diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,108534 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.1219932390859013,
+  "eval_steps": 500,
+  "global_step": 155000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 7.238666058618717e-05,
+      "grad_norm": 0.5545526742935181,
+      "learning_rate": 4.999934852005473e-06,
+      "loss": 1.5857,
+      "step": 10
+    },
+    {
+      "epoch": 0.00014477332117237434,
+      "grad_norm": 0.44032129645347595,
+      "learning_rate": 4.999862465344887e-06,
+      "loss": 1.5257,
+      "step": 20
+    },
+    {
+      "epoch": 0.00021715998175856154,
+      "grad_norm": 0.40077871084213257,
+      "learning_rate": 4.999790078684301e-06,
+      "loss": 1.4955,
+      "step": 30
+    },
+    {
+      "epoch": 0.0002895466423447487,
+      "grad_norm": 0.3221718966960907,
+      "learning_rate": 4.999717692023714e-06,
+      "loss": 1.4903,
+      "step": 40
+    },
+    {
+      "epoch": 0.00036193330293093586,
+      "grad_norm": 0.3151167631149292,
+      "learning_rate": 4.999645305363128e-06,
+      "loss": 1.4603,
+      "step": 50
+    },
+    {
+      "epoch": 0.0004343199635171231,
+      "grad_norm": 0.27057284116744995,
+      "learning_rate": 4.9995729187025424e-06,
+      "loss": 1.4502,
+      "step": 60
+    },
+    {
+      "epoch": 0.0005067066241033103,
+      "grad_norm": 0.3064720928668976,
+      "learning_rate": 4.999500532041956e-06,
+      "loss": 1.4473,
+      "step": 70
+    },
+    {
+      "epoch": 0.0005790932846894974,
+      "grad_norm": 0.2519373595714569,
+      "learning_rate": 4.99942814538137e-06,
+      "loss": 1.4421,
+      "step": 80
+    },
+    {
+      "epoch": 0.0006514799452756846,
+      "grad_norm": 0.23665937781333923,
+      "learning_rate": 4.999355758720783e-06,
+      "loss": 1.4143,
+      "step": 90
+    },
+    {
+      "epoch": 0.0007238666058618717,
+      "grad_norm": 0.26254650950431824,
+      "learning_rate": 4.999283372060198e-06,
+      "loss": 1.4186,
+      "step": 100
+    },
+    {
+      "epoch": 0.000796253266448059,
+      "grad_norm": 0.21838980913162231,
+      "learning_rate": 4.9992109853996105e-06,
+      "loss": 1.3901,
+      "step": 110
+    },
+    {
+      "epoch": 0.0008686399270342462,
+      "grad_norm": 0.22787261009216309,
+      "learning_rate": 4.999138598739024e-06,
+      "loss": 1.393,
+      "step": 120
+    },
+    {
+      "epoch": 0.0009410265876204333,
+      "grad_norm": 0.24213093519210815,
+      "learning_rate": 4.999066212078439e-06,
+      "loss": 1.3777,
+      "step": 130
+    },
+    {
+      "epoch": 0.0010134132482066205,
+      "grad_norm": 0.31372249126434326,
+      "learning_rate": 4.998993825417852e-06,
+      "loss": 1.3858,
+      "step": 140
+    },
+    {
+      "epoch": 0.0010857999087928076,
+      "grad_norm": 0.23661810159683228,
+      "learning_rate": 4.998921438757266e-06,
+      "loss": 1.3904,
+      "step": 150
+    },
+    {
+      "epoch": 0.0011581865693789948,
+      "grad_norm": 0.245035320520401,
+      "learning_rate": 4.9988490520966794e-06,
+      "loss": 1.3719,
+      "step": 160
+    },
+    {
+      "epoch": 0.001230573229965182,
+      "grad_norm": 0.2323773056268692,
+      "learning_rate": 4.998776665436094e-06,
+      "loss": 1.3662,
+      "step": 170
+    },
+    {
+      "epoch": 0.0013029598905513692,
+      "grad_norm": 0.20842741429805756,
+      "learning_rate": 4.9987042787755075e-06,
+      "loss": 1.3549,
+      "step": 180
+    },
+    {
+      "epoch": 0.0013753465511375563,
+      "grad_norm": 0.2077568769454956,
+      "learning_rate": 4.998631892114921e-06,
+      "loss": 1.351,
+      "step": 190
+    },
+    {
+      "epoch": 0.0014477332117237434,
+      "grad_norm": 0.2298651784658432,
+      "learning_rate": 4.998559505454335e-06,
+      "loss": 1.3661,
+      "step": 200
+    },
+    {
+      "epoch": 0.0015201198723099308,
+      "grad_norm": 0.23133844137191772,
+      "learning_rate": 4.998487118793749e-06,
+      "loss": 1.349,
+      "step": 210
+    },
+    {
+      "epoch": 0.001592506532896118,
+      "grad_norm": 0.2279757559299469,
+      "learning_rate": 4.998414732133163e-06,
+      "loss": 1.3391,
+      "step": 220
+    },
+    {
+      "epoch": 0.001664893193482305,
+      "grad_norm": 0.22140687704086304,
+      "learning_rate": 4.9983423454725765e-06,
+      "loss": 1.3491,
+      "step": 230
+    },
+    {
+      "epoch": 0.0017372798540684924,
+      "grad_norm": 0.2116747796535492,
+      "learning_rate": 4.99826995881199e-06,
+      "loss": 1.3488,
+      "step": 240
+    },
+    {
+      "epoch": 0.0018096665146546795,
+      "grad_norm": 0.2300599366426468,
+      "learning_rate": 4.9981975721514045e-06,
+      "loss": 1.3383,
+      "step": 250
+    },
+    {
+      "epoch": 0.0018820531752408666,
+      "grad_norm": 0.2189347743988037,
+      "learning_rate": 4.998125185490818e-06,
+      "loss": 1.3214,
+      "step": 260
+    },
+    {
+      "epoch": 0.001954439835827054,
+      "grad_norm": 0.2163819819688797,
+      "learning_rate": 4.998052798830232e-06,
+      "loss": 1.331,
+      "step": 270
+    },
+    {
+      "epoch": 0.002026826496413241,
+      "grad_norm": 0.24370183050632477,
+      "learning_rate": 4.997980412169645e-06,
+      "loss": 1.3307,
+      "step": 280
+    },
+    {
+      "epoch": 0.002099213156999428,
+      "grad_norm": 0.21121959388256073,
+      "learning_rate": 4.99790802550906e-06,
+      "loss": 1.326,
+      "step": 290
+    },
+    {
+      "epoch": 0.0021715998175856153,
+      "grad_norm": 0.2269212156534195,
+      "learning_rate": 4.9978356388484735e-06,
+      "loss": 1.3192,
+      "step": 300
+    },
+    {
+      "epoch": 0.0022439864781718024,
+      "grad_norm": 0.19639967381954193,
+      "learning_rate": 4.997763252187887e-06,
+      "loss": 1.3322,
+      "step": 310
+    },
+    {
+      "epoch": 0.0023163731387579895,
+      "grad_norm": 0.2111475169658661,
+      "learning_rate": 4.997690865527301e-06,
+      "loss": 1.3282,
+      "step": 320
+    },
+    {
+      "epoch": 0.002388759799344177,
+      "grad_norm": 0.21942846477031708,
+      "learning_rate": 4.997618478866715e-06,
+      "loss": 1.3358,
+      "step": 330
+    },
+    {
+      "epoch": 0.002461146459930364,
+      "grad_norm": 0.24269279837608337,
+      "learning_rate": 4.997546092206129e-06,
+      "loss": 1.2974,
+      "step": 340
+    },
+    {
+      "epoch": 0.0025335331205165513,
+      "grad_norm": 0.2175322324037552,
+      "learning_rate": 4.997473705545542e-06,
+      "loss": 1.3239,
+      "step": 350
+    },
+    {
+      "epoch": 0.0026059197811027384,
+      "grad_norm": 0.2417760044336319,
+      "learning_rate": 4.997401318884956e-06,
+      "loss": 1.3016,
+      "step": 360
+    },
+    {
+      "epoch": 0.0026783064416889255,
+      "grad_norm": 0.21008457243442535,
+      "learning_rate": 4.9973289322243705e-06,
+      "loss": 1.3162,
+      "step": 370
+    },
+    {
+      "epoch": 0.0027506931022751127,
+      "grad_norm": 0.21002964675426483,
+      "learning_rate": 4.997256545563784e-06,
+      "loss": 1.2806,
+      "step": 380
+    },
+    {
+      "epoch": 0.0028230797628612998,
+      "grad_norm": 0.2070387601852417,
+      "learning_rate": 4.997184158903198e-06,
+      "loss": 1.3012,
+      "step": 390
+    },
+    {
+      "epoch": 0.002895466423447487,
+      "grad_norm": 0.22568874061107635,
+      "learning_rate": 4.997111772242611e-06,
+      "loss": 1.2917,
+      "step": 400
+    },
+    {
+      "epoch": 0.0029678530840336744,
+      "grad_norm": 0.21147729456424713,
+      "learning_rate": 4.997039385582026e-06,
+      "loss": 1.3087,
+      "step": 410
+    },
+    {
+      "epoch": 0.0030402397446198616,
+      "grad_norm": 0.2232261747121811,
+      "learning_rate": 4.996966998921439e-06,
+      "loss": 1.3038,
+      "step": 420
+    },
+    {
+      "epoch": 0.0031126264052060487,
+      "grad_norm": 0.2272012084722519,
+      "learning_rate": 4.996894612260853e-06,
+      "loss": 1.2891,
+      "step": 430
+    },
+    {
+      "epoch": 0.003185013065792236,
+      "grad_norm": 0.20684392750263214,
+      "learning_rate": 4.996822225600267e-06,
+      "loss": 1.2942,
+      "step": 440
+    },
+    {
+      "epoch": 0.003257399726378423,
+      "grad_norm": 0.20101316273212433,
+      "learning_rate": 4.996749838939681e-06,
+      "loss": 1.2889,
+      "step": 450
+    },
+    {
+      "epoch": 0.00332978638696461,
+      "grad_norm": 0.22801408171653748,
+      "learning_rate": 4.996677452279095e-06,
+      "loss": 1.285,
+      "step": 460
+    },
+    {
+      "epoch": 0.003402173047550797,
+      "grad_norm": 0.222897469997406,
+      "learning_rate": 4.996605065618508e-06,
+      "loss": 1.2837,
+      "step": 470
+    },
+    {
+      "epoch": 0.0034745597081369847,
+      "grad_norm": 0.2286527454853058,
+      "learning_rate": 4.996532678957922e-06,
+      "loss": 1.2894,
+      "step": 480
+    },
+    {
+      "epoch": 0.003546946368723172,
+      "grad_norm": 0.23111595213413239,
+      "learning_rate": 4.9964602922973356e-06,
+      "loss": 1.2842,
+      "step": 490
+    },
+    {
+      "epoch": 0.003619333029309359,
+      "grad_norm": 0.20559774339199066,
+      "learning_rate": 4.99638790563675e-06,
+      "loss": 1.2789,
+      "step": 500
+    },
+    {
+      "epoch": 0.003691719689895546,
+      "grad_norm": 0.2076191008090973,
+      "learning_rate": 4.996315518976164e-06,
+      "loss": 1.2871,
+      "step": 510
+    },
+    {
+      "epoch": 0.003764106350481733,
+      "grad_norm": 0.1894315779209137,
+      "learning_rate": 4.996243132315577e-06,
+      "loss": 1.2697,
+      "step": 520
+    },
+    {
+      "epoch": 0.0038364930110679203,
+      "grad_norm": 0.21035927534103394,
+      "learning_rate": 4.996170745654991e-06,
+      "loss": 1.2744,
+      "step": 530
+    },
+    {
+      "epoch": 0.003908879671654108,
+      "grad_norm": 0.22824759781360626,
+      "learning_rate": 4.996098358994405e-06,
+      "loss": 1.2881,
+      "step": 540
+    },
+    {
+      "epoch": 0.003981266332240295,
+      "grad_norm": 0.21655461192131042,
+      "learning_rate": 4.996025972333819e-06,
+      "loss": 1.273,
+      "step": 550
+    },
+    {
+      "epoch": 0.004053652992826482,
+      "grad_norm": 0.2505953907966614,
+      "learning_rate": 4.9959535856732326e-06,
+      "loss": 1.2879,
+      "step": 560
+    },
+    {
+      "epoch": 0.004126039653412669,
+      "grad_norm": 0.21900534629821777,
+      "learning_rate": 4.995881199012646e-06,
+      "loss": 1.2763,
+      "step": 570
+    },
+    {
+      "epoch": 0.004198426313998856,
+      "grad_norm": 0.21123918890953064,
+      "learning_rate": 4.995808812352061e-06,
+      "loss": 1.2833,
+      "step": 580
+    },
+    {
+      "epoch": 0.0042708129745850434,
+      "grad_norm": 0.20754094421863556,
+      "learning_rate": 4.995736425691474e-06,
+      "loss": 1.2692,
+      "step": 590
+    },
+    {
+      "epoch": 0.0043431996351712306,
+      "grad_norm": 0.23484139144420624,
+      "learning_rate": 4.995664039030888e-06,
+      "loss": 1.2596,
+      "step": 600
+    },
+    {
+      "epoch": 0.004415586295757418,
+      "grad_norm": 0.2571958303451538,
+      "learning_rate": 4.9955916523703015e-06,
+      "loss": 1.2714,
+      "step": 610
+    },
+    {
+      "epoch": 0.004487972956343605,
+      "grad_norm": 0.21375253796577454,
+      "learning_rate": 4.995519265709716e-06,
+      "loss": 1.2718,
+      "step": 620
+    },
+    {
+      "epoch": 0.004560359616929792,
+      "grad_norm": 0.20837347209453583,
+      "learning_rate": 4.9954468790491296e-06,
+      "loss": 1.2555,
+      "step": 630
+    },
+    {
+      "epoch": 0.004632746277515979,
+      "grad_norm": 0.21926173567771912,
+      "learning_rate": 4.995374492388543e-06,
+      "loss": 1.2565,
+      "step": 640
+    },
+    {
+      "epoch": 0.004705132938102166,
+      "grad_norm": 0.21038663387298584,
+      "learning_rate": 4.995302105727957e-06,
+      "loss": 1.2713,
+      "step": 650
+    },
+    {
+      "epoch": 0.004777519598688354,
+      "grad_norm": 0.2313099056482315,
+      "learning_rate": 4.99522971906737e-06,
+      "loss": 1.2441,
+      "step": 660
+    },
+    {
+      "epoch": 0.004849906259274541,
+      "grad_norm": 0.2143174558877945,
+      "learning_rate": 4.995157332406784e-06,
+      "loss": 1.2497,
+      "step": 670
+    },
+    {
+      "epoch": 0.004922292919860728,
+      "grad_norm": 0.21607981622219086,
+      "learning_rate": 4.995084945746198e-06,
+      "loss": 1.2455,
+      "step": 680
+    },
+    {
+      "epoch": 0.0049946795804469155,
+      "grad_norm": 0.2096307873725891,
+      "learning_rate": 4.995012559085612e-06,
+      "loss": 1.2614,
+      "step": 690
+    },
+    {
+      "epoch": 0.005067066241033103,
+      "grad_norm": 0.20711645483970642,
+      "learning_rate": 4.994940172425026e-06,
+      "loss": 1.2529,
+      "step": 700
+    },
+    {
+      "epoch": 0.00513945290161929,
+      "grad_norm": 0.2200125753879547,
+      "learning_rate": 4.994867785764439e-06,
+      "loss": 1.2626,
+      "step": 710
+    },
+    {
+      "epoch": 0.005211839562205477,
+      "grad_norm": 0.24058933556079865,
+      "learning_rate": 4.994795399103853e-06,
+      "loss": 1.2615,
+      "step": 720
+    },
+    {
+      "epoch": 0.005284226222791664,
+      "grad_norm": 0.21051721274852753,
+      "learning_rate": 4.9947230124432674e-06,
+      "loss": 1.2608,
+      "step": 730
+    },
+    {
+      "epoch": 0.005356612883377851,
+      "grad_norm": 0.18748053908348083,
+      "learning_rate": 4.994650625782681e-06,
+      "loss": 1.2579,
+      "step": 740
+    },
+    {
+      "epoch": 0.005428999543964038,
+      "grad_norm": 0.4778112769126892,
+      "learning_rate": 4.994578239122095e-06,
+      "loss": 1.2514,
+      "step": 750
+    },
+    {
+      "epoch": 0.005501386204550225,
+      "grad_norm": 0.22301283478736877,
+      "learning_rate": 4.994505852461508e-06,
+      "loss": 1.2391,
+      "step": 760
+    },
+    {
+      "epoch": 0.0055737728651364124,
+      "grad_norm": 0.21762309968471527,
+      "learning_rate": 4.994433465800923e-06,
+      "loss": 1.2562,
+      "step": 770
+    },
+    {
+      "epoch": 0.0056461595257225996,
+      "grad_norm": 0.216139554977417,
+      "learning_rate": 4.994361079140336e-06,
+      "loss": 1.2587,
+      "step": 780
+    },
+    {
+      "epoch": 0.005718546186308787,
+      "grad_norm": 0.19104599952697754,
+      "learning_rate": 4.99428869247975e-06,
+      "loss": 1.2622,
+      "step": 790
+    },
+    {
+      "epoch": 0.005790932846894974,
+      "grad_norm": 0.205087810754776,
+      "learning_rate": 4.994216305819164e-06,
+      "loss": 1.2465,
+      "step": 800
+    },
+    {
+      "epoch": 0.005863319507481162,
+      "grad_norm": 0.20174795389175415,
+      "learning_rate": 4.994143919158578e-06,
+      "loss": 1.2285,
+      "step": 810
+    },
+    {
+      "epoch": 0.005935706168067349,
+      "grad_norm": 0.1900462955236435,
+      "learning_rate": 4.994071532497992e-06,
+      "loss": 1.2433,
+      "step": 820
+    },
+    {
+      "epoch": 0.006008092828653536,
+      "grad_norm": 0.21302379667758942,
+      "learning_rate": 4.993999145837405e-06,
+      "loss": 1.2355,
+      "step": 830
+    },
+    {
+      "epoch": 0.006080479489239723,
+      "grad_norm": 0.19449925422668457,
+      "learning_rate": 4.993926759176819e-06,
+      "loss": 1.2538,
+      "step": 840
+    },
+    {
+      "epoch": 0.00615286614982591,
+      "grad_norm": 0.2037796676158905,
+      "learning_rate": 4.993854372516233e-06,
+      "loss": 1.2643,
+      "step": 850
+    },
+    {
+      "epoch": 0.006225252810412097,
+      "grad_norm": 0.22023706138134003,
+      "learning_rate": 4.993781985855647e-06,
+      "loss": 1.2358,
+      "step": 860
+    },
+    {
+      "epoch": 0.0062976394709982845,
+      "grad_norm": 0.1888815462589264,
+      "learning_rate": 4.993709599195061e-06,
+      "loss": 1.2499,
+      "step": 870
+    },
+    {
+      "epoch": 0.006370026131584472,
+      "grad_norm": 0.18520642817020416,
+      "learning_rate": 4.993637212534474e-06,
+      "loss": 1.2459,
+      "step": 880
+    },
+    {
+      "epoch": 0.006442412792170659,
+      "grad_norm": 0.1910271942615509,
+      "learning_rate": 4.993564825873889e-06,
+      "loss": 1.2406,
+      "step": 890
+    },
+    {
+      "epoch": 0.006514799452756846,
+      "grad_norm": 0.1944640874862671,
+      "learning_rate": 4.993492439213302e-06,
+      "loss": 1.2331,
+      "step": 900
+    },
+    {
+      "epoch": 0.006587186113343033,
+      "grad_norm": 0.19614140689373016,
+      "learning_rate": 4.993420052552716e-06,
+      "loss": 1.2319,
+      "step": 910
+    },
+    {
+      "epoch": 0.00665957277392922,
+      "grad_norm": 0.20917336642742157,
+      "learning_rate": 4.9933476658921295e-06,
+      "loss": 1.2433,
+      "step": 920
+    },
+    {
+      "epoch": 0.006731959434515407,
+      "grad_norm": 0.18660324811935425,
+      "learning_rate": 4.993275279231544e-06,
+      "loss": 1.2266,
+      "step": 930
+    },
+    {
+      "epoch": 0.006804346095101594,
+      "grad_norm": 0.20741026103496552,
+      "learning_rate": 4.993202892570958e-06,
+      "loss": 1.2376,
+      "step": 940
+    },
+    {
+      "epoch": 0.006876732755687781,
+      "grad_norm": 0.20696569979190826,
+      "learning_rate": 4.993130505910371e-06,
+      "loss": 1.2479,
+      "step": 950
+    },
+    {
+      "epoch": 0.006949119416273969,
+      "grad_norm": 0.19880063831806183,
+      "learning_rate": 4.993058119249785e-06,
+      "loss": 1.246,
+      "step": 960
+    },
+    {
+      "epoch": 0.0070215060768601565,
+      "grad_norm": 0.21153710782527924,
+      "learning_rate": 4.992985732589199e-06,
+      "loss": 1.2399,
+      "step": 970
+    },
+    {
+      "epoch": 0.007093892737446344,
+      "grad_norm": 0.22593656182289124,
+      "learning_rate": 4.992913345928613e-06,
+      "loss": 1.2423,
+      "step": 980
+    },
+    {
+      "epoch": 0.007166279398032531,
+      "grad_norm": 0.22546051442623138,
+      "learning_rate": 4.9928409592680265e-06,
+      "loss": 1.2099,
+      "step": 990
+    },
+    {
+      "epoch": 0.007238666058618718,
+      "grad_norm": 0.2021579146385193,
+      "learning_rate": 4.99276857260744e-06,
+      "loss": 1.2264,
+      "step": 1000
+    },
+    {
+      "epoch": 0.007311052719204905,
+      "grad_norm": 0.2759735584259033,
+      "learning_rate": 4.992696185946855e-06,
+      "loss": 1.2178,
+      "step": 1010
+    },
+    {
+      "epoch": 0.007383439379791092,
+      "grad_norm": 0.20149949193000793,
+      "learning_rate": 4.992623799286268e-06,
+      "loss": 1.2289,
+      "step": 1020
+    },
+    {
+      "epoch": 0.007455826040377279,
+      "grad_norm": 0.1932690441608429,
+      "learning_rate": 4.992551412625682e-06,
+      "loss": 1.2506,
+      "step": 1030
+    },
+    {
+      "epoch": 0.007528212700963466,
+      "grad_norm": 0.2068718671798706,
+      "learning_rate": 4.9924790259650955e-06,
+      "loss": 1.215,
+      "step": 1040
+    },
+    {
+      "epoch": 0.0076005993615496535,
+      "grad_norm": 0.22313687205314636,
+      "learning_rate": 4.99240663930451e-06,
+      "loss": 1.2356,
+      "step": 1050
+    },
+    {
+      "epoch": 0.007672986022135841,
+      "grad_norm": 0.2034347951412201,
+      "learning_rate": 4.9923342526439235e-06,
+      "loss": 1.2195,
+      "step": 1060
+    },
+    {
+      "epoch": 0.007745372682722028,
+      "grad_norm": 0.2164289355278015,
+      "learning_rate": 4.992261865983337e-06,
+      "loss": 1.2069,
+      "step": 1070
+    },
+    {
+      "epoch": 0.007817759343308216,
+      "grad_norm": 0.20517635345458984,
+      "learning_rate": 4.992189479322751e-06,
+      "loss": 1.2181,
+      "step": 1080
+    },
+    {
+      "epoch": 0.007890146003894403,
+      "grad_norm": 0.19486500322818756,
+      "learning_rate": 4.992117092662165e-06,
+      "loss": 1.229,
+      "step": 1090
+    },
+    {
+      "epoch": 0.00796253266448059,
+      "grad_norm": 0.20392891764640808,
+      "learning_rate": 4.992044706001579e-06,
+      "loss": 1.2325,
+      "step": 1100
+    },
+    {
+      "epoch": 0.008034919325066777,
+      "grad_norm": 0.19018247723579407,
+      "learning_rate": 4.9919723193409925e-06,
+      "loss": 1.2242,
+      "step": 1110
+    },
+    {
+      "epoch": 0.008107305985652964,
+      "grad_norm": 0.2185811698436737,
+      "learning_rate": 4.991899932680406e-06,
+      "loss": 1.2287,
+      "step": 1120
+    },
+    {
+      "epoch": 0.008179692646239151,
+      "grad_norm": 0.20160779356956482,
+      "learning_rate": 4.99182754601982e-06,
+      "loss": 1.2131,
+      "step": 1130
+    },
+    {
+      "epoch": 0.008252079306825338,
+      "grad_norm": 0.19665688276290894,
+      "learning_rate": 4.991755159359234e-06,
+      "loss": 1.219,
+      "step": 1140
+    },
+    {
+      "epoch": 0.008324465967411526,
+      "grad_norm": 0.19984516501426697,
+      "learning_rate": 4.991682772698648e-06,
+      "loss": 1.241,
+      "step": 1150
+    },
+    {
+      "epoch": 0.008396852627997713,
+      "grad_norm": 0.19308070838451385,
+      "learning_rate": 4.991610386038061e-06,
+      "loss": 1.213,
+      "step": 1160
+    },
+    {
+      "epoch": 0.0084692392885839,
+      "grad_norm": 0.2210283875465393,
+      "learning_rate": 4.991537999377475e-06,
+      "loss": 1.2214,
+      "step": 1170
+    },
+    {
+      "epoch": 0.008541625949170087,
+      "grad_norm": 0.19817933440208435,
+      "learning_rate": 4.991465612716889e-06,
+      "loss": 1.2446,
+      "step": 1180
+    },
+    {
+      "epoch": 0.008614012609756274,
+      "grad_norm": 0.198905810713768,
+      "learning_rate": 4.991393226056302e-06,
+      "loss": 1.2252,
+      "step": 1190
+    },
+    {
+      "epoch": 0.008686399270342461,
+      "grad_norm": 0.21226166188716888,
+      "learning_rate": 4.991320839395717e-06,
+      "loss": 1.2189,
+      "step": 1200
+    },
+    {
+      "epoch": 0.008758785930928648,
+      "grad_norm": 0.20412813127040863,
+      "learning_rate": 4.99124845273513e-06,
+      "loss": 1.2299,
+      "step": 1210
+    },
+    {
+      "epoch": 0.008831172591514835,
+      "grad_norm": 0.18132169544696808,
+      "learning_rate": 4.991176066074544e-06,
+      "loss": 1.2236,
+      "step": 1220
+    },
+    {
+      "epoch": 0.008903559252101022,
+      "grad_norm": 0.2028530240058899,
+      "learning_rate": 4.9911036794139576e-06,
+      "loss": 1.2205,
+      "step": 1230
+    },
+    {
+      "epoch": 0.00897594591268721,
+      "grad_norm": 0.21371830999851227,
+      "learning_rate": 4.991031292753372e-06,
+      "loss": 1.213,
+      "step": 1240
+    },
+    {
+      "epoch": 0.009048332573273397,
+      "grad_norm": 0.20845739543437958,
+      "learning_rate": 4.990958906092786e-06,
+      "loss": 1.2153,
+      "step": 1250
+    },
+    {
+      "epoch": 0.009120719233859584,
+      "grad_norm": 0.19972443580627441,
+      "learning_rate": 4.990886519432199e-06,
+      "loss": 1.2211,
+      "step": 1260
+    },
+    {
+      "epoch": 0.009193105894445771,
+      "grad_norm": 0.21171468496322632,
+      "learning_rate": 4.990814132771613e-06,
+      "loss": 1.2159,
+      "step": 1270
+    },
+    {
+      "epoch": 0.009265492555031958,
+      "grad_norm": 0.18781454861164093,
+      "learning_rate": 4.9907417461110265e-06,
+      "loss": 1.2305,
+      "step": 1280
+    },
+    {
+      "epoch": 0.009337879215618145,
+      "grad_norm": 0.20194801688194275,
+      "learning_rate": 4.990669359450441e-06,
+      "loss": 1.2119,
+      "step": 1290
+    },
+    {
+      "epoch": 0.009410265876204332,
+      "grad_norm": 0.20740105211734772,
+      "learning_rate": 4.9905969727898546e-06,
+      "loss": 1.2171,
+      "step": 1300
+    },
+    {
+      "epoch": 0.00948265253679052,
+      "grad_norm": 0.19401483237743378,
+      "learning_rate": 4.990524586129268e-06,
+      "loss": 1.2198,
+      "step": 1310
+    },
+    {
+      "epoch": 0.009555039197376708,
+      "grad_norm": 0.21922868490219116,
+      "learning_rate": 4.990452199468682e-06,
+      "loss": 1.2135,
+      "step": 1320
+    },
+    {
+      "epoch": 0.009627425857962895,
+      "grad_norm": 0.19779783487319946,
+      "learning_rate": 4.990379812808096e-06,
+      "loss": 1.2111,
+      "step": 1330
+    },
+    {
+      "epoch": 0.009699812518549083,
+      "grad_norm": 0.23003704845905304,
+      "learning_rate": 4.99030742614751e-06,
+      "loss": 1.2131,
+      "step": 1340
+    },
+    {
+      "epoch": 0.00977219917913527,
+      "grad_norm": 0.2075931429862976,
+      "learning_rate": 4.9902350394869235e-06,
+      "loss": 1.2128,
+      "step": 1350
+    },
+    {
+      "epoch": 0.009844585839721457,
+      "grad_norm": 0.19751273095607758,
+      "learning_rate": 4.990162652826337e-06,
+      "loss": 1.199,
+      "step": 1360
+    },
+    {
+      "epoch": 0.009916972500307644,
+      "grad_norm": 0.20674757659435272,
+      "learning_rate": 4.9900902661657516e-06,
+      "loss": 1.2362,
+      "step": 1370
+    },
+    {
+      "epoch": 0.009989359160893831,
+      "grad_norm": 0.19264833629131317,
+      "learning_rate": 4.990017879505165e-06,
+      "loss": 1.229,
+      "step": 1380
+    },
+    {
+      "epoch": 0.010061745821480018,
+      "grad_norm": 0.19390186667442322,
+      "learning_rate": 4.989945492844579e-06,
+      "loss": 1.2308,
+      "step": 1390
+    },
+    {
+      "epoch": 0.010134132482066205,
+      "grad_norm": 0.20780542492866516,
+      "learning_rate": 4.989873106183992e-06,
+      "loss": 1.1996,
+      "step": 1400
+    },
+    {
+      "epoch": 0.010206519142652392,
+      "grad_norm": 0.20205456018447876,
+      "learning_rate": 4.989800719523407e-06,
+      "loss": 1.2062,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01027890580323858,
+      "grad_norm": 0.2114095389842987,
+      "learning_rate": 4.9897283328628205e-06,
+      "loss": 1.2147,
+      "step": 1420
+    },
+    {
+      "epoch": 0.010351292463824767,
+      "grad_norm": 0.2318940907716751,
+      "learning_rate": 4.989655946202234e-06,
+      "loss": 1.2316,
+      "step": 1430
+    },
+    {
+      "epoch": 0.010423679124410954,
+      "grad_norm": 0.20557528734207153,
+      "learning_rate": 4.989583559541648e-06,
+      "loss": 1.2094,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01049606578499714,
+      "grad_norm": 0.23943296074867249,
+      "learning_rate": 4.989511172881062e-06,
+      "loss": 1.2044,
+      "step": 1450
+    },
+    {
+      "epoch": 0.010568452445583328,
+      "grad_norm": 0.19559936225414276,
+      "learning_rate": 4.989438786220476e-06,
+      "loss": 1.2108,
+      "step": 1460
+    },
+    {
+      "epoch": 0.010640839106169515,
+      "grad_norm": 0.22766897082328796,
+      "learning_rate": 4.9893663995598894e-06,
+      "loss": 1.2156,
+      "step": 1470
+    },
+    {
+      "epoch": 0.010713225766755702,
+      "grad_norm": 0.1910102516412735,
+      "learning_rate": 4.989294012899303e-06,
+      "loss": 1.2147,
+      "step": 1480
+    },
+    {
+      "epoch": 0.01078561242734189,
+      "grad_norm": 0.20596715807914734,
+      "learning_rate": 4.9892216262387175e-06,
+      "loss": 1.2146,
+      "step": 1490
+    },
+    {
+      "epoch": 0.010857999087928076,
+      "grad_norm": 0.21965187788009644,
+      "learning_rate": 4.989149239578131e-06,
+      "loss": 1.2115,
+      "step": 1500
+    },
+    {
+      "epoch": 0.010930385748514264,
+      "grad_norm": 0.2052324265241623,
+      "learning_rate": 4.989076852917545e-06,
+      "loss": 1.2269,
+      "step": 1510
+    },
+    {
+      "epoch": 0.01100277240910045,
+      "grad_norm": 0.19294176995754242,
+      "learning_rate": 4.989004466256958e-06,
+      "loss": 1.2152,
+      "step": 1520
+    },
+    {
+      "epoch": 0.011075159069686638,
+      "grad_norm": 0.20509877800941467,
+      "learning_rate": 4.988932079596373e-06,
+      "loss": 1.2036,
+      "step": 1530
+    },
+    {
+      "epoch": 0.011147545730272825,
+      "grad_norm": 0.20573221147060394,
+      "learning_rate": 4.9888596929357864e-06,
+      "loss": 1.2033,
+      "step": 1540
+    },
+    {
+      "epoch": 0.011219932390859012,
+      "grad_norm": 0.1944323480129242,
+      "learning_rate": 4.9887873062752e-06,
+      "loss": 1.2002,
+      "step": 1550
+    },
+    {
+      "epoch": 0.011292319051445199,
+      "grad_norm": 0.20238761603832245,
+      "learning_rate": 4.988714919614614e-06,
+      "loss": 1.1859,
+      "step": 1560
+    },
+    {
+      "epoch": 0.011364705712031386,
+      "grad_norm": 0.21487738192081451,
+      "learning_rate": 4.988642532954028e-06,
+      "loss": 1.1908,
+      "step": 1570
+    },
+    {
+      "epoch": 0.011437092372617573,
+      "grad_norm": 0.19267559051513672,
+      "learning_rate": 4.988570146293442e-06,
+      "loss": 1.2108,
+      "step": 1580
+    },
+    {
+      "epoch": 0.01150947903320376,
+      "grad_norm": 0.21015968918800354,
+      "learning_rate": 4.988497759632855e-06,
+      "loss": 1.1977,
+      "step": 1590
+    },
+    {
+      "epoch": 0.011581865693789948,
+      "grad_norm": 0.1899135857820511,
+      "learning_rate": 4.988425372972269e-06,
+      "loss": 1.1996,
+      "step": 1600
+    },
+    {
+      "epoch": 0.011654252354376135,
+      "grad_norm": 0.21136519312858582,
+      "learning_rate": 4.9883529863116834e-06,
+      "loss": 1.2074,
+      "step": 1610
+    },
+    {
+      "epoch": 0.011726639014962324,
+      "grad_norm": 0.18629726767539978,
+      "learning_rate": 4.988280599651097e-06,
+      "loss": 1.1995,
+      "step": 1620
+    },
+    {
+      "epoch": 0.01179902567554851,
+      "grad_norm": 0.18262408673763275,
+      "learning_rate": 4.988208212990511e-06,
+      "loss": 1.1972,
+      "step": 1630
+    },
+    {
+      "epoch": 0.011871412336134698,
+      "grad_norm": 0.216887965798378,
+      "learning_rate": 4.988135826329924e-06,
+      "loss": 1.1925,
+      "step": 1640
+    },
+    {
+      "epoch": 0.011943798996720885,
+      "grad_norm": 0.18681691586971283,
+      "learning_rate": 4.988063439669339e-06,
+      "loss": 1.21,
+      "step": 1650
+    },
+    {
+      "epoch": 0.012016185657307072,
+      "grad_norm": 0.199030801653862,
+      "learning_rate": 4.987991053008752e-06,
+      "loss": 1.2014,
+      "step": 1660
+    },
+    {
+      "epoch": 0.01208857231789326,
+      "grad_norm": 0.23438072204589844,
+      "learning_rate": 4.987918666348166e-06,
+      "loss": 1.1913,
+      "step": 1670
+    },
+    {
+      "epoch": 0.012160958978479446,
+      "grad_norm": 0.18666905164718628,
+      "learning_rate": 4.98784627968758e-06,
+      "loss": 1.2013,
+      "step": 1680
+    },
+    {
+      "epoch": 0.012233345639065633,
+      "grad_norm": 0.20759981870651245,
+      "learning_rate": 4.987773893026994e-06,
+      "loss": 1.2058,
+      "step": 1690
+    },
+    {
+      "epoch": 0.01230573229965182,
+      "grad_norm": 0.20983868837356567,
+      "learning_rate": 4.987701506366407e-06,
+      "loss": 1.2153,
+      "step": 1700
+    },
+    {
+      "epoch": 0.012378118960238008,
+      "grad_norm": 0.21793246269226074,
+      "learning_rate": 4.9876291197058205e-06,
+      "loss": 1.1994,
+      "step": 1710
+    },
+    {
+      "epoch": 0.012450505620824195,
+      "grad_norm": 0.20397932827472687,
+      "learning_rate": 4.987556733045235e-06,
+      "loss": 1.2112,
+      "step": 1720
+    },
+    {
+      "epoch": 0.012522892281410382,
+      "grad_norm": 0.202115997672081,
+      "learning_rate": 4.9874843463846485e-06,
+      "loss": 1.2132,
+      "step": 1730
+    },
+    {
+      "epoch": 0.012595278941996569,
+      "grad_norm": 0.1943732500076294,
+      "learning_rate": 4.987411959724062e-06,
+      "loss": 1.1877,
+      "step": 1740
+    },
+    {
+      "epoch": 0.012667665602582756,
+      "grad_norm": 0.20084722340106964,
+      "learning_rate": 4.987339573063476e-06,
+      "loss": 1.1773,
+      "step": 1750
+    },
+    {
+      "epoch": 0.012740052263168943,
+      "grad_norm": 0.19970197975635529,
+      "learning_rate": 4.98726718640289e-06,
+      "loss": 1.1826,
+      "step": 1760
+    },
+    {
+      "epoch": 0.01281243892375513,
+      "grad_norm": 0.20870880782604218,
+      "learning_rate": 4.987194799742304e-06,
+      "loss": 1.1833,
+      "step": 1770
+    },
+    {
+      "epoch": 0.012884825584341317,
+      "grad_norm": 0.2071852833032608,
+      "learning_rate": 4.9871224130817175e-06,
+      "loss": 1.1955,
+      "step": 1780
+    },
+    {
+      "epoch": 0.012957212244927505,
+      "grad_norm": 0.21110813319683075,
+      "learning_rate": 4.987050026421131e-06,
+      "loss": 1.1876,
+      "step": 1790
+    },
+    {
+      "epoch": 0.013029598905513692,
+      "grad_norm": 0.18131721019744873,
+      "learning_rate": 4.9869776397605455e-06,
+      "loss": 1.1867,
+      "step": 1800
+    },
+    {
+      "epoch": 0.013101985566099879,
+      "grad_norm": 0.22368177771568298,
+      "learning_rate": 4.986905253099959e-06,
+      "loss": 1.199,
+      "step": 1810
+    },
+    {
+      "epoch": 0.013174372226686066,
+      "grad_norm": 0.19988280534744263,
+      "learning_rate": 4.986832866439373e-06,
+      "loss": 1.1972,
+      "step": 1820
+    },
+    {
+      "epoch": 0.013246758887272253,
+      "grad_norm": 0.2156451791524887,
+      "learning_rate": 4.986760479778786e-06,
+      "loss": 1.1999,
+      "step": 1830
+    },
+    {
+      "epoch": 0.01331914554785844,
+      "grad_norm": 0.1884111911058426,
+      "learning_rate": 4.986688093118201e-06,
+      "loss": 1.2085,
+      "step": 1840
+    },
+    {
+      "epoch": 0.013391532208444627,
+      "grad_norm": 0.20055466890335083,
+      "learning_rate": 4.9866157064576145e-06,
+      "loss": 1.2014,
+      "step": 1850
+    },
+    {
+      "epoch": 0.013463918869030814,
+      "grad_norm": 0.22643277049064636,
+      "learning_rate": 4.986543319797028e-06,
+      "loss": 1.194,
+      "step": 1860
+    },
+    {
+      "epoch": 0.013536305529617002,
+      "grad_norm": 0.2079644799232483,
+      "learning_rate": 4.986470933136442e-06,
+      "loss": 1.1998,
+      "step": 1870
+    },
+    {
+      "epoch": 0.013608692190203189,
+      "grad_norm": 0.19740234315395355,
+      "learning_rate": 4.986398546475856e-06,
+      "loss": 1.1917,
+      "step": 1880
+    },
+    {
+      "epoch": 0.013681078850789376,
+      "grad_norm": 0.18155385553836823,
+      "learning_rate": 4.98632615981527e-06,
+      "loss": 1.1904,
+      "step": 1890
+    },
+    {
+      "epoch": 0.013753465511375563,
+      "grad_norm": 0.1898881047964096,
+      "learning_rate": 4.986253773154683e-06,
+      "loss": 1.2094,
+      "step": 1900
+    },
+    {
+      "epoch": 0.013825852171961752,
+      "grad_norm": 0.19113953411579132,
+      "learning_rate": 4.986181386494097e-06,
+      "loss": 1.1973,
+      "step": 1910
+    },
+    {
+      "epoch": 0.013898238832547939,
+      "grad_norm": 0.18314586579799652,
+      "learning_rate": 4.986108999833511e-06,
+      "loss": 1.1954,
+      "step": 1920
+    },
+    {
+      "epoch": 0.013970625493134126,
+      "grad_norm": 0.21783554553985596,
+      "learning_rate": 4.986036613172925e-06,
+      "loss": 1.1947,
+      "step": 1930
+    },
+    {
+      "epoch": 0.014043012153720313,
+      "grad_norm": 0.18775740265846252,
+      "learning_rate": 4.985964226512339e-06,
+      "loss": 1.1773,
+      "step": 1940
+    },
+    {
+      "epoch": 0.0141153988143065,
+      "grad_norm": 0.1981116682291031,
+      "learning_rate": 4.985891839851752e-06,
+      "loss": 1.1887,
+      "step": 1950
+    },
+    {
+      "epoch": 0.014187785474892687,
+      "grad_norm": 0.19518138468265533,
+      "learning_rate": 4.985819453191166e-06,
+      "loss": 1.1832,
+      "step": 1960
+    },
+    {
+      "epoch": 0.014260172135478874,
+      "grad_norm": 0.18999353051185608,
+      "learning_rate": 4.98574706653058e-06,
+      "loss": 1.1854,
+      "step": 1970
+    },
+    {
+      "epoch": 0.014332558796065062,
+      "grad_norm": 0.18030518293380737,
+      "learning_rate": 4.985674679869994e-06,
+      "loss": 1.1873,
+      "step": 1980
+    },
+    {
+      "epoch": 0.014404945456651249,
+      "grad_norm": 0.19327767193317413,
+      "learning_rate": 4.985602293209408e-06,
+      "loss": 1.1978,
+      "step": 1990
+    },
+    {
+      "epoch": 0.014477332117237436,
+      "grad_norm": 0.18995395302772522,
+      "learning_rate": 4.985529906548821e-06,
+      "loss": 1.1941,
+      "step": 2000
+    },
+    {
+      "epoch": 0.014549718777823623,
+      "grad_norm": 0.19528520107269287,
+      "learning_rate": 4.985457519888236e-06,
+      "loss": 1.1841,
+      "step": 2010
+    },
+    {
+      "epoch": 0.01462210543840981,
+      "grad_norm": 0.20308247208595276,
+      "learning_rate": 4.985385133227649e-06,
+      "loss": 1.19,
+      "step": 2020
+    },
+    {
+      "epoch": 0.014694492098995997,
+      "grad_norm": 0.22538486123085022,
+      "learning_rate": 4.985312746567063e-06,
+      "loss": 1.1979,
+      "step": 2030
+    },
+    {
+      "epoch": 0.014766878759582184,
+      "grad_norm": 0.19460053741931915,
+      "learning_rate": 4.9852403599064766e-06,
+      "loss": 1.1837,
+      "step": 2040
+    },
+    {
+      "epoch": 0.014839265420168371,
+      "grad_norm": 0.25119996070861816,
+      "learning_rate": 4.985167973245891e-06,
+      "loss": 1.1851,
+      "step": 2050
+    },
+    {
+      "epoch": 0.014911652080754558,
+      "grad_norm": 0.1847628504037857,
+      "learning_rate": 4.985095586585305e-06,
+      "loss": 1.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.014984038741340746,
+      "grad_norm": 0.18682530522346497,
+      "learning_rate": 4.985023199924718e-06,
+      "loss": 1.1957,
+      "step": 2070
+    },
+    {
+      "epoch": 0.015056425401926933,
+      "grad_norm": 0.18913131952285767,
+      "learning_rate": 4.984950813264132e-06,
+      "loss": 1.1825,
+      "step": 2080
+    },
+    {
+      "epoch": 0.01512881206251312,
+      "grad_norm": 0.18790295720100403,
+      "learning_rate": 4.984878426603546e-06,
+      "loss": 1.1794,
+      "step": 2090
+    },
+    {
+      "epoch": 0.015201198723099307,
+      "grad_norm": 0.17898432910442352,
+      "learning_rate": 4.98480603994296e-06,
+      "loss": 1.1643,
+      "step": 2100
+    },
+    {
+      "epoch": 0.015273585383685494,
+      "grad_norm": 0.18662859499454498,
+      "learning_rate": 4.9847336532823736e-06,
+      "loss": 1.1796,
+      "step": 2110
+    },
+    {
+      "epoch": 0.015345972044271681,
+      "grad_norm": 0.1902911514043808,
+      "learning_rate": 4.984661266621787e-06,
+      "loss": 1.1846,
+      "step": 2120
+    },
+    {
+      "epoch": 0.015418358704857868,
+      "grad_norm": 0.21089226007461548,
+      "learning_rate": 4.984588879961202e-06,
+      "loss": 1.1826,
+      "step": 2130
+    },
+    {
+      "epoch": 0.015490745365444055,
+      "grad_norm": 0.19871971011161804,
+      "learning_rate": 4.984516493300615e-06,
+      "loss": 1.1881,
+      "step": 2140
+    },
+    {
+      "epoch": 0.015563132026030243,
+      "grad_norm": 0.1826990395784378,
+      "learning_rate": 4.984444106640029e-06,
+      "loss": 1.1654,
+      "step": 2150
+    },
+    {
+      "epoch": 0.01563551868661643,
+      "grad_norm": 0.22978819906711578,
+      "learning_rate": 4.9843717199794425e-06,
+      "loss": 1.1748,
+      "step": 2160
+    },
+    {
+      "epoch": 0.01570790534720262,
+      "grad_norm": 0.22456589341163635,
+      "learning_rate": 4.984299333318857e-06,
+      "loss": 1.1922,
+      "step": 2170
+    },
+    {
+      "epoch": 0.015780292007788806,
+      "grad_norm": 0.19744843244552612,
+      "learning_rate": 4.984226946658271e-06,
+      "loss": 1.1794,
+      "step": 2180
+    },
+    {
+      "epoch": 0.015852678668374993,
+      "grad_norm": 0.19209572672843933,
+      "learning_rate": 4.984154559997684e-06,
+      "loss": 1.1828,
+      "step": 2190
+    },
+    {
+      "epoch": 0.01592506532896118,
+      "grad_norm": 0.22542759776115417,
+      "learning_rate": 4.984082173337098e-06,
+      "loss": 1.1772,
+      "step": 2200
+    },
+    {
+      "epoch": 0.015997451989547367,
+      "grad_norm": 0.21402914822101593,
+      "learning_rate": 4.984009786676512e-06,
+      "loss": 1.1847,
+      "step": 2210
+    },
+    {
+      "epoch": 0.016069838650133554,
+      "grad_norm": 0.3712976574897766,
+      "learning_rate": 4.983937400015926e-06,
+      "loss": 1.1727,
+      "step": 2220
+    },
+    {
+      "epoch": 0.01614222531071974,
+      "grad_norm": 0.21875779330730438,
+      "learning_rate": 4.983865013355339e-06,
+      "loss": 1.1844,
+      "step": 2230
+    },
+    {
+      "epoch": 0.01621461197130593,
+      "grad_norm": 0.1886206865310669,
+      "learning_rate": 4.983792626694753e-06,
+      "loss": 1.1875,
+      "step": 2240
+    },
+    {
+      "epoch": 0.016286998631892115,
+      "grad_norm": 0.2019553780555725,
+      "learning_rate": 4.983720240034167e-06,
+      "loss": 1.1729,
+      "step": 2250
+    },
+    {
+      "epoch": 0.016359385292478303,
+      "grad_norm": 0.21517504751682281,
+      "learning_rate": 4.98364785337358e-06,
+      "loss": 1.1713,
+      "step": 2260
+    },
+    {
+      "epoch": 0.01643177195306449,
+      "grad_norm": 0.18494921922683716,
+      "learning_rate": 4.983575466712994e-06,
+      "loss": 1.1786,
+      "step": 2270
+    },
+    {
+      "epoch": 0.016504158613650677,
+      "grad_norm": 0.3039611876010895,
+      "learning_rate": 4.9835030800524084e-06,
+      "loss": 1.1819,
+      "step": 2280
+    },
+    {
+      "epoch": 0.016576545274236864,
+      "grad_norm": 0.23030070960521698,
+      "learning_rate": 4.983430693391822e-06,
+      "loss": 1.1915,
+      "step": 2290
+    },
+    {
+      "epoch": 0.01664893193482305,
+      "grad_norm": 0.1895866096019745,
+      "learning_rate": 4.983358306731236e-06,
+      "loss": 1.1782,
+      "step": 2300
+    },
+    {
+      "epoch": 0.016721318595409238,
+      "grad_norm": 0.19012758135795593,
+      "learning_rate": 4.983285920070649e-06,
+      "loss": 1.1894,
+      "step": 2310
+    },
+    {
+      "epoch": 0.016793705255995425,
+      "grad_norm": 0.18805542588233948,
+      "learning_rate": 4.983213533410064e-06,
+      "loss": 1.1802,
+      "step": 2320
+    },
+    {
+      "epoch": 0.016866091916581612,
+      "grad_norm": 0.19105587899684906,
+      "learning_rate": 4.983141146749477e-06,
+      "loss": 1.1797,
+      "step": 2330
+    },
+    {
+      "epoch": 0.0169384785771678,
+      "grad_norm": 0.22354061901569366,
+      "learning_rate": 4.983068760088891e-06,
+      "loss": 1.1898,
+      "step": 2340
+    },
+    {
+      "epoch": 0.017010865237753987,
+      "grad_norm": 0.1948034018278122,
+      "learning_rate": 4.982996373428305e-06,
+      "loss": 1.166,
+      "step": 2350
+    },
+    {
+      "epoch": 0.017083251898340174,
+      "grad_norm": 0.18683524429798126,
+      "learning_rate": 4.982923986767719e-06,
+      "loss": 1.1826,
+      "step": 2360
+    },
+    {
+      "epoch": 0.01715563855892636,
+      "grad_norm": 0.19329576194286346,
+      "learning_rate": 4.982851600107133e-06,
+      "loss": 1.1853,
+      "step": 2370
+    },
+    {
+      "epoch": 0.017228025219512548,
+      "grad_norm": 0.2141970992088318,
+      "learning_rate": 4.982779213446546e-06,
+      "loss": 1.172,
+      "step": 2380
+    },
+    {
+      "epoch": 0.017300411880098735,
+      "grad_norm": 0.20139099657535553,
+      "learning_rate": 4.98270682678596e-06,
+      "loss": 1.178,
+      "step": 2390
+    },
+    {
+      "epoch": 0.017372798540684922,
+      "grad_norm": 0.22393347322940826,
+      "learning_rate": 4.982634440125374e-06,
+      "loss": 1.1724,
+      "step": 2400
+    },
+    {
+      "epoch": 0.01744518520127111,
+      "grad_norm": 0.18857796490192413,
+      "learning_rate": 4.982562053464788e-06,
+      "loss": 1.1663,
+      "step": 2410
+    },
+    {
+      "epoch": 0.017517571861857296,
+      "grad_norm": 0.18142379820346832,
+      "learning_rate": 4.982489666804202e-06,
+      "loss": 1.1972,
+      "step": 2420
+    },
+    {
+      "epoch": 0.017589958522443484,
+      "grad_norm": 0.20119883120059967,
+      "learning_rate": 4.982417280143615e-06,
+      "loss": 1.1719,
+      "step": 2430
+    },
+    {
+      "epoch": 0.01766234518302967,
+      "grad_norm": 0.20527637004852295,
+      "learning_rate": 4.98234489348303e-06,
+      "loss": 1.1612,
+      "step": 2440
+    },
+    {
+      "epoch": 0.017734731843615858,
+      "grad_norm": 0.19073791801929474,
+      "learning_rate": 4.982272506822443e-06,
+      "loss": 1.1752,
+      "step": 2450
+    },
+    {
+      "epoch": 0.017807118504202045,
+      "grad_norm": 0.19779928028583527,
+      "learning_rate": 4.982200120161857e-06,
+      "loss": 1.1877,
+      "step": 2460
+    },
+    {
+      "epoch": 0.017879505164788232,
+      "grad_norm": 0.2000398486852646,
+      "learning_rate": 4.9821277335012705e-06,
+      "loss": 1.169,
+      "step": 2470
+    },
+    {
+      "epoch": 0.01795189182537442,
+      "grad_norm": 0.19572487473487854,
+      "learning_rate": 4.982055346840685e-06,
+      "loss": 1.1985,
+      "step": 2480
+    },
+    {
+      "epoch": 0.018024278485960606,
+      "grad_norm": 0.2484935224056244,
+      "learning_rate": 4.981982960180099e-06,
+      "loss": 1.1692,
+      "step": 2490
+    },
+    {
+      "epoch": 0.018096665146546793,
+      "grad_norm": 0.18102866411209106,
+      "learning_rate": 4.981910573519512e-06,
+      "loss": 1.1632,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01816905180713298,
+      "grad_norm": 0.18723329901695251,
+      "learning_rate": 4.981838186858926e-06,
+      "loss": 1.1763,
+      "step": 2510
+    },
+    {
+      "epoch": 0.018241438467719168,
+      "grad_norm": 0.201924130320549,
+      "learning_rate": 4.98176580019834e-06,
+      "loss": 1.1712,
+      "step": 2520
+    },
+    {
+      "epoch": 0.018313825128305355,
+      "grad_norm": 0.2887810468673706,
+      "learning_rate": 4.981693413537754e-06,
+      "loss": 1.1868,
+      "step": 2530
+    },
+    {
+      "epoch": 0.018386211788891542,
+      "grad_norm": 0.17751596868038177,
+      "learning_rate": 4.9816210268771675e-06,
+      "loss": 1.1815,
+      "step": 2540
+    },
+    {
+      "epoch": 0.01845859844947773,
+      "grad_norm": 0.20777058601379395,
+      "learning_rate": 4.981548640216581e-06,
+      "loss": 1.1607,
+      "step": 2550
+    },
+    {
+      "epoch": 0.018530985110063916,
+      "grad_norm": 0.2111022174358368,
+      "learning_rate": 4.981476253555995e-06,
+      "loss": 1.1709,
+      "step": 2560
+    },
+    {
+      "epoch": 0.018603371770650103,
+      "grad_norm": 0.1838046908378601,
+      "learning_rate": 4.981403866895409e-06,
+      "loss": 1.1658,
+      "step": 2570
+    },
+    {
+      "epoch": 0.01867575843123629,
+      "grad_norm": 0.20409898459911346,
+      "learning_rate": 4.981331480234823e-06,
+      "loss": 1.161,
+      "step": 2580
+    },
+    {
+      "epoch": 0.018748145091822477,
+      "grad_norm": 0.25109410285949707,
+      "learning_rate": 4.9812590935742365e-06,
+      "loss": 1.1483,
+      "step": 2590
+    },
+    {
+      "epoch": 0.018820531752408665,
+      "grad_norm": 0.25918787717819214,
+      "learning_rate": 4.98118670691365e-06,
+      "loss": 1.1781,
+      "step": 2600
+    },
+    {
+      "epoch": 0.01889291841299485,
+      "grad_norm": 0.18770872056484222,
+      "learning_rate": 4.9811143202530645e-06,
+      "loss": 1.1732,
+      "step": 2610
+    },
+    {
+      "epoch": 0.01896530507358104,
+      "grad_norm": 0.195766419172287,
+      "learning_rate": 4.981041933592478e-06,
+      "loss": 1.1673,
+      "step": 2620
+    },
+    {
+      "epoch": 0.019037691734167226,
+      "grad_norm": 0.19628261029720306,
+      "learning_rate": 4.980969546931892e-06,
+      "loss": 1.1687,
+      "step": 2630
+    },
+    {
+      "epoch": 0.019110078394753417,
+      "grad_norm": 0.22135888040065765,
+      "learning_rate": 4.980897160271305e-06,
+      "loss": 1.1664,
+      "step": 2640
+    },
+    {
+      "epoch": 0.019182465055339604,
+      "grad_norm": 0.19807446002960205,
+      "learning_rate": 4.98082477361072e-06,
+      "loss": 1.1762,
+      "step": 2650
+    },
+    {
+      "epoch": 0.01925485171592579,
+      "grad_norm": 0.3103126883506775,
+      "learning_rate": 4.9807523869501335e-06,
+      "loss": 1.1723,
+      "step": 2660
+    },
+    {
+      "epoch": 0.019327238376511978,
+      "grad_norm": 0.18378371000289917,
+      "learning_rate": 4.980680000289547e-06,
+      "loss": 1.185,
+      "step": 2670
+    },
+    {
+      "epoch": 0.019399625037098165,
+      "grad_norm": 0.1914234310388565,
+      "learning_rate": 4.980607613628961e-06,
+      "loss": 1.17,
+      "step": 2680
+    },
+    {
+      "epoch": 0.019472011697684352,
+      "grad_norm": 0.18758605420589447,
+      "learning_rate": 4.980535226968375e-06,
+      "loss": 1.1857,
+      "step": 2690
+    },
+    {
+      "epoch": 0.01954439835827054,
+      "grad_norm": 0.19784751534461975,
+      "learning_rate": 4.980462840307789e-06,
+      "loss": 1.1627,
+      "step": 2700
+    },
+    {
+      "epoch": 0.019616785018856726,
+      "grad_norm": 0.2056863158941269,
+      "learning_rate": 4.980390453647202e-06,
+      "loss": 1.1744,
+      "step": 2710
+    },
+    {
+      "epoch": 0.019689171679442913,
+      "grad_norm": 0.23083150386810303,
+      "learning_rate": 4.980318066986616e-06,
+      "loss": 1.1682,
+      "step": 2720
+    },
+    {
+      "epoch": 0.0197615583400291,
+      "grad_norm": 0.2089979201555252,
+      "learning_rate": 4.9802456803260305e-06,
+      "loss": 1.171,
+      "step": 2730
+    },
+    {
+      "epoch": 0.019833945000615288,
+      "grad_norm": 0.36265090107917786,
+      "learning_rate": 4.980173293665444e-06,
+      "loss": 1.1738,
+      "step": 2740
+    },
+    {
+      "epoch": 0.019906331661201475,
+      "grad_norm": 0.21013633906841278,
+      "learning_rate": 4.980100907004858e-06,
+      "loss": 1.1591,
+      "step": 2750
+    },
+    {
+      "epoch": 0.019978718321787662,
+      "grad_norm": 0.190397247672081,
+      "learning_rate": 4.980028520344271e-06,
+      "loss": 1.1761,
+      "step": 2760
+    },
+    {
+      "epoch": 0.02005110498237385,
+      "grad_norm": 0.2045404314994812,
+      "learning_rate": 4.979956133683685e-06,
+      "loss": 1.1701,
+      "step": 2770
+    },
+    {
+      "epoch": 0.020123491642960036,
+      "grad_norm": 0.18745577335357666,
+      "learning_rate": 4.9798837470230986e-06,
+      "loss": 1.1736,
+      "step": 2780
+    },
+    {
+      "epoch": 0.020195878303546223,
+      "grad_norm": 0.2338806539773941,
+      "learning_rate": 4.979811360362512e-06,
+      "loss": 1.1565,
+      "step": 2790
+    },
+    {
+      "epoch": 0.02026826496413241,
+      "grad_norm": 0.21875229477882385,
+      "learning_rate": 4.979738973701927e-06,
+      "loss": 1.1576,
+      "step": 2800
+    },
+    {
+      "epoch": 0.020340651624718598,
+      "grad_norm": 0.18577270209789276,
+      "learning_rate": 4.97966658704134e-06,
+      "loss": 1.1519,
+      "step": 2810
+    },
+    {
+      "epoch": 0.020413038285304785,
+      "grad_norm": 0.20935316383838654,
+      "learning_rate": 4.979594200380754e-06,
+      "loss": 1.168,
+      "step": 2820
+    },
+    {
+      "epoch": 0.020485424945890972,
+      "grad_norm": 0.2035789042711258,
+      "learning_rate": 4.9795218137201675e-06,
+      "loss": 1.1657,
+      "step": 2830
+    },
+    {
+      "epoch": 0.02055781160647716,
+      "grad_norm": 0.1842500865459442,
+      "learning_rate": 4.979449427059582e-06,
+      "loss": 1.1898,
+      "step": 2840
+    },
+    {
+      "epoch": 0.020630198267063346,
+      "grad_norm": 0.18557128310203552,
+      "learning_rate": 4.9793770403989956e-06,
+      "loss": 1.1708,
+      "step": 2850
+    },
+    {
+      "epoch": 0.020702584927649533,
+      "grad_norm": 0.19317933917045593,
+      "learning_rate": 4.979304653738409e-06,
+      "loss": 1.1572,
+      "step": 2860
+    },
+    {
+      "epoch": 0.02077497158823572,
+      "grad_norm": 0.17931902408599854,
+      "learning_rate": 4.979232267077823e-06,
+      "loss": 1.1581,
+      "step": 2870
+    },
+    {
+      "epoch": 0.020847358248821907,
+      "grad_norm": 0.20684286952018738,
+      "learning_rate": 4.979159880417237e-06,
+      "loss": 1.1828,
+      "step": 2880
+    },
+    {
+      "epoch": 0.020919744909408095,
+      "grad_norm": 0.19323229789733887,
+      "learning_rate": 4.979087493756651e-06,
+      "loss": 1.1456,
+      "step": 2890
+    },
+    {
+      "epoch": 0.02099213156999428,
+      "grad_norm": 0.19767233729362488,
+      "learning_rate": 4.9790151070960645e-06,
+      "loss": 1.1742,
+      "step": 2900
+    },
+    {
+      "epoch": 0.02106451823058047,
+      "grad_norm": 0.2398158460855484,
+      "learning_rate": 4.978942720435478e-06,
+      "loss": 1.1721,
+      "step": 2910
+    },
+    {
+      "epoch": 0.021136904891166656,
+      "grad_norm": 0.20536300539970398,
+      "learning_rate": 4.978870333774893e-06,
+      "loss": 1.158,
+      "step": 2920
+    },
+    {
+      "epoch": 0.021209291551752843,
+      "grad_norm": 0.20128795504570007,
+      "learning_rate": 4.978797947114306e-06,
+      "loss": 1.1632,
+      "step": 2930
+    },
+    {
+      "epoch": 0.02128167821233903,
+      "grad_norm": 0.19307374954223633,
+      "learning_rate": 4.97872556045372e-06,
+      "loss": 1.1718,
+      "step": 2940
+    },
+    {
+      "epoch": 0.021354064872925217,
+      "grad_norm": 0.19423744082450867,
+      "learning_rate": 4.9786531737931334e-06,
+      "loss": 1.1568,
+      "step": 2950
+    },
+    {
+      "epoch": 0.021426451533511404,
+      "grad_norm": 0.2026156485080719,
+      "learning_rate": 4.978580787132548e-06,
+      "loss": 1.168,
+      "step": 2960
+    },
+    {
+      "epoch": 0.02149883819409759,
+      "grad_norm": 0.21222802996635437,
+      "learning_rate": 4.9785084004719615e-06,
+      "loss": 1.1588,
+      "step": 2970
+    },
+    {
+      "epoch": 0.02157122485468378,
+      "grad_norm": 0.1813318431377411,
+      "learning_rate": 4.978436013811375e-06,
+      "loss": 1.1752,
+      "step": 2980
+    },
+    {
+      "epoch": 0.021643611515269966,
+      "grad_norm": 0.1816568523645401,
+      "learning_rate": 4.978363627150789e-06,
+      "loss": 1.1586,
+      "step": 2990
+    },
+    {
+      "epoch": 0.021715998175856153,
+      "grad_norm": 0.18530064821243286,
+      "learning_rate": 4.978291240490203e-06,
+      "loss": 1.1684,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02178838483644234,
+      "grad_norm": 0.2098444104194641,
+      "learning_rate": 4.978218853829617e-06,
+      "loss": 1.1685,
+      "step": 3010
+    },
+    {
+      "epoch": 0.021860771497028527,
+      "grad_norm": 0.17709197103977203,
+      "learning_rate": 4.9781464671690304e-06,
+      "loss": 1.1557,
+      "step": 3020
+    },
+    {
+      "epoch": 0.021933158157614714,
+      "grad_norm": 0.21442793309688568,
+      "learning_rate": 4.978074080508444e-06,
+      "loss": 1.1649,
+      "step": 3030
+    },
+    {
+      "epoch": 0.0220055448182009,
+      "grad_norm": 0.1951083093881607,
+      "learning_rate": 4.9780016938478585e-06,
+      "loss": 1.1499,
+      "step": 3040
+    },
+    {
+      "epoch": 0.02207793147878709,
+      "grad_norm": 0.18640479445457458,
+      "learning_rate": 4.977929307187272e-06,
+      "loss": 1.1664,
+      "step": 3050
+    },
+    {
+      "epoch": 0.022150318139373276,
+      "grad_norm": 0.18357819318771362,
+      "learning_rate": 4.977856920526686e-06,
+      "loss": 1.1744,
+      "step": 3060
+    },
+    {
+      "epoch": 0.022222704799959463,
+      "grad_norm": 0.23373299837112427,
+      "learning_rate": 4.977784533866099e-06,
+      "loss": 1.1594,
+      "step": 3070
+    },
+    {
+      "epoch": 0.02229509146054565,
+      "grad_norm": 0.2267906218767166,
+      "learning_rate": 4.977712147205514e-06,
+      "loss": 1.161,
+      "step": 3080
+    },
+    {
+      "epoch": 0.022367478121131837,
+      "grad_norm": 0.19738420844078064,
+      "learning_rate": 4.9776397605449274e-06,
+      "loss": 1.1524,
+      "step": 3090
+    },
+    {
+      "epoch": 0.022439864781718024,
+      "grad_norm": 0.22471626102924347,
+      "learning_rate": 4.977567373884341e-06,
+      "loss": 1.1426,
+      "step": 3100
+    },
+    {
+      "epoch": 0.02251225144230421,
+      "grad_norm": 0.18693824112415314,
+      "learning_rate": 4.977494987223755e-06,
+      "loss": 1.1516,
+      "step": 3110
+    },
+    {
+      "epoch": 0.022584638102890398,
+      "grad_norm": 0.29740896821022034,
+      "learning_rate": 4.977422600563169e-06,
+      "loss": 1.1545,
+      "step": 3120
+    },
+    {
+      "epoch": 0.022657024763476585,
+      "grad_norm": 0.19464527070522308,
+      "learning_rate": 4.977350213902583e-06,
+      "loss": 1.1474,
+      "step": 3130
+    },
+    {
+      "epoch": 0.022729411424062772,
+      "grad_norm": 0.19471614062786102,
+      "learning_rate": 4.977277827241996e-06,
+      "loss": 1.1441,
+      "step": 3140
+    },
+    {
+      "epoch": 0.02280179808464896,
+      "grad_norm": 0.1917818784713745,
+      "learning_rate": 4.97720544058141e-06,
+      "loss": 1.1588,
+      "step": 3150
+    },
+    {
+      "epoch": 0.022874184745235147,
+      "grad_norm": 0.20581530034542084,
+      "learning_rate": 4.977133053920824e-06,
+      "loss": 1.1644,
+      "step": 3160
