638015374e-05,
+      "loss": 0.0048,
+      "step": 1890
+    },
+    {
+      "epoch": 1.3277428371767994,
+      "grad_norm": 0.02374901808798313,
+      "learning_rate": 1.4691823899371069e-05,
+      "loss": 0.081,
+      "step": 1900
+    },
+    {
+      "epoch": 1.3347309573724668,
+      "grad_norm": 0.03886565566062927,
+      "learning_rate": 1.46638714185884e-05,
+      "loss": 0.005,
+      "step": 1910
+    },
+    {
+      "epoch": 1.3417190775681342,
+      "grad_norm": 0.16429579257965088,
+      "learning_rate": 1.463591893780573e-05,
+      "loss": 0.0154,
+      "step": 1920
+    },
+    {
+      "epoch": 1.3487071977638014,
+      "grad_norm": 0.15556152164936066,
+      "learning_rate": 1.4607966457023063e-05,
+      "loss": 0.0659,
+      "step": 1930
+    },
+    {
+      "epoch": 1.3556953179594688,
+      "grad_norm": 0.022907955572009087,
+      "learning_rate": 1.4580013976240393e-05,
+      "loss": 0.068,
+      "step": 1940
+    },
+    {
+      "epoch": 1.3626834381551363,
+      "grad_norm": 0.028623567894101143,
+      "learning_rate": 1.4552061495457724e-05,
+      "loss": 0.0023,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3696715583508037,
+      "grad_norm": 7.010847091674805,
+      "learning_rate": 1.4524109014675054e-05,
+      "loss": 0.147,
+      "step": 1960
+    },
+    {
+      "epoch": 1.376659678546471,
+      "grad_norm": 0.03581179678440094,
+      "learning_rate": 1.4496156533892385e-05,
+      "loss": 0.1209,
+      "step": 1970
+    },
+    {
+      "epoch": 1.3836477987421385,
+      "grad_norm": 0.019916867837309837,
+      "learning_rate": 1.4468204053109715e-05,
+      "loss": 0.0233,
+      "step": 1980
+    },
+    {
+      "epoch": 1.3906359189378057,
+      "grad_norm": 10.045747756958008,
+      "learning_rate": 1.4440251572327044e-05,
+      "loss": 0.0416,
+      "step": 1990
+    },
+    {
+      "epoch": 1.397624039133473,
+      "grad_norm": 0.04035484418272972,
+      "learning_rate": 1.4412299091544375e-05,
+      "loss": 0.0265,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4046121593291405,
+      "grad_norm": 0.7727859616279602,
+      "learning_rate": 1.4384346610761705e-05,
+      "loss": 0.0331,
+      "step": 2010
+    },
+    {
+      "epoch": 1.4116002795248077,
+      "grad_norm": 0.018749171867966652,
+      "learning_rate": 1.4356394129979036e-05,
+      "loss": 0.0749,
+      "step": 2020
+    },
+    {
+      "epoch": 1.4185883997204751,
+      "grad_norm": 5.582619667053223,
+      "learning_rate": 1.4328441649196368e-05,
+      "loss": 0.0622,
+      "step": 2030
+    },
+    {
+      "epoch": 1.4255765199161425,
+      "grad_norm": 0.03347029909491539,
+      "learning_rate": 1.4300489168413699e-05,
+      "loss": 0.0021,
+      "step": 2040
+    },
+    {
+      "epoch": 1.43256464011181,
+      "grad_norm": 0.1696143001317978,
+      "learning_rate": 1.427253668763103e-05,
+      "loss": 0.0021,
+      "step": 2050
+    },
+    {
+      "epoch": 1.4395527603074774,
+      "grad_norm": 0.021968642249703407,
+      "learning_rate": 1.424458420684836e-05,
+      "loss": 0.0025,
+      "step": 2060
+    },
+    {
+      "epoch": 1.4465408805031448,
+      "grad_norm": 0.007907007820904255,
+      "learning_rate": 1.421663172606569e-05,
+      "loss": 0.0478,
+      "step": 2070
+    },
+    {
+      "epoch": 1.453529000698812,
+      "grad_norm": 0.035767070949077606,
+      "learning_rate": 1.418867924528302e-05,
+      "loss": 0.0012,
+      "step": 2080
+    },
+    {
+      "epoch": 1.4605171208944794,
+      "grad_norm": 0.011390830390155315,
+      "learning_rate": 1.416072676450035e-05,
+      "loss": 0.1533,
+      "step": 2090
+    },
+    {
+      "epoch": 1.4675052410901468,
+      "grad_norm": 7.848016738891602,
+      "learning_rate": 1.413277428371768e-05,
+      "loss": 0.0675,
+      "step": 2100
+    },
+    {
+      "epoch": 1.474493361285814,
+      "grad_norm": 0.011802544817328453,
+      "learning_rate": 1.4104821802935011e-05,
+      "loss": 0.0261,
+      "step": 2110
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 2.9148340225219727,
+      "learning_rate": 1.4076869322152342e-05,
+      "loss": 0.2831,
+      "step": 2120
+    },
+    {
+      "epoch": 1.4884696016771488,
+      "grad_norm": 0.08199399709701538,
+      "learning_rate": 1.4048916841369674e-05,
+      "loss": 0.0088,
+      "step": 2130
+    },
+    {
+      "epoch": 1.4954577218728162,
+      "grad_norm": 0.01199902594089508,
+      "learning_rate": 1.4020964360587004e-05,
+      "loss": 0.0693,
+      "step": 2140
+    },
+    {
+      "epoch": 1.5024458420684836,
+      "grad_norm": 0.07146008312702179,
+      "learning_rate": 1.3993011879804335e-05,
+      "loss": 0.0746,
+      "step": 2150
+    },
+    {
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.02118564024567604,
+      "learning_rate": 1.3965059399021666e-05,
+      "loss": 0.003,
+      "step": 2160
+    },
+    {
+      "epoch": 1.5164220824598185,
+      "grad_norm": 0.016972247511148453,
+      "learning_rate": 1.3937106918238994e-05,
+      "loss": 0.0106,
+      "step": 2170
+    },
+    {
+      "epoch": 1.5234102026554857,
+      "grad_norm": 0.023427631705999374,
+      "learning_rate": 1.3909154437456325e-05,
+      "loss": 0.0444,
+      "step": 2180
+    },
+    {
+      "epoch": 1.530398322851153,
+      "grad_norm": 0.03280281275510788,
+      "learning_rate": 1.3881201956673656e-05,
+      "loss": 0.0026,
+      "step": 2190
+    },
+    {
+      "epoch": 1.5373864430468203,
+      "grad_norm": 0.06073632091283798,
+      "learning_rate": 1.3853249475890986e-05,
+      "loss": 0.194,
+      "step": 2200
+    },
+    {
+      "epoch": 1.5443745632424877,
+      "grad_norm": 0.008114098571240902,
+      "learning_rate": 1.3825296995108317e-05,
+      "loss": 0.0016,
+      "step": 2210
+    },
+    {
+      "epoch": 1.551362683438155,
+      "grad_norm": 0.0467991828918457,
+      "learning_rate": 1.3797344514325647e-05,
+      "loss": 0.1305,
+      "step": 2220
+    },
+    {
+      "epoch": 1.5583508036338225,
+      "grad_norm": 0.06984475255012512,
+      "learning_rate": 1.376939203354298e-05,
+      "loss": 0.1386,
+      "step": 2230
+    },
+    {
+      "epoch": 1.56533892382949,
+      "grad_norm": 0.014983494766056538,
+      "learning_rate": 1.374143955276031e-05,
+      "loss": 0.1399,
+      "step": 2240
+    },
+    {
+      "epoch": 1.5723270440251573,
+      "grad_norm": 0.01568700559437275,
+      "learning_rate": 1.3713487071977637e-05,
+      "loss": 0.1084,
+      "step": 2250
+    },
+    {
+      "epoch": 1.5793151642208247,
+      "grad_norm": 10.549098014831543,
+      "learning_rate": 1.368553459119497e-05,
+      "loss": 0.0082,
+      "step": 2260
+    },
+    {
+      "epoch": 1.586303284416492,
+      "grad_norm": 0.02495586685836315,
+      "learning_rate": 1.36575821104123e-05,
+      "loss": 0.0019,
+      "step": 2270
+    },
+    {
+      "epoch": 1.5932914046121593,
+      "grad_norm": 0.01876898854970932,
+      "learning_rate": 1.362962962962963e-05,
+      "loss": 0.1002,
+      "step": 2280
+    },
+    {
+      "epoch": 1.6002795248078265,
+      "grad_norm": 0.044464971870183945,
+      "learning_rate": 1.3601677148846961e-05,
+      "loss": 0.0019,
+      "step": 2290
+    },
+    {
+      "epoch": 1.607267645003494,
+      "grad_norm": 0.04108593240380287,
+      "learning_rate": 1.3573724668064292e-05,
+      "loss": 0.0753,
+      "step": 2300
+    },
+    {
+      "epoch": 1.6142557651991614,
+      "grad_norm": 7.812976837158203,
+      "learning_rate": 1.3545772187281622e-05,
+      "loss": 0.2875,
+      "step": 2310
+    },
+    {
+      "epoch": 1.6212438853948288,
+      "grad_norm": 4.261682987213135,
+      "learning_rate": 1.3517819706498953e-05,
+      "loss": 0.1059,
+      "step": 2320
+    },
+    {
+      "epoch": 1.6282320055904962,
+      "grad_norm": 0.03309008479118347,
+      "learning_rate": 1.3489867225716285e-05,
+      "loss": 0.0051,
+      "step": 2330
+    },
+    {
+      "epoch": 1.6352201257861636,
+      "grad_norm": 9.983598709106445,
+      "learning_rate": 1.3461914744933612e-05,
+      "loss": 0.0249,
+      "step": 2340
+    },
+    {
+      "epoch": 1.642208245981831,
+      "grad_norm": 4.973055839538574,
+      "learning_rate": 1.3433962264150943e-05,
+      "loss": 0.1441,
+      "step": 2350
+    },
+    {
+      "epoch": 1.6491963661774982,
+      "grad_norm": 9.258291244506836,
+      "learning_rate": 1.3406009783368275e-05,
+      "loss": 0.1151,
+      "step": 2360
+    },
+    {
+      "epoch": 1.6561844863731656,
+      "grad_norm": 0.015135078690946102,
+      "learning_rate": 1.3378057302585606e-05,