+    },
+    {
+      "epoch": 0.022946571405821334,
+      "grad_norm": 0.19986987113952637,
+      "learning_rate": 4.977060667260238e-06,
+      "loss": 1.153,
+      "step": 3170
+    },
+    {
+      "epoch": 0.02301895806640752,
+      "grad_norm": 0.195315420627594,
+      "learning_rate": 4.976988280599652e-06,
+      "loss": 1.1563,
+      "step": 3180
+    },
+    {
+      "epoch": 0.023091344726993708,
+      "grad_norm": 0.2056894749403,
+      "learning_rate": 4.976915893939065e-06,
+      "loss": 1.1605,
+      "step": 3190
+    },
+    {
+      "epoch": 0.023163731387579895,
+      "grad_norm": 0.19704991579055786,
+      "learning_rate": 4.976843507278479e-06,
+      "loss": 1.1635,
+      "step": 3200
+    },
+    {
+      "epoch": 0.023236118048166082,
+      "grad_norm": 0.2168043702840805,
+      "learning_rate": 4.976771120617893e-06,
+      "loss": 1.165,
+      "step": 3210
+    },
+    {
+      "epoch": 0.02330850470875227,
+      "grad_norm": 0.21489858627319336,
+      "learning_rate": 4.976698733957307e-06,
+      "loss": 1.1592,
+      "step": 3220
+    },
+    {
+      "epoch": 0.02338089136933846,
+      "grad_norm": 0.18227837979793549,
+      "learning_rate": 4.976626347296721e-06,
+      "loss": 1.1573,
+      "step": 3230
+    },
+    {
+      "epoch": 0.023453278029924647,
+      "grad_norm": 0.20387127995491028,
+      "learning_rate": 4.976553960636134e-06,
+      "loss": 1.1611,
+      "step": 3240
+    },
+    {
+      "epoch": 0.023525664690510834,
+      "grad_norm": 0.17616231739521027,
+      "learning_rate": 4.976481573975549e-06,
+      "loss": 1.1614,
+      "step": 3250
+    },
+    {
+      "epoch": 0.02359805135109702,
+      "grad_norm": 0.1885817050933838,
+      "learning_rate": 4.976409187314962e-06,
+      "loss": 1.152,
+      "step": 3260
+    },
+    {
+      "epoch": 0.02367043801168321,
+      "grad_norm": 0.22119595110416412,
+      "learning_rate": 4.976336800654376e-06,
+      "loss": 1.1541,
+      "step": 3270
+    },
+    {
+      "epoch": 0.023742824672269396,
+      "grad_norm": 0.24721552431583405,
+      "learning_rate": 4.9762644139937895e-06,
+      "loss": 1.1419,
+      "step": 3280
+    },
+    {
+      "epoch": 0.023815211332855583,
+      "grad_norm": 0.20840491354465485,
+      "learning_rate": 4.976192027333203e-06,
+      "loss": 1.157,
+      "step": 3290
+    },
+    {
+      "epoch": 0.02388759799344177,
+      "grad_norm": 0.1903543770313263,
+      "learning_rate": 4.976119640672617e-06,
+      "loss": 1.1492,
+      "step": 3300
+    },
+    {
+      "epoch": 0.023959984654027957,
+      "grad_norm": 0.1851651966571808,
+      "learning_rate": 4.976047254012031e-06,
+      "loss": 1.1473,
+      "step": 3310
+    },
+    {
+      "epoch": 0.024032371314614144,
+      "grad_norm": 0.18348610401153564,
+      "learning_rate": 4.975974867351445e-06,
+      "loss": 1.1615,
+      "step": 3320
+    },
+    {
+      "epoch": 0.02410475797520033,
+      "grad_norm": 0.19050352275371552,
+      "learning_rate": 4.9759024806908585e-06,
+      "loss": 1.1705,
+      "step": 3330
+    },
+    {
+      "epoch": 0.02417714463578652,
+      "grad_norm": 0.2005050927400589,
+      "learning_rate": 4.975830094030272e-06,
+      "loss": 1.1582,
+      "step": 3340
+    },
+    {
+      "epoch": 0.024249531296372705,
+      "grad_norm": 0.19381307065486908,
+      "learning_rate": 4.975757707369686e-06,
+      "loss": 1.165,
+      "step": 3350
+    },
+    {
+      "epoch": 0.024321917956958893,
+      "grad_norm": 0.1775844395160675,
+      "learning_rate": 4.9756853207091e-06,
+      "loss": 1.1466,
+      "step": 3360
+    },
+    {
+      "epoch": 0.02439430461754508,
+      "grad_norm": 0.1731746643781662,
+      "learning_rate": 4.975612934048514e-06,
+      "loss": 1.1399,
+      "step": 3370
+    },
+    {
+      "epoch": 0.024466691278131267,
+      "grad_norm": 0.2276047319173813,
+      "learning_rate": 4.975540547387927e-06,
+      "loss": 1.1665,
+      "step": 3380
+    },
+    {
+      "epoch": 0.024539077938717454,
+      "grad_norm": 0.19553668797016144,
+      "learning_rate": 4.975468160727341e-06,
+      "loss": 1.1632,
+      "step": 3390
+    },
+    {
+      "epoch": 0.02461146459930364,
+      "grad_norm": 0.19757987558841705,
+      "learning_rate": 4.9753957740667555e-06,
+      "loss": 1.1549,
+      "step": 3400
+    },
+    {
+      "epoch": 0.024683851259889828,
+      "grad_norm": 0.19470013678073883,
+      "learning_rate": 4.975323387406169e-06,
+      "loss": 1.1677,
+      "step": 3410
+    },
+    {
+      "epoch": 0.024756237920476015,
+      "grad_norm": 0.20019254088401794,
+      "learning_rate": 4.975251000745583e-06,
+      "loss": 1.1456,
+      "step": 3420
+    },
+    {
+      "epoch": 0.024828624581062202,
+      "grad_norm": 0.20071542263031006,
+      "learning_rate": 4.975178614084996e-06,
+      "loss": 1.149,
+      "step": 3430
+    },
+    {
+      "epoch": 0.02490101124164839,
+      "grad_norm": 0.18767118453979492,
+      "learning_rate": 4.975106227424411e-06,
+      "loss": 1.1633,
+      "step": 3440
+    },
+    {
+      "epoch": 0.024973397902234577,
+      "grad_norm": 0.19544526934623718,
+      "learning_rate": 4.975033840763824e-06,
+      "loss": 1.145,
+      "step": 3450
+    },
+    {
+      "epoch": 0.025045784562820764,
+      "grad_norm": 0.18626125156879425,
+      "learning_rate": 4.974961454103238e-06,
+      "loss": 1.1457,
+      "step": 3460
+    },
+    {
+      "epoch": 0.02511817122340695,
+      "grad_norm": 0.19342540204524994,
+      "learning_rate": 4.974889067442652e-06,
+      "loss": 1.1648,
+      "step": 3470
+    },
+    {
+      "epoch": 0.025190557883993138,
+      "grad_norm": 0.20179611444473267,
+      "learning_rate": 4.974816680782066e-06,
+      "loss": 1.1574,
+      "step": 3480
+    },
+    {
+      "epoch": 0.025262944544579325,
+      "grad_norm": 0.20184293389320374,
+      "learning_rate": 4.97474429412148e-06,
+      "loss": 1.1542,
+      "step": 3490
+    },
+    {
+      "epoch": 0.025335331205165512,
+      "grad_norm": 0.18145424127578735,
+      "learning_rate": 4.974671907460893e-06,
+      "loss": 1.1478,
+      "step": 3500
+    },
+    {
+      "epoch": 0.0254077178657517,
+      "grad_norm": 0.2179315686225891,
+      "learning_rate": 4.974599520800307e-06,
+      "loss": 1.1552,
+      "step": 3510
+    },
+    {
+      "epoch": 0.025480104526337886,
+      "grad_norm": 0.20705650746822357,
+      "learning_rate": 4.974527134139721e-06,
+      "loss": 1.1547,
+      "step": 3520
+    },
+    {
+      "epoch": 0.025552491186924074,
+      "grad_norm": 0.19656804203987122,
+      "learning_rate": 4.974454747479135e-06,
+      "loss": 1.1541,
+      "step": 3530
+    },
+    {
+      "epoch": 0.02562487784751026,
+      "grad_norm": 0.18042577803134918,
+      "learning_rate": 4.974382360818549e-06,
+      "loss": 1.1442,
+      "step": 3540
+    },
+    {
+      "epoch": 0.025697264508096448,
+      "grad_norm": 0.20010463893413544,
+      "learning_rate": 4.974309974157962e-06,
+      "loss": 1.1454,
+      "step": 3550
+    },
+    {
+      "epoch": 0.025769651168682635,
+      "grad_norm": 0.17687994241714478,
+      "learning_rate": 4.974237587497377e-06,
+      "loss": 1.1548,
+      "step": 3560
+    },
+    {
+      "epoch": 0.025842037829268822,
+      "grad_norm": 0.17953871190547943,
+      "learning_rate": 4.97416520083679e-06,
+      "loss": 1.1388,
+      "step": 3570
+    },
+    {
+      "epoch": 0.02591442448985501,
+      "grad_norm": 0.1789717972278595,
+      "learning_rate": 4.974092814176204e-06,
+      "loss": 1.1522,
+      "step": 3580
+    },
+    {
+      "epoch": 0.025986811150441196,
+      "grad_norm": 0.1992194801568985,
+      "learning_rate": 4.9740204275156176e-06,
+      "loss": 1.1413,
+      "step": 3590
+    },
+    {
+      "epoch": 0.026059197811027383,
+      "grad_norm": 0.18657977879047394,
+      "learning_rate": 4.973948040855032e-06,
+      "loss": 1.1566,
+      "step": 3600
+    },
+    {
+      "epoch": 0.02613158447161357,
+      "grad_norm": 0.1899375021457672,
+      "learning_rate": 4.973875654194446e-06,
+      "loss": 1.1589,
+      "step": 3610
+    },
+    {
+      "epoch": 0.026203971132199758,
+      "grad_norm": 0.18726755678653717,
+      "learning_rate": 4.973803267533859e-06,
+      "loss": 1.1479,
+      "step": 3620
+    },
+    {
+      "epoch": 0.026276357792785945,
+      "grad_norm": 0.19596411287784576,
+      "learning_rate": 4.973730880873273e-06,
+      "loss": 1.1555,
+      "step": 3630
+    },
+    {
+      "epoch": 0.026348744453372132,
+      "grad_norm": 0.19859635829925537,
+      "learning_rate": 4.973658494212687e-06,
+      "loss": 1.1535,
+      "step": 3640
+    },
+    {
+      "epoch": 0.02642113111395832,
+      "grad_norm": 0.2296416461467743,
+      "learning_rate": 4.973586107552101e-06,
+      "loss": 1.129,
+      "step": 3650
+    },
+    {
+      "epoch": 0.026493517774544506,
+      "grad_norm": 0.1784886121749878,
+      "learning_rate": 4.973513720891515e-06,
+      "loss": 1.148,
+      "step": 3660
+    },
+    {
+      "epoch": 0.026565904435130693,
+      "grad_norm": 0.19031654298305511,
+      "learning_rate": 4.973441334230928e-06,
+      "loss": 1.1486,
+      "step": 3670
+    },
+    {
+      "epoch": 0.02663829109571688,
+      "grad_norm": 0.192035511136055,
+      "learning_rate": 4.973368947570343e-06,
+      "loss": 1.1436,
+      "step": 3680
+    },
+    {
+      "epoch": 0.026710677756303067,
+      "grad_norm": 0.19924819469451904,
+      "learning_rate": 4.973296560909756e-06,
+      "loss": 1.1522,
+      "step": 3690
+    },
+    {
+      "epoch": 0.026783064416889255,
+      "grad_norm": 0.17904998362064362,
+      "learning_rate": 4.97322417424917e-06,
+      "loss": 1.1435,
+      "step": 3700
+    },
+    {
+      "epoch": 0.02685545107747544,
+      "grad_norm": 0.17476427555084229,
+      "learning_rate": 4.9731517875885835e-06,
+      "loss": 1.1375,
+      "step": 3710
+    },
+    {
+      "epoch": 0.02692783773806163,
+      "grad_norm": 0.18144498765468597,
+      "learning_rate": 4.973079400927998e-06,
+      "loss": 1.1413,
+      "step": 3720
+    },
+    {
+      "epoch": 0.027000224398647816,
+      "grad_norm": 0.29269012808799744,
+      "learning_rate": 4.973007014267412e-06,
+      "loss": 1.1434,
+      "step": 3730
+    },
+    {
+      "epoch": 0.027072611059234003,
+      "grad_norm": 0.3143029510974884,
+      "learning_rate": 4.972934627606825e-06,
+      "loss": 1.127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.02714499771982019,
+      "grad_norm": 0.2110947221517563,
+      "learning_rate": 4.972862240946239e-06,
+      "loss": 1.1477,
+      "step": 3750
+    },
+    {
+      "epoch": 0.027217384380406377,
+      "grad_norm": 0.18889427185058594,
+      "learning_rate": 4.972789854285653e-06,
+      "loss": 1.1458,
+      "step": 3760
+    },
+    {
+      "epoch": 0.027289771040992564,
+      "grad_norm": 0.19204457104206085,
+      "learning_rate": 4.972717467625067e-06,
+      "loss": 1.1528,
+      "step": 3770
+    },
+    {
+      "epoch": 0.02736215770157875,
+      "grad_norm": 0.19252029061317444,
+      "learning_rate": 4.9726450809644805e-06,
+      "loss": 1.1411,
+      "step": 3780
+    },
+    {
+      "epoch": 0.02743454436216494,
+      "grad_norm": 0.1872684508562088,
+      "learning_rate": 4.972572694303894e-06,
+      "loss": 1.1486,
+      "step": 3790
+    },
+    {
+      "epoch": 0.027506931022751126,
+      "grad_norm": 0.189093217253685,
+      "learning_rate": 4.972500307643308e-06,
+      "loss": 1.151,
+      "step": 3800
+    },
+    {
+      "epoch": 0.027579317683337316,
+      "grad_norm": 0.19076119363307953,
+      "learning_rate": 4.972427920982722e-06,
+      "loss": 1.1442,
+      "step": 3810
+    },
+    {
+      "epoch": 0.027651704343923503,
+      "grad_norm": 0.2054983228445053,
+      "learning_rate": 4.972355534322135e-06,
+      "loss": 1.149,
+      "step": 3820
+    },
+    {
+      "epoch": 0.02772409100450969,
+      "grad_norm": 0.19762204587459564,
+      "learning_rate": 4.9722831476615494e-06,
+      "loss": 1.1373,
+      "step": 3830
+    },
+    {
+      "epoch": 0.027796477665095878,
+      "grad_norm": 0.20214678347110748,
+      "learning_rate": 4.972210761000963e-06,
+      "loss": 1.1302,
+      "step": 3840
+    },
+    {
+      "epoch": 0.027868864325682065,
+      "grad_norm": 0.19313126802444458,
+      "learning_rate": 4.972138374340377e-06,
+      "loss": 1.152,
+      "step": 3850
+    },
+    {
+      "epoch": 0.027941250986268252,
+      "grad_norm": 0.19163060188293457,
+      "learning_rate": 4.97206598767979e-06,
+      "loss": 1.1401,
+      "step": 3860
+    },
+    {
+      "epoch": 0.02801363764685444,
+      "grad_norm": 0.1946219801902771,
+      "learning_rate": 4.971993601019205e-06,
+      "loss": 1.1586,
+      "step": 3870
+    },
+    {
+      "epoch": 0.028086024307440626,
+      "grad_norm": 0.18935218453407288,
+      "learning_rate": 4.971921214358618e-06,
+      "loss": 1.1657,
+      "step": 3880
+    },
+    {
+      "epoch": 0.028158410968026813,
+      "grad_norm": 0.864253044128418,
+      "learning_rate": 4.971848827698032e-06,
+      "loss": 1.1543,
+      "step": 3890
+    },
+    {
+      "epoch": 0.028230797628613,
+      "grad_norm": 0.1859940141439438,
+      "learning_rate": 4.971776441037446e-06,
+      "loss": 1.1487,
+      "step": 3900
+    },
+    {
+      "epoch": 0.028303184289199187,
+      "grad_norm": 0.18587446212768555,
+      "learning_rate": 4.97170405437686e-06,
+      "loss": 1.1423,
+      "step": 3910
+    },
+    {
+      "epoch": 0.028375570949785375,
+      "grad_norm": 0.22516964375972748,
+      "learning_rate": 4.971631667716274e-06,
+      "loss": 1.1224,
+      "step": 3920
+    },
+    {
+      "epoch": 0.02844795761037156,
+      "grad_norm": 0.19674985110759735,
+      "learning_rate": 4.971559281055687e-06,
+      "loss": 1.1492,
+      "step": 3930
+    },
+    {
+      "epoch": 0.02852034427095775,
+      "grad_norm": 0.1964777559041977,
+      "learning_rate": 4.971486894395101e-06,
+      "loss": 1.1433,
+      "step": 3940
+    },
+    {
+      "epoch": 0.028592730931543936,
+      "grad_norm": 0.1828773319721222,
+      "learning_rate": 4.9714145077345145e-06,
+      "loss": 1.1462,
+      "step": 3950
+    },
+    {
+      "epoch": 0.028665117592130123,
+      "grad_norm": 0.19454605877399445,
+      "learning_rate": 4.971342121073929e-06,
+      "loss": 1.137,
+      "step": 3960
+    },
+    {
+      "epoch": 0.02873750425271631,
+      "grad_norm": 0.20798833668231964,
+      "learning_rate": 4.971269734413343e-06,
+      "loss": 1.1352,
+      "step": 3970
+    },
+    {
+      "epoch": 0.028809890913302497,
+      "grad_norm": 0.1934082806110382,
+      "learning_rate": 4.971197347752756e-06,
+      "loss": 1.1256,
+      "step": 3980
+    },
+    {
+      "epoch": 0.028882277573888684,
+      "grad_norm": 0.19269952178001404,
+      "learning_rate": 4.97112496109217e-06,
+      "loss": 1.147,
+      "step": 3990
+    },
+    {
+      "epoch": 0.02895466423447487,
+      "grad_norm": 0.1940053403377533,
+      "learning_rate": 4.971052574431584e-06,
+      "loss": 1.1449,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02902705089506106,
+      "grad_norm": 0.18359413743019104,
+      "learning_rate": 4.970980187770998e-06,
+      "loss": 1.1507,
+      "step": 4010
+    },
+    {
+      "epoch": 0.029099437555647246,
+      "grad_norm": 0.19637279212474823,
+      "learning_rate": 4.9709078011104115e-06,
+      "loss": 1.14,
+      "step": 4020
+    },
+    {
+      "epoch": 0.029171824216233433,
+      "grad_norm": 0.19355088472366333,
+      "learning_rate": 4.970835414449825e-06,
+      "loss": 1.1521,
+      "step": 4030
+    },
+    {
+      "epoch": 0.02924421087681962,
+      "grad_norm": 0.18721257150173187,
+      "learning_rate": 4.97076302778924e-06,
+      "loss": 1.1391,
+      "step": 4040
+    },
+    {
+      "epoch": 0.029316597537405807,
+      "grad_norm": 0.2009786069393158,
+      "learning_rate": 4.970690641128653e-06,
+      "loss": 1.1205,
+      "step": 4050
+    },
+    {
+      "epoch": 0.029388984197991994,
+      "grad_norm": 0.19062365591526031,
+      "learning_rate": 4.970618254468067e-06,
+      "loss": 1.1506,
+      "step": 4060
+    },
+    {
+      "epoch": 0.02946137085857818,
+      "grad_norm": 0.2137022167444229,
+      "learning_rate": 4.9705458678074805e-06,
+      "loss": 1.1407,
+      "step": 4070
+    },
+    {
+      "epoch": 0.02953375751916437,
+      "grad_norm": 0.18308189511299133,
+      "learning_rate": 4.970473481146895e-06,
+      "loss": 1.1483,
+      "step": 4080
+    },
+    {
+      "epoch": 0.029606144179750556,
+      "grad_norm": 0.1955224871635437,
+      "learning_rate": 4.9704010944863085e-06,
+      "loss": 1.1315,
+      "step": 4090
+    },
+    {
+      "epoch": 0.029678530840336743,
+      "grad_norm": 0.1963161826133728,
+      "learning_rate": 4.970328707825722e-06,
+      "loss": 1.1519,
+      "step": 4100
+    },
+    {
+      "epoch": 0.02975091750092293,
+      "grad_norm": 0.20369768142700195,
+      "learning_rate": 4.970256321165136e-06,
+      "loss": 1.1286,
+      "step": 4110
+    },
+    {
+      "epoch": 0.029823304161509117,
+      "grad_norm": 0.18953342735767365,
+      "learning_rate": 4.97018393450455e-06,
+      "loss": 1.1546,
+      "step": 4120
+    },
+    {
+      "epoch": 0.029895690822095304,
+      "grad_norm": 0.22297842800617218,
+      "learning_rate": 4.970111547843964e-06,
+      "loss": 1.1373,
+      "step": 4130
+    },
+    {
+      "epoch": 0.02996807748268149,
+      "grad_norm": 0.23974451422691345,
+      "learning_rate": 4.9700391611833775e-06,
+      "loss": 1.1384,
+      "step": 4140
+    },
+    {
+      "epoch": 0.03004046414326768,
+      "grad_norm": 0.2132536619901657,
+      "learning_rate": 4.969966774522791e-06,
+      "loss": 1.145,
+      "step": 4150
+    },
+    {
+      "epoch": 0.030112850803853865,
+      "grad_norm": 0.18748712539672852,
+      "learning_rate": 4.9698943878622056e-06,
+      "loss": 1.129,
+      "step": 4160
+    },
+    {
+      "epoch": 0.030185237464440053,
+      "grad_norm": 0.18784211575984955,
+      "learning_rate": 4.969822001201619e-06,
+      "loss": 1.1376,
+      "step": 4170
+    },
+    {
+      "epoch": 0.03025762412502624,
+      "grad_norm": 0.19498911499977112,
+      "learning_rate": 4.969749614541033e-06,
+      "loss": 1.1282,
+      "step": 4180
+    },
+    {
+      "epoch": 0.030330010785612427,
+      "grad_norm": 0.19395703077316284,
+      "learning_rate": 4.969677227880446e-06,
+      "loss": 1.1319,
+      "step": 4190
+    },
+    {
+      "epoch": 0.030402397446198614,
+      "grad_norm": 0.1815170794725418,
+      "learning_rate": 4.969604841219861e-06,
+      "loss": 1.135,
+      "step": 4200
+    },
+    {
+      "epoch": 0.0304747841067848,
+      "grad_norm": 0.18967507779598236,
+      "learning_rate": 4.9695324545592745e-06,
+      "loss": 1.1281,
+      "step": 4210
+    },
+    {
+      "epoch": 0.030547170767370988,
+      "grad_norm": 0.19109271466732025,
+      "learning_rate": 4.969460067898688e-06,
+      "loss": 1.1424,
+      "step": 4220
+    },
+    {
+      "epoch": 0.030619557427957175,
+      "grad_norm": 0.23208124935626984,
+      "learning_rate": 4.969387681238102e-06,
+      "loss": 1.1408,
+      "step": 4230
+    },
+    {
+      "epoch": 0.030691944088543362,
+      "grad_norm": 0.18746989965438843,
+      "learning_rate": 4.969315294577516e-06,
+      "loss": 1.143,
+      "step": 4240
+    },
+    {
+      "epoch": 0.03076433074912955,
+      "grad_norm": 0.21662504971027374,
+      "learning_rate": 4.96924290791693e-06,
+      "loss": 1.1353,
+      "step": 4250
+    },
+    {
+      "epoch": 0.030836717409715737,
+      "grad_norm": 0.2007625848054886,
+      "learning_rate": 4.969170521256343e-06,
+      "loss": 1.1345,
+      "step": 4260
+    },
+    {
+      "epoch": 0.030909104070301924,
+      "grad_norm": 0.18364295363426208,
+      "learning_rate": 4.969098134595757e-06,
+      "loss": 1.1357,
+      "step": 4270
+    },
+    {
+      "epoch": 0.03098149073088811,
+      "grad_norm": 0.19245269894599915,
+      "learning_rate": 4.9690257479351715e-06,
+      "loss": 1.1326,
+      "step": 4280
+    },
+    {
+      "epoch": 0.031053877391474298,
+      "grad_norm": 0.1952354460954666,
+      "learning_rate": 4.968953361274585e-06,
+      "loss": 1.1253,
+      "step": 4290
+    },
+    {
+      "epoch": 0.031126264052060485,
+      "grad_norm": 0.18464967608451843,
+      "learning_rate": 4.968880974613999e-06,
+      "loss": 1.1324,
+      "step": 4300
+    },
+    {
+      "epoch": 0.031198650712646672,
+      "grad_norm": 0.19587458670139313,
+      "learning_rate": 4.968808587953412e-06,
+      "loss": 1.1399,
+      "step": 4310
+    },
+    {
+      "epoch": 0.03127103737323286,
+      "grad_norm": 0.1744762659072876,
+      "learning_rate": 4.968736201292827e-06,
+      "loss": 1.1389,
+      "step": 4320
+    },
+    {
+      "epoch": 0.031343424033819046,
+      "grad_norm": 0.17935672402381897,
+      "learning_rate": 4.96866381463224e-06,
+      "loss": 1.1392,
+      "step": 4330
+    },
+    {
+      "epoch": 0.03141581069440524,
+      "grad_norm": 0.19451332092285156,
+      "learning_rate": 4.968591427971654e-06,
+      "loss": 1.1409,
+      "step": 4340
+    },
+    {
+      "epoch": 0.03148819735499142,
+      "grad_norm": 0.20897798240184784,
+      "learning_rate": 4.968519041311068e-06,
+      "loss": 1.1205,
+      "step": 4350
+    },
+    {
+      "epoch": 0.03156058401557761,
+      "grad_norm": 0.20010434091091156,
+      "learning_rate": 4.968446654650481e-06,
+      "loss": 1.1408,
+      "step": 4360
+    },
+    {
+      "epoch": 0.031632970676163795,
+      "grad_norm": 0.19705431163311005,
+      "learning_rate": 4.968374267989895e-06,
+      "loss": 1.1392,
+      "step": 4370
+    },
+    {
+      "epoch": 0.031705357336749986,
+      "grad_norm": 0.19512054324150085,
+      "learning_rate": 4.9683018813293085e-06,
+      "loss": 1.1481,
+      "step": 4380
+    },
+    {
+      "epoch": 0.03177774399733617,
+      "grad_norm": 0.2135220468044281,
+      "learning_rate": 4.968229494668723e-06,
+      "loss": 1.1294,
+      "step": 4390
+    },
+    {
+      "epoch": 0.03185013065792236,
+      "grad_norm": 0.18849720060825348,
+      "learning_rate": 4.9681571080081366e-06,
+      "loss": 1.137,
+      "step": 4400
+    },
+    {
+      "epoch": 0.03192251731850854,
+      "grad_norm": 0.18328508734703064,
+      "learning_rate": 4.96808472134755e-06,
+      "loss": 1.1302,
+      "step": 4410
+    },
+    {
+      "epoch": 0.031994903979094734,
+      "grad_norm": 0.19263693690299988,
+      "learning_rate": 4.968012334686964e-06,
+      "loss": 1.1344,
+      "step": 4420
+    },
+    {
+      "epoch": 0.03206729063968092,
+      "grad_norm": 0.17624565958976746,
+      "learning_rate": 4.967939948026378e-06,
+      "loss": 1.1283,
+      "step": 4430
+    },
+    {
+      "epoch": 0.03213967730026711,
+      "grad_norm": 0.2214570790529251,
+      "learning_rate": 4.967867561365792e-06,
+      "loss": 1.1212,
+      "step": 4440
+    },
+    {
+      "epoch": 0.03221206396085329,
+      "grad_norm": 0.19538843631744385,
+      "learning_rate": 4.9677951747052055e-06,
+      "loss": 1.1519,
+      "step": 4450
+    },
+    {
+      "epoch": 0.03228445062143948,
+      "grad_norm": 0.19535070657730103,
+      "learning_rate": 4.967722788044619e-06,
+      "loss": 1.1206,
+      "step": 4460
+    },
+    {
+      "epoch": 0.032356837282025666,
+      "grad_norm": 0.20394045114517212,
+      "learning_rate": 4.967650401384034e-06,
+      "loss": 1.1546,
+      "step": 4470
+    },
+    {
+      "epoch": 0.03242922394261186,
+      "grad_norm": 0.18652796745300293,
+      "learning_rate": 4.967578014723447e-06,
+      "loss": 1.1371,
+      "step": 4480
+    },
+    {
+      "epoch": 0.03250161060319804,
+      "grad_norm": 0.18646922707557678,
+      "learning_rate": 4.967505628062861e-06,
+      "loss": 1.1479,
+      "step": 4490
+    },
+    {
+      "epoch": 0.03257399726378423,
+      "grad_norm": 0.1993158608675003,
+      "learning_rate": 4.9674332414022744e-06,
+      "loss": 1.1402,
+      "step": 4500
+    },
+    {
+      "epoch": 0.032646383924370415,
+      "grad_norm": 0.18076984584331512,
+      "learning_rate": 4.967360854741689e-06,
+      "loss": 1.1291,
+      "step": 4510
+    },
+    {
+      "epoch": 0.032718770584956605,
+      "grad_norm": 0.9336857199668884,
+      "learning_rate": 4.9672884680811025e-06,
+      "loss": 1.136,
+      "step": 4520
+    },
+    {
+      "epoch": 0.03279115724554279,
+      "grad_norm": 0.1966332644224167,
+      "learning_rate": 4.967216081420516e-06,
+      "loss": 1.1248,
+      "step": 4530
+    },
+    {
+      "epoch": 0.03286354390612898,
+      "grad_norm": 0.20412597060203552,
+      "learning_rate": 4.96714369475993e-06,
+      "loss": 1.1375,
+      "step": 4540
+    },
+    {
+      "epoch": 0.03293593056671516,
+      "grad_norm": 0.1918439120054245,
+      "learning_rate": 4.967071308099344e-06,
+      "loss": 1.1298,
+      "step": 4550
+    },
+    {
+      "epoch": 0.033008317227301354,
+      "grad_norm": 0.18617042899131775,
+      "learning_rate": 4.966998921438758e-06,
+      "loss": 1.1299,
+      "step": 4560
+    },
+    {
+      "epoch": 0.03308070388788754,
+      "grad_norm": 0.18576890230178833,
+      "learning_rate": 4.9669265347781714e-06,
+      "loss": 1.127,
+      "step": 4570
+    },
+    {
+      "epoch": 0.03315309054847373,
+      "grad_norm": 0.18330788612365723,
+      "learning_rate": 4.966854148117585e-06,
+      "loss": 1.1222,
+      "step": 4580
+    },
+    {
+      "epoch": 0.03322547720905991,
+      "grad_norm": 0.19359582662582397,
+      "learning_rate": 4.966781761456999e-06,
+      "loss": 1.1218,
+      "step": 4590
+    },
+    {
+      "epoch": 0.0332978638696461,
+      "grad_norm": 0.19401496648788452,
+      "learning_rate": 4.966709374796413e-06,
+      "loss": 1.1386,
+      "step": 4600
+    },
+    {
+      "epoch": 0.033370250530232286,
+      "grad_norm": 0.18727423250675201,
+      "learning_rate": 4.966636988135827e-06,
+      "loss": 1.1225,
+      "step": 4610
+    },
+    {
+      "epoch": 0.033442637190818476,
+      "grad_norm": 0.19162550568580627,
+      "learning_rate": 4.96656460147524e-06,
+      "loss": 1.1226,
+      "step": 4620
+    },
+    {
+      "epoch": 0.03351502385140466,
+      "grad_norm": 0.18989378213882446,
+      "learning_rate": 4.966492214814654e-06,
+      "loss": 1.1409,
+      "step": 4630
+    },
+    {
+      "epoch": 0.03358741051199085,
+      "grad_norm": 0.17028242349624634,
+      "learning_rate": 4.9664198281540684e-06,
+      "loss": 1.1353,
+      "step": 4640
+    },
+    {
+      "epoch": 0.033659797172577034,
+      "grad_norm": 0.18227070569992065,
+      "learning_rate": 4.966347441493482e-06,
+      "loss": 1.136,
+      "step": 4650
+    },
+    {
+      "epoch": 0.033732183833163225,
+      "grad_norm": 0.1902882605791092,
+      "learning_rate": 4.966275054832896e-06,
+      "loss": 1.1392,
+      "step": 4660
+    },
+    {
+      "epoch": 0.03380457049374941,
+      "grad_norm": 0.17739436030387878,
+      "learning_rate": 4.966202668172309e-06,
+      "loss": 1.1258,
+      "step": 4670
+    },
+    {
+      "epoch": 0.0338769571543356,
+      "grad_norm": 0.187397763133049,
+      "learning_rate": 4.966130281511724e-06,
+      "loss": 1.1261,
+      "step": 4680
+    },
+    {
+      "epoch": 0.03394934381492179,
+      "grad_norm": 0.17765313386917114,
+      "learning_rate": 4.966057894851137e-06,
+      "loss": 1.1252,
+      "step": 4690
+    },
+    {
+      "epoch": 0.03402173047550797,
+      "grad_norm": 0.18505384027957916,
+      "learning_rate": 4.965985508190551e-06,
+      "loss": 1.1257,
+      "step": 4700
+    },
+    {
+      "epoch": 0.034094117136094164,
+      "grad_norm": 0.19399258494377136,
+      "learning_rate": 4.965913121529965e-06,
+      "loss": 1.1236,
+      "step": 4710
+    },
+    {
+      "epoch": 0.03416650379668035,
+      "grad_norm": 0.17900791764259338,
+      "learning_rate": 4.965840734869379e-06,
+      "loss": 1.13,
+      "step": 4720
+    },
+    {
+      "epoch": 0.03423889045726654,
+      "grad_norm": 0.17912939190864563,
+      "learning_rate": 4.965768348208793e-06,
+      "loss": 1.1416,
+      "step": 4730
+    },
+    {
+      "epoch": 0.03431127711785272,
+      "grad_norm": 0.1887941062450409,
+      "learning_rate": 4.965695961548206e-06,
+      "loss": 1.1308,
+      "step": 4740
+    },
+    {
+      "epoch": 0.03438366377843891,
+      "grad_norm": 0.19085676968097687,
+      "learning_rate": 4.96562357488762e-06,
+      "loss": 1.1374,
+      "step": 4750
+    },
+    {
+      "epoch": 0.034456050439025096,
+      "grad_norm": 0.1932060867547989,
+      "learning_rate": 4.965551188227034e-06,
+      "loss": 1.1412,
+      "step": 4760
+    },
+    {
+      "epoch": 0.03452843709961129,
+      "grad_norm": 0.1827419102191925,
+      "learning_rate": 4.965478801566448e-06,
+      "loss": 1.1221,
+      "step": 4770
+    },
+    {
+      "epoch": 0.03460082376019747,
+      "grad_norm": 0.31538698077201843,
+      "learning_rate": 4.965406414905862e-06,
+      "loss": 1.1423,
+      "step": 4780
+    },
+    {
+      "epoch": 0.03467321042078366,
+      "grad_norm": 0.23767271637916565,
+      "learning_rate": 4.965334028245275e-06,
+      "loss": 1.1166,
+      "step": 4790
+    },
+    {
+      "epoch": 0.034745597081369844,
+      "grad_norm": 0.1903272569179535,
+      "learning_rate": 4.96526164158469e-06,
+      "loss": 1.1551,
+      "step": 4800
+    },
+    {
+      "epoch": 0.034817983741956035,
+      "grad_norm": 0.1915259212255478,
+      "learning_rate": 4.965189254924103e-06,
+      "loss": 1.1427,
+      "step": 4810
+    },
+    {
+      "epoch": 0.03489037040254222,
+      "grad_norm": 0.18546457588672638,
+      "learning_rate": 4.965116868263517e-06,
+      "loss": 1.1335,
+      "step": 4820
+    },
+    {
+      "epoch": 0.03496275706312841,
+      "grad_norm": 0.17920000851154327,
+      "learning_rate": 4.9650444816029305e-06,
+      "loss": 1.1362,
+      "step": 4830
+    },
+    {
+      "epoch": 0.03503514372371459,
+      "grad_norm": 0.1886454075574875,
+      "learning_rate": 4.964972094942345e-06,
+      "loss": 1.1328,
+      "step": 4840
+    },
+    {
+      "epoch": 0.035107530384300784,
+      "grad_norm": 0.18787150084972382,
+      "learning_rate": 4.964899708281759e-06,
+      "loss": 1.1359,
+      "step": 4850
+    },
+    {
+      "epoch": 0.03517991704488697,
+      "grad_norm": 0.18325144052505493,
+      "learning_rate": 4.964827321621172e-06,
+      "loss": 1.1281,
+      "step": 4860
+    },
+    {
+      "epoch": 0.03525230370547316,
+      "grad_norm": 0.17951847612857819,
+      "learning_rate": 4.964754934960586e-06,
+      "loss": 1.1304,
+      "step": 4870
+    },
+    {
+      "epoch": 0.03532469036605934,
+      "grad_norm": 0.17793092131614685,
+      "learning_rate": 4.9646825482999995e-06,
+      "loss": 1.1348,
+      "step": 4880
+    },
+    {
+      "epoch": 0.03539707702664553,
+      "grad_norm": 0.19709095358848572,
+      "learning_rate": 4.964610161639413e-06,
+      "loss": 1.1046,
+      "step": 4890
+    },
+    {
+      "epoch": 0.035469463687231716,
+      "grad_norm": 0.2029866725206375,
+      "learning_rate": 4.964537774978827e-06,
+      "loss": 1.1269,
+      "step": 4900
+    },
+    {
+      "epoch": 0.035541850347817906,
+      "grad_norm": 0.1873743087053299,
+      "learning_rate": 4.964465388318241e-06,
+      "loss": 1.1214,
+      "step": 4910
+    },
+    {
+      "epoch": 0.03561423700840409,
+      "grad_norm": 0.18011152744293213,
+      "learning_rate": 4.964393001657655e-06,
+      "loss": 1.1244,
+      "step": 4920
+    },
+    {
+      "epoch": 0.03568662366899028,
+      "grad_norm": 0.2078821361064911,
+      "learning_rate": 4.964320614997068e-06,
+      "loss": 1.123,
+      "step": 4930
+    },
+    {
+      "epoch": 0.035759010329576464,
+      "grad_norm": 0.22960899770259857,
+      "learning_rate": 4.964248228336482e-06,
+      "loss": 1.1354,
+      "step": 4940
+    },
+    {
+      "epoch": 0.035831396990162655,
+      "grad_norm": 0.19076983630657196,
+      "learning_rate": 4.9641758416758965e-06,
+      "loss": 1.1237,
+      "step": 4950
+    },
+    {
+      "epoch": 0.03590378365074884,
+      "grad_norm": 0.19131101667881012,
+      "learning_rate": 4.96410345501531e-06,
+      "loss": 1.1253,
+      "step": 4960
+    },
+    {
+      "epoch": 0.03597617031133503,
+      "grad_norm": 0.1835022121667862,
+      "learning_rate": 4.964031068354724e-06,
+      "loss": 1.1338,
+      "step": 4970
+    },
+    {
+      "epoch": 0.03604855697192121,
+      "grad_norm": 0.17867615818977356,
+      "learning_rate": 4.963958681694137e-06,
+      "loss": 1.112,
+      "step": 4980
+    },
+    {
+      "epoch": 0.0361209436325074,
+      "grad_norm": 0.18737174570560455,
+      "learning_rate": 4.963886295033552e-06,
+      "loss": 1.1277,
+      "step": 4990
+    },
+    {
+      "epoch": 0.03619333029309359,
+      "grad_norm": 0.21853433549404144,
+      "learning_rate": 4.963813908372965e-06,
+      "loss": 1.1324,
+      "step": 5000
+    },
+    {
+      "epoch": 0.03626571695367978,
+      "grad_norm": 0.18206347525119781,
+      "learning_rate": 4.963741521712379e-06,
+      "loss": 1.1275,
+      "step": 5010
+    },
+    {
+      "epoch": 0.03633810361426596,
+      "grad_norm": 0.20839586853981018,
+      "learning_rate": 4.963669135051793e-06,
+      "loss": 1.1235,
+      "step": 5020
+    },
+    {
+      "epoch": 0.03641049027485215,
+      "grad_norm": 0.208791121840477,
+      "learning_rate": 4.963596748391207e-06,
+      "loss": 1.1197,
+      "step": 5030
+    },
+    {
+      "epoch": 0.036482876935438335,
+      "grad_norm": 0.18400295078754425,
+      "learning_rate": 4.963524361730621e-06,
+      "loss": 1.1413,
+      "step": 5040
+    },
+    {
+      "epoch": 0.036555263596024526,
+      "grad_norm": 0.1914309412240982,
+      "learning_rate": 4.963451975070034e-06,
+      "loss": 1.1389,
+      "step": 5050
+    },
+    {
+      "epoch": 0.03662765025661071,
+      "grad_norm": 0.18529996275901794,
+      "learning_rate": 4.963379588409448e-06,
+      "loss": 1.1152,
+      "step": 5060
+    },
+    {
+      "epoch": 0.0367000369171969,
+      "grad_norm": 0.1869628131389618,
+      "learning_rate": 4.963307201748862e-06,
+      "loss": 1.1276,
+      "step": 5070
+    },
+    {
+      "epoch": 0.036772423577783084,
+      "grad_norm": 0.1835557073354721,
+      "learning_rate": 4.963234815088276e-06,
+      "loss": 1.1408,
+      "step": 5080
+    },
+    {
+      "epoch": 0.036844810238369274,
+      "grad_norm": 0.19675718247890472,
+      "learning_rate": 4.96316242842769e-06,
+      "loss": 1.1053,
+      "step": 5090
+    },
+    {
+      "epoch": 0.03691719689895546,
+      "grad_norm": 0.18781916797161102,
+      "learning_rate": 4.963090041767103e-06,
+      "loss": 1.1331,
+      "step": 5100
+    },
+    {
+      "epoch": 0.03698958355954165,
+      "grad_norm": 0.24024321138858795,
+      "learning_rate": 4.963017655106518e-06,
+      "loss": 1.1233,
+      "step": 5110
+    },
+    {
+      "epoch": 0.03706197022012783,
+      "grad_norm": 0.17619359493255615,
+      "learning_rate": 4.962945268445931e-06,
+      "loss": 1.123,
+      "step": 5120
+    },
+    {
+      "epoch": 0.03713435688071402,
+      "grad_norm": 0.19211329519748688,
+      "learning_rate": 4.962872881785345e-06,
+      "loss": 1.1162,
+      "step": 5130
+    },
+    {
+      "epoch": 0.037206743541300207,
+      "grad_norm": 0.18998998403549194,
+      "learning_rate": 4.9628004951247586e-06,
+      "loss": 1.1298,
+      "step": 5140
+    },
+    {
+      "epoch": 0.0372791302018864,
+      "grad_norm": 0.18230006098747253,
+      "learning_rate": 4.962728108464173e-06,
+      "loss": 1.1313,
+      "step": 5150
+    },
+    {
+      "epoch": 0.03735151686247258,
+      "grad_norm": 0.18055368959903717,
+      "learning_rate": 4.962655721803587e-06,
+      "loss": 1.1395,
+      "step": 5160
+    },
+    {
+      "epoch": 0.03742390352305877,
+      "grad_norm": 0.1782815009355545,
+      "learning_rate": 4.962583335143e-06,
+      "loss": 1.1295,
+      "step": 5170
+    },
+    {
+      "epoch": 0.037496290183644955,
+      "grad_norm": 0.21609275043010712,
+      "learning_rate": 4.962510948482414e-06,
+      "loss": 1.1273,
+      "step": 5180
+    },
+    {
+      "epoch": 0.037568676844231146,
+      "grad_norm": 0.19607985019683838,
+      "learning_rate": 4.962438561821828e-06,
+      "loss": 1.116,
+      "step": 5190
+    },
+    {
+      "epoch": 0.03764106350481733,
+      "grad_norm": 0.26767560839653015,
+      "learning_rate": 4.962366175161242e-06,
+      "loss": 1.1085,
+      "step": 5200
+    },
+    {
+      "epoch": 0.03771345016540352,
+      "grad_norm": 0.2847156524658203,
+      "learning_rate": 4.962293788500656e-06,
+      "loss": 1.1067,
+      "step": 5210
+    },
+    {
+      "epoch": 0.0377858368259897,
+      "grad_norm": 0.20309562981128693,
+      "learning_rate": 4.962221401840069e-06,
+      "loss": 1.1204,
+      "step": 5220
+    },
+    {
+      "epoch": 0.037858223486575894,
+      "grad_norm": 0.20030486583709717,
+      "learning_rate": 4.962149015179483e-06,
+      "loss": 1.1203,
+      "step": 5230
+    },
+    {
+      "epoch": 0.03793061014716208,
+      "grad_norm": 0.16900622844696045,
+      "learning_rate": 4.962076628518897e-06,
+      "loss": 1.1398,
+      "step": 5240
+    },
+    {
+      "epoch": 0.03800299680774827,
+      "grad_norm": 0.18107786774635315,
+      "learning_rate": 4.962004241858311e-06,
+      "loss": 1.1224,
+      "step": 5250
+    },
+    {
+      "epoch": 0.03807538346833445,
+      "grad_norm": 0.19224514067173004,
+      "learning_rate": 4.9619318551977245e-06,
+      "loss": 1.1198,
+      "step": 5260
+    },
+    {
+      "epoch": 0.03814777012892064,
+      "grad_norm": 0.18875695765018463,
+      "learning_rate": 4.961859468537138e-06,
+      "loss": 1.1281,
+      "step": 5270
+    },
+    {
+      "epoch": 0.03822015678950683,
+      "grad_norm": 0.20869621634483337,
+      "learning_rate": 4.961787081876553e-06,
+      "loss": 1.0947,
+      "step": 5280
+    },
+    {
+      "epoch": 0.03829254345009302,
+      "grad_norm": 0.18768785893917084,
+      "learning_rate": 4.961714695215966e-06,
+      "loss": 1.1132,
+      "step": 5290
+    },
+    {
+      "epoch": 0.03836493011067921,
+      "grad_norm": 0.19424548745155334,
+      "learning_rate": 4.96164230855538e-06,
+      "loss": 1.1311,
+      "step": 5300
+    },
+    {
+      "epoch": 0.03843731677126539,
+      "grad_norm": 0.18979331851005554,
+      "learning_rate": 4.9615699218947934e-06,
+      "loss": 1.1254,
+      "step": 5310
+    },
+    {
+      "epoch": 0.03850970343185158,
+      "grad_norm": 0.18203093111515045,
+      "learning_rate": 4.961497535234208e-06,
+      "loss": 1.133,
+      "step": 5320
+    },
+    {
+      "epoch": 0.038582090092437765,
+      "grad_norm": 0.1737833172082901,
+      "learning_rate": 4.9614251485736215e-06,
+      "loss": 1.1132,
+      "step": 5330
+    },
+    {
+      "epoch": 0.038654476753023956,
+      "grad_norm": 0.18952320516109467,
+      "learning_rate": 4.961352761913035e-06,
+      "loss": 1.1311,
+      "step": 5340
+    },
+    {
+      "epoch": 0.03872686341361014,
+      "grad_norm": 0.25203409790992737,
+      "learning_rate": 4.961280375252449e-06,
+      "loss": 1.1111,
+      "step": 5350
+    },
+    {
+      "epoch": 0.03879925007419633,
+      "grad_norm": 0.20354746282100677,
+      "learning_rate": 4.961207988591863e-06,
+      "loss": 1.1149,
+      "step": 5360
+    },
+    {
+      "epoch": 0.038871636734782514,
+      "grad_norm": 0.21158255636692047,
+      "learning_rate": 4.961135601931277e-06,
+      "loss": 1.1342,
+      "step": 5370
+    },
+    {
+      "epoch": 0.038944023395368704,
+      "grad_norm": 0.19399100542068481,
+      "learning_rate": 4.9610632152706904e-06,
+      "loss": 1.1173,
+      "step": 5380
+    },
+    {
+      "epoch": 0.03901641005595489,
+      "grad_norm": 0.18055978417396545,
+      "learning_rate": 4.960990828610104e-06,
+      "loss": 1.1228,
+      "step": 5390
+    },
+    {
+      "epoch": 0.03908879671654108,
+      "grad_norm": 0.17884358763694763,
+      "learning_rate": 4.9609184419495185e-06,
+      "loss": 1.1214,
+      "step": 5400
+    },
+    {
+      "epoch": 0.03916118337712726,
+      "grad_norm": 0.1775633841753006,
+      "learning_rate": 4.960846055288932e-06,
+      "loss": 1.121,
+      "step": 5410
+    },
+    {
+      "epoch": 0.03923357003771345,
+      "grad_norm": 0.18697598576545715,
+      "learning_rate": 4.960773668628345e-06,
+      "loss": 1.147,
+      "step": 5420
+    },
+    {
+      "epoch": 0.039305956698299636,
+      "grad_norm": 0.8238439559936523,
+      "learning_rate": 4.960701281967759e-06,
+      "loss": 1.1239,
+      "step": 5430
+    },
+    {
+      "epoch": 0.03937834335888583,
+      "grad_norm": 0.18985433876514435,
+      "learning_rate": 4.960628895307173e-06,
+      "loss": 1.1123,
+      "step": 5440
+    },
+    {
+      "epoch": 0.03945073001947201,
+      "grad_norm": 0.18031606078147888,
+      "learning_rate": 4.960556508646587e-06,
+      "loss": 1.1203,
+      "step": 5450
+    },
+    {
+      "epoch": 0.0395231166800582,
+      "grad_norm": 0.16933594644069672,
+      "learning_rate": 4.960484121986e-06,
+      "loss": 1.118,
+      "step": 5460
+    },
+    {
+      "epoch": 0.039595503340644385,
+      "grad_norm": 0.1860426515340805,
+      "learning_rate": 4.960411735325415e-06,
+      "loss": 1.1193,
+      "step": 5470
+    },
+    {
+      "epoch": 0.039667890001230575,
+      "grad_norm": 0.18198014795780182,
+      "learning_rate": 4.960339348664828e-06,
+      "loss": 1.1229,
+      "step": 5480
+    },
+    {
+      "epoch": 0.03974027666181676,
+      "grad_norm": 0.19671514630317688,
+      "learning_rate": 4.960266962004242e-06,
+      "loss": 1.1263,
+      "step": 5490
+    },
+    {
+      "epoch": 0.03981266332240295,
+      "grad_norm": 0.18220868706703186,
+      "learning_rate": 4.9601945753436555e-06,
+      "loss": 1.122,
+      "step": 5500
+    },
+    {
+      "epoch": 0.03988504998298913,
+      "grad_norm": 0.19989323616027832,
+      "learning_rate": 4.96012218868307e-06,
+      "loss": 1.1202,
+      "step": 5510
+    },
+    {
+      "epoch": 0.039957436643575324,
+      "grad_norm": 0.1794479936361313,
+      "learning_rate": 4.960049802022484e-06,
+      "loss": 1.1221,
+      "step": 5520
+    },
+    {
+      "epoch": 0.04002982330416151,
+      "grad_norm": 0.23208436369895935,
+      "learning_rate": 4.959977415361897e-06,
+      "loss": 1.1197,
+      "step": 5530
+    },
+    {
+      "epoch": 0.0401022099647477,
+      "grad_norm": 0.20182034373283386,
+      "learning_rate": 4.959905028701311e-06,
+      "loss": 1.1272,
+      "step": 5540
+    },
+    {
+      "epoch": 0.04017459662533388,
+      "grad_norm": 0.18753409385681152,
+      "learning_rate": 4.959832642040725e-06,
+      "loss": 1.1078,
+      "step": 5550
+    },
+    {
+      "epoch": 0.04024698328592007,
+      "grad_norm": 0.20286722481250763,
+      "learning_rate": 4.959760255380139e-06,
+      "loss": 1.1251,
+      "step": 5560
+    },
+    {
+      "epoch": 0.040319369946506256,
+      "grad_norm": 0.21133583784103394,
+      "learning_rate": 4.9596878687195525e-06,
+      "loss": 1.1299,
+      "step": 5570
+    },
+    {
+      "epoch": 0.04039175660709245,
+      "grad_norm": 0.1748242825269699,
+      "learning_rate": 4.959615482058966e-06,
+      "loss": 1.1263,
+      "step": 5580
+    },
+    {
+      "epoch": 0.04046414326767863,
+      "grad_norm": 0.19774052500724792,
+      "learning_rate": 4.959543095398381e-06,
+      "loss": 1.1134,
+      "step": 5590
+    },
+    {
+      "epoch": 0.04053652992826482,
+      "grad_norm": 0.2021917700767517,
+      "learning_rate": 4.959470708737794e-06,
+      "loss": 1.1222,
+      "step": 5600
+    },
+    {
+      "epoch": 0.040608916588851005,
+      "grad_norm": 0.20801125466823578,
+      "learning_rate": 4.959398322077208e-06,
+      "loss": 1.1187,
+      "step": 5610
+    },
+    {
+      "epoch": 0.040681303249437195,
+      "grad_norm": 0.17507006227970123,
+      "learning_rate": 4.9593259354166215e-06,
+      "loss": 1.1209,
+      "step": 5620
+    },
+    {
+      "epoch": 0.04075368991002338,
+      "grad_norm": 0.1864837110042572,
+      "learning_rate": 4.959253548756036e-06,
+      "loss": 1.1278,
+      "step": 5630
+    },
+    {
+      "epoch": 0.04082607657060957,
+      "grad_norm": 0.17824417352676392,
+      "learning_rate": 4.9591811620954495e-06,
+      "loss": 1.114,
+      "step": 5640
+    },
+    {
+      "epoch": 0.04089846323119575,
+      "grad_norm": 0.18606680631637573,
+      "learning_rate": 4.959108775434863e-06,
+      "loss": 1.133,
+      "step": 5650
+    },
+    {
+      "epoch": 0.040970849891781944,
+      "grad_norm": 0.1736188679933548,
+      "learning_rate": 4.959036388774277e-06,
+      "loss": 1.1071,
+      "step": 5660
+    },
+    {
+      "epoch": 0.04104323655236813,
+      "grad_norm": 0.1926575005054474,
+      "learning_rate": 4.958964002113691e-06,
+      "loss": 1.1257,
+      "step": 5670
+    },
+    {
+      "epoch": 0.04111562321295432,
+      "grad_norm": 0.1755741983652115,
+      "learning_rate": 4.958891615453105e-06,
+      "loss": 1.108,
+      "step": 5680
+    },
+    {
+      "epoch": 0.0411880098735405,
+      "grad_norm": 0.19758310914039612,
+      "learning_rate": 4.9588192287925185e-06,
+      "loss": 1.1305,
+      "step": 5690
+    },
+    {
+      "epoch": 0.04126039653412669,
+      "grad_norm": 0.21883004903793335,
+      "learning_rate": 4.958746842131932e-06,
+      "loss": 1.1153,
+      "step": 5700
+    },
+    {
+      "epoch": 0.041332783194712876,
+      "grad_norm": 0.19593234360218048,
+      "learning_rate": 4.9586744554713466e-06,
+      "loss": 1.1249,
+      "step": 5710
+    },
+    {
+      "epoch": 0.041405169855299066,
+      "grad_norm": 0.19684365391731262,
+      "learning_rate": 4.95860206881076e-06,
+      "loss": 1.1193,
+      "step": 5720
+    },
+    {
+      "epoch": 0.04147755651588525,
+      "grad_norm": 0.19557367265224457,
+      "learning_rate": 4.958529682150174e-06,
+      "loss": 1.1251,
+      "step": 5730
+    },
+    {
+      "epoch": 0.04154994317647144,
+      "grad_norm": 0.21626479923725128,
+      "learning_rate": 4.958457295489587e-06,
+      "loss": 1.1123,
+      "step": 5740
+    },
+    {
+      "epoch": 0.041622329837057624,
+      "grad_norm": 0.18775388598442078,
+      "learning_rate": 4.958384908829002e-06,
+      "loss": 1.128,
+      "step": 5750
+    },
+    {
+      "epoch": 0.041694716497643815,
+      "grad_norm": 0.19671566784381866,
+      "learning_rate": 4.9583125221684155e-06,
+      "loss": 1.1164,
+      "step": 5760
+    },
+    {
+      "epoch": 0.04176710315823,
+      "grad_norm": 0.23783642053604126,
+      "learning_rate": 4.958240135507829e-06,
+      "loss": 1.1256,
+      "step": 5770
+    },
+    {
+      "epoch": 0.04183948981881619,
+      "grad_norm": 0.18829865753650665,
+      "learning_rate": 4.958167748847243e-06,
+      "loss": 1.1186,
+      "step": 5780
+    },
+    {
+      "epoch": 0.04191187647940237,
+      "grad_norm": 0.1892513930797577,
+      "learning_rate": 4.958095362186657e-06,
+      "loss": 1.1372,
+      "step": 5790
+    },
+    {
+      "epoch": 0.04198426313998856,
+      "grad_norm": 0.2102990746498108,
+      "learning_rate": 4.958022975526071e-06,
+      "loss": 1.1055,
+      "step": 5800
+    },
+    {
+      "epoch": 0.04205664980057475,
+      "grad_norm": 0.18884964287281036,
+      "learning_rate": 4.957950588865484e-06,
+      "loss": 1.1284,
+      "step": 5810
+    },
+    {
+      "epoch": 0.04212903646116094,
+      "grad_norm": 0.17858710885047913,
+      "learning_rate": 4.957878202204898e-06,
+      "loss": 1.1122,
+      "step": 5820
+    },
+    {
+      "epoch": 0.04220142312174712,
+      "grad_norm": 0.1783481240272522,
+      "learning_rate": 4.957805815544312e-06,
+      "loss": 1.1289,
+      "step": 5830
+    },
+    {
+      "epoch": 0.04227380978233331,
+      "grad_norm": 0.18287035822868347,
+      "learning_rate": 4.957733428883726e-06,
+      "loss": 1.1005,
+      "step": 5840
+    },
+    {
+      "epoch": 0.042346196442919495,
+      "grad_norm": 0.20345322787761688,
+      "learning_rate": 4.95766104222314e-06,
+      "loss": 1.1115,
+      "step": 5850
+    },
+    {
+      "epoch": 0.042418583103505686,
+      "grad_norm": 0.1760493516921997,
+      "learning_rate": 4.957588655562553e-06,
+      "loss": 1.1119,
+      "step": 5860
+    },
+    {
+      "epoch": 0.04249096976409188,
+      "grad_norm": 0.1835738718509674,
+      "learning_rate": 4.957516268901967e-06,
+      "loss": 1.1132,
+      "step": 5870
+    },
+    {
+      "epoch": 0.04256335642467806,
+      "grad_norm": 0.17547370493412018,
+      "learning_rate": 4.957443882241381e-06,
+      "loss": 1.1409,
+      "step": 5880
+    },
+    {
+      "epoch": 0.04263574308526425,
+      "grad_norm": 0.17572353780269623,
+      "learning_rate": 4.957371495580795e-06,
+      "loss": 1.1132,
+      "step": 5890
+    },
+    {
+      "epoch": 0.042708129745850434,
+      "grad_norm": 0.18828994035720825,
+      "learning_rate": 4.957299108920209e-06,
+      "loss": 1.1331,
+      "step": 5900
+    },
+    {
+      "epoch": 0.042780516406436625,
+      "grad_norm": 0.20104007422924042,
+      "learning_rate": 4.957226722259622e-06,
+      "loss": 1.0998,
+      "step": 5910
+    },
+    {
+      "epoch": 0.04285290306702281,
+      "grad_norm": 0.1787911206483841,
+      "learning_rate": 4.957154335599037e-06,
+      "loss": 1.105,
+      "step": 5920
+    },
+    {
+      "epoch": 0.042925289727609,
+      "grad_norm": 0.1929769665002823,
+      "learning_rate": 4.95708194893845e-06,
+      "loss": 1.1265,
+      "step": 5930
+    },
+    {
+      "epoch": 0.04299767638819518,
+      "grad_norm": 0.18230368196964264,
+      "learning_rate": 4.957009562277864e-06,
+      "loss": 1.1114,
+      "step": 5940
+    },
+    {
+      "epoch": 0.043070063048781373,
+      "grad_norm": 0.18805713951587677,
+      "learning_rate": 4.956937175617278e-06,
+      "loss": 1.1119,
+      "step": 5950
+    },
+    {
+      "epoch": 0.04314244970936756,
+      "grad_norm": 0.178778737783432,
+      "learning_rate": 4.956864788956691e-06,
+      "loss": 1.105,
+      "step": 5960
+    },
+    {
+      "epoch": 0.04321483636995375,
+      "grad_norm": 0.19835583865642548,
+      "learning_rate": 4.956792402296105e-06,
+      "loss": 1.1146,
+      "step": 5970
+    },
+    {
+      "epoch": 0.04328722303053993,
+      "grad_norm": 0.18930941820144653,
+      "learning_rate": 4.956720015635519e-06,
+      "loss": 1.1172,
+      "step": 5980
+    },
+    {
+      "epoch": 0.04335960969112612,
+      "grad_norm": 0.18036046624183655,
+      "learning_rate": 4.956647628974933e-06,
+      "loss": 1.1233,
+      "step": 5990
+    },
+    {
+      "epoch": 0.043431996351712306,
+      "grad_norm": 0.18780753016471863,
+      "learning_rate": 4.9565752423143465e-06,
+      "loss": 1.1101,
+      "step": 6000
+    },
+    {
+      "epoch": 0.043504383012298496,
+      "grad_norm": 0.2190426141023636,
+      "learning_rate": 4.95650285565376e-06,
+      "loss": 1.1199,
+      "step": 6010
+    },
+    {
+      "epoch": 0.04357676967288468,
+      "grad_norm": 0.20030994713306427,
+      "learning_rate": 4.956430468993174e-06,
+      "loss": 1.1194,
+      "step": 6020
+    },
+    {
+      "epoch": 0.04364915633347087,
+      "grad_norm": 0.17827999591827393,
+      "learning_rate": 4.956358082332588e-06,
+      "loss": 1.1142,
+      "step": 6030
+    },
+    {
+      "epoch": 0.043721542994057054,
+      "grad_norm": 0.22356738150119781,
+      "learning_rate": 4.956285695672002e-06,
+      "loss": 1.1218,
+      "step": 6040
+    },
+    {
+      "epoch": 0.043793929654643245,
+      "grad_norm": 0.18917842209339142,
+      "learning_rate": 4.9562133090114154e-06,
+      "loss": 1.1133,
+      "step": 6050
+    },
+    {
+      "epoch": 0.04386631631522943,
+      "grad_norm": 0.18079480528831482,
+      "learning_rate": 4.956140922350829e-06,
+      "loss": 1.1137,
+      "step": 6060
+    },
+    {
+      "epoch": 0.04393870297581562,
+      "grad_norm": 0.18289583921432495,
+      "learning_rate": 4.9560685356902435e-06,
+      "loss": 1.1261,
+      "step": 6070
+    },
+    {
+      "epoch": 0.0440110896364018,
+      "grad_norm": 0.2138090580701828,
+      "learning_rate": 4.955996149029657e-06,
+      "loss": 1.1005,
+      "step": 6080
+    },
+    {
+      "epoch": 0.04408347629698799,
+      "grad_norm": 0.20163343846797943,
+      "learning_rate": 4.955923762369071e-06,
+      "loss": 1.1079,
+      "step": 6090
+    },
+    {
+      "epoch": 0.04415586295757418,
+      "grad_norm": 0.20078091323375702,
+      "learning_rate": 4.955851375708484e-06,
+      "loss": 1.0869,
+      "step": 6100
+    },
+    {
+      "epoch": 0.04422824961816037,
+      "grad_norm": 0.2066076546907425,
+      "learning_rate": 4.955778989047899e-06,
+      "loss": 1.1208,
+      "step": 6110
+    },
+    {
+      "epoch": 0.04430063627874655,
+      "grad_norm": 0.18639633059501648,
+      "learning_rate": 4.9557066023873124e-06,
+      "loss": 1.1256,
+      "step": 6120
+    },
+    {
+      "epoch": 0.04437302293933274,
+      "grad_norm": 0.1844189614057541,
+      "learning_rate": 4.955634215726726e-06,
+      "loss": 1.1258,
+      "step": 6130
+    },
+    {
+      "epoch": 0.044445409599918925,
+      "grad_norm": 0.1896459013223648,
+      "learning_rate": 4.95556182906614e-06,
+      "loss": 1.1393,
+      "step": 6140
+    },
+    {
+      "epoch": 0.044517796260505116,
+      "grad_norm": 0.17254671454429626,
+      "learning_rate": 4.955489442405554e-06,
+      "loss": 1.1274,
+      "step": 6150
+    },
+    {
+      "epoch": 0.0445901829210913,
+      "grad_norm": 0.22296860814094543,
+      "learning_rate": 4.955417055744968e-06,
+      "loss": 1.125,
+      "step": 6160
+    },
+    {
+      "epoch": 0.04466256958167749,
+      "grad_norm": 0.20404869318008423,
+      "learning_rate": 4.955344669084381e-06,
+      "loss": 1.1093,
+      "step": 6170
+    },
+    {
+      "epoch": 0.044734956242263674,
+      "grad_norm": 0.18639911711215973,
+      "learning_rate": 4.955272282423795e-06,
+      "loss": 1.1051,
+      "step": 6180
+    },
+    {
+      "epoch": 0.044807342902849864,
+      "grad_norm": 0.1889095902442932,
+      "learning_rate": 4.9551998957632095e-06,
+      "loss": 1.1199,
+      "step": 6190
+    },
+    {
+      "epoch": 0.04487972956343605,
+      "grad_norm": 0.19607719779014587,
+      "learning_rate": 4.955127509102623e-06,
+      "loss": 1.1052,
+      "step": 6200
+    },
+    {
+      "epoch": 0.04495211622402224,
+      "grad_norm": 0.18083210289478302,
+      "learning_rate": 4.955055122442037e-06,
+      "loss": 1.129,
+      "step": 6210
+    },
+    {
+      "epoch": 0.04502450288460842,
+      "grad_norm": 0.2071446180343628,
+      "learning_rate": 4.95498273578145e-06,
+      "loss": 1.0987,
+      "step": 6220
+    },
+    {
+      "epoch": 0.04509688954519461,
+      "grad_norm": 0.1783336102962494,
+      "learning_rate": 4.954910349120865e-06,
+      "loss": 1.1306,
+      "step": 6230
+    },
+    {
+      "epoch": 0.045169276205780796,
+      "grad_norm": 0.18706344068050385,
+      "learning_rate": 4.954837962460278e-06,
+      "loss": 1.1217,
+      "step": 6240
+    },
+    {
+      "epoch": 0.04524166286636699,
+      "grad_norm": 0.19343775510787964,
+      "learning_rate": 4.954765575799692e-06,
+      "loss": 1.1079,
+      "step": 6250
+    },
+    {
+      "epoch": 0.04531404952695317,
+      "grad_norm": 0.18519844114780426,
+      "learning_rate": 4.954693189139106e-06,
+      "loss": 1.1148,
+      "step": 6260
+    },
+    {
+      "epoch": 0.04538643618753936,
+      "grad_norm": 0.2034037858247757,
+      "learning_rate": 4.95462080247852e-06,
+      "loss": 1.1,
+      "step": 6270
+    },
+    {
+      "epoch": 0.045458822848125545,
+      "grad_norm": 0.22080808877944946,
+      "learning_rate": 4.954548415817934e-06,
+      "loss": 1.1046,
+      "step": 6280
+    },
+    {
+      "epoch": 0.045531209508711736,
+      "grad_norm": 0.19020305573940277,
+      "learning_rate": 4.954476029157347e-06,
+      "loss": 1.1003,
+      "step": 6290
+    },
+    {
+      "epoch": 0.04560359616929792,
+      "grad_norm": 0.18018420040607452,
+      "learning_rate": 4.954403642496761e-06,
+      "loss": 1.1247,
+      "step": 6300
+    },
+    {
+      "epoch": 0.04567598282988411,
+      "grad_norm": 0.18034905195236206,
+      "learning_rate": 4.954331255836175e-06,
+      "loss": 1.103,
+      "step": 6310
+    },
+    {
+      "epoch": 0.04574836949047029,
+      "grad_norm": 0.18934369087219238,
+      "learning_rate": 4.954258869175589e-06,
+      "loss": 1.1229,
+      "step": 6320
+    },
+    {
+      "epoch": 0.045820756151056484,
+      "grad_norm": 0.18742787837982178,
+      "learning_rate": 4.954186482515003e-06,
+      "loss": 1.1131,
+      "step": 6330
+    },
+    {
+      "epoch": 0.04589314281164267,
+      "grad_norm": 0.1858285516500473,