+      "loss": 0.0458,
+      "step": 2370
+    },
+    {
+      "epoch": 1.663172606568833,
+      "grad_norm": 0.027922656387090683,
+      "learning_rate": 1.3350104821802936e-05,
+      "loss": 0.012,
+      "step": 2380
+    },
+    {
+      "epoch": 1.6701607267645002,
+      "grad_norm": 0.043312493711709976,
+      "learning_rate": 1.3322152341020267e-05,
+      "loss": 0.1028,
+      "step": 2390
+    },
+    {
+      "epoch": 1.6771488469601676,
+      "grad_norm": 0.012087292037904263,
+      "learning_rate": 1.3294199860237597e-05,
+      "loss": 0.0475,
+      "step": 2400
+    },
+    {
+      "epoch": 1.684136967155835,
+      "grad_norm": 0.5556979775428772,
+      "learning_rate": 1.3266247379454928e-05,
+      "loss": 0.0142,
+      "step": 2410
+    },
+    {
+      "epoch": 1.6911250873515025,
+      "grad_norm": 5.196042537689209,
+      "learning_rate": 1.3238294898672259e-05,
+      "loss": 0.1002,
+      "step": 2420
+    },
+    {
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.010861317627131939,
+      "learning_rate": 1.3210342417889587e-05,
+      "loss": 0.0953,
+      "step": 2430
+    },
+    {
+      "epoch": 1.7051013277428373,
+      "grad_norm": 0.017955880612134933,
+      "learning_rate": 1.3182389937106918e-05,
+      "loss": 0.0642,
+      "step": 2440
+    },
+    {
+      "epoch": 1.7120894479385047,
+      "grad_norm": 0.028966449201107025,
+      "learning_rate": 1.3154437456324249e-05,
+      "loss": 0.1619,
+      "step": 2450
+    },
+    {
+      "epoch": 1.719077568134172,
+      "grad_norm": 0.8712167739868164,
+      "learning_rate": 1.312648497554158e-05,
+      "loss": 0.0324,
+      "step": 2460
+    },
+    {
+      "epoch": 1.7260656883298393,
+      "grad_norm": 0.2874550521373749,
+      "learning_rate": 1.3098532494758911e-05,
+      "loss": 0.026,
+      "step": 2470
+    },
+    {
+      "epoch": 1.7330538085255065,
+      "grad_norm": 0.010579444468021393,
+      "learning_rate": 1.3070580013976242e-05,
+      "loss": 0.0048,
+      "step": 2480
+    },
+    {
+      "epoch": 1.740041928721174,
+      "grad_norm": 0.15773366391658783,
+      "learning_rate": 1.3042627533193573e-05,
+      "loss": 0.108,
+      "step": 2490
+    },
+    {
+      "epoch": 1.7470300489168413,
+      "grad_norm": 0.10269400477409363,
+      "learning_rate": 1.3014675052410903e-05,
+      "loss": 0.0884,
+      "step": 2500
+    },
+    {
+      "epoch": 1.7540181691125087,
+      "grad_norm": 0.012210669927299023,
+      "learning_rate": 1.2986722571628234e-05,
+      "loss": 0.0025,
+      "step": 2510
+    },
+    {
+      "epoch": 1.7610062893081762,
+      "grad_norm": 0.046242229640483856,
+      "learning_rate": 1.2958770090845563e-05,
+      "loss": 0.0124,
+      "step": 2520
+    },
+    {
+      "epoch": 1.7679944095038436,
+      "grad_norm": 12.532265663146973,
+      "learning_rate": 1.2930817610062893e-05,
+      "loss": 0.1502,
+      "step": 2530
+    },
+    {
+      "epoch": 1.774982529699511,
+      "grad_norm": 0.018337048590183258,
+      "learning_rate": 1.2902865129280224e-05,
+      "loss": 0.0402,
+      "step": 2540
+    },
+    {
+      "epoch": 1.7819706498951782,
+      "grad_norm": 0.9082455039024353,
+      "learning_rate": 1.2874912648497554e-05,
+      "loss": 0.0719,
+      "step": 2550
+    },
+    {
+      "epoch": 1.7889587700908456,
+      "grad_norm": 0.6127817034721375,
+      "learning_rate": 1.2846960167714887e-05,
+      "loss": 0.011,
+      "step": 2560
+    },
+    {
+      "epoch": 1.7959468902865128,
+      "grad_norm": 0.02741195261478424,
+      "learning_rate": 1.2819007686932217e-05,
+      "loss": 0.0436,
+      "step": 2570
+    },
+    {
+      "epoch": 1.8029350104821802,
+      "grad_norm": 0.012224216014146805,
+      "learning_rate": 1.2791055206149548e-05,
+      "loss": 0.0026,
+      "step": 2580
+    },
+    {
+      "epoch": 1.8099231306778476,
+      "grad_norm": 0.01689975894987583,
+      "learning_rate": 1.2763102725366878e-05,
+      "loss": 0.0067,
+      "step": 2590
+    },
+    {
+      "epoch": 1.816911250873515,
+      "grad_norm": 0.010701341554522514,
+      "learning_rate": 1.2735150244584207e-05,
+      "loss": 0.008,
+      "step": 2600
+    },
+    {
+      "epoch": 1.8238993710691824,
+      "grad_norm": 0.12163686007261276,
+      "learning_rate": 1.2707197763801538e-05,
+      "loss": 0.0767,
+      "step": 2610
+    },
+    {
+      "epoch": 1.8308874912648498,
+      "grad_norm": 0.07165365666151047,
+      "learning_rate": 1.2679245283018868e-05,
+      "loss": 0.0028,
+      "step": 2620
+    },
+    {
+      "epoch": 1.8378756114605173,
+      "grad_norm": 0.007849736139178276,
+      "learning_rate": 1.2651292802236199e-05,
+      "loss": 0.0488,
+      "step": 2630
+    },
+    {
+      "epoch": 1.8448637316561844,
+      "grad_norm": 0.01707429252564907,
+      "learning_rate": 1.262334032145353e-05,
+      "loss": 0.0022,
+      "step": 2640
+    },
+    {
+      "epoch": 1.8518518518518519,
+      "grad_norm": 6.567415237426758,
+      "learning_rate": 1.259538784067086e-05,
+      "loss": 0.0677,
+      "step": 2650
+    },
+    {
+      "epoch": 1.858839972047519,
+      "grad_norm": 0.005324068479239941,
+      "learning_rate": 1.2567435359888192e-05,
+      "loss": 0.0057,
+      "step": 2660
+    },
+    {
+      "epoch": 1.8658280922431865,
+      "grad_norm": 11.791135787963867,
+      "learning_rate": 1.2539482879105523e-05,
+      "loss": 0.0764,
+      "step": 2670
+    },
+    {
+      "epoch": 1.8728162124388539,
+      "grad_norm": 0.005892631132155657,
+      "learning_rate": 1.2511530398322853e-05,
+      "loss": 0.0736,
+      "step": 2680
+    },
+    {
+      "epoch": 1.8798043326345213,
+      "grad_norm": 0.04005246236920357,
+      "learning_rate": 1.2483577917540182e-05,
+      "loss": 0.0617,
+      "step": 2690
+    },
+    {
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.010624396614730358,
+      "learning_rate": 1.2455625436757513e-05,
+      "loss": 0.0011,
+      "step": 2700
+    },
+    {
+      "epoch": 1.8937805730258561,
+      "grad_norm": 0.01761520840227604,
+      "learning_rate": 1.2427672955974843e-05,
+      "loss": 0.1574,
+      "step": 2710
+    },
+    {
+      "epoch": 1.9007686932215235,
+      "grad_norm": 0.05212152749300003,
+      "learning_rate": 1.2399720475192174e-05,
+      "loss": 0.0618,
+      "step": 2720
+    },
+    {
+      "epoch": 1.9077568134171907,
+      "grad_norm": 0.012297810055315495,
+      "learning_rate": 1.2371767994409504e-05,
+      "loss": 0.0802,
+      "step": 2730
+    },
+    {
+      "epoch": 1.9147449336128581,
+      "grad_norm": 1.4671796560287476,
+      "learning_rate": 1.2343815513626835e-05,
+      "loss": 0.0351,
+      "step": 2740
+    },
+    {
+      "epoch": 1.9217330538085255,
+      "grad_norm": 1.28997802734375,
+      "learning_rate": 1.2315863032844166e-05,
+      "loss": 0.0069,
+      "step": 2750
+    },
+    {
+      "epoch": 1.9287211740041927,
+      "grad_norm": 0.15810908377170563,
+      "learning_rate": 1.2287910552061498e-05,
+      "loss": 0.0037,
+      "step": 2760
+    },
+    {
+      "epoch": 1.9357092941998602,
+      "grad_norm": 0.01939568482339382,
+      "learning_rate": 1.2259958071278828e-05,
+      "loss": 0.1375,
+      "step": 2770
+    },
+    {
+      "epoch": 1.9426974143955276,
+      "grad_norm": 0.023233506828546524,
+      "learning_rate": 1.2232005590496157e-05,
+      "loss": 0.0016,
+      "step": 2780
+    },
+    {
+      "epoch": 1.949685534591195,
+      "grad_norm": 13.808353424072266,
+      "learning_rate": 1.2204053109713488e-05,
+      "loss": 0.1251,
+      "step": 2790
+    },
+    {
+      "epoch": 1.9566736547868624,
+      "grad_norm": 0.007430393248796463,
+      "learning_rate": 1.2176100628930818e-05,
+      "loss": 0.0957,
+      "step": 2800
+    },
+    {
+      "epoch": 1.9636617749825298,
+      "grad_norm": 0.015328769572079182,
+      "learning_rate": 1.2148148148148149e-05,
+      "loss": 0.0019,
+      "step": 2810
+    },
+    {
+      "epoch": 1.9706498951781972,
+      "grad_norm": 0.05328751727938652,
+      "learning_rate": 1.212019566736548e-05,
+      "loss": 0.2193,
+      "step": 2820
+    },
+    {
+      "epoch": 1.9776380153738644,
+      "grad_norm": 0.05544736981391907,
+      "learning_rate": 1.209224318658281e-05,
+      "loss": 0.011,
+      "step": 2830
+    },
+    {
+      "epoch": 1.9846261355695318,
+      "grad_norm": 0.006596289575099945,
+      "learning_rate": 1.206429070580014e-05,
+      "loss": 0.0454,
+      "step": 2840
+    },
+    {
+      "epoch": 1.991614255765199,
+      "grad_norm": 7.7824015617370605,
+      "learning_rate": 1.2036338225017471e-05,