+      "learning_rate": 4.954114095854416e-06,
+      "loss": 1.1242,
+      "step": 6340
+    },
+    {
+      "epoch": 0.04596552947222886,
+      "grad_norm": 0.17056511342525482,
+      "learning_rate": 4.954041709193831e-06,
+      "loss": 1.0997,
+      "step": 6350
+    },
+    {
+      "epoch": 0.04603791613281504,
+      "grad_norm": 0.2107161581516266,
+      "learning_rate": 4.953969322533244e-06,
+      "loss": 1.1075,
+      "step": 6360
+    },
+    {
+      "epoch": 0.04611030279340123,
+      "grad_norm": 0.17084477841854095,
+      "learning_rate": 4.953896935872658e-06,
+      "loss": 1.1113,
+      "step": 6370
+    },
+    {
+      "epoch": 0.046182689453987416,
+      "grad_norm": 0.18742406368255615,
+      "learning_rate": 4.9538245492120715e-06,
+      "loss": 1.115,
+      "step": 6380
+    },
+    {
+      "epoch": 0.04625507611457361,
+      "grad_norm": 0.18658484518527985,
+      "learning_rate": 4.953752162551486e-06,
+      "loss": 1.112,
+      "step": 6390
+    },
+    {
+      "epoch": 0.04632746277515979,
+      "grad_norm": 0.19460472464561462,
+      "learning_rate": 4.9536797758909e-06,
+      "loss": 1.1297,
+      "step": 6400
+    },
+    {
+      "epoch": 0.04639984943574598,
+      "grad_norm": 0.1897081881761551,
+      "learning_rate": 4.953607389230313e-06,
+      "loss": 1.1045,
+      "step": 6410
+    },
+    {
+      "epoch": 0.046472236096332165,
+      "grad_norm": 0.1974371373653412,
+      "learning_rate": 4.953535002569727e-06,
+      "loss": 1.106,
+      "step": 6420
+    },
+    {
+      "epoch": 0.046544622756918355,
+      "grad_norm": 0.18914976716041565,
+      "learning_rate": 4.953462615909141e-06,
+      "loss": 1.1199,
+      "step": 6430
+    },
+    {
+      "epoch": 0.04661700941750454,
+      "grad_norm": 0.18014445900917053,
+      "learning_rate": 4.953390229248555e-06,
+      "loss": 1.1011,
+      "step": 6440
+    },
+    {
+      "epoch": 0.04668939607809073,
+      "grad_norm": 0.17030683159828186,
+      "learning_rate": 4.9533178425879686e-06,
+      "loss": 1.1098,
+      "step": 6450
+    },
+    {
+      "epoch": 0.04676178273867692,
+      "grad_norm": 0.19075506925582886,
+      "learning_rate": 4.953245455927382e-06,
+      "loss": 1.1064,
+      "step": 6460
+    },
+    {
+      "epoch": 0.046834169399263104,
+      "grad_norm": 0.19481076300144196,
+      "learning_rate": 4.953173069266796e-06,
+      "loss": 1.1031,
+      "step": 6470
+    },
+    {
+      "epoch": 0.046906556059849294,
+      "grad_norm": 0.1836860328912735,
+      "learning_rate": 4.953100682606209e-06,
+      "loss": 1.1134,
+      "step": 6480
+    },
+    {
+      "epoch": 0.04697894272043548,
+      "grad_norm": 0.18626393377780914,
+      "learning_rate": 4.953028295945623e-06,
+      "loss": 1.1165,
+      "step": 6490
+    },
+    {
+      "epoch": 0.04705132938102167,
+      "grad_norm": 0.17867518961429596,
+      "learning_rate": 4.9529559092850375e-06,
+      "loss": 1.1102,
+      "step": 6500
+    },
+    {
+      "epoch": 0.04712371604160785,
+      "grad_norm": 0.1940499097108841,
+      "learning_rate": 4.952883522624451e-06,
+      "loss": 1.1063,
+      "step": 6510
+    },
+    {
+      "epoch": 0.04719610270219404,
+      "grad_norm": 0.17317171394824982,
+      "learning_rate": 4.952811135963865e-06,
+      "loss": 1.1254,
+      "step": 6520
+    },
+    {
+      "epoch": 0.047268489362780226,
+      "grad_norm": 0.19576644897460938,
+      "learning_rate": 4.952738749303278e-06,
+      "loss": 1.1251,
+      "step": 6530
+    },
+    {
+      "epoch": 0.04734087602336642,
+      "grad_norm": 0.17316751182079315,
+      "learning_rate": 4.952666362642693e-06,
+      "loss": 1.1097,
+      "step": 6540
+    },
+    {
+      "epoch": 0.0474132626839526,
+      "grad_norm": 0.1811741143465042,
+      "learning_rate": 4.952593975982106e-06,
+      "loss": 1.122,
+      "step": 6550
+    },
+    {
+      "epoch": 0.04748564934453879,
+      "grad_norm": 0.1905275136232376,
+      "learning_rate": 4.95252158932152e-06,
+      "loss": 1.1074,
+      "step": 6560
+    },
+    {
+      "epoch": 0.047558036005124975,
+      "grad_norm": 0.19564856588840485,
+      "learning_rate": 4.952449202660934e-06,
+      "loss": 1.1139,
+      "step": 6570
+    },
+    {
+      "epoch": 0.047630422665711165,
+      "grad_norm": 0.17164203524589539,
+      "learning_rate": 4.952376816000348e-06,
+      "loss": 1.1148,
+      "step": 6580
+    },
+    {
+      "epoch": 0.04770280932629735,
+      "grad_norm": 0.19059689342975616,
+      "learning_rate": 4.952304429339762e-06,
+      "loss": 1.1193,
+      "step": 6590
+    },
+    {
+      "epoch": 0.04777519598688354,
+      "grad_norm": 0.18997924029827118,
+      "learning_rate": 4.952232042679175e-06,
+      "loss": 1.1091,
+      "step": 6600
+    },
+    {
+      "epoch": 0.04784758264746972,
+      "grad_norm": 0.2226713001728058,
+      "learning_rate": 4.952159656018589e-06,
+      "loss": 1.1131,
+      "step": 6610
+    },
+    {
+      "epoch": 0.047919969308055914,
+      "grad_norm": 0.18311481177806854,
+      "learning_rate": 4.9520872693580026e-06,
+      "loss": 1.1152,
+      "step": 6620
+    },
+    {
+      "epoch": 0.0479923559686421,
+      "grad_norm": 0.18285690248012543,
+      "learning_rate": 4.952014882697417e-06,
+      "loss": 1.1209,
+      "step": 6630
+    },
+    {
+      "epoch": 0.04806474262922829,
+      "grad_norm": 0.18869620561599731,
+      "learning_rate": 4.951942496036831e-06,
+      "loss": 1.0976,
+      "step": 6640
+    },
+    {
+      "epoch": 0.04813712928981447,
+      "grad_norm": 0.17298665642738342,
+      "learning_rate": 4.951870109376244e-06,
+      "loss": 1.1105,
+      "step": 6650
+    },
+    {
+      "epoch": 0.04820951595040066,
+      "grad_norm": 0.1727328598499298,
+      "learning_rate": 4.951797722715658e-06,
+      "loss": 1.1042,
+      "step": 6660
+    },
+    {
+      "epoch": 0.048281902610986846,
+      "grad_norm": 0.2000247985124588,
+      "learning_rate": 4.951725336055072e-06,
+      "loss": 1.1056,
+      "step": 6670
+    },
+    {
+      "epoch": 0.04835428927157304,
+      "grad_norm": 0.18341891467571259,
+      "learning_rate": 4.951652949394486e-06,
+      "loss": 1.1081,
+      "step": 6680
+    },
+    {
+      "epoch": 0.04842667593215922,
+      "grad_norm": 0.17036312818527222,
+      "learning_rate": 4.9515805627339e-06,
+      "loss": 1.1055,
+      "step": 6690
+    },
+    {
+      "epoch": 0.04849906259274541,
+      "grad_norm": 0.1787406951189041,
+      "learning_rate": 4.951508176073313e-06,
+      "loss": 1.1104,
+      "step": 6700
+    },
+    {
+      "epoch": 0.048571449253331594,
+      "grad_norm": 0.21522028744220734,
+      "learning_rate": 4.951435789412728e-06,
+      "loss": 1.1306,
+      "step": 6710
+    },
+    {
+      "epoch": 0.048643835913917785,
+      "grad_norm": 0.18588437139987946,
+      "learning_rate": 4.951363402752141e-06,
+      "loss": 1.1115,
+      "step": 6720
+    },
+    {
+      "epoch": 0.04871622257450397,
+      "grad_norm": 0.18174554407596588,
+      "learning_rate": 4.951291016091555e-06,
+      "loss": 1.1072,
+      "step": 6730
+    },
+    {
+      "epoch": 0.04878860923509016,
+      "grad_norm": 0.20772916078567505,
+      "learning_rate": 4.9512186294309685e-06,
+      "loss": 1.1141,
+      "step": 6740
+    },
+    {
+      "epoch": 0.04886099589567634,
+      "grad_norm": 0.1825239658355713,
+      "learning_rate": 4.951146242770383e-06,
+      "loss": 1.0956,
+      "step": 6750
+    },
+    {
+      "epoch": 0.048933382556262534,
+      "grad_norm": 0.1815953105688095,
+      "learning_rate": 4.951073856109797e-06,
+      "loss": 1.1061,
+      "step": 6760
+    },
+    {
+      "epoch": 0.04900576921684872,
+      "grad_norm": 0.20214617252349854,
+      "learning_rate": 4.95100146944921e-06,
+      "loss": 1.1076,
+      "step": 6770
+    },
+    {
+      "epoch": 0.04907815587743491,
+      "grad_norm": 0.18477530777454376,
+      "learning_rate": 4.950929082788624e-06,
+      "loss": 1.1158,
+      "step": 6780
+    },
+    {
+      "epoch": 0.04915054253802109,
+      "grad_norm": 0.18946325778961182,
+      "learning_rate": 4.950856696128038e-06,
+      "loss": 1.0897,
+      "step": 6790
+    },
+    {
+      "epoch": 0.04922292919860728,
+      "grad_norm": 0.18577060103416443,
+      "learning_rate": 4.950784309467452e-06,
+      "loss": 1.1187,
+      "step": 6800
+    },
+    {
+      "epoch": 0.049295315859193466,
+      "grad_norm": 0.21096892654895782,
+      "learning_rate": 4.9507119228068655e-06,
+      "loss": 1.117,
+      "step": 6810
+    },
+    {
+      "epoch": 0.049367702519779656,
+      "grad_norm": 0.18540360033512115,
+      "learning_rate": 4.950639536146279e-06,
+      "loss": 1.1034,
+      "step": 6820
+    },
+    {
+      "epoch": 0.04944008918036584,
+      "grad_norm": 0.18726834654808044,
+      "learning_rate": 4.950567149485694e-06,
+      "loss": 1.1083,
+      "step": 6830
+    },
+    {
+      "epoch": 0.04951247584095203,
+      "grad_norm": 0.1806015968322754,
+      "learning_rate": 4.950494762825107e-06,
+      "loss": 1.1017,
+      "step": 6840
+    },
+    {
+      "epoch": 0.049584862501538214,
+      "grad_norm": 0.19054830074310303,
+      "learning_rate": 4.950422376164521e-06,
+      "loss": 1.1093,
+      "step": 6850
+    },
+    {
+      "epoch": 0.049657249162124405,
+      "grad_norm": 0.18223927915096283,
+      "learning_rate": 4.9503499895039344e-06,
+      "loss": 1.1138,
+      "step": 6860
+    },
+    {
+      "epoch": 0.04972963582271059,
+      "grad_norm": 0.18125773966312408,
+      "learning_rate": 4.950277602843349e-06,
+      "loss": 1.0931,
+      "step": 6870
+    },
+    {
+      "epoch": 0.04980202248329678,
+      "grad_norm": 0.18000420928001404,
+      "learning_rate": 4.9502052161827625e-06,
+      "loss": 1.1019,
+      "step": 6880
+    },
+    {
+      "epoch": 0.04987440914388296,
+      "grad_norm": 0.21712656319141388,
+      "learning_rate": 4.950132829522176e-06,
+      "loss": 1.1233,
+      "step": 6890
+    },
+    {
+      "epoch": 0.04994679580446915,
+      "grad_norm": 0.20299050211906433,
+      "learning_rate": 4.95006044286159e-06,
+      "loss": 1.1055,
+      "step": 6900
+    },
+    {
+      "epoch": 0.05001918246505534,
+      "grad_norm": 0.1789911538362503,
+      "learning_rate": 4.949988056201004e-06,
+      "loss": 1.0976,
+      "step": 6910
+    },
+    {
+      "epoch": 0.05009156912564153,
+      "grad_norm": 0.18480534851551056,
+      "learning_rate": 4.949915669540418e-06,
+      "loss": 1.1133,
+      "step": 6920
+    },
+    {
+      "epoch": 0.05016395578622771,
+      "grad_norm": 0.17870941758155823,
+      "learning_rate": 4.9498432828798315e-06,
+      "loss": 1.1187,
+      "step": 6930
+    },
+    {
+      "epoch": 0.0502363424468139,
+      "grad_norm": 0.18759605288505554,
+      "learning_rate": 4.949770896219245e-06,
+      "loss": 1.0889,
+      "step": 6940
+    },
+    {
+      "epoch": 0.050308729107400085,
+      "grad_norm": 0.19115033745765686,
+      "learning_rate": 4.9496985095586595e-06,
+      "loss": 1.1194,
+      "step": 6950
+    },
+    {
+      "epoch": 0.050381115767986276,
+      "grad_norm": 0.19017775356769562,
+      "learning_rate": 4.949626122898073e-06,
+      "loss": 1.1183,
+      "step": 6960
+    },
+    {
+      "epoch": 0.05045350242857246,
+      "grad_norm": 0.16906103491783142,
+      "learning_rate": 4.949553736237487e-06,
+      "loss": 1.1165,
+      "step": 6970
+    },
+    {
+      "epoch": 0.05052588908915865,
+      "grad_norm": 0.18214966356754303,
+      "learning_rate": 4.9494813495769e-06,
+      "loss": 1.0915,
+      "step": 6980
+    },
+    {
+      "epoch": 0.050598275749744834,
+      "grad_norm": 0.19243939220905304,
+      "learning_rate": 4.949408962916315e-06,
+      "loss": 1.1123,
+      "step": 6990
+    },
+    {
+      "epoch": 0.050670662410331024,
+      "grad_norm": 0.1790931522846222,
+      "learning_rate": 4.9493365762557285e-06,
+      "loss": 1.1139,
+      "step": 7000
+    },
+    {
+      "epoch": 0.05074304907091721,
+      "grad_norm": 0.1852511316537857,
+      "learning_rate": 4.949264189595141e-06,
+      "loss": 1.1077,
+      "step": 7010
+    },
+    {
+      "epoch": 0.0508154357315034,
+      "grad_norm": 0.24233588576316833,
+      "learning_rate": 4.949191802934556e-06,
+      "loss": 1.1023,
+      "step": 7020
+    },
+    {
+      "epoch": 0.05088782239208959,
+      "grad_norm": 0.22311051189899445,
+      "learning_rate": 4.949119416273969e-06,
+      "loss": 1.0985,
+      "step": 7030
+    },
+    {
+      "epoch": 0.05096020905267577,
+      "grad_norm": 0.21093599498271942,
+      "learning_rate": 4.949047029613383e-06,
+      "loss": 1.0985,
+      "step": 7040
+    },
+    {
+      "epoch": 0.05103259571326196,
+      "grad_norm": 0.19992297887802124,
+      "learning_rate": 4.9489746429527965e-06,
+      "loss": 1.1149,
+      "step": 7050
+    },
+    {
+      "epoch": 0.05110498237384815,
+      "grad_norm": 0.23999755084514618,
+      "learning_rate": 4.948902256292211e-06,
+      "loss": 1.1069,
+      "step": 7060
+    },
+    {
+      "epoch": 0.05117736903443434,
+      "grad_norm": 0.1987914890050888,
+      "learning_rate": 4.948829869631625e-06,
+      "loss": 1.1077,
+      "step": 7070
+    },
+    {
+      "epoch": 0.05124975569502052,
+      "grad_norm": 0.2037319839000702,
+      "learning_rate": 4.948757482971038e-06,
+      "loss": 1.107,
+      "step": 7080
+    },
+    {
+      "epoch": 0.05132214235560671,
+      "grad_norm": 0.1777563840150833,
+      "learning_rate": 4.948685096310452e-06,
+      "loss": 1.1182,
+      "step": 7090
+    },
+    {
+      "epoch": 0.051394529016192896,
+      "grad_norm": 0.18838225305080414,
+      "learning_rate": 4.948612709649866e-06,
+      "loss": 1.1226,
+      "step": 7100
+    },
+    {
+      "epoch": 0.051466915676779086,
+      "grad_norm": 0.19094857573509216,
+      "learning_rate": 4.94854032298928e-06,
+      "loss": 1.0974,
+      "step": 7110
+    },
+    {
+      "epoch": 0.05153930233736527,
+      "grad_norm": 0.22589007019996643,
+      "learning_rate": 4.9484679363286935e-06,
+      "loss": 1.1015,
+      "step": 7120
+    },
+    {
+      "epoch": 0.05161168899795146,
+      "grad_norm": 0.1975346952676773,
+      "learning_rate": 4.948395549668107e-06,
+      "loss": 1.1143,
+      "step": 7130
+    },
+    {
+      "epoch": 0.051684075658537644,
+      "grad_norm": 0.17119309306144714,
+      "learning_rate": 4.948323163007522e-06,
+      "loss": 1.1089,
+      "step": 7140
+    },
+    {
+      "epoch": 0.051756462319123835,
+      "grad_norm": 0.1853228211402893,
+      "learning_rate": 4.948250776346935e-06,
+      "loss": 1.1202,
+      "step": 7150
+    },
+    {
+      "epoch": 0.05182884897971002,
+      "grad_norm": 0.21450157463550568,
+      "learning_rate": 4.948178389686349e-06,
+      "loss": 1.1093,
+      "step": 7160
+    },
+    {
+      "epoch": 0.05190123564029621,
+      "grad_norm": 0.1728169471025467,
+      "learning_rate": 4.9481060030257625e-06,
+      "loss": 1.0914,
+      "step": 7170
+    },
+    {
+      "epoch": 0.05197362230088239,
+      "grad_norm": 0.21489717066287994,
+      "learning_rate": 4.948033616365177e-06,
+      "loss": 1.0987,
+      "step": 7180
+    },
+    {
+      "epoch": 0.05204600896146858,
+      "grad_norm": 0.19044430553913116,
+      "learning_rate": 4.9479612297045906e-06,
+      "loss": 1.1089,
+      "step": 7190
+    },
+    {
+      "epoch": 0.05211839562205477,
+      "grad_norm": 0.18932093679904938,
+      "learning_rate": 4.947888843044004e-06,
+      "loss": 1.1093,
+      "step": 7200
+    },
+    {
+      "epoch": 0.05219078228264096,
+      "grad_norm": 0.18193960189819336,
+      "learning_rate": 4.947816456383418e-06,
+      "loss": 1.1116,
+      "step": 7210
+    },
+    {
+      "epoch": 0.05226316894322714,
+      "grad_norm": 0.18055914342403412,
+      "learning_rate": 4.947744069722832e-06,
+      "loss": 1.1052,
+      "step": 7220
+    },
+    {
+      "epoch": 0.05233555560381333,
+      "grad_norm": 0.18837417662143707,
+      "learning_rate": 4.947671683062246e-06,
+      "loss": 1.0917,
+      "step": 7230
+    },
+    {
+      "epoch": 0.052407942264399515,
+      "grad_norm": 0.18194933235645294,
+      "learning_rate": 4.9475992964016595e-06,
+      "loss": 1.0992,
+      "step": 7240
+    },
+    {
+      "epoch": 0.052480328924985706,
+      "grad_norm": 0.2614051103591919,
+      "learning_rate": 4.947526909741073e-06,
+      "loss": 1.0997,
+      "step": 7250
+    },
+    {
+      "epoch": 0.05255271558557189,
+      "grad_norm": 0.1798228770494461,
+      "learning_rate": 4.947454523080487e-06,
+      "loss": 1.1035,
+      "step": 7260
+    },
+    {
+      "epoch": 0.05262510224615808,
+      "grad_norm": 0.1857793927192688,
+      "learning_rate": 4.947382136419901e-06,
+      "loss": 1.099,
+      "step": 7270
+    },
+    {
+      "epoch": 0.052697488906744264,
+      "grad_norm": 0.1768251359462738,
+      "learning_rate": 4.947309749759315e-06,
+      "loss": 1.0987,
+      "step": 7280
+    },
+    {
+      "epoch": 0.052769875567330454,
+      "grad_norm": 0.18754634261131287,
+      "learning_rate": 4.947237363098728e-06,
+      "loss": 1.098,
+      "step": 7290
+    },
+    {
+      "epoch": 0.05284226222791664,
+      "grad_norm": 0.18987387418746948,
+      "learning_rate": 4.947164976438142e-06,
+      "loss": 1.1097,
+      "step": 7300
+    },
+    {
+      "epoch": 0.05291464888850283,
+      "grad_norm": 0.17656712234020233,
+      "learning_rate": 4.9470925897775565e-06,
+      "loss": 1.1031,
+      "step": 7310
+    },
+    {
+      "epoch": 0.05298703554908901,
+      "grad_norm": 0.20419563353061676,
+      "learning_rate": 4.94702020311697e-06,
+      "loss": 1.1133,
+      "step": 7320
+    },
+    {
+      "epoch": 0.0530594222096752,
+      "grad_norm": 0.19128115475177765,
+      "learning_rate": 4.946947816456384e-06,
+      "loss": 1.1022,
+      "step": 7330
+    },
+    {
+      "epoch": 0.053131808870261386,
+      "grad_norm": 0.17749454081058502,
+      "learning_rate": 4.946875429795797e-06,
+      "loss": 1.1134,
+      "step": 7340
+    },
+    {
+      "epoch": 0.05320419553084758,
+      "grad_norm": 0.18553034961223602,
+      "learning_rate": 4.946803043135212e-06,
+      "loss": 1.1059,
+      "step": 7350
+    },
+    {
+      "epoch": 0.05327658219143376,
+      "grad_norm": 0.2216414511203766,
+      "learning_rate": 4.946730656474625e-06,
+      "loss": 1.1099,
+      "step": 7360
+    },
+    {
+      "epoch": 0.05334896885201995,
+      "grad_norm": 0.1996411681175232,
+      "learning_rate": 4.946658269814039e-06,
+      "loss": 1.0844,
+      "step": 7370
+    },
+    {
+      "epoch": 0.053421355512606135,
+      "grad_norm": 0.17310328781604767,
+      "learning_rate": 4.946585883153453e-06,
+      "loss": 1.0895,
+      "step": 7380
+    },
+    {
+      "epoch": 0.053493742173192325,
+      "grad_norm": 0.18325687944889069,
+      "learning_rate": 4.946513496492867e-06,
+      "loss": 1.1017,
+      "step": 7390
+    },
+    {
+      "epoch": 0.05356612883377851,
+      "grad_norm": 0.1925450563430786,
+      "learning_rate": 4.946441109832281e-06,
+      "loss": 1.1152,
+      "step": 7400
+    },
+    {
+      "epoch": 0.0536385154943647,
+      "grad_norm": 0.20546835660934448,
+      "learning_rate": 4.946368723171694e-06,
+      "loss": 1.0937,
+      "step": 7410
+    },
+    {
+      "epoch": 0.05371090215495088,
+      "grad_norm": 0.19936460256576538,
+      "learning_rate": 4.946296336511108e-06,
+      "loss": 1.0876,
+      "step": 7420
+    },
+    {
+      "epoch": 0.053783288815537074,
+      "grad_norm": 0.18377749621868134,
+      "learning_rate": 4.9462239498505224e-06,
+      "loss": 1.1009,
+      "step": 7430
+    },
+    {
+      "epoch": 0.05385567547612326,
+      "grad_norm": 0.18654756247997284,
+      "learning_rate": 4.946151563189936e-06,
+      "loss": 1.1017,
+      "step": 7440
+    },
+    {
+      "epoch": 0.05392806213670945,
+      "grad_norm": 0.17584462463855743,
+      "learning_rate": 4.94607917652935e-06,
+      "loss": 1.1111,
+      "step": 7450
+    },
+    {
+      "epoch": 0.05400044879729563,
+      "grad_norm": 0.1823701709508896,
+      "learning_rate": 4.946006789868763e-06,
+      "loss": 1.0983,
+      "step": 7460
+    },
+    {
+      "epoch": 0.05407283545788182,
+      "grad_norm": 0.17409732937812805,
+      "learning_rate": 4.945934403208178e-06,
+      "loss": 1.1101,
+      "step": 7470
+    },
+    {
+      "epoch": 0.054145222118468006,
+      "grad_norm": 0.19750286638736725,
+      "learning_rate": 4.945862016547591e-06,
+      "loss": 1.1043,
+      "step": 7480
+    },
+    {
+      "epoch": 0.0542176087790542,
+      "grad_norm": 0.18154850602149963,
+      "learning_rate": 4.945789629887005e-06,
+      "loss": 1.1078,
+      "step": 7490
+    },
+    {
+      "epoch": 0.05428999543964038,
+      "grad_norm": 0.17519652843475342,
+      "learning_rate": 4.945717243226419e-06,
+      "loss": 1.0995,
+      "step": 7500
+    },
+    {
+      "epoch": 0.05436238210022657,
+      "grad_norm": 0.18698933720588684,
+      "learning_rate": 4.945644856565833e-06,
+      "loss": 1.0991,
+      "step": 7510
+    },
+    {
+      "epoch": 0.054434768760812755,
+      "grad_norm": 0.1779969483613968,
+      "learning_rate": 4.945572469905247e-06,
+      "loss": 1.1029,
+      "step": 7520
+    },
+    {
+      "epoch": 0.054507155421398945,
+      "grad_norm": 0.1794944554567337,
+      "learning_rate": 4.94550008324466e-06,
+      "loss": 1.0927,
+      "step": 7530
+    },
+    {
+      "epoch": 0.05457954208198513,
+      "grad_norm": 0.1742369830608368,
+      "learning_rate": 4.945427696584074e-06,
+      "loss": 1.1085,
+      "step": 7540
+    },
+    {
+      "epoch": 0.05465192874257132,
+      "grad_norm": 0.17996317148208618,
+      "learning_rate": 4.9453553099234875e-06,
+      "loss": 1.1053,
+      "step": 7550
+    },
+    {
+      "epoch": 0.0547243154031575,
+      "grad_norm": 0.18489129841327667,
+      "learning_rate": 4.945282923262901e-06,
+      "loss": 1.0965,
+      "step": 7560
+    },
+    {
+      "epoch": 0.054796702063743694,
+      "grad_norm": 0.2375306636095047,
+      "learning_rate": 4.945210536602315e-06,
+      "loss": 1.1016,
+      "step": 7570
+    },
+    {
+      "epoch": 0.05486908872432988,
+      "grad_norm": 0.18841832876205444,
+      "learning_rate": 4.945138149941729e-06,
+      "loss": 1.1089,
+      "step": 7580
+    },
+    {
+      "epoch": 0.05494147538491607,
+      "grad_norm": 0.17497693002223969,
+      "learning_rate": 4.945065763281143e-06,
+      "loss": 1.0913,
+      "step": 7590
+    },
+    {
+      "epoch": 0.05501386204550225,
+      "grad_norm": 0.20578007400035858,
+      "learning_rate": 4.9449933766205564e-06,
+      "loss": 1.0975,
+      "step": 7600
+    },
+    {
+      "epoch": 0.05508624870608844,
+      "grad_norm": 0.1815917193889618,
+      "learning_rate": 4.94492098995997e-06,
+      "loss": 1.0931,
+      "step": 7610
+    },
+    {
+      "epoch": 0.05515863536667463,
+      "grad_norm": 0.18187057971954346,
+      "learning_rate": 4.9448486032993845e-06,
+      "loss": 1.0993,
+      "step": 7620
+    },
+    {
+      "epoch": 0.055231022027260816,
+      "grad_norm": 0.1740749329328537,
+      "learning_rate": 4.944776216638798e-06,
+      "loss": 1.0941,
+      "step": 7630
+    },
+    {
+      "epoch": 0.05530340868784701,
+      "grad_norm": 0.17215760052204132,
+      "learning_rate": 4.944703829978212e-06,
+      "loss": 1.1036,
+      "step": 7640
+    },
+    {
+      "epoch": 0.05537579534843319,
+      "grad_norm": 0.21924471855163574,
+      "learning_rate": 4.944631443317625e-06,
+      "loss": 1.0897,
+      "step": 7650
+    },
+    {
+      "epoch": 0.05544818200901938,
+      "grad_norm": 0.1766793131828308,
+      "learning_rate": 4.94455905665704e-06,
+      "loss": 1.0977,
+      "step": 7660
+    },
+    {
+      "epoch": 0.055520568669605565,
+      "grad_norm": 0.20563791692256927,
+      "learning_rate": 4.9444866699964535e-06,
+      "loss": 1.0981,
+      "step": 7670
+    },
+    {
+      "epoch": 0.055592955330191755,
+      "grad_norm": 0.17593294382095337,
+      "learning_rate": 4.944414283335867e-06,
+      "loss": 1.0932,
+      "step": 7680
+    },
+    {
+      "epoch": 0.05566534199077794,
+      "grad_norm": 0.18366055190563202,
+      "learning_rate": 4.944341896675281e-06,
+      "loss": 1.1067,
+      "step": 7690
+    },
+    {
+      "epoch": 0.05573772865136413,
+      "grad_norm": 0.19099287688732147,
+      "learning_rate": 4.944269510014695e-06,
+      "loss": 1.0877,
+      "step": 7700
+    },
+    {
+      "epoch": 0.05581011531195031,
+      "grad_norm": 0.17935815453529358,
+      "learning_rate": 4.944197123354109e-06,
+      "loss": 1.0943,
+      "step": 7710
+    },
+    {
+      "epoch": 0.055882501972536504,
+      "grad_norm": 0.18561848998069763,
+      "learning_rate": 4.944124736693522e-06,
+      "loss": 1.1099,
+      "step": 7720
+    },
+    {
+      "epoch": 0.05595488863312269,
+      "grad_norm": 0.18966755270957947,
+      "learning_rate": 4.944052350032936e-06,
+      "loss": 1.0981,
+      "step": 7730
+    },
+    {
+      "epoch": 0.05602727529370888,
+      "grad_norm": 0.18499694764614105,
+      "learning_rate": 4.9439799633723505e-06,
+      "loss": 1.103,
+      "step": 7740
+    },
+    {
+      "epoch": 0.05609966195429506,
+      "grad_norm": 0.17918939888477325,
+      "learning_rate": 4.943907576711764e-06,
+      "loss": 1.1007,
+      "step": 7750
+    },
+    {
+      "epoch": 0.05617204861488125,
+      "grad_norm": 0.19508133828639984,
+      "learning_rate": 4.943835190051178e-06,
+      "loss": 1.1104,
+      "step": 7760
+    },
+    {
+      "epoch": 0.056244435275467436,
+      "grad_norm": 0.19462929666042328,
+      "learning_rate": 4.943762803390591e-06,
+      "loss": 1.1089,
+      "step": 7770
+    },
+    {
+      "epoch": 0.056316821936053627,
+      "grad_norm": 0.3181241750717163,
+      "learning_rate": 4.943690416730006e-06,
+      "loss": 1.1036,
+      "step": 7780
+    },
+    {
+      "epoch": 0.05638920859663981,
+      "grad_norm": 0.175667867064476,
+      "learning_rate": 4.943618030069419e-06,
+      "loss": 1.1137,
+      "step": 7790
+    },
+    {
+      "epoch": 0.056461595257226,
+      "grad_norm": 0.17442381381988525,
+      "learning_rate": 4.943545643408833e-06,
+      "loss": 1.0801,
+      "step": 7800
+    },
+    {
+      "epoch": 0.056533981917812184,
+      "grad_norm": 0.17181989550590515,
+      "learning_rate": 4.943473256748247e-06,
+      "loss": 1.1102,
+      "step": 7810
+    },
+    {
+      "epoch": 0.056606368578398375,
+      "grad_norm": 0.19522684812545776,
+      "learning_rate": 4.943400870087661e-06,
+      "loss": 1.103,
+      "step": 7820
+    },
+    {
+      "epoch": 0.05667875523898456,
+      "grad_norm": 0.3031037747859955,
+      "learning_rate": 4.943328483427075e-06,
+      "loss": 1.1028,
+      "step": 7830
+    },
+    {
+      "epoch": 0.05675114189957075,
+      "grad_norm": 0.21572640538215637,
+      "learning_rate": 4.943256096766488e-06,
+      "loss": 1.0969,
+      "step": 7840
+    },
+    {
+      "epoch": 0.05682352856015693,
+      "grad_norm": 0.1849983185529709,
+      "learning_rate": 4.943183710105902e-06,
+      "loss": 1.0988,
+      "step": 7850
+    },
+    {
+      "epoch": 0.05689591522074312,
+      "grad_norm": 0.17931009829044342,
+      "learning_rate": 4.9431113234453155e-06,
+      "loss": 1.0871,
+      "step": 7860
+    },
+    {
+      "epoch": 0.05696830188132931,
+      "grad_norm": 0.19952841103076935,
+      "learning_rate": 4.94303893678473e-06,
+      "loss": 1.1064,
+      "step": 7870
+    },
+    {
+      "epoch": 0.0570406885419155,
+      "grad_norm": 0.19907739758491516,
+      "learning_rate": 4.942966550124144e-06,
+      "loss": 1.0837,
+      "step": 7880
+    },
+    {
+      "epoch": 0.05711307520250168,
+      "grad_norm": 0.1972847580909729,
+      "learning_rate": 4.942894163463557e-06,
+      "loss": 1.0981,
+      "step": 7890
+    },
+    {
+      "epoch": 0.05718546186308787,
+      "grad_norm": 0.17437316477298737,
+      "learning_rate": 4.942821776802971e-06,
+      "loss": 1.0794,
+      "step": 7900
+    },
+    {
+      "epoch": 0.057257848523674056,
+      "grad_norm": 0.24157637357711792,
+      "learning_rate": 4.942749390142385e-06,
+      "loss": 1.0933,
+      "step": 7910
+    },
+    {
+      "epoch": 0.057330235184260246,
+      "grad_norm": 0.17602761089801788,
+      "learning_rate": 4.942677003481799e-06,
+      "loss": 1.1007,
+      "step": 7920
+    },
+    {
+      "epoch": 0.05740262184484643,
+      "grad_norm": 0.17508484423160553,
+      "learning_rate": 4.9426046168212126e-06,
+      "loss": 1.1054,
+      "step": 7930
+    },
+    {
+      "epoch": 0.05747500850543262,
+      "grad_norm": 0.18290212750434875,
+      "learning_rate": 4.942532230160626e-06,
+      "loss": 1.0949,
+      "step": 7940
+    },
+    {
+      "epoch": 0.057547395166018804,
+      "grad_norm": 0.17786473035812378,
+      "learning_rate": 4.942459843500041e-06,
+      "loss": 1.1041,
+      "step": 7950
+    },
+    {
+      "epoch": 0.057619781826604995,
+      "grad_norm": 0.19254614412784576,
+      "learning_rate": 4.942387456839454e-06,
+      "loss": 1.0975,
+      "step": 7960
+    },
+    {
+      "epoch": 0.05769216848719118,
+      "grad_norm": 0.16799108684062958,
+      "learning_rate": 4.942315070178868e-06,
+      "loss": 1.0947,
+      "step": 7970
+    },
+    {
+      "epoch": 0.05776455514777737,
+      "grad_norm": 0.21218262612819672,
+      "learning_rate": 4.9422426835182815e-06,
+      "loss": 1.0826,
+      "step": 7980
+    },
+    {
+      "epoch": 0.05783694180836355,
+      "grad_norm": 0.1867387741804123,
+      "learning_rate": 4.942170296857696e-06,
+      "loss": 1.0918,
+      "step": 7990
+    },
+    {
+      "epoch": 0.05790932846894974,
+      "grad_norm": 0.19001053273677826,
+      "learning_rate": 4.9420979101971096e-06,
+      "loss": 1.113,
+      "step": 8000
+    },
+    {
+      "epoch": 0.05798171512953593,
+      "grad_norm": 0.23286646604537964,
+      "learning_rate": 4.942025523536523e-06,
+      "loss": 1.0957,
+      "step": 8010
+    },
+    {
+      "epoch": 0.05805410179012212,
+      "grad_norm": 0.184943288564682,
+      "learning_rate": 4.941953136875937e-06,
+      "loss": 1.1023,
+      "step": 8020
+    },
+    {
+      "epoch": 0.0581264884507083,
+      "grad_norm": 0.20318259298801422,
+      "learning_rate": 4.941880750215351e-06,
+      "loss": 1.1059,
+      "step": 8030
+    },
+    {
+      "epoch": 0.05819887511129449,
+      "grad_norm": 0.17356903851032257,
+      "learning_rate": 4.941808363554765e-06,
+      "loss": 1.0779,
+      "step": 8040
+    },
+    {
+      "epoch": 0.058271261771880675,
+      "grad_norm": 0.19621440768241882,
+      "learning_rate": 4.9417359768941785e-06,
+      "loss": 1.0961,
+      "step": 8050
+    },
+    {
+      "epoch": 0.058343648432466866,
+      "grad_norm": 0.18054349720478058,
+      "learning_rate": 4.941663590233592e-06,
+      "loss": 1.0977,
+      "step": 8060
+    },
+    {
+      "epoch": 0.05841603509305305,
+      "grad_norm": 0.18659920990467072,
+      "learning_rate": 4.941591203573006e-06,
+      "loss": 1.0989,
+      "step": 8070
+    },
+    {
+      "epoch": 0.05848842175363924,
+      "grad_norm": 0.1870632916688919,
+      "learning_rate": 4.941518816912419e-06,
+      "loss": 1.1024,
+      "step": 8080
+    },
+    {
+      "epoch": 0.058560808414225424,
+      "grad_norm": 0.18549518287181854,
+      "learning_rate": 4.941446430251833e-06,
+      "loss": 1.0826,
+      "step": 8090
+    },
+    {
+      "epoch": 0.058633195074811614,
+      "grad_norm": 0.1786888837814331,
+      "learning_rate": 4.941374043591247e-06,
+      "loss": 1.0928,
+      "step": 8100
+    },
+    {
+      "epoch": 0.0587055817353978,
+      "grad_norm": 0.1912764608860016,
+      "learning_rate": 4.941301656930661e-06,
+      "loss": 1.0815,
+      "step": 8110
+    },
+    {
+      "epoch": 0.05877796839598399,
+      "grad_norm": 0.17277614772319794,
+      "learning_rate": 4.941229270270075e-06,
+      "loss": 1.0938,
+      "step": 8120
+    },
+    {
+      "epoch": 0.05885035505657017,
+      "grad_norm": 0.18441365659236908,
+      "learning_rate": 4.941156883609488e-06,
+      "loss": 1.1084,
+      "step": 8130
+    },
+    {
+      "epoch": 0.05892274171715636,
+      "grad_norm": 0.24493847787380219,
+      "learning_rate": 4.941084496948903e-06,
+      "loss": 1.0924,
+      "step": 8140
+    },
+    {
+      "epoch": 0.058995128377742546,
+      "grad_norm": 0.1992846131324768,
+      "learning_rate": 4.941012110288316e-06,
+      "loss": 1.0969,
+      "step": 8150
+    },
+    {
+      "epoch": 0.05906751503832874,
+      "grad_norm": 0.20065009593963623,
+      "learning_rate": 4.94093972362773e-06,
+      "loss": 1.1019,
+      "step": 8160
+    },
+    {
+      "epoch": 0.05913990169891492,
+      "grad_norm": 0.17659001052379608,
+      "learning_rate": 4.940867336967144e-06,
+      "loss": 1.0926,
+      "step": 8170
+    },
+    {
+      "epoch": 0.05921228835950111,
+      "grad_norm": 0.19576965272426605,
+      "learning_rate": 4.940794950306558e-06,
+      "loss": 1.1049,
+      "step": 8180
+    },
+    {
+      "epoch": 0.059284675020087295,
+      "grad_norm": 0.20090840756893158,
+      "learning_rate": 4.940722563645972e-06,
+      "loss": 1.1027,
+      "step": 8190
+    },
+    {
+      "epoch": 0.059357061680673485,
+      "grad_norm": 0.16997766494750977,
+      "learning_rate": 4.940650176985385e-06,
+      "loss": 1.0915,
+      "step": 8200
+    },
+    {
+      "epoch": 0.059429448341259676,
+      "grad_norm": 0.20561368763446808,
+      "learning_rate": 4.940577790324799e-06,
+      "loss": 1.0983,
+      "step": 8210
+    },
+    {
+      "epoch": 0.05950183500184586,
+      "grad_norm": 0.20617088675498962,
+      "learning_rate": 4.940505403664213e-06,
+      "loss": 1.0889,
+      "step": 8220
+    },
+    {
+      "epoch": 0.05957422166243205,
+      "grad_norm": 0.1871093362569809,
+      "learning_rate": 4.940433017003627e-06,
+      "loss": 1.0964,
+      "step": 8230
+    },
+    {
+      "epoch": 0.059646608323018234,
+      "grad_norm": 0.1626369059085846,
+      "learning_rate": 4.940360630343041e-06,
+      "loss": 1.0938,
+      "step": 8240
+    },
+    {
+      "epoch": 0.059718994983604425,
+      "grad_norm": 0.18901404738426208,
+      "learning_rate": 4.940288243682454e-06,
+      "loss": 1.0823,
+      "step": 8250
+    },
+    {
+      "epoch": 0.05979138164419061,
+      "grad_norm": 0.18634545803070068,
+      "learning_rate": 4.940215857021869e-06,
+      "loss": 1.0863,
+      "step": 8260
+    },
+    {
+      "epoch": 0.0598637683047768,
+      "grad_norm": 0.1944178342819214,
+      "learning_rate": 4.940143470361282e-06,
+      "loss": 1.0806,
+      "step": 8270
+    },
+    {
+      "epoch": 0.05993615496536298,
+      "grad_norm": 0.1961442083120346,
+      "learning_rate": 4.940071083700696e-06,
+      "loss": 1.0876,
+      "step": 8280
+    },
+    {
+      "epoch": 0.06000854162594917,
+      "grad_norm": 0.17273077368736267,
+      "learning_rate": 4.9399986970401095e-06,
+      "loss": 1.091,
+      "step": 8290
+    },
+    {
+      "epoch": 0.06008092828653536,
+      "grad_norm": 0.24428923428058624,
+      "learning_rate": 4.939926310379524e-06,
+      "loss": 1.0987,
+      "step": 8300
+    },
+    {
+      "epoch": 0.06015331494712155,
+      "grad_norm": 0.18236348032951355,
+      "learning_rate": 4.939853923718938e-06,
+      "loss": 1.0951,
+      "step": 8310
+    },
+    {
+      "epoch": 0.06022570160770773,
+      "grad_norm": 0.18725799024105072,
+      "learning_rate": 4.939781537058351e-06,
+      "loss": 1.0989,
+      "step": 8320
+    },
+    {
+      "epoch": 0.06029808826829392,
+      "grad_norm": 0.21253715455532074,
+      "learning_rate": 4.939709150397765e-06,
+      "loss": 1.1017,
+      "step": 8330
+    },
+    {
+      "epoch": 0.060370474928880105,
+      "grad_norm": 0.17409272491931915,
+      "learning_rate": 4.939636763737179e-06,
+      "loss": 1.0809,
+      "step": 8340
+    },
+    {
+      "epoch": 0.060442861589466296,
+      "grad_norm": 0.21270409226417542,
+      "learning_rate": 4.939564377076593e-06,
+      "loss": 1.084,
+      "step": 8350
+    },
+    {
+      "epoch": 0.06051524825005248,
+      "grad_norm": 0.17901867628097534,
+      "learning_rate": 4.9394919904160065e-06,
+      "loss": 1.0841,
+      "step": 8360
+    },
+    {
+      "epoch": 0.06058763491063867,
+      "grad_norm": 0.18769490718841553,
+      "learning_rate": 4.93941960375542e-06,
+      "loss": 1.0789,
+      "step": 8370
+    },
+    {
+      "epoch": 0.060660021571224854,
+      "grad_norm": 0.18529854714870453,
+      "learning_rate": 4.939347217094835e-06,
+      "loss": 1.0867,
+      "step": 8380
+    },
+    {
+      "epoch": 0.060732408231811044,
+      "grad_norm": 0.18796664476394653,
+      "learning_rate": 4.939274830434248e-06,
+      "loss": 1.0871,
+      "step": 8390
+    },
+    {
+      "epoch": 0.06080479489239723,
+      "grad_norm": 0.1806957870721817,
+      "learning_rate": 4.939202443773662e-06,
+      "loss": 1.1102,
+      "step": 8400
+    },
+    {
+      "epoch": 0.06087718155298342,
+      "grad_norm": 0.19365540146827698,
+      "learning_rate": 4.9391300571130754e-06,
+      "loss": 1.0842,
+      "step": 8410
+    },
+    {
+      "epoch": 0.0609495682135696,
+      "grad_norm": 0.18744197487831116,
+      "learning_rate": 4.93905767045249e-06,
+      "loss": 1.1065,
+      "step": 8420
+    },
+    {
+      "epoch": 0.06102195487415579,
+      "grad_norm": 0.1846928596496582,
+      "learning_rate": 4.9389852837919035e-06,
+      "loss": 1.0979,
+      "step": 8430
+    },
+    {
+      "epoch": 0.061094341534741976,
+      "grad_norm": 0.17853565514087677,
+      "learning_rate": 4.938912897131317e-06,
+      "loss": 1.0965,
+      "step": 8440
+    },
+    {
+      "epoch": 0.06116672819532817,
+      "grad_norm": 0.18652281165122986,
+      "learning_rate": 4.938840510470731e-06,
+      "loss": 1.0835,
+      "step": 8450
+    },
+    {
+      "epoch": 0.06123911485591435,
+      "grad_norm": 0.2018202245235443,
+      "learning_rate": 4.938768123810145e-06,
+      "loss": 1.0853,
+      "step": 8460
+    },
+    {
+      "epoch": 0.06131150151650054,
+      "grad_norm": 0.17888695001602173,
+      "learning_rate": 4.938695737149559e-06,
+      "loss": 1.0994,
+      "step": 8470
+    },
+    {
+      "epoch": 0.061383888177086725,
+      "grad_norm": 0.18451561033725739,
+      "learning_rate": 4.9386233504889725e-06,
+      "loss": 1.0874,
+      "step": 8480
+    },
+    {
+      "epoch": 0.061456274837672915,
+      "grad_norm": 0.18623459339141846,
+      "learning_rate": 4.938550963828386e-06,
+      "loss": 1.108,
+      "step": 8490
+    },
+    {
+      "epoch": 0.0615286614982591,
+      "grad_norm": 0.1710231751203537,
+      "learning_rate": 4.9384785771678e-06,
+      "loss": 1.0908,
+      "step": 8500
+    },
+    {
+      "epoch": 0.06160104815884529,
+      "grad_norm": 0.2325315922498703,
+      "learning_rate": 4.938406190507214e-06,
+      "loss": 1.0932,
+      "step": 8510
+    },
+    {
+      "epoch": 0.06167343481943147,
+      "grad_norm": 0.17885486781597137,
+      "learning_rate": 4.938333803846628e-06,
+      "loss": 1.088,
+      "step": 8520
+    },
+    {
+      "epoch": 0.061745821480017664,
+      "grad_norm": 0.18864506483078003,
+      "learning_rate": 4.938261417186041e-06,
+      "loss": 1.0981,
+      "step": 8530
+    },
+    {
+      "epoch": 0.06181820814060385,
+      "grad_norm": 0.18470600247383118,
+      "learning_rate": 4.938189030525455e-06,
+      "loss": 1.0974,
+      "step": 8540
+    },
+    {
+      "epoch": 0.06189059480119004,
+      "grad_norm": 0.18703562021255493,
+      "learning_rate": 4.9381166438648695e-06,
+      "loss": 1.0987,
+      "step": 8550
+    },
+    {
+      "epoch": 0.06196298146177622,
+      "grad_norm": 0.18582236766815186,
+      "learning_rate": 4.938044257204283e-06,
+      "loss": 1.0925,
+      "step": 8560
+    },
+    {
+      "epoch": 0.06203536812236241,
+      "grad_norm": 0.2140107899904251,
+      "learning_rate": 4.937971870543697e-06,
+      "loss": 1.0907,
+      "step": 8570
+    },
+    {
+      "epoch": 0.062107754782948596,
+      "grad_norm": 0.19368021190166473,
+      "learning_rate": 4.93789948388311e-06,
+      "loss": 1.0877,
+      "step": 8580
+    },
+    {
+      "epoch": 0.06218014144353479,
+      "grad_norm": 0.17043587565422058,
+      "learning_rate": 4.937827097222525e-06,
+      "loss": 1.0851,
+      "step": 8590
+    },
+    {
+      "epoch": 0.06225252810412097,
+      "grad_norm": 0.18179139494895935,
+      "learning_rate": 4.9377547105619375e-06,
+      "loss": 1.095,
+      "step": 8600
+    },
+    {
+      "epoch": 0.06232491476470716,
+      "grad_norm": 0.19978579878807068,
+      "learning_rate": 4.937682323901352e-06,
+      "loss": 1.0989,
+      "step": 8610
+    },
+    {
+      "epoch": 0.062397301425293344,
+      "grad_norm": 0.1800697147846222,
+      "learning_rate": 4.937609937240766e-06,
+      "loss": 1.1074,
+      "step": 8620
+    },
+    {
+      "epoch": 0.062469688085879535,
+      "grad_norm": 0.176766499876976,
+      "learning_rate": 4.937537550580179e-06,
+      "loss": 1.0897,
+      "step": 8630
+    },
+    {
+      "epoch": 0.06254207474646573,
+      "grad_norm": 0.19286136329174042,
+      "learning_rate": 4.937465163919593e-06,
+      "loss": 1.1041,
+      "step": 8640
+    },
+    {
+      "epoch": 0.0626144614070519,
+      "grad_norm": 0.2166859209537506,
+      "learning_rate": 4.937392777259007e-06,
+      "loss": 1.0827,
+      "step": 8650
+    },
+    {
+      "epoch": 0.06268684806763809,
+      "grad_norm": 0.17568807303905487,
+      "learning_rate": 4.937320390598421e-06,
+      "loss": 1.1016,
+      "step": 8660
+    },
+    {
+      "epoch": 0.06275923472822428,
+      "grad_norm": 0.19446556270122528,
+      "learning_rate": 4.9372480039378346e-06,
+      "loss": 1.0951,
+      "step": 8670
+    },
+    {
+      "epoch": 0.06283162138881047,
+      "grad_norm": 0.1830175518989563,
+      "learning_rate": 4.937175617277248e-06,
+      "loss": 1.0891,
+      "step": 8680
+    },
+    {
+      "epoch": 0.06290400804939665,
+      "grad_norm": 0.170004740357399,
+      "learning_rate": 4.937103230616662e-06,
+      "loss": 1.0872,
+      "step": 8690
+    },
+    {
+      "epoch": 0.06297639470998284,
+      "grad_norm": 0.1958620399236679,
+      "learning_rate": 4.937030843956076e-06,
+      "loss": 1.0987,
+      "step": 8700
+    },
+    {
+      "epoch": 0.06304878137056903,
+      "grad_norm": 0.19869394600391388,
+      "learning_rate": 4.93695845729549e-06,
+      "loss": 1.0727,
+      "step": 8710
+    },
+    {
+      "epoch": 0.06312116803115522,
+      "grad_norm": 0.17475168406963348,
+      "learning_rate": 4.9368860706349035e-06,
+      "loss": 1.0991,
+      "step": 8720
+    },
+    {
+      "epoch": 0.0631935546917414,
+      "grad_norm": 0.19080884754657745,
+      "learning_rate": 4.936813683974317e-06,
+      "loss": 1.0988,
+      "step": 8730
+    },
+    {
+      "epoch": 0.06326594135232759,
+      "grad_norm": 0.17002183198928833,
+      "learning_rate": 4.9367412973137316e-06,
+      "loss": 1.0859,
+      "step": 8740
+    },
+    {
+      "epoch": 0.06333832801291378,
+      "grad_norm": 0.19736173748970032,
+      "learning_rate": 4.936668910653145e-06,
+      "loss": 1.0757,
+      "step": 8750
+    },
+    {
+      "epoch": 0.06341071467349997,
+      "grad_norm": 0.20306764543056488,
+      "learning_rate": 4.936596523992559e-06,
+      "loss": 1.091,
+      "step": 8760
+    },
+    {
+      "epoch": 0.06348310133408615,
+      "grad_norm": 0.16739881038665771,
+      "learning_rate": 4.936524137331972e-06,
+      "loss": 1.0796,
+      "step": 8770
+    },
+    {
+      "epoch": 0.06355548799467234,
+      "grad_norm": 0.18593810498714447,
+      "learning_rate": 4.936451750671387e-06,
+      "loss": 1.0916,
+      "step": 8780
+    },
+    {
+      "epoch": 0.06362787465525853,
+      "grad_norm": 0.17287929356098175,
+      "learning_rate": 4.9363793640108005e-06,
+      "loss": 1.0747,
+      "step": 8790
+    },
+    {
+      "epoch": 0.06370026131584472,
+      "grad_norm": 0.19802750647068024,
+      "learning_rate": 4.936306977350214e-06,
+      "loss": 1.1005,
+      "step": 8800
+    },
+    {
+      "epoch": 0.06377264797643091,
+      "grad_norm": 0.17878587543964386,
+      "learning_rate": 4.936234590689628e-06,
+      "loss": 1.075,
+      "step": 8810
+    },
+    {
+      "epoch": 0.06384503463701709,
+      "grad_norm": 0.1824096143245697,
+      "learning_rate": 4.936162204029042e-06,
+      "loss": 1.0706,
+      "step": 8820
+    },
+    {
+      "epoch": 0.06391742129760328,
+      "grad_norm": 0.20461733639240265,
+      "learning_rate": 4.936089817368456e-06,
+      "loss": 1.0816,
+      "step": 8830
+    },
+    {
+      "epoch": 0.06398980795818947,
+      "grad_norm": 0.17579731345176697,
+      "learning_rate": 4.936017430707869e-06,
+      "loss": 1.0805,
+      "step": 8840
+    },
+    {
+      "epoch": 0.06406219461877566,
+      "grad_norm": 0.18281228840351105,
+      "learning_rate": 4.935945044047283e-06,
+      "loss": 1.0857,
+      "step": 8850
+    },
+    {
+      "epoch": 0.06413458127936184,
+      "grad_norm": 0.21296283602714539,
+      "learning_rate": 4.9358726573866975e-06,
+      "loss": 1.098,
+      "step": 8860
+    },
+    {
+      "epoch": 0.06420696793994803,
+      "grad_norm": 0.28200826048851013,
+      "learning_rate": 4.935800270726111e-06,
+      "loss": 1.0917,
+      "step": 8870
+    },
+    {
+      "epoch": 0.06427935460053422,
+      "grad_norm": 0.1753077208995819,
+      "learning_rate": 4.935727884065525e-06,
+      "loss": 1.0908,
+      "step": 8880
+    },
+    {
+      "epoch": 0.06435174126112041,
+      "grad_norm": 0.17760993540287018,
+      "learning_rate": 4.935655497404938e-06,
+      "loss": 1.0828,
+      "step": 8890
+    },
+    {
+      "epoch": 0.06442412792170658,
+      "grad_norm": 0.1909879595041275,
+      "learning_rate": 4.935583110744353e-06,
+      "loss": 1.0867,
+      "step": 8900
+    },
+    {
+      "epoch": 0.06449651458229277,
+      "grad_norm": 0.19046929478645325,
+      "learning_rate": 4.9355107240837664e-06,
+      "loss": 1.0913,
+      "step": 8910
+    },
+    {
+      "epoch": 0.06456890124287896,
+      "grad_norm": 0.19228294491767883,
+      "learning_rate": 4.93543833742318e-06,
+      "loss": 1.0848,
+      "step": 8920
+    },
+    {
+      "epoch": 0.06464128790346516,
+      "grad_norm": 0.1879161149263382,
+      "learning_rate": 4.935365950762594e-06,
+      "loss": 1.0996,
+      "step": 8930
+    },
+    {
+      "epoch": 0.06471367456405133,
+      "grad_norm": 0.18242213129997253,
+      "learning_rate": 4.935293564102008e-06,
+      "loss": 1.0873,
+      "step": 8940
+    },
+    {
+      "epoch": 0.06478606122463752,
+      "grad_norm": 0.19306018948554993,
+      "learning_rate": 4.935221177441422e-06,
+      "loss": 1.092,
+      "step": 8950
+    },
+    {
+      "epoch": 0.06485844788522371,
+      "grad_norm": 0.18336234986782074,
+      "learning_rate": 4.935148790780835e-06,
+      "loss": 1.1006,
+      "step": 8960
+    },
+    {
+      "epoch": 0.0649308345458099,
+      "grad_norm": 0.1992163509130478,
+      "learning_rate": 4.935076404120249e-06,
+      "loss": 1.0969,
+      "step": 8970
+    },
+    {
+      "epoch": 0.06500322120639608,
+      "grad_norm": 0.187247171998024,
+      "learning_rate": 4.9350040174596634e-06,
+      "loss": 1.0934,
+      "step": 8980
+    },
+    {
+      "epoch": 0.06507560786698227,
+      "grad_norm": 0.35581186413764954,
+      "learning_rate": 4.934931630799077e-06,
+      "loss": 1.0897,
+      "step": 8990
+    },
+    {
+      "epoch": 0.06514799452756846,
+      "grad_norm": 0.18312859535217285,
+      "learning_rate": 4.934859244138491e-06,
+      "loss": 1.0841,
+      "step": 9000
+    },
+    {
+      "epoch": 0.06522038118815465,
+      "grad_norm": 0.18880075216293335,
+      "learning_rate": 4.934786857477904e-06,
+      "loss": 1.0709,
+      "step": 9010
+    },
+    {
+      "epoch": 0.06529276784874083,
+      "grad_norm": 0.1736619919538498,
+      "learning_rate": 4.934714470817319e-06,
+      "loss": 1.0891,
+      "step": 9020
+    },
+    {
+      "epoch": 0.06536515450932702,
+      "grad_norm": 0.1821936070919037,
+      "learning_rate": 4.934642084156732e-06,
+      "loss": 1.0867,
+      "step": 9030
+    },
+    {
+      "epoch": 0.06543754116991321,
+      "grad_norm": 0.16732348501682281,
+      "learning_rate": 4.934569697496146e-06,
+      "loss": 1.0913,
+      "step": 9040
+    },
+    {
+      "epoch": 0.0655099278304994,
+      "grad_norm": 0.17859619855880737,
+      "learning_rate": 4.93449731083556e-06,
+      "loss": 1.0942,
+      "step": 9050
+    },
+    {
+      "epoch": 0.06558231449108558,
+      "grad_norm": 0.1915702074766159,
+      "learning_rate": 4.934424924174974e-06,
+      "loss": 1.0917,
+      "step": 9060
+    },
+    {
+      "epoch": 0.06565470115167177,
+      "grad_norm": 0.17785701155662537,
+      "learning_rate": 4.934352537514388e-06,
+      "loss": 1.0749,
+      "step": 9070
+    },
+    {
+      "epoch": 0.06572708781225796,
+      "grad_norm": 0.17183399200439453,
+      "learning_rate": 4.934280150853801e-06,
+      "loss": 1.0896,
+      "step": 9080
+    },
+    {
+      "epoch": 0.06579947447284415,
+      "grad_norm": 0.16996833682060242,
+      "learning_rate": 4.934207764193215e-06,
+      "loss": 1.0842,
+      "step": 9090
+    },
+    {
+      "epoch": 0.06587186113343033,
+      "grad_norm": 0.17851723730564117,
+      "learning_rate": 4.934135377532629e-06,
+      "loss": 1.0824,
+      "step": 9100
+    },
+    {
+      "epoch": 0.06594424779401652,
+      "grad_norm": 0.18523283302783966,
+      "learning_rate": 4.934062990872043e-06,
+      "loss": 1.081,
+      "step": 9110
+    },
+    {
+      "epoch": 0.06601663445460271,
+      "grad_norm": 0.1726391464471817,
+      "learning_rate": 4.933990604211457e-06,
+      "loss": 1.0866,
+      "step": 9120
+    },
+    {
+      "epoch": 0.0660890211151889,
+      "grad_norm": 0.18292652070522308,
+      "learning_rate": 4.93391821755087e-06,
+      "loss": 1.0933,
+      "step": 9130
+    },
+    {
+      "epoch": 0.06616140777577507,
+      "grad_norm": 0.18601487576961517,
+      "learning_rate": 4.933845830890284e-06,
+      "loss": 1.0834,
+      "step": 9140
+    },
+    {
+      "epoch": 0.06623379443636127,
+      "grad_norm": 0.16572198271751404,
+      "learning_rate": 4.9337734442296974e-06,
+      "loss": 1.0698,
+      "step": 9150
+    },
+    {
+      "epoch": 0.06630618109694746,
+      "grad_norm": 0.18648765981197357,
+      "learning_rate": 4.933701057569111e-06,
+      "loss": 1.0965,
+      "step": 9160
+    },
+    {
+      "epoch": 0.06637856775753365,
+      "grad_norm": 0.1860000491142273,
+      "learning_rate": 4.9336286709085255e-06,
+      "loss": 1.0782,
+      "step": 9170
+    },
+    {
+      "epoch": 0.06645095441811982,
+      "grad_norm": 0.2106804996728897,
+      "learning_rate": 4.933556284247939e-06,
+      "loss": 1.0899,
+      "step": 9180
+    },
+    {
+      "epoch": 0.06652334107870601,
+      "grad_norm": 0.16858477890491486,
+      "learning_rate": 4.933483897587353e-06,
+      "loss": 1.0902,
+      "step": 9190
+    },
+    {
+      "epoch": 0.0665957277392922,
+      "grad_norm": 0.17290134727954865,
+      "learning_rate": 4.933411510926766e-06,
+      "loss": 1.0991,
+      "step": 9200
+    },
+    {
+      "epoch": 0.0666681143998784,
+      "grad_norm": 0.18629100918769836,
+      "learning_rate": 4.933339124266181e-06,
+      "loss": 1.0849,
+      "step": 9210
+    },
+    {
+      "epoch": 0.06674050106046457,
+      "grad_norm": 0.18276071548461914,
+      "learning_rate": 4.9332667376055945e-06,
+      "loss": 1.08,
+      "step": 9220
+    },
+    {
+      "epoch": 0.06681288772105076,
+      "grad_norm": 0.2082335501909256,
+      "learning_rate": 4.933194350945008e-06,
+      "loss": 1.0747,
+      "step": 9230
+    },
+    {
+      "epoch": 0.06688527438163695,
+      "grad_norm": 0.20962758362293243,
+      "learning_rate": 4.933121964284422e-06,
+      "loss": 1.0784,
+      "step": 9240
+    },
+    {
+      "epoch": 0.06695766104222314,
+      "grad_norm": 0.20550638437271118,
+      "learning_rate": 4.933049577623836e-06,
+      "loss": 1.0856,
+      "step": 9250
+    },
+    {
+      "epoch": 0.06703004770280932,
+      "grad_norm": 0.17833620309829712,
+      "learning_rate": 4.93297719096325e-06,
+      "loss": 1.0797,
+      "step": 9260
+    },
+    {
+      "epoch": 0.06710243436339551,
+      "grad_norm": 0.17725925147533417,
+      "learning_rate": 4.932904804302663e-06,
+      "loss": 1.0872,
+      "step": 9270
+    },
+    {
+      "epoch": 0.0671748210239817,
+      "grad_norm": 0.1763269156217575,
+      "learning_rate": 4.932832417642077e-06,
+      "loss": 1.0956,
+      "step": 9280
+    },
+    {
+      "epoch": 0.06724720768456789,
+      "grad_norm": 0.17126107215881348,
+      "learning_rate": 4.932760030981491e-06,
+      "loss": 1.0727,
+      "step": 9290
+    },
+    {
+      "epoch": 0.06731959434515407,
+      "grad_norm": 0.16544747352600098,
+      "learning_rate": 4.932687644320905e-06,
+      "loss": 1.0782,
+      "step": 9300
+    },
+    {
+      "epoch": 0.06739198100574026,
+      "grad_norm": 0.175096794962883,
+      "learning_rate": 4.932615257660319e-06,
+      "loss": 1.0884,
+      "step": 9310
+    },
+    {
+      "epoch": 0.06746436766632645,
+      "grad_norm": 0.1752748042345047,
+      "learning_rate": 4.932542870999732e-06,
+      "loss": 1.0963,
+      "step": 9320
+    },
+    {
+      "epoch": 0.06753675432691264,
+      "grad_norm": 0.17110379040241241,
+      "learning_rate": 4.932470484339146e-06,
+      "loss": 1.1028,
+      "step": 9330
+    },
+    {
+      "epoch": 0.06760914098749882,
+      "grad_norm": 0.20907478034496307,
+      "learning_rate": 4.93239809767856e-06,
+      "loss": 1.0764,
+      "step": 9340
+    },
+    {
+      "epoch": 0.06768152764808501,
+      "grad_norm": 0.19506807625293732,
+      "learning_rate": 4.932325711017974e-06,
+      "loss": 1.065,
+      "step": 9350
+    },
+    {
+      "epoch": 0.0677539143086712,
+      "grad_norm": 0.1756713092327118,
+      "learning_rate": 4.932253324357388e-06,
+      "loss": 1.1012,
+      "step": 9360
+    },
+    {
+      "epoch": 0.06782630096925739,
+      "grad_norm": 0.18884608149528503,
+      "learning_rate": 4.932180937696801e-06,
+      "loss": 1.0739,
+      "step": 9370
+    },
+    {
+      "epoch": 0.06789868762984358,
+      "grad_norm": 0.19393788278102875,
+      "learning_rate": 4.932108551036216e-06,
+      "loss": 1.0933,
+      "step": 9380
+    },
+    {
+      "epoch": 0.06797107429042976,
+      "grad_norm": 0.18844221532344818,
+      "learning_rate": 4.932036164375629e-06,
+      "loss": 1.0796,
+      "step": 9390
+    },
+    {
+      "epoch": 0.06804346095101595,
+      "grad_norm": 0.17534899711608887,
+      "learning_rate": 4.931963777715043e-06,
+      "loss": 1.0807,
+      "step": 9400
+    },
+    {
+      "epoch": 0.06811584761160214,
+      "grad_norm": 0.1886800080537796,
+      "learning_rate": 4.9318913910544566e-06,
+      "loss": 1.0834,
+      "step": 9410
+    },
+    {
+      "epoch": 0.06818823427218833,
+      "grad_norm": 0.20587556064128876,
+      "learning_rate": 4.931819004393871e-06,
+      "loss": 1.1054,
+      "step": 9420
+    },
+    {
+      "epoch": 0.0682606209327745,
+      "grad_norm": 0.18019132316112518,
+      "learning_rate": 4.931746617733285e-06,
+      "loss": 1.076,
+      "step": 9430
+    },
+    {
+      "epoch": 0.0683330075933607,
+      "grad_norm": 0.18654106557369232,
+      "learning_rate": 4.931674231072698e-06,
+      "loss": 1.0768,
+      "step": 9440
+    },
+    {
+      "epoch": 0.06840539425394689,
+      "grad_norm": 0.1790568232536316,
+      "learning_rate": 4.931601844412112e-06,
+      "loss": 1.0751,
+      "step": 9450
+    },
+    {
+      "epoch": 0.06847778091453308,
+      "grad_norm": 0.17355284094810486,
+      "learning_rate": 4.931529457751526e-06,
+      "loss": 1.0755,
+      "step": 9460
+    },
+    {
+      "epoch": 0.06855016757511925,
+      "grad_norm": 0.18090367317199707,
+      "learning_rate": 4.93145707109094e-06,
+      "loss": 1.0927,
+      "step": 9470
+    },
+    {
+      "epoch": 0.06862255423570544,
+      "grad_norm": 0.18429531157016754,
+      "learning_rate": 4.9313846844303536e-06,
+      "loss": 1.0857,
+      "step": 9480
+    },
+    {
+      "epoch": 0.06869494089629163,
+      "grad_norm": 0.16996966302394867,
+      "learning_rate": 4.931312297769767e-06,
+      "loss": 1.0955,
+      "step": 9490
+    },
+    {
+      "epoch": 0.06876732755687782,
+      "grad_norm": 0.17073926329612732,
+      "learning_rate": 4.931239911109182e-06,
+      "loss": 1.0934,
+      "step": 9500
+    },
+    {
+      "epoch": 0.068839714217464,
+      "grad_norm": 0.19295451045036316,
+      "learning_rate": 4.931167524448595e-06,
+      "loss": 1.0843,
+      "step": 9510
+    },
+    {
+      "epoch": 0.06891210087805019,
+      "grad_norm": 0.19370391964912415,
+      "learning_rate": 4.931095137788009e-06,
+      "loss": 1.0939,
+      "step": 9520
+    },
+    {
+      "epoch": 0.06898448753863638,
+      "grad_norm": 0.18708017468452454,
+      "learning_rate": 4.9310227511274225e-06,
+      "loss": 1.0761,
+      "step": 9530
+    },
+    {
+      "epoch": 0.06905687419922257,
+      "grad_norm": 0.16977204382419586,
+      "learning_rate": 4.930950364466837e-06,
+      "loss": 1.0833,
+      "step": 9540
+    },
+    {
+      "epoch": 0.06912926085980875,
+      "grad_norm": 0.1928476095199585,
+      "learning_rate": 4.9308779778062506e-06,
+      "loss": 1.0773,
+      "step": 9550
+    },
+    {
+      "epoch": 0.06920164752039494,
+      "grad_norm": 0.19466859102249146,
+      "learning_rate": 4.930805591145664e-06,
+      "loss": 1.07,
+      "step": 9560
+    },
+    {
+      "epoch": 0.06927403418098113,
+      "grad_norm": 0.18574132025241852,
+      "learning_rate": 4.930733204485078e-06,
+      "loss": 1.0887,
+      "step": 9570
+    },
+    {
+      "epoch": 0.06934642084156732,
+      "grad_norm": 0.17645835876464844,
+      "learning_rate": 4.930660817824492e-06,
+      "loss": 1.0841,
+      "step": 9580
+    },
+    {
+      "epoch": 0.0694188075021535,
+      "grad_norm": 0.17621047794818878,
+      "learning_rate": 4.930588431163906e-06,
+      "loss": 1.0909,
+      "step": 9590
+    },
+    {
+      "epoch": 0.06949119416273969,
+      "grad_norm": 0.1808473765850067,
+      "learning_rate": 4.9305160445033195e-06,
+      "loss": 1.0885,
+      "step": 9600
+    },
+    {
+      "epoch": 0.06956358082332588,
+      "grad_norm": 0.1906941682100296,
+      "learning_rate": 4.930443657842733e-06,
+      "loss": 1.0699,
+      "step": 9610
+    },
+    {
+      "epoch": 0.06963596748391207,
+      "grad_norm": 0.18440672755241394,
+      "learning_rate": 4.9303712711821476e-06,
+      "loss": 1.0753,
+      "step": 9620
+    },
+    {
+      "epoch": 0.06970835414449825,
+      "grad_norm": 0.31207403540611267,
+      "learning_rate": 4.930298884521561e-06,
+      "loss": 1.0821,
+      "step": 9630
+    },
+    {
+      "epoch": 0.06978074080508444,
+      "grad_norm": 0.178600013256073,
+      "learning_rate": 4.930226497860975e-06,
+      "loss": 1.0827,
+      "step": 9640
+    },
+    {
+      "epoch": 0.06985312746567063,
+      "grad_norm": 0.1899259239435196,
+      "learning_rate": 4.930154111200388e-06,
+      "loss": 1.0673,
+      "step": 9650
+    },
+    {
+      "epoch": 0.06992551412625682,
+      "grad_norm": 0.1868361085653305,
+      "learning_rate": 4.930081724539802e-06,
+      "loss": 1.0838,
+      "step": 9660
+    },
+    {
+      "epoch": 0.069997900786843,
+      "grad_norm": 0.19484713673591614,
+      "learning_rate": 4.930009337879216e-06,
+      "loss": 1.0806,
+      "step": 9670
+    },
+    {
+      "epoch": 0.07007028744742919,
+      "grad_norm": 0.1849289834499359,
+      "learning_rate": 4.929936951218629e-06,
+      "loss": 1.0847,
+      "step": 9680
+    },
+    {
+      "epoch": 0.07014267410801538,
+      "grad_norm": 0.1799035519361496,
+      "learning_rate": 4.929864564558044e-06,
+      "loss": 1.0791,
+      "step": 9690
+    },
+    {
+      "epoch": 0.07021506076860157,
+      "grad_norm": 0.19453303515911102,
+      "learning_rate": 4.929792177897457e-06,
+      "loss": 1.0923,
+      "step": 9700
+    },
+    {
+      "epoch": 0.07028744742918774,
+      "grad_norm": 0.20945830643177032,
+      "learning_rate": 4.929719791236871e-06,
+      "loss": 1.1022,
+      "step": 9710
+    },
+    {
+      "epoch": 0.07035983408977393,
+      "grad_norm": 0.1825156956911087,
+      "learning_rate": 4.929647404576285e-06,
+      "loss": 1.0819,
+      "step": 9720
+    },
+    {
+      "epoch": 0.07043222075036012,
+      "grad_norm": 0.18250644207000732,
+      "learning_rate": 4.929575017915699e-06,
+      "loss": 1.0681,
+      "step": 9730
+    },
+    {
+      "epoch": 0.07050460741094632,
+      "grad_norm": 0.20941105484962463,
+      "learning_rate": 4.929502631255113e-06,
+      "loss": 1.0651,
+      "step": 9740
+    },
+    {
+      "epoch": 0.07057699407153249,
+      "grad_norm": 0.1895821988582611,
+      "learning_rate": 4.929430244594526e-06,
+      "loss": 1.0919,
+      "step": 9750
+    },
+    {
+      "epoch": 0.07064938073211868,
+      "grad_norm": 0.16658397018909454,
+      "learning_rate": 4.92935785793394e-06,
+      "loss": 1.0787,
+      "step": 9760
+    },
+    {
+      "epoch": 0.07072176739270487,
+      "grad_norm": 0.19257685542106628,
+      "learning_rate": 4.929285471273354e-06,
+      "loss": 1.0788,
+      "step": 9770
+    },
+    {
+      "epoch": 0.07079415405329106,
+      "grad_norm": 0.17767976224422455,
+      "learning_rate": 4.929213084612768e-06,
+      "loss": 1.0762,
+      "step": 9780
+    },
+    {
+      "epoch": 0.07086654071387724,
+      "grad_norm": 0.2234843373298645,
+      "learning_rate": 4.929140697952182e-06,
+      "loss": 1.0751,
+      "step": 9790
+    },
+    {
+      "epoch": 0.07093892737446343,
+      "grad_norm": 0.2335948348045349,
+      "learning_rate": 4.929068311291595e-06,
+      "loss": 1.0819,
+      "step": 9800
+    },
+    {
+      "epoch": 0.07101131403504962,
+      "grad_norm": 0.1768556535243988,
+      "learning_rate": 4.92899592463101e-06,
+      "loss": 1.0837,
+      "step": 9810
+    },
+    {
+      "epoch": 0.07108370069563581,
+      "grad_norm": 0.17308947443962097,
+      "learning_rate": 4.928923537970423e-06,
+      "loss": 1.074,
+      "step": 9820
+    },
+    {
+      "epoch": 0.07115608735622199,
+      "grad_norm": 0.18633995950222015,
+      "learning_rate": 4.928851151309837e-06,
+      "loss": 1.0755,
+      "step": 9830
+    },
+    {
+      "epoch": 0.07122847401680818,
+      "grad_norm": 0.19569677114486694,
+      "learning_rate": 4.9287787646492505e-06,
+      "loss": 1.0815,
+      "step": 9840
+    },
+    {
+      "epoch": 0.07130086067739437,
+      "grad_norm": 0.19831666350364685,
+      "learning_rate": 4.928706377988665e-06,
+      "loss": 1.084,
+      "step": 9850
+    },
+    {
+      "epoch": 0.07137324733798056,
+      "grad_norm": 0.18258638679981232,
+      "learning_rate": 4.928633991328079e-06,
+      "loss": 1.0666,
+      "step": 9860
+    },
+    {
+      "epoch": 0.07144563399856674,
+      "grad_norm": 0.16318519413471222,
+      "learning_rate": 4.928561604667492e-06,
+      "loss": 1.0754,
+      "step": 9870
+    },
+    {
+      "epoch": 0.07151802065915293,
+      "grad_norm": 0.17054226994514465,
+      "learning_rate": 4.928489218006906e-06,
+      "loss": 1.0815,
+      "step": 9880
+    },
+    {
+      "epoch": 0.07159040731973912,
+      "grad_norm": 0.1713060736656189,
+      "learning_rate": 4.92841683134632e-06,
+      "loss": 1.0822,
+      "step": 9890
+    },
+    {
+      "epoch": 0.07166279398032531,
+      "grad_norm": 0.17956681549549103,
+      "learning_rate": 4.928344444685734e-06,
+      "loss": 1.0747,
+      "step": 9900
+    },
+    {
+      "epoch": 0.07173518064091149,
+      "grad_norm": 0.19176030158996582,
+      "learning_rate": 4.9282720580251475e-06,
+      "loss": 1.074,
+      "step": 9910
+    },
+    {
+      "epoch": 0.07180756730149768,
+      "grad_norm": 0.17614677548408508,
+      "learning_rate": 4.928199671364561e-06,
+      "loss": 1.0894,
+      "step": 9920
+    },
+    {
+      "epoch": 0.07187995396208387,
+      "grad_norm": 0.1864258348941803,
+      "learning_rate": 4.928127284703975e-06,
+      "loss": 1.0826,
+      "step": 9930
+    },
+    {
+      "epoch": 0.07195234062267006,
+      "grad_norm": 0.18494221568107605,
+      "learning_rate": 4.928054898043389e-06,
+      "loss": 1.0613,
+      "step": 9940
+    },
+    {
+      "epoch": 0.07202472728325623,
+      "grad_norm": 0.17159555852413177,
+      "learning_rate": 4.927982511382803e-06,
+      "loss": 1.073,
+      "step": 9950
+    },
+    {
+      "epoch": 0.07209711394384243,
+      "grad_norm": 0.18387450277805328,
+      "learning_rate": 4.9279101247222165e-06,
+      "loss": 1.0729,
+      "step": 9960
+    },
+    {
+      "epoch": 0.07216950060442862,
+      "grad_norm": 0.1786179393529892,
+      "learning_rate": 4.92783773806163e-06,
+      "loss": 1.0867,
+      "step": 9970
+    },
+    {
+      "epoch": 0.0722418872650148,
+      "grad_norm": 0.1791893094778061,
+      "learning_rate": 4.9277653514010445e-06,
+      "loss": 1.0783,
+      "step": 9980
+    },
+    {
+      "epoch": 0.072314273925601,
+      "grad_norm": 0.18308618664741516,
+      "learning_rate": 4.927692964740458e-06,
+      "loss": 1.0667,
+      "step": 9990
+    },
+    {
+      "epoch": 0.07238666058618717,
+      "grad_norm": 0.18215444684028625,
+      "learning_rate": 4.927620578079872e-06,
+      "loss": 1.0804,
+      "step": 10000
+    },
+    {
+      "epoch": 0.07245904724677336,
+      "grad_norm": 0.1691853404045105,
+      "learning_rate": 4.927548191419285e-06,
+      "loss": 1.0837,
+      "step": 10010
+    },
+    {
+      "epoch": 0.07253143390735955,
+      "grad_norm": 0.1819877028465271,
+      "learning_rate": 4.9274758047587e-06,
+      "loss": 1.0799,
+      "step": 10020
+    },
+    {
+      "epoch": 0.07260382056794575,
+      "grad_norm": 0.17936040461063385,
+      "learning_rate": 4.9274034180981135e-06,
+      "loss": 1.0675,
+      "step": 10030
+    },
+    {
+      "epoch": 0.07267620722853192,
+      "grad_norm": 0.17777326703071594,
+      "learning_rate": 4.927331031437527e-06,
+      "loss": 1.0674,
+      "step": 10040
+    },
+    {
+      "epoch": 0.07274859388911811,
+      "grad_norm": 0.22706569731235504,
+      "learning_rate": 4.927258644776941e-06,
+      "loss": 1.0901,
+      "step": 10050
+    },
+    {
+      "epoch": 0.0728209805497043,
+      "grad_norm": 0.17453064024448395,
+      "learning_rate": 4.927186258116355e-06,
+      "loss": 1.0665,
+      "step": 10060
+    },
+    {
+      "epoch": 0.0728933672102905,
+      "grad_norm": 0.1731829047203064,
+      "learning_rate": 4.927113871455769e-06,
+      "loss": 1.0734,
+      "step": 10070
+    },
+    {
+      "epoch": 0.07296575387087667,
+      "grad_norm": 0.1648159921169281,
+      "learning_rate": 4.927041484795182e-06,
+      "loss": 1.0762,
+      "step": 10080
+    },
+    {
+      "epoch": 0.07303814053146286,
+      "grad_norm": 0.19320763647556305,
+      "learning_rate": 4.926969098134596e-06,
+      "loss": 1.0672,
+      "step": 10090
+    },
+    {
+      "epoch": 0.07311052719204905,
+      "grad_norm": 0.17869828641414642,
+      "learning_rate": 4.9268967114740105e-06,
+      "loss": 1.0748,
+      "step": 10100
+    },
+    {
+      "epoch": 0.07318291385263524,
+      "grad_norm": 0.17317935824394226,
+      "learning_rate": 4.926824324813424e-06,
+      "loss": 1.0725,
+      "step": 10110
+    },
+    {
+      "epoch": 0.07325530051322142,
+      "grad_norm": 0.18840764462947845,
+      "learning_rate": 4.926751938152838e-06,
+      "loss": 1.0902,
+      "step": 10120
+    },
+    {
+      "epoch": 0.07332768717380761,
+      "grad_norm": 0.18961584568023682,
+      "learning_rate": 4.926679551492251e-06,
+      "loss": 1.0677,
+      "step": 10130
+    },
+    {
+      "epoch": 0.0734000738343938,
+      "grad_norm": 0.1739116609096527,
+      "learning_rate": 4.926607164831666e-06,
+      "loss": 1.0744,
+      "step": 10140
+    },
+    {
+      "epoch": 0.07347246049497999,
+      "grad_norm": 0.19770686328411102,
+      "learning_rate": 4.926534778171079e-06,
+      "loss": 1.0968,
+      "step": 10150
+    },
+    {
+      "epoch": 0.07354484715556617,
+      "grad_norm": 0.17552542686462402,
+      "learning_rate": 4.926462391510493e-06,
+      "loss": 1.0851,
+      "step": 10160
+    },
+    {
+      "epoch": 0.07361723381615236,
+      "grad_norm": 0.18830542266368866,
+      "learning_rate": 4.926390004849907e-06,
+      "loss": 1.0855,
+      "step": 10170
+    },
+    {
+      "epoch": 0.07368962047673855,
+      "grad_norm": 0.2055501937866211,
+      "learning_rate": 4.926317618189321e-06,
+      "loss": 1.0803,
+      "step": 10180
+    },
+    {
+      "epoch": 0.07376200713732474,
+      "grad_norm": 0.18813519179821014,
+      "learning_rate": 4.926245231528734e-06,
+      "loss": 1.071,
+      "step": 10190
+    },
+    {
+      "epoch": 0.07383439379791092,
+      "grad_norm": 0.2169976532459259,
+      "learning_rate": 4.9261728448681475e-06,
+      "loss": 1.0872,
+      "step": 10200
+    },
+    {
+      "epoch": 0.0739067804584971,
+      "grad_norm": 0.2091887891292572,
+      "learning_rate": 4.926100458207562e-06,
+      "loss": 1.0811,
+      "step": 10210
+    },
+    {
+      "epoch": 0.0739791671190833,
+      "grad_norm": 0.17298966646194458,
+      "learning_rate": 4.9260280715469756e-06,
+      "loss": 1.0846,
+      "step": 10220
+    },
+    {
+      "epoch": 0.07405155377966949,
+      "grad_norm": 0.19214802980422974,
+      "learning_rate": 4.925955684886389e-06,
+      "loss": 1.0804,
+      "step": 10230
+    },
+    {
+      "epoch": 0.07412394044025566,
+      "grad_norm": 0.18159951269626617,
+      "learning_rate": 4.925883298225803e-06,
+      "loss": 1.0818,
+      "step": 10240
+    },
+    {
+      "epoch": 0.07419632710084186,
+      "grad_norm": 0.2073182910680771,
+      "learning_rate": 4.925810911565217e-06,
+      "loss": 1.0717,
+      "step": 10250
+    },
+    {
+      "epoch": 0.07426871376142805,
+      "grad_norm": 0.17683623731136322,
+      "learning_rate": 4.925738524904631e-06,
+      "loss": 1.0691,
+      "step": 10260
+    },
+    {
+      "epoch": 0.07434110042201424,
+      "grad_norm": 0.17853355407714844,
+      "learning_rate": 4.9256661382440445e-06,
+      "loss": 1.0807,
+      "step": 10270
+    },
+    {
+      "epoch": 0.07441348708260041,
+      "grad_norm": 0.1861315667629242,
+      "learning_rate": 4.925593751583458e-06,
+      "loss": 1.075,
+      "step": 10280
+    },
+    {
+      "epoch": 0.0744858737431866,
+      "grad_norm": 0.22674742341041565,
+      "learning_rate": 4.9255213649228726e-06,
+      "loss": 1.0824,
+      "step": 10290
+    },
+    {
+      "epoch": 0.0745582604037728,
+      "grad_norm": 0.1941600888967514,
+      "learning_rate": 4.925448978262286e-06,
+      "loss": 1.0663,
+      "step": 10300
+    },
+    {
+      "epoch": 0.07463064706435898,
+      "grad_norm": 0.1759897917509079,
+      "learning_rate": 4.9253765916017e-06,
+      "loss": 1.0761,
+      "step": 10310
+    },
+    {
+      "epoch": 0.07470303372494516,
+      "grad_norm": 0.1744421124458313,
+      "learning_rate": 4.925304204941113e-06,
+      "loss": 1.0705,
+      "step": 10320
+    },
+    {
+      "epoch": 0.07477542038553135,
+      "grad_norm": 0.1926373541355133,
+      "learning_rate": 4.925231818280528e-06,
+      "loss": 1.0714,
+      "step": 10330
+    },
+    {
+      "epoch": 0.07484780704611754,
+      "grad_norm": 0.19619183242321014,
+      "learning_rate": 4.9251594316199415e-06,
+      "loss": 1.0728,
+      "step": 10340
+    },
+    {
+      "epoch": 0.07492019370670373,
+      "grad_norm": 0.23807357251644135,
+      "learning_rate": 4.925087044959355e-06,
+      "loss": 1.0772,
+      "step": 10350
+    },
+    {
+      "epoch": 0.07499258036728991,
+      "grad_norm": 0.1988050788640976,
+      "learning_rate": 4.925014658298769e-06,
+      "loss": 1.0829,
+      "step": 10360
+    },
+    {
+      "epoch": 0.0750649670278761,
+      "grad_norm": 0.16706973314285278,
+      "learning_rate": 4.924942271638183e-06,
+      "loss": 1.0642,
+      "step": 10370
+    },
+    {
+      "epoch": 0.07513735368846229,
+      "grad_norm": 0.19122160971164703,
+      "learning_rate": 4.924869884977597e-06,
+      "loss": 1.0973,
+      "step": 10380
+    },
+    {
+      "epoch": 0.07520974034904848,
+      "grad_norm": 0.1846979856491089,
+      "learning_rate": 4.92479749831701e-06,
+      "loss": 1.0775,
+      "step": 10390
+    },
+    {
+      "epoch": 0.07528212700963466,
+      "grad_norm": 0.17939844727516174,
+      "learning_rate": 4.924725111656424e-06,
+      "loss": 1.0644,
+      "step": 10400
+    },
+    {
+      "epoch": 0.07535451367022085,
+      "grad_norm": 0.17780403792858124,
+      "learning_rate": 4.9246527249958385e-06,
+      "loss": 1.0655,
+      "step": 10410
+    },
+    {
+      "epoch": 0.07542690033080704,
+      "grad_norm": 0.2274876832962036,
+      "learning_rate": 4.924580338335252e-06,
+      "loss": 1.0866,
+      "step": 10420
+    },
+    {
+      "epoch": 0.07549928699139323,
+      "grad_norm": 0.23748233914375305,
+      "learning_rate": 4.924507951674666e-06,
+      "loss": 1.0637,
+      "step": 10430
+    },
+    {
+      "epoch": 0.0755716736519794,
+      "grad_norm": 0.22824735939502716,
+      "learning_rate": 4.924435565014079e-06,
+      "loss": 1.0908,
+      "step": 10440
+    },
+    {
+      "epoch": 0.0756440603125656,
+      "grad_norm": 0.1775079220533371,
+      "learning_rate": 4.924363178353494e-06,
+      "loss": 1.0736,
+      "step": 10450
+    },
+    {
+      "epoch": 0.07571644697315179,
+      "grad_norm": 0.1669524759054184,
+      "learning_rate": 4.9242907916929074e-06,
+      "loss": 1.0772,
+      "step": 10460
+    },
+    {
+      "epoch": 0.07578883363373798,
+      "grad_norm": 0.19081354141235352,
+      "learning_rate": 4.924218405032321e-06,
+      "loss": 1.0668,
+      "step": 10470
+    },
+    {
+      "epoch": 0.07586122029432416,
+      "grad_norm": 0.18212305009365082,
+      "learning_rate": 4.924146018371735e-06,
+      "loss": 1.0824,
+      "step": 10480
+    },
+    {
+      "epoch": 0.07593360695491035,
+      "grad_norm": 0.19392240047454834,
+      "learning_rate": 4.924073631711149e-06,
+      "loss": 1.086,
+      "step": 10490
+    },
+    {
+      "epoch": 0.07600599361549654,
+      "grad_norm": 0.17266540229320526,
+      "learning_rate": 4.924001245050563e-06,
+      "loss": 1.0713,
+      "step": 10500
+    },
+    {
+      "epoch": 0.07607838027608273,
+      "grad_norm": 0.16641221940517426,
+      "learning_rate": 4.923928858389976e-06,
+      "loss": 1.0734,
+      "step": 10510
+    },
+    {
+      "epoch": 0.0761507669366689,
+      "grad_norm": 0.19562187790870667,
+      "learning_rate": 4.92385647172939e-06,
+      "loss": 1.0696,
+      "step": 10520
+    },
+    {
+      "epoch": 0.0762231535972551,
+      "grad_norm": 0.1799498200416565,
+      "learning_rate": 4.923784085068804e-06,
+      "loss": 1.0691,
+      "step": 10530
+    },
+    {
+      "epoch": 0.07629554025784129,
+      "grad_norm": 0.21454447507858276,
+      "learning_rate": 4.923711698408218e-06,
+      "loss": 1.0651,
+      "step": 10540
+    },
+    {
+      "epoch": 0.07636792691842748,
+      "grad_norm": 0.16438615322113037,
+      "learning_rate": 4.923639311747632e-06,
+      "loss": 1.0758,
+      "step": 10550
+    },
+    {
+      "epoch": 0.07644031357901367,
+      "grad_norm": 0.17331558465957642,
+      "learning_rate": 4.923566925087045e-06,
+      "loss": 1.092,
+      "step": 10560
+    },
+    {
+      "epoch": 0.07651270023959984,
+      "grad_norm": 0.17129570245742798,
+      "learning_rate": 4.923494538426459e-06,
+      "loss": 1.0644,
+      "step": 10570
+    },
+    {
+      "epoch": 0.07658508690018603,
+      "grad_norm": 0.17949286103248596,
+      "learning_rate": 4.923422151765873e-06,
+      "loss": 1.0765,
+      "step": 10580
+    },
+    {
+      "epoch": 0.07665747356077222,
+      "grad_norm": 0.20298872888088226,
+      "learning_rate": 4.923349765105287e-06,
+      "loss": 1.0813,
+      "step": 10590
+    },
+    {
+      "epoch": 0.07672986022135841,
+      "grad_norm": 0.17841701209545135,
+      "learning_rate": 4.923277378444701e-06,
+      "loss": 1.0763,
+      "step": 10600
+    },
+    {
+      "epoch": 0.07680224688194459,
+      "grad_norm": 0.1926995813846588,
+      "learning_rate": 4.923204991784114e-06,
+      "loss": 1.0666,
+      "step": 10610
+    },
+    {
+      "epoch": 0.07687463354253078,
+      "grad_norm": 0.20982502400875092,
+      "learning_rate": 4.923132605123529e-06,
+      "loss": 1.0674,
+      "step": 10620
+    },
+    {
+      "epoch": 0.07694702020311697,
+      "grad_norm": 0.18079523742198944,
+      "learning_rate": 4.923060218462942e-06,
+      "loss": 1.0782,
+      "step": 10630
+    },
+    {
+      "epoch": 0.07701940686370316,
+      "grad_norm": 0.18885524570941925,
+      "learning_rate": 4.922987831802356e-06,
+      "loss": 1.0675,
+      "step": 10640
+    },
+    {
+      "epoch": 0.07709179352428934,
+      "grad_norm": 0.1845076084136963,
+      "learning_rate": 4.9229154451417695e-06,
+      "loss": 1.0691,
+      "step": 10650
+    },
+    {
+      "epoch": 0.07716418018487553,
+      "grad_norm": 0.18656429648399353,
+      "learning_rate": 4.922843058481184e-06,
+      "loss": 1.0698,
+      "step": 10660
+    },
+    {
+      "epoch": 0.07723656684546172,
+      "grad_norm": 0.1843183934688568,
+      "learning_rate": 4.922770671820598e-06,
+      "loss": 1.0685,
+      "step": 10670
+    },
+    {
+      "epoch": 0.07730895350604791,
+      "grad_norm": 0.22134332358837128,
+      "learning_rate": 4.922698285160011e-06,
+      "loss": 1.0716,
+      "step": 10680
+    },
+    {
+      "epoch": 0.07738134016663409,
+      "grad_norm": 0.23194286227226257,
+      "learning_rate": 4.922625898499425e-06,
+      "loss": 1.0655,
+      "step": 10690
+    },
+    {
+      "epoch": 0.07745372682722028,
+      "grad_norm": 0.18523964285850525,
+      "learning_rate": 4.922553511838839e-06,
+      "loss": 1.0793,
+      "step": 10700
+    },
+    {
+      "epoch": 0.07752611348780647,
+      "grad_norm": 0.19442328810691833,
+      "learning_rate": 4.922481125178253e-06,
+      "loss": 1.0722,
+      "step": 10710
+    },
+    {
+      "epoch": 0.07759850014839266,
+      "grad_norm": 0.17840822041034698,
+      "learning_rate": 4.922408738517666e-06,
+      "loss": 1.0698,
+      "step": 10720
+    },
+    {
+      "epoch": 0.07767088680897884,
+      "grad_norm": 0.1624884307384491,
+      "learning_rate": 4.92233635185708e-06,
+      "loss": 1.0767,
+      "step": 10730
+    },
+    {
+      "epoch": 0.07774327346956503,
+      "grad_norm": 0.1825413703918457,
+      "learning_rate": 4.922263965196494e-06,
+      "loss": 1.0707,
+      "step": 10740
+    },
+    {
+      "epoch": 0.07781566013015122,
+      "grad_norm": 0.18537434935569763,
+      "learning_rate": 4.922191578535907e-06,
+      "loss": 1.0802,
+      "step": 10750
+    },
+    {
+      "epoch": 0.07788804679073741,
+      "grad_norm": 0.16724461317062378,
+      "learning_rate": 4.922119191875321e-06,
+      "loss": 1.0879,
+      "step": 10760
+    },
+    {
+      "epoch": 0.07796043345132359,
+      "grad_norm": 0.20884546637535095,
+      "learning_rate": 4.9220468052147355e-06,
+      "loss": 1.0533,
+      "step": 10770
+    },
+    {
+      "epoch": 0.07803282011190978,
+      "grad_norm": 0.19194482266902924,
+      "learning_rate": 4.921974418554149e-06,
+      "loss": 1.083,
+      "step": 10780
+    },
+    {
+      "epoch": 0.07810520677249597,
+      "grad_norm": 0.17105208337306976,
+      "learning_rate": 4.921902031893563e-06,
+      "loss": 1.0832,
+      "step": 10790
+    },
+    {
+      "epoch": 0.07817759343308216,
+      "grad_norm": 0.18788471817970276,
+      "learning_rate": 4.921829645232976e-06,
+      "loss": 1.0891,
+      "step": 10800
+    },
+    {
+      "epoch": 0.07824998009366833,
+      "grad_norm": 0.20326700806617737,
+      "learning_rate": 4.921757258572391e-06,
+      "loss": 1.0925,
+      "step": 10810
+    },
+    {
+      "epoch": 0.07832236675425452,
+      "grad_norm": 0.17056359350681305,
+      "learning_rate": 4.921684871911804e-06,
+      "loss": 1.06,
+      "step": 10820
+    },
+    {
+      "epoch": 0.07839475341484071,
+      "grad_norm": 0.17205092310905457,
+      "learning_rate": 4.921612485251218e-06,
+      "loss": 1.0682,
+      "step": 10830
+    },
+    {
+      "epoch": 0.0784671400754269,
+      "grad_norm": 0.17760831117630005,
+      "learning_rate": 4.921540098590632e-06,
+      "loss": 1.079,
+      "step": 10840
+    },
+    {
+      "epoch": 0.07853952673601308,
+      "grad_norm": 0.2105284333229065,
+      "learning_rate": 4.921467711930046e-06,
+      "loss": 1.0787,
+      "step": 10850
+    },
+    {
+      "epoch": 0.07861191339659927,
+      "grad_norm": 0.1877516359090805,
+      "learning_rate": 4.92139532526946e-06,
+      "loss": 1.0756,
+      "step": 10860
+    },
+    {
+      "epoch": 0.07868430005718546,
+      "grad_norm": 0.15930220484733582,
+      "learning_rate": 4.921322938608873e-06,
+      "loss": 1.0785,
+      "step": 10870
+    },
+    {
+      "epoch": 0.07875668671777165,
+      "grad_norm": 0.19607755541801453,
+      "learning_rate": 4.921250551948287e-06,
+      "loss": 1.0908,
+      "step": 10880
+    },
+    {
+      "epoch": 0.07882907337835783,
+      "grad_norm": 0.18164999783039093,
+      "learning_rate": 4.921178165287701e-06,
+      "loss": 1.049,
+      "step": 10890
+    },
+    {
+      "epoch": 0.07890146003894402,
+      "grad_norm": 0.17521744966506958,
+      "learning_rate": 4.921105778627115e-06,
+      "loss": 1.0801,
+      "step": 10900
+    },
+    {
+      "epoch": 0.07897384669953021,
+      "grad_norm": 0.1904889941215515,
+      "learning_rate": 4.921033391966529e-06,
+      "loss": 1.0697,
+      "step": 10910
+    },
+    {
+      "epoch": 0.0790462333601164,
+      "grad_norm": 0.1861046850681305,
+      "learning_rate": 4.920961005305942e-06,
+      "loss": 1.0619,
+      "step": 10920
+    },
+    {
+      "epoch": 0.07911862002070258,
+      "grad_norm": 0.18160459399223328,
+      "learning_rate": 4.920888618645357e-06,
+      "loss": 1.075,
+      "step": 10930
+    },
+    {
+      "epoch": 0.07919100668128877,
+      "grad_norm": 0.1644423007965088,
+      "learning_rate": 4.92081623198477e-06,
+      "loss": 1.0704,
+      "step": 10940
+    },
+    {
+      "epoch": 0.07926339334187496,
+      "grad_norm": 0.192913219332695,
+      "learning_rate": 4.920743845324184e-06,
+      "loss": 1.0905,
+      "step": 10950
+    },
+    {
+      "epoch": 0.07933578000246115,
+      "grad_norm": 0.18490082025527954,
+      "learning_rate": 4.9206714586635976e-06,
+      "loss": 1.0704,
+      "step": 10960
+    },
+    {
+      "epoch": 0.07940816666304733,
+      "grad_norm": 0.17639735341072083,
+      "learning_rate": 4.920599072003012e-06,
+      "loss": 1.0789,
+      "step": 10970
+    },
+    {
+      "epoch": 0.07948055332363352,
+      "grad_norm": 0.1718800812959671,
+      "learning_rate": 4.920526685342426e-06,
+      "loss": 1.0849,
+      "step": 10980
+    },
+    {
+      "epoch": 0.07955293998421971,
+      "grad_norm": 0.1801319122314453,
+      "learning_rate": 4.920454298681839e-06,
+      "loss": 1.0621,
+      "step": 10990
+    },
+    {
+      "epoch": 0.0796253266448059,
+      "grad_norm": 0.17438428103923798,
+      "learning_rate": 4.920381912021253e-06,
+      "loss": 1.0694,
+      "step": 11000
+    },
+    {
+      "epoch": 0.07969771330539208,
+      "grad_norm": 0.19011586904525757,
+      "learning_rate": 4.920309525360667e-06,
+      "loss": 1.0672,
+      "step": 11010
+    },
+    {
+      "epoch": 0.07977009996597827,
+      "grad_norm": 0.17366407811641693,
+      "learning_rate": 4.920237138700081e-06,
+      "loss": 1.0692,
+      "step": 11020
+    },
+    {
+      "epoch": 0.07984248662656446,
+      "grad_norm": 0.18017128109931946,
+      "learning_rate": 4.9201647520394946e-06,
+      "loss": 1.0839,
+      "step": 11030
+    },
+    {
+      "epoch": 0.07991487328715065,
+      "grad_norm": 0.18107180297374725,
+      "learning_rate": 4.920092365378908e-06,
+      "loss": 1.0789,
+      "step": 11040
+    },
+    {
+      "epoch": 0.07998725994773682,
+      "grad_norm": 0.1829778552055359,
+      "learning_rate": 4.920019978718323e-06,
+      "loss": 1.0716,
+      "step": 11050
+    },
+    {
+      "epoch": 0.08005964660832302,
+      "grad_norm": 0.17643770575523376,
+      "learning_rate": 4.919947592057736e-06,
+      "loss": 1.0778,
+      "step": 11060
+    },
+    {
+      "epoch": 0.0801320332689092,
+      "grad_norm": 0.17540425062179565,
+      "learning_rate": 4.91987520539715e-06,
+      "loss": 1.0685,
+      "step": 11070
+    },
+    {
+      "epoch": 0.0802044199294954,
+      "grad_norm": 0.18910111486911774,
+      "learning_rate": 4.9198028187365635e-06,
+      "loss": 1.0639,
+      "step": 11080
+    },
+    {
+      "epoch": 0.08027680659008157,
+      "grad_norm": 0.17153383791446686,
+      "learning_rate": 4.919730432075978e-06,
+      "loss": 1.0872,
+      "step": 11090
+    },
+    {
+      "epoch": 0.08034919325066776,
+      "grad_norm": 0.16545794904232025,
+      "learning_rate": 4.9196580454153916e-06,
+      "loss": 1.0681,
+      "step": 11100
+    },
+    {
+      "epoch": 0.08042157991125395,
+      "grad_norm": 0.19072282314300537,
+      "learning_rate": 4.919585658754805e-06,
+      "loss": 1.0723,
+      "step": 11110
+    },
+    {
+      "epoch": 0.08049396657184014,
+      "grad_norm": 0.190969780087471,
+      "learning_rate": 4.919513272094219e-06,
+      "loss": 1.0853,
+      "step": 11120
+    },
+    {
+      "epoch": 0.08056635323242634,
+      "grad_norm": 0.17297950387001038,
+      "learning_rate": 4.919440885433633e-06,
+      "loss": 1.071,
+      "step": 11130
+    },
+    {
+      "epoch": 0.08063873989301251,
+      "grad_norm": 0.19058924913406372,
+      "learning_rate": 4.919368498773047e-06,
+      "loss": 1.0879,
+      "step": 11140
+    },
+    {
+      "epoch": 0.0807111265535987,
+      "grad_norm": 0.19707483053207397,
+      "learning_rate": 4.9192961121124605e-06,
+      "loss": 1.0816,
+      "step": 11150
+    },
+    {
+      "epoch": 0.0807835132141849,
+      "grad_norm": 0.1789708286523819,
+      "learning_rate": 4.919223725451874e-06,
+      "loss": 1.0836,
+      "step": 11160
+    },
+    {
+      "epoch": 0.08085589987477108,
+      "grad_norm": 0.17154861986637115,
+      "learning_rate": 4.919151338791288e-06,
+      "loss": 1.0631,
+      "step": 11170
+    },
+    {
+      "epoch": 0.08092828653535726,
+      "grad_norm": 0.1575436145067215,
+      "learning_rate": 4.919078952130702e-06,
+      "loss": 1.0772,
+      "step": 11180
+    },
+    {
+      "epoch": 0.08100067319594345,
+      "grad_norm": 0.18095077574253082,
+      "learning_rate": 4.919006565470116e-06,
+      "loss": 1.0588,
+      "step": 11190
+    },
+    {
+      "epoch": 0.08107305985652964,
+      "grad_norm": 0.18087738752365112,
+      "learning_rate": 4.9189341788095294e-06,
+      "loss": 1.0859,
+      "step": 11200
+    },
+    {
+      "epoch": 0.08114544651711583,
+      "grad_norm": 0.41991138458251953,
+      "learning_rate": 4.918861792148943e-06,
+      "loss": 1.0671,
+      "step": 11210
+    },
+    {
+      "epoch": 0.08121783317770201,
+      "grad_norm": 0.19493845105171204,
+      "learning_rate": 4.9187894054883575e-06,
+      "loss": 1.0691,
+      "step": 11220
+    },
+    {
+      "epoch": 0.0812902198382882,
+      "grad_norm": 0.16449686884880066,
+      "learning_rate": 4.918717018827771e-06,
+      "loss": 1.0713,
+      "step": 11230
+    },
+    {
+      "epoch": 0.08136260649887439,
+      "grad_norm": 0.16916941106319427,
+      "learning_rate": 4.918644632167185e-06,
+      "loss": 1.0779,
+      "step": 11240
+    },
+    {
+      "epoch": 0.08143499315946058,
+      "grad_norm": 0.1702130287885666,
+      "learning_rate": 4.918572245506598e-06,
+      "loss": 1.0677,
+      "step": 11250
+    },
+    {
+      "epoch": 0.08150737982004676,
+      "grad_norm": 0.1953752636909485,
+      "learning_rate": 4.918499858846012e-06,
+      "loss": 1.0766,
+      "step": 11260
+    },
+    {
+      "epoch": 0.08157976648063295,
+      "grad_norm": 0.18306571245193481,
+      "learning_rate": 4.918427472185426e-06,
+      "loss": 1.0737,
+      "step": 11270
+    },
+    {
+      "epoch": 0.08165215314121914,
+      "grad_norm": 0.15860667824745178,
+      "learning_rate": 4.91835508552484e-06,
+      "loss": 1.0552,
+      "step": 11280
+    },
+    {
+      "epoch": 0.08172453980180533,
+      "grad_norm": 0.1841406524181366,
+      "learning_rate": 4.918282698864254e-06,
+      "loss": 1.0779,
+      "step": 11290
+    },
+    {
+      "epoch": 0.0817969264623915,
+      "grad_norm": 0.22922413051128387,
+      "learning_rate": 4.918210312203667e-06,
+      "loss": 1.0727,
+      "step": 11300
+    },
+    {
+      "epoch": 0.0818693131229777,
+      "grad_norm": 0.17833411693572998,
+      "learning_rate": 4.918137925543081e-06,
+      "loss": 1.0751,
+      "step": 11310
+    },
+    {
+      "epoch": 0.08194169978356389,
+      "grad_norm": 0.1674196720123291,
+      "learning_rate": 4.9180655388824945e-06,
+      "loss": 1.0797,
+      "step": 11320
+    },
+    {
+      "epoch": 0.08201408644415008,
+      "grad_norm": 0.18351003527641296,
+      "learning_rate": 4.917993152221909e-06,
+      "loss": 1.0893,
+      "step": 11330
+    },
+    {
+      "epoch": 0.08208647310473625,
+      "grad_norm": 0.17393292486667633,
+      "learning_rate": 4.917920765561323e-06,
+      "loss": 1.073,
+      "step": 11340
+    },
+    {
+      "epoch": 0.08215885976532245,
+      "grad_norm": 0.18493473529815674,
+      "learning_rate": 4.917848378900736e-06,
+      "loss": 1.0783,
+      "step": 11350
+    },
+    {
+      "epoch": 0.08223124642590864,
+      "grad_norm": 0.18359433114528656,
+      "learning_rate": 4.91777599224015e-06,
+      "loss": 1.0538,
+      "step": 11360
+    },
+    {
+      "epoch": 0.08230363308649483,
+      "grad_norm": 0.1751643419265747,
+      "learning_rate": 4.917703605579564e-06,
+      "loss": 1.0779,
+      "step": 11370
+    },
+    {
+      "epoch": 0.082376019747081,
+      "grad_norm": 0.16881217062473297,
+      "learning_rate": 4.917631218918978e-06,
+      "loss": 1.0722,
+      "step": 11380
+    },
+    {
+      "epoch": 0.0824484064076672,
+      "grad_norm": 0.19173048436641693,
+      "learning_rate": 4.9175588322583915e-06,
+      "loss": 1.0616,
+      "step": 11390
+    },
+    {
+      "epoch": 0.08252079306825338,
+      "grad_norm": 0.1824052780866623,
+      "learning_rate": 4.917486445597805e-06,
+      "loss": 1.0691,
+      "step": 11400
+    },
+    {
+      "epoch": 0.08259317972883957,
+      "grad_norm": 0.1803470402956009,
+      "learning_rate": 4.91741405893722e-06,
+      "loss": 1.0615,
+      "step": 11410
+    },
+    {
+      "epoch": 0.08266556638942575,
+      "grad_norm": 0.18563689291477203,
+      "learning_rate": 4.917341672276633e-06,
+      "loss": 1.0701,
+      "step": 11420
+    },
+    {
+      "epoch": 0.08273795305001194,
+      "grad_norm": 0.17787225544452667,
+      "learning_rate": 4.917269285616047e-06,
+      "loss": 1.0812,
+      "step": 11430
+    },
+    {
+      "epoch": 0.08281033971059813,
+      "grad_norm": 0.1732112318277359,
+      "learning_rate": 4.9171968989554605e-06,
+      "loss": 1.0754,
+      "step": 11440
+    },
+    {
+      "epoch": 0.08288272637118432,
+      "grad_norm": 0.19006459414958954,
+      "learning_rate": 4.917124512294875e-06,
+      "loss": 1.0577,
+      "step": 11450
+    },
+    {
+      "epoch": 0.0829551130317705,
+      "grad_norm": 0.17567865550518036,
+      "learning_rate": 4.9170521256342885e-06,
+      "loss": 1.0775,
+      "step": 11460
+    },
+    {
+      "epoch": 0.08302749969235669,
+      "grad_norm": 0.1794678121805191,
+      "learning_rate": 4.916979738973702e-06,
+      "loss": 1.0655,
+      "step": 11470
+    },
+    {
+      "epoch": 0.08309988635294288,
+      "grad_norm": 0.20485854148864746,
+      "learning_rate": 4.916907352313116e-06,
+      "loss": 1.065,
+      "step": 11480
+    },
+    {
+      "epoch": 0.08317227301352907,
+      "grad_norm": 0.2057936191558838,
+      "learning_rate": 4.91683496565253e-06,
+      "loss": 1.0769,
+      "step": 11490
+    },
+    {
+      "epoch": 0.08324465967411525,
+      "grad_norm": 0.17958250641822815,
+      "learning_rate": 4.916762578991944e-06,
+      "loss": 1.0644,
+      "step": 11500
+    },
+    {
+      "epoch": 0.08331704633470144,
+      "grad_norm": 0.16068291664123535,
+      "learning_rate": 4.9166901923313575e-06,
+      "loss": 1.0761,
+      "step": 11510
+    },
+    {
+      "epoch": 0.08338943299528763,
+      "grad_norm": 0.18293248116970062,
+      "learning_rate": 4.916617805670771e-06,
+      "loss": 1.0632,
+      "step": 11520
+    },
+    {
+      "epoch": 0.08346181965587382,
+      "grad_norm": 0.19636189937591553,
+      "learning_rate": 4.9165454190101855e-06,
+      "loss": 1.0827,
+      "step": 11530
+    },
+    {
+      "epoch": 0.08353420631646,
+      "grad_norm": 0.18141409754753113,
+      "learning_rate": 4.916473032349599e-06,
+      "loss": 1.077,
+      "step": 11540
+    },
+    {
+      "epoch": 0.08360659297704619,
+      "grad_norm": 0.18197061121463776,
+      "learning_rate": 4.916400645689013e-06,
+      "loss": 1.0639,
+      "step": 11550
+    },
+    {
+      "epoch": 0.08367897963763238,
+      "grad_norm": 0.18665997684001923,
+      "learning_rate": 4.916328259028426e-06,
+      "loss": 1.0709,
+      "step": 11560
+    },
+    {
+      "epoch": 0.08375136629821857,
+      "grad_norm": 0.18202146887779236,
+      "learning_rate": 4.916255872367841e-06,
+      "loss": 1.0797,
+      "step": 11570
+    },
+    {
+      "epoch": 0.08382375295880475,
+      "grad_norm": 0.16882570087909698,
+      "learning_rate": 4.9161834857072545e-06,
+      "loss": 1.0683,
+      "step": 11580
+    },
+    {
+      "epoch": 0.08389613961939094,
+      "grad_norm": 0.17632512748241425,
+      "learning_rate": 4.916111099046668e-06,
+      "loss": 1.0526,
+      "step": 11590
+    },
+    {
+      "epoch": 0.08396852627997713,
+      "grad_norm": 0.1826392412185669,
+      "learning_rate": 4.916038712386082e-06,
+      "loss": 1.0694,
+      "step": 11600
+    },
+    {
+      "epoch": 0.08404091294056332,
+      "grad_norm": 0.17934630811214447,
+      "learning_rate": 4.915966325725496e-06,
+      "loss": 1.0621,
+      "step": 11610
+    },
+    {
+      "epoch": 0.0841132996011495,
+      "grad_norm": 0.23041057586669922,
+      "learning_rate": 4.91589393906491e-06,
+      "loss": 1.0716,
+      "step": 11620
+    },
+    {
+      "epoch": 0.08418568626173568,
+      "grad_norm": 0.18051432073116302,
+      "learning_rate": 4.915821552404323e-06,
+      "loss": 1.0688,
+      "step": 11630
+    },
+    {
+      "epoch": 0.08425807292232187,
+      "grad_norm": 0.18257080018520355,
+      "learning_rate": 4.915749165743737e-06,
+      "loss": 1.0637,
+      "step": 11640
+    },
+    {
+      "epoch": 0.08433045958290807,
+      "grad_norm": 0.17362256348133087,
+      "learning_rate": 4.9156767790831515e-06,
+      "loss": 1.0703,
+      "step": 11650
+    },
+    {
+      "epoch": 0.08440284624349424,
+      "grad_norm": 0.1855854094028473,
+      "learning_rate": 4.915604392422565e-06,
+      "loss": 1.0711,
+      "step": 11660
+    },
+    {
+      "epoch": 0.08447523290408043,
+      "grad_norm": 0.17203544080257416,
+      "learning_rate": 4.915532005761979e-06,
+      "loss": 1.0666,
+      "step": 11670
+    },
+    {
+      "epoch": 0.08454761956466662,
+      "grad_norm": 0.18600568175315857,
+      "learning_rate": 4.915459619101392e-06,
+      "loss": 1.0635,
+      "step": 11680
+    },
+    {
+      "epoch": 0.08462000622525281,
+      "grad_norm": 0.1972518265247345,
+      "learning_rate": 4.915387232440807e-06,
+      "loss": 1.0761,
+      "step": 11690
+    },
+    {
+      "epoch": 0.08469239288583899,
+      "grad_norm": 0.1868593841791153,
+      "learning_rate": 4.91531484578022e-06,
+      "loss": 1.0943,
+      "step": 11700
+    },
+    {
+      "epoch": 0.08476477954642518,
+      "grad_norm": 0.18218737840652466,
+      "learning_rate": 4.915242459119634e-06,
+      "loss": 1.0787,
+      "step": 11710
+    },
+    {
+      "epoch": 0.08483716620701137,
+      "grad_norm": 0.19324436783790588,
+      "learning_rate": 4.915170072459048e-06,
+      "loss": 1.0549,
+      "step": 11720
+    },
+    {
+      "epoch": 0.08490955286759756,
+      "grad_norm": 0.16747941076755524,
+      "learning_rate": 4.915097685798462e-06,
+      "loss": 1.0751,
+      "step": 11730
+    },
+    {
+      "epoch": 0.08498193952818375,
+      "grad_norm": 0.1799047440290451,
+      "learning_rate": 4.915025299137876e-06,
+      "loss": 1.072,
+      "step": 11740
+    },
+    {
+      "epoch": 0.08505432618876993,
+      "grad_norm": 0.17832840979099274,
+      "learning_rate": 4.914952912477289e-06,
+      "loss": 1.0627,
+      "step": 11750
+    },
+    {
+      "epoch": 0.08512671284935612,
+      "grad_norm": 0.21101170778274536,
+      "learning_rate": 4.914880525816703e-06,
+      "loss": 1.0887,
+      "step": 11760
+    },
+    {
+      "epoch": 0.08519909950994231,
+      "grad_norm": 0.20275020599365234,
+      "learning_rate": 4.9148081391561166e-06,
+      "loss": 1.0752,
+      "step": 11770
+    },
+    {
+      "epoch": 0.0852714861705285,
+      "grad_norm": 0.173346146941185,
+      "learning_rate": 4.91473575249553e-06,
+      "loss": 1.0769,
+      "step": 11780
+    },
+    {
+      "epoch": 0.08534387283111468,
+      "grad_norm": 0.19621768593788147,
+      "learning_rate": 4.914663365834944e-06,
+      "loss": 1.0652,
+      "step": 11790
+    },
+    {
+      "epoch": 0.08541625949170087,
+      "grad_norm": 0.19705529510974884,
+      "learning_rate": 4.914590979174358e-06,
+      "loss": 1.0729,
+      "step": 11800
+    },
+    {
+      "epoch": 0.08548864615228706,
+      "grad_norm": 0.1750185489654541,
+      "learning_rate": 4.914518592513772e-06,
+      "loss": 1.0859,
+      "step": 11810
+    },
+    {
+      "epoch": 0.08556103281287325,
+      "grad_norm": 0.17165139317512512,
+      "learning_rate": 4.9144462058531855e-06,
+      "loss": 1.0737,
+      "step": 11820
+    },
+    {
+      "epoch": 0.08563341947345943,
+      "grad_norm": 0.18447789549827576,
+      "learning_rate": 4.914373819192599e-06,
+      "loss": 1.0659,
+      "step": 11830
+    },
+    {
+      "epoch": 0.08570580613404562,
+      "grad_norm": 0.18128694593906403,
+      "learning_rate": 4.9143014325320136e-06,
+      "loss": 1.0664,
+      "step": 11840
+    },
+    {
+      "epoch": 0.08577819279463181,
+      "grad_norm": 0.1879102736711502,
+      "learning_rate": 4.914229045871427e-06,
+      "loss": 1.0826,
+      "step": 11850
+    },
+    {
+      "epoch": 0.085850579455218,
+      "grad_norm": 0.1664225310087204,
+      "learning_rate": 4.914156659210841e-06,
+      "loss": 1.0587,
+      "step": 11860
+    },
+    {
+      "epoch": 0.08592296611580418,
+      "grad_norm": 0.18422931432724,
+      "learning_rate": 4.914084272550254e-06,
+      "loss": 1.0793,
+      "step": 11870
+    },
+    {
+      "epoch": 0.08599535277639037,
+      "grad_norm": 0.17721515893936157,
+      "learning_rate": 4.914011885889669e-06,
+      "loss": 1.0602,
+      "step": 11880
+    },
+    {
+      "epoch": 0.08606773943697656,
+      "grad_norm": 0.17775796353816986,
+      "learning_rate": 4.9139394992290825e-06,
+      "loss": 1.0734,
+      "step": 11890
+    },
+    {
+      "epoch": 0.08614012609756275,
+      "grad_norm": 0.19729198515415192,
+      "learning_rate": 4.913867112568496e-06,
+      "loss": 1.0694,
+      "step": 11900
+    },
+    {
+      "epoch": 0.08621251275814892,
+      "grad_norm": 0.17630425095558167,
+      "learning_rate": 4.91379472590791e-06,
+      "loss": 1.0646,
+      "step": 11910
+    },
+    {
+      "epoch": 0.08628489941873511,
+      "grad_norm": 0.1867901086807251,
+      "learning_rate": 4.913722339247324e-06,
+      "loss": 1.0823,
+      "step": 11920
+    },
+    {
+      "epoch": 0.0863572860793213,
+      "grad_norm": 0.17957784235477448,
+      "learning_rate": 4.913649952586738e-06,
+      "loss": 1.0621,
+      "step": 11930
+    },
+    {
+      "epoch": 0.0864296727399075,
+      "grad_norm": 0.18513678014278412,
+      "learning_rate": 4.9135775659261514e-06,
+      "loss": 1.0737,
+      "step": 11940
+    },
+    {
+      "epoch": 0.08650205940049367,
+      "grad_norm": 0.18073897063732147,
+      "learning_rate": 4.913505179265565e-06,
+      "loss": 1.0567,
+      "step": 11950
+    },
+    {
+      "epoch": 0.08657444606107986,
+      "grad_norm": 0.19034330546855927,
+      "learning_rate": 4.913432792604979e-06,
+      "loss": 1.0701,
+      "step": 11960
+    },
+    {
+      "epoch": 0.08664683272166605,
+      "grad_norm": 0.19168923795223236,
+      "learning_rate": 4.913360405944393e-06,
+      "loss": 1.0592,
+      "step": 11970
+    },
+    {
+      "epoch": 0.08671921938225224,
+      "grad_norm": 0.1833886355161667,
+      "learning_rate": 4.913288019283807e-06,
+      "loss": 1.0812,
+      "step": 11980
+    },
+    {
+      "epoch": 0.08679160604283842,
+      "grad_norm": 0.1789073795080185,
+      "learning_rate": 4.91321563262322e-06,
+      "loss": 1.0813,
+      "step": 11990
+    },
+    {
+      "epoch": 0.08686399270342461,
+      "grad_norm": 0.19956181943416595,
+      "learning_rate": 4.913143245962634e-06,
+      "loss": 1.0602,
+      "step": 12000
+    },
+    {
+      "epoch": 0.0869363793640108,
+      "grad_norm": 0.16924233734607697,
+      "learning_rate": 4.9130708593020484e-06,
+      "loss": 1.0682,
+      "step": 12010
+    },
+    {
+      "epoch": 0.08700876602459699,
+      "grad_norm": 0.20101603865623474,
+      "learning_rate": 4.912998472641462e-06,
+      "loss": 1.0896,
+      "step": 12020
+    },
+    {
+      "epoch": 0.08708115268518317,
+      "grad_norm": 0.17488181591033936,
+      "learning_rate": 4.912926085980876e-06,
+      "loss": 1.0551,
+      "step": 12030
+    },
+    {
+      "epoch": 0.08715353934576936,
+      "grad_norm": 0.18475006520748138,
+      "learning_rate": 4.912853699320289e-06,
+      "loss": 1.0825,
+      "step": 12040
+    },
+    {
+      "epoch": 0.08722592600635555,
+      "grad_norm": 0.2061099112033844,
+      "learning_rate": 4.912781312659704e-06,
+      "loss": 1.0836,
+      "step": 12050
+    },
+    {
+      "epoch": 0.08729831266694174,
+      "grad_norm": 0.7837763428688049,
+      "learning_rate": 4.912708925999117e-06,
+      "loss": 1.0716,
+      "step": 12060
+    },
+    {
+      "epoch": 0.08737069932752792,
+      "grad_norm": 0.2470492273569107,
+      "learning_rate": 4.912636539338531e-06,
+      "loss": 1.0686,
+      "step": 12070
+    },
+    {
+      "epoch": 0.08744308598811411,
+      "grad_norm": 0.18343280255794525,
+      "learning_rate": 4.912564152677945e-06,
+      "loss": 1.0704,
+      "step": 12080
+    },
+    {
+      "epoch": 0.0875154726487003,
+      "grad_norm": 0.18261706829071045,
+      "learning_rate": 4.912491766017359e-06,
+      "loss": 1.0632,
+      "step": 12090
+    },
+    {
+      "epoch": 0.08758785930928649,
+      "grad_norm": 0.25385862588882446,
+      "learning_rate": 4.912419379356773e-06,
+      "loss": 1.0686,
+      "step": 12100
+    },
+    {
+      "epoch": 0.08766024596987267,
+      "grad_norm": 0.1851423978805542,
+      "learning_rate": 4.912346992696186e-06,
+      "loss": 1.0647,
+      "step": 12110
+    },
+    {
+      "epoch": 0.08773263263045886,
+      "grad_norm": 0.16765649616718292,
+      "learning_rate": 4.9122746060356e-06,
+      "loss": 1.0625,
+      "step": 12120
+    },
+    {
+      "epoch": 0.08780501929104505,
+      "grad_norm": 0.17632648348808289,
+      "learning_rate": 4.912202219375014e-06,
+      "loss": 1.0767,
+      "step": 12130
+    },
+    {
+      "epoch": 0.08787740595163124,
+      "grad_norm": 0.20516112446784973,
+      "learning_rate": 4.912129832714428e-06,
+      "loss": 1.0764,
+      "step": 12140
+    },
+    {
+      "epoch": 0.08794979261221741,
+      "grad_norm": 0.19637328386306763,
+      "learning_rate": 4.912057446053842e-06,
+      "loss": 1.0819,
+      "step": 12150
+    },
+    {
+      "epoch": 0.0880221792728036,
+      "grad_norm": 0.17330960929393768,
+      "learning_rate": 4.911985059393255e-06,
+      "loss": 1.0731,
+      "step": 12160
+    },
+    {
+      "epoch": 0.0880945659333898,
+      "grad_norm": 0.17274819314479828,
+      "learning_rate": 4.91191267273267e-06,
+      "loss": 1.0811,
+      "step": 12170
+    },
+    {
+      "epoch": 0.08816695259397599,
+      "grad_norm": 0.18579480051994324,
+      "learning_rate": 4.911840286072083e-06,
+      "loss": 1.0769,
+      "step": 12180
+    },
+    {
+      "epoch": 0.08823933925456216,
+      "grad_norm": 0.18373064696788788,
+      "learning_rate": 4.911767899411497e-06,
+      "loss": 1.0749,
+      "step": 12190
+    },
+    {
+      "epoch": 0.08831172591514835,
+      "grad_norm": 0.1849633753299713,
+      "learning_rate": 4.9116955127509105e-06,
+      "loss": 1.0789,
+      "step": 12200
+    },
+    {
+      "epoch": 0.08838411257573454,
+      "grad_norm": 0.1739869862794876,
+      "learning_rate": 4.911623126090325e-06,
+      "loss": 1.0805,
+      "step": 12210
+    },
+    {
+      "epoch": 0.08845649923632073,
+      "grad_norm": 0.18374896049499512,
+      "learning_rate": 4.911550739429739e-06,
+      "loss": 1.0491,
+      "step": 12220
+    },
+    {
+      "epoch": 0.08852888589690691,
+      "grad_norm": 0.20483864843845367,
+      "learning_rate": 4.911478352769152e-06,
+      "loss": 1.0698,
+      "step": 12230
+    },
+    {
+      "epoch": 0.0886012725574931,
+      "grad_norm": 0.17698989808559418,
+      "learning_rate": 4.911405966108566e-06,
+      "loss": 1.0757,
+      "step": 12240
+    },
+    {
+      "epoch": 0.08867365921807929,
+      "grad_norm": 0.16468937695026398,
+      "learning_rate": 4.91133357944798e-06,
+      "loss": 1.0577,
+      "step": 12250
+    },
+    {
+      "epoch": 0.08874604587866548,
+      "grad_norm": 0.18862901628017426,
+      "learning_rate": 4.911261192787394e-06,
+      "loss": 1.0779,
+      "step": 12260
+    },
+    {
+      "epoch": 0.08881843253925166,
+      "grad_norm": 0.17492325603961945,
+      "learning_rate": 4.9111888061268075e-06,
+      "loss": 1.0662,
+      "step": 12270
+    },
+    {
+      "epoch": 0.08889081919983785,
+      "grad_norm": 0.16824495792388916,
+      "learning_rate": 4.911116419466221e-06,
+      "loss": 1.0617,
+      "step": 12280
+    },
+    {
+      "epoch": 0.08896320586042404,
+      "grad_norm": 0.17155864834785461,
+      "learning_rate": 4.911044032805636e-06,
+      "loss": 1.058,
+      "step": 12290
+    },
+    {
+      "epoch": 0.08903559252101023,
+      "grad_norm": 0.17374593019485474,
+      "learning_rate": 4.910971646145049e-06,
+      "loss": 1.0538,
+      "step": 12300
+    },
+    {
+      "epoch": 0.08910797918159642,
+      "grad_norm": 0.170820415019989,
+      "learning_rate": 4.910899259484462e-06,
+      "loss": 1.0719,
+      "step": 12310
+    },
+    {
+      "epoch": 0.0891803658421826,
+      "grad_norm": 0.1844668984413147,
+      "learning_rate": 4.9108268728238765e-06,
+      "loss": 1.058,
+      "step": 12320
+    },
+    {
+      "epoch": 0.08925275250276879,
+      "grad_norm": 0.17644762992858887,
+      "learning_rate": 4.91075448616329e-06,
+      "loss": 1.0533,
+      "step": 12330
+    },
+    {
+      "epoch": 0.08932513916335498,
+      "grad_norm": 0.19445423781871796,
+      "learning_rate": 4.910682099502704e-06,
+      "loss": 1.0617,
+      "step": 12340
+    },
+    {
+      "epoch": 0.08939752582394117,
+      "grad_norm": 0.20080770552158356,
+      "learning_rate": 4.910609712842117e-06,
+      "loss": 1.0592,
+      "step": 12350
+    },
+    {
+      "epoch": 0.08946991248452735,
+      "grad_norm": 0.17183251678943634,
+      "learning_rate": 4.910537326181532e-06,
+      "loss": 1.0629,
+      "step": 12360
+    },
+    {
+      "epoch": 0.08954229914511354,
+      "grad_norm": 0.17015236616134644,
+      "learning_rate": 4.910464939520945e-06,
+      "loss": 1.0571,
+      "step": 12370
+    },
+    {
+      "epoch": 0.08961468580569973,
+      "grad_norm": 0.19391165673732758,
+      "learning_rate": 4.910392552860359e-06,
+      "loss": 1.0634,
+      "step": 12380
+    },
+    {
+      "epoch": 0.08968707246628592,
+      "grad_norm": 0.21788957715034485,
+      "learning_rate": 4.910320166199773e-06,
+      "loss": 1.0633,
+      "step": 12390
+    },
+    {
+      "epoch": 0.0897594591268721,
+      "grad_norm": 0.1834946870803833,
+      "learning_rate": 4.910247779539187e-06,
+      "loss": 1.0594,
+      "step": 12400
+    },
+    {
+      "epoch": 0.08983184578745829,
+      "grad_norm": 0.1719583421945572,
+      "learning_rate": 4.910175392878601e-06,
+      "loss": 1.0485,
+      "step": 12410
+    },
+    {
+      "epoch": 0.08990423244804448,
+      "grad_norm": 0.21385973691940308,
+      "learning_rate": 4.910103006218014e-06,
+      "loss": 1.068,
+      "step": 12420
+    },
+    {
+      "epoch": 0.08997661910863067,
+      "grad_norm": 0.1749151051044464,
+      "learning_rate": 4.910030619557428e-06,
+      "loss": 1.0658,
+      "step": 12430
+    },
+    {
+      "epoch": 0.09004900576921684,
+      "grad_norm": 0.17845019698143005,
+      "learning_rate": 4.909958232896842e-06,
+      "loss": 1.0693,
+      "step": 12440
+    },
+    {
+      "epoch": 0.09012139242980303,
+      "grad_norm": 0.16557128727436066,
+      "learning_rate": 4.909885846236256e-06,
+      "loss": 1.05,
+      "step": 12450
+    },
+    {
+      "epoch": 0.09019377909038923,
+      "grad_norm": 0.1660662740468979,
+      "learning_rate": 4.90981345957567e-06,
+      "loss": 1.0756,
+      "step": 12460
+    },
+    {
+      "epoch": 0.09026616575097542,
+      "grad_norm": 0.17898871004581451,
+      "learning_rate": 4.909741072915083e-06,
+      "loss": 1.0692,
+      "step": 12470
+    },
+    {
+      "epoch": 0.09033855241156159,
+      "grad_norm": 0.18854671716690063,
+      "learning_rate": 4.909668686254498e-06,
+      "loss": 1.0648,
+      "step": 12480
+    },
+    {
+      "epoch": 0.09041093907214778,
+      "grad_norm": 0.18120893836021423,
+      "learning_rate": 4.909596299593911e-06,
+      "loss": 1.0646,
+      "step": 12490
+    },
+    {
+      "epoch": 0.09048332573273397,
+      "grad_norm": 0.1960660070180893,
+      "learning_rate": 4.909523912933325e-06,
+      "loss": 1.067,
+      "step": 12500
+    },
+    {
+      "epoch": 0.09055571239332016,
+      "grad_norm": 0.2270880788564682,
+      "learning_rate": 4.9094515262727386e-06,
+      "loss": 1.057,
+      "step": 12510
+    },
+    {
+      "epoch": 0.09062809905390634,
+      "grad_norm": 0.19658519327640533,
+      "learning_rate": 4.909379139612153e-06,
+      "loss": 1.0789,
+      "step": 12520
+    },
+    {
+      "epoch": 0.09070048571449253,
+      "grad_norm": 0.1924048811197281,
+      "learning_rate": 4.909306752951567e-06,
+      "loss": 1.0608,
+      "step": 12530
+    },
+    {
+      "epoch": 0.09077287237507872,
+      "grad_norm": 0.1948341578245163,
+      "learning_rate": 4.90923436629098e-06,
+      "loss": 1.0594,
+      "step": 12540
+    },
+    {
+      "epoch": 0.09084525903566491,
+      "grad_norm": 0.21154290437698364,
+      "learning_rate": 4.909161979630394e-06,
+      "loss": 1.0635,
+      "step": 12550
+    },
+    {
+      "epoch": 0.09091764569625109,
+      "grad_norm": 0.18303339183330536,
+      "learning_rate": 4.909089592969808e-06,
+      "loss": 1.0802,
+      "step": 12560
+    },
+    {
+      "epoch": 0.09099003235683728,
+      "grad_norm": 0.17858447134494781,
+      "learning_rate": 4.909017206309222e-06,
+      "loss": 1.065,
+      "step": 12570
+    },
+    {
+      "epoch": 0.09106241901742347,
+      "grad_norm": 0.186380535364151,
+      "learning_rate": 4.9089448196486356e-06,
+      "loss": 1.0611,
+      "step": 12580
+    },
+    {
+      "epoch": 0.09113480567800966,
+      "grad_norm": 0.17138642072677612,
+      "learning_rate": 4.908872432988049e-06,
+      "loss": 1.0598,
+      "step": 12590
+    },
+    {
+      "epoch": 0.09120719233859584,
+      "grad_norm": 0.21660137176513672,
+      "learning_rate": 4.908800046327463e-06,
+      "loss": 1.0604,
+      "step": 12600
+    },
+    {
+      "epoch": 0.09127957899918203,
+      "grad_norm": 0.17236590385437012,
+      "learning_rate": 4.908727659666877e-06,
+      "loss": 1.0554,
+      "step": 12610
+    },
+    {
+      "epoch": 0.09135196565976822,
+      "grad_norm": 0.16949652135372162,
+      "learning_rate": 4.908655273006291e-06,
+      "loss": 1.0523,
+      "step": 12620
+    },
+    {
+      "epoch": 0.09142435232035441,
+      "grad_norm": 0.18573276698589325,
+      "learning_rate": 4.9085828863457045e-06,
+      "loss": 1.0574,
+      "step": 12630
+    },
+    {
+      "epoch": 0.09149673898094059,
+      "grad_norm": 0.17692014575004578,
+      "learning_rate": 4.908510499685118e-06,
+      "loss": 1.0683,
+      "step": 12640
+    },
+    {
+      "epoch": 0.09156912564152678,
+      "grad_norm": 0.17320817708969116,
+      "learning_rate": 4.908438113024533e-06,
+      "loss": 1.0725,
+      "step": 12650
+    },
+    {
+      "epoch": 0.09164151230211297,
+      "grad_norm": 0.18194791674613953,
+      "learning_rate": 4.908365726363946e-06,
+      "loss": 1.0769,
+      "step": 12660
+    },
+    {
+      "epoch": 0.09171389896269916,
+      "grad_norm": 0.17334748804569244,
+      "learning_rate": 4.90829333970336e-06,
+      "loss": 1.0648,
+      "step": 12670
+    },
+    {
+      "epoch": 0.09178628562328534,
+      "grad_norm": 0.18237444758415222,
+      "learning_rate": 4.9082209530427734e-06,
+      "loss": 1.0701,
+      "step": 12680
+    },
+    {
+      "epoch": 0.09185867228387153,