+      "loss": 0.024,
+      "step": 2850
+    },
+    {
+      "epoch": 1.9986023759608664,
+      "grad_norm": 0.024772530421614647,
+      "learning_rate": 1.2008385744234804e-05,
+      "loss": 0.0109,
+      "step": 2860
+    },
+    {
+      "epoch": 2.005590496156534,
+      "grad_norm": 0.026197150349617004,
+      "learning_rate": 1.198043326345213e-05,
+      "loss": 0.0362,
+      "step": 2870
+    },
+    {
+      "epoch": 2.0125786163522013,
+      "grad_norm": 18.572996139526367,
+      "learning_rate": 1.1952480782669463e-05,
+      "loss": 0.0935,
+      "step": 2880
+    },
+    {
+      "epoch": 2.0195667365478687,
+      "grad_norm": 0.17002998292446136,
+      "learning_rate": 1.1924528301886794e-05,
+      "loss": 0.0343,
+      "step": 2890
+    },
+    {
+      "epoch": 2.026554856743536,
+      "grad_norm": 0.029017208144068718,
+      "learning_rate": 1.1896575821104124e-05,
+      "loss": 0.0013,
+      "step": 2900
+    },
+    {
+      "epoch": 2.0335429769392035,
+      "grad_norm": 0.03208720684051514,
+      "learning_rate": 1.1868623340321455e-05,
+      "loss": 0.0595,
+      "step": 2910
+    },
+    {
+      "epoch": 2.040531097134871,
+      "grad_norm": 0.03342962637543678,
+      "learning_rate": 1.1840670859538785e-05,
+      "loss": 0.0025,
+      "step": 2920
+    },
+    {
+      "epoch": 2.047519217330538,
+      "grad_norm": 0.00888112373650074,
+      "learning_rate": 1.1812718378756116e-05,
+      "loss": 0.0007,
+      "step": 2930
+    },
+    {
+      "epoch": 2.0545073375262053,
+      "grad_norm": 0.14470069110393524,
+      "learning_rate": 1.1784765897973446e-05,
+      "loss": 0.1185,
+      "step": 2940
+    },
+    {
+      "epoch": 2.0614954577218727,
+      "grad_norm": 0.006698206998407841,
+      "learning_rate": 1.1756813417190777e-05,
+      "loss": 0.0027,
+      "step": 2950
+    },
+    {
+      "epoch": 2.06848357791754,
+      "grad_norm": 0.013084967620670795,
+      "learning_rate": 1.1728860936408106e-05,
+      "loss": 0.1188,
+      "step": 2960
+    },
+    {
+      "epoch": 2.0754716981132075,
+      "grad_norm": 4.599226951599121,
+      "learning_rate": 1.1700908455625436e-05,
+      "loss": 0.192,
+      "step": 2970
+    },
+    {
+      "epoch": 2.082459818308875,
+      "grad_norm": 0.04704582691192627,
+      "learning_rate": 1.1672955974842769e-05,
+      "loss": 0.0214,
+      "step": 2980
+    },
+    {
+      "epoch": 2.0894479385045424,
+      "grad_norm": 0.00998607650399208,
+      "learning_rate": 1.16450034940601e-05,
+      "loss": 0.0642,
+      "step": 2990
+    },
+    {
+      "epoch": 2.0964360587002098,
+      "grad_norm": 11.119775772094727,
+      "learning_rate": 1.161705101327743e-05,
+      "loss": 0.0075,
+      "step": 3000
+    },
+    {
+      "epoch": 2.103424178895877,
+      "grad_norm": 10.634467124938965,
+      "learning_rate": 1.158909853249476e-05,
+      "loss": 0.1313,
+      "step": 3010
+    },
+    {
+      "epoch": 2.1104122990915446,
+      "grad_norm": 0.05366509407758713,
+      "learning_rate": 1.1561146051712091e-05,
+      "loss": 0.0151,
+      "step": 3020
+    },
+    {
+      "epoch": 2.1174004192872116,
+      "grad_norm": 0.014267387799918652,
+      "learning_rate": 1.1533193570929421e-05,
+      "loss": 0.1542,
+      "step": 3030
+    },
+    {
+      "epoch": 2.124388539482879,
+      "grad_norm": 0.008031471632421017,
+      "learning_rate": 1.150524109014675e-05,
+      "loss": 0.0015,
+      "step": 3040
+    },
+    {
+      "epoch": 2.1313766596785464,
+      "grad_norm": 8.725931167602539,
+      "learning_rate": 1.1477288609364081e-05,
+      "loss": 0.0981,
+      "step": 3050
+    },
+    {
+      "epoch": 2.138364779874214,
+      "grad_norm": 0.010948359034955502,
+      "learning_rate": 1.1449336128581411e-05,
+      "loss": 0.0513,
+      "step": 3060
+    },
+    {
+      "epoch": 2.145352900069881,
+      "grad_norm": 0.22167250514030457,
+      "learning_rate": 1.1421383647798742e-05,
+      "loss": 0.0295,
+      "step": 3070
+    },
+    {
+      "epoch": 2.1523410202655486,
+      "grad_norm": 0.038010116666555405,
+      "learning_rate": 1.1393431167016074e-05,
+      "loss": 0.001,
+      "step": 3080
+    },
+    {
+      "epoch": 2.159329140461216,
+      "grad_norm": 0.014428159222006798,
+      "learning_rate": 1.1365478686233405e-05,
+      "loss": 0.0015,
+      "step": 3090
+    },
+    {
+      "epoch": 2.1663172606568835,
+      "grad_norm": 0.027289502322673798,
+      "learning_rate": 1.1337526205450735e-05,
+      "loss": 0.0036,
+      "step": 3100
+    },
+    {
+      "epoch": 2.1733053808525504,
+      "grad_norm": 1.7627710103988647,
+      "learning_rate": 1.1309573724668066e-05,
+      "loss": 0.0314,
+      "step": 3110
+    },
+    {
+      "epoch": 2.180293501048218,
+      "grad_norm": 0.006555517669767141,
+      "learning_rate": 1.1281621243885397e-05,
+      "loss": 0.0017,
+      "step": 3120
+    },
+    {
+      "epoch": 2.1872816212438853,
+      "grad_norm": 0.006436652038246393,
+      "learning_rate": 1.1253668763102725e-05,
+      "loss": 0.006,
+      "step": 3130
+    },
+    {
+      "epoch": 2.1942697414395527,
+      "grad_norm": 0.010117010213434696,
+      "learning_rate": 1.1225716282320056e-05,
+      "loss": 0.0127,
+      "step": 3140
+    },
+    {
+      "epoch": 2.20125786163522,
+      "grad_norm": 0.01584050804376602,
+      "learning_rate": 1.1197763801537387e-05,
+      "loss": 0.0011,
+      "step": 3150
+    },
+    {
+      "epoch": 2.2082459818308875,
+      "grad_norm": 3.1698734760284424,
+      "learning_rate": 1.1169811320754717e-05,
+      "loss": 0.0643,
+      "step": 3160
+    },
+    {
+      "epoch": 2.215234102026555,
+      "grad_norm": 20.464344024658203,
+      "learning_rate": 1.1141858839972048e-05,
+      "loss": 0.1347,
+      "step": 3170
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.04433240368962288,
+      "learning_rate": 1.111390635918938e-05,
+      "loss": 0.0079,
+      "step": 3180
+    },
+    {
+      "epoch": 2.2292103424178897,
+      "grad_norm": 0.04871518909931183,
+      "learning_rate": 1.108595387840671e-05,
+      "loss": 0.1057,
+      "step": 3190
+    },
+    {
+      "epoch": 2.236198462613557,
+      "grad_norm": 21.398836135864258,
+      "learning_rate": 1.1058001397624041e-05,
+      "loss": 0.0091,
+      "step": 3200
+    },
+    {
+      "epoch": 2.243186582809224,
+      "grad_norm": 0.023466596379876137,
+      "learning_rate": 1.1030048916841372e-05,
+      "loss": 0.0057,
+      "step": 3210
+    },
+    {
+      "epoch": 2.2501747030048915,
+      "grad_norm": 0.011096821166574955,
+      "learning_rate": 1.10020964360587e-05,
+      "loss": 0.0048,
+      "step": 3220
+    },
+    {
+      "epoch": 2.257162823200559,
+      "grad_norm": 7.414328098297119,
+      "learning_rate": 1.0974143955276031e-05,
+      "loss": 0.0135,
+      "step": 3230
+    },
+    {
+      "epoch": 2.2641509433962264,
+      "grad_norm": 0.03595978021621704,
+      "learning_rate": 1.0946191474493362e-05,
+      "loss": 0.001,
+      "step": 3240
+    },
+    {
+      "epoch": 2.2711390635918938,
+      "grad_norm": 17.89633560180664,
+      "learning_rate": 1.0918238993710692e-05,
+      "loss": 0.0867,
+      "step": 3250
+    },
+    {
+      "epoch": 2.278127183787561,
+      "grad_norm": 0.009794089011847973,
+      "learning_rate": 1.0890286512928023e-05,
+      "loss": 0.0324,
+      "step": 3260
+    },
+    {
+      "epoch": 2.2851153039832286,
+      "grad_norm": 0.04615940898656845,
+      "learning_rate": 1.0862334032145353e-05,
+      "loss": 0.0013,
+      "step": 3270
+    },
+    {
+      "epoch": 2.292103424178896,
+      "grad_norm": 0.14327336847782135,
+      "learning_rate": 1.0834381551362686e-05,
+      "loss": 0.0009,
+      "step": 3280
+    },
+    {
+      "epoch": 2.2990915443745634,
+      "grad_norm": 0.003527725813910365,
+      "learning_rate": 1.0806429070580016e-05,
+      "loss": 0.0751,
+      "step": 3290
+    },
+    {
+      "epoch": 2.3060796645702304,
+      "grad_norm": 0.013885865919291973,
+      "learning_rate": 1.0778476589797347e-05,
+      "loss": 0.0537,
+      "step": 3300
+    },
+    {
+      "epoch": 2.313067784765898,
+      "grad_norm": 0.01316741295158863,
+      "learning_rate": 1.0750524109014676e-05,
+      "loss": 0.0008,
+      "step": 3310
+    },
+    {
+      "epoch": 2.320055904961565,
+      "grad_norm": 0.0060685062780976295,
+      "learning_rate": 1.0722571628232006e-05,
+      "loss": 0.0673,
+      "step": 3320
+    },
+    {
+      "epoch": 2.3270440251572326,
+      "grad_norm": 0.021814968436956406,
+      "learning_rate": 1.0694619147449337e-05,
+      "loss": 0.0014,
+      "step": 3330
+    },
+    {
+      "epoch": 2.3340321453529,
+      "grad_norm": 0.013100036419928074,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.0756,
+      "step": 3340
+    },
+    {
+      "epoch": 2.3410202655485675,
+      "grad_norm": 0.01045132428407669,
+      "learning_rate": 1.0638714185883998e-05,
+      "loss": 0.0459,
+      "step": 3350
+    },
+    {
+      "epoch": 2.348008385744235,
+      "grad_norm": 0.006095814984291792,
+      "learning_rate": 1.0610761705101328e-05,
+      "loss": 0.0551,
+      "step": 3360
+    },
+    {
+      "epoch": 2.3549965059399023,
+      "grad_norm": 0.011639703065156937,
+      "learning_rate": 1.0582809224318659e-05,
+      "loss": 0.0032,
+      "step": 3370
+    },
+    {
+      "epoch": 2.3619846261355697,
+      "grad_norm": 0.0661187469959259,
+      "learning_rate": 1.0554856743535991e-05,
+      "loss": 0.0009,
+      "step": 3380
+    },
+    {
+      "epoch": 2.368972746331237,
+      "grad_norm": 0.030025649815797806,
+      "learning_rate": 1.0526904262753318e-05,
+      "loss": 0.0096,
+      "step": 3390
+    },
+    {
+      "epoch": 2.375960866526904,
+      "grad_norm": 0.05209723860025406,
+      "learning_rate": 1.049895178197065e-05,
+      "loss": 0.1485,
+      "step": 3400
+    },
+    {
+      "epoch": 2.3829489867225715,
+      "grad_norm": 0.4318474233150482,
+      "learning_rate": 1.0470999301187981e-05,
+      "loss": 0.0011,
+      "step": 3410
+    },
+    {
+      "epoch": 2.389937106918239,
+      "grad_norm": 0.011814878322184086,
+      "learning_rate": 1.0443046820405312e-05,
+      "loss": 0.0019,
+      "step": 3420
+    },
+    {
+      "epoch": 2.3969252271139063,
+      "grad_norm": 27.78424644470215,
+      "learning_rate": 1.0415094339622642e-05,
+      "loss": 0.1376,
+      "step": 3430
+    },
+    {
+      "epoch": 2.4039133473095737,
+      "grad_norm": 0.27662497758865356,
+      "learning_rate": 1.0387141858839973e-05,
+      "loss": 0.0009,
+      "step": 3440
+    },
+    {
+      "epoch": 2.410901467505241,
+      "grad_norm": 0.0203730296343565,
+      "learning_rate": 1.0359189378057304e-05,
+      "loss": 0.001,
+      "step": 3450
+    },
+    {
+      "epoch": 2.4178895877009086,
+      "grad_norm": 0.005023602396249771,
+      "learning_rate": 1.0331236897274634e-05,
+      "loss": 0.0026,
+      "step": 3460
+    },
+    {
+      "epoch": 2.424877707896576,
+      "grad_norm": 0.005766034591943026,
+      "learning_rate": 1.0303284416491965e-05,
+      "loss": 0.0043,
+      "step": 3470
+    },
+    {
+      "epoch": 2.431865828092243,
+      "grad_norm": 0.0078241853043437,
+      "learning_rate": 1.0275331935709294e-05,
+      "loss": 0.0666,
+      "step": 3480
+    },
+    {
+      "epoch": 2.4388539482879104,
+      "grad_norm": 0.13950783014297485,
+      "learning_rate": 1.0247379454926624e-05,
+      "loss": 0.0299,
+      "step": 3490
+    },
+    {
+      "epoch": 2.4458420684835778,
+      "grad_norm": 0.03160367161035538,
+      "learning_rate": 1.0219426974143956e-05,
+      "loss": 0.003,
+      "step": 3500
+    },
+    {
+      "epoch": 2.452830188679245,
+      "grad_norm": 0.01961313560605049,
+      "learning_rate": 1.0191474493361287e-05,
+      "loss": 0.0103,
+      "step": 3510
+    },
+    {
+      "epoch": 2.4598183088749126,
+      "grad_norm": 0.007672517094761133,
+      "learning_rate": 1.0163522012578618e-05,
+      "loss": 0.0037,
+      "step": 3520
+    },
+    {
+      "epoch": 2.46680642907058,
+      "grad_norm": 0.005141376983374357,
+      "learning_rate": 1.0135569531795948e-05,
+      "loss": 0.0005,
+      "step": 3530
+    },
+    {
+      "epoch": 2.4737945492662474,
+      "grad_norm": 0.027929285541176796,
+      "learning_rate": 1.0107617051013279e-05,
+      "loss": 0.0248,
+      "step": 3540
+    },
+    {
+      "epoch": 2.480782669461915,
+      "grad_norm": 1.3310028314590454,
+      "learning_rate": 1.007966457023061e-05,
+      "loss": 0.124,
+      "step": 3550
+    },
+    {
+      "epoch": 2.4877707896575822,
+      "grad_norm": 0.06481662392616272,
+      "learning_rate": 1.005171208944794e-05,
+      "loss": 0.0024,
+      "step": 3560
+    },
+    {
+      "epoch": 2.4947589098532497,
+      "grad_norm": 0.008162762969732285,
+      "learning_rate": 1.0023759608665269e-05,
+      "loss": 0.0017,
+      "step": 3570
+    },
+    {
+      "epoch": 2.501747030048917,
+      "grad_norm": 7.582877159118652,
+      "learning_rate": 9.995807127882601e-06,
+      "loss": 0.0838,
+      "step": 3580
+    },
+    {
+      "epoch": 2.508735150244584,
+      "grad_norm": 0.06521458923816681,
+      "learning_rate": 9.96785464709993e-06,
+      "loss": 0.0009,
+      "step": 3590
+    },
+    {
+      "epoch": 2.5157232704402515,
+      "grad_norm": 0.005363748874515295,
+      "learning_rate": 9.939902166317262e-06,
+      "loss": 0.0797,
+      "step": 3600
+    },
+    {
+      "epoch": 2.522711390635919,
+      "grad_norm": 0.007538496516644955,
+      "learning_rate": 9.911949685534593e-06,
+      "loss": 0.039,
+      "step": 3610
+    },
+    {
+      "epoch": 2.5296995108315863,
+      "grad_norm": 0.007935012690722942,
+      "learning_rate": 9.883997204751923e-06,
+      "loss": 0.0039,
+      "step": 3620
+    },
+    {
+      "epoch": 2.5366876310272537,
+      "grad_norm": 0.005138902924954891,
+      "learning_rate": 9.856044723969254e-06,
+      "loss": 0.0006,
+      "step": 3630
+    },
+    {
+      "epoch": 2.543675751222921,
+      "grad_norm": 0.012060786597430706,
+      "learning_rate": 9.828092243186583e-06,
+      "loss": 0.0013,
+      "step": 3640
+    },
+    {
+      "epoch": 2.5506638714185885,
+      "grad_norm": 0.09823895990848541,
+      "learning_rate": 9.800139762403915e-06,
+      "loss": 0.0319,
+      "step": 3650
+    },
+    {
+      "epoch": 2.5576519916142555,
+      "grad_norm": 0.009398439899086952,
+      "learning_rate": 9.772187281621245e-06,
+      "loss": 0.0012,
+      "step": 3660
+    },
+    {
+      "epoch": 2.564640111809923,
+      "grad_norm": 0.009316815994679928,
+      "learning_rate": 9.744234800838576e-06,
+      "loss": 0.0008,
+      "step": 3670
+    },
+    {
+      "epoch": 2.5716282320055903,
+      "grad_norm": 0.01164779718965292,
+      "learning_rate": 9.716282320055905e-06,
+      "loss": 0.0014,
+      "step": 3680
+    },
+    {
+      "epoch": 2.5786163522012577,
+      "grad_norm": 7.828653335571289,
+      "learning_rate": 9.688329839273235e-06,
+      "loss": 0.0452,
+      "step": 3690
+    },
+    {
+      "epoch": 2.585604472396925,
+      "grad_norm": 0.003899825969710946,
+      "learning_rate": 9.660377358490568e-06,
+      "loss": 0.0004,
+      "step": 3700
+    },
+    {
+      "epoch": 2.5925925925925926,
+      "grad_norm": 6.803162097930908,
+      "learning_rate": 9.632424877707898e-06,
+      "loss": 0.0892,
+      "step": 3710
+    },
+    {
+      "epoch": 2.59958071278826,
+      "grad_norm": 0.00998039823025465,
+      "learning_rate": 9.604472396925227e-06,
+      "loss": 0.123,
+      "step": 3720
+    },
+    {
+      "epoch": 2.6065688329839274,
+      "grad_norm": 0.0067862533032894135,
+      "learning_rate": 9.576519916142558e-06,
+      "loss": 0.001,
+      "step": 3730
+    },
+    {
+      "epoch": 2.613556953179595,
+      "grad_norm": 0.004215400665998459,
+      "learning_rate": 9.548567435359888e-06,
+      "loss": 0.177,
+      "step": 3740
+    },
+    {
+      "epoch": 2.620545073375262,
+      "grad_norm": 0.02502995729446411,
+      "learning_rate": 9.52061495457722e-06,
+      "loss": 0.008,
+      "step": 3750
+    },
+    {
+      "epoch": 2.6275331935709296,
+      "grad_norm": 5.30055046081543,
+      "learning_rate": 9.492662473794551e-06,
+      "loss": 0.1303,
+      "step": 3760
+    },
+    {
+      "epoch": 2.634521313766597,
+      "grad_norm": 0.01970675028860569,
+      "learning_rate": 9.46470999301188e-06,
+      "loss": 0.0012,
+      "step": 3770
+    },
+    {
+      "epoch": 2.641509433962264,
+      "grad_norm": 0.004884845577180386,
+      "learning_rate": 9.43675751222921e-06,
+      "loss": 0.0307,
+      "step": 3780
+    },
+    {
+      "epoch": 2.6484975541579314,
+      "grad_norm": 0.08335088938474655,
+      "learning_rate": 9.408805031446541e-06,
+      "loss": 0.002,
+      "step": 3790
+    },
+    {
+      "epoch": 2.655485674353599,
+      "grad_norm": 0.012866518460214138,
+      "learning_rate": 9.380852550663873e-06,
+      "loss": 0.0006,
+      "step": 3800
+    },
+    {
+      "epoch": 2.6624737945492662,
+      "grad_norm": 0.024261610582470894,
+      "learning_rate": 9.352900069881202e-06,