+      "grad_norm": 0.18935726583003998,
+      "learning_rate": 4.908148566382188e-06,
+      "loss": 1.0576,
+      "step": 12690
+    },
+    {
+      "epoch": 0.09193105894445772,
+      "grad_norm": 0.17094674706459045,
+      "learning_rate": 4.9080761797216015e-06,
+      "loss": 1.0695,
+      "step": 12700
+    },
+    {
+      "epoch": 0.09200344560504391,
+      "grad_norm": 0.1917349398136139,
+      "learning_rate": 4.908003793061015e-06,
+      "loss": 1.061,
+      "step": 12710
+    },
+    {
+      "epoch": 0.09207583226563008,
+      "grad_norm": 0.18850326538085938,
+      "learning_rate": 4.907931406400429e-06,
+      "loss": 1.0772,
+      "step": 12720
+    },
+    {
+      "epoch": 0.09214821892621627,
+      "grad_norm": 0.20601466298103333,
+      "learning_rate": 4.907859019739843e-06,
+      "loss": 1.0626,
+      "step": 12730
+    },
+    {
+      "epoch": 0.09222060558680246,
+      "grad_norm": 0.1699492633342743,
+      "learning_rate": 4.907786633079257e-06,
+      "loss": 1.054,
+      "step": 12740
+    },
+    {
+      "epoch": 0.09229299224738866,
+      "grad_norm": 0.16967886686325073,
+      "learning_rate": 4.9077142464186704e-06,
+      "loss": 1.0633,
+      "step": 12750
+    },
+    {
+      "epoch": 0.09236537890797483,
+      "grad_norm": 0.16590727865695953,
+      "learning_rate": 4.907641859758084e-06,
+      "loss": 1.0436,
+      "step": 12760
+    },
+    {
+      "epoch": 0.09243776556856102,
+      "grad_norm": 0.18494842946529388,
+      "learning_rate": 4.9075694730974985e-06,
+      "loss": 1.0639,
+      "step": 12770
+    },
+    {
+      "epoch": 0.09251015222914721,
+      "grad_norm": 0.19467203319072723,
+      "learning_rate": 4.907497086436912e-06,
+      "loss": 1.059,
+      "step": 12780
+    },
+    {
+      "epoch": 0.0925825388897334,
+      "grad_norm": 0.1718284636735916,
+      "learning_rate": 4.907424699776326e-06,
+      "loss": 1.0615,
+      "step": 12790
+    },
+    {
+      "epoch": 0.09265492555031958,
+      "grad_norm": 0.16538646817207336,
+      "learning_rate": 4.907352313115739e-06,
+      "loss": 1.0653,
+      "step": 12800
+    },
+    {
+      "epoch": 0.09272731221090577,
+      "grad_norm": 0.17664389312267303,
+      "learning_rate": 4.907279926455154e-06,
+      "loss": 1.0595,
+      "step": 12810
+    },
+    {
+      "epoch": 0.09279969887149196,
+      "grad_norm": 0.1798931062221527,
+      "learning_rate": 4.9072075397945674e-06,
+      "loss": 1.0578,
+      "step": 12820
+    },
+    {
+      "epoch": 0.09287208553207815,
+      "grad_norm": 0.20928549766540527,
+      "learning_rate": 4.907135153133981e-06,
+      "loss": 1.0699,
+      "step": 12830
+    },
+    {
+      "epoch": 0.09294447219266433,
+      "grad_norm": 0.1914556324481964,
+      "learning_rate": 4.907062766473395e-06,
+      "loss": 1.078,
+      "step": 12840
+    },
+    {
+      "epoch": 0.09301685885325052,
+      "grad_norm": 0.17535455524921417,
+      "learning_rate": 4.906990379812808e-06,
+      "loss": 1.062,
+      "step": 12850
+    },
+    {
+      "epoch": 0.09308924551383671,
+      "grad_norm": 0.209900364279747,
+      "learning_rate": 4.906917993152222e-06,
+      "loss": 1.0651,
+      "step": 12860
+    },
+    {
+      "epoch": 0.0931616321744229,
+      "grad_norm": 0.19271911680698395,
+      "learning_rate": 4.9068456064916355e-06,
+      "loss": 1.0695,
+      "step": 12870
+    },
+    {
+      "epoch": 0.09323401883500908,
+      "grad_norm": 0.17206203937530518,
+      "learning_rate": 4.90677321983105e-06,
+      "loss": 1.054,
+      "step": 12880
+    },
+    {
+      "epoch": 0.09330640549559527,
+      "grad_norm": 0.17323225736618042,
+      "learning_rate": 4.906700833170464e-06,
+      "loss": 1.0545,
+      "step": 12890
+    },
+    {
+      "epoch": 0.09337879215618146,
+      "grad_norm": 0.18585805594921112,
+      "learning_rate": 4.906628446509877e-06,
+      "loss": 1.0638,
+      "step": 12900
+    },
+    {
+      "epoch": 0.09345117881676765,
+      "grad_norm": 0.18131840229034424,
+      "learning_rate": 4.906556059849291e-06,
+      "loss": 1.0575,
+      "step": 12910
+    },
+    {
+      "epoch": 0.09352356547735384,
+      "grad_norm": 0.18972773849964142,
+      "learning_rate": 4.906483673188705e-06,
+      "loss": 1.0561,
+      "step": 12920
+    },
+    {
+      "epoch": 0.09359595213794002,
+      "grad_norm": 0.17242597043514252,
+      "learning_rate": 4.906411286528119e-06,
+      "loss": 1.0754,
+      "step": 12930
+    },
+    {
+      "epoch": 0.09366833879852621,
+      "grad_norm": 0.17874747514724731,
+      "learning_rate": 4.9063388998675325e-06,
+      "loss": 1.0769,
+      "step": 12940
+    },
+    {
+      "epoch": 0.0937407254591124,
+      "grad_norm": 0.1628153920173645,
+      "learning_rate": 4.906266513206946e-06,
+      "loss": 1.0701,
+      "step": 12950
+    },
+    {
+      "epoch": 0.09381311211969859,
+      "grad_norm": 0.1775561273097992,
+      "learning_rate": 4.906194126546361e-06,
+      "loss": 1.0496,
+      "step": 12960
+    },
+    {
+      "epoch": 0.09388549878028477,
+      "grad_norm": 0.17346498370170593,
+      "learning_rate": 4.906121739885774e-06,
+      "loss": 1.0592,
+      "step": 12970
+    },
+    {
+      "epoch": 0.09395788544087096,
+      "grad_norm": 0.1621558666229248,
+      "learning_rate": 4.906049353225188e-06,
+      "loss": 1.0639,
+      "step": 12980
+    },
+    {
+      "epoch": 0.09403027210145715,
+      "grad_norm": 0.16274607181549072,
+      "learning_rate": 4.9059769665646015e-06,
+      "loss": 1.0593,
+      "step": 12990
+    },
+    {
+      "epoch": 0.09410265876204334,
+      "grad_norm": 0.24660325050354004,
+      "learning_rate": 4.905904579904016e-06,
+      "loss": 1.0687,
+      "step": 13000
+    },
+    {
+      "epoch": 0.09417504542262951,
+      "grad_norm": 0.17829085886478424,
+      "learning_rate": 4.9058321932434295e-06,
+      "loss": 1.061,
+      "step": 13010
+    },
+    {
+      "epoch": 0.0942474320832157,
+      "grad_norm": 0.19128848612308502,
+      "learning_rate": 4.905759806582843e-06,
+      "loss": 1.0612,
+      "step": 13020
+    },
+    {
+      "epoch": 0.0943198187438019,
+      "grad_norm": 0.1639917492866516,
+      "learning_rate": 4.905687419922257e-06,
+      "loss": 1.0514,
+      "step": 13030
+    },
+    {
+      "epoch": 0.09439220540438809,
+      "grad_norm": 0.18937060236930847,
+      "learning_rate": 4.905615033261671e-06,
+      "loss": 1.0471,
+      "step": 13040
+    },
+    {
+      "epoch": 0.09446459206497426,
+      "grad_norm": 0.20053894817829132,
+      "learning_rate": 4.905542646601085e-06,
+      "loss": 1.051,
+      "step": 13050
+    },
+    {
+      "epoch": 0.09453697872556045,
+      "grad_norm": 0.17322196066379547,
+      "learning_rate": 4.9054702599404985e-06,
+      "loss": 1.0626,
+      "step": 13060
+    },
+    {
+      "epoch": 0.09460936538614664,
+      "grad_norm": 0.18326787650585175,
+      "learning_rate": 4.905397873279912e-06,
+      "loss": 1.0556,
+      "step": 13070
+    },
+    {
+      "epoch": 0.09468175204673283,
+      "grad_norm": 0.18732202053070068,
+      "learning_rate": 4.9053254866193265e-06,
+      "loss": 1.0697,
+      "step": 13080
+    },
+    {
+      "epoch": 0.09475413870731901,
+      "grad_norm": 0.1722288727760315,
+      "learning_rate": 4.90525309995874e-06,
+      "loss": 1.0558,
+      "step": 13090
+    },
+    {
+      "epoch": 0.0948265253679052,
+      "grad_norm": 0.1705879271030426,
+      "learning_rate": 4.905180713298154e-06,
+      "loss": 1.0436,
+      "step": 13100
+    },
+    {
+      "epoch": 0.09489891202849139,
+      "grad_norm": 0.19368794560432434,
+      "learning_rate": 4.905108326637567e-06,
+      "loss": 1.0541,
+      "step": 13110
+    },
+    {
+      "epoch": 0.09497129868907758,
+      "grad_norm": 0.16703177988529205,
+      "learning_rate": 4.905035939976982e-06,
+      "loss": 1.061,
+      "step": 13120
+    },
+    {
+      "epoch": 0.09504368534966376,
+      "grad_norm": 0.1757340431213379,
+      "learning_rate": 4.9049635533163955e-06,
+      "loss": 1.054,
+      "step": 13130
+    },
+    {
+      "epoch": 0.09511607201024995,
+      "grad_norm": 0.1705889105796814,
+      "learning_rate": 4.904891166655809e-06,
+      "loss": 1.0499,
+      "step": 13140
+    },
+    {
+      "epoch": 0.09518845867083614,
+      "grad_norm": 0.18532103300094604,
+      "learning_rate": 4.904818779995223e-06,
+      "loss": 1.073,
+      "step": 13150
+    },
+    {
+      "epoch": 0.09526084533142233,
+      "grad_norm": 0.17054064571857452,
+      "learning_rate": 4.904746393334637e-06,
+      "loss": 1.0548,
+      "step": 13160
+    },
+    {
+      "epoch": 0.09533323199200851,
+      "grad_norm": 0.18268738687038422,
+      "learning_rate": 4.904674006674051e-06,
+      "loss": 1.0489,
+      "step": 13170
+    },
+    {
+      "epoch": 0.0954056186525947,
+      "grad_norm": 0.1656617373228073,
+      "learning_rate": 4.904601620013464e-06,
+      "loss": 1.048,
+      "step": 13180
+    },
+    {
+      "epoch": 0.09547800531318089,
+      "grad_norm": 0.16891372203826904,
+      "learning_rate": 4.904529233352878e-06,
+      "loss": 1.0719,
+      "step": 13190
+    },
+    {
+      "epoch": 0.09555039197376708,
+      "grad_norm": 0.19069869816303253,
+      "learning_rate": 4.904456846692292e-06,
+      "loss": 1.0673,
+      "step": 13200
+    },
+    {
+      "epoch": 0.09562277863435326,
+      "grad_norm": 0.17400185763835907,
+      "learning_rate": 4.904384460031706e-06,
+      "loss": 1.0586,
+      "step": 13210
+    },
+    {
+      "epoch": 0.09569516529493945,
+      "grad_norm": 0.17865540087223053,
+      "learning_rate": 4.90431207337112e-06,
+      "loss": 1.0509,
+      "step": 13220
+    },
+    {
+      "epoch": 0.09576755195552564,
+      "grad_norm": 0.22971458733081818,
+      "learning_rate": 4.904239686710533e-06,
+      "loss": 1.0619,
+      "step": 13230
+    },
+    {
+      "epoch": 0.09583993861611183,
+      "grad_norm": 0.16705693304538727,
+      "learning_rate": 4.904167300049947e-06,
+      "loss": 1.0616,
+      "step": 13240
+    },
+    {
+      "epoch": 0.095912325276698,
+      "grad_norm": 0.17722009122371674,
+      "learning_rate": 4.904094913389361e-06,
+      "loss": 1.046,
+      "step": 13250
+    },
+    {
+      "epoch": 0.0959847119372842,
+      "grad_norm": 0.18104061484336853,
+      "learning_rate": 4.904022526728775e-06,
+      "loss": 1.0516,
+      "step": 13260
+    },
+    {
+      "epoch": 0.09605709859787039,
+      "grad_norm": 0.1646488904953003,
+      "learning_rate": 4.903950140068189e-06,
+      "loss": 1.0612,
+      "step": 13270
+    },
+    {
+      "epoch": 0.09612948525845658,
+      "grad_norm": 0.17843876779079437,
+      "learning_rate": 4.903877753407602e-06,
+      "loss": 1.0584,
+      "step": 13280
+    },
+    {
+      "epoch": 0.09620187191904275,
+      "grad_norm": 0.18090327084064484,
+      "learning_rate": 4.903805366747017e-06,
+      "loss": 1.063,
+      "step": 13290
+    },
+    {
+      "epoch": 0.09627425857962894,
+      "grad_norm": 0.17758344113826752,
+      "learning_rate": 4.90373298008643e-06,
+      "loss": 1.0633,
+      "step": 13300
+    },
+    {
+      "epoch": 0.09634664524021513,
+      "grad_norm": 0.17629143595695496,
+      "learning_rate": 4.903660593425844e-06,
+      "loss": 1.0546,
+      "step": 13310
+    },
+    {
+      "epoch": 0.09641903190080132,
+      "grad_norm": 0.17930828034877777,
+      "learning_rate": 4.9035882067652576e-06,
+      "loss": 1.0679,
+      "step": 13320
+    },
+    {
+      "epoch": 0.0964914185613875,
+      "grad_norm": 0.22573702037334442,
+      "learning_rate": 4.903515820104672e-06,
+      "loss": 1.062,
+      "step": 13330
+    },
+    {
+      "epoch": 0.09656380522197369,
+      "grad_norm": 0.19880887866020203,
+      "learning_rate": 4.903443433444086e-06,
+      "loss": 1.0614,
+      "step": 13340
+    },
+    {
+      "epoch": 0.09663619188255988,
+      "grad_norm": 0.20379464328289032,
+      "learning_rate": 4.903371046783499e-06,
+      "loss": 1.0585,
+      "step": 13350
+    },
+    {
+      "epoch": 0.09670857854314607,
+      "grad_norm": 0.1923993080854416,
+      "learning_rate": 4.903298660122913e-06,
+      "loss": 1.0647,
+      "step": 13360
+    },
+    {
+      "epoch": 0.09678096520373225,
+      "grad_norm": 0.18714511394500732,
+      "learning_rate": 4.9032262734623265e-06,
+      "loss": 1.0439,
+      "step": 13370
+    },
+    {
+      "epoch": 0.09685335186431844,
+      "grad_norm": 0.20776358246803284,
+      "learning_rate": 4.90315388680174e-06,
+      "loss": 1.0717,
+      "step": 13380
+    },
+    {
+      "epoch": 0.09692573852490463,
+      "grad_norm": 0.1841840147972107,
+      "learning_rate": 4.903081500141154e-06,
+      "loss": 1.0716,
+      "step": 13390
+    },
+    {
+      "epoch": 0.09699812518549082,
+      "grad_norm": 0.1891954392194748,
+      "learning_rate": 4.903009113480568e-06,
+      "loss": 1.0572,
+      "step": 13400
+    },
+    {
+      "epoch": 0.097070511846077,
+      "grad_norm": 0.2061716616153717,
+      "learning_rate": 4.902936726819982e-06,
+      "loss": 1.0587,
+      "step": 13410
+    },
+    {
+      "epoch": 0.09714289850666319,
+      "grad_norm": 0.1809329241514206,
+      "learning_rate": 4.9028643401593954e-06,
+      "loss": 1.0772,
+      "step": 13420
+    },
+    {
+      "epoch": 0.09721528516724938,
+      "grad_norm": 0.18182168900966644,
+      "learning_rate": 4.902791953498809e-06,
+      "loss": 1.0707,
+      "step": 13430
+    },
+    {
+      "epoch": 0.09728767182783557,
+      "grad_norm": 0.16881795227527618,
+      "learning_rate": 4.9027195668382235e-06,
+      "loss": 1.0617,
+      "step": 13440
+    },
+    {
+      "epoch": 0.09736005848842175,
+      "grad_norm": 0.18087784945964813,
+      "learning_rate": 4.902647180177637e-06,
+      "loss": 1.0566,
+      "step": 13450
+    },
+    {
+      "epoch": 0.09743244514900794,
+      "grad_norm": 0.19071871042251587,
+      "learning_rate": 4.902574793517051e-06,
+      "loss": 1.0479,
+      "step": 13460
+    },
+    {
+      "epoch": 0.09750483180959413,
+      "grad_norm": 0.18914766609668732,
+      "learning_rate": 4.902502406856464e-06,
+      "loss": 1.0657,
+      "step": 13470
+    },
+    {
+      "epoch": 0.09757721847018032,
+      "grad_norm": 0.16709379851818085,
+      "learning_rate": 4.902430020195879e-06,
+      "loss": 1.0522,
+      "step": 13480
+    },
+    {
+      "epoch": 0.09764960513076651,
+      "grad_norm": 0.178353950381279,
+      "learning_rate": 4.9023576335352924e-06,
+      "loss": 1.0385,
+      "step": 13490
+    },
+    {
+      "epoch": 0.09772199179135269,
+      "grad_norm": 0.18520021438598633,
+      "learning_rate": 4.902285246874706e-06,
+      "loss": 1.0431,
+      "step": 13500
+    },
+    {
+      "epoch": 0.09779437845193888,
+      "grad_norm": 0.1669263392686844,
+      "learning_rate": 4.90221286021412e-06,
+      "loss": 1.0552,
+      "step": 13510
+    },
+    {
+      "epoch": 0.09786676511252507,
+      "grad_norm": 0.1864238828420639,
+      "learning_rate": 4.902140473553534e-06,
+      "loss": 1.0536,
+      "step": 13520
+    },
+    {
+      "epoch": 0.09793915177311126,
+      "grad_norm": 0.1777682602405548,
+      "learning_rate": 4.902068086892948e-06,
+      "loss": 1.0604,
+      "step": 13530
+    },
+    {
+      "epoch": 0.09801153843369743,
+      "grad_norm": 0.1826665848493576,
+      "learning_rate": 4.901995700232361e-06,
+      "loss": 1.0642,
+      "step": 13540
+    },
+    {
+      "epoch": 0.09808392509428362,
+      "grad_norm": 0.17385782301425934,
+      "learning_rate": 4.901923313571775e-06,
+      "loss": 1.0643,
+      "step": 13550
+    },
+    {
+      "epoch": 0.09815631175486982,
+      "grad_norm": 0.18327641487121582,
+      "learning_rate": 4.9018509269111894e-06,
+      "loss": 1.0438,
+      "step": 13560
+    },
+    {
+      "epoch": 0.098228698415456,
+      "grad_norm": 0.18838289380073547,
+      "learning_rate": 4.901778540250603e-06,
+      "loss": 1.0529,
+      "step": 13570
+    },
+    {
+      "epoch": 0.09830108507604218,
+      "grad_norm": 0.1839076280593872,
+      "learning_rate": 4.901706153590017e-06,
+      "loss": 1.0553,
+      "step": 13580
+    },
+    {
+      "epoch": 0.09837347173662837,
+      "grad_norm": 0.21451406180858612,
+      "learning_rate": 4.90163376692943e-06,
+      "loss": 1.0549,
+      "step": 13590
+    },
+    {
+      "epoch": 0.09844585839721456,
+      "grad_norm": 0.1959172636270523,
+      "learning_rate": 4.901561380268845e-06,
+      "loss": 1.0824,
+      "step": 13600
+    },
+    {
+      "epoch": 0.09851824505780075,
+      "grad_norm": 0.17677393555641174,
+      "learning_rate": 4.901488993608258e-06,
+      "loss": 1.0723,
+      "step": 13610
+    },
+    {
+      "epoch": 0.09859063171838693,
+      "grad_norm": 0.17665037512779236,
+      "learning_rate": 4.901416606947672e-06,
+      "loss": 1.0616,
+      "step": 13620
+    },
+    {
+      "epoch": 0.09866301837897312,
+      "grad_norm": 0.20014891028404236,
+      "learning_rate": 4.901344220287086e-06,
+      "loss": 1.0484,
+      "step": 13630
+    },
+    {
+      "epoch": 0.09873540503955931,
+      "grad_norm": 0.18088600039482117,
+      "learning_rate": 4.9012718336265e-06,
+      "loss": 1.0507,
+      "step": 13640
+    },
+    {
+      "epoch": 0.0988077917001455,
+      "grad_norm": 0.19113531708717346,
+      "learning_rate": 4.901199446965914e-06,
+      "loss": 1.063,
+      "step": 13650
+    },
+    {
+      "epoch": 0.09888017836073168,
+      "grad_norm": 0.18885605037212372,
+      "learning_rate": 4.901127060305327e-06,
+      "loss": 1.0525,
+      "step": 13660
+    },
+    {
+      "epoch": 0.09895256502131787,
+      "grad_norm": 0.21783652901649475,
+      "learning_rate": 4.901054673644741e-06,
+      "loss": 1.0417,
+      "step": 13670
+    },
+    {
+      "epoch": 0.09902495168190406,
+      "grad_norm": 0.17940004169940948,
+      "learning_rate": 4.900982286984155e-06,
+      "loss": 1.0507,
+      "step": 13680
+    },
+    {
+      "epoch": 0.09909733834249025,
+      "grad_norm": 0.18767113983631134,
+      "learning_rate": 4.900909900323569e-06,
+      "loss": 1.0595,
+      "step": 13690
+    },
+    {
+      "epoch": 0.09916972500307643,
+      "grad_norm": 0.21696849167346954,
+      "learning_rate": 4.900837513662983e-06,
+      "loss": 1.0526,
+      "step": 13700
+    },
+    {
+      "epoch": 0.09924211166366262,
+      "grad_norm": 0.18748457729816437,
+      "learning_rate": 4.900765127002396e-06,
+      "loss": 1.0464,
+      "step": 13710
+    },
+    {
+      "epoch": 0.09931449832424881,
+      "grad_norm": 0.16807325184345245,
+      "learning_rate": 4.900692740341811e-06,
+      "loss": 1.0509,
+      "step": 13720
+    },
+    {
+      "epoch": 0.099386884984835,
+      "grad_norm": 0.19087384641170502,
+      "learning_rate": 4.900620353681224e-06,
+      "loss": 1.0599,
+      "step": 13730
+    },
+    {
+      "epoch": 0.09945927164542118,
+      "grad_norm": 0.2038954794406891,
+      "learning_rate": 4.900547967020638e-06,
+      "loss": 1.064,
+      "step": 13740
+    },
+    {
+      "epoch": 0.09953165830600737,
+      "grad_norm": 0.18865635991096497,
+      "learning_rate": 4.9004755803600515e-06,
+      "loss": 1.0605,
+      "step": 13750
+    },
+    {
+      "epoch": 0.09960404496659356,
+      "grad_norm": 0.1682719886302948,
+      "learning_rate": 4.900403193699466e-06,
+      "loss": 1.053,
+      "step": 13760
+    },
+    {
+      "epoch": 0.09967643162717975,
+      "grad_norm": 0.17732466757297516,
+      "learning_rate": 4.90033080703888e-06,
+      "loss": 1.0541,
+      "step": 13770
+    },
+    {
+      "epoch": 0.09974881828776593,
+      "grad_norm": 0.18870952725410461,
+      "learning_rate": 4.900258420378293e-06,
+      "loss": 1.0635,
+      "step": 13780
+    },
+    {
+      "epoch": 0.09982120494835212,
+      "grad_norm": 0.17982196807861328,
+      "learning_rate": 4.900186033717707e-06,
+      "loss": 1.064,
+      "step": 13790
+    },
+    {
+      "epoch": 0.0998935916089383,
+      "grad_norm": 0.17786167562007904,
+      "learning_rate": 4.900113647057121e-06,
+      "loss": 1.0619,
+      "step": 13800
+    },
+    {
+      "epoch": 0.0999659782695245,
+      "grad_norm": 0.1713530421257019,
+      "learning_rate": 4.900041260396535e-06,
+      "loss": 1.051,
+      "step": 13810
+    },
+    {
+      "epoch": 0.10003836493011067,
+      "grad_norm": 0.17207489907741547,
+      "learning_rate": 4.8999688737359485e-06,
+      "loss": 1.0744,
+      "step": 13820
+    },
+    {
+      "epoch": 0.10011075159069686,
+      "grad_norm": 0.18473108112812042,
+      "learning_rate": 4.899896487075362e-06,
+      "loss": 1.0505,
+      "step": 13830
+    },
+    {
+      "epoch": 0.10018313825128305,
+      "grad_norm": 0.16173668205738068,
+      "learning_rate": 4.899824100414776e-06,
+      "loss": 1.054,
+      "step": 13840
+    },
+    {
+      "epoch": 0.10025552491186925,
+      "grad_norm": 0.18166185915470123,
+      "learning_rate": 4.89975171375419e-06,
+      "loss": 1.0566,
+      "step": 13850
+    },
+    {
+      "epoch": 0.10032791157245542,
+      "grad_norm": 0.1847759187221527,
+      "learning_rate": 4.899679327093604e-06,
+      "loss": 1.0589,
+      "step": 13860
+    },
+    {
+      "epoch": 0.10040029823304161,
+      "grad_norm": 0.23197191953659058,
+      "learning_rate": 4.8996069404330175e-06,
+      "loss": 1.0468,
+      "step": 13870
+    },
+    {
+      "epoch": 0.1004726848936278,
+      "grad_norm": 0.1927204728126526,
+      "learning_rate": 4.899534553772431e-06,
+      "loss": 1.0602,
+      "step": 13880
+    },
+    {
+      "epoch": 0.100545071554214,
+      "grad_norm": 0.18343763053417206,
+      "learning_rate": 4.8994621671118456e-06,
+      "loss": 1.0436,
+      "step": 13890
+    },
+    {
+      "epoch": 0.10061745821480017,
+      "grad_norm": 0.18282248079776764,
+      "learning_rate": 4.899389780451258e-06,
+      "loss": 1.0505,
+      "step": 13900
+    },
+    {
+      "epoch": 0.10068984487538636,
+      "grad_norm": 0.17692866921424866,
+      "learning_rate": 4.899317393790673e-06,
+      "loss": 1.0826,
+      "step": 13910
+    },
+    {
+      "epoch": 0.10076223153597255,
+      "grad_norm": 0.17446507513523102,
+      "learning_rate": 4.899245007130086e-06,
+      "loss": 1.06,
+      "step": 13920
+    },
+    {
+      "epoch": 0.10083461819655874,
+      "grad_norm": 0.1633169800043106,
+      "learning_rate": 4.8991726204695e-06,
+      "loss": 1.0358,
+      "step": 13930
+    },
+    {
+      "epoch": 0.10090700485714492,
+      "grad_norm": 0.17512056231498718,
+      "learning_rate": 4.899100233808914e-06,
+      "loss": 1.0552,
+      "step": 13940
+    },
+    {
+      "epoch": 0.10097939151773111,
+      "grad_norm": 0.1872796267271042,
+      "learning_rate": 4.899027847148328e-06,
+      "loss": 1.0467,
+      "step": 13950
+    },
+    {
+      "epoch": 0.1010517781783173,
+      "grad_norm": 0.17610350251197815,
+      "learning_rate": 4.898955460487742e-06,
+      "loss": 1.0536,
+      "step": 13960
+    },
+    {
+      "epoch": 0.10112416483890349,
+      "grad_norm": 0.16492041945457458,
+      "learning_rate": 4.898883073827155e-06,
+      "loss": 1.0724,
+      "step": 13970
+    },
+    {
+      "epoch": 0.10119655149948967,
+      "grad_norm": 0.16583912074565887,
+      "learning_rate": 4.898810687166569e-06,
+      "loss": 1.0532,
+      "step": 13980
+    },
+    {
+      "epoch": 0.10126893816007586,
+      "grad_norm": 0.20664115250110626,
+      "learning_rate": 4.8987383005059826e-06,
+      "loss": 1.0643,
+      "step": 13990
+    },
+    {
+      "epoch": 0.10134132482066205,
+      "grad_norm": 0.17064620554447174,
+      "learning_rate": 4.898665913845397e-06,
+      "loss": 1.042,
+      "step": 14000
+    },
+    {
+      "epoch": 0.10141371148124824,
+      "grad_norm": 0.17799104750156403,
+      "learning_rate": 4.898593527184811e-06,
+      "loss": 1.0575,
+      "step": 14010
+    },
+    {
+      "epoch": 0.10148609814183442,
+      "grad_norm": 0.16274908185005188,
+      "learning_rate": 4.898521140524224e-06,
+      "loss": 1.0514,
+      "step": 14020
+    },
+    {
+      "epoch": 0.1015584848024206,
+      "grad_norm": 0.1757798045873642,
+      "learning_rate": 4.898448753863638e-06,
+      "loss": 1.0512,
+      "step": 14030
+    },
+    {
+      "epoch": 0.1016308714630068,
+      "grad_norm": 0.18405269086360931,
+      "learning_rate": 4.898376367203052e-06,
+      "loss": 1.0545,
+      "step": 14040
+    },
+    {
+      "epoch": 0.10170325812359299,
+      "grad_norm": 0.1723915934562683,
+      "learning_rate": 4.898303980542466e-06,
+      "loss": 1.0563,
+      "step": 14050
+    },
+    {
+      "epoch": 0.10177564478417918,
+      "grad_norm": 0.18672221899032593,
+      "learning_rate": 4.8982315938818796e-06,
+      "loss": 1.0542,
+      "step": 14060
+    },
+    {
+      "epoch": 0.10184803144476536,
+      "grad_norm": 0.20487704873085022,
+      "learning_rate": 4.898159207221293e-06,
+      "loss": 1.0567,
+      "step": 14070
+    },
+    {
+      "epoch": 0.10192041810535155,
+      "grad_norm": 0.1787646859884262,
+      "learning_rate": 4.898086820560708e-06,
+      "loss": 1.0683,
+      "step": 14080
+    },
+    {
+      "epoch": 0.10199280476593774,
+      "grad_norm": 0.18574008345603943,
+      "learning_rate": 4.898014433900121e-06,
+      "loss": 1.0517,
+      "step": 14090
+    },
+    {
+      "epoch": 0.10206519142652393,
+      "grad_norm": 0.17313173413276672,
+      "learning_rate": 4.897942047239535e-06,
+      "loss": 1.0451,
+      "step": 14100
+    },
+    {
+      "epoch": 0.1021375780871101,
+      "grad_norm": 0.18771642446517944,
+      "learning_rate": 4.8978696605789485e-06,
+      "loss": 1.0599,
+      "step": 14110
+    },
+    {
+      "epoch": 0.1022099647476963,
+      "grad_norm": 0.22109109163284302,
+      "learning_rate": 4.897797273918363e-06,
+      "loss": 1.0622,
+      "step": 14120
+    },
+    {
+      "epoch": 0.10228235140828248,
+      "grad_norm": 0.1844063252210617,
+      "learning_rate": 4.8977248872577766e-06,
+      "loss": 1.0602,
+      "step": 14130
+    },
+    {
+      "epoch": 0.10235473806886868,
+      "grad_norm": 0.18542051315307617,
+      "learning_rate": 4.89765250059719e-06,
+      "loss": 1.0554,
+      "step": 14140
+    },
+    {
+      "epoch": 0.10242712472945485,
+      "grad_norm": 0.17631708085536957,
+      "learning_rate": 4.897580113936604e-06,
+      "loss": 1.053,
+      "step": 14150
+    },
+    {
+      "epoch": 0.10249951139004104,
+      "grad_norm": 0.17057740688323975,
+      "learning_rate": 4.897507727276018e-06,
+      "loss": 1.0535,
+      "step": 14160
+    },
+    {
+      "epoch": 0.10257189805062723,
+      "grad_norm": 0.17483165860176086,
+      "learning_rate": 4.897435340615432e-06,
+      "loss": 1.0433,
+      "step": 14170
+    },
+    {
+      "epoch": 0.10264428471121342,
+      "grad_norm": 0.18016666173934937,
+      "learning_rate": 4.8973629539548455e-06,
+      "loss": 1.031,
+      "step": 14180
+    },
+    {
+      "epoch": 0.1027166713717996,
+      "grad_norm": 0.1845754235982895,
+      "learning_rate": 4.897290567294259e-06,
+      "loss": 1.0498,
+      "step": 14190
+    },
+    {
+      "epoch": 0.10278905803238579,
+      "grad_norm": 0.1925055980682373,
+      "learning_rate": 4.897218180633674e-06,
+      "loss": 1.0588,
+      "step": 14200
+    },
+    {
+      "epoch": 0.10286144469297198,
+      "grad_norm": 0.19144387543201447,
+      "learning_rate": 4.897145793973087e-06,
+      "loss": 1.0514,
+      "step": 14210
+    },
+    {
+      "epoch": 0.10293383135355817,
+      "grad_norm": 0.1973779946565628,
+      "learning_rate": 4.897073407312501e-06,
+      "loss": 1.0468,
+      "step": 14220
+    },
+    {
+      "epoch": 0.10300621801414435,
+      "grad_norm": 0.16684280335903168,
+      "learning_rate": 4.8970010206519144e-06,
+      "loss": 1.0578,
+      "step": 14230
+    },
+    {
+      "epoch": 0.10307860467473054,
+      "grad_norm": 0.22712835669517517,
+      "learning_rate": 4.896928633991329e-06,
+      "loss": 1.055,
+      "step": 14240
+    },
+    {
+      "epoch": 0.10315099133531673,
+      "grad_norm": 0.189208522439003,
+      "learning_rate": 4.8968562473307425e-06,
+      "loss": 1.0338,
+      "step": 14250
+    },
+    {
+      "epoch": 0.10322337799590292,
+      "grad_norm": 0.18186455965042114,
+      "learning_rate": 4.896783860670156e-06,
+      "loss": 1.0497,
+      "step": 14260
+    },
+    {
+      "epoch": 0.1032957646564891,
+      "grad_norm": 0.1724977344274521,
+      "learning_rate": 4.89671147400957e-06,
+      "loss": 1.0475,
+      "step": 14270
+    },
+    {
+      "epoch": 0.10336815131707529,
+      "grad_norm": 0.1740335077047348,
+      "learning_rate": 4.896639087348984e-06,
+      "loss": 1.0406,
+      "step": 14280
+    },
+    {
+      "epoch": 0.10344053797766148,
+      "grad_norm": 0.21872451901435852,
+      "learning_rate": 4.896566700688398e-06,
+      "loss": 1.0514,
+      "step": 14290
+    },
+    {
+      "epoch": 0.10351292463824767,
+      "grad_norm": 0.18021239340305328,
+      "learning_rate": 4.8964943140278114e-06,
+      "loss": 1.0619,
+      "step": 14300
+    },
+    {
+      "epoch": 0.10358531129883385,
+      "grad_norm": 0.17084094882011414,
+      "learning_rate": 4.896421927367225e-06,
+      "loss": 1.0567,
+      "step": 14310
+    },
+    {
+      "epoch": 0.10365769795942004,
+      "grad_norm": 0.18873821198940277,
+      "learning_rate": 4.8963495407066395e-06,
+      "loss": 1.0565,
+      "step": 14320
+    },
+    {
+      "epoch": 0.10373008462000623,
+      "grad_norm": 0.16363616287708282,
+      "learning_rate": 4.896277154046053e-06,
+      "loss": 1.0444,
+      "step": 14330
+    },
+    {
+      "epoch": 0.10380247128059242,
+      "grad_norm": 0.20980127155780792,
+      "learning_rate": 4.896204767385467e-06,
+      "loss": 1.0515,
+      "step": 14340
+    },
+    {
+      "epoch": 0.1038748579411786,
+      "grad_norm": 0.17421722412109375,
+      "learning_rate": 4.89613238072488e-06,
+      "loss": 1.0513,
+      "step": 14350
+    },
+    {
+      "epoch": 0.10394724460176478,
+      "grad_norm": 0.20598512887954712,
+      "learning_rate": 4.896059994064295e-06,
+      "loss": 1.0549,
+      "step": 14360
+    },
+    {
+      "epoch": 0.10401963126235098,
+      "grad_norm": 0.1713586449623108,
+      "learning_rate": 4.8959876074037084e-06,
+      "loss": 1.0456,
+      "step": 14370
+    },
+    {
+      "epoch": 0.10409201792293717,
+      "grad_norm": 0.1770513504743576,
+      "learning_rate": 4.895915220743122e-06,
+      "loss": 1.0492,
+      "step": 14380
+    },
+    {
+      "epoch": 0.10416440458352334,
+      "grad_norm": 0.18118791282176971,
+      "learning_rate": 4.895842834082536e-06,
+      "loss": 1.0547,
+      "step": 14390
+    },
+    {
+      "epoch": 0.10423679124410953,
+      "grad_norm": 0.18534211814403534,
+      "learning_rate": 4.89577044742195e-06,
+      "loss": 1.057,
+      "step": 14400
+    },
+    {
+      "epoch": 0.10430917790469572,
+      "grad_norm": 0.20665541291236877,
+      "learning_rate": 4.895698060761364e-06,
+      "loss": 1.0455,
+      "step": 14410
+    },
+    {
+      "epoch": 0.10438156456528191,
+      "grad_norm": 0.17217250168323517,
+      "learning_rate": 4.895625674100777e-06,
+      "loss": 1.0524,
+      "step": 14420
+    },
+    {
+      "epoch": 0.10445395122586809,
+      "grad_norm": 0.1958875209093094,
+      "learning_rate": 4.895553287440191e-06,
+      "loss": 1.0457,
+      "step": 14430
+    },
+    {
+      "epoch": 0.10452633788645428,
+      "grad_norm": 0.18812295794487,
+      "learning_rate": 4.895480900779605e-06,
+      "loss": 1.0523,
+      "step": 14440
+    },
+    {
+      "epoch": 0.10459872454704047,
+      "grad_norm": 0.16309994459152222,
+      "learning_rate": 4.895408514119018e-06,
+      "loss": 1.0446,
+      "step": 14450
+    },
+    {
+      "epoch": 0.10467111120762666,
+      "grad_norm": 0.1937844455242157,
+      "learning_rate": 4.895336127458432e-06,
+      "loss": 1.0534,
+      "step": 14460
+    },
+    {
+      "epoch": 0.10474349786821284,
+      "grad_norm": 0.17462307214736938,
+      "learning_rate": 4.895263740797846e-06,
+      "loss": 1.0362,
+      "step": 14470
+    },
+    {
+      "epoch": 0.10481588452879903,
+      "grad_norm": 0.17586156725883484,
+      "learning_rate": 4.89519135413726e-06,
+      "loss": 1.034,
+      "step": 14480
+    },
+    {
+      "epoch": 0.10488827118938522,
+      "grad_norm": 0.1698175072669983,
+      "learning_rate": 4.8951189674766735e-06,
+      "loss": 1.0439,
+      "step": 14490
+    },
+    {
+      "epoch": 0.10496065784997141,
+      "grad_norm": 0.18947486579418182,
+      "learning_rate": 4.895046580816087e-06,
+      "loss": 1.0507,
+      "step": 14500
+    },
+    {
+      "epoch": 0.10503304451055759,
+      "grad_norm": 0.17008043825626373,
+      "learning_rate": 4.894974194155502e-06,
+      "loss": 1.0368,
+      "step": 14510
+    },
+    {
+      "epoch": 0.10510543117114378,
+      "grad_norm": 0.18763844668865204,
+      "learning_rate": 4.894901807494915e-06,
+      "loss": 1.0634,
+      "step": 14520
+    },
+    {
+      "epoch": 0.10517781783172997,
+      "grad_norm": 0.18135419487953186,
+      "learning_rate": 4.894829420834329e-06,
+      "loss": 1.0502,
+      "step": 14530
+    },
+    {
+      "epoch": 0.10525020449231616,
+      "grad_norm": 0.18853351473808289,
+      "learning_rate": 4.8947570341737425e-06,
+      "loss": 1.0538,
+      "step": 14540
+    },
+    {
+      "epoch": 0.10532259115290234,
+      "grad_norm": 0.17148560285568237,
+      "learning_rate": 4.894684647513157e-06,
+      "loss": 1.0524,
+      "step": 14550
+    },
+    {
+      "epoch": 0.10539497781348853,
+      "grad_norm": 0.18300309777259827,
+      "learning_rate": 4.8946122608525705e-06,
+      "loss": 1.0528,
+      "step": 14560
+    },
+    {
+      "epoch": 0.10546736447407472,
+      "grad_norm": 0.17703156173229218,
+      "learning_rate": 4.894539874191984e-06,
+      "loss": 1.057,
+      "step": 14570
+    },
+    {
+      "epoch": 0.10553975113466091,
+      "grad_norm": 0.18114346265792847,
+      "learning_rate": 4.894467487531398e-06,
+      "loss": 1.0477,
+      "step": 14580
+    },
+    {
+      "epoch": 0.10561213779524709,
+      "grad_norm": 0.18363863229751587,
+      "learning_rate": 4.894395100870812e-06,
+      "loss": 1.0485,
+      "step": 14590
+    },
+    {
+      "epoch": 0.10568452445583328,
+      "grad_norm": 0.17243093252182007,
+      "learning_rate": 4.894322714210226e-06,
+      "loss": 1.0512,
+      "step": 14600
+    },
+    {
+      "epoch": 0.10575691111641947,
+      "grad_norm": 0.16274864971637726,
+      "learning_rate": 4.8942503275496395e-06,
+      "loss": 1.0321,
+      "step": 14610
+    },
+    {
+      "epoch": 0.10582929777700566,
+      "grad_norm": 0.1914278268814087,
+      "learning_rate": 4.894177940889053e-06,
+      "loss": 1.0623,
+      "step": 14620
+    },
+    {
+      "epoch": 0.10590168443759183,
+      "grad_norm": 0.187461256980896,
+      "learning_rate": 4.894105554228467e-06,
+      "loss": 1.0426,
+      "step": 14630
+    },
+    {
+      "epoch": 0.10597407109817802,
+      "grad_norm": 0.17549605667591095,
+      "learning_rate": 4.894033167567881e-06,
+      "loss": 1.0341,
+      "step": 14640
+    },
+    {
+      "epoch": 0.10604645775876421,
+      "grad_norm": 0.17880699038505554,
+      "learning_rate": 4.893960780907295e-06,
+      "loss": 1.0549,
+      "step": 14650
+    },
+    {
+      "epoch": 0.1061188444193504,
+      "grad_norm": 0.17565539479255676,
+      "learning_rate": 4.893888394246708e-06,
+      "loss": 1.048,
+      "step": 14660
+    },
+    {
+      "epoch": 0.1061912310799366,
+      "grad_norm": 0.17139199376106262,
+      "learning_rate": 4.893816007586122e-06,
+      "loss": 1.0557,
+      "step": 14670
+    },
+    {
+      "epoch": 0.10626361774052277,
+      "grad_norm": 0.18786922097206116,
+      "learning_rate": 4.8937436209255365e-06,
+      "loss": 1.059,
+      "step": 14680
+    },
+    {
+      "epoch": 0.10633600440110896,
+      "grad_norm": 0.18301083147525787,
+      "learning_rate": 4.89367123426495e-06,
+      "loss": 1.0397,
+      "step": 14690
+    },
+    {
+      "epoch": 0.10640839106169515,
+      "grad_norm": 0.1840226799249649,
+      "learning_rate": 4.893598847604364e-06,
+      "loss": 1.0548,
+      "step": 14700
+    },
+    {
+      "epoch": 0.10648077772228134,
+      "grad_norm": 0.18379652500152588,
+      "learning_rate": 4.893526460943777e-06,
+      "loss": 1.0441,
+      "step": 14710
+    },
+    {
+      "epoch": 0.10655316438286752,
+      "grad_norm": 0.17776694893836975,
+      "learning_rate": 4.893454074283192e-06,
+      "loss": 1.0594,
+      "step": 14720
+    },
+    {
+      "epoch": 0.10662555104345371,
+      "grad_norm": 0.17674773931503296,
+      "learning_rate": 4.893381687622605e-06,
+      "loss": 1.0594,
+      "step": 14730
+    },
+    {
+      "epoch": 0.1066979377040399,
+      "grad_norm": 0.17039507627487183,
+      "learning_rate": 4.893309300962019e-06,
+      "loss": 1.0525,
+      "step": 14740
+    },
+    {
+      "epoch": 0.10677032436462609,
+      "grad_norm": 0.16657809913158417,
+      "learning_rate": 4.893236914301433e-06,
+      "loss": 1.0669,
+      "step": 14750
+    },
+    {
+      "epoch": 0.10684271102521227,
+      "grad_norm": 0.17446079850196838,
+      "learning_rate": 4.893164527640847e-06,
+      "loss": 1.0464,
+      "step": 14760
+    },
+    {
+      "epoch": 0.10691509768579846,
+      "grad_norm": 0.16530625522136688,
+      "learning_rate": 4.893092140980261e-06,
+      "loss": 1.0426,
+      "step": 14770
+    },
+    {
+      "epoch": 0.10698748434638465,
+      "grad_norm": 0.180899515748024,
+      "learning_rate": 4.893019754319674e-06,
+      "loss": 1.0426,
+      "step": 14780
+    },
+    {
+      "epoch": 0.10705987100697084,
+      "grad_norm": 0.18140670657157898,
+      "learning_rate": 4.892947367659088e-06,
+      "loss": 1.0468,
+      "step": 14790
+    },
+    {
+      "epoch": 0.10713225766755702,
+      "grad_norm": 0.1905767023563385,
+      "learning_rate": 4.892874980998502e-06,
+      "loss": 1.0518,
+      "step": 14800
+    },
+    {
+      "epoch": 0.10720464432814321,
+      "grad_norm": 0.17752863466739655,
+      "learning_rate": 4.892802594337916e-06,
+      "loss": 1.048,
+      "step": 14810
+    },
+    {
+      "epoch": 0.1072770309887294,
+      "grad_norm": 0.16803748905658722,
+      "learning_rate": 4.89273020767733e-06,
+      "loss": 1.0472,
+      "step": 14820
+    },
+    {
+      "epoch": 0.10734941764931559,
+      "grad_norm": 0.17945846915245056,
+      "learning_rate": 4.892657821016743e-06,
+      "loss": 1.0396,
+      "step": 14830
+    },
+    {
+      "epoch": 0.10742180430990177,
+      "grad_norm": 0.19865509867668152,
+      "learning_rate": 4.892585434356158e-06,
+      "loss": 1.0592,
+      "step": 14840
+    },
+    {
+      "epoch": 0.10749419097048796,
+      "grad_norm": 0.18445666134357452,
+      "learning_rate": 4.892513047695571e-06,
+      "loss": 1.0411,
+      "step": 14850
+    },
+    {
+      "epoch": 0.10756657763107415,
+      "grad_norm": 0.1645526885986328,
+      "learning_rate": 4.892440661034985e-06,
+      "loss": 1.0588,
+      "step": 14860
+    },
+    {
+      "epoch": 0.10763896429166034,
+      "grad_norm": 0.1656254678964615,
+      "learning_rate": 4.8923682743743986e-06,
+      "loss": 1.0547,
+      "step": 14870
+    },
+    {
+      "epoch": 0.10771135095224652,
+      "grad_norm": 0.1796533614397049,
+      "learning_rate": 4.892295887713813e-06,
+      "loss": 1.0504,
+      "step": 14880
+    },
+    {
+      "epoch": 0.1077837376128327,
+      "grad_norm": 0.17627054452896118,
+      "learning_rate": 4.892223501053227e-06,
+      "loss": 1.064,
+      "step": 14890
+    },
+    {
+      "epoch": 0.1078561242734189,
+      "grad_norm": 0.1647876799106598,
+      "learning_rate": 4.89215111439264e-06,
+      "loss": 1.0305,
+      "step": 14900
+    },
+    {
+      "epoch": 0.10792851093400509,
+      "grad_norm": 0.17485207319259644,
+      "learning_rate": 4.892078727732054e-06,
+      "loss": 1.058,
+      "step": 14910
+    },
+    {
+      "epoch": 0.10800089759459126,
+      "grad_norm": 0.18584197759628296,
+      "learning_rate": 4.892006341071468e-06,
+      "loss": 1.0459,
+      "step": 14920
+    },
+    {
+      "epoch": 0.10807328425517745,
+      "grad_norm": 0.17022201418876648,
+      "learning_rate": 4.891933954410882e-06,
+      "loss": 1.0758,
+      "step": 14930
+    },
+    {
+      "epoch": 0.10814567091576364,
+      "grad_norm": 0.159866064786911,
+      "learning_rate": 4.891861567750296e-06,
+      "loss": 1.0397,
+      "step": 14940
+    },
+    {
+      "epoch": 0.10821805757634984,
+      "grad_norm": 0.16421310603618622,
+      "learning_rate": 4.891789181089709e-06,
+      "loss": 1.0472,
+      "step": 14950
+    },
+    {
+      "epoch": 0.10829044423693601,
+      "grad_norm": 0.17276668548583984,
+      "learning_rate": 4.891716794429123e-06,
+      "loss": 1.0433,
+      "step": 14960
+    },
+    {
+      "epoch": 0.1083628308975222,
+      "grad_norm": 0.1813632696866989,
+      "learning_rate": 4.8916444077685364e-06,
+      "loss": 1.0355,
+      "step": 14970
+    },
+    {
+      "epoch": 0.1084352175581084,
+      "grad_norm": 0.16496697068214417,
+      "learning_rate": 4.89157202110795e-06,
+      "loss": 1.0368,
+      "step": 14980
+    },
+    {
+      "epoch": 0.10850760421869458,
+      "grad_norm": 0.1636180728673935,
+      "learning_rate": 4.8914996344473645e-06,
+      "loss": 1.042,
+      "step": 14990
+    },
+    {
+      "epoch": 0.10857999087928076,
+      "grad_norm": 0.1825898438692093,
+      "learning_rate": 4.891427247786778e-06,
+      "loss": 1.0519,
+      "step": 15000
+    },
+    {
+      "epoch": 0.10865237753986695,
+      "grad_norm": 0.19418931007385254,
+      "learning_rate": 4.891354861126192e-06,
+      "loss": 1.0613,
+      "step": 15010
+    },
+    {
+      "epoch": 0.10872476420045314,
+      "grad_norm": 0.1902722269296646,
+      "learning_rate": 4.891282474465605e-06,
+      "loss": 1.0433,
+      "step": 15020
+    },
+    {
+      "epoch": 0.10879715086103933,
+      "grad_norm": 0.1725703924894333,
+      "learning_rate": 4.89121008780502e-06,
+      "loss": 1.0537,
+      "step": 15030
+    },
+    {
+      "epoch": 0.10886953752162551,
+      "grad_norm": 0.2096461057662964,
+      "learning_rate": 4.8911377011444334e-06,
+      "loss": 1.0416,
+      "step": 15040
+    },
+    {
+      "epoch": 0.1089419241822117,
+      "grad_norm": 0.1624659150838852,
+      "learning_rate": 4.891065314483847e-06,
+      "loss": 1.0572,
+      "step": 15050
+    },
+    {
+      "epoch": 0.10901431084279789,
+      "grad_norm": 0.1570434719324112,
+      "learning_rate": 4.890992927823261e-06,
+      "loss": 1.0396,
+      "step": 15060
+    },
+    {
+      "epoch": 0.10908669750338408,
+      "grad_norm": 0.17029106616973877,
+      "learning_rate": 4.890920541162675e-06,
+      "loss": 1.0596,
+      "step": 15070
+    },
+    {
+      "epoch": 0.10915908416397026,
+      "grad_norm": 0.16916610300540924,
+      "learning_rate": 4.890848154502089e-06,
+      "loss": 1.0534,
+      "step": 15080
+    },
+    {
+      "epoch": 0.10923147082455645,
+      "grad_norm": 0.18326231837272644,
+      "learning_rate": 4.890775767841502e-06,
+      "loss": 1.0517,
+      "step": 15090
+    },
+    {
+      "epoch": 0.10930385748514264,
+      "grad_norm": 0.2638685405254364,
+      "learning_rate": 4.890703381180916e-06,
+      "loss": 1.0556,
+      "step": 15100
+    },
+    {
+      "epoch": 0.10937624414572883,
+      "grad_norm": 0.17363910377025604,
+      "learning_rate": 4.8906309945203304e-06,
+      "loss": 1.0463,
+      "step": 15110
+    },
+    {
+      "epoch": 0.109448630806315,
+      "grad_norm": 0.17666375637054443,
+      "learning_rate": 4.890558607859744e-06,
+      "loss": 1.0603,
+      "step": 15120
+    },
+    {
+      "epoch": 0.1095210174669012,
+      "grad_norm": 0.16066348552703857,
+      "learning_rate": 4.890486221199158e-06,
+      "loss": 1.0578,
+      "step": 15130
+    },
+    {
+      "epoch": 0.10959340412748739,
+      "grad_norm": 0.18254084885120392,
+      "learning_rate": 4.890413834538571e-06,
+      "loss": 1.0439,
+      "step": 15140
+    },
+    {
+      "epoch": 0.10966579078807358,
+      "grad_norm": 0.18753020465373993,
+      "learning_rate": 4.890341447877986e-06,
+      "loss": 1.0529,
+      "step": 15150
+    },
+    {
+      "epoch": 0.10973817744865975,
+      "grad_norm": 0.21900314092636108,
+      "learning_rate": 4.890269061217399e-06,
+      "loss": 1.0389,
+      "step": 15160
+    },
+    {
+      "epoch": 0.10981056410924595,
+      "grad_norm": 0.17943327128887177,
+      "learning_rate": 4.890196674556813e-06,
+      "loss": 1.0558,
+      "step": 15170
+    },
+    {
+      "epoch": 0.10988295076983214,
+      "grad_norm": 0.16995948553085327,
+      "learning_rate": 4.890124287896227e-06,
+      "loss": 1.0474,
+      "step": 15180
+    },
+    {
+      "epoch": 0.10995533743041833,
+      "grad_norm": 0.17363294959068298,
+      "learning_rate": 4.890051901235641e-06,
+      "loss": 1.0629,
+      "step": 15190
+    },
+    {
+      "epoch": 0.1100277240910045,
+      "grad_norm": 0.1756804883480072,
+      "learning_rate": 4.889979514575055e-06,
+      "loss": 1.0547,
+      "step": 15200
+    },
+    {
+      "epoch": 0.1101001107515907,
+      "grad_norm": 0.1712425798177719,
+      "learning_rate": 4.889907127914468e-06,
+      "loss": 1.047,
+      "step": 15210
+    },
+    {
+      "epoch": 0.11017249741217688,
+      "grad_norm": 0.1906772404909134,
+      "learning_rate": 4.889834741253882e-06,
+      "loss": 1.0554,
+      "step": 15220
+    },
+    {
+      "epoch": 0.11024488407276307,
+      "grad_norm": 0.1886027455329895,
+      "learning_rate": 4.8897623545932955e-06,
+      "loss": 1.0429,
+      "step": 15230
+    },
+    {
+      "epoch": 0.11031727073334927,
+      "grad_norm": 0.17662407457828522,
+      "learning_rate": 4.88968996793271e-06,
+      "loss": 1.0659,
+      "step": 15240
+    },
+    {
+      "epoch": 0.11038965739393544,
+      "grad_norm": 0.1786973476409912,
+      "learning_rate": 4.889617581272124e-06,
+      "loss": 1.0657,
+      "step": 15250
+    },
+    {
+      "epoch": 0.11046204405452163,
+      "grad_norm": 0.18255244195461273,
+      "learning_rate": 4.889545194611537e-06,
+      "loss": 1.0411,
+      "step": 15260
+    },
+    {
+      "epoch": 0.11053443071510782,
+      "grad_norm": 0.18926455080509186,
+      "learning_rate": 4.889472807950951e-06,
+      "loss": 1.0492,
+      "step": 15270
+    },
+    {
+      "epoch": 0.11060681737569401,
+      "grad_norm": 0.17178624868392944,
+      "learning_rate": 4.889400421290365e-06,
+      "loss": 1.0424,
+      "step": 15280
+    },
+    {
+      "epoch": 0.11067920403628019,
+      "grad_norm": 0.17576183378696442,
+      "learning_rate": 4.889328034629779e-06,
+      "loss": 1.0464,
+      "step": 15290
+    },
+    {
+      "epoch": 0.11075159069686638,
+      "grad_norm": 0.19050145149230957,
+      "learning_rate": 4.8892556479691925e-06,
+      "loss": 1.0531,
+      "step": 15300
+    },
+    {
+      "epoch": 0.11082397735745257,
+      "grad_norm": 0.17656132578849792,
+      "learning_rate": 4.889183261308606e-06,
+      "loss": 1.0312,
+      "step": 15310
+    },
+    {
+      "epoch": 0.11089636401803876,
+      "grad_norm": 0.20985843241214752,
+      "learning_rate": 4.889110874648021e-06,
+      "loss": 1.0405,
+      "step": 15320
+    },
+    {
+      "epoch": 0.11096875067862494,
+      "grad_norm": 0.17128346860408783,
+      "learning_rate": 4.889038487987434e-06,
+      "loss": 1.0542,
+      "step": 15330
+    },
+    {
+      "epoch": 0.11104113733921113,
+      "grad_norm": 0.18605564534664154,
+      "learning_rate": 4.888966101326848e-06,
+      "loss": 1.0492,
+      "step": 15340
+    },
+    {
+      "epoch": 0.11111352399979732,
+      "grad_norm": 0.17718972265720367,
+      "learning_rate": 4.8888937146662615e-06,
+      "loss": 1.0379,
+      "step": 15350
+    },
+    {
+      "epoch": 0.11118591066038351,
+      "grad_norm": 0.263354629278183,
+      "learning_rate": 4.888821328005676e-06,
+      "loss": 1.0562,
+      "step": 15360
+    },
+    {
+      "epoch": 0.11125829732096969,
+      "grad_norm": 0.2226659655570984,
+      "learning_rate": 4.8887489413450895e-06,
+      "loss": 1.0404,
+      "step": 15370
+    },
+    {
+      "epoch": 0.11133068398155588,
+      "grad_norm": 0.18347443640232086,
+      "learning_rate": 4.888676554684503e-06,
+      "loss": 1.0458,
+      "step": 15380
+    },
+    {
+      "epoch": 0.11140307064214207,
+      "grad_norm": 0.2046961486339569,
+      "learning_rate": 4.888604168023917e-06,
+      "loss": 1.0678,
+      "step": 15390
+    },
+    {
+      "epoch": 0.11147545730272826,
+      "grad_norm": 0.18448476493358612,
+      "learning_rate": 4.888531781363331e-06,
+      "loss": 1.0627,
+      "step": 15400
+    },
+    {
+      "epoch": 0.11154784396331444,
+      "grad_norm": 0.17062750458717346,
+      "learning_rate": 4.888459394702745e-06,
+      "loss": 1.0404,
+      "step": 15410
+    },
+    {
+      "epoch": 0.11162023062390063,
+      "grad_norm": 0.18066982924938202,
+      "learning_rate": 4.8883870080421585e-06,
+      "loss": 1.0386,
+      "step": 15420
+    },
+    {
+      "epoch": 0.11169261728448682,
+      "grad_norm": 0.18624179065227509,
+      "learning_rate": 4.888314621381572e-06,
+      "loss": 1.0547,
+      "step": 15430
+    },
+    {
+      "epoch": 0.11176500394507301,
+      "grad_norm": 0.23639068007469177,
+      "learning_rate": 4.8882422347209866e-06,
+      "loss": 1.0424,
+      "step": 15440
+    },
+    {
+      "epoch": 0.11183739060565918,
+      "grad_norm": 0.2034512162208557,
+      "learning_rate": 4.8881698480604e-06,
+      "loss": 1.0349,
+      "step": 15450
+    },
+    {
+      "epoch": 0.11190977726624537,
+      "grad_norm": 0.1779569387435913,
+      "learning_rate": 4.888097461399814e-06,
+      "loss": 1.0517,
+      "step": 15460
+    },
+    {
+      "epoch": 0.11198216392683157,
+      "grad_norm": 0.1710672378540039,
+      "learning_rate": 4.888025074739227e-06,
+      "loss": 1.0476,
+      "step": 15470
+    },
+    {
+      "epoch": 0.11205455058741776,
+      "grad_norm": 0.1749650537967682,
+      "learning_rate": 4.887952688078642e-06,
+      "loss": 1.0481,
+      "step": 15480
+    },
+    {
+      "epoch": 0.11212693724800393,
+      "grad_norm": 0.17475520074367523,
+      "learning_rate": 4.887880301418055e-06,
+      "loss": 1.0611,
+      "step": 15490
+    },
+    {
+      "epoch": 0.11219932390859012,
+      "grad_norm": 0.16181065142154694,
+      "learning_rate": 4.887807914757468e-06,
+      "loss": 1.0518,
+      "step": 15500
+    },
+    {
+      "epoch": 0.11227171056917631,
+      "grad_norm": 0.2130517065525055,
+      "learning_rate": 4.887735528096883e-06,
+      "loss": 1.0441,
+      "step": 15510
+    },
+    {
+      "epoch": 0.1123440972297625,
+      "grad_norm": 0.18250152468681335,
+      "learning_rate": 4.887663141436296e-06,
+      "loss": 1.0524,
+      "step": 15520
+    },
+    {
+      "epoch": 0.11241648389034868,
+      "grad_norm": 0.17363341152668,
+      "learning_rate": 4.88759075477571e-06,
+      "loss": 1.0465,
+      "step": 15530
+    },
+    {
+      "epoch": 0.11248887055093487,
+      "grad_norm": 0.17302308976650238,
+      "learning_rate": 4.8875183681151236e-06,
+      "loss": 1.0467,
+      "step": 15540
+    },
+    {
+      "epoch": 0.11256125721152106,
+      "grad_norm": 0.18607738614082336,
+      "learning_rate": 4.887445981454538e-06,
+      "loss": 1.0449,
+      "step": 15550
+    },
+    {
+      "epoch": 0.11263364387210725,
+      "grad_norm": 0.2096407413482666,
+      "learning_rate": 4.887373594793952e-06,
+      "loss": 1.0541,
+      "step": 15560
+    },
+    {
+      "epoch": 0.11270603053269343,
+      "grad_norm": 0.17433767020702362,
+      "learning_rate": 4.887301208133365e-06,
+      "loss": 1.052,
+      "step": 15570
+    },
+    {
+      "epoch": 0.11277841719327962,
+      "grad_norm": 0.16963066160678864,
+      "learning_rate": 4.887228821472779e-06,
+      "loss": 1.0411,
+      "step": 15580
+    },
+    {
+      "epoch": 0.11285080385386581,
+      "grad_norm": 0.19870556890964508,
+      "learning_rate": 4.887156434812193e-06,
+      "loss": 1.0467,
+      "step": 15590
+    },
+    {
+      "epoch": 0.112923190514452,
+      "grad_norm": 0.17876464128494263,
+      "learning_rate": 4.887084048151607e-06,
+      "loss": 1.0401,
+      "step": 15600
+    },
+    {
+      "epoch": 0.11299557717503818,
+      "grad_norm": 0.15952976047992706,
+      "learning_rate": 4.8870116614910206e-06,
+      "loss": 1.0339,
+      "step": 15610
+    },
+    {
+      "epoch": 0.11306796383562437,
+      "grad_norm": 0.17333056032657623,
+      "learning_rate": 4.886939274830434e-06,
+      "loss": 1.0465,
+      "step": 15620
+    },
+    {
+      "epoch": 0.11314035049621056,
+      "grad_norm": 0.1765410602092743,
+      "learning_rate": 4.886866888169849e-06,
+      "loss": 1.0402,
+      "step": 15630
+    },
+    {
+      "epoch": 0.11321273715679675,
+      "grad_norm": 0.17773103713989258,
+      "learning_rate": 4.886794501509262e-06,
+      "loss": 1.0483,
+      "step": 15640
+    },
+    {
+      "epoch": 0.11328512381738293,
+      "grad_norm": 0.18877708911895752,
+      "learning_rate": 4.886722114848676e-06,
+      "loss": 1.0528,
+      "step": 15650
+    },
+    {
+      "epoch": 0.11335751047796912,
+      "grad_norm": 0.20413969457149506,
+      "learning_rate": 4.8866497281880895e-06,
+      "loss": 1.0667,
+      "step": 15660
+    },
+    {
+      "epoch": 0.11342989713855531,
+      "grad_norm": 0.16380807757377625,
+      "learning_rate": 4.886577341527504e-06,
+      "loss": 1.0583,
+      "step": 15670
+    },
+    {
+      "epoch": 0.1135022837991415,
+      "grad_norm": 0.17399290204048157,
+      "learning_rate": 4.886504954866918e-06,
+      "loss": 1.0675,
+      "step": 15680
+    },
+    {
+      "epoch": 0.11357467045972768,
+      "grad_norm": 0.18681904673576355,
+      "learning_rate": 4.886432568206331e-06,
+      "loss": 1.0452,
+      "step": 15690
+    },
+    {
+      "epoch": 0.11364705712031387,
+      "grad_norm": 0.15756523609161377,
+      "learning_rate": 4.886360181545745e-06,
+      "loss": 1.0391,
+      "step": 15700
+    },
+    {
+      "epoch": 0.11371944378090006,
+      "grad_norm": 0.18316251039505005,
+      "learning_rate": 4.886287794885159e-06,
+      "loss": 1.0474,
+      "step": 15710
+    },
+    {
+      "epoch": 0.11379183044148625,
+      "grad_norm": 0.19340310990810394,
+      "learning_rate": 4.886215408224573e-06,
+      "loss": 1.0438,
+      "step": 15720
+    },
+    {
+      "epoch": 0.11386421710207242,
+      "grad_norm": 0.16754977405071259,
+      "learning_rate": 4.8861430215639865e-06,
+      "loss": 1.0435,
+      "step": 15730
+    },
+    {
+      "epoch": 0.11393660376265861,
+      "grad_norm": 0.16273082792758942,
+      "learning_rate": 4.8860706349034e-06,
+      "loss": 1.0434,
+      "step": 15740
+    },
+    {
+      "epoch": 0.1140089904232448,
+      "grad_norm": 0.19182349741458893,
+      "learning_rate": 4.885998248242815e-06,
+      "loss": 1.0516,
+      "step": 15750
+    },
+    {
+      "epoch": 0.114081377083831,
+      "grad_norm": 0.2044076770544052,
+      "learning_rate": 4.885925861582228e-06,
+      "loss": 1.042,
+      "step": 15760
+    },
+    {
+      "epoch": 0.11415376374441717,
+      "grad_norm": 0.17020849883556366,
+      "learning_rate": 4.885853474921642e-06,
+      "loss": 1.0367,
+      "step": 15770
+    },
+    {
+      "epoch": 0.11422615040500336,
+      "grad_norm": 0.17558561265468597,
+      "learning_rate": 4.8857810882610554e-06,
+      "loss": 1.0528,
+      "step": 15780
+    },
+    {
+      "epoch": 0.11429853706558955,
+      "grad_norm": 0.20141629874706268,
+      "learning_rate": 4.88570870160047e-06,
+      "loss": 1.0485,
+      "step": 15790
+    },
+    {
+      "epoch": 0.11437092372617574,
+      "grad_norm": 0.17599183320999146,
+      "learning_rate": 4.8856363149398835e-06,
+      "loss": 1.0499,
+      "step": 15800
+    },
+    {
+      "epoch": 0.11444331038676192,
+      "grad_norm": 0.17712847888469696,
+      "learning_rate": 4.885563928279297e-06,
+      "loss": 1.0386,
+      "step": 15810
+    },
+    {
+      "epoch": 0.11451569704734811,
+      "grad_norm": 0.19977372884750366,
+      "learning_rate": 4.885491541618711e-06,
+      "loss": 1.0372,
+      "step": 15820
+    },
+    {
+      "epoch": 0.1145880837079343,
+      "grad_norm": 0.17353418469429016,
+      "learning_rate": 4.885419154958125e-06,
+      "loss": 1.0647,
+      "step": 15830
+    },
+    {
+      "epoch": 0.11466047036852049,
+      "grad_norm": 0.1808134913444519,
+      "learning_rate": 4.885346768297539e-06,
+      "loss": 1.0505,
+      "step": 15840
+    },
+    {
+      "epoch": 0.11473285702910668,
+      "grad_norm": 0.1848040372133255,
+      "learning_rate": 4.8852743816369524e-06,
+      "loss": 1.0363,
+      "step": 15850
+    },
+    {
+      "epoch": 0.11480524368969286,
+      "grad_norm": 0.1654992550611496,
+      "learning_rate": 4.885201994976366e-06,
+      "loss": 1.0361,
+      "step": 15860
+    },
+    {
+      "epoch": 0.11487763035027905,
+      "grad_norm": 0.17101933062076569,
+      "learning_rate": 4.88512960831578e-06,
+      "loss": 1.0407,
+      "step": 15870
+    },
+    {
+      "epoch": 0.11495001701086524,
+      "grad_norm": 0.2501929998397827,
+      "learning_rate": 4.885057221655194e-06,
+      "loss": 1.0494,
+      "step": 15880
+    },
+    {
+      "epoch": 0.11502240367145143,
+      "grad_norm": 0.17349718511104584,
+      "learning_rate": 4.884984834994608e-06,
+      "loss": 1.0417,
+      "step": 15890
+    },
+    {
+      "epoch": 0.11509479033203761,
+      "grad_norm": 0.20508350431919098,
+      "learning_rate": 4.884912448334021e-06,
+      "loss": 1.0481,
+      "step": 15900
+    },
+    {
+      "epoch": 0.1151671769926238,
+      "grad_norm": 0.17247234284877777,
+      "learning_rate": 4.884840061673435e-06,
+      "loss": 1.0362,
+      "step": 15910
+    },
+    {
+      "epoch": 0.11523956365320999,
+      "grad_norm": 0.16966472566127777,
+      "learning_rate": 4.8847676750128495e-06,
+      "loss": 1.0501,
+      "step": 15920
+    },
+    {
+      "epoch": 0.11531195031379618,
+      "grad_norm": 0.18211546540260315,
+      "learning_rate": 4.884695288352263e-06,
+      "loss": 1.0437,
+      "step": 15930
+    },
+    {
+      "epoch": 0.11538433697438236,
+      "grad_norm": 0.1784241944551468,
+      "learning_rate": 4.884622901691677e-06,
+      "loss": 1.0439,
+      "step": 15940
+    },
+    {
+      "epoch": 0.11545672363496855,
+      "grad_norm": 0.18033133447170258,
+      "learning_rate": 4.88455051503109e-06,
+      "loss": 1.0441,
+      "step": 15950
+    },
+    {
+      "epoch": 0.11552911029555474,
+      "grad_norm": 0.1717543601989746,
+      "learning_rate": 4.884478128370505e-06,
+      "loss": 1.032,
+      "step": 15960
+    },
+    {
+      "epoch": 0.11560149695614093,
+      "grad_norm": 0.18930160999298096,
+      "learning_rate": 4.884405741709918e-06,
+      "loss": 1.0425,
+      "step": 15970
+    },
+    {
+      "epoch": 0.1156738836167271,
+      "grad_norm": 0.17767852544784546,
+      "learning_rate": 4.884333355049332e-06,
+      "loss": 1.039,
+      "step": 15980
+    },
+    {
+      "epoch": 0.1157462702773133,
+      "grad_norm": 0.20390869677066803,
+      "learning_rate": 4.884260968388746e-06,
+      "loss": 1.0554,
+      "step": 15990
+    },
+    {
+      "epoch": 0.11581865693789949,
+      "grad_norm": 0.17186413705348969,
+      "learning_rate": 4.88418858172816e-06,
+      "loss": 1.0445,
+      "step": 16000
+    },
+    {
+      "epoch": 0.11589104359848568,
+      "grad_norm": 0.17237228155136108,
+      "learning_rate": 4.884116195067574e-06,
+      "loss": 1.0376,
+      "step": 16010
+    },
+    {
+      "epoch": 0.11596343025907185,
+      "grad_norm": 0.17973795533180237,
+      "learning_rate": 4.884043808406987e-06,
+      "loss": 1.0304,
+      "step": 16020
+    },
+    {
+      "epoch": 0.11603581691965804,
+      "grad_norm": 0.16578522324562073,
+      "learning_rate": 4.883971421746401e-06,
+      "loss": 1.0475,
+      "step": 16030
+    },
+    {
+      "epoch": 0.11610820358024423,
+      "grad_norm": 0.16535358130931854,
+      "learning_rate": 4.8838990350858145e-06,
+      "loss": 1.0501,
+      "step": 16040
+    },
+    {
+      "epoch": 0.11618059024083043,
+      "grad_norm": 0.18786129355430603,
+      "learning_rate": 4.883826648425228e-06,
+      "loss": 1.05,
+      "step": 16050
+    },
+    {
+      "epoch": 0.1162529769014166,
+      "grad_norm": 0.19526006281375885,
+      "learning_rate": 4.883754261764642e-06,
+      "loss": 1.0365,
+      "step": 16060
+    },
+    {
+      "epoch": 0.11632536356200279,
+      "grad_norm": 0.166265606880188,
+      "learning_rate": 4.883681875104056e-06,
+      "loss": 1.0481,
+      "step": 16070
+    },
+    {
+      "epoch": 0.11639775022258898,
+      "grad_norm": 0.18460646271705627,
+      "learning_rate": 4.88360948844347e-06,
+      "loss": 1.0422,
+      "step": 16080
+    },
+    {
+      "epoch": 0.11647013688317517,
+      "grad_norm": 0.16810445487499237,
+      "learning_rate": 4.8835371017828835e-06,
+      "loss": 1.0535,
+      "step": 16090
+    },
+    {
+      "epoch": 0.11654252354376135,
+      "grad_norm": 0.469312459230423,
+      "learning_rate": 4.883464715122297e-06,
+      "loss": 1.0527,
+      "step": 16100
+    },
+    {
+      "epoch": 0.11661491020434754,
+      "grad_norm": 0.1745699644088745,
+      "learning_rate": 4.8833923284617115e-06,
+      "loss": 1.0331,
+      "step": 16110
+    },
+    {
+      "epoch": 0.11668729686493373,
+      "grad_norm": 0.17050036787986755,
+      "learning_rate": 4.883319941801125e-06,
+      "loss": 1.0449,
+      "step": 16120
+    },
+    {
+      "epoch": 0.11675968352551992,
+      "grad_norm": 0.18037088215351105,
+      "learning_rate": 4.883247555140539e-06,
+      "loss": 1.044,
+      "step": 16130
+    },
+    {
+      "epoch": 0.1168320701861061,
+      "grad_norm": 0.23608429729938507,
+      "learning_rate": 4.883175168479952e-06,
+      "loss": 1.0464,
+      "step": 16140
+    },
+    {
+      "epoch": 0.11690445684669229,
+      "grad_norm": 0.175536647439003,
+      "learning_rate": 4.883102781819367e-06,
+      "loss": 1.051,
+      "step": 16150
+    },
+    {
+      "epoch": 0.11697684350727848,
+      "grad_norm": 0.17264996469020844,
+      "learning_rate": 4.8830303951587805e-06,
+      "loss": 1.0428,
+      "step": 16160
+    },
+    {
+      "epoch": 0.11704923016786467,
+      "grad_norm": 0.17153437435626984,
+      "learning_rate": 4.882958008498194e-06,
+      "loss": 1.035,
+      "step": 16170
+    },
+    {
+      "epoch": 0.11712161682845085,
+      "grad_norm": 0.16962124407291412,
+      "learning_rate": 4.882885621837608e-06,
+      "loss": 1.0428,
+      "step": 16180
+    },
+    {
+      "epoch": 0.11719400348903704,
+      "grad_norm": 0.17661970853805542,
+      "learning_rate": 4.882813235177022e-06,
+      "loss": 1.0442,
+      "step": 16190
+    },
+    {
+      "epoch": 0.11726639014962323,
+      "grad_norm": 0.1694341003894806,
+      "learning_rate": 4.882740848516436e-06,
+      "loss": 1.0409,
+      "step": 16200
+    },
+    {
+      "epoch": 0.11733877681020942,
+      "grad_norm": 0.1740081012248993,
+      "learning_rate": 4.882668461855849e-06,
+      "loss": 1.042,
+      "step": 16210
+    },
+    {
+      "epoch": 0.1174111634707956,
+      "grad_norm": 0.20887626707553864,
+      "learning_rate": 4.882596075195263e-06,
+      "loss": 1.0573,
+      "step": 16220
+    },
+    {
+      "epoch": 0.11748355013138179,
+      "grad_norm": 0.1686887890100479,
+      "learning_rate": 4.8825236885346775e-06,
+      "loss": 1.0396,
+      "step": 16230
+    },
+    {
+      "epoch": 0.11755593679196798,
+      "grad_norm": 0.654279351234436,
+      "learning_rate": 4.882451301874091e-06,
+      "loss": 1.0511,
+      "step": 16240
+    },
+    {
+      "epoch": 0.11762832345255417,
+      "grad_norm": 0.18497861921787262,
+      "learning_rate": 4.882378915213505e-06,
+      "loss": 1.0568,
+      "step": 16250
+    },
+    {
+      "epoch": 0.11770071011314034,
+      "grad_norm": 0.1842462718486786,
+      "learning_rate": 4.882306528552918e-06,
+      "loss": 1.0534,
+      "step": 16260
+    },
+    {
+      "epoch": 0.11777309677372653,
+      "grad_norm": 0.16614331305027008,
+      "learning_rate": 4.882234141892333e-06,
+      "loss": 1.0546,
+      "step": 16270
+    },
+    {
+      "epoch": 0.11784548343431273,
+      "grad_norm": 0.17406275868415833,
+      "learning_rate": 4.882161755231746e-06,
+      "loss": 1.0451,
+      "step": 16280
+    },
+    {
+      "epoch": 0.11791787009489892,
+      "grad_norm": 0.18107841908931732,
+      "learning_rate": 4.88208936857116e-06,
+      "loss": 1.0556,
+      "step": 16290
+    },
+    {
+      "epoch": 0.11799025675548509,
+      "grad_norm": 0.17502503097057343,
+      "learning_rate": 4.882016981910574e-06,
+      "loss": 1.0464,
+      "step": 16300
+    },
+    {
+      "epoch": 0.11806264341607128,
+      "grad_norm": 0.18071506917476654,
+      "learning_rate": 4.881944595249988e-06,
+      "loss": 1.0354,
+      "step": 16310
+    },
+    {
+      "epoch": 0.11813503007665747,
+      "grad_norm": 0.16707313060760498,
+      "learning_rate": 4.881872208589402e-06,
+      "loss": 1.0548,
+      "step": 16320
+    },
+    {
+      "epoch": 0.11820741673724366,
+      "grad_norm": 0.18432946503162384,
+      "learning_rate": 4.881799821928815e-06,
+      "loss": 1.0413,
+      "step": 16330
+    },
+    {
+      "epoch": 0.11827980339782984,
+      "grad_norm": 0.19203749299049377,
+      "learning_rate": 4.881727435268229e-06,
+      "loss": 1.0507,
+      "step": 16340
+    },
+    {
+      "epoch": 0.11835219005841603,
+      "grad_norm": 0.20910878479480743,
+      "learning_rate": 4.881655048607643e-06,
+      "loss": 1.0511,
+      "step": 16350
+    },
+    {
+      "epoch": 0.11842457671900222,
+      "grad_norm": 0.16124652326107025,
+      "learning_rate": 4.881582661947057e-06,
+      "loss": 1.0418,
+      "step": 16360
+    },
+    {
+      "epoch": 0.11849696337958841,
+      "grad_norm": 0.1836594045162201,
+      "learning_rate": 4.881510275286471e-06,
+      "loss": 1.0411,
+      "step": 16370
+    },
+    {
+      "epoch": 0.11856935004017459,
+      "grad_norm": 0.1791921854019165,
+      "learning_rate": 4.881437888625884e-06,
+      "loss": 1.0399,
+      "step": 16380
+    },
+    {
+      "epoch": 0.11864173670076078,
+      "grad_norm": 0.1730756163597107,
+      "learning_rate": 4.881365501965299e-06,
+      "loss": 1.0393,
+      "step": 16390
+    },
+    {
+      "epoch": 0.11871412336134697,
+      "grad_norm": 0.17524994909763336,
+      "learning_rate": 4.881293115304712e-06,
+      "loss": 1.0573,
+      "step": 16400
+    },
+    {
+      "epoch": 0.11878651002193316,
+      "grad_norm": 0.17954038083553314,
+      "learning_rate": 4.881220728644126e-06,
+      "loss": 1.0378,
+      "step": 16410
+    },
+    {
+      "epoch": 0.11885889668251935,
+      "grad_norm": 0.17099116742610931,
+      "learning_rate": 4.88114834198354e-06,
+      "loss": 1.047,
+      "step": 16420
+    },
+    {
+      "epoch": 0.11893128334310553,
+      "grad_norm": 0.1616821140050888,
+      "learning_rate": 4.881075955322954e-06,
+      "loss": 1.0512,
+      "step": 16430
+    },
+    {
+      "epoch": 0.11900367000369172,
+      "grad_norm": 0.18617095053195953,
+      "learning_rate": 4.881003568662368e-06,
+      "loss": 1.0634,
+      "step": 16440
+    },
+    {
+      "epoch": 0.11907605666427791,
+      "grad_norm": 0.19976121187210083,
+      "learning_rate": 4.880931182001781e-06,
+      "loss": 1.0464,
+      "step": 16450
+    },
+    {
+      "epoch": 0.1191484433248641,
+      "grad_norm": 0.17326410114765167,
+      "learning_rate": 4.880858795341195e-06,
+      "loss": 1.0619,
+      "step": 16460
+    },
+    {
+      "epoch": 0.11922082998545028,
+      "grad_norm": 0.16668163239955902,
+      "learning_rate": 4.880786408680609e-06,
+      "loss": 1.03,
+      "step": 16470
+    },
+    {
+      "epoch": 0.11929321664603647,
+      "grad_norm": 0.16474226117134094,
+      "learning_rate": 4.880714022020023e-06,
+      "loss": 1.0532,
+      "step": 16480
+    },
+    {
+      "epoch": 0.11936560330662266,
+      "grad_norm": 0.18454322218894958,
+      "learning_rate": 4.880641635359437e-06,
+      "loss": 1.0407,
+      "step": 16490
+    },
+    {
+      "epoch": 0.11943798996720885,
+      "grad_norm": 0.18234814703464508,
+      "learning_rate": 4.88056924869885e-06,
+      "loss": 1.0338,
+      "step": 16500
+    },
+    {
+      "epoch": 0.11951037662779503,
+      "grad_norm": 0.2152404487133026,
+      "learning_rate": 4.880496862038264e-06,
+      "loss": 1.0384,
+      "step": 16510
+    },
+    {
+      "epoch": 0.11958276328838122,
+      "grad_norm": 0.1833188831806183,
+      "learning_rate": 4.880424475377678e-06,
+      "loss": 1.052,
+      "step": 16520
+    },
+    {
+      "epoch": 0.11965514994896741,
+      "grad_norm": 0.16804316639900208,
+      "learning_rate": 4.880352088717092e-06,
+      "loss": 1.0423,
+      "step": 16530
+    },
+    {
+      "epoch": 0.1197275366095536,
+      "grad_norm": 0.21003463864326477,
+      "learning_rate": 4.8802797020565055e-06,
+      "loss": 1.0444,
+      "step": 16540
+    },
+    {
+      "epoch": 0.11979992327013977,
+      "grad_norm": 0.17130064964294434,
+      "learning_rate": 4.880207315395919e-06,
+      "loss": 1.0572,
+      "step": 16550
+    },
+    {
+      "epoch": 0.11987230993072596,
+      "grad_norm": 0.17350070178508759,
+      "learning_rate": 4.880134928735333e-06,
+      "loss": 1.0547,
+      "step": 16560
+    },
+    {
+      "epoch": 0.11994469659131216,
+      "grad_norm": 0.17210082709789276,
+      "learning_rate": 4.880062542074746e-06,
+      "loss": 1.0394,
+      "step": 16570
+    },
+    {
+      "epoch": 0.12001708325189835,
+      "grad_norm": 0.17453570663928986,
+      "learning_rate": 4.879990155414161e-06,
+      "loss": 1.044,
+      "step": 16580
+    },
+    {
+      "epoch": 0.12008946991248452,
+      "grad_norm": 0.16361692547798157,
+      "learning_rate": 4.8799177687535744e-06,
+      "loss": 1.0393,
+      "step": 16590
+    },
+    {
+      "epoch": 0.12016185657307071,
+      "grad_norm": 0.17388883233070374,
+      "learning_rate": 4.879845382092988e-06,
+      "loss": 1.0354,
+      "step": 16600
+    },
+    {
+      "epoch": 0.1202342432336569,
+      "grad_norm": 0.18685205280780792,
+      "learning_rate": 4.879772995432402e-06,
+      "loss": 1.0414,
+      "step": 16610
+    },
+    {
+      "epoch": 0.1203066298942431,
+      "grad_norm": 0.20947107672691345,
+      "learning_rate": 4.879700608771816e-06,
+      "loss": 1.0482,
+      "step": 16620
+    },
+    {
+      "epoch": 0.12037901655482927,
+      "grad_norm": 0.16179795563220978,
+      "learning_rate": 4.87962822211123e-06,
+      "loss": 1.0336,
+      "step": 16630
+    },
+    {
+      "epoch": 0.12045140321541546,
+      "grad_norm": 0.18635103106498718,
+      "learning_rate": 4.879555835450643e-06,
+      "loss": 1.0425,
+      "step": 16640
+    },
+    {
+      "epoch": 0.12052378987600165,
+      "grad_norm": 0.17454616725444794,
+      "learning_rate": 4.879483448790057e-06,
+      "loss": 1.0383,
+      "step": 16650
+    },
+    {
+      "epoch": 0.12059617653658784,
+      "grad_norm": 0.17645971477031708,
+      "learning_rate": 4.879411062129471e-06,
+      "loss": 1.036,
+      "step": 16660
+    },
+    {
+      "epoch": 0.12066856319717402,
+      "grad_norm": 0.20391476154327393,
+      "learning_rate": 4.879338675468885e-06,
+      "loss": 1.0259,
+      "step": 16670
+    },
+    {
+      "epoch": 0.12074094985776021,
+      "grad_norm": 0.1724318414926529,
+      "learning_rate": 4.879266288808299e-06,
+      "loss": 1.0425,
+      "step": 16680
+    },
+    {
+      "epoch": 0.1208133365183464,
+      "grad_norm": 0.17643970251083374,
+      "learning_rate": 4.879193902147712e-06,
+      "loss": 1.0654,
+      "step": 16690
+    },
+    {
+      "epoch": 0.12088572317893259,
+      "grad_norm": 0.339347243309021,
+      "learning_rate": 4.879121515487126e-06,
+      "loss": 1.048,
+      "step": 16700
+    },
+    {
+      "epoch": 0.12095810983951877,
+      "grad_norm": 0.17284011840820312,
+      "learning_rate": 4.87904912882654e-06,
+      "loss": 1.0377,
+      "step": 16710
+    },
+    {
+      "epoch": 0.12103049650010496,
+      "grad_norm": 0.1662074625492096,
+      "learning_rate": 4.878976742165954e-06,
+      "loss": 1.0272,
+      "step": 16720
+    },
+    {
+      "epoch": 0.12110288316069115,
+      "grad_norm": 0.1813957393169403,
+      "learning_rate": 4.878904355505368e-06,
+      "loss": 1.0351,
+      "step": 16730
+    },
+    {
+      "epoch": 0.12117526982127734,
+      "grad_norm": 0.17036522924900055,
+      "learning_rate": 4.878831968844781e-06,
+      "loss": 1.0373,
+      "step": 16740
+    },
+    {
+      "epoch": 0.12124765648186352,
+      "grad_norm": 0.17929202318191528,
+      "learning_rate": 4.878759582184196e-06,
+      "loss": 1.0419,
+      "step": 16750
+    },
+    {
+      "epoch": 0.12132004314244971,
+      "grad_norm": 0.173272043466568,
+      "learning_rate": 4.878687195523609e-06,
+      "loss": 1.0535,
+      "step": 16760
+    },
+    {
+      "epoch": 0.1213924298030359,
+      "grad_norm": 0.18325883150100708,
+      "learning_rate": 4.878614808863023e-06,
+      "loss": 1.0403,
+      "step": 16770
+    },
+    {
+      "epoch": 0.12146481646362209,
+      "grad_norm": 0.18894407153129578,
+      "learning_rate": 4.8785424222024365e-06,
+      "loss": 1.0472,
+      "step": 16780
+    },
+    {
+      "epoch": 0.12153720312420827,
+      "grad_norm": 0.17379698157310486,
+      "learning_rate": 4.878470035541851e-06,
+      "loss": 1.0422,
+      "step": 16790
+    },
+    {
+      "epoch": 0.12160958978479446,
+      "grad_norm": 0.17601001262664795,
+      "learning_rate": 4.878397648881265e-06,
+      "loss": 1.0497,
+      "step": 16800
+    },
+    {
+      "epoch": 0.12168197644538065,
+      "grad_norm": 0.16135278344154358,
+      "learning_rate": 4.878325262220678e-06,
+      "loss": 1.0349,
+      "step": 16810
+    },
+    {
+      "epoch": 0.12175436310596684,
+      "grad_norm": 0.17922668159008026,
+      "learning_rate": 4.878252875560092e-06,
+      "loss": 1.0377,
+      "step": 16820
+    },
+    {
+      "epoch": 0.12182674976655301,
+      "grad_norm": 0.17484663426876068,
+      "learning_rate": 4.878180488899506e-06,
+      "loss": 1.0463,
+      "step": 16830
+    },
+    {
+      "epoch": 0.1218991364271392,
+      "grad_norm": 0.16210487484931946,
+      "learning_rate": 4.87810810223892e-06,
+      "loss": 1.0429,
+      "step": 16840
+    },
+    {
+      "epoch": 0.1219715230877254,
+      "grad_norm": 0.17364494502544403,
+      "learning_rate": 4.8780357155783335e-06,
+      "loss": 1.0426,
+      "step": 16850
+    },
+    {
+      "epoch": 0.12204390974831159,
+      "grad_norm": 0.1732948124408722,
+      "learning_rate": 4.877963328917747e-06,
+      "loss": 1.0417,
+      "step": 16860
+    },
+    {
+      "epoch": 0.12211629640889776,
+      "grad_norm": 0.18654009699821472,
+      "learning_rate": 4.877890942257162e-06,
+      "loss": 1.0351,
+      "step": 16870
+    },
+    {
+      "epoch": 0.12218868306948395,
+      "grad_norm": 0.17695607244968414,
+      "learning_rate": 4.877818555596575e-06,
+      "loss": 1.0412,
+      "step": 16880
+    },
+    {
+      "epoch": 0.12226106973007014,
+      "grad_norm": 0.173911452293396,
+      "learning_rate": 4.877746168935989e-06,
+      "loss": 1.0419,
+      "step": 16890
+    },
+    {
+      "epoch": 0.12233345639065633,
+      "grad_norm": 0.17585676908493042,
+      "learning_rate": 4.8776737822754025e-06,
+      "loss": 1.0414,
+      "step": 16900
+    },
+    {
+      "epoch": 0.12240584305124251,
+      "grad_norm": 0.179653599858284,
+      "learning_rate": 4.877601395614817e-06,
+      "loss": 1.0412,
+      "step": 16910
+    },
+    {
+      "epoch": 0.1224782297118287,
+      "grad_norm": 0.16631865501403809,
+      "learning_rate": 4.8775290089542306e-06,
+      "loss": 1.0408,
+      "step": 16920
+    },
+    {
+      "epoch": 0.12255061637241489,
+      "grad_norm": 0.17161540687084198,
+      "learning_rate": 4.877456622293644e-06,
+      "loss": 1.0323,
+      "step": 16930
+    },
+    {
+      "epoch": 0.12262300303300108,
+      "grad_norm": 0.1741219311952591,
+      "learning_rate": 4.877384235633058e-06,
+      "loss": 1.0435,
+      "step": 16940
+    },
+    {
+      "epoch": 0.12269538969358726,
+      "grad_norm": 0.2024671733379364,
+      "learning_rate": 4.877311848972472e-06,
+      "loss": 1.0446,
+      "step": 16950
+    },
+    {
+      "epoch": 0.12276777635417345,
+      "grad_norm": 0.1737942099571228,
+      "learning_rate": 4.877239462311886e-06,
+      "loss": 1.0464,
+      "step": 16960
+    },
+    {
+      "epoch": 0.12284016301475964,
+      "grad_norm": 0.21153798699378967,
+      "learning_rate": 4.8771670756512995e-06,
+      "loss": 1.0438,
+      "step": 16970
+    },
+    {
+      "epoch": 0.12291254967534583,
+      "grad_norm": 0.17885242402553558,
+      "learning_rate": 4.877094688990713e-06,
+      "loss": 1.0402,
+      "step": 16980
+    },
+    {
+      "epoch": 0.12298493633593202,
+      "grad_norm": 0.18069690465927124,
+      "learning_rate": 4.8770223023301276e-06,
+      "loss": 1.0519,
+      "step": 16990
+    },
+    {
+      "epoch": 0.1230573229965182,
+      "grad_norm": 0.17245696485042572,
+      "learning_rate": 4.876949915669541e-06,
+      "loss": 1.0395,
+      "step": 17000
+    },
+    {
+      "epoch": 0.12312970965710439,
+      "grad_norm": 0.16451863944530487,
+      "learning_rate": 4.876877529008955e-06,
+      "loss": 1.0378,
+      "step": 17010
+    },
+    {
+      "epoch": 0.12320209631769058,
+      "grad_norm": 0.18624931573867798,
+      "learning_rate": 4.876805142348368e-06,
+      "loss": 1.0513,
+      "step": 17020
+    },
+    {
+      "epoch": 0.12327448297827677,
+      "grad_norm": 0.16225986182689667,
+      "learning_rate": 4.876732755687783e-06,
+      "loss": 1.0374,
+      "step": 17030
+    },
+    {
+      "epoch": 0.12334686963886295,
+      "grad_norm": 0.19967621564865112,
+      "learning_rate": 4.8766603690271965e-06,
+      "loss": 1.0455,
+      "step": 17040
+    },
+    {
+      "epoch": 0.12341925629944914,
+      "grad_norm": 0.18386250734329224,
+      "learning_rate": 4.87658798236661e-06,
+      "loss": 1.0349,
+      "step": 17050
+    },
+    {
+      "epoch": 0.12349164296003533,
+      "grad_norm": 0.17461447417736053,
+      "learning_rate": 4.876515595706024e-06,
+      "loss": 1.0374,
+      "step": 17060
+    },
+    {
+      "epoch": 0.12356402962062152,
+      "grad_norm": 0.16911447048187256,
+      "learning_rate": 4.876443209045438e-06,
+      "loss": 1.0312,
+      "step": 17070
+    },
+    {
+      "epoch": 0.1236364162812077,
+      "grad_norm": 0.17958052456378937,
+      "learning_rate": 4.876370822384851e-06,
+      "loss": 1.038,
+      "step": 17080
+    },
+    {
+      "epoch": 0.12370880294179389,
+      "grad_norm": 0.16912564635276794,
+      "learning_rate": 4.8762984357242646e-06,
+      "loss": 1.0268,
+      "step": 17090
+    },
+    {
+      "epoch": 0.12378118960238008,
+      "grad_norm": 0.18362846970558167,
+      "learning_rate": 4.876226049063679e-06,
+      "loss": 1.0457,
+      "step": 17100
+    },
+    {
+      "epoch": 0.12385357626296627,
+      "grad_norm": 0.18593281507492065,
+      "learning_rate": 4.876153662403093e-06,
+      "loss": 1.0296,
+      "step": 17110
+    },
+    {
+      "epoch": 0.12392596292355244,
+      "grad_norm": 0.1746986359357834,
+      "learning_rate": 4.876081275742506e-06,
+      "loss": 1.0429,
+      "step": 17120
+    },
+    {
+      "epoch": 0.12399834958413863,
+      "grad_norm": 0.16610664129257202,
+      "learning_rate": 4.87600888908192e-06,
+      "loss": 1.0432,
+      "step": 17130
+    },
+    {
+      "epoch": 0.12407073624472482,
+      "grad_norm": 0.1753586083650589,
+      "learning_rate": 4.875936502421334e-06,
+      "loss": 1.0408,
+      "step": 17140
+    },
+    {
+      "epoch": 0.12414312290531102,
+      "grad_norm": 0.1812318116426468,
+      "learning_rate": 4.875864115760748e-06,
+      "loss": 1.0482,
+      "step": 17150
+    },
+    {
+      "epoch": 0.12421550956589719,
+      "grad_norm": 0.18393771350383759,
+      "learning_rate": 4.875791729100162e-06,
+      "loss": 1.0439,
+      "step": 17160
+    },
+    {
+      "epoch": 0.12428789622648338,
+      "grad_norm": 0.2628847062587738,
+      "learning_rate": 4.875719342439575e-06,
+      "loss": 1.0498,
+      "step": 17170
+    },
+    {
+      "epoch": 0.12436028288706957,
+      "grad_norm": 0.18083707988262177,
+      "learning_rate": 4.87564695577899e-06,
+      "loss": 1.0428,
+      "step": 17180
+    },
+    {
+      "epoch": 0.12443266954765576,
+      "grad_norm": 0.19616307318210602,
+      "learning_rate": 4.875574569118403e-06,
+      "loss": 1.0391,
+      "step": 17190
+    },
+    {
+      "epoch": 0.12450505620824194,
+      "grad_norm": 0.16640251874923706,
+      "learning_rate": 4.875502182457817e-06,
+      "loss": 1.0412,
+      "step": 17200
+    },
+    {
+      "epoch": 0.12457744286882813,
+      "grad_norm": 0.2135634571313858,
+      "learning_rate": 4.8754297957972305e-06,
+      "loss": 1.0415,
+      "step": 17210
+    },
+    {
+      "epoch": 0.12464982952941432,
+      "grad_norm": 0.17053596675395966,
+      "learning_rate": 4.875357409136645e-06,
+      "loss": 1.0467,
+      "step": 17220
+    },
+    {
+      "epoch": 0.12472221619000051,
+      "grad_norm": 0.16077913343906403,
+      "learning_rate": 4.875285022476059e-06,
+      "loss": 1.0309,
+      "step": 17230
+    },
+    {
+      "epoch": 0.12479460285058669,
+      "grad_norm": 0.17686612904071808,
+      "learning_rate": 4.875212635815472e-06,
+      "loss": 1.0304,
+      "step": 17240
+    },
+    {
+      "epoch": 0.12486698951117288,
+      "grad_norm": 0.1883876919746399,
+      "learning_rate": 4.875140249154886e-06,
+      "loss": 1.0413,
+      "step": 17250
+    },
+    {
+      "epoch": 0.12493937617175907,
+      "grad_norm": 0.20872871577739716,
+      "learning_rate": 4.8750678624943e-06,
+      "loss": 1.0434,
+      "step": 17260
+    },
+    {
+      "epoch": 0.12501176283234525,
+      "grad_norm": 0.17581568658351898,
+      "learning_rate": 4.874995475833714e-06,
+      "loss": 1.0488,
+      "step": 17270
+    },
+    {
+      "epoch": 0.12508414949293145,
+      "grad_norm": 0.17273147404193878,
+      "learning_rate": 4.8749230891731275e-06,
+      "loss": 1.0498,
+      "step": 17280
+    },
+    {
+      "epoch": 0.12515653615351763,
+      "grad_norm": 0.21244202554225922,
+      "learning_rate": 4.874850702512541e-06,
+      "loss": 1.0332,
+      "step": 17290
+    },
+    {
+      "epoch": 0.1252289228141038,
+      "grad_norm": 0.17866666615009308,
+      "learning_rate": 4.874778315851955e-06,
+      "loss": 1.0346,
+      "step": 17300
+    },
+    {
+      "epoch": 0.12530130947469,
+      "grad_norm": 0.1749383807182312,
+      "learning_rate": 4.874705929191369e-06,
+      "loss": 1.0452,
+      "step": 17310
+    },
+    {
+      "epoch": 0.12537369613527619,
+      "grad_norm": 0.175296813249588,
+      "learning_rate": 4.874633542530783e-06,
+      "loss": 1.0376,
+      "step": 17320
+    },
+    {
+      "epoch": 0.1254460827958624,
+      "grad_norm": 0.18073537945747375,
+      "learning_rate": 4.8745611558701964e-06,
+      "loss": 1.0316,
+      "step": 17330
+    },
+    {
+      "epoch": 0.12551846945644857,
+      "grad_norm": 0.18828389048576355,
+      "learning_rate": 4.87448876920961e-06,
+      "loss": 1.0376,
+      "step": 17340
+    },
+    {
+      "epoch": 0.12559085611703474,
+      "grad_norm": 0.25895076990127563,
+      "learning_rate": 4.8744163825490245e-06,
+      "loss": 1.0266,
+      "step": 17350
+    },
+    {
+      "epoch": 0.12566324277762095,
+      "grad_norm": 0.200386181473732,
+      "learning_rate": 4.874343995888438e-06,
+      "loss": 1.0438,
+      "step": 17360
+    },
+    {
+      "epoch": 0.12573562943820712,
+      "grad_norm": 0.1703101396560669,
+      "learning_rate": 4.874271609227852e-06,
+      "loss": 1.0397,
+      "step": 17370
+    },
+    {
+      "epoch": 0.1258080160987933,
+      "grad_norm": 0.18138134479522705,
+      "learning_rate": 4.874199222567265e-06,
+      "loss": 1.0447,
+      "step": 17380
+    },
+    {
+      "epoch": 0.1258804027593795,
+      "grad_norm": 0.16725878417491913,
+      "learning_rate": 4.87412683590668e-06,
+      "loss": 1.0437,
+      "step": 17390
+    },
+    {
+      "epoch": 0.12595278941996568,
+      "grad_norm": 0.19038145244121552,
+      "learning_rate": 4.8740544492460934e-06,
+      "loss": 1.0377,
+      "step": 17400
+    },
+    {
+      "epoch": 0.1260251760805519,
+      "grad_norm": 0.20098719000816345,
+      "learning_rate": 4.873982062585507e-06,
+      "loss": 1.0406,
+      "step": 17410
+    },
+    {
+      "epoch": 0.12609756274113806,
+      "grad_norm": 0.17163150012493134,
+      "learning_rate": 4.873909675924921e-06,
+      "loss": 1.0316,
+      "step": 17420
+    },
+    {
+      "epoch": 0.12616994940172424,
+      "grad_norm": 0.17717736959457397,
+      "learning_rate": 4.873837289264335e-06,
+      "loss": 1.0378,
+      "step": 17430
+    },
+    {
+      "epoch": 0.12624233606231045,
+      "grad_norm": 0.1882064789533615,
+      "learning_rate": 4.873764902603749e-06,
+      "loss": 1.0284,
+      "step": 17440
+    },
+    {
+      "epoch": 0.12631472272289662,
+      "grad_norm": 0.1787366420030594,
+      "learning_rate": 4.873692515943162e-06,
+      "loss": 1.0418,
+      "step": 17450
+    },
+    {
+      "epoch": 0.1263871093834828,
+      "grad_norm": 0.17708836495876312,
+      "learning_rate": 4.873620129282576e-06,
+      "loss": 1.047,
+      "step": 17460
+    },
+    {
+      "epoch": 0.126459496044069,
+      "grad_norm": 0.1702563464641571,
+      "learning_rate": 4.8735477426219905e-06,
+      "loss": 1.038,
+      "step": 17470
+    },
+    {
+      "epoch": 0.12653188270465518,
+      "grad_norm": 0.1806415468454361,
+      "learning_rate": 4.873475355961404e-06,
+      "loss": 1.0358,
+      "step": 17480
+    },
+    {
+      "epoch": 0.12660426936524138,
+      "grad_norm": 0.24682064354419708,
+      "learning_rate": 4.873402969300818e-06,
+      "loss": 1.0479,
+      "step": 17490
+    },
+    {
+      "epoch": 0.12667665602582756,
+      "grad_norm": 0.18831071257591248,
+      "learning_rate": 4.873330582640231e-06,
+      "loss": 1.0333,
+      "step": 17500
+    },
+    {
+      "epoch": 0.12674904268641374,
+      "grad_norm": 0.17342287302017212,
+      "learning_rate": 4.873258195979646e-06,
+      "loss": 1.0253,
+      "step": 17510
+    },
+    {
+      "epoch": 0.12682142934699994,
+      "grad_norm": 0.17534632980823517,
+      "learning_rate": 4.873185809319059e-06,
+      "loss": 1.0478,
+      "step": 17520
+    },
+    {
+      "epoch": 0.12689381600758612,
+      "grad_norm": 0.16948920488357544,
+      "learning_rate": 4.873113422658473e-06,
+      "loss": 1.0396,
+      "step": 17530
+    },
+    {
+      "epoch": 0.1269662026681723,
+      "grad_norm": 0.19890448451042175,
+      "learning_rate": 4.873041035997887e-06,
+      "loss": 1.035,
+      "step": 17540
+    },
+    {
+      "epoch": 0.1270385893287585,
+      "grad_norm": 0.17209777235984802,
+      "learning_rate": 4.872968649337301e-06,
+      "loss": 1.031,
+      "step": 17550
+    },
+    {
+      "epoch": 0.12711097598934468,
+      "grad_norm": 0.1808318793773651,
+      "learning_rate": 4.872896262676715e-06,
+      "loss": 1.0491,
+      "step": 17560
+    },
+    {
+      "epoch": 0.12718336264993088,
+      "grad_norm": 0.24130041897296906,
+      "learning_rate": 4.872823876016128e-06,
+      "loss": 1.0217,
+      "step": 17570
+    },
+    {
+      "epoch": 0.12725574931051706,
+      "grad_norm": 0.22134606540203094,
+      "learning_rate": 4.872751489355542e-06,
+      "loss": 1.0452,
+      "step": 17580
+    },
+    {
+      "epoch": 0.12732813597110323,
+      "grad_norm": 0.18467597663402557,
+      "learning_rate": 4.872679102694956e-06,
+      "loss": 1.0515,
+      "step": 17590
+    },
+    {
+      "epoch": 0.12740052263168944,
+      "grad_norm": 0.1730322241783142,
+      "learning_rate": 4.87260671603437e-06,
+      "loss": 1.0315,
+      "step": 17600
+    },
+    {
+      "epoch": 0.12747290929227562,
+      "grad_norm": 0.1717580109834671,
+      "learning_rate": 4.872534329373783e-06,
+      "loss": 1.0351,
+      "step": 17610
+    },
+    {
+      "epoch": 0.12754529595286182,
+      "grad_norm": 0.17170915007591248,
+      "learning_rate": 4.872461942713197e-06,
+      "loss": 1.033,
+      "step": 17620
+    },
+    {
+      "epoch": 0.127617682613448,
+      "grad_norm": 0.17856451869010925,
+      "learning_rate": 4.872389556052611e-06,
+      "loss": 1.0123,
+      "step": 17630
+    },
+    {
+      "epoch": 0.12769006927403417,
+      "grad_norm": 0.19595351815223694,
+      "learning_rate": 4.8723171693920245e-06,
+      "loss": 1.0247,
+      "step": 17640
+    },
+    {
+      "epoch": 0.12776245593462038,
+      "grad_norm": 0.16732360422611237,
+      "learning_rate": 4.872244782731438e-06,
+      "loss": 1.0421,
+      "step": 17650
+    },
+    {
+      "epoch": 0.12783484259520655,
+      "grad_norm": 0.18419161438941956,
+      "learning_rate": 4.8721723960708526e-06,
+      "loss": 1.0563,
+      "step": 17660
+    },
+    {
+      "epoch": 0.12790722925579273,
+      "grad_norm": 0.20935377478599548,
+      "learning_rate": 4.872100009410266e-06,
+      "loss": 1.0111,
+      "step": 17670
+    },
+    {
+      "epoch": 0.12797961591637894,
+      "grad_norm": 0.1993681639432907,
+      "learning_rate": 4.87202762274968e-06,
+      "loss": 1.0407,
+      "step": 17680
+    },
+    {
+      "epoch": 0.1280520025769651,
+      "grad_norm": 0.16351039707660675,
+      "learning_rate": 4.871955236089093e-06,
+      "loss": 1.0276,
+      "step": 17690
+    },
+    {
+      "epoch": 0.12812438923755132,
+      "grad_norm": 0.17967389523983002,
+      "learning_rate": 4.871882849428508e-06,
+      "loss": 1.0442,
+      "step": 17700
+    },
+    {
+      "epoch": 0.1281967758981375,
+      "grad_norm": 0.1645398736000061,
+      "learning_rate": 4.8718104627679215e-06,
+      "loss": 1.0285,
+      "step": 17710
+    },
+    {
+      "epoch": 0.12826916255872367,
+      "grad_norm": 0.16454453766345978,
+      "learning_rate": 4.871738076107335e-06,
+      "loss": 1.0418,
+      "step": 17720
+    },
+    {
+      "epoch": 0.12834154921930988,
+      "grad_norm": 0.16446729004383087,
+      "learning_rate": 4.871665689446749e-06,
+      "loss": 1.0381,
+      "step": 17730
+    },
+    {
+      "epoch": 0.12841393587989605,
+      "grad_norm": 0.16874778270721436,
+      "learning_rate": 4.871593302786163e-06,
+      "loss": 1.0239,
+      "step": 17740
+    },
+    {
+      "epoch": 0.12848632254048223,
+      "grad_norm": 0.17436982691287994,
+      "learning_rate": 4.871520916125577e-06,
+      "loss": 1.0345,
+      "step": 17750
+    },
+    {
+      "epoch": 0.12855870920106843,
+      "grad_norm": 0.18663202226161957,
+      "learning_rate": 4.87144852946499e-06,
+      "loss": 1.0282,
+      "step": 17760
+    },
+    {
+      "epoch": 0.1286310958616546,
+      "grad_norm": 0.1899365931749344,
+      "learning_rate": 4.871376142804404e-06,
+      "loss": 1.0335,
+      "step": 17770
+    },
+    {
+      "epoch": 0.12870348252224081,
+      "grad_norm": 0.17862582206726074,
+      "learning_rate": 4.8713037561438185e-06,
+      "loss": 1.0453,
+      "step": 17780
+    },
+    {
+      "epoch": 0.128775869182827,
+      "grad_norm": 0.1662217378616333,
+      "learning_rate": 4.871231369483232e-06,
+      "loss": 1.0389,
+      "step": 17790
+    },
+    {
+      "epoch": 0.12884825584341317,
+      "grad_norm": 0.1717958003282547,
+      "learning_rate": 4.871158982822646e-06,
+      "loss": 1.0385,
+      "step": 17800
+    },
+    {
+      "epoch": 0.12892064250399937,
+      "grad_norm": 0.17093853652477264,
+      "learning_rate": 4.871086596162059e-06,
+      "loss": 1.0347,
+      "step": 17810
+    },
+    {
+      "epoch": 0.12899302916458555,
+      "grad_norm": 0.212626650929451,
+      "learning_rate": 4.871014209501474e-06,
+      "loss": 1.0299,
+      "step": 17820
+    },
+    {
+      "epoch": 0.12906541582517173,
+      "grad_norm": 0.16492265462875366,
+      "learning_rate": 4.870941822840887e-06,
+      "loss": 1.0366,
+      "step": 17830
+    },
+    {
+      "epoch": 0.12913780248575793,
+      "grad_norm": 0.17654773592948914,
+      "learning_rate": 4.870869436180301e-06,
+      "loss": 1.0359,
+      "step": 17840
+    },
+    {
+      "epoch": 0.1292101891463441,
+      "grad_norm": 0.17455124855041504,
+      "learning_rate": 4.870797049519715e-06,
+      "loss": 1.0263,
+      "step": 17850
+    },
+    {
+      "epoch": 0.1292825758069303,
+      "grad_norm": 0.16440659761428833,
+      "learning_rate": 4.870724662859129e-06,
+      "loss": 1.0193,
+      "step": 17860
+    },
+    {
+      "epoch": 0.1293549624675165,
+      "grad_norm": 0.1920333057641983,
+      "learning_rate": 4.870652276198543e-06,
+      "loss": 1.0347,
+      "step": 17870
+    },
+    {
+      "epoch": 0.12942734912810266,
+      "grad_norm": 0.18916349112987518,
+      "learning_rate": 4.870579889537956e-06,
+      "loss": 1.036,
+      "step": 17880
+    },
+    {
+      "epoch": 0.12949973578868887,
+      "grad_norm": 0.20579378306865692,
+      "learning_rate": 4.87050750287737e-06,
+      "loss": 1.0291,
+      "step": 17890
+    },
+    {
+      "epoch": 0.12957212244927505,
+      "grad_norm": 0.16314570605754852,
+      "learning_rate": 4.870435116216784e-06,
+      "loss": 1.0316,
+      "step": 17900
+    },
+    {
+      "epoch": 0.12964450910986122,
+      "grad_norm": 0.18022793531417847,
+      "learning_rate": 4.870362729556198e-06,
+      "loss": 1.0444,
+      "step": 17910
+    },
+    {
+      "epoch": 0.12971689577044743,
+      "grad_norm": 0.1798829734325409,
+      "learning_rate": 4.870290342895612e-06,
+      "loss": 1.0351,
+      "step": 17920
+    },
+    {
+      "epoch": 0.1297892824310336,
+      "grad_norm": 0.1796688586473465,
+      "learning_rate": 4.870217956235025e-06,
+      "loss": 1.0431,
+      "step": 17930
+    },
+    {
+      "epoch": 0.1298616690916198,
+      "grad_norm": 0.1828147917985916,
+      "learning_rate": 4.870145569574439e-06,
+      "loss": 1.0343,
+      "step": 17940
+    },
+    {
+      "epoch": 0.12993405575220598,
+      "grad_norm": 0.21611171960830688,
+      "learning_rate": 4.870073182913853e-06,
+      "loss": 1.0343,
+      "step": 17950
+    },
+    {
+      "epoch": 0.13000644241279216,
+      "grad_norm": 0.17004896700382233,
+      "learning_rate": 4.870000796253267e-06,
+      "loss": 1.0468,
+      "step": 17960
+    },
+    {
+      "epoch": 0.13007882907337837,
+      "grad_norm": 0.18028271198272705,
+      "learning_rate": 4.869928409592681e-06,
+      "loss": 1.0504,
+      "step": 17970
+    },
+    {
+      "epoch": 0.13015121573396454,
+      "grad_norm": 0.18797443807125092,
+      "learning_rate": 4.869856022932094e-06,
+      "loss": 1.0274,
+      "step": 17980
+    },
+    {
+      "epoch": 0.13022360239455072,
+      "grad_norm": 0.18303246796131134,
+      "learning_rate": 4.869783636271509e-06,
+      "loss": 1.0087,
+      "step": 17990
+    },
+    {
+      "epoch": 0.13029598905513692,
+      "grad_norm": 0.18918801844120026,
+      "learning_rate": 4.869711249610922e-06,
+      "loss": 1.0417,
+      "step": 18000
+    },
+    {
+      "epoch": 0.1303683757157231,
+      "grad_norm": 0.16673238575458527,
+      "learning_rate": 4.869638862950336e-06,
+      "loss": 1.0255,
+      "step": 18010
+    },
+    {
+      "epoch": 0.1304407623763093,
+      "grad_norm": 0.16749219596385956,
+      "learning_rate": 4.8695664762897495e-06,
+      "loss": 1.0267,
+      "step": 18020
+    },
+    {
+      "epoch": 0.13051314903689548,
+      "grad_norm": 0.1816338449716568,
+      "learning_rate": 4.869494089629164e-06,
+      "loss": 1.0432,
+      "step": 18030
+    },
+    {
+      "epoch": 0.13058553569748166,
+      "grad_norm": 0.1622341275215149,
+      "learning_rate": 4.869421702968578e-06,
+      "loss": 1.023,
+      "step": 18040
+    },
+    {
+      "epoch": 0.13065792235806786,
+      "grad_norm": 0.1742738038301468,
+      "learning_rate": 4.869349316307991e-06,
+      "loss": 1.0301,
+      "step": 18050
+    },
+    {
+      "epoch": 0.13073030901865404,
+      "grad_norm": 0.16464455425739288,
+      "learning_rate": 4.869276929647405e-06,
+      "loss": 1.0151,
+      "step": 18060
+    },
+    {
+      "epoch": 0.13080269567924022,
+      "grad_norm": 0.16827940940856934,
+      "learning_rate": 4.869204542986819e-06,
+      "loss": 1.0237,
+      "step": 18070
+    },
+    {
+      "epoch": 0.13087508233982642,
+      "grad_norm": 0.17623087763786316,
+      "learning_rate": 4.869132156326233e-06,
+      "loss": 1.0521,
+      "step": 18080
+    },
+    {
+      "epoch": 0.1309474690004126,
+      "grad_norm": 0.1728028506040573,
+      "learning_rate": 4.8690597696656465e-06,
+      "loss": 1.0373,
+      "step": 18090
+    },
+    {
+      "epoch": 0.1310198556609988,
+      "grad_norm": 0.17844252288341522,
+      "learning_rate": 4.86898738300506e-06,
+      "loss": 1.0461,
+      "step": 18100
+    },
+    {
+      "epoch": 0.13109224232158498,
+      "grad_norm": 0.16047002375125885,
+      "learning_rate": 4.868914996344475e-06,
+      "loss": 1.0312,
+      "step": 18110
+    },
+    {
+      "epoch": 0.13116462898217116,
+      "grad_norm": 0.18902169167995453,
+      "learning_rate": 4.868842609683888e-06,
+      "loss": 1.0386,
+      "step": 18120
+    },
+    {
+      "epoch": 0.13123701564275736,
+      "grad_norm": 0.18937572836875916,
+      "learning_rate": 4.868770223023302e-06,
+      "loss": 1.0421,
+      "step": 18130
+    },
+    {
+      "epoch": 0.13130940230334354,
+      "grad_norm": 0.17996446788311005,
+      "learning_rate": 4.8686978363627154e-06,
+      "loss": 1.0457,
+      "step": 18140
+    },
+    {
+      "epoch": 0.1313817889639297,
+      "grad_norm": 0.1748812198638916,
+      "learning_rate": 4.868625449702129e-06,
+      "loss": 1.0539,
+      "step": 18150
+    },
+    {
+      "epoch": 0.13145417562451592,
+      "grad_norm": 0.17151296138763428,
+      "learning_rate": 4.868553063041543e-06,
+      "loss": 1.0268,
+      "step": 18160
+    },
+    {
+      "epoch": 0.1315265622851021,
+      "grad_norm": 0.17143696546554565,
+      "learning_rate": 4.868480676380956e-06,
+      "loss": 1.0318,
+      "step": 18170
+    },
+    {
+      "epoch": 0.1315989489456883,
+      "grad_norm": 0.167258158326149,
+      "learning_rate": 4.868408289720371e-06,
+      "loss": 1.0264,
+      "step": 18180
+    },
+    {
+      "epoch": 0.13167133560627448,
+      "grad_norm": 0.17534078657627106,
+      "learning_rate": 4.868335903059784e-06,
+      "loss": 1.0408,
+      "step": 18190
+    },
+    {
+      "epoch": 0.13174372226686065,
+      "grad_norm": 0.17414483428001404,
+      "learning_rate": 4.868263516399198e-06,
+      "loss": 1.0265,
+      "step": 18200
+    },
+    {
+      "epoch": 0.13181610892744686,
+      "grad_norm": 0.1946071982383728,
+      "learning_rate": 4.868191129738612e-06,
+      "loss": 1.0186,
+      "step": 18210
+    },
+    {
+      "epoch": 0.13188849558803303,
+      "grad_norm": 0.1678534746170044,
+      "learning_rate": 4.868118743078026e-06,
+      "loss": 1.0336,
+      "step": 18220
+    },
+    {
+      "epoch": 0.13196088224861924,
+      "grad_norm": 0.1735198199748993,
+      "learning_rate": 4.86804635641744e-06,
+      "loss": 1.0279,
+      "step": 18230
+    },
+    {
+      "epoch": 0.13203326890920541,
+      "grad_norm": 0.19873082637786865,
+      "learning_rate": 4.867973969756853e-06,
+      "loss": 1.0294,
+      "step": 18240
+    },
+    {
+      "epoch": 0.1321056555697916,
+      "grad_norm": 0.17502813041210175,
+      "learning_rate": 4.867901583096267e-06,
+      "loss": 1.0235,
+      "step": 18250
+    },
+    {
+      "epoch": 0.1321780422303778,
+      "grad_norm": 0.18205420672893524,
+      "learning_rate": 4.867829196435681e-06,
+      "loss": 1.0364,
+      "step": 18260
+    },
+    {
+      "epoch": 0.13225042889096397,
+      "grad_norm": 0.16406495869159698,
+      "learning_rate": 4.867756809775095e-06,
+      "loss": 1.0416,
+      "step": 18270
+    },
+    {
+      "epoch": 0.13232281555155015,
+      "grad_norm": 0.2004392445087433,
+      "learning_rate": 4.867684423114509e-06,
+      "loss": 1.0369,
+      "step": 18280
+    },
+    {
+      "epoch": 0.13239520221213635,
+      "grad_norm": 0.16325534880161285,
+      "learning_rate": 4.867612036453922e-06,
+      "loss": 1.0315,
+      "step": 18290
+    },
+    {
+      "epoch": 0.13246758887272253,
+      "grad_norm": 0.17578786611557007,
+      "learning_rate": 4.867539649793337e-06,
+      "loss": 1.0262,
+      "step": 18300
+    },
+    {
+      "epoch": 0.13253997553330873,
+      "grad_norm": 0.18025441467761993,
+      "learning_rate": 4.86746726313275e-06,
+      "loss": 1.0369,
+      "step": 18310
+    },
+    {
+      "epoch": 0.1326123621938949,
+      "grad_norm": 0.1580292284488678,
+      "learning_rate": 4.867394876472164e-06,
+      "loss": 1.023,
+      "step": 18320
+    },
+    {
+      "epoch": 0.1326847488544811,
+      "grad_norm": 0.16052091121673584,
+      "learning_rate": 4.8673224898115775e-06,
+      "loss": 1.0332,
+      "step": 18330
+    },
+    {
+      "epoch": 0.1327571355150673,
+      "grad_norm": 0.17163227498531342,
+      "learning_rate": 4.867250103150992e-06,
+      "loss": 1.0384,
+      "step": 18340
+    },
+    {
+      "epoch": 0.13282952217565347,
+      "grad_norm": 0.17826439440250397,
+      "learning_rate": 4.867177716490406e-06,
+      "loss": 1.0355,
+      "step": 18350
+    },
+    {
+      "epoch": 0.13290190883623965,
+      "grad_norm": 0.16754212975502014,
+      "learning_rate": 4.867105329829819e-06,
+      "loss": 1.044,
+      "step": 18360
+    },
+    {
+      "epoch": 0.13297429549682585,
+      "grad_norm": 0.17139877378940582,
+      "learning_rate": 4.867032943169233e-06,
+      "loss": 1.0249,
+      "step": 18370
+    },
+    {
+      "epoch": 0.13304668215741203,
+      "grad_norm": 0.1751633733510971,
+      "learning_rate": 4.866960556508647e-06,
+      "loss": 1.0354,
+      "step": 18380
+    },
+    {
+      "epoch": 0.13311906881799823,
+      "grad_norm": 0.1777397096157074,
+      "learning_rate": 4.866888169848061e-06,
+      "loss": 1.0224,
+      "step": 18390
+    },
+    {
+      "epoch": 0.1331914554785844,
+      "grad_norm": 0.1866157501935959,
+      "learning_rate": 4.8668157831874746e-06,
+      "loss": 1.0231,
+      "step": 18400
+    },
+    {
+      "epoch": 0.13326384213917059,
+      "grad_norm": 0.2078569382429123,
+      "learning_rate": 4.866743396526888e-06,
+      "loss": 1.0317,
+      "step": 18410
+    },
+    {
+      "epoch": 0.1333362287997568,
+      "grad_norm": 0.1714392751455307,
+      "learning_rate": 4.866671009866303e-06,
+      "loss": 1.0303,
+      "step": 18420
+    },
+    {
+      "epoch": 0.13340861546034297,
+      "grad_norm": 0.1851416379213333,
+      "learning_rate": 4.866598623205716e-06,
+      "loss": 1.0321,
+      "step": 18430
+    },
+    {
+      "epoch": 0.13348100212092914,
+      "grad_norm": 0.17238068580627441,
+      "learning_rate": 4.86652623654513e-06,
+      "loss": 1.0313,
+      "step": 18440
+    },
+    {
+      "epoch": 0.13355338878151535,
+      "grad_norm": 0.18135380744934082,
+      "learning_rate": 4.8664538498845435e-06,
+      "loss": 1.0359,
+      "step": 18450
+    },
+    {
+      "epoch": 0.13362577544210152,
+      "grad_norm": 0.1726706624031067,
+      "learning_rate": 4.866381463223958e-06,
+      "loss": 1.0164,
+      "step": 18460
+    },
+    {
+      "epoch": 0.13369816210268773,
+      "grad_norm": 0.1773628294467926,
+      "learning_rate": 4.8663090765633716e-06,
+      "loss": 1.0397,
+      "step": 18470
+    },
+    {
+      "epoch": 0.1337705487632739,
+      "grad_norm": 0.17249688506126404,
+      "learning_rate": 4.866236689902785e-06,
+      "loss": 1.0314,
+      "step": 18480
+    },
+    {
+      "epoch": 0.13384293542386008,
+      "grad_norm": 0.17601031064987183,
+      "learning_rate": 4.866164303242199e-06,
+      "loss": 1.0412,
+      "step": 18490
+    },
+    {
+      "epoch": 0.1339153220844463,
+      "grad_norm": 0.17362290620803833,
+      "learning_rate": 4.866091916581613e-06,
+      "loss": 1.0335,
+      "step": 18500
+    },
+    {
+      "epoch": 0.13398770874503246,
+      "grad_norm": 0.17129851877689362,
+      "learning_rate": 4.866019529921027e-06,
+      "loss": 1.0282,
+      "step": 18510
+    },
+    {
+      "epoch": 0.13406009540561864,
+      "grad_norm": 0.18305164575576782,
+      "learning_rate": 4.8659471432604405e-06,
+      "loss": 1.0276,
+      "step": 18520
+    },
+    {
+      "epoch": 0.13413248206620484,
+      "grad_norm": 0.16306072473526,
+      "learning_rate": 4.865874756599854e-06,
+      "loss": 1.026,
+      "step": 18530
+    },
+    {
+      "epoch": 0.13420486872679102,
+      "grad_norm": 0.17902541160583496,
+      "learning_rate": 4.865802369939268e-06,
+      "loss": 1.0372,
+      "step": 18540
+    },
+    {
+      "epoch": 0.13427725538737723,
+      "grad_norm": 0.17102956771850586,
+      "learning_rate": 4.865729983278682e-06,
+      "loss": 1.0272,
+      "step": 18550
+    },
+    {
+      "epoch": 0.1343496420479634,
+      "grad_norm": 0.1690719872713089,
+      "learning_rate": 4.865657596618096e-06,
+      "loss": 1.0411,
+      "step": 18560
+    },
+    {
+      "epoch": 0.13442202870854958,
+      "grad_norm": 0.18555520474910736,
+      "learning_rate": 4.865585209957509e-06,
+      "loss": 1.0406,
+      "step": 18570
+    },
+    {
+      "epoch": 0.13449441536913578,
+      "grad_norm": 0.17667360603809357,
+      "learning_rate": 4.865512823296923e-06,
+      "loss": 1.0216,
+      "step": 18580
+    },
+    {
+      "epoch": 0.13456680202972196,
+      "grad_norm": 0.15590178966522217,
+      "learning_rate": 4.8654404366363375e-06,
+      "loss": 1.025,
+      "step": 18590
+    },
+    {
+      "epoch": 0.13463918869030814,
+      "grad_norm": 0.17240816354751587,
+      "learning_rate": 4.865368049975751e-06,
+      "loss": 1.0211,
+      "step": 18600
+    },
+    {
+      "epoch": 0.13471157535089434,
+      "grad_norm": 0.16866736114025116,
+      "learning_rate": 4.865295663315165e-06,
+      "loss": 1.0422,
+      "step": 18610
+    },
+    {
+      "epoch": 0.13478396201148052,
+      "grad_norm": 0.17667965590953827,
+      "learning_rate": 4.865223276654578e-06,
+      "loss": 1.0312,
+      "step": 18620
+    },
+    {
+      "epoch": 0.13485634867206672,
+      "grad_norm": 0.176561638712883,
+      "learning_rate": 4.865150889993993e-06,
+      "loss": 1.0363,
+      "step": 18630
+    },
+    {
+      "epoch": 0.1349287353326529,
+      "grad_norm": 0.18116410076618195,
+      "learning_rate": 4.8650785033334064e-06,
+      "loss": 1.0333,
+      "step": 18640
+    },
+    {
+      "epoch": 0.13500112199323908,
+      "grad_norm": 0.17595870792865753,
+      "learning_rate": 4.86500611667282e-06,
+      "loss": 1.0195,
+      "step": 18650
+    },
+    {
+      "epoch": 0.13507350865382528,
+      "grad_norm": 0.17091989517211914,
+      "learning_rate": 4.864933730012234e-06,
+      "loss": 1.0266,
+      "step": 18660
+    },
+    {
+      "epoch": 0.13514589531441146,
+      "grad_norm": 0.18030595779418945,
+      "learning_rate": 4.864861343351647e-06,
+      "loss": 1.0377,
+      "step": 18670
+    },
+    {
+      "epoch": 0.13521828197499763,
+      "grad_norm": 0.1966775506734848,
+      "learning_rate": 4.864788956691061e-06,
+      "loss": 1.0233,
+      "step": 18680
+    },
+    {
+      "epoch": 0.13529066863558384,
+      "grad_norm": 0.1909208595752716,
+      "learning_rate": 4.8647165700304745e-06,
+      "loss": 1.0294,
+      "step": 18690
+    },
+    {
+      "epoch": 0.13536305529617002,
+      "grad_norm": 0.17203876376152039,
+      "learning_rate": 4.864644183369889e-06,
+      "loss": 1.0422,
+      "step": 18700
+    },
+    {
+      "epoch": 0.13543544195675622,
+      "grad_norm": 0.164755716919899,
+      "learning_rate": 4.864571796709303e-06,
+      "loss": 1.0307,
+      "step": 18710
+    },
+    {
+      "epoch": 0.1355078286173424,
+      "grad_norm": 0.17235048115253448,
+      "learning_rate": 4.864499410048716e-06,
+      "loss": 1.025,
+      "step": 18720
+    },
+    {
+      "epoch": 0.13558021527792857,
+      "grad_norm": 0.17906124889850616,
+      "learning_rate": 4.86442702338813e-06,
+      "loss": 1.0251,
+      "step": 18730
+    },
+    {
+      "epoch": 0.13565260193851478,
+      "grad_norm": 0.1699845939874649,
+      "learning_rate": 4.864354636727544e-06,
+      "loss": 1.0513,
+      "step": 18740
+    },
+    {
+      "epoch": 0.13572498859910095,
+      "grad_norm": 0.2316533476114273,
+      "learning_rate": 4.864282250066958e-06,
+      "loss": 1.0299,
+      "step": 18750
+    },
+    {
+      "epoch": 0.13579737525968716,
+      "grad_norm": 0.1772417426109314,
+      "learning_rate": 4.8642098634063715e-06,
+      "loss": 1.014,
+      "step": 18760
+    },
+    {
+      "epoch": 0.13586976192027334,
+      "grad_norm": 0.17594146728515625,
+      "learning_rate": 4.864137476745785e-06,
+      "loss": 1.0376,
+      "step": 18770
+    },
+    {
+      "epoch": 0.1359421485808595,
+      "grad_norm": 0.1800992637872696,
+      "learning_rate": 4.8640650900852e-06,
+      "loss": 1.0218,
+      "step": 18780
+    },
+    {
+      "epoch": 0.13601453524144572,
+      "grad_norm": 0.2082536369562149,
+      "learning_rate": 4.863992703424613e-06,
+      "loss": 1.029,
+      "step": 18790
+    },
+    {
+      "epoch": 0.1360869219020319,
+      "grad_norm": 0.17295007407665253,
+      "learning_rate": 4.863920316764027e-06,
+      "loss": 1.0321,
+      "step": 18800
+    },
+    {
+      "epoch": 0.13615930856261807,
+      "grad_norm": 0.2445904165506363,
+      "learning_rate": 4.8638479301034404e-06,
+      "loss": 1.034,
+      "step": 18810
+    },
+    {
+      "epoch": 0.13623169522320427,
+      "grad_norm": 0.16870233416557312,
+      "learning_rate": 4.863775543442855e-06,
+      "loss": 1.0237,
+      "step": 18820
+    },
+    {
+      "epoch": 0.13630408188379045,
+      "grad_norm": 0.1675586849451065,
+      "learning_rate": 4.8637031567822685e-06,
+      "loss": 1.0238,
+      "step": 18830
+    },
+    {
+      "epoch": 0.13637646854437666,
+      "grad_norm": 0.1609523743391037,
+      "learning_rate": 4.863630770121682e-06,
+      "loss": 1.0141,
+      "step": 18840
+    },
+    {
+      "epoch": 0.13644885520496283,
+      "grad_norm": 0.1770259588956833,
+      "learning_rate": 4.863558383461096e-06,
+      "loss": 1.0507,
+      "step": 18850
+    },
+    {
+      "epoch": 0.136521241865549,
+      "grad_norm": 0.17467375099658966,
+      "learning_rate": 4.86348599680051e-06,
+      "loss": 1.0226,
+      "step": 18860
+    },
+    {
+      "epoch": 0.1365936285261352,
+      "grad_norm": 0.17384403944015503,
+      "learning_rate": 4.863413610139924e-06,
+      "loss": 1.0342,
+      "step": 18870
+    },
+    {
+      "epoch": 0.1366660151867214,
+      "grad_norm": 0.18507815897464752,
+      "learning_rate": 4.8633412234793374e-06,
+      "loss": 1.0338,
+      "step": 18880
+    },
+    {
+      "epoch": 0.13673840184730757,
+      "grad_norm": 0.173783078789711,
+      "learning_rate": 4.863268836818751e-06,
+      "loss": 1.0354,
+      "step": 18890
+    },
+    {
+      "epoch": 0.13681078850789377,
+      "grad_norm": 0.1754179745912552,
+      "learning_rate": 4.8631964501581655e-06,
+      "loss": 1.031,
+      "step": 18900
+    },
+    {
+      "epoch": 0.13688317516847995,
+      "grad_norm": 0.17524704337120056,
+      "learning_rate": 4.863124063497579e-06,
+      "loss": 1.0221,
+      "step": 18910
+    },
+    {
+      "epoch": 0.13695556182906615,
+      "grad_norm": 0.1778000146150589,
+      "learning_rate": 4.863051676836993e-06,
+      "loss": 1.0122,
+      "step": 18920
+    },
+    {
+      "epoch": 0.13702794848965233,
+      "grad_norm": 0.19337837398052216,
+      "learning_rate": 4.862979290176406e-06,
+      "loss": 1.0321,
+      "step": 18930
+    },
+    {
+      "epoch": 0.1371003351502385,
+      "grad_norm": 0.16937635838985443,
+      "learning_rate": 4.862906903515821e-06,
+      "loss": 1.0281,
+      "step": 18940
+    },
+    {
+      "epoch": 0.1371727218108247,
+      "grad_norm": 0.161665141582489,
+      "learning_rate": 4.8628345168552345e-06,
+      "loss": 1.0256,
+      "step": 18950
+    },
+    {
+      "epoch": 0.1372451084714109,
+      "grad_norm": 0.18332736194133759,
+      "learning_rate": 4.862762130194648e-06,
+      "loss": 1.034,
+      "step": 18960
+    },
+    {
+      "epoch": 0.13731749513199706,
+      "grad_norm": 0.17856566607952118,
+      "learning_rate": 4.862689743534062e-06,
+      "loss": 1.0334,
+      "step": 18970
+    },
+    {
+      "epoch": 0.13738988179258327,
+      "grad_norm": 0.16888518631458282,
+      "learning_rate": 4.862617356873476e-06,
+      "loss": 1.0146,
+      "step": 18980
+    },
+    {
+      "epoch": 0.13746226845316944,
+      "grad_norm": 0.18556654453277588,
+      "learning_rate": 4.86254497021289e-06,
+      "loss": 1.0241,
+      "step": 18990
+    },
+    {
+      "epoch": 0.13753465511375565,
+      "grad_norm": 0.21083232760429382,
+      "learning_rate": 4.862472583552303e-06,
+      "loss": 1.0315,
+      "step": 19000
+    },
+    {
+      "epoch": 0.13760704177434183,
+      "grad_norm": 0.194644495844841,
+      "learning_rate": 4.862400196891717e-06,
+      "loss": 1.0312,
+      "step": 19010
+    },
+    {
+      "epoch": 0.137679428434928,
+      "grad_norm": 0.19104093313217163,
+      "learning_rate": 4.8623278102311315e-06,