+      "loss": 0.0012,
+      "step": 3810
+    },
+    {
+      "epoch": 2.6694619147449337,
+      "grad_norm": 0.00420184712857008,
+      "learning_rate": 9.324947589098533e-06,
+      "loss": 0.0076,
+      "step": 3820
+    },
+    {
+      "epoch": 2.676450034940601,
+      "grad_norm": 0.13258372247219086,
+      "learning_rate": 9.296995108315863e-06,
+      "loss": 0.0254,
+      "step": 3830
+    },
+    {
+      "epoch": 2.6834381551362685,
+      "grad_norm": 0.05237485095858574,
+      "learning_rate": 9.269042627533194e-06,
+      "loss": 0.2556,
+      "step": 3840
+    },
+    {
+      "epoch": 2.6904262753319355,
+      "grad_norm": 0.007042170502245426,
+      "learning_rate": 9.241090146750526e-06,
+      "loss": 0.0451,
+      "step": 3850
+    },
+    {
+      "epoch": 2.697414395527603,
+      "grad_norm": 0.0603499561548233,
+      "learning_rate": 9.213137665967855e-06,
+      "loss": 0.018,
+      "step": 3860
+    },
+    {
+      "epoch": 2.7044025157232703,
+      "grad_norm": 0.005806710571050644,
+      "learning_rate": 9.185185185185186e-06,
+      "loss": 0.0268,
+      "step": 3870
+    },
+    {
+      "epoch": 2.7113906359189377,
+      "grad_norm": 0.019446423277258873,
+      "learning_rate": 9.157232704402516e-06,
+      "loss": 0.0013,
+      "step": 3880
+    },
+    {
+      "epoch": 2.718378756114605,
+      "grad_norm": 0.046024855226278305,
+      "learning_rate": 9.129280223619847e-06,
+      "loss": 0.0094,
+      "step": 3890
+    },
+    {
+      "epoch": 2.7253668763102725,
+      "grad_norm": 0.2116568386554718,
+      "learning_rate": 9.101327742837177e-06,
+      "loss": 0.0079,
+      "step": 3900
+    },
+    {
+      "epoch": 2.73235499650594,
+      "grad_norm": 0.014955777674913406,
+      "learning_rate": 9.073375262054508e-06,
+      "loss": 0.0007,
+      "step": 3910
+    },
+    {
+      "epoch": 2.7393431167016074,
+      "grad_norm": 0.7381608486175537,
+      "learning_rate": 9.045422781271838e-06,
+      "loss": 0.0017,
+      "step": 3920
+    },
+    {
+      "epoch": 2.7463312368972748,
+      "grad_norm": 0.0042509473860263824,
+      "learning_rate": 9.017470300489169e-06,
+      "loss": 0.0328,
+      "step": 3930
+    },
+    {
+      "epoch": 2.753319357092942,
+      "grad_norm": 0.008610788732767105,
+      "learning_rate": 8.9895178197065e-06,
+      "loss": 0.0033,
+      "step": 3940
+    },
+    {
+      "epoch": 2.7603074772886096,
+      "grad_norm": 0.032417889684438705,
+      "learning_rate": 8.96156533892383e-06,
+      "loss": 0.0007,
+      "step": 3950
+    },
+    {
+      "epoch": 2.767295597484277,
+      "grad_norm": 0.003039875766262412,
+      "learning_rate": 8.93361285814116e-06,
+      "loss": 0.0006,
+      "step": 3960
+    },
+    {
+      "epoch": 2.774283717679944,
+      "grad_norm": 0.1048172265291214,
+      "learning_rate": 8.905660377358491e-06,
+      "loss": 0.0213,
+      "step": 3970
+    },
+    {
+      "epoch": 2.7812718378756114,
+      "grad_norm": 0.0031095799058675766,
+      "learning_rate": 8.877707896575822e-06,
+      "loss": 0.0358,
+      "step": 3980
+    },
+    {
+      "epoch": 2.788259958071279,
+      "grad_norm": 0.08362710475921631,
+      "learning_rate": 8.849755415793152e-06,
+      "loss": 0.0433,
+      "step": 3990
+    },
+    {
+      "epoch": 2.795248078266946,
+      "grad_norm": 0.006064527668058872,
+      "learning_rate": 8.821802935010483e-06,
+      "loss": 0.0009,
+      "step": 4000
+    },
+    {
+      "epoch": 2.8022361984626136,
+      "grad_norm": 0.009575744159519672,
+      "learning_rate": 8.793850454227814e-06,
+      "loss": 0.0003,
+      "step": 4010
+    },
+    {
+      "epoch": 2.809224318658281,
+      "grad_norm": 0.017887070775032043,
+      "learning_rate": 8.765897973445144e-06,
+      "loss": 0.0152,
+      "step": 4020
+    },
+    {
+      "epoch": 2.8162124388539485,
+      "grad_norm": 0.0015486754709854722,
+      "learning_rate": 8.737945492662475e-06,
+      "loss": 0.0692,
+      "step": 4030
+    },
+    {
+      "epoch": 2.8232005590496154,
+      "grad_norm": 0.00923781655728817,
+      "learning_rate": 8.709993011879805e-06,
+      "loss": 0.0006,
+      "step": 4040
+    },
+    {
+      "epoch": 2.830188679245283,
+      "grad_norm": 14.54555892944336,
+      "learning_rate": 8.682040531097136e-06,
+      "loss": 0.0804,
+      "step": 4050
+    },
+    {
+      "epoch": 2.8371767994409502,
+      "grad_norm": 0.035101618617773056,
+      "learning_rate": 8.654088050314466e-06,
+      "loss": 0.0077,
+      "step": 4060
+    },
+    {
+      "epoch": 2.8441649196366177,
+      "grad_norm": 0.22155028581619263,
+      "learning_rate": 8.626135569531797e-06,
+      "loss": 0.0257,
+      "step": 4070
+    },
+    {
+      "epoch": 2.851153039832285,
+      "grad_norm": 0.013333957642316818,
+      "learning_rate": 8.598183088749128e-06,
+      "loss": 0.0033,
+      "step": 4080
+    },
+    {
+      "epoch": 2.8581411600279525,
+      "grad_norm": 0.029933998361229897,
+      "learning_rate": 8.570230607966458e-06,
+      "loss": 0.0018,
+      "step": 4090
+    },
+    {
+      "epoch": 2.86512928022362,
+      "grad_norm": 0.021358752623200417,
+      "learning_rate": 8.542278127183789e-06,
+      "loss": 0.0004,
+      "step": 4100
+    },
+    {
+      "epoch": 2.8721174004192873,
+      "grad_norm": 0.0044855596497654915,
+      "learning_rate": 8.51432564640112e-06,
+      "loss": 0.0007,
+      "step": 4110
+    },
+    {
+      "epoch": 2.8791055206149547,
+      "grad_norm": 4.667630672454834,
+      "learning_rate": 8.486373165618448e-06,
+      "loss": 0.0307,
+      "step": 4120
+    },
+    {
+      "epoch": 2.886093640810622,
+      "grad_norm": 0.0070539130829274654,
+      "learning_rate": 8.45842068483578e-06,
+      "loss": 0.0006,
+      "step": 4130
+    },
+    {
+      "epoch": 2.8930817610062896,
+      "grad_norm": 0.029348071664571762,
+      "learning_rate": 8.430468204053111e-06,
+      "loss": 0.0014,
+      "step": 4140
+    },
+    {
+      "epoch": 2.9000698812019565,
+      "grad_norm": 0.12070006877183914,
+      "learning_rate": 8.402515723270442e-06,
+      "loss": 0.0043,
+      "step": 4150
+    },
+    {
+      "epoch": 2.907058001397624,
+      "grad_norm": 0.00619628606364131,
+      "learning_rate": 8.37456324248777e-06,
+      "loss": 0.0004,
+      "step": 4160
+    },
+    {
+      "epoch": 2.9140461215932913,
+      "grad_norm": 15.53851318359375,
+      "learning_rate": 8.346610761705101e-06,
+      "loss": 0.0074,
+      "step": 4170
+    },
+    {
+      "epoch": 2.9210342417889588,
+      "grad_norm": 9.77151870727539,
+      "learning_rate": 8.318658280922433e-06,
+      "loss": 0.1214,
+      "step": 4180
+    },
+    {
+      "epoch": 2.928022361984626,
+      "grad_norm": 0.0042813620530068874,
+      "learning_rate": 8.290705800139764e-06,
+      "loss": 0.0853,
+      "step": 4190
+    },
+    {
+      "epoch": 2.9350104821802936,
+      "grad_norm": 0.0018544025951996446,
+      "learning_rate": 8.262753319357094e-06,
+      "loss": 0.0183,
+      "step": 4200
+    },
+    {
+      "epoch": 2.941998602375961,
+      "grad_norm": 0.0065402681939303875,
+      "learning_rate": 8.234800838574423e-06,
+      "loss": 0.0746,
+      "step": 4210
+    },
+    {
+      "epoch": 2.948986722571628,
+      "grad_norm": 0.0039556859992444515,
+      "learning_rate": 8.206848357791754e-06,
+      "loss": 0.0101,
+      "step": 4220
+    },
+    {
+      "epoch": 2.9559748427672954,
+      "grad_norm": 0.8577067255973816,
+      "learning_rate": 8.178895877009086e-06,
+      "loss": 0.0323,
+      "step": 4230
+    },
+    {
+      "epoch": 2.962962962962963,
+      "grad_norm": 0.013186552561819553,
+      "learning_rate": 8.150943396226417e-06,
+      "loss": 0.0135,
+      "step": 4240
+    },
+    {
+      "epoch": 2.96995108315863,
+      "grad_norm": 0.008083418011665344,
+      "learning_rate": 8.122990915443745e-06,
+      "loss": 0.0133,
+      "step": 4250
+    },
+    {
+      "epoch": 2.9769392033542976,
+      "grad_norm": 13.401435852050781,
+      "learning_rate": 8.095038434661076e-06,
+      "loss": 0.0557,
+      "step": 4260
+    },
+    {
+      "epoch": 2.983927323549965,
+      "grad_norm": 0.00372733804397285,
+      "learning_rate": 8.067085953878407e-06,
+      "loss": 0.0125,
+      "step": 4270
+    },
+    {
+      "epoch": 2.9909154437456325,
+      "grad_norm": 0.21677538752555847,
+      "learning_rate": 8.039133473095739e-06,
+      "loss": 0.0026,
+      "step": 4280
+    },
+    {
+      "epoch": 2.9979035639413,
+      "grad_norm": 0.005610453896224499,
+      "learning_rate": 8.011180992313068e-06,
+      "loss": 0.0241,
+      "step": 4290
+    },
+    {
+      "epoch": 3.0048916841369673,
+      "grad_norm": 2.1690850257873535,
+      "learning_rate": 7.983228511530398e-06,
+      "loss": 0.0026,
+      "step": 4300
+    },
+    {
+      "epoch": 3.0118798043326347,
+      "grad_norm": 0.029843270778656006,
+      "learning_rate": 7.955276030747729e-06,
+      "loss": 0.0008,
+      "step": 4310
+    },
+    {
+      "epoch": 3.018867924528302,
+      "grad_norm": 0.0014904079725965858,
+      "learning_rate": 7.92732354996506e-06,
+      "loss": 0.0224,
+      "step": 4320
+    },
+    {
+      "epoch": 3.025856044723969,
+      "grad_norm": 3.553492546081543,
+      "learning_rate": 7.899371069182392e-06,
+      "loss": 0.0045,
+      "step": 4330
+    },
+    {
+      "epoch": 3.0328441649196365,
+      "grad_norm": 12.98043155670166,
+      "learning_rate": 7.87141858839972e-06,
+      "loss": 0.0501,
+      "step": 4340
+    },
+    {
+      "epoch": 3.039832285115304,
+      "grad_norm": 0.0013916384195908904,
+      "learning_rate": 7.843466107617051e-06,
+      "loss": 0.0005,
+      "step": 4350
+    },
+    {
+      "epoch": 3.0468204053109713,
+      "grad_norm": 0.0015694483881816268,
+      "learning_rate": 7.815513626834382e-06,
+      "loss": 0.1447,
+      "step": 4360
+    },
+    {
+      "epoch": 3.0538085255066387,
+      "grad_norm": 0.0035848869010806084,
+      "learning_rate": 7.787561146051712e-06,
+      "loss": 0.0015,
+      "step": 4370
+    },
+    {
+      "epoch": 3.060796645702306,
+      "grad_norm": 0.005436725448817015,
+      "learning_rate": 7.759608665269043e-06,
+      "loss": 0.0619,
+      "step": 4380
+    },
+    {
+      "epoch": 3.0677847658979736,
+      "grad_norm": 0.01864534057676792,
+      "learning_rate": 7.731656184486373e-06,
+      "loss": 0.013,
+      "step": 4390
+    },
+    {
+      "epoch": 3.074772886093641,
+      "grad_norm": 0.004284723661839962,
+      "learning_rate": 7.703703703703704e-06,
+      "loss": 0.0883,
+      "step": 4400
+    },
+    {
+      "epoch": 3.0817610062893084,
+      "grad_norm": 0.06042485311627388,
+      "learning_rate": 7.675751222921035e-06,
+      "loss": 0.001,
+      "step": 4410
+    },
+    {
+      "epoch": 3.0887491264849753,
+      "grad_norm": 0.005046827718615532,
+      "learning_rate": 7.647798742138365e-06,
+      "loss": 0.0218,
+      "step": 4420
+    },
+    {
+      "epoch": 3.0957372466806428,
+      "grad_norm": 6.6962361335754395,
+      "learning_rate": 7.619846261355696e-06,
+      "loss": 0.0592,
+      "step": 4430
+    },
+    {
+      "epoch": 3.10272536687631,
+      "grad_norm": 0.021092260256409645,
+      "learning_rate": 7.591893780573026e-06,
+      "loss": 0.0005,
+      "step": 4440
+    },
+    {
+      "epoch": 3.1097134870719776,
+      "grad_norm": 0.009671575389802456,
+      "learning_rate": 7.563941299790357e-06,
+      "loss": 0.0008,
+      "step": 4450
+    },
+    {
+      "epoch": 3.116701607267645,
+      "grad_norm": 0.011312313377857208,
+      "learning_rate": 7.535988819007688e-06,
+      "loss": 0.1175,
+      "step": 4460
+    },
+    {
+      "epoch": 3.1236897274633124,
+      "grad_norm": 0.009435279294848442,
+      "learning_rate": 7.508036338225018e-06,
+      "loss": 0.2328,
+      "step": 4470
+    },
+    {
+      "epoch": 3.13067784765898,
+      "grad_norm": 0.014669501222670078,
+      "learning_rate": 7.4800838574423485e-06,
+      "loss": 0.0003,
+      "step": 4480
+    },
+    {
+      "epoch": 3.1376659678546472,
+      "grad_norm": 0.005058380775153637,
+      "learning_rate": 7.452131376659679e-06,
+      "loss": 0.0698,
+      "step": 4490
+    },
+    {
+      "epoch": 3.1446540880503147,
+      "grad_norm": 0.004547697491943836,
+      "learning_rate": 7.42417889587701e-06,
+      "loss": 0.0962,
+      "step": 4500
+    },
+    {
+      "epoch": 3.1516422082459816,
+      "grad_norm": 0.030391795560717583,
+      "learning_rate": 7.396226415094339e-06,
+      "loss": 0.0011,
+      "step": 4510
+    },
+    {
+      "epoch": 3.158630328441649,
+      "grad_norm": 0.00580151192843914,
+      "learning_rate": 7.368273934311671e-06,
+      "loss": 0.0006,
+      "step": 4520
+    },
+    {
+      "epoch": 3.1656184486373165,
+      "grad_norm": 0.08100881427526474,
+      "learning_rate": 7.340321453529001e-06,
+      "loss": 0.11,
+      "step": 4530
+    },
+    {
+      "epoch": 3.172606568832984,
+      "grad_norm": 0.17724710702896118,
+      "learning_rate": 7.312368972746332e-06,
+      "loss": 0.0141,
+      "step": 4540
+    },
+    {
+      "epoch": 3.1795946890286513,
+      "grad_norm": 0.015520211309194565,
+      "learning_rate": 7.2844164919636625e-06,
+      "loss": 0.0051,
+      "step": 4550
+    },
+    {
+      "epoch": 3.1865828092243187,
+      "grad_norm": 0.06920566409826279,
+      "learning_rate": 7.256464011180992e-06,
+      "loss": 0.002,
+      "step": 4560
+    },
+    {
+      "epoch": 3.193570929419986,
+      "grad_norm": 0.036474499851465225,
+      "learning_rate": 7.228511530398324e-06,
+      "loss": 0.0007,
+      "step": 4570
+    },
+    {
+      "epoch": 3.2005590496156535,
+      "grad_norm": 0.0034632752649486065,
+      "learning_rate": 7.200559049615654e-06,
+      "loss": 0.0201,
+      "step": 4580
+    },
+    {
+      "epoch": 3.207547169811321,
+      "grad_norm": 0.01172909326851368,
+      "learning_rate": 7.172606568832985e-06,
+      "loss": 0.0072,
+      "step": 4590
+    },
+    {
+      "epoch": 3.214535290006988,
+      "grad_norm": 0.010598461143672466,
+      "learning_rate": 7.1446540880503145e-06,
+      "loss": 0.0035,
+      "step": 4600
+    },
+    {
+      "epoch": 3.2215234102026553,
+      "grad_norm": 0.0054387301206588745,
+      "learning_rate": 7.116701607267645e-06,
+      "loss": 0.0021,
+      "step": 4610
+    },
+    {
+      "epoch": 3.2285115303983227,
+      "grad_norm": 0.004587017931044102,
+      "learning_rate": 7.0887491264849764e-06,
+      "loss": 0.0556,
+      "step": 4620
+    },
+    {
+      "epoch": 3.23549965059399,
+      "grad_norm": 0.541749894618988,
+      "learning_rate": 7.060796645702307e-06,
+      "loss": 0.0277,
+      "step": 4630
+    },
+    {
+      "epoch": 3.2424877707896576,
+      "grad_norm": 0.020434614270925522,
+      "learning_rate": 7.032844164919638e-06,
+      "loss": 0.0006,
+      "step": 4640
+    },
+    {
+      "epoch": 3.249475890985325,
+      "grad_norm": 0.08088918775320053,
+      "learning_rate": 7.004891684136967e-06,
+      "loss": 0.0017,
+      "step": 4650
+    },
+    {
+      "epoch": 3.2564640111809924,
+      "grad_norm": 0.00651569152250886,
+      "learning_rate": 6.976939203354298e-06,
+      "loss": 0.0048,
+      "step": 4660
+    },
+    {
+      "epoch": 3.26345213137666,
+      "grad_norm": 0.012207509018480778,
+      "learning_rate": 6.948986722571629e-06,
+      "loss": 0.0527,
+      "step": 4670
+    },
+    {
+      "epoch": 3.270440251572327,
+      "grad_norm": 0.40066081285476685,
+      "learning_rate": 6.92103424178896e-06,
+      "loss": 0.0013,
+      "step": 4680
+    },
+    {
+      "epoch": 3.2774283717679946,
+      "grad_norm": 0.037062227725982666,
+      "learning_rate": 6.8930817610062896e-06,
+      "loss": 0.0767,
+      "step": 4690
+    },
+    {
+      "epoch": 3.2844164919636616,
+      "grad_norm": 0.002083055442199111,
+      "learning_rate": 6.86512928022362e-06,
+      "loss": 0.0313,
+      "step": 4700
+    },
+    {
+      "epoch": 3.291404612159329,
+      "grad_norm": 0.01235890295356512,
+      "learning_rate": 6.837176799440951e-06,
+      "loss": 0.0003,
+      "step": 4710
+    },
+    {
+      "epoch": 3.2983927323549964,
+      "grad_norm": 0.0049998946487903595,
+      "learning_rate": 6.809224318658282e-06,
+      "loss": 0.0005,
+      "step": 4720
+    },
+    {
+      "epoch": 3.305380852550664,
+      "grad_norm": 0.017238592728972435,
+      "learning_rate": 6.781271837875612e-06,
+      "loss": 0.0005,
+      "step": 4730
+    },
+    {
+      "epoch": 3.3123689727463312,
+      "grad_norm": 0.0085734399035573,
+      "learning_rate": 6.753319357092942e-06,
+      "loss": 0.0008,
+      "step": 4740
+    },
+    {
+      "epoch": 3.3193570929419987,
+      "grad_norm": 0.02260555699467659,
+      "learning_rate": 6.725366876310273e-06,
+      "loss": 0.0006,
+      "step": 4750
+    },
+    {
+      "epoch": 3.326345213137666,
+      "grad_norm": 0.006640274077653885,
+      "learning_rate": 6.6974143955276035e-06,
+      "loss": 0.0547,
+      "step": 4760
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.006710624787956476,
+      "learning_rate": 6.669461914744935e-06,
+      "loss": 0.0517,
+      "step": 4770
+    },