+      "loss": 1.0296,
+      "step": 19020
+    },
+    {
+      "epoch": 0.1377518150955142,
+      "grad_norm": 0.16693130135536194,
+      "learning_rate": 4.862255423570545e-06,
+      "loss": 1.0217,
+      "step": 19030
+    },
+    {
+      "epoch": 0.13782420175610038,
+      "grad_norm": 0.18064992129802704,
+      "learning_rate": 4.862183036909959e-06,
+      "loss": 1.0308,
+      "step": 19040
+    },
+    {
+      "epoch": 0.13789658841668656,
+      "grad_norm": 0.19371187686920166,
+      "learning_rate": 4.862110650249372e-06,
+      "loss": 1.032,
+      "step": 19050
+    },
+    {
+      "epoch": 0.13796897507727277,
+      "grad_norm": 0.17446281015872955,
+      "learning_rate": 4.862038263588787e-06,
+      "loss": 1.0323,
+      "step": 19060
+    },
+    {
+      "epoch": 0.13804136173785894,
+      "grad_norm": 0.17750053107738495,
+      "learning_rate": 4.8619658769282e-06,
+      "loss": 1.0269,
+      "step": 19070
+    },
+    {
+      "epoch": 0.13811374839844515,
+      "grad_norm": 0.1809011548757553,
+      "learning_rate": 4.861893490267614e-06,
+      "loss": 1.0242,
+      "step": 19080
+    },
+    {
+      "epoch": 0.13818613505903132,
+      "grad_norm": 0.19166511297225952,
+      "learning_rate": 4.861821103607028e-06,
+      "loss": 1.0228,
+      "step": 19090
+    },
+    {
+      "epoch": 0.1382585217196175,
+      "grad_norm": 0.17810720205307007,
+      "learning_rate": 4.861748716946442e-06,
+      "loss": 1.038,
+      "step": 19100
+    },
+    {
+      "epoch": 0.1383309083802037,
+      "grad_norm": 0.16615994274616241,
+      "learning_rate": 4.861676330285856e-06,
+      "loss": 1.027,
+      "step": 19110
+    },
+    {
+      "epoch": 0.13840329504078988,
+      "grad_norm": 0.16814115643501282,
+      "learning_rate": 4.861603943625269e-06,
+      "loss": 1.0254,
+      "step": 19120
+    },
+    {
+      "epoch": 0.13847568170137606,
+      "grad_norm": 0.1832701861858368,
+      "learning_rate": 4.861531556964683e-06,
+      "loss": 1.029,
+      "step": 19130
+    },
+    {
+      "epoch": 0.13854806836196226,
+      "grad_norm": 0.17128700017929077,
+      "learning_rate": 4.861459170304097e-06,
+      "loss": 1.0295,
+      "step": 19140
+    },
+    {
+      "epoch": 0.13862045502254844,
+      "grad_norm": 0.1750701367855072,
+      "learning_rate": 4.861386783643511e-06,
+      "loss": 1.0251,
+      "step": 19150
+    },
+    {
+      "epoch": 0.13869284168313464,
+      "grad_norm": 0.2172747105360031,
+      "learning_rate": 4.861314396982925e-06,
+      "loss": 1.022,
+      "step": 19160
+    },
+    {
+      "epoch": 0.13876522834372082,
+      "grad_norm": 0.17995861172676086,
+      "learning_rate": 4.861242010322338e-06,
+      "loss": 1.0308,
+      "step": 19170
+    },
+    {
+      "epoch": 0.138837615004307,
+      "grad_norm": 0.17807912826538086,
+      "learning_rate": 4.861169623661752e-06,
+      "loss": 1.0221,
+      "step": 19180
+    },
+    {
+      "epoch": 0.1389100016648932,
+      "grad_norm": 0.17291560769081116,
+      "learning_rate": 4.861097237001166e-06,
+      "loss": 1.023,
+      "step": 19190
+    },
+    {
+      "epoch": 0.13898238832547938,
+      "grad_norm": 0.21126574277877808,
+      "learning_rate": 4.861024850340579e-06,
+      "loss": 1.0267,
+      "step": 19200
+    },
+    {
+      "epoch": 0.13905477498606555,
+      "grad_norm": 0.178102508187294,
+      "learning_rate": 4.8609524636799936e-06,
+      "loss": 1.028,
+      "step": 19210
+    },
+    {
+      "epoch": 0.13912716164665176,
+      "grad_norm": 0.18170644342899323,
+      "learning_rate": 4.860880077019407e-06,
+      "loss": 1.0209,
+      "step": 19220
+    },
+    {
+      "epoch": 0.13919954830723794,
+      "grad_norm": 0.17773956060409546,
+      "learning_rate": 4.860807690358821e-06,
+      "loss": 1.0319,
+      "step": 19230
+    },
+    {
+      "epoch": 0.13927193496782414,
+      "grad_norm": 0.25084516406059265,
+      "learning_rate": 4.860735303698234e-06,
+      "loss": 1.027,
+      "step": 19240
+    },
+    {
+      "epoch": 0.13934432162841032,
+      "grad_norm": 0.18037767708301544,
+      "learning_rate": 4.860662917037649e-06,
+      "loss": 1.0325,
+      "step": 19250
+    },
+    {
+      "epoch": 0.1394167082889965,
+      "grad_norm": 0.1826416552066803,
+      "learning_rate": 4.8605905303770625e-06,
+      "loss": 1.0292,
+      "step": 19260
+    },
+    {
+      "epoch": 0.1394890949495827,
+      "grad_norm": 0.16938439011573792,
+      "learning_rate": 4.860518143716476e-06,
+      "loss": 1.0303,
+      "step": 19270
+    },
+    {
+      "epoch": 0.13956148161016887,
+      "grad_norm": 0.18216344714164734,
+      "learning_rate": 4.86044575705589e-06,
+      "loss": 1.0351,
+      "step": 19280
+    },
+    {
+      "epoch": 0.13963386827075505,
+      "grad_norm": 0.17035633325576782,
+      "learning_rate": 4.860373370395304e-06,
+      "loss": 1.0357,
+      "step": 19290
+    },
+    {
+      "epoch": 0.13970625493134126,
+      "grad_norm": 0.17245067656040192,
+      "learning_rate": 4.860300983734718e-06,
+      "loss": 1.0377,
+      "step": 19300
+    },
+    {
+      "epoch": 0.13977864159192743,
+      "grad_norm": 0.17423754930496216,
+      "learning_rate": 4.860228597074131e-06,
+      "loss": 1.0384,
+      "step": 19310
+    },
+    {
+      "epoch": 0.13985102825251364,
+      "grad_norm": 0.20927661657333374,
+      "learning_rate": 4.860156210413545e-06,
+      "loss": 1.0141,
+      "step": 19320
+    },
+    {
+      "epoch": 0.13992341491309981,
+      "grad_norm": 0.18298327922821045,
+      "learning_rate": 4.860083823752959e-06,
+      "loss": 1.0114,
+      "step": 19330
+    },
+    {
+      "epoch": 0.139995801573686,
+      "grad_norm": 0.1714993566274643,
+      "learning_rate": 4.860011437092373e-06,
+      "loss": 1.0275,
+      "step": 19340
+    },
+    {
+      "epoch": 0.1400681882342722,
+      "grad_norm": 0.1708700954914093,
+      "learning_rate": 4.859939050431787e-06,
+      "loss": 1.0364,
+      "step": 19350
+    },
+    {
+      "epoch": 0.14014057489485837,
+      "grad_norm": 0.18635700643062592,
+      "learning_rate": 4.8598666637712e-06,
+      "loss": 1.034,
+      "step": 19360
+    },
+    {
+      "epoch": 0.14021296155544458,
+      "grad_norm": 0.17284467816352844,
+      "learning_rate": 4.859794277110614e-06,
+      "loss": 1.0332,
+      "step": 19370
+    },
+    {
+      "epoch": 0.14028534821603075,
+      "grad_norm": 0.17173859477043152,
+      "learning_rate": 4.859721890450028e-06,
+      "loss": 1.0151,
+      "step": 19380
+    },
+    {
+      "epoch": 0.14035773487661693,
+      "grad_norm": 0.17994263768196106,
+      "learning_rate": 4.859649503789442e-06,
+      "loss": 1.0139,
+      "step": 19390
+    },
+    {
+      "epoch": 0.14043012153720313,
+      "grad_norm": 0.21482296288013458,
+      "learning_rate": 4.859577117128856e-06,
+      "loss": 1.0116,
+      "step": 19400
+    },
+    {
+      "epoch": 0.1405025081977893,
+      "grad_norm": 0.17747275531291962,
+      "learning_rate": 4.859504730468269e-06,
+      "loss": 1.0267,
+      "step": 19410
+    },
+    {
+      "epoch": 0.1405748948583755,
+      "grad_norm": 0.17190630733966827,
+      "learning_rate": 4.859432343807684e-06,
+      "loss": 1.0314,
+      "step": 19420
+    },
+    {
+      "epoch": 0.1406472815189617,
+      "grad_norm": 0.1733713150024414,
+      "learning_rate": 4.859359957147097e-06,
+      "loss": 1.0189,
+      "step": 19430
+    },
+    {
+      "epoch": 0.14071966817954787,
+      "grad_norm": 0.18654613196849823,
+      "learning_rate": 4.859287570486511e-06,
+      "loss": 1.0353,
+      "step": 19440
+    },
+    {
+      "epoch": 0.14079205484013407,
+      "grad_norm": 0.19242019951343536,
+      "learning_rate": 4.859215183825925e-06,
+      "loss": 1.0235,
+      "step": 19450
+    },
+    {
+      "epoch": 0.14086444150072025,
+      "grad_norm": 0.17690154910087585,
+      "learning_rate": 4.859142797165339e-06,
+      "loss": 1.0262,
+      "step": 19460
+    },
+    {
+      "epoch": 0.14093682816130643,
+      "grad_norm": 0.17208655178546906,
+      "learning_rate": 4.859070410504753e-06,
+      "loss": 1.026,
+      "step": 19470
+    },
+    {
+      "epoch": 0.14100921482189263,
+      "grad_norm": 0.1762927770614624,
+      "learning_rate": 4.858998023844166e-06,
+      "loss": 1.0367,
+      "step": 19480
+    },
+    {
+      "epoch": 0.1410816014824788,
+      "grad_norm": 0.18810449540615082,
+      "learning_rate": 4.85892563718358e-06,
+      "loss": 1.0267,
+      "step": 19490
+    },
+    {
+      "epoch": 0.14115398814306498,
+      "grad_norm": 0.23858876526355743,
+      "learning_rate": 4.858853250522994e-06,
+      "loss": 1.033,
+      "step": 19500
+    },
+    {
+      "epoch": 0.1412263748036512,
+      "grad_norm": 0.186942458152771,
+      "learning_rate": 4.858780863862408e-06,
+      "loss": 1.0223,
+      "step": 19510
+    },
+    {
+      "epoch": 0.14129876146423737,
+      "grad_norm": 0.1946224570274353,
+      "learning_rate": 4.858708477201822e-06,
+      "loss": 1.0169,
+      "step": 19520
+    },
+    {
+      "epoch": 0.14137114812482357,
+      "grad_norm": 0.1714710295200348,
+      "learning_rate": 4.858636090541235e-06,
+      "loss": 1.0199,
+      "step": 19530
+    },
+    {
+      "epoch": 0.14144353478540975,
+      "grad_norm": 0.17258310317993164,
+      "learning_rate": 4.85856370388065e-06,
+      "loss": 1.0274,
+      "step": 19540
+    },
+    {
+      "epoch": 0.14151592144599592,
+      "grad_norm": 0.18983785808086395,
+      "learning_rate": 4.858491317220063e-06,
+      "loss": 1.0382,
+      "step": 19550
+    },
+    {
+      "epoch": 0.14158830810658213,
+      "grad_norm": 0.17009888589382172,
+      "learning_rate": 4.858418930559477e-06,
+      "loss": 1.0138,
+      "step": 19560
+    },
+    {
+      "epoch": 0.1416606947671683,
+      "grad_norm": 0.17958617210388184,
+      "learning_rate": 4.8583465438988905e-06,
+      "loss": 1.0258,
+      "step": 19570
+    },
+    {
+      "epoch": 0.14173308142775448,
+      "grad_norm": 0.17548643052577972,
+      "learning_rate": 4.858274157238305e-06,
+      "loss": 1.0213,
+      "step": 19580
+    },
+    {
+      "epoch": 0.14180546808834069,
+      "grad_norm": 0.16976386308670044,
+      "learning_rate": 4.858201770577719e-06,
+      "loss": 1.0292,
+      "step": 19590
+    },
+    {
+      "epoch": 0.14187785474892686,
+      "grad_norm": 0.16231513023376465,
+      "learning_rate": 4.858129383917132e-06,
+      "loss": 1.0299,
+      "step": 19600
+    },
+    {
+      "epoch": 0.14195024140951307,
+      "grad_norm": 0.17872096598148346,
+      "learning_rate": 4.858056997256546e-06,
+      "loss": 1.0275,
+      "step": 19610
+    },
+    {
+      "epoch": 0.14202262807009924,
+      "grad_norm": 0.17281471192836761,
+      "learning_rate": 4.85798461059596e-06,
+      "loss": 1.0325,
+      "step": 19620
+    },
+    {
+      "epoch": 0.14209501473068542,
+      "grad_norm": 0.19203288853168488,
+      "learning_rate": 4.857912223935374e-06,
+      "loss": 1.0231,
+      "step": 19630
+    },
+    {
+      "epoch": 0.14216740139127163,
+      "grad_norm": 0.1869240552186966,
+      "learning_rate": 4.8578398372747875e-06,
+      "loss": 1.0303,
+      "step": 19640
+    },
+    {
+      "epoch": 0.1422397880518578,
+      "grad_norm": 0.18349246680736542,
+      "learning_rate": 4.857767450614201e-06,
+      "loss": 1.051,
+      "step": 19650
+    },
+    {
+      "epoch": 0.14231217471244398,
+      "grad_norm": 0.16620957851409912,
+      "learning_rate": 4.857695063953616e-06,
+      "loss": 1.0302,
+      "step": 19660
+    },
+    {
+      "epoch": 0.14238456137303018,
+      "grad_norm": 0.16156761348247528,
+      "learning_rate": 4.857622677293029e-06,
+      "loss": 1.0181,
+      "step": 19670
+    },
+    {
+      "epoch": 0.14245694803361636,
+      "grad_norm": 0.19458822906017303,
+      "learning_rate": 4.857550290632443e-06,
+      "loss": 1.027,
+      "step": 19680
+    },
+    {
+      "epoch": 0.14252933469420256,
+      "grad_norm": 0.17083300650119781,
+      "learning_rate": 4.8574779039718565e-06,
+      "loss": 1.0365,
+      "step": 19690
+    },
+    {
+      "epoch": 0.14260172135478874,
+      "grad_norm": 0.1720832884311676,
+      "learning_rate": 4.857405517311271e-06,
+      "loss": 1.0303,
+      "step": 19700
+    },
+    {
+      "epoch": 0.14267410801537492,
+      "grad_norm": 0.17224328219890594,
+      "learning_rate": 4.8573331306506845e-06,
+      "loss": 1.0318,
+      "step": 19710
+    },
+    {
+      "epoch": 0.14274649467596112,
+      "grad_norm": 0.16815780103206635,
+      "learning_rate": 4.857260743990098e-06,
+      "loss": 1.041,
+      "step": 19720
+    },
+    {
+      "epoch": 0.1428188813365473,
+      "grad_norm": 0.19083335995674133,
+      "learning_rate": 4.857188357329512e-06,
+      "loss": 1.0303,
+      "step": 19730
+    },
+    {
+      "epoch": 0.14289126799713348,
+      "grad_norm": 0.167524054646492,
+      "learning_rate": 4.857115970668925e-06,
+      "loss": 1.0343,
+      "step": 19740
+    },
+    {
+      "epoch": 0.14296365465771968,
+      "grad_norm": 0.17465034127235413,
+      "learning_rate": 4.857043584008339e-06,
+      "loss": 1.0161,
+      "step": 19750
+    },
+    {
+      "epoch": 0.14303604131830586,
+      "grad_norm": 0.18341541290283203,
+      "learning_rate": 4.856971197347753e-06,
+      "loss": 1.0134,
+      "step": 19760
+    },
+    {
+      "epoch": 0.14310842797889206,
+      "grad_norm": 0.1703953891992569,
+      "learning_rate": 4.856898810687167e-06,
+      "loss": 1.0333,
+      "step": 19770
+    },
+    {
+      "epoch": 0.14318081463947824,
+      "grad_norm": 0.17636318504810333,
+      "learning_rate": 4.856826424026581e-06,
+      "loss": 1.0271,
+      "step": 19780
+    },
+    {
+      "epoch": 0.14325320130006441,
+      "grad_norm": 0.17411428689956665,
+      "learning_rate": 4.856754037365994e-06,
+      "loss": 1.0399,
+      "step": 19790
+    },
+    {
+      "epoch": 0.14332558796065062,
+      "grad_norm": 0.18185856938362122,
+      "learning_rate": 4.856681650705408e-06,
+      "loss": 1.0305,
+      "step": 19800
+    },
+    {
+      "epoch": 0.1433979746212368,
+      "grad_norm": 0.17615829408168793,
+      "learning_rate": 4.856609264044822e-06,
+      "loss": 1.0266,
+      "step": 19810
+    },
+    {
+      "epoch": 0.14347036128182297,
+      "grad_norm": 0.19287104904651642,
+      "learning_rate": 4.856536877384236e-06,
+      "loss": 1.012,
+      "step": 19820
+    },
+    {
+      "epoch": 0.14354274794240918,
+      "grad_norm": 0.17633087933063507,
+      "learning_rate": 4.85646449072365e-06,
+      "loss": 1.0288,
+      "step": 19830
+    },
+    {
+      "epoch": 0.14361513460299535,
+      "grad_norm": 0.16220209002494812,
+      "learning_rate": 4.856392104063063e-06,
+      "loss": 1.0151,
+      "step": 19840
+    },
+    {
+      "epoch": 0.14368752126358156,
+      "grad_norm": 0.16982801258563995,
+      "learning_rate": 4.856319717402478e-06,
+      "loss": 1.0261,
+      "step": 19850
+    },
+    {
+      "epoch": 0.14375990792416773,
+      "grad_norm": 0.1811235100030899,
+      "learning_rate": 4.856247330741891e-06,
+      "loss": 1.0323,
+      "step": 19860
+    },
+    {
+      "epoch": 0.1438322945847539,
+      "grad_norm": 0.17505569756031036,
+      "learning_rate": 4.856174944081305e-06,
+      "loss": 1.0212,
+      "step": 19870
+    },
+    {
+      "epoch": 0.14390468124534012,
+      "grad_norm": 0.17915651202201843,
+      "learning_rate": 4.8561025574207186e-06,
+      "loss": 1.0347,
+      "step": 19880
+    },
+    {
+      "epoch": 0.1439770679059263,
+      "grad_norm": 0.16607633233070374,
+      "learning_rate": 4.856030170760133e-06,
+      "loss": 1.0331,
+      "step": 19890
+    },
+    {
+      "epoch": 0.14404945456651247,
+      "grad_norm": 0.1832243949174881,
+      "learning_rate": 4.855957784099547e-06,
+      "loss": 1.0218,
+      "step": 19900
+    },
+    {
+      "epoch": 0.14412184122709867,
+      "grad_norm": 0.17596963047981262,
+      "learning_rate": 4.85588539743896e-06,
+      "loss": 1.0321,
+      "step": 19910
+    },
+    {
+      "epoch": 0.14419422788768485,
+      "grad_norm": 0.17509303987026215,
+      "learning_rate": 4.855813010778374e-06,
+      "loss": 1.0224,
+      "step": 19920
+    },
+    {
+      "epoch": 0.14426661454827105,
+      "grad_norm": 0.159066841006279,
+      "learning_rate": 4.855740624117788e-06,
+      "loss": 1.0205,
+      "step": 19930
+    },
+    {
+      "epoch": 0.14433900120885723,
+      "grad_norm": 0.18700556457042694,
+      "learning_rate": 4.855668237457202e-06,
+      "loss": 1.0297,
+      "step": 19940
+    },
+    {
+      "epoch": 0.1444113878694434,
+      "grad_norm": 0.17160135507583618,
+      "learning_rate": 4.8555958507966156e-06,
+      "loss": 1.034,
+      "step": 19950
+    },
+    {
+      "epoch": 0.1444837745300296,
+      "grad_norm": 0.17142252624034882,
+      "learning_rate": 4.855523464136029e-06,
+      "loss": 1.0225,
+      "step": 19960
+    },
+    {
+      "epoch": 0.1445561611906158,
+      "grad_norm": 0.18951663374900818,
+      "learning_rate": 4.855451077475443e-06,
+      "loss": 1.0276,
+      "step": 19970
+    },
+    {
+      "epoch": 0.144628547851202,
+      "grad_norm": 0.1778341382741928,
+      "learning_rate": 4.855378690814857e-06,
+      "loss": 1.0116,
+      "step": 19980
+    },
+    {
+      "epoch": 0.14470093451178817,
+      "grad_norm": 0.1711929589509964,
+      "learning_rate": 4.855306304154271e-06,
+      "loss": 1.0469,
+      "step": 19990
+    },
+    {
+      "epoch": 0.14477332117237435,
+      "grad_norm": 0.1792709082365036,
+      "learning_rate": 4.8552339174936845e-06,
+      "loss": 1.0369,
+      "step": 20000
+    },
+    {
+      "epoch": 0.14484570783296055,
+      "grad_norm": 0.1821603775024414,
+      "learning_rate": 4.855161530833098e-06,
+      "loss": 1.0374,
+      "step": 20010
+    },
+    {
+      "epoch": 0.14491809449354673,
+      "grad_norm": 0.17707721889019012,
+      "learning_rate": 4.8550891441725126e-06,
+      "loss": 1.0276,
+      "step": 20020
+    },
+    {
+      "epoch": 0.1449904811541329,
+      "grad_norm": 0.18702900409698486,
+      "learning_rate": 4.855016757511926e-06,
+      "loss": 1.0198,
+      "step": 20030
+    },
+    {
+      "epoch": 0.1450628678147191,
+      "grad_norm": 0.16952428221702576,
+      "learning_rate": 4.85494437085134e-06,
+      "loss": 1.0333,
+      "step": 20040
+    },
+    {
+      "epoch": 0.1451352544753053,
+      "grad_norm": 0.17305567860603333,
+      "learning_rate": 4.854871984190753e-06,
+      "loss": 1.0139,
+      "step": 20050
+    },
+    {
+      "epoch": 0.1452076411358915,
+      "grad_norm": 0.17391686141490936,
+      "learning_rate": 4.854799597530168e-06,
+      "loss": 1.029,
+      "step": 20060
+    },
+    {
+      "epoch": 0.14528002779647767,
+      "grad_norm": 0.17623725533485413,
+      "learning_rate": 4.8547272108695815e-06,
+      "loss": 1.0245,
+      "step": 20070
+    },
+    {
+      "epoch": 0.14535241445706384,
+      "grad_norm": 0.17333059012889862,
+      "learning_rate": 4.854654824208995e-06,
+      "loss": 1.0297,
+      "step": 20080
+    },
+    {
+      "epoch": 0.14542480111765005,
+      "grad_norm": 0.1761116087436676,
+      "learning_rate": 4.854582437548409e-06,
+      "loss": 1.0236,
+      "step": 20090
+    },
+    {
+      "epoch": 0.14549718777823623,
+      "grad_norm": 0.16909906268119812,
+      "learning_rate": 4.854510050887823e-06,
+      "loss": 1.0219,
+      "step": 20100
+    },
+    {
+      "epoch": 0.1455695744388224,
+      "grad_norm": 0.1722634732723236,
+      "learning_rate": 4.854437664227237e-06,
+      "loss": 1.0271,
+      "step": 20110
+    },
+    {
+      "epoch": 0.1456419610994086,
+      "grad_norm": 0.32046079635620117,
+      "learning_rate": 4.85436527756665e-06,
+      "loss": 1.0361,
+      "step": 20120
+    },
+    {
+      "epoch": 0.14571434775999478,
+      "grad_norm": 0.18109475076198578,
+      "learning_rate": 4.854292890906064e-06,
+      "loss": 1.0265,
+      "step": 20130
+    },
+    {
+      "epoch": 0.145786734420581,
+      "grad_norm": 0.17990946769714355,
+      "learning_rate": 4.8542205042454785e-06,
+      "loss": 1.033,
+      "step": 20140
+    },
+    {
+      "epoch": 0.14585912108116716,
+      "grad_norm": 0.17502890527248383,
+      "learning_rate": 4.854148117584892e-06,
+      "loss": 1.0363,
+      "step": 20150
+    },
+    {
+      "epoch": 0.14593150774175334,
+      "grad_norm": 0.17126423120498657,
+      "learning_rate": 4.854075730924306e-06,
+      "loss": 1.0339,
+      "step": 20160
+    },
+    {
+      "epoch": 0.14600389440233955,
+      "grad_norm": 0.1773746907711029,
+      "learning_rate": 4.854003344263719e-06,
+      "loss": 1.038,
+      "step": 20170
+    },
+    {
+      "epoch": 0.14607628106292572,
+      "grad_norm": 0.1729922592639923,
+      "learning_rate": 4.853930957603134e-06,
+      "loss": 1.0151,
+      "step": 20180
+    },
+    {
+      "epoch": 0.1461486677235119,
+      "grad_norm": 0.1790771633386612,
+      "learning_rate": 4.8538585709425474e-06,
+      "loss": 1.0436,
+      "step": 20190
+    },
+    {
+      "epoch": 0.1462210543840981,
+      "grad_norm": 0.17333535850048065,
+      "learning_rate": 4.853786184281961e-06,
+      "loss": 1.0241,
+      "step": 20200
+    },
+    {
+      "epoch": 0.14629344104468428,
+      "grad_norm": 0.19140805304050446,
+      "learning_rate": 4.853713797621375e-06,
+      "loss": 1.0242,
+      "step": 20210
+    },
+    {
+      "epoch": 0.14636582770527048,
+      "grad_norm": 0.16760674118995667,
+      "learning_rate": 4.853641410960789e-06,
+      "loss": 1.0257,
+      "step": 20220
+    },
+    {
+      "epoch": 0.14643821436585666,
+      "grad_norm": 0.17233441770076752,
+      "learning_rate": 4.853569024300203e-06,
+      "loss": 1.0338,
+      "step": 20230
+    },
+    {
+      "epoch": 0.14651060102644284,
+      "grad_norm": 0.17265230417251587,
+      "learning_rate": 4.853496637639616e-06,
+      "loss": 1.037,
+      "step": 20240
+    },
+    {
+      "epoch": 0.14658298768702904,
+      "grad_norm": 0.17247281968593597,
+      "learning_rate": 4.85342425097903e-06,
+      "loss": 1.0141,
+      "step": 20250
+    },
+    {
+      "epoch": 0.14665537434761522,
+      "grad_norm": 0.1745438128709793,
+      "learning_rate": 4.853351864318444e-06,
+      "loss": 1.023,
+      "step": 20260
+    },
+    {
+      "epoch": 0.1467277610082014,
+      "grad_norm": 0.15753747522830963,
+      "learning_rate": 4.853279477657857e-06,
+      "loss": 1.0326,
+      "step": 20270
+    },
+    {
+      "epoch": 0.1468001476687876,
+      "grad_norm": 0.16413787007331848,
+      "learning_rate": 4.853207090997271e-06,
+      "loss": 1.021,
+      "step": 20280
+    },
+    {
+      "epoch": 0.14687253432937378,
+      "grad_norm": 0.1920785754919052,
+      "learning_rate": 4.853134704336685e-06,
+      "loss": 1.039,
+      "step": 20290
+    },
+    {
+      "epoch": 0.14694492098995998,
+      "grad_norm": 0.1672215759754181,
+      "learning_rate": 4.853062317676099e-06,
+      "loss": 1.0227,
+      "step": 20300
+    },
+    {
+      "epoch": 0.14701730765054616,
+      "grad_norm": 0.17362061142921448,
+      "learning_rate": 4.8529899310155125e-06,
+      "loss": 1.0167,
+      "step": 20310
+    },
+    {
+      "epoch": 0.14708969431113234,
+      "grad_norm": 0.22914515435695648,
+      "learning_rate": 4.852917544354926e-06,
+      "loss": 1.0369,
+      "step": 20320
+    },
+    {
+      "epoch": 0.14716208097171854,
+      "grad_norm": 0.2375790923833847,
+      "learning_rate": 4.852845157694341e-06,
+      "loss": 1.0304,
+      "step": 20330
+    },
+    {
+      "epoch": 0.14723446763230472,
+      "grad_norm": 0.1776515245437622,
+      "learning_rate": 4.852772771033754e-06,
+      "loss": 1.0327,
+      "step": 20340
+    },
+    {
+      "epoch": 0.1473068542928909,
+      "grad_norm": 0.18703965842723846,
+      "learning_rate": 4.852700384373168e-06,
+      "loss": 1.0377,
+      "step": 20350
+    },
+    {
+      "epoch": 0.1473792409534771,
+      "grad_norm": 0.18549805879592896,
+      "learning_rate": 4.8526279977125814e-06,
+      "loss": 1.0297,
+      "step": 20360
+    },
+    {
+      "epoch": 0.14745162761406327,
+      "grad_norm": 0.1732458919286728,
+      "learning_rate": 4.852555611051996e-06,
+      "loss": 1.0432,
+      "step": 20370
+    },
+    {
+      "epoch": 0.14752401427464948,
+      "grad_norm": 0.1937321275472641,
+      "learning_rate": 4.8524832243914095e-06,
+      "loss": 1.0219,
+      "step": 20380
+    },
+    {
+      "epoch": 0.14759640093523566,
+      "grad_norm": 0.2144630402326584,
+      "learning_rate": 4.852410837730823e-06,
+      "loss": 1.025,
+      "step": 20390
+    },
+    {
+      "epoch": 0.14766878759582183,
+      "grad_norm": 0.1818886548280716,
+      "learning_rate": 4.852338451070237e-06,
+      "loss": 1.0292,
+      "step": 20400
+    },
+    {
+      "epoch": 0.14774117425640804,
+      "grad_norm": 0.1875666230916977,
+      "learning_rate": 4.852266064409651e-06,
+      "loss": 1.0262,
+      "step": 20410
+    },
+    {
+      "epoch": 0.1478135609169942,
+      "grad_norm": 0.15505819022655487,
+      "learning_rate": 4.852193677749065e-06,
+      "loss": 1.0158,
+      "step": 20420
+    },
+    {
+      "epoch": 0.1478859475775804,
+      "grad_norm": 0.16805370151996613,
+      "learning_rate": 4.8521212910884785e-06,
+      "loss": 1.0337,
+      "step": 20430
+    },
+    {
+      "epoch": 0.1479583342381666,
+      "grad_norm": 0.18419931828975677,
+      "learning_rate": 4.852048904427892e-06,
+      "loss": 1.0314,
+      "step": 20440
+    },
+    {
+      "epoch": 0.14803072089875277,
+      "grad_norm": 0.1946050375699997,
+      "learning_rate": 4.8519765177673065e-06,
+      "loss": 1.0217,
+      "step": 20450
+    },
+    {
+      "epoch": 0.14810310755933898,
+      "grad_norm": 0.1928732991218567,
+      "learning_rate": 4.85190413110672e-06,
+      "loss": 1.0272,
+      "step": 20460
+    },
+    {
+      "epoch": 0.14817549421992515,
+      "grad_norm": 0.20039114356040955,
+      "learning_rate": 4.851831744446134e-06,
+      "loss": 1.0294,
+      "step": 20470
+    },
+    {
+      "epoch": 0.14824788088051133,
+      "grad_norm": 0.1852181851863861,
+      "learning_rate": 4.851759357785547e-06,
+      "loss": 1.0188,
+      "step": 20480
+    },
+    {
+      "epoch": 0.14832026754109753,
+      "grad_norm": 0.16493256390094757,
+      "learning_rate": 4.851686971124962e-06,
+      "loss": 1.0354,
+      "step": 20490
+    },
+    {
+      "epoch": 0.1483926542016837,
+      "grad_norm": 0.17907851934432983,
+      "learning_rate": 4.8516145844643755e-06,
+      "loss": 1.0362,
+      "step": 20500
+    },
+    {
+      "epoch": 0.14846504086226991,
+      "grad_norm": 0.19379720091819763,
+      "learning_rate": 4.851542197803789e-06,
+      "loss": 1.0367,
+      "step": 20510
+    },
+    {
+      "epoch": 0.1485374275228561,
+      "grad_norm": 0.1685412973165512,
+      "learning_rate": 4.851469811143203e-06,
+      "loss": 1.0292,
+      "step": 20520
+    },
+    {
+      "epoch": 0.14860981418344227,
+      "grad_norm": 0.17223286628723145,
+      "learning_rate": 4.851397424482617e-06,
+      "loss": 1.0302,
+      "step": 20530
+    },
+    {
+      "epoch": 0.14868220084402847,
+      "grad_norm": 0.17199258506298065,
+      "learning_rate": 4.851325037822031e-06,
+      "loss": 1.0299,
+      "step": 20540
+    },
+    {
+      "epoch": 0.14875458750461465,
+      "grad_norm": 0.1630372256040573,
+      "learning_rate": 4.851252651161444e-06,
+      "loss": 1.0267,
+      "step": 20550
+    },
+    {
+      "epoch": 0.14882697416520083,
+      "grad_norm": 0.17937202751636505,
+      "learning_rate": 4.851180264500858e-06,
+      "loss": 1.0288,
+      "step": 20560
+    },
+    {
+      "epoch": 0.14889936082578703,
+      "grad_norm": 0.1685437560081482,
+      "learning_rate": 4.851107877840272e-06,
+      "loss": 1.0382,
+      "step": 20570
+    },
+    {
+      "epoch": 0.1489717474863732,
+      "grad_norm": 0.7421558499336243,
+      "learning_rate": 4.851035491179686e-06,
+      "loss": 1.0395,
+      "step": 20580
+    },
+    {
+      "epoch": 0.1490441341469594,
+      "grad_norm": 0.1732497662305832,
+      "learning_rate": 4.8509631045191e-06,
+      "loss": 1.024,
+      "step": 20590
+    },
+    {
+      "epoch": 0.1491165208075456,
+      "grad_norm": 0.15808378159999847,
+      "learning_rate": 4.850890717858513e-06,
+      "loss": 1.0197,
+      "step": 20600
+    },
+    {
+      "epoch": 0.14918890746813177,
+      "grad_norm": 0.177708238363266,
+      "learning_rate": 4.850818331197927e-06,
+      "loss": 1.026,
+      "step": 20610
+    },
+    {
+      "epoch": 0.14926129412871797,
+      "grad_norm": 0.16309256851673126,
+      "learning_rate": 4.850745944537341e-06,
+      "loss": 1.0317,
+      "step": 20620
+    },
+    {
+      "epoch": 0.14933368078930415,
+      "grad_norm": 0.16510923206806183,
+      "learning_rate": 4.850673557876755e-06,
+      "loss": 1.0207,
+      "step": 20630
+    },
+    {
+      "epoch": 0.14940606744989032,
+      "grad_norm": 0.18781432509422302,
+      "learning_rate": 4.850601171216169e-06,
+      "loss": 1.0201,
+      "step": 20640
+    },
+    {
+      "epoch": 0.14947845411047653,
+      "grad_norm": 0.19044159352779388,
+      "learning_rate": 4.850528784555582e-06,
+      "loss": 1.0272,
+      "step": 20650
+    },
+    {
+      "epoch": 0.1495508407710627,
+      "grad_norm": 0.17361673712730408,
+      "learning_rate": 4.850456397894997e-06,
+      "loss": 1.0183,
+      "step": 20660
+    },
+    {
+      "epoch": 0.1496232274316489,
+      "grad_norm": 0.1650736778974533,
+      "learning_rate": 4.85038401123441e-06,
+      "loss": 1.0283,
+      "step": 20670
+    },
+    {
+      "epoch": 0.14969561409223509,
+      "grad_norm": 0.16908712685108185,
+      "learning_rate": 4.850311624573824e-06,
+      "loss": 1.0159,
+      "step": 20680
+    },
+    {
+      "epoch": 0.14976800075282126,
+      "grad_norm": 0.17872673273086548,
+      "learning_rate": 4.8502392379132376e-06,
+      "loss": 1.0202,
+      "step": 20690
+    },
+    {
+      "epoch": 0.14984038741340747,
+      "grad_norm": 0.16816575825214386,
+      "learning_rate": 4.850166851252652e-06,
+      "loss": 1.0228,
+      "step": 20700
+    },
+    {
+      "epoch": 0.14991277407399364,
+      "grad_norm": 0.16545483469963074,
+      "learning_rate": 4.850094464592066e-06,
+      "loss": 1.026,
+      "step": 20710
+    },
+    {
+      "epoch": 0.14998516073457982,
+      "grad_norm": 0.17488573491573334,
+      "learning_rate": 4.850022077931479e-06,
+      "loss": 1.0393,
+      "step": 20720
+    },
+    {
+      "epoch": 0.15005754739516602,
+      "grad_norm": 0.16280938684940338,
+      "learning_rate": 4.849949691270893e-06,
+      "loss": 1.0309,
+      "step": 20730
+    },
+    {
+      "epoch": 0.1501299340557522,
+      "grad_norm": 0.18939673900604248,
+      "learning_rate": 4.849877304610307e-06,
+      "loss": 1.034,
+      "step": 20740
+    },
+    {
+      "epoch": 0.1502023207163384,
+      "grad_norm": 0.16721650958061218,
+      "learning_rate": 4.849804917949721e-06,
+      "loss": 1.0227,
+      "step": 20750
+    },
+    {
+      "epoch": 0.15027470737692458,
+      "grad_norm": 0.1724279224872589,
+      "learning_rate": 4.8497325312891346e-06,
+      "loss": 1.0207,
+      "step": 20760
+    },
+    {
+      "epoch": 0.15034709403751076,
+      "grad_norm": 0.1731773465871811,
+      "learning_rate": 4.849660144628548e-06,
+      "loss": 1.032,
+      "step": 20770
+    },
+    {
+      "epoch": 0.15041948069809696,
+      "grad_norm": 0.15489158034324646,
+      "learning_rate": 4.849587757967963e-06,
+      "loss": 1.0216,
+      "step": 20780
+    },
+    {
+      "epoch": 0.15049186735868314,
+      "grad_norm": 0.1700417697429657,
+      "learning_rate": 4.849515371307375e-06,
+      "loss": 1.0307,
+      "step": 20790
+    },
+    {
+      "epoch": 0.15056425401926932,
+      "grad_norm": 0.19111177325248718,
+      "learning_rate": 4.849442984646789e-06,
+      "loss": 1.0101,
+      "step": 20800
+    },
+    {
+      "epoch": 0.15063664067985552,
+      "grad_norm": 0.17551042139530182,
+      "learning_rate": 4.8493705979862035e-06,
+      "loss": 1.0204,
+      "step": 20810
+    },
+    {
+      "epoch": 0.1507090273404417,
+      "grad_norm": 0.18322902917861938,
+      "learning_rate": 4.849298211325617e-06,
+      "loss": 1.0231,
+      "step": 20820
+    },
+    {
+      "epoch": 0.1507814140010279,
+      "grad_norm": 0.23962242901325226,
+      "learning_rate": 4.849225824665031e-06,
+      "loss": 1.0276,
+      "step": 20830
+    },
+    {
+      "epoch": 0.15085380066161408,
+      "grad_norm": 0.24165092408657074,
+      "learning_rate": 4.849153438004444e-06,
+      "loss": 1.0114,
+      "step": 20840
+    },
+    {
+      "epoch": 0.15092618732220026,
+      "grad_norm": 0.16115596890449524,
+      "learning_rate": 4.849081051343859e-06,
+      "loss": 1.0203,
+      "step": 20850
+    },
+    {
+      "epoch": 0.15099857398278646,
+      "grad_norm": 0.1693454384803772,
+      "learning_rate": 4.849008664683272e-06,
+      "loss": 1.0206,
+      "step": 20860
+    },
+    {
+      "epoch": 0.15107096064337264,
+      "grad_norm": 0.1848178654909134,
+      "learning_rate": 4.848936278022686e-06,
+      "loss": 1.0206,
+      "step": 20870
+    },
+    {
+      "epoch": 0.1511433473039588,
+      "grad_norm": 0.17669790983200073,
+      "learning_rate": 4.8488638913621e-06,
+      "loss": 1.0219,
+      "step": 20880
+    },
+    {
+      "epoch": 0.15121573396454502,
+      "grad_norm": 0.17501038312911987,
+      "learning_rate": 4.848791504701514e-06,
+      "loss": 1.0268,
+      "step": 20890
+    },
+    {
+      "epoch": 0.1512881206251312,
+      "grad_norm": 0.17689432203769684,
+      "learning_rate": 4.848719118040928e-06,
+      "loss": 1.0276,
+      "step": 20900
+    },
+    {
+      "epoch": 0.1513605072857174,
+      "grad_norm": 0.19285716116428375,
+      "learning_rate": 4.848646731380341e-06,
+      "loss": 1.0148,
+      "step": 20910
+    },
+    {
+      "epoch": 0.15143289394630358,
+      "grad_norm": 0.17144864797592163,
+      "learning_rate": 4.848574344719755e-06,
+      "loss": 1.0195,
+      "step": 20920
+    },
+    {
+      "epoch": 0.15150528060688975,
+      "grad_norm": 0.18200930953025818,
+      "learning_rate": 4.8485019580591694e-06,
+      "loss": 1.0325,
+      "step": 20930
+    },
+    {
+      "epoch": 0.15157766726747596,
+      "grad_norm": 0.17781402170658112,
+      "learning_rate": 4.848429571398583e-06,
+      "loss": 1.028,
+      "step": 20940
+    },
+    {
+      "epoch": 0.15165005392806213,
+      "grad_norm": 0.16629651188850403,
+      "learning_rate": 4.848357184737997e-06,
+      "loss": 1.0354,
+      "step": 20950
+    },
+    {
+      "epoch": 0.1517224405886483,
+      "grad_norm": 0.1842113435268402,
+      "learning_rate": 4.84828479807741e-06,
+      "loss": 1.0181,
+      "step": 20960
+    },
+    {
+      "epoch": 0.15179482724923452,
+      "grad_norm": 0.16665078699588776,
+      "learning_rate": 4.848212411416825e-06,
+      "loss": 1.0333,
+      "step": 20970
+    },
+    {
+      "epoch": 0.1518672139098207,
+      "grad_norm": 0.17545108497142792,
+      "learning_rate": 4.848140024756238e-06,
+      "loss": 1.0224,
+      "step": 20980
+    },
+    {
+      "epoch": 0.1519396005704069,
+      "grad_norm": 0.17534632980823517,
+      "learning_rate": 4.848067638095652e-06,
+      "loss": 1.035,
+      "step": 20990
+    },
+    {
+      "epoch": 0.15201198723099307,
+      "grad_norm": 0.16326074302196503,
+      "learning_rate": 4.847995251435066e-06,
+      "loss": 1.0243,
+      "step": 21000
+    },
+    {
+      "epoch": 0.15208437389157925,
+      "grad_norm": 0.16196899116039276,
+      "learning_rate": 4.84792286477448e-06,
+      "loss": 1.0239,
+      "step": 21010
+    },
+    {
+      "epoch": 0.15215676055216545,
+      "grad_norm": 0.17832061648368835,
+      "learning_rate": 4.847850478113894e-06,
+      "loss": 1.0319,
+      "step": 21020
+    },
+    {
+      "epoch": 0.15222914721275163,
+      "grad_norm": 0.16528917849063873,
+      "learning_rate": 4.847778091453307e-06,
+      "loss": 1.024,
+      "step": 21030
+    },
+    {
+      "epoch": 0.1523015338733378,
+      "grad_norm": 0.18502266705036163,
+      "learning_rate": 4.847705704792721e-06,
+      "loss": 1.0254,
+      "step": 21040
+    },
+    {
+      "epoch": 0.152373920533924,
+      "grad_norm": 0.21463727951049805,
+      "learning_rate": 4.847633318132135e-06,
+      "loss": 1.0331,
+      "step": 21050
+    },
+    {
+      "epoch": 0.1524463071945102,
+      "grad_norm": 0.1813105046749115,
+      "learning_rate": 4.847560931471549e-06,
+      "loss": 1.022,
+      "step": 21060
+    },
+    {
+      "epoch": 0.1525186938550964,
+      "grad_norm": 0.17173029482364655,
+      "learning_rate": 4.847488544810963e-06,
+      "loss": 1.0358,
+      "step": 21070
+    },
+    {
+      "epoch": 0.15259108051568257,
+      "grad_norm": 0.17391446232795715,
+      "learning_rate": 4.847416158150376e-06,
+      "loss": 1.0353,
+      "step": 21080
+    },
+    {
+      "epoch": 0.15266346717626875,
+      "grad_norm": 0.18049128353595734,
+      "learning_rate": 4.847343771489791e-06,
+      "loss": 1.0328,
+      "step": 21090
+    },
+    {
+      "epoch": 0.15273585383685495,
+      "grad_norm": 0.16343386471271515,
+      "learning_rate": 4.847271384829204e-06,
+      "loss": 1.0207,
+      "step": 21100
+    },
+    {
+      "epoch": 0.15280824049744113,
+      "grad_norm": 0.17857620120048523,
+      "learning_rate": 4.847198998168618e-06,
+      "loss": 1.0256,
+      "step": 21110
+    },
+    {
+      "epoch": 0.15288062715802733,
+      "grad_norm": 0.17615753412246704,
+      "learning_rate": 4.8471266115080315e-06,
+      "loss": 1.0197,
+      "step": 21120
+    },
+    {
+      "epoch": 0.1529530138186135,
+      "grad_norm": 0.26710131764411926,
+      "learning_rate": 4.847054224847446e-06,
+      "loss": 1.0344,
+      "step": 21130
+    },
+    {
+      "epoch": 0.15302540047919969,
+      "grad_norm": 0.17339998483657837,
+      "learning_rate": 4.84698183818686e-06,
+      "loss": 1.0031,
+      "step": 21140
+    },
+    {
+      "epoch": 0.1530977871397859,
+      "grad_norm": 0.1859639585018158,
+      "learning_rate": 4.846909451526273e-06,
+      "loss": 1.0236,
+      "step": 21150
+    },
+    {
+      "epoch": 0.15317017380037207,
+      "grad_norm": 0.16950179636478424,
+      "learning_rate": 4.846837064865687e-06,
+      "loss": 1.0251,
+      "step": 21160
+    },
+    {
+      "epoch": 0.15324256046095824,
+      "grad_norm": 0.18173764646053314,
+      "learning_rate": 4.846764678205101e-06,
+      "loss": 1.0126,
+      "step": 21170
+    },
+    {
+      "epoch": 0.15331494712154445,
+      "grad_norm": 0.1840379685163498,
+      "learning_rate": 4.846692291544515e-06,
+      "loss": 1.0327,
+      "step": 21180
+    },
+    {
+      "epoch": 0.15338733378213062,
+      "grad_norm": 0.1827646642923355,
+      "learning_rate": 4.8466199048839285e-06,
+      "loss": 1.0287,
+      "step": 21190
+    },
+    {
+      "epoch": 0.15345972044271683,
+      "grad_norm": 0.20936237275600433,
+      "learning_rate": 4.846547518223342e-06,
+      "loss": 1.0154,
+      "step": 21200
+    },
+    {
+      "epoch": 0.153532107103303,
+      "grad_norm": 0.20079828798770905,
+      "learning_rate": 4.846475131562756e-06,
+      "loss": 1.0166,
+      "step": 21210
+    },
+    {
+      "epoch": 0.15360449376388918,
+      "grad_norm": 0.1706727296113968,
+      "learning_rate": 4.84640274490217e-06,
+      "loss": 1.0317,
+      "step": 21220
+    },
+    {
+      "epoch": 0.1536768804244754,
+      "grad_norm": 0.1730235069990158,
+      "learning_rate": 4.846330358241584e-06,
+      "loss": 1.0206,
+      "step": 21230
+    },
+    {
+      "epoch": 0.15374926708506156,
+      "grad_norm": 0.17368663847446442,
+      "learning_rate": 4.8462579715809975e-06,
+      "loss": 1.026,
+      "step": 21240
+    },
+    {
+      "epoch": 0.15382165374564774,
+      "grad_norm": 0.19320766627788544,
+      "learning_rate": 4.846185584920411e-06,
+      "loss": 1.0455,
+      "step": 21250
+    },
+    {
+      "epoch": 0.15389404040623395,
+      "grad_norm": 0.17245323956012726,
+      "learning_rate": 4.8461131982598255e-06,
+      "loss": 1.0254,
+      "step": 21260
+    },
+    {
+      "epoch": 0.15396642706682012,
+      "grad_norm": 0.28055688738822937,
+      "learning_rate": 4.846040811599239e-06,
+      "loss": 1.0158,
+      "step": 21270
+    },
+    {
+      "epoch": 0.15403881372740633,
+      "grad_norm": 0.1800316721200943,
+      "learning_rate": 4.845968424938653e-06,
+      "loss": 1.0278,
+      "step": 21280
+    },
+    {
+      "epoch": 0.1541112003879925,
+      "grad_norm": 0.16145852208137512,
+      "learning_rate": 4.845896038278066e-06,
+      "loss": 1.0221,
+      "step": 21290
+    },
+    {
+      "epoch": 0.15418358704857868,
+      "grad_norm": 0.16589269042015076,
+      "learning_rate": 4.845823651617481e-06,
+      "loss": 1.029,
+      "step": 21300
+    },
+    {
+      "epoch": 0.15425597370916488,
+      "grad_norm": 0.17117495834827423,
+      "learning_rate": 4.8457512649568945e-06,
+      "loss": 1.012,
+      "step": 21310
+    },
+    {
+      "epoch": 0.15432836036975106,
+      "grad_norm": 0.16930687427520752,
+      "learning_rate": 4.845678878296308e-06,
+      "loss": 1.0132,
+      "step": 21320
+    },
+    {
+      "epoch": 0.15440074703033724,
+      "grad_norm": 0.15988081693649292,
+      "learning_rate": 4.845606491635722e-06,
+      "loss": 1.0226,
+      "step": 21330
+    },
+    {
+      "epoch": 0.15447313369092344,
+      "grad_norm": 0.16869573295116425,
+      "learning_rate": 4.845534104975135e-06,
+      "loss": 1.0183,
+      "step": 21340
+    },
+    {
+      "epoch": 0.15454552035150962,
+      "grad_norm": 0.1683894544839859,
+      "learning_rate": 4.845461718314549e-06,
+      "loss": 1.02,
+      "step": 21350
+    },
+    {
+      "epoch": 0.15461790701209582,
+      "grad_norm": 0.16887131333351135,
+      "learning_rate": 4.8453893316539625e-06,
+      "loss": 1.0191,
+      "step": 21360
+    },
+    {
+      "epoch": 0.154690293672682,
+      "grad_norm": 0.1669628620147705,
+      "learning_rate": 4.845316944993377e-06,
+      "loss": 1.0215,
+      "step": 21370
+    },
+    {
+      "epoch": 0.15476268033326818,
+      "grad_norm": 0.16394099593162537,
+      "learning_rate": 4.845244558332791e-06,
+      "loss": 1.0435,
+      "step": 21380
+    },
+    {
+      "epoch": 0.15483506699385438,
+      "grad_norm": 0.298396497964859,
+      "learning_rate": 4.845172171672204e-06,
+      "loss": 1.0062,
+      "step": 21390
+    },
+    {
+      "epoch": 0.15490745365444056,
+      "grad_norm": 0.1778578758239746,
+      "learning_rate": 4.845099785011618e-06,
+      "loss": 1.0383,
+      "step": 21400
+    },
+    {
+      "epoch": 0.15497984031502673,
+      "grad_norm": 0.16672207415103912,
+      "learning_rate": 4.845027398351032e-06,
+      "loss": 1.0257,
+      "step": 21410
+    },
+    {
+      "epoch": 0.15505222697561294,
+      "grad_norm": 0.1904023140668869,
+      "learning_rate": 4.844955011690446e-06,
+      "loss": 1.0183,
+      "step": 21420
+    },
+    {
+      "epoch": 0.15512461363619912,
+      "grad_norm": 0.19985061883926392,
+      "learning_rate": 4.8448826250298596e-06,
+      "loss": 1.0243,
+      "step": 21430
+    },
+    {
+      "epoch": 0.15519700029678532,
+      "grad_norm": 0.18588073551654816,
+      "learning_rate": 4.844810238369273e-06,
+      "loss": 1.0255,
+      "step": 21440
+    },
+    {
+      "epoch": 0.1552693869573715,
+      "grad_norm": 0.18870490789413452,
+      "learning_rate": 4.844737851708688e-06,
+      "loss": 1.0333,
+      "step": 21450
+    },
+    {
+      "epoch": 0.15534177361795767,
+      "grad_norm": 0.1897239238023758,
+      "learning_rate": 4.844665465048101e-06,
+      "loss": 1.0228,
+      "step": 21460
+    },
+    {
+      "epoch": 0.15541416027854388,
+      "grad_norm": 0.2040310651063919,
+      "learning_rate": 4.844593078387515e-06,
+      "loss": 1.0144,
+      "step": 21470
+    },
+    {
+      "epoch": 0.15548654693913005,
+      "grad_norm": 0.18327052891254425,
+      "learning_rate": 4.8445206917269285e-06,
+      "loss": 1.0259,
+      "step": 21480
+    },
+    {
+      "epoch": 0.15555893359971623,
+      "grad_norm": 0.2118920236825943,
+      "learning_rate": 4.844448305066343e-06,
+      "loss": 1.0299,
+      "step": 21490
+    },
+    {
+      "epoch": 0.15563132026030244,
+      "grad_norm": 0.17221951484680176,
+      "learning_rate": 4.8443759184057566e-06,
+      "loss": 1.0316,
+      "step": 21500
+    },
+    {
+      "epoch": 0.1557037069208886,
+      "grad_norm": 0.1662265807390213,
+      "learning_rate": 4.84430353174517e-06,
+      "loss": 1.0094,
+      "step": 21510
+    },
+    {
+      "epoch": 0.15577609358147482,
+      "grad_norm": 0.18612660467624664,
+      "learning_rate": 4.844231145084584e-06,
+      "loss": 1.0185,
+      "step": 21520
+    },
+    {
+      "epoch": 0.155848480242061,
+      "grad_norm": 0.17847616970539093,
+      "learning_rate": 4.844158758423998e-06,
+      "loss": 1.0138,
+      "step": 21530
+    },
+    {
+      "epoch": 0.15592086690264717,
+      "grad_norm": 0.17742417752742767,
+      "learning_rate": 4.844086371763412e-06,
+      "loss": 1.0129,
+      "step": 21540
+    },
+    {
+      "epoch": 0.15599325356323337,
+      "grad_norm": 0.17097236216068268,
+      "learning_rate": 4.8440139851028255e-06,
+      "loss": 1.0194,
+      "step": 21550
+    },
+    {
+      "epoch": 0.15606564022381955,
+      "grad_norm": 0.2134924679994583,
+      "learning_rate": 4.843941598442239e-06,
+      "loss": 1.0192,
+      "step": 21560
+    },
+    {
+      "epoch": 0.15613802688440573,
+      "grad_norm": 0.17918936908245087,
+      "learning_rate": 4.8438692117816536e-06,
+      "loss": 1.0286,
+      "step": 21570
+    },
+    {
+      "epoch": 0.15621041354499193,
+      "grad_norm": 0.16962049901485443,
+      "learning_rate": 4.843796825121067e-06,
+      "loss": 1.0243,
+      "step": 21580
+    },
+    {
+      "epoch": 0.1562828002055781,
+      "grad_norm": 0.18653210997581482,
+      "learning_rate": 4.843724438460481e-06,
+      "loss": 1.0321,
+      "step": 21590
+    },
+    {
+      "epoch": 0.15635518686616431,
+      "grad_norm": 0.1704837828874588,
+      "learning_rate": 4.843652051799894e-06,
+      "loss": 1.031,
+      "step": 21600
+    },
+    {
+      "epoch": 0.1564275735267505,
+      "grad_norm": 0.18651624023914337,
+      "learning_rate": 4.843579665139309e-06,
+      "loss": 1.0029,
+      "step": 21610
+    },
+    {
+      "epoch": 0.15649996018733667,
+      "grad_norm": 0.18372347950935364,
+      "learning_rate": 4.8435072784787225e-06,
+      "loss": 1.0356,
+      "step": 21620
+    },
+    {
+      "epoch": 0.15657234684792287,
+      "grad_norm": 0.18019145727157593,
+      "learning_rate": 4.843434891818136e-06,
+      "loss": 1.0289,
+      "step": 21630
+    },
+    {
+      "epoch": 0.15664473350850905,
+      "grad_norm": 0.1659417301416397,
+      "learning_rate": 4.84336250515755e-06,
+      "loss": 1.0187,
+      "step": 21640
+    },
+    {
+      "epoch": 0.15671712016909523,
+      "grad_norm": 0.18040567636489868,
+      "learning_rate": 4.843290118496964e-06,
+      "loss": 1.0321,
+      "step": 21650
+    },
+    {
+      "epoch": 0.15678950682968143,
+      "grad_norm": 0.1708545982837677,
+      "learning_rate": 4.843217731836378e-06,
+      "loss": 1.0207,
+      "step": 21660
+    },
+    {
+      "epoch": 0.1568618934902676,
+      "grad_norm": 0.16297031939029694,
+      "learning_rate": 4.8431453451757914e-06,
+      "loss": 1.0146,
+      "step": 21670
+    },
+    {
+      "epoch": 0.1569342801508538,
+      "grad_norm": 0.19065815210342407,
+      "learning_rate": 4.843072958515205e-06,
+      "loss": 1.0256,
+      "step": 21680
+    },
+    {
+      "epoch": 0.15700666681144,
+      "grad_norm": 0.17626740038394928,
+      "learning_rate": 4.8430005718546195e-06,
+      "loss": 1.025,
+      "step": 21690
+    },
+    {
+      "epoch": 0.15707905347202616,
+      "grad_norm": 0.1907099336385727,
+      "learning_rate": 4.842928185194033e-06,
+      "loss": 1.0163,
+      "step": 21700
+    },
+    {
+      "epoch": 0.15715144013261237,
+      "grad_norm": 0.1677362322807312,
+      "learning_rate": 4.842855798533447e-06,
+      "loss": 1.0341,
+      "step": 21710
+    },
+    {
+      "epoch": 0.15722382679319855,
+      "grad_norm": 0.18757909536361694,
+      "learning_rate": 4.84278341187286e-06,
+      "loss": 1.0201,
+      "step": 21720
+    },
+    {
+      "epoch": 0.15729621345378475,
+      "grad_norm": 0.16965140402317047,
+      "learning_rate": 4.842711025212275e-06,
+      "loss": 1.0217,
+      "step": 21730
+    },
+    {
+      "epoch": 0.15736860011437093,
+      "grad_norm": 0.16649499535560608,
+      "learning_rate": 4.8426386385516884e-06,
+      "loss": 1.0142,
+      "step": 21740
+    },
+    {
+      "epoch": 0.1574409867749571,
+      "grad_norm": 0.1895909309387207,
+      "learning_rate": 4.842566251891102e-06,
+      "loss": 1.017,
+      "step": 21750
+    },
+    {
+      "epoch": 0.1575133734355433,
+      "grad_norm": 0.17109711468219757,
+      "learning_rate": 4.842493865230516e-06,
+      "loss": 1.0203,
+      "step": 21760
+    },
+    {
+      "epoch": 0.15758576009612948,
+      "grad_norm": 0.16221435368061066,
+      "learning_rate": 4.84242147856993e-06,
+      "loss": 1.0098,
+      "step": 21770
+    },
+    {
+      "epoch": 0.15765814675671566,
+      "grad_norm": 0.17318131029605865,
+      "learning_rate": 4.842349091909344e-06,
+      "loss": 1.0307,
+      "step": 21780
+    },
+    {
+      "epoch": 0.15773053341730187,
+      "grad_norm": 0.17340058088302612,
+      "learning_rate": 4.842276705248757e-06,
+      "loss": 1.021,
+      "step": 21790
+    },
+    {
+      "epoch": 0.15780292007788804,
+      "grad_norm": 0.16992861032485962,
+      "learning_rate": 4.842204318588171e-06,
+      "loss": 1.0237,
+      "step": 21800
+    },
+    {
+      "epoch": 0.15787530673847425,
+      "grad_norm": 0.17952974140644073,
+      "learning_rate": 4.842131931927585e-06,
+      "loss": 1.0169,
+      "step": 21810
+    },
+    {
+      "epoch": 0.15794769339906042,
+      "grad_norm": 0.19869837164878845,
+      "learning_rate": 4.842059545266999e-06,
+      "loss": 1.0267,
+      "step": 21820
+    },
+    {
+      "epoch": 0.1580200800596466,
+      "grad_norm": 0.1871790587902069,
+      "learning_rate": 4.841987158606413e-06,
+      "loss": 1.017,
+      "step": 21830
+    },
+    {
+      "epoch": 0.1580924667202328,
+      "grad_norm": 0.183842271566391,
+      "learning_rate": 4.841914771945826e-06,
+      "loss": 1.0101,
+      "step": 21840
+    },
+    {
+      "epoch": 0.15816485338081898,
+      "grad_norm": 0.17067475616931915,
+      "learning_rate": 4.84184238528524e-06,
+      "loss": 1.0251,
+      "step": 21850
+    },
+    {
+      "epoch": 0.15823724004140516,
+      "grad_norm": 0.17248547077178955,
+      "learning_rate": 4.8417699986246535e-06,
+      "loss": 1.0154,
+      "step": 21860
+    },
+    {
+      "epoch": 0.15830962670199136,
+      "grad_norm": 0.17312128841876984,
+      "learning_rate": 4.841697611964067e-06,
+      "loss": 1.0098,
+      "step": 21870
+    },
+    {
+      "epoch": 0.15838201336257754,
+      "grad_norm": 0.17725808918476105,
+      "learning_rate": 4.841625225303482e-06,
+      "loss": 1.016,
+      "step": 21880
+    },
+    {
+      "epoch": 0.15845440002316374,
+      "grad_norm": 0.16956385970115662,
+      "learning_rate": 4.841552838642895e-06,
+      "loss": 1.0214,
+      "step": 21890
+    },
+    {
+      "epoch": 0.15852678668374992,
+      "grad_norm": 0.16064795851707458,
+      "learning_rate": 4.841480451982309e-06,
+      "loss": 1.0117,
+      "step": 21900
+    },
+    {
+      "epoch": 0.1585991733443361,
+      "grad_norm": 0.1749652624130249,
+      "learning_rate": 4.8414080653217225e-06,
+      "loss": 1.0257,
+      "step": 21910
+    },
+    {
+      "epoch": 0.1586715600049223,
+      "grad_norm": 0.18944776058197021,
+      "learning_rate": 4.841335678661137e-06,
+      "loss": 0.9973,
+      "step": 21920
+    },
+    {
+      "epoch": 0.15874394666550848,
+      "grad_norm": 0.24421219527721405,
+      "learning_rate": 4.8412632920005505e-06,
+      "loss": 1.0302,
+      "step": 21930
+    },
+    {
+      "epoch": 0.15881633332609466,
+      "grad_norm": 0.1638653576374054,
+      "learning_rate": 4.841190905339964e-06,
+      "loss": 1.0135,
+      "step": 21940
+    },
+    {
+      "epoch": 0.15888871998668086,
+      "grad_norm": 0.18156132102012634,
+      "learning_rate": 4.841118518679378e-06,
+      "loss": 1.0098,
+      "step": 21950
+    },
+    {
+      "epoch": 0.15896110664726704,
+      "grad_norm": 0.170707106590271,
+      "learning_rate": 4.841046132018792e-06,
+      "loss": 1.0142,
+      "step": 21960
+    },
+    {
+      "epoch": 0.15903349330785324,
+      "grad_norm": 0.17522519826889038,
+      "learning_rate": 4.840973745358206e-06,
+      "loss": 1.0156,
+      "step": 21970
+    },
+    {
+      "epoch": 0.15910587996843942,
+      "grad_norm": 0.17544645071029663,
+      "learning_rate": 4.8409013586976195e-06,
+      "loss": 1.0216,
+      "step": 21980
+    },
+    {
+      "epoch": 0.1591782666290256,
+      "grad_norm": 0.16112075746059418,
+      "learning_rate": 4.840828972037033e-06,
+      "loss": 1.0137,
+      "step": 21990
+    },
+    {
+      "epoch": 0.1592506532896118,
+      "grad_norm": 0.166812926530838,
+      "learning_rate": 4.840756585376447e-06,
+      "loss": 1.0234,
+      "step": 22000
+    },
+    {
+      "epoch": 0.15932303995019798,
+      "grad_norm": 0.19277237355709076,
+      "learning_rate": 4.840684198715861e-06,
+      "loss": 1.0112,
+      "step": 22010
+    },
+    {
+      "epoch": 0.15939542661078415,
+      "grad_norm": 0.18085378408432007,
+      "learning_rate": 4.840611812055275e-06,
+      "loss": 1.0177,
+      "step": 22020
+    },
+    {
+      "epoch": 0.15946781327137036,
+      "grad_norm": 0.2057359516620636,
+      "learning_rate": 4.840539425394688e-06,
+      "loss": 1.0294,
+      "step": 22030
+    },
+    {
+      "epoch": 0.15954019993195653,
+      "grad_norm": 0.165378138422966,
+      "learning_rate": 4.840467038734102e-06,
+      "loss": 1.0127,
+      "step": 22040
+    },
+    {
+      "epoch": 0.15961258659254274,
+      "grad_norm": 0.18065567314624786,
+      "learning_rate": 4.8403946520735165e-06,
+      "loss": 1.0238,
+      "step": 22050
+    },
+    {
+      "epoch": 0.15968497325312891,
+      "grad_norm": 0.1699487715959549,
+      "learning_rate": 4.84032226541293e-06,
+      "loss": 1.0217,
+      "step": 22060
+    },
+    {
+      "epoch": 0.1597573599137151,
+      "grad_norm": 0.22719596326351166,
+      "learning_rate": 4.840249878752344e-06,
+      "loss": 1.032,
+      "step": 22070
+    },
+    {
+      "epoch": 0.1598297465743013,
+      "grad_norm": 0.17852869629859924,
+      "learning_rate": 4.840177492091757e-06,
+      "loss": 1.0022,
+      "step": 22080
+    },
+    {
+      "epoch": 0.15990213323488747,
+      "grad_norm": 0.18880029022693634,
+      "learning_rate": 4.840105105431172e-06,
+      "loss": 1.0214,
+      "step": 22090
+    },
+    {
+      "epoch": 0.15997451989547365,
+      "grad_norm": 0.15838246047496796,
+      "learning_rate": 4.840032718770585e-06,
+      "loss": 1.0162,
+      "step": 22100
+    },
+    {
+      "epoch": 0.16004690655605985,
+      "grad_norm": 0.15795861184597015,
+      "learning_rate": 4.839960332109999e-06,
+      "loss": 1.0199,
+      "step": 22110
+    },
+    {
+      "epoch": 0.16011929321664603,
+      "grad_norm": 0.2182161509990692,
+      "learning_rate": 4.839887945449413e-06,
+      "loss": 1.0107,
+      "step": 22120
+    },
+    {
+      "epoch": 0.16019167987723223,
+      "grad_norm": 0.16949397325515747,
+      "learning_rate": 4.839815558788827e-06,
+      "loss": 1.0171,
+      "step": 22130
+    },
+    {
+      "epoch": 0.1602640665378184,
+      "grad_norm": 0.16689515113830566,
+      "learning_rate": 4.839743172128241e-06,
+      "loss": 1.0128,
+      "step": 22140
+    },
+    {
+      "epoch": 0.1603364531984046,
+      "grad_norm": 0.19180209934711456,
+      "learning_rate": 4.839670785467654e-06,
+      "loss": 1.0162,
+      "step": 22150
+    },
+    {
+      "epoch": 0.1604088398589908,
+      "grad_norm": 0.16757026314735413,
+      "learning_rate": 4.839598398807068e-06,
+      "loss": 1.0052,
+      "step": 22160
+    },
+    {
+      "epoch": 0.16048122651957697,
+      "grad_norm": 0.1716785430908203,
+      "learning_rate": 4.839526012146482e-06,
+      "loss": 1.0202,
+      "step": 22170
+    },
+    {
+      "epoch": 0.16055361318016315,
+      "grad_norm": 0.17269423604011536,
+      "learning_rate": 4.839453625485896e-06,
+      "loss": 1.0103,
+      "step": 22180
+    },
+    {
+      "epoch": 0.16062599984074935,
+      "grad_norm": 0.22084921598434448,
+      "learning_rate": 4.83938123882531e-06,