+    {
+      "epoch": 3.3403214535290005,
+      "grad_norm": 0.015057160519063473,
+      "learning_rate": 6.641509433962265e-06,
+      "loss": 0.0007,
+      "step": 4780
+    },
+    {
+      "epoch": 3.347309573724668,
+      "grad_norm": 0.01974908635020256,
+      "learning_rate": 6.613556953179595e-06,
+      "loss": 0.0005,
+      "step": 4790
+    },
+    {
+      "epoch": 3.3542976939203353,
+      "grad_norm": 0.005637224763631821,
+      "learning_rate": 6.585604472396926e-06,
+      "loss": 0.0327,
+      "step": 4800
+    },
+    {
+      "epoch": 3.3612858141160027,
+      "grad_norm": 0.0025775651447474957,
+      "learning_rate": 6.557651991614256e-06,
+      "loss": 0.0791,
+      "step": 4810
+    },
+    {
+      "epoch": 3.36827393431167,
+      "grad_norm": 0.005838762037456036,
+      "learning_rate": 6.529699510831586e-06,
+      "loss": 0.0004,
+      "step": 4820
+    },
+    {
+      "epoch": 3.3752620545073375,
+      "grad_norm": 0.3478299379348755,
+      "learning_rate": 6.5017470300489175e-06,
+      "loss": 0.0011,
+      "step": 4830
+    },
+    {
+      "epoch": 3.382250174703005,
+      "grad_norm": 0.005249143578112125,
+      "learning_rate": 6.473794549266248e-06,
+      "loss": 0.0006,
+      "step": 4840
+    },
+    {
+      "epoch": 3.3892382948986723,
+      "grad_norm": 1.6451876163482666,
+      "learning_rate": 6.445842068483579e-06,
+      "loss": 0.0598,
+      "step": 4850
+    },
+    {
+      "epoch": 3.3962264150943398,
+      "grad_norm": 0.005776327569037676,
+      "learning_rate": 6.417889587700909e-06,
+      "loss": 0.0008,
+      "step": 4860
+    },
+    {
+      "epoch": 3.403214535290007,
+      "grad_norm": 0.0033416105434298515,
+      "learning_rate": 6.389937106918239e-06,
+      "loss": 0.0572,
+      "step": 4870
+    },
+    {
+      "epoch": 3.4102026554856746,
+      "grad_norm": 0.002429103944450617,
+      "learning_rate": 6.36198462613557e-06,
+      "loss": 0.1072,
+      "step": 4880
+    },
+    {
+      "epoch": 3.4171907756813416,
+      "grad_norm": 8.708498001098633,
+      "learning_rate": 6.334032145352901e-06,
+      "loss": 0.1464,
+      "step": 4890
+    },
+    {
+      "epoch": 3.424178895877009,
+      "grad_norm": 0.0021170196123421192,
+      "learning_rate": 6.3060796645702315e-06,
+      "loss": 0.0004,
+      "step": 4900
+    },
+    {
+      "epoch": 3.4311670160726764,
+      "grad_norm": 0.012093136087059975,
+      "learning_rate": 6.278127183787561e-06,
+      "loss": 0.0619,
+      "step": 4910
+    },
+    {
+      "epoch": 3.438155136268344,
+      "grad_norm": 0.005036477465182543,
+      "learning_rate": 6.250174703004892e-06,
+      "loss": 0.0004,
+      "step": 4920
+    },
+    {
+      "epoch": 3.445143256464011,
+      "grad_norm": 0.004095244687050581,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 0.0004,
+      "step": 4930
+    },
+    {
+      "epoch": 3.4521313766596786,
+      "grad_norm": 0.019728029146790504,
+      "learning_rate": 6.194269741439554e-06,
+      "loss": 0.0406,
+      "step": 4940
+    },
+    {
+      "epoch": 3.459119496855346,
+      "grad_norm": 0.025431908667087555,
+      "learning_rate": 6.1663172606568835e-06,
+      "loss": 0.0018,
+      "step": 4950
+    },
+    {
+      "epoch": 3.4661076170510134,
+      "grad_norm": 0.0277637280523777,
+      "learning_rate": 6.138364779874214e-06,
+      "loss": 0.0029,
+      "step": 4960
+    },
+    {
+      "epoch": 3.4730957372466804,
+      "grad_norm": 0.4063994288444519,
+      "learning_rate": 6.110412299091545e-06,
+      "loss": 0.0864,
+      "step": 4970
+    },
+    {
+      "epoch": 3.480083857442348,
+      "grad_norm": 0.06079300493001938,
+      "learning_rate": 6.082459818308876e-06,
+      "loss": 0.003,
+      "step": 4980
+    },
+    {
+      "epoch": 3.4870719776380152,
+      "grad_norm": 30.48621368408203,
+      "learning_rate": 6.0545073375262066e-06,
+      "loss": 0.0506,
+      "step": 4990
+    },
+    {
+      "epoch": 3.4940600978336827,
+      "grad_norm": 0.0112813301384449,
+      "learning_rate": 6.026554856743536e-06,
+      "loss": 0.0003,
+      "step": 5000
+    },
+    {
+      "epoch": 3.50104821802935,
+      "grad_norm": 0.005831410177052021,
+      "learning_rate": 5.998602375960867e-06,
+      "loss": 0.0006,
+      "step": 5010
+    },
+    {
+      "epoch": 3.5080363382250175,
+      "grad_norm": 0.0054182917810976505,
+      "learning_rate": 5.970649895178197e-06,
+      "loss": 0.0004,
+      "step": 5020
+    },
+    {
+      "epoch": 3.515024458420685,
+      "grad_norm": 0.12203137576580048,
+      "learning_rate": 5.942697414395529e-06,
+      "loss": 0.1176,
+      "step": 5030
+    },
+    {
+      "epoch": 3.5220125786163523,
+      "grad_norm": 0.002840982051566243,
+      "learning_rate": 5.9147449336128585e-06,
+      "loss": 0.0133,
+      "step": 5040
+    },
+    {
+      "epoch": 3.5290006988120197,
+      "grad_norm": 0.026787206530570984,
+      "learning_rate": 5.886792452830189e-06,
+      "loss": 0.089,
+      "step": 5050
+    },
+    {
+      "epoch": 3.535988819007687,
+      "grad_norm": 0.06147437170147896,
+      "learning_rate": 5.85883997204752e-06,
+      "loss": 0.0035,
+      "step": 5060
+    },
+    {
+      "epoch": 3.5429769392033545,
+      "grad_norm": 9.293349266052246,
+      "learning_rate": 5.83088749126485e-06,
+      "loss": 0.0678,
+      "step": 5070
+    },
+    {
+      "epoch": 3.5499650593990215,
+      "grad_norm": 0.006805673241615295,
+      "learning_rate": 5.80293501048218e-06,
+      "loss": 0.0056,
+      "step": 5080
+    },
+    {
+      "epoch": 3.556953179594689,
+      "grad_norm": 0.006203502882272005,
+      "learning_rate": 5.774982529699511e-06,
+      "loss": 0.06,
+      "step": 5090
+    },
+    {
+      "epoch": 3.5639412997903563,
+      "grad_norm": 9.436474800109863,
+      "learning_rate": 5.747030048916842e-06,
+      "loss": 0.1251,
+      "step": 5100
+    },
+    {
+      "epoch": 3.5709294199860238,
+      "grad_norm": 0.007246874738484621,
+      "learning_rate": 5.7190775681341725e-06,
+      "loss": 0.0004,
+      "step": 5110
+    },
+    {
+      "epoch": 3.577917540181691,
+      "grad_norm": 0.0030432320199906826,
+      "learning_rate": 5.691125087351503e-06,
+      "loss": 0.0005,
+      "step": 5120
+    },
+    {
+      "epoch": 3.5849056603773586,
+      "grad_norm": 0.011605805717408657,
+      "learning_rate": 5.663172606568833e-06,
+      "loss": 0.0247,
+      "step": 5130
+    },
+    {
+      "epoch": 3.591893780573026,
+      "grad_norm": 0.06202976033091545,
+      "learning_rate": 5.635220125786164e-06,
+      "loss": 0.0597,
+      "step": 5140
+    },
+    {
+      "epoch": 3.598881900768693,
+      "grad_norm": 0.0037594842724502087,
+      "learning_rate": 5.607267645003495e-06,
+      "loss": 0.0011,
+      "step": 5150
+    },
+    {
+      "epoch": 3.6058700209643604,
+      "grad_norm": 0.006377144251018763,
+      "learning_rate": 5.579315164220825e-06,
+      "loss": 0.0016,
+      "step": 5160
+    },
+    {
+      "epoch": 3.612858141160028,
+      "grad_norm": 0.006404062733054161,
+      "learning_rate": 5.551362683438155e-06,
+      "loss": 0.1364,
+      "step": 5170
+    },
+    {
+      "epoch": 3.619846261355695,
+      "grad_norm": 0.00531811686232686,
+      "learning_rate": 5.523410202655486e-06,
+      "loss": 0.0012,
+      "step": 5180
+    },
+    {
+      "epoch": 3.6268343815513626,
+      "grad_norm": 0.002592286095023155,
+      "learning_rate": 5.495457721872817e-06,
+      "loss": 0.0004,
+      "step": 5190
+    },
+    {
+      "epoch": 3.63382250174703,
+      "grad_norm": 0.013801293447613716,
+      "learning_rate": 5.467505241090148e-06,
+      "loss": 0.0106,
+      "step": 5200
+    },
+    {
+      "epoch": 3.6408106219426974,
+      "grad_norm": 0.0015551424585282803,
+      "learning_rate": 5.439552760307478e-06,
+      "loss": 0.0004,
+      "step": 5210
+    },
+    {
+      "epoch": 3.647798742138365,
+      "grad_norm": 0.013982684351503849,
+      "learning_rate": 5.411600279524808e-06,
+      "loss": 0.0004,
+      "step": 5220
+    },
+    {
+      "epoch": 3.6547868623340323,
+      "grad_norm": 0.007577819749712944,
+      "learning_rate": 5.3836477987421385e-06,
+      "loss": 0.0201,
+      "step": 5230
+    },
+    {
+      "epoch": 3.6617749825296997,
+      "grad_norm": 0.0024994234554469585,
+      "learning_rate": 5.35569531795947e-06,
+      "loss": 0.0014,
+      "step": 5240
+    },
+    {
+      "epoch": 3.668763102725367,
+      "grad_norm": 0.303627073764801,
+      "learning_rate": 5.3277428371768004e-06,
+      "loss": 0.0006,
+      "step": 5250
+    },
+    {
+      "epoch": 3.6757512229210345,
+      "grad_norm": 0.0030300321523100138,
+      "learning_rate": 5.29979035639413e-06,
+      "loss": 0.0011,
+      "step": 5260
+    },
+    {
+      "epoch": 3.6827393431167015,
+      "grad_norm": 0.02062870003283024,
+      "learning_rate": 5.271837875611461e-06,
+      "loss": 0.0583,
+      "step": 5270
+    },
+    {
+      "epoch": 3.689727463312369,
+      "grad_norm": 0.005427168216556311,
+      "learning_rate": 5.243885394828791e-06,
+      "loss": 0.0355,
+      "step": 5280
+    },
+    {
+      "epoch": 3.6967155835080363,
+      "grad_norm": 0.14737765491008759,
+      "learning_rate": 5.215932914046123e-06,
+      "loss": 0.0011,
+      "step": 5290
+    },
+    {
+      "epoch": 3.7037037037037037,
+      "grad_norm": 0.047865480184555054,
+      "learning_rate": 5.1879804332634524e-06,
+      "loss": 0.0003,
+      "step": 5300
+    },
+    {
+      "epoch": 3.710691823899371,
+      "grad_norm": 0.005643247161060572,
+      "learning_rate": 5.160027952480783e-06,
+      "loss": 0.0703,
+      "step": 5310
+    },
+    {
+      "epoch": 3.7176799440950385,
+      "grad_norm": 0.002993800677359104,
+      "learning_rate": 5.1320754716981136e-06,
+      "loss": 0.0003,
+      "step": 5320
+    },
+    {
+      "epoch": 3.724668064290706,
+      "grad_norm": 0.0012869026977568865,
+      "learning_rate": 5.104122990915444e-06,
+      "loss": 0.0002,
+      "step": 5330
+    },
+    {
+      "epoch": 3.731656184486373,
+      "grad_norm": 0.015380630269646645,
+      "learning_rate": 5.0761705101327755e-06,
+      "loss": 0.0116,
+      "step": 5340
+    },
+    {
+      "epoch": 3.7386443046820403,
+      "grad_norm": 0.012133465148508549,
+      "learning_rate": 5.048218029350105e-06,
+      "loss": 0.0437,
+      "step": 5350
+    },
+    {
+      "epoch": 3.7456324248777078,
+      "grad_norm": 0.004281846806406975,
+      "learning_rate": 5.020265548567436e-06,
+      "loss": 0.0008,
+      "step": 5360
+    },
+    {
+      "epoch": 3.752620545073375,
+      "grad_norm": 7.5166401863098145,
+      "learning_rate": 4.992313067784766e-06,
+      "loss": 0.0455,
+      "step": 5370
+    },
+    {
+      "epoch": 3.7596086652690426,
+      "grad_norm": 0.04558609798550606,
+      "learning_rate": 4.964360587002097e-06,
+      "loss": 0.0005,
+      "step": 5380
+    },
+    {
+      "epoch": 3.76659678546471,
+      "grad_norm": 0.006235541310161352,
+      "learning_rate": 4.9364081062194275e-06,
+      "loss": 0.0413,
+      "step": 5390
+    },
+    {
+      "epoch": 3.7735849056603774,
+      "grad_norm": 0.003967160824686289,
+      "learning_rate": 4.908455625436758e-06,
+      "loss": 0.0004,
+      "step": 5400
+    },
+    {
+      "epoch": 3.780573025856045,
+      "grad_norm": 0.006097977515310049,
+      "learning_rate": 4.880503144654089e-06,
+      "loss": 0.0004,
+      "step": 5410
+    },
+    {
+      "epoch": 3.7875611460517122,
+      "grad_norm": 0.0025531253777444363,
+      "learning_rate": 4.852550663871418e-06,
+      "loss": 0.071,
+      "step": 5420
+    },
+    {
+      "epoch": 3.7945492662473796,
+      "grad_norm": 0.058856215327978134,
+      "learning_rate": 4.82459818308875e-06,
+      "loss": 0.0088,
+      "step": 5430
+    },
+    {
+      "epoch": 3.801537386443047,
+      "grad_norm": 0.0011401502415537834,
+      "learning_rate": 4.79664570230608e-06,
+      "loss": 0.0007,
+      "step": 5440
+    },
+    {
+      "epoch": 3.808525506638714,
+      "grad_norm": 0.18255279958248138,
+      "learning_rate": 4.768693221523411e-06,
+      "loss": 0.0009,
+      "step": 5450
+    },
+    {
+      "epoch": 3.8155136268343814,
+      "grad_norm": 0.0024249760899692774,
+      "learning_rate": 4.7407407407407415e-06,
+      "loss": 0.0001,
+      "step": 5460
+    },
+    {
+      "epoch": 3.822501747030049,
+      "grad_norm": 0.002625074004754424,
+      "learning_rate": 4.712788259958071e-06,
+      "loss": 0.0173,
+      "step": 5470
+    },
+    {
+      "epoch": 3.8294898672257163,
+      "grad_norm": 0.00629691407084465,
+      "learning_rate": 4.684835779175403e-06,
+      "loss": 0.0003,
+      "step": 5480
+    },
+    {
+      "epoch": 3.8364779874213837,
+      "grad_norm": 0.007325862068682909,
+      "learning_rate": 4.656883298392732e-06,
+      "loss": 0.0002,
+      "step": 5490
+    },
+    {
+      "epoch": 3.843466107617051,
+      "grad_norm": 1.3780497312545776,
+      "learning_rate": 4.628930817610064e-06,
+      "loss": 0.0025,
+      "step": 5500
+    },
+    {
+      "epoch": 3.8504542278127185,
+      "grad_norm": 0.002903768327087164,
+      "learning_rate": 4.6009783368273935e-06,
+      "loss": 0.0613,
+      "step": 5510
+    },
+    {
+      "epoch": 3.8574423480083855,
+      "grad_norm": 3.889617443084717,
+      "learning_rate": 4.573025856044724e-06,
+      "loss": 0.0677,
+      "step": 5520
+    },
+    {
+      "epoch": 3.864430468204053,
+      "grad_norm": 0.0012939589796587825,
+      "learning_rate": 4.545073375262055e-06,
+      "loss": 0.0002,
+      "step": 5530
+    },
+    {
+      "epoch": 3.8714185883997203,
+      "grad_norm": 0.0032597167883068323,
+      "learning_rate": 4.517120894479385e-06,
+      "loss": 0.0685,
+      "step": 5540
+    },
+    {
+      "epoch": 3.8784067085953877,
+      "grad_norm": 0.004127015825361013,
+      "learning_rate": 4.489168413696717e-06,
+      "loss": 0.006,
+      "step": 5550
+    },
+    {
+      "epoch": 3.885394828791055,
+      "grad_norm": 0.005040404852479696,
+      "learning_rate": 4.461215932914046e-06,
+      "loss": 0.0003,
+      "step": 5560
+    },
+    {
+      "epoch": 3.8923829489867225,
+      "grad_norm": 0.006766372825950384,
+      "learning_rate": 4.433263452131377e-06,
+      "loss": 0.0013,
+      "step": 5570
+    },
+    {
+      "epoch": 3.89937106918239,
+      "grad_norm": 0.014896499924361706,
+      "learning_rate": 4.4053109713487075e-06,
+      "loss": 0.1589,
+      "step": 5580
+    },
+    {
+      "epoch": 3.9063591893780574,
+      "grad_norm": 0.008297051303088665,
+      "learning_rate": 4.377358490566038e-06,
+      "loss": 0.0012,
+      "step": 5590
+    },
+    {
+      "epoch": 3.913347309573725,
+      "grad_norm": 0.015372090972959995,
+      "learning_rate": 4.349406009783369e-06,
+      "loss": 0.0002,
+      "step": 5600
+    },
+    {
+      "epoch": 3.920335429769392,
+      "grad_norm": 0.008451149798929691,
+      "learning_rate": 4.321453529000699e-06,
+      "loss": 0.0004,
+      "step": 5610
+    },
+    {
+      "epoch": 3.9273235499650596,
+      "grad_norm": 0.014892960898578167,
+      "learning_rate": 4.29350104821803e-06,
+      "loss": 0.0779,
+      "step": 5620
+    },
+    {
+      "epoch": 3.934311670160727,
+      "grad_norm": 0.008910837583243847,
+      "learning_rate": 4.26554856743536e-06,
+      "loss": 0.0491,
+      "step": 5630
+    },
+    {
+      "epoch": 3.941299790356394,
+      "grad_norm": 0.004672807175666094,
+      "learning_rate": 4.237596086652691e-06,
+      "loss": 0.0003,
+      "step": 5640
+    },
+    {
+      "epoch": 3.9482879105520614,
+      "grad_norm": 0.002980695106089115,
+      "learning_rate": 4.209643605870021e-06,
+      "loss": 0.0008,
+      "step": 5650
+    },
+    {
+      "epoch": 3.955276030747729,
+      "grad_norm": 0.0031832789536565542,
+      "learning_rate": 4.181691125087352e-06,
+      "loss": 0.0776,
+      "step": 5660
+    },
+    {
+      "epoch": 3.9622641509433962,
+      "grad_norm": 0.00444215489551425,
+      "learning_rate": 4.1537386443046825e-06,
+      "loss": 0.0002,
+      "step": 5670
+    },
+    {
+      "epoch": 3.9692522711390636,
+      "grad_norm": 0.005608571693301201,
+      "learning_rate": 4.125786163522013e-06,
+      "loss": 0.031,
+      "step": 5680
+    },
+    {
+      "epoch": 3.976240391334731,
+      "grad_norm": 0.009694025851786137,
+      "learning_rate": 4.097833682739344e-06,
+      "loss": 0.1584,
+      "step": 5690
+    },
+    {
+      "epoch": 3.9832285115303985,
+      "grad_norm": 26.081716537475586,
+      "learning_rate": 4.069881201956674e-06,
+      "loss": 0.0181,
+      "step": 5700
+    },
+    {
+      "epoch": 3.9902166317260654,
+      "grad_norm": 0.0018772291950881481,
+      "learning_rate": 4.041928721174004e-06,
+      "loss": 0.0108,
+      "step": 5710
+    },
+    {
+      "epoch": 3.997204751921733,
+      "grad_norm": 0.0026039350777864456,
+      "learning_rate": 4.013976240391335e-06,
+      "loss": 0.0536,
+      "step": 5720
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7155,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6187647906816000.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}