+      "loss": 1.0143,
+      "step": 22190
+    },
+    {
+      "epoch": 0.16069838650133553,
+      "grad_norm": 0.18779946863651276,
+      "learning_rate": 4.839308852164723e-06,
+      "loss": 1.016,
+      "step": 22200
+    },
+    {
+      "epoch": 0.16077077316192173,
+      "grad_norm": 0.17556487023830414,
+      "learning_rate": 4.839236465504138e-06,
+      "loss": 1.0178,
+      "step": 22210
+    },
+    {
+      "epoch": 0.1608431598225079,
+      "grad_norm": 0.164267897605896,
+      "learning_rate": 4.839164078843551e-06,
+      "loss": 1.0224,
+      "step": 22220
+    },
+    {
+      "epoch": 0.16091554648309409,
+      "grad_norm": 0.20508578419685364,
+      "learning_rate": 4.839091692182965e-06,
+      "loss": 1.0087,
+      "step": 22230
+    },
+    {
+      "epoch": 0.1609879331436803,
+      "grad_norm": 0.46417108178138733,
+      "learning_rate": 4.8390193055223786e-06,
+      "loss": 1.0203,
+      "step": 22240
+    },
+    {
+      "epoch": 0.16106031980426647,
+      "grad_norm": 0.19800300896167755,
+      "learning_rate": 4.838946918861793e-06,
+      "loss": 1.0203,
+      "step": 22250
+    },
+    {
+      "epoch": 0.16113270646485267,
+      "grad_norm": 0.17719148099422455,
+      "learning_rate": 4.838874532201207e-06,
+      "loss": 1.0306,
+      "step": 22260
+    },
+    {
+      "epoch": 0.16120509312543885,
+      "grad_norm": 0.16901084780693054,
+      "learning_rate": 4.83880214554062e-06,
+      "loss": 1.014,
+      "step": 22270
+    },
+    {
+      "epoch": 0.16127747978602502,
+      "grad_norm": 0.1782984584569931,
+      "learning_rate": 4.838729758880034e-06,
+      "loss": 1.0268,
+      "step": 22280
+    },
+    {
+      "epoch": 0.16134986644661123,
+      "grad_norm": 0.18408305943012238,
+      "learning_rate": 4.838657372219448e-06,
+      "loss": 1.0163,
+      "step": 22290
+    },
+    {
+      "epoch": 0.1614222531071974,
+      "grad_norm": 0.16512493789196014,
+      "learning_rate": 4.838584985558862e-06,
+      "loss": 1.0251,
+      "step": 22300
+    },
+    {
+      "epoch": 0.16149463976778358,
+      "grad_norm": 0.180845245718956,
+      "learning_rate": 4.8385125988982756e-06,
+      "loss": 1.019,
+      "step": 22310
+    },
+    {
+      "epoch": 0.1615670264283698,
+      "grad_norm": 0.17334243655204773,
+      "learning_rate": 4.838440212237689e-06,
+      "loss": 1.0123,
+      "step": 22320
+    },
+    {
+      "epoch": 0.16163941308895596,
+      "grad_norm": 0.16331173479557037,
+      "learning_rate": 4.838367825577104e-06,
+      "loss": 1.0139,
+      "step": 22330
+    },
+    {
+      "epoch": 0.16171179974954217,
+      "grad_norm": 0.17761701345443726,
+      "learning_rate": 4.838295438916517e-06,
+      "loss": 1.0381,
+      "step": 22340
+    },
+    {
+      "epoch": 0.16178418641012834,
+      "grad_norm": 0.162800133228302,
+      "learning_rate": 4.838223052255931e-06,
+      "loss": 1.026,
+      "step": 22350
+    },
+    {
+      "epoch": 0.16185657307071452,
+      "grad_norm": 0.18399173021316528,
+      "learning_rate": 4.8381506655953445e-06,
+      "loss": 1.0155,
+      "step": 22360
+    },
+    {
+      "epoch": 0.16192895973130073,
+      "grad_norm": 0.18210694193840027,
+      "learning_rate": 4.838078278934759e-06,
+      "loss": 1.0326,
+      "step": 22370
+    },
+    {
+      "epoch": 0.1620013463918869,
+      "grad_norm": 0.19371944665908813,
+      "learning_rate": 4.838005892274173e-06,
+      "loss": 1.0091,
+      "step": 22380
+    },
+    {
+      "epoch": 0.16207373305247308,
+      "grad_norm": 0.2175704389810562,
+      "learning_rate": 4.837933505613585e-06,
+      "loss": 1.0094,
+      "step": 22390
+    },
+    {
+      "epoch": 0.16214611971305928,
+      "grad_norm": 0.17889146506786346,
+      "learning_rate": 4.837861118953e-06,
+      "loss": 1.0349,
+      "step": 22400
+    },
+    {
+      "epoch": 0.16221850637364546,
+      "grad_norm": 0.17176759243011475,
+      "learning_rate": 4.8377887322924134e-06,
+      "loss": 1.0191,
+      "step": 22410
+    },
+    {
+      "epoch": 0.16229089303423166,
+      "grad_norm": 0.17913545668125153,
+      "learning_rate": 4.837716345631827e-06,
+      "loss": 1.0136,
+      "step": 22420
+    },
+    {
+      "epoch": 0.16236327969481784,
+      "grad_norm": 0.17989039421081543,
+      "learning_rate": 4.837643958971241e-06,
+      "loss": 1.0442,
+      "step": 22430
+    },
+    {
+      "epoch": 0.16243566635540402,
+      "grad_norm": 0.165592759847641,
+      "learning_rate": 4.837571572310655e-06,
+      "loss": 1.0248,
+      "step": 22440
+    },
+    {
+      "epoch": 0.16250805301599022,
+      "grad_norm": 0.179484561085701,
+      "learning_rate": 4.837499185650069e-06,
+      "loss": 1.0218,
+      "step": 22450
+    },
+    {
+      "epoch": 0.1625804396765764,
+      "grad_norm": 0.17062394320964813,
+      "learning_rate": 4.837426798989482e-06,
+      "loss": 1.0168,
+      "step": 22460
+    },
+    {
+      "epoch": 0.16265282633716258,
+      "grad_norm": 0.1736491620540619,
+      "learning_rate": 4.837354412328896e-06,
+      "loss": 1.0122,
+      "step": 22470
+    },
+    {
+      "epoch": 0.16272521299774878,
+      "grad_norm": 0.17264223098754883,
+      "learning_rate": 4.8372820256683104e-06,
+      "loss": 1.0215,
+      "step": 22480
+    },
+    {
+      "epoch": 0.16279759965833496,
+      "grad_norm": 0.18589048087596893,
+      "learning_rate": 4.837209639007724e-06,
+      "loss": 1.0197,
+      "step": 22490
+    },
+    {
+      "epoch": 0.16286998631892116,
+      "grad_norm": 0.16932156682014465,
+      "learning_rate": 4.837137252347138e-06,
+      "loss": 1.0257,
+      "step": 22500
+    },
+    {
+      "epoch": 0.16294237297950734,
+      "grad_norm": 0.17345447838306427,
+      "learning_rate": 4.837064865686551e-06,
+      "loss": 1.0228,
+      "step": 22510
+    },
+    {
+      "epoch": 0.16301475964009352,
+      "grad_norm": 0.16551077365875244,
+      "learning_rate": 4.836992479025966e-06,
+      "loss": 1.0266,
+      "step": 22520
+    },
+    {
+      "epoch": 0.16308714630067972,
+      "grad_norm": 0.1778588742017746,
+      "learning_rate": 4.836920092365379e-06,
+      "loss": 1.0194,
+      "step": 22530
+    },
+    {
+      "epoch": 0.1631595329612659,
+      "grad_norm": 0.16605743765830994,
+      "learning_rate": 4.836847705704793e-06,
+      "loss": 1.0151,
+      "step": 22540
+    },
+    {
+      "epoch": 0.16323191962185207,
+      "grad_norm": 0.18560338020324707,
+      "learning_rate": 4.836775319044207e-06,
+      "loss": 1.017,
+      "step": 22550
+    },
+    {
+      "epoch": 0.16330430628243828,
+      "grad_norm": 0.1800818145275116,
+      "learning_rate": 4.836702932383621e-06,
+      "loss": 1.0375,
+      "step": 22560
+    },
+    {
+      "epoch": 0.16337669294302445,
+      "grad_norm": 0.20372332632541656,
+      "learning_rate": 4.836630545723035e-06,
+      "loss": 1.0101,
+      "step": 22570
+    },
+    {
+      "epoch": 0.16344907960361066,
+      "grad_norm": 0.17202037572860718,
+      "learning_rate": 4.836558159062448e-06,
+      "loss": 1.0169,
+      "step": 22580
+    },
+    {
+      "epoch": 0.16352146626419684,
+      "grad_norm": 0.17716839909553528,
+      "learning_rate": 4.836485772401862e-06,
+      "loss": 1.0207,
+      "step": 22590
+    },
+    {
+      "epoch": 0.163593852924783,
+      "grad_norm": 0.18483179807662964,
+      "learning_rate": 4.836413385741276e-06,
+      "loss": 1.004,
+      "step": 22600
+    },
+    {
+      "epoch": 0.16366623958536922,
+      "grad_norm": 0.18177522718906403,
+      "learning_rate": 4.83634099908069e-06,
+      "loss": 1.005,
+      "step": 22610
+    },
+    {
+      "epoch": 0.1637386262459554,
+      "grad_norm": 0.1648116558790207,
+      "learning_rate": 4.836268612420104e-06,
+      "loss": 1.0109,
+      "step": 22620
+    },
+    {
+      "epoch": 0.16381101290654157,
+      "grad_norm": 0.169694185256958,
+      "learning_rate": 4.836196225759517e-06,
+      "loss": 1.0065,
+      "step": 22630
+    },
+    {
+      "epoch": 0.16388339956712777,
+      "grad_norm": 0.1901792734861374,
+      "learning_rate": 4.836123839098931e-06,
+      "loss": 1.0206,
+      "step": 22640
+    },
+    {
+      "epoch": 0.16395578622771395,
+      "grad_norm": 0.1724863350391388,
+      "learning_rate": 4.836051452438345e-06,
+      "loss": 1.0276,
+      "step": 22650
+    },
+    {
+      "epoch": 0.16402817288830016,
+      "grad_norm": 0.17653776705265045,
+      "learning_rate": 4.835979065777759e-06,
+      "loss": 1.0317,
+      "step": 22660
+    },
+    {
+      "epoch": 0.16410055954888633,
+      "grad_norm": 0.1690499484539032,
+      "learning_rate": 4.8359066791171725e-06,
+      "loss": 1.0252,
+      "step": 22670
+    },
+    {
+      "epoch": 0.1641729462094725,
+      "grad_norm": 0.1637679785490036,
+      "learning_rate": 4.835834292456586e-06,
+      "loss": 1.0172,
+      "step": 22680
+    },
+    {
+      "epoch": 0.1642453328700587,
+      "grad_norm": 0.17211398482322693,
+      "learning_rate": 4.835761905796001e-06,
+      "loss": 1.018,
+      "step": 22690
+    },
+    {
+      "epoch": 0.1643177195306449,
+      "grad_norm": 0.18994399905204773,
+      "learning_rate": 4.835689519135414e-06,
+      "loss": 1.0102,
+      "step": 22700
+    },
+    {
+      "epoch": 0.16439010619123107,
+      "grad_norm": 0.17697584629058838,
+      "learning_rate": 4.835617132474828e-06,
+      "loss": 1.0071,
+      "step": 22710
+    },
+    {
+      "epoch": 0.16446249285181727,
+      "grad_norm": 0.17998354136943817,
+      "learning_rate": 4.8355447458142415e-06,
+      "loss": 1.0074,
+      "step": 22720
+    },
+    {
+      "epoch": 0.16453487951240345,
+      "grad_norm": 0.17211481928825378,
+      "learning_rate": 4.835472359153656e-06,
+      "loss": 1.0121,
+      "step": 22730
+    },
+    {
+      "epoch": 0.16460726617298965,
+      "grad_norm": 0.19927141070365906,
+      "learning_rate": 4.8353999724930695e-06,
+      "loss": 1.026,
+      "step": 22740
+    },
+    {
+      "epoch": 0.16467965283357583,
+      "grad_norm": 0.18247176706790924,
+      "learning_rate": 4.835327585832483e-06,
+      "loss": 1.0036,
+      "step": 22750
+    },
+    {
+      "epoch": 0.164752039494162,
+      "grad_norm": 0.1697113811969757,
+      "learning_rate": 4.835255199171897e-06,
+      "loss": 1.011,
+      "step": 22760
+    },
+    {
+      "epoch": 0.1648244261547482,
+      "grad_norm": 0.1692165732383728,
+      "learning_rate": 4.835182812511311e-06,
+      "loss": 1.0219,
+      "step": 22770
+    },
+    {
+      "epoch": 0.1648968128153344,
+      "grad_norm": 0.18225090205669403,
+      "learning_rate": 4.835110425850725e-06,
+      "loss": 1.0236,
+      "step": 22780
+    },
+    {
+      "epoch": 0.16496919947592056,
+      "grad_norm": 0.17765475809574127,
+      "learning_rate": 4.8350380391901385e-06,
+      "loss": 1.0097,
+      "step": 22790
+    },
+    {
+      "epoch": 0.16504158613650677,
+      "grad_norm": 0.17133235931396484,
+      "learning_rate": 4.834965652529552e-06,
+      "loss": 1.0139,
+      "step": 22800
+    },
+    {
+      "epoch": 0.16511397279709294,
+      "grad_norm": 0.1649719774723053,
+      "learning_rate": 4.8348932658689665e-06,
+      "loss": 1.0256,
+      "step": 22810
+    },
+    {
+      "epoch": 0.16518635945767915,
+      "grad_norm": 0.1826830953359604,
+      "learning_rate": 4.83482087920838e-06,
+      "loss": 1.0307,
+      "step": 22820
+    },
+    {
+      "epoch": 0.16525874611826533,
+      "grad_norm": 0.16446372866630554,
+      "learning_rate": 4.834748492547794e-06,
+      "loss": 1.0148,
+      "step": 22830
+    },
+    {
+      "epoch": 0.1653311327788515,
+      "grad_norm": 0.16916634142398834,
+      "learning_rate": 4.834676105887207e-06,
+      "loss": 1.0137,
+      "step": 22840
+    },
+    {
+      "epoch": 0.1654035194394377,
+      "grad_norm": 0.18869148194789886,
+      "learning_rate": 4.834603719226622e-06,
+      "loss": 1.0212,
+      "step": 22850
+    },
+    {
+      "epoch": 0.16547590610002388,
+      "grad_norm": 0.2165064662694931,
+      "learning_rate": 4.8345313325660355e-06,
+      "loss": 1.0222,
+      "step": 22860
+    },
+    {
+      "epoch": 0.1655482927606101,
+      "grad_norm": 0.17655514180660248,
+      "learning_rate": 4.834458945905449e-06,
+      "loss": 1.0227,
+      "step": 22870
+    },
+    {
+      "epoch": 0.16562067942119627,
+      "grad_norm": 0.17349255084991455,
+      "learning_rate": 4.834386559244863e-06,
+      "loss": 1.0176,
+      "step": 22880
+    },
+    {
+      "epoch": 0.16569306608178244,
+      "grad_norm": 0.1745917946100235,
+      "learning_rate": 4.834314172584277e-06,
+      "loss": 1.0105,
+      "step": 22890
+    },
+    {
+      "epoch": 0.16576545274236865,
+      "grad_norm": 0.17585240304470062,
+      "learning_rate": 4.834241785923691e-06,
+      "loss": 1.0113,
+      "step": 22900
+    },
+    {
+      "epoch": 0.16583783940295482,
+      "grad_norm": 0.1638375222682953,
+      "learning_rate": 4.834169399263104e-06,
+      "loss": 1.0205,
+      "step": 22910
+    },
+    {
+      "epoch": 0.165910226063541,
+      "grad_norm": 0.24316856265068054,
+      "learning_rate": 4.834097012602518e-06,
+      "loss": 1.0187,
+      "step": 22920
+    },
+    {
+      "epoch": 0.1659826127241272,
+      "grad_norm": 0.18205313384532928,
+      "learning_rate": 4.834024625941932e-06,
+      "loss": 1.0046,
+      "step": 22930
+    },
+    {
+      "epoch": 0.16605499938471338,
+      "grad_norm": 0.16292521357536316,
+      "learning_rate": 4.833952239281345e-06,
+      "loss": 1.0175,
+      "step": 22940
+    },
+    {
+      "epoch": 0.16612738604529959,
+      "grad_norm": 0.1718023717403412,
+      "learning_rate": 4.833879852620759e-06,
+      "loss": 1.0144,
+      "step": 22950
+    },
+    {
+      "epoch": 0.16619977270588576,
+      "grad_norm": 0.17306740581989288,
+      "learning_rate": 4.833807465960173e-06,
+      "loss": 1.0277,
+      "step": 22960
+    },
+    {
+      "epoch": 0.16627215936647194,
+      "grad_norm": 0.2184477001428604,
+      "learning_rate": 4.833735079299587e-06,
+      "loss": 1.0086,
+      "step": 22970
+    },
+    {
+      "epoch": 0.16634454602705814,
+      "grad_norm": 0.18039968609809875,
+      "learning_rate": 4.8336626926390006e-06,
+      "loss": 1.0144,
+      "step": 22980
+    },
+    {
+      "epoch": 0.16641693268764432,
+      "grad_norm": 0.1612757295370102,
+      "learning_rate": 4.833590305978414e-06,
+      "loss": 1.0096,
+      "step": 22990
+    },
+    {
+      "epoch": 0.1664893193482305,
+      "grad_norm": 0.16724608838558197,
+      "learning_rate": 4.833517919317829e-06,
+      "loss": 1.0224,
+      "step": 23000
+    },
+    {
+      "epoch": 0.1665617060088167,
+      "grad_norm": 0.19500578939914703,
+      "learning_rate": 4.833445532657242e-06,
+      "loss": 1.0279,
+      "step": 23010
+    },
+    {
+      "epoch": 0.16663409266940288,
+      "grad_norm": 0.1689131259918213,
+      "learning_rate": 4.833373145996656e-06,
+      "loss": 1.0225,
+      "step": 23020
+    },
+    {
+      "epoch": 0.16670647932998908,
+      "grad_norm": 0.18741093575954437,
+      "learning_rate": 4.8333007593360695e-06,
+      "loss": 1.0067,
+      "step": 23030
+    },
+    {
+      "epoch": 0.16677886599057526,
+      "grad_norm": 0.16117849946022034,
+      "learning_rate": 4.833228372675484e-06,
+      "loss": 1.0135,
+      "step": 23040
+    },
+    {
+      "epoch": 0.16685125265116144,
+      "grad_norm": 0.17075563967227936,
+      "learning_rate": 4.8331559860148976e-06,
+      "loss": 1.0091,
+      "step": 23050
+    },
+    {
+      "epoch": 0.16692363931174764,
+      "grad_norm": 0.16465437412261963,
+      "learning_rate": 4.833083599354311e-06,
+      "loss": 1.013,
+      "step": 23060
+    },
+    {
+      "epoch": 0.16699602597233382,
+      "grad_norm": 0.17264023423194885,
+      "learning_rate": 4.833011212693725e-06,
+      "loss": 1.0298,
+      "step": 23070
+    },
+    {
+      "epoch": 0.16706841263292,
+      "grad_norm": 0.27394869923591614,
+      "learning_rate": 4.832938826033139e-06,
+      "loss": 1.0207,
+      "step": 23080
+    },
+    {
+      "epoch": 0.1671407992935062,
+      "grad_norm": 0.16792115569114685,
+      "learning_rate": 4.832866439372553e-06,
+      "loss": 1.0246,
+      "step": 23090
+    },
+    {
+      "epoch": 0.16721318595409237,
+      "grad_norm": 0.17486481368541718,
+      "learning_rate": 4.8327940527119665e-06,
+      "loss": 0.9999,
+      "step": 23100
+    },
+    {
+      "epoch": 0.16728557261467858,
+      "grad_norm": 0.18093885481357574,
+      "learning_rate": 4.83272166605138e-06,
+      "loss": 1.0158,
+      "step": 23110
+    },
+    {
+      "epoch": 0.16735795927526476,
+      "grad_norm": 0.16081741452217102,
+      "learning_rate": 4.8326492793907946e-06,
+      "loss": 1.0061,
+      "step": 23120
+    },
+    {
+      "epoch": 0.16743034593585093,
+      "grad_norm": 0.1673981100320816,
+      "learning_rate": 4.832576892730208e-06,
+      "loss": 1.0152,
+      "step": 23130
+    },
+    {
+      "epoch": 0.16750273259643714,
+      "grad_norm": 0.1669563502073288,
+      "learning_rate": 4.832504506069622e-06,
+      "loss": 1.0121,
+      "step": 23140
+    },
+    {
+      "epoch": 0.16757511925702331,
+      "grad_norm": 0.19718532264232635,
+      "learning_rate": 4.8324321194090354e-06,
+      "loss": 1.0223,
+      "step": 23150
+    },
+    {
+      "epoch": 0.1676475059176095,
+      "grad_norm": 0.2987012565135956,
+      "learning_rate": 4.83235973274845e-06,
+      "loss": 1.0097,
+      "step": 23160
+    },
+    {
+      "epoch": 0.1677198925781957,
+      "grad_norm": 0.16320352256298065,
+      "learning_rate": 4.8322873460878635e-06,
+      "loss": 1.006,
+      "step": 23170
+    },
+    {
+      "epoch": 0.16779227923878187,
+      "grad_norm": 0.16298076510429382,
+      "learning_rate": 4.832214959427277e-06,
+      "loss": 1.0035,
+      "step": 23180
+    },
+    {
+      "epoch": 0.16786466589936808,
+      "grad_norm": 0.1666620373725891,
+      "learning_rate": 4.832142572766691e-06,
+      "loss": 1.0136,
+      "step": 23190
+    },
+    {
+      "epoch": 0.16793705255995425,
+      "grad_norm": 0.16718272864818573,
+      "learning_rate": 4.832070186106105e-06,
+      "loss": 1.0126,
+      "step": 23200
+    },
+    {
+      "epoch": 0.16800943922054043,
+      "grad_norm": 0.2104543298482895,
+      "learning_rate": 4.831997799445519e-06,
+      "loss": 1.0133,
+      "step": 23210
+    },
+    {
+      "epoch": 0.16808182588112663,
+      "grad_norm": 0.17877444624900818,
+      "learning_rate": 4.8319254127849324e-06,
+      "loss": 1.0188,
+      "step": 23220
+    },
+    {
+      "epoch": 0.1681542125417128,
+      "grad_norm": 0.16311542689800262,
+      "learning_rate": 4.831853026124346e-06,
+      "loss": 1.0098,
+      "step": 23230
+    },
+    {
+      "epoch": 0.168226599202299,
+      "grad_norm": 0.1771518886089325,
+      "learning_rate": 4.83178063946376e-06,
+      "loss": 1.0004,
+      "step": 23240
+    },
+    {
+      "epoch": 0.1682989858628852,
+      "grad_norm": 0.20852121710777283,
+      "learning_rate": 4.831708252803174e-06,
+      "loss": 1.014,
+      "step": 23250
+    },
+    {
+      "epoch": 0.16837137252347137,
+      "grad_norm": 0.17537963390350342,
+      "learning_rate": 4.831635866142588e-06,
+      "loss": 1.0188,
+      "step": 23260
+    },
+    {
+      "epoch": 0.16844375918405757,
+      "grad_norm": 0.16795283555984497,
+      "learning_rate": 4.831563479482001e-06,
+      "loss": 1.0063,
+      "step": 23270
+    },
+    {
+      "epoch": 0.16851614584464375,
+      "grad_norm": 0.18271319568157196,
+      "learning_rate": 4.831491092821415e-06,
+      "loss": 1.0066,
+      "step": 23280
+    },
+    {
+      "epoch": 0.16858853250522993,
+      "grad_norm": 0.16966833174228668,
+      "learning_rate": 4.8314187061608294e-06,
+      "loss": 1.0109,
+      "step": 23290
+    },
+    {
+      "epoch": 0.16866091916581613,
+      "grad_norm": 0.1862793117761612,
+      "learning_rate": 4.831346319500243e-06,
+      "loss": 1.0195,
+      "step": 23300
+    },
+    {
+      "epoch": 0.1687333058264023,
+      "grad_norm": 0.17365515232086182,
+      "learning_rate": 4.831273932839657e-06,
+      "loss": 1.04,
+      "step": 23310
+    },
+    {
+      "epoch": 0.16880569248698848,
+      "grad_norm": 0.17782513797283173,
+      "learning_rate": 4.83120154617907e-06,
+      "loss": 1.0078,
+      "step": 23320
+    },
+    {
+      "epoch": 0.1688780791475747,
+      "grad_norm": 0.16336967051029205,
+      "learning_rate": 4.831129159518485e-06,
+      "loss": 1.0081,
+      "step": 23330
+    },
+    {
+      "epoch": 0.16895046580816087,
+      "grad_norm": 0.18843859434127808,
+      "learning_rate": 4.831056772857898e-06,
+      "loss": 1.0103,
+      "step": 23340
+    },
+    {
+      "epoch": 0.16902285246874707,
+      "grad_norm": 0.1720830351114273,
+      "learning_rate": 4.830984386197312e-06,
+      "loss": 1.0168,
+      "step": 23350
+    },
+    {
+      "epoch": 0.16909523912933325,
+      "grad_norm": 0.24049195647239685,
+      "learning_rate": 4.830911999536726e-06,
+      "loss": 1.0181,
+      "step": 23360
+    },
+    {
+      "epoch": 0.16916762578991942,
+      "grad_norm": 0.17097869515419006,
+      "learning_rate": 4.83083961287614e-06,
+      "loss": 1.0042,
+      "step": 23370
+    },
+    {
+      "epoch": 0.16924001245050563,
+      "grad_norm": 0.2235918790102005,
+      "learning_rate": 4.830767226215554e-06,
+      "loss": 1.0094,
+      "step": 23380
+    },
+    {
+      "epoch": 0.1693123991110918,
+      "grad_norm": 0.16128119826316833,
+      "learning_rate": 4.830694839554967e-06,
+      "loss": 1.0185,
+      "step": 23390
+    },
+    {
+      "epoch": 0.16938478577167798,
+      "grad_norm": 0.16598336398601532,
+      "learning_rate": 4.830622452894381e-06,
+      "loss": 1.0118,
+      "step": 23400
+    },
+    {
+      "epoch": 0.16945717243226419,
+      "grad_norm": 0.16268615424633026,
+      "learning_rate": 4.830550066233795e-06,
+      "loss": 1.0171,
+      "step": 23410
+    },
+    {
+      "epoch": 0.16952955909285036,
+      "grad_norm": 0.1621619611978531,
+      "learning_rate": 4.830477679573209e-06,
+      "loss": 1.0176,
+      "step": 23420
+    },
+    {
+      "epoch": 0.16960194575343657,
+      "grad_norm": 0.1687730848789215,
+      "learning_rate": 4.830405292912623e-06,
+      "loss": 1.0058,
+      "step": 23430
+    },
+    {
+      "epoch": 0.16967433241402274,
+      "grad_norm": 0.17967501282691956,
+      "learning_rate": 4.830332906252036e-06,
+      "loss": 1.0141,
+      "step": 23440
+    },
+    {
+      "epoch": 0.16974671907460892,
+      "grad_norm": 0.23088906705379486,
+      "learning_rate": 4.83026051959145e-06,
+      "loss": 1.0283,
+      "step": 23450
+    },
+    {
+      "epoch": 0.16981910573519512,
+      "grad_norm": 0.16651782393455505,
+      "learning_rate": 4.8301881329308635e-06,
+      "loss": 1.0155,
+      "step": 23460
+    },
+    {
+      "epoch": 0.1698914923957813,
+      "grad_norm": 0.16330918669700623,
+      "learning_rate": 4.830115746270277e-06,
+      "loss": 1.0126,
+      "step": 23470
+    },
+    {
+      "epoch": 0.1699638790563675,
+      "grad_norm": 0.1758406162261963,
+      "learning_rate": 4.8300433596096915e-06,
+      "loss": 1.007,
+      "step": 23480
+    },
+    {
+      "epoch": 0.17003626571695368,
+      "grad_norm": 0.18241257965564728,
+      "learning_rate": 4.829970972949105e-06,
+      "loss": 1.0032,
+      "step": 23490
+    },
+    {
+      "epoch": 0.17010865237753986,
+      "grad_norm": 0.18768127262592316,
+      "learning_rate": 4.829898586288519e-06,
+      "loss": 1.0193,
+      "step": 23500
+    },
+    {
+      "epoch": 0.17018103903812606,
+      "grad_norm": 0.1854964941740036,
+      "learning_rate": 4.829826199627932e-06,
+      "loss": 1.0254,
+      "step": 23510
+    },
+    {
+      "epoch": 0.17025342569871224,
+      "grad_norm": 0.164667010307312,
+      "learning_rate": 4.829753812967347e-06,
+      "loss": 1.0093,
+      "step": 23520
+    },
+    {
+      "epoch": 0.17032581235929842,
+      "grad_norm": 0.1953144520521164,
+      "learning_rate": 4.8296814263067605e-06,
+      "loss": 1.0138,
+      "step": 23530
+    },
+    {
+      "epoch": 0.17039819901988462,
+      "grad_norm": 0.170278862118721,
+      "learning_rate": 4.829609039646174e-06,
+      "loss": 1.0114,
+      "step": 23540
+    },
+    {
+      "epoch": 0.1704705856804708,
+      "grad_norm": 0.19750620424747467,
+      "learning_rate": 4.829536652985588e-06,
+      "loss": 1.0215,
+      "step": 23550
+    },
+    {
+      "epoch": 0.170542972341057,
+      "grad_norm": 0.19091200828552246,
+      "learning_rate": 4.829464266325002e-06,
+      "loss": 1.0166,
+      "step": 23560
+    },
+    {
+      "epoch": 0.17061535900164318,
+      "grad_norm": 0.1851140260696411,
+      "learning_rate": 4.829391879664416e-06,
+      "loss": 1.0062,
+      "step": 23570
+    },
+    {
+      "epoch": 0.17068774566222936,
+      "grad_norm": 0.19524310529232025,
+      "learning_rate": 4.829319493003829e-06,
+      "loss": 1.0217,
+      "step": 23580
+    },
+    {
+      "epoch": 0.17076013232281556,
+      "grad_norm": 0.17677630484104156,
+      "learning_rate": 4.829247106343243e-06,
+      "loss": 1.0066,
+      "step": 23590
+    },
+    {
+      "epoch": 0.17083251898340174,
+      "grad_norm": 0.1704902946949005,
+      "learning_rate": 4.8291747196826575e-06,
+      "loss": 1.0033,
+      "step": 23600
+    },
+    {
+      "epoch": 0.17090490564398791,
+      "grad_norm": 0.16769947111606598,
+      "learning_rate": 4.829102333022071e-06,
+      "loss": 0.9979,
+      "step": 23610
+    },
+    {
+      "epoch": 0.17097729230457412,
+      "grad_norm": 0.16838715970516205,
+      "learning_rate": 4.829029946361485e-06,
+      "loss": 1.0022,
+      "step": 23620
+    },
+    {
+      "epoch": 0.1710496789651603,
+      "grad_norm": 0.18555735051631927,
+      "learning_rate": 4.828957559700898e-06,
+      "loss": 1.0036,
+      "step": 23630
+    },
+    {
+      "epoch": 0.1711220656257465,
+      "grad_norm": 0.17811746895313263,
+      "learning_rate": 4.828885173040313e-06,
+      "loss": 1.0122,
+      "step": 23640
+    },
+    {
+      "epoch": 0.17119445228633268,
+      "grad_norm": 0.17292849719524384,
+      "learning_rate": 4.828812786379726e-06,
+      "loss": 0.9983,
+      "step": 23650
+    },
+    {
+      "epoch": 0.17126683894691885,
+      "grad_norm": 0.18011000752449036,
+      "learning_rate": 4.82874039971914e-06,
+      "loss": 1.0149,
+      "step": 23660
+    },
+    {
+      "epoch": 0.17133922560750506,
+      "grad_norm": 0.1814233809709549,
+      "learning_rate": 4.828668013058554e-06,
+      "loss": 1.0192,
+      "step": 23670
+    },
+    {
+      "epoch": 0.17141161226809123,
+      "grad_norm": 0.17261269688606262,
+      "learning_rate": 4.828595626397968e-06,
+      "loss": 1.0306,
+      "step": 23680
+    },
+    {
+      "epoch": 0.1714839989286774,
+      "grad_norm": 0.18298877775669098,
+      "learning_rate": 4.828523239737382e-06,
+      "loss": 1.0157,
+      "step": 23690
+    },
+    {
+      "epoch": 0.17155638558926362,
+      "grad_norm": 0.16827493906021118,
+      "learning_rate": 4.828450853076795e-06,
+      "loss": 1.0056,
+      "step": 23700
+    },
+    {
+      "epoch": 0.1716287722498498,
+      "grad_norm": 0.21955229341983795,
+      "learning_rate": 4.828378466416209e-06,
+      "loss": 1.0198,
+      "step": 23710
+    },
+    {
+      "epoch": 0.171701158910436,
+      "grad_norm": 0.17503350973129272,
+      "learning_rate": 4.828306079755623e-06,
+      "loss": 1.0191,
+      "step": 23720
+    },
+    {
+      "epoch": 0.17177354557102217,
+      "grad_norm": 0.1629331111907959,
+      "learning_rate": 4.828233693095037e-06,
+      "loss": 1.0135,
+      "step": 23730
+    },
+    {
+      "epoch": 0.17184593223160835,
+      "grad_norm": 0.1784103661775589,
+      "learning_rate": 4.828161306434451e-06,
+      "loss": 1.0184,
+      "step": 23740
+    },
+    {
+      "epoch": 0.17191831889219455,
+      "grad_norm": 0.17589734494686127,
+      "learning_rate": 4.828088919773864e-06,
+      "loss": 1.0039,
+      "step": 23750
+    },
+    {
+      "epoch": 0.17199070555278073,
+      "grad_norm": 0.17863091826438904,
+      "learning_rate": 4.828016533113279e-06,
+      "loss": 0.9991,
+      "step": 23760
+    },
+    {
+      "epoch": 0.1720630922133669,
+      "grad_norm": 0.16851244866847992,
+      "learning_rate": 4.827944146452692e-06,
+      "loss": 1.0054,
+      "step": 23770
+    },
+    {
+      "epoch": 0.1721354788739531,
+      "grad_norm": 0.1785619854927063,
+      "learning_rate": 4.827871759792106e-06,
+      "loss": 1.0108,
+      "step": 23780
+    },
+    {
+      "epoch": 0.1722078655345393,
+      "grad_norm": 0.18070313334465027,
+      "learning_rate": 4.8277993731315196e-06,
+      "loss": 1.0221,
+      "step": 23790
+    },
+    {
+      "epoch": 0.1722802521951255,
+      "grad_norm": 0.16967083513736725,
+      "learning_rate": 4.827726986470934e-06,
+      "loss": 1.0178,
+      "step": 23800
+    },
+    {
+      "epoch": 0.17235263885571167,
+      "grad_norm": 0.19012100994586945,
+      "learning_rate": 4.827654599810348e-06,
+      "loss": 1.0037,
+      "step": 23810
+    },
+    {
+      "epoch": 0.17242502551629785,
+      "grad_norm": 0.18341712653636932,
+      "learning_rate": 4.827582213149761e-06,
+      "loss": 1.0065,
+      "step": 23820
+    },
+    {
+      "epoch": 0.17249741217688405,
+      "grad_norm": 0.1624983698129654,
+      "learning_rate": 4.827509826489175e-06,
+      "loss": 1.0026,
+      "step": 23830
+    },
+    {
+      "epoch": 0.17256979883747023,
+      "grad_norm": 0.1717177778482437,
+      "learning_rate": 4.827437439828589e-06,
+      "loss": 1.0206,
+      "step": 23840
+    },
+    {
+      "epoch": 0.1726421854980564,
+      "grad_norm": 0.18346939980983734,
+      "learning_rate": 4.827365053168003e-06,
+      "loss": 1.0212,
+      "step": 23850
+    },
+    {
+      "epoch": 0.1727145721586426,
+      "grad_norm": 0.2184140831232071,
+      "learning_rate": 4.8272926665074166e-06,
+      "loss": 1.0239,
+      "step": 23860
+    },
+    {
+      "epoch": 0.1727869588192288,
+      "grad_norm": 0.15121838450431824,
+      "learning_rate": 4.82722027984683e-06,
+      "loss": 0.9996,
+      "step": 23870
+    },
+    {
+      "epoch": 0.172859345479815,
+      "grad_norm": 0.1719536930322647,
+      "learning_rate": 4.827147893186244e-06,
+      "loss": 1.0067,
+      "step": 23880
+    },
+    {
+      "epoch": 0.17293173214040117,
+      "grad_norm": 0.18498043715953827,
+      "learning_rate": 4.827075506525658e-06,
+      "loss": 1.011,
+      "step": 23890
+    },
+    {
+      "epoch": 0.17300411880098734,
+      "grad_norm": 0.16693322360515594,
+      "learning_rate": 4.827003119865072e-06,
+      "loss": 1.001,
+      "step": 23900
+    },
+    {
+      "epoch": 0.17307650546157355,
+      "grad_norm": 0.19714364409446716,
+      "learning_rate": 4.8269307332044855e-06,
+      "loss": 1.0202,
+      "step": 23910
+    },
+    {
+      "epoch": 0.17314889212215973,
+      "grad_norm": 0.1966882348060608,
+      "learning_rate": 4.826858346543899e-06,
+      "loss": 1.0124,
+      "step": 23920
+    },
+    {
+      "epoch": 0.1732212787827459,
+      "grad_norm": 0.1639029085636139,
+      "learning_rate": 4.826785959883314e-06,
+      "loss": 1.012,
+      "step": 23930
+    },
+    {
+      "epoch": 0.1732936654433321,
+      "grad_norm": 0.18476232886314392,
+      "learning_rate": 4.826713573222727e-06,
+      "loss": 1.0252,
+      "step": 23940
+    },
+    {
+      "epoch": 0.17336605210391828,
+      "grad_norm": 0.1863066405057907,
+      "learning_rate": 4.826641186562141e-06,
+      "loss": 1.0282,
+      "step": 23950
+    },
+    {
+      "epoch": 0.1734384387645045,
+      "grad_norm": 0.1653691977262497,
+      "learning_rate": 4.8265687999015544e-06,
+      "loss": 1.0146,
+      "step": 23960
+    },
+    {
+      "epoch": 0.17351082542509066,
+      "grad_norm": 0.18651318550109863,
+      "learning_rate": 4.826496413240969e-06,
+      "loss": 1.0117,
+      "step": 23970
+    },
+    {
+      "epoch": 0.17358321208567684,
+      "grad_norm": 0.16590899229049683,
+      "learning_rate": 4.826424026580382e-06,
+      "loss": 1.0129,
+      "step": 23980
+    },
+    {
+      "epoch": 0.17365559874626305,
+      "grad_norm": 0.15830731391906738,
+      "learning_rate": 4.826351639919796e-06,
+      "loss": 1.0247,
+      "step": 23990
+    },
+    {
+      "epoch": 0.17372798540684922,
+      "grad_norm": 0.20496408641338348,
+      "learning_rate": 4.82627925325921e-06,
+      "loss": 1.0094,
+      "step": 24000
+    },
+    {
+      "epoch": 0.1738003720674354,
+      "grad_norm": 0.17806655168533325,
+      "learning_rate": 4.826206866598623e-06,
+      "loss": 1.0311,
+      "step": 24010
+    },
+    {
+      "epoch": 0.1738727587280216,
+      "grad_norm": 0.18243646621704102,
+      "learning_rate": 4.826134479938037e-06,
+      "loss": 1.001,
+      "step": 24020
+    },
+    {
+      "epoch": 0.17394514538860778,
+      "grad_norm": 0.16999110579490662,
+      "learning_rate": 4.826062093277451e-06,
+      "loss": 1.0067,
+      "step": 24030
+    },
+    {
+      "epoch": 0.17401753204919398,
+      "grad_norm": 0.16639530658721924,
+      "learning_rate": 4.825989706616865e-06,
+      "loss": 0.9953,
+      "step": 24040
+    },
+    {
+      "epoch": 0.17408991870978016,
+      "grad_norm": 0.20730364322662354,
+      "learning_rate": 4.825917319956279e-06,
+      "loss": 1.0058,
+      "step": 24050
+    },
+    {
+      "epoch": 0.17416230537036634,
+      "grad_norm": 0.17685486376285553,
+      "learning_rate": 4.825844933295692e-06,
+      "loss": 1.005,
+      "step": 24060
+    },
+    {
+      "epoch": 0.17423469203095254,
+      "grad_norm": 0.16955289244651794,
+      "learning_rate": 4.825772546635106e-06,
+      "loss": 1.0101,
+      "step": 24070
+    },
+    {
+      "epoch": 0.17430707869153872,
+      "grad_norm": 0.1870541274547577,
+      "learning_rate": 4.82570015997452e-06,
+      "loss": 1.0102,
+      "step": 24080
+    },
+    {
+      "epoch": 0.17437946535212492,
+      "grad_norm": 0.22333884239196777,
+      "learning_rate": 4.825627773313934e-06,
+      "loss": 1.0087,
+      "step": 24090
+    },
+    {
+      "epoch": 0.1744518520127111,
+      "grad_norm": 0.18519848585128784,
+      "learning_rate": 4.825555386653348e-06,
+      "loss": 1.005,
+      "step": 24100
+    },
+    {
+      "epoch": 0.17452423867329728,
+      "grad_norm": 0.18657277524471283,
+      "learning_rate": 4.825482999992761e-06,
+      "loss": 1.0142,
+      "step": 24110
+    },
+    {
+      "epoch": 0.17459662533388348,
+      "grad_norm": 0.20050521194934845,
+      "learning_rate": 4.825410613332176e-06,
+      "loss": 1.0069,
+      "step": 24120
+    },
+    {
+      "epoch": 0.17466901199446966,
+      "grad_norm": 0.18028657138347626,
+      "learning_rate": 4.825338226671589e-06,
+      "loss": 1.0045,
+      "step": 24130
+    },
+    {
+      "epoch": 0.17474139865505584,
+      "grad_norm": 0.17001697421073914,
+      "learning_rate": 4.825265840011003e-06,
+      "loss": 1.011,
+      "step": 24140
+    },
+    {
+      "epoch": 0.17481378531564204,
+      "grad_norm": 0.16994266211986542,
+      "learning_rate": 4.8251934533504165e-06,
+      "loss": 1.0136,
+      "step": 24150
+    },
+    {
+      "epoch": 0.17488617197622822,
+      "grad_norm": 0.17373910546302795,
+      "learning_rate": 4.825121066689831e-06,
+      "loss": 1.0038,
+      "step": 24160
+    },
+    {
+      "epoch": 0.17495855863681442,
+      "grad_norm": 0.17860816419124603,
+      "learning_rate": 4.825048680029245e-06,
+      "loss": 1.0161,
+      "step": 24170
+    },
+    {
+      "epoch": 0.1750309452974006,
+      "grad_norm": 0.16933861374855042,
+      "learning_rate": 4.824976293368658e-06,
+      "loss": 1.0084,
+      "step": 24180
+    },
+    {
+      "epoch": 0.17510333195798677,
+      "grad_norm": 0.18152697384357452,
+      "learning_rate": 4.824903906708072e-06,
+      "loss": 0.9913,
+      "step": 24190
+    },
+    {
+      "epoch": 0.17517571861857298,
+      "grad_norm": 0.15832629799842834,
+      "learning_rate": 4.824831520047486e-06,
+      "loss": 1.0034,
+      "step": 24200
+    },
+    {
+      "epoch": 0.17524810527915916,
+      "grad_norm": 0.1804628223180771,
+      "learning_rate": 4.8247591333869e-06,
+      "loss": 1.0056,
+      "step": 24210
+    },
+    {
+      "epoch": 0.17532049193974533,
+      "grad_norm": 0.1735430359840393,
+      "learning_rate": 4.8246867467263135e-06,
+      "loss": 1.0073,
+      "step": 24220
+    },
+    {
+      "epoch": 0.17539287860033154,
+      "grad_norm": 0.176886186003685,
+      "learning_rate": 4.824614360065727e-06,
+      "loss": 1.0106,
+      "step": 24230
+    },
+    {
+      "epoch": 0.1754652652609177,
+      "grad_norm": 0.16210952401161194,
+      "learning_rate": 4.824541973405142e-06,
+      "loss": 1.0199,
+      "step": 24240
+    },
+    {
+      "epoch": 0.17553765192150392,
+      "grad_norm": 0.16794593632221222,
+      "learning_rate": 4.824469586744555e-06,
+      "loss": 0.9955,
+      "step": 24250
+    },
+    {
+      "epoch": 0.1756100385820901,
+      "grad_norm": 0.18100325763225555,
+      "learning_rate": 4.824397200083969e-06,
+      "loss": 1.0069,
+      "step": 24260
+    },
+    {
+      "epoch": 0.17568242524267627,
+      "grad_norm": 0.16271761059761047,
+      "learning_rate": 4.8243248134233825e-06,
+      "loss": 1.0122,
+      "step": 24270
+    },
+    {
+      "epoch": 0.17575481190326248,
+      "grad_norm": 0.16619524359703064,
+      "learning_rate": 4.824252426762797e-06,
+      "loss": 1.0177,
+      "step": 24280
+    },
+    {
+      "epoch": 0.17582719856384865,
+      "grad_norm": 0.18534304201602936,
+      "learning_rate": 4.8241800401022105e-06,
+      "loss": 1.0133,
+      "step": 24290
+    },
+    {
+      "epoch": 0.17589958522443483,
+      "grad_norm": 0.179828941822052,
+      "learning_rate": 4.824107653441624e-06,
+      "loss": 1.0115,
+      "step": 24300
+    },
+    {
+      "epoch": 0.17597197188502103,
+      "grad_norm": 0.19218121469020844,
+      "learning_rate": 4.824035266781038e-06,
+      "loss": 1.0002,
+      "step": 24310
+    },
+    {
+      "epoch": 0.1760443585456072,
+      "grad_norm": 0.1730458289384842,
+      "learning_rate": 4.823962880120452e-06,
+      "loss": 1.0093,
+      "step": 24320
+    },
+    {
+      "epoch": 0.17611674520619341,
+      "grad_norm": 0.1910993903875351,
+      "learning_rate": 4.823890493459866e-06,
+      "loss": 1.0133,
+      "step": 24330
+    },
+    {
+      "epoch": 0.1761891318667796,
+      "grad_norm": 0.1970120370388031,
+      "learning_rate": 4.8238181067992795e-06,
+      "loss": 1.0102,
+      "step": 24340
+    },
+    {
+      "epoch": 0.17626151852736577,
+      "grad_norm": 0.17400617897510529,
+      "learning_rate": 4.823745720138693e-06,
+      "loss": 1.0102,
+      "step": 24350
+    },
+    {
+      "epoch": 0.17633390518795197,
+      "grad_norm": 0.17096106708049774,
+      "learning_rate": 4.8236733334781075e-06,
+      "loss": 1.0161,
+      "step": 24360
+    },
+    {
+      "epoch": 0.17640629184853815,
+      "grad_norm": 0.185289666056633,
+      "learning_rate": 4.823600946817521e-06,
+      "loss": 1.0013,
+      "step": 24370
+    },
+    {
+      "epoch": 0.17647867850912433,
+      "grad_norm": 0.17100411653518677,
+      "learning_rate": 4.823528560156935e-06,
+      "loss": 0.994,
+      "step": 24380
+    },
+    {
+      "epoch": 0.17655106516971053,
+      "grad_norm": 0.1876966655254364,
+      "learning_rate": 4.823456173496348e-06,
+      "loss": 1.0041,
+      "step": 24390
+    },
+    {
+      "epoch": 0.1766234518302967,
+      "grad_norm": 0.1617269068956375,
+      "learning_rate": 4.823383786835763e-06,
+      "loss": 0.997,
+      "step": 24400
+    },
+    {
+      "epoch": 0.1766958384908829,
+      "grad_norm": 0.17999807000160217,
+      "learning_rate": 4.8233114001751765e-06,
+      "loss": 1.0085,
+      "step": 24410
+    },
+    {
+      "epoch": 0.1767682251514691,
+      "grad_norm": 0.16914209723472595,
+      "learning_rate": 4.82323901351459e-06,
+      "loss": 1.0218,
+      "step": 24420
+    },
+    {
+      "epoch": 0.17684061181205527,
+      "grad_norm": 0.20446030795574188,
+      "learning_rate": 4.823166626854004e-06,
+      "loss": 1.0115,
+      "step": 24430
+    },
+    {
+      "epoch": 0.17691299847264147,
+      "grad_norm": 0.16619598865509033,
+      "learning_rate": 4.823094240193418e-06,
+      "loss": 1.0197,
+      "step": 24440
+    },
+    {
+      "epoch": 0.17698538513322765,
+      "grad_norm": 0.173335999250412,
+      "learning_rate": 4.823021853532832e-06,
+      "loss": 1.0128,
+      "step": 24450
+    },
+    {
+      "epoch": 0.17705777179381382,
+      "grad_norm": 0.17360666394233704,
+      "learning_rate": 4.822949466872245e-06,
+      "loss": 1.0073,
+      "step": 24460
+    },
+    {
+      "epoch": 0.17713015845440003,
+      "grad_norm": 0.16045695543289185,
+      "learning_rate": 4.822877080211659e-06,
+      "loss": 0.999,
+      "step": 24470
+    },
+    {
+      "epoch": 0.1772025451149862,
+      "grad_norm": 0.17634133994579315,
+      "learning_rate": 4.822804693551073e-06,
+      "loss": 1.0113,
+      "step": 24480
+    },
+    {
+      "epoch": 0.1772749317755724,
+      "grad_norm": 0.15489248931407928,
+      "learning_rate": 4.822732306890487e-06,
+      "loss": 1.0033,
+      "step": 24490
+    },
+    {
+      "epoch": 0.17734731843615859,
+      "grad_norm": 0.17164482176303864,
+      "learning_rate": 4.822659920229901e-06,
+      "loss": 1.0036,
+      "step": 24500
+    },
+    {
+      "epoch": 0.17741970509674476,
+      "grad_norm": 0.17662885785102844,
+      "learning_rate": 4.822587533569314e-06,
+      "loss": 1.0162,
+      "step": 24510
+    },
+    {
+      "epoch": 0.17749209175733097,
+      "grad_norm": 0.1665455847978592,
+      "learning_rate": 4.822515146908728e-06,
+      "loss": 1.0061,
+      "step": 24520
+    },
+    {
+      "epoch": 0.17756447841791714,
+      "grad_norm": 0.17285911738872528,
+      "learning_rate": 4.8224427602481416e-06,
+      "loss": 1.0259,
+      "step": 24530
+    },
+    {
+      "epoch": 0.17763686507850332,
+      "grad_norm": 0.1706143617630005,
+      "learning_rate": 4.822370373587555e-06,
+      "loss": 1.0258,
+      "step": 24540
+    },
+    {
+      "epoch": 0.17770925173908952,
+      "grad_norm": 0.17605163156986237,
+      "learning_rate": 4.82229798692697e-06,
+      "loss": 1.0184,
+      "step": 24550
+    },
+    {
+      "epoch": 0.1777816383996757,
+      "grad_norm": 0.1811869591474533,
+      "learning_rate": 4.822225600266383e-06,
+      "loss": 1.018,
+      "step": 24560
+    },
+    {
+      "epoch": 0.1778540250602619,
+      "grad_norm": 0.16938798129558563,
+      "learning_rate": 4.822153213605797e-06,
+      "loss": 1.0101,
+      "step": 24570
+    },
+    {
+      "epoch": 0.17792641172084808,
+      "grad_norm": 0.17794403433799744,
+      "learning_rate": 4.8220808269452105e-06,
+      "loss": 1.0068,
+      "step": 24580
+    },
+    {
+      "epoch": 0.17799879838143426,
+      "grad_norm": 0.1750117838382721,
+      "learning_rate": 4.822008440284625e-06,
+      "loss": 1.0183,
+      "step": 24590
+    },
+    {
+      "epoch": 0.17807118504202046,
+      "grad_norm": 0.19311663508415222,
+      "learning_rate": 4.8219360536240386e-06,
+      "loss": 1.0059,
+      "step": 24600
+    },
+    {
+      "epoch": 0.17814357170260664,
+      "grad_norm": 0.17027834057807922,
+      "learning_rate": 4.821863666963452e-06,
+      "loss": 1.0117,
+      "step": 24610
+    },
+    {
+      "epoch": 0.17821595836319284,
+      "grad_norm": 0.16142655909061432,
+      "learning_rate": 4.821791280302866e-06,
+      "loss": 1.0039,
+      "step": 24620
+    },
+    {
+      "epoch": 0.17828834502377902,
+      "grad_norm": 0.20361775159835815,
+      "learning_rate": 4.82171889364228e-06,
+      "loss": 0.9984,
+      "step": 24630
+    },
+    {
+      "epoch": 0.1783607316843652,
+      "grad_norm": 0.1685153841972351,
+      "learning_rate": 4.821646506981694e-06,
+      "loss": 1.0199,
+      "step": 24640
+    },
+    {
+      "epoch": 0.1784331183449514,
+      "grad_norm": 0.17604348063468933,
+      "learning_rate": 4.8215741203211075e-06,
+      "loss": 1.0111,
+      "step": 24650
+    },
+    {
+      "epoch": 0.17850550500553758,
+      "grad_norm": 0.17381654679775238,
+      "learning_rate": 4.821501733660521e-06,
+      "loss": 1.0142,
+      "step": 24660
+    },
+    {
+      "epoch": 0.17857789166612376,
+      "grad_norm": 0.1625945121049881,
+      "learning_rate": 4.821429346999935e-06,
+      "loss": 1.0036,
+      "step": 24670
+    },
+    {
+      "epoch": 0.17865027832670996,
+      "grad_norm": 0.16681507229804993,
+      "learning_rate": 4.821356960339349e-06,
+      "loss": 1.0174,
+      "step": 24680
+    },
+    {
+      "epoch": 0.17872266498729614,
+      "grad_norm": 0.17053383588790894,
+      "learning_rate": 4.821284573678763e-06,
+      "loss": 1.0073,
+      "step": 24690
+    },
+    {
+      "epoch": 0.17879505164788234,
+      "grad_norm": 0.16991731524467468,
+      "learning_rate": 4.8212121870181764e-06,
+      "loss": 1.0232,
+      "step": 24700
+    },
+    {
+      "epoch": 0.17886743830846852,
+      "grad_norm": 0.18369825184345245,
+      "learning_rate": 4.82113980035759e-06,
+      "loss": 1.0144,
+      "step": 24710
+    },
+    {
+      "epoch": 0.1789398249690547,
+      "grad_norm": 0.18356874585151672,
+      "learning_rate": 4.8210674136970045e-06,
+      "loss": 1.0159,
+      "step": 24720
+    },
+    {
+      "epoch": 0.1790122116296409,
+      "grad_norm": 0.21696501970291138,
+      "learning_rate": 4.820995027036418e-06,
+      "loss": 1.001,
+      "step": 24730
+    },
+    {
+      "epoch": 0.17908459829022708,
+      "grad_norm": 0.1677326261997223,
+      "learning_rate": 4.820922640375832e-06,
+      "loss": 1.02,
+      "step": 24740
+    },
+    {
+      "epoch": 0.17915698495081325,
+      "grad_norm": 0.17450115084648132,
+      "learning_rate": 4.820850253715245e-06,
+      "loss": 1.0215,
+      "step": 24750
+    },
+    {
+      "epoch": 0.17922937161139946,
+      "grad_norm": 0.1796158254146576,
+      "learning_rate": 4.82077786705466e-06,
+      "loss": 1.0213,
+      "step": 24760
+    },
+    {
+      "epoch": 0.17930175827198563,
+      "grad_norm": 0.1706731915473938,
+      "learning_rate": 4.8207054803940734e-06,
+      "loss": 1.0065,
+      "step": 24770
+    },
+    {
+      "epoch": 0.17937414493257184,
+      "grad_norm": 0.16157269477844238,
+      "learning_rate": 4.820633093733487e-06,
+      "loss": 1.0122,
+      "step": 24780
+    },
+    {
+      "epoch": 0.17944653159315802,
+      "grad_norm": 0.17014949023723602,
+      "learning_rate": 4.820560707072901e-06,
+      "loss": 1.0163,
+      "step": 24790
+    },
+    {
+      "epoch": 0.1795189182537442,
+      "grad_norm": 0.17551864683628082,
+      "learning_rate": 4.820488320412315e-06,
+      "loss": 1.001,
+      "step": 24800
+    },
+    {
+      "epoch": 0.1795913049143304,
+      "grad_norm": 0.17576205730438232,
+      "learning_rate": 4.820415933751729e-06,
+      "loss": 1.0058,
+      "step": 24810
+    },
+    {
+      "epoch": 0.17966369157491657,
+      "grad_norm": 0.16241174936294556,
+      "learning_rate": 4.820343547091142e-06,
+      "loss": 1.001,
+      "step": 24820
+    },
+    {
+      "epoch": 0.17973607823550275,
+      "grad_norm": 0.2229170948266983,
+      "learning_rate": 4.820271160430556e-06,
+      "loss": 1.0258,
+      "step": 24830
+    },
+    {
+      "epoch": 0.17980846489608895,
+      "grad_norm": 0.16988316178321838,
+      "learning_rate": 4.8201987737699704e-06,
+      "loss": 1.013,
+      "step": 24840
+    },
+    {
+      "epoch": 0.17988085155667513,
+      "grad_norm": 0.2061333805322647,
+      "learning_rate": 4.820126387109384e-06,
+      "loss": 1.0025,
+      "step": 24850
+    },
+    {
+      "epoch": 0.17995323821726134,
+      "grad_norm": 0.19015344977378845,
+      "learning_rate": 4.820054000448798e-06,
+      "loss": 1.014,
+      "step": 24860
+    },
+    {
+      "epoch": 0.1800256248778475,
+      "grad_norm": 0.17161577939987183,
+      "learning_rate": 4.819981613788211e-06,
+      "loss": 0.9984,
+      "step": 24870
+    },
+    {
+      "epoch": 0.1800980115384337,
+      "grad_norm": 0.17109227180480957,
+      "learning_rate": 4.819909227127626e-06,
+      "loss": 1.0087,
+      "step": 24880
+    },
+    {
+      "epoch": 0.1801703981990199,
+      "grad_norm": 0.1876406967639923,
+      "learning_rate": 4.819836840467039e-06,
+      "loss": 1.0225,
+      "step": 24890
+    },
+    {
+      "epoch": 0.18024278485960607,
+      "grad_norm": 0.17785070836544037,
+      "learning_rate": 4.819764453806453e-06,
+      "loss": 1.0034,
+      "step": 24900
+    },
+    {
+      "epoch": 0.18031517152019225,
+      "grad_norm": 0.17659418284893036,
+      "learning_rate": 4.819692067145867e-06,
+      "loss": 1.0196,
+      "step": 24910
+    },
+    {
+      "epoch": 0.18038755818077845,
+      "grad_norm": 0.19268226623535156,
+      "learning_rate": 4.819619680485281e-06,
+      "loss": 1.0012,
+      "step": 24920
+    },
+    {
+      "epoch": 0.18045994484136463,
+      "grad_norm": 0.1869351863861084,
+      "learning_rate": 4.819547293824695e-06,
+      "loss": 1.0112,
+      "step": 24930
+    },
+    {
+      "epoch": 0.18053233150195083,
+      "grad_norm": 0.17009639739990234,
+      "learning_rate": 4.819474907164108e-06,
+      "loss": 1.0161,
+      "step": 24940
+    },
+    {
+      "epoch": 0.180604718162537,
+      "grad_norm": 0.20268967747688293,
+      "learning_rate": 4.819402520503522e-06,
+      "loss": 1.0077,
+      "step": 24950
+    },
+    {
+      "epoch": 0.18067710482312319,
+      "grad_norm": 0.1628827154636383,
+      "learning_rate": 4.819330133842936e-06,
+      "loss": 1.0127,
+      "step": 24960
+    },
+    {
+      "epoch": 0.1807494914837094,
+      "grad_norm": 0.18746638298034668,
+      "learning_rate": 4.81925774718235e-06,
+      "loss": 1.015,
+      "step": 24970
+    },
+    {
+      "epoch": 0.18082187814429557,
+      "grad_norm": 0.16319945454597473,
+      "learning_rate": 4.819185360521764e-06,
+      "loss": 0.9972,
+      "step": 24980
+    },
+    {
+      "epoch": 0.18089426480488174,
+      "grad_norm": 0.1594797521829605,
+      "learning_rate": 4.819112973861177e-06,
+      "loss": 1.0161,
+      "step": 24990
+    },
+    {
+      "epoch": 0.18096665146546795,
+      "grad_norm": 0.20105616748332977,
+      "learning_rate": 4.819040587200592e-06,
+      "loss": 1.014,
+      "step": 25000
+    },
+    {
+      "epoch": 0.18103903812605412,
+      "grad_norm": 0.1811130940914154,
+      "learning_rate": 4.818968200540005e-06,
+      "loss": 1.0114,
+      "step": 25010
+    },
+    {
+      "epoch": 0.18111142478664033,
+      "grad_norm": 0.17586207389831543,
+      "learning_rate": 4.818895813879419e-06,
+      "loss": 1.0274,
+      "step": 25020
+    },
+    {
+      "epoch": 0.1811838114472265,
+      "grad_norm": 0.19107623398303986,
+      "learning_rate": 4.8188234272188325e-06,
+      "loss": 1.0176,
+      "step": 25030
+    },
+    {
+      "epoch": 0.18125619810781268,
+      "grad_norm": 0.16346633434295654,
+      "learning_rate": 4.818751040558246e-06,
+      "loss": 1.0137,
+      "step": 25040
+    },
+    {
+      "epoch": 0.1813285847683989,
+      "grad_norm": 0.17446771264076233,
+      "learning_rate": 4.81867865389766e-06,
+      "loss": 0.989,
+      "step": 25050
+    },
+    {
+      "epoch": 0.18140097142898506,
+      "grad_norm": 0.1719013899564743,
+      "learning_rate": 4.818606267237073e-06,
+      "loss": 1.0155,
+      "step": 25060
+    },
+    {
+      "epoch": 0.18147335808957124,
+      "grad_norm": 0.1678323745727539,
+      "learning_rate": 4.818533880576488e-06,
+      "loss": 1.0097,
+      "step": 25070
+    },
+    {
+      "epoch": 0.18154574475015745,
+      "grad_norm": 0.1604251116514206,
+      "learning_rate": 4.8184614939159015e-06,
+      "loss": 1.0109,
+      "step": 25080
+    },
+    {
+      "epoch": 0.18161813141074362,
+      "grad_norm": 0.20874586701393127,
+      "learning_rate": 4.818389107255315e-06,
+      "loss": 1.0054,
+      "step": 25090
+    },
+    {
+      "epoch": 0.18169051807132983,
+      "grad_norm": 0.17557349801063538,
+      "learning_rate": 4.818316720594729e-06,
+      "loss": 1.0165,
+      "step": 25100
+    },
+    {
+      "epoch": 0.181762904731916,
+      "grad_norm": 0.17090950906276703,
+      "learning_rate": 4.818244333934143e-06,
+      "loss": 1.0049,
+      "step": 25110
+    },
+    {
+      "epoch": 0.18183529139250218,
+      "grad_norm": 0.20330511033535004,
+      "learning_rate": 4.818171947273557e-06,
+      "loss": 1.0095,
+      "step": 25120
+    },
+    {
+      "epoch": 0.18190767805308838,
+      "grad_norm": 0.1726309359073639,
+      "learning_rate": 4.81809956061297e-06,
+      "loss": 1.0183,
+      "step": 25130
+    },
+    {
+      "epoch": 0.18198006471367456,
+      "grad_norm": 0.18963780999183655,
+      "learning_rate": 4.818027173952384e-06,
+      "loss": 0.9986,
+      "step": 25140
+    },
+    {
+      "epoch": 0.18205245137426074,
+      "grad_norm": 0.18106186389923096,
+      "learning_rate": 4.8179547872917985e-06,
+      "loss": 1.0016,
+      "step": 25150
+    },
+    {
+      "epoch": 0.18212483803484694,
+      "grad_norm": 0.17316314578056335,
+      "learning_rate": 4.817882400631212e-06,
+      "loss": 0.9989,
+      "step": 25160
+    },
+    {
+      "epoch": 0.18219722469543312,
+      "grad_norm": 0.17553074657917023,
+      "learning_rate": 4.817810013970626e-06,
+      "loss": 0.9927,
+      "step": 25170
+    },
+    {
+      "epoch": 0.18226961135601932,
+      "grad_norm": 0.21649324893951416,
+      "learning_rate": 4.817737627310039e-06,
+      "loss": 1.0065,
+      "step": 25180
+    },
+    {
+      "epoch": 0.1823419980166055,
+      "grad_norm": 0.19755081832408905,
+      "learning_rate": 4.817665240649454e-06,
+      "loss": 1.0105,
+      "step": 25190
+    },
+    {
+      "epoch": 0.18241438467719168,
+      "grad_norm": 0.15995916724205017,
+      "learning_rate": 4.817592853988867e-06,
+      "loss": 1.0064,
+      "step": 25200
+    },
+    {
+      "epoch": 0.18248677133777788,
+      "grad_norm": 0.17277465760707855,
+      "learning_rate": 4.817520467328281e-06,
+      "loss": 1.0104,
+      "step": 25210
+    },
+    {
+      "epoch": 0.18255915799836406,
+      "grad_norm": 0.17675970494747162,
+      "learning_rate": 4.817448080667695e-06,
+      "loss": 0.9952,
+      "step": 25220
+    },
+    {
+      "epoch": 0.18263154465895026,
+      "grad_norm": 0.18293781578540802,
+      "learning_rate": 4.817375694007109e-06,
+      "loss": 0.9949,
+      "step": 25230
+    },
+    {
+      "epoch": 0.18270393131953644,
+      "grad_norm": 0.18516694009304047,
+      "learning_rate": 4.817303307346523e-06,
+      "loss": 1.0042,
+      "step": 25240
+    },
+    {
+      "epoch": 0.18277631798012262,
+      "grad_norm": 0.1832907646894455,
+      "learning_rate": 4.817230920685936e-06,
+      "loss": 1.0153,
+      "step": 25250
+    },
+    {
+      "epoch": 0.18284870464070882,
+      "grad_norm": 0.17651574313640594,
+      "learning_rate": 4.81715853402535e-06,
+      "loss": 1.0011,
+      "step": 25260
+    },
+    {
+      "epoch": 0.182921091301295,
+      "grad_norm": 0.17321477830410004,
+      "learning_rate": 4.8170861473647636e-06,
+      "loss": 0.9985,
+      "step": 25270
+    },
+    {
+      "epoch": 0.18299347796188117,
+      "grad_norm": 0.16472671926021576,
+      "learning_rate": 4.817013760704178e-06,
+      "loss": 1.0067,
+      "step": 25280
+    },
+    {
+      "epoch": 0.18306586462246738,
+      "grad_norm": 0.16202005743980408,
+      "learning_rate": 4.816941374043592e-06,
+      "loss": 1.0109,
+      "step": 25290
+    },
+    {
+      "epoch": 0.18313825128305355,
+      "grad_norm": 0.19252698123455048,
+      "learning_rate": 4.816868987383005e-06,
+      "loss": 0.9949,
+      "step": 25300
+    },
+    {
+      "epoch": 0.18321063794363976,
+      "grad_norm": 0.18542930483818054,
+      "learning_rate": 4.816796600722419e-06,
+      "loss": 1.0235,
+      "step": 25310
+    },
+    {
+      "epoch": 0.18328302460422594,
+      "grad_norm": 0.17838090658187866,
+      "learning_rate": 4.816724214061833e-06,
+      "loss": 1.0196,
+      "step": 25320
+    },
+    {
+      "epoch": 0.1833554112648121,
+      "grad_norm": 0.1758023202419281,
+      "learning_rate": 4.816651827401247e-06,
+      "loss": 1.0076,
+      "step": 25330
+    },
+    {
+      "epoch": 0.18342779792539832,
+      "grad_norm": 0.16187690198421478,
+      "learning_rate": 4.8165794407406606e-06,
+      "loss": 1.0086,
+      "step": 25340
+    },
+    {
+      "epoch": 0.1835001845859845,
+      "grad_norm": 0.16450618207454681,
+      "learning_rate": 4.816507054080074e-06,
+      "loss": 1.0065,
+      "step": 25350
+    },
+    {
+      "epoch": 0.18357257124657067,
+      "grad_norm": 0.18235477805137634,
+      "learning_rate": 4.816434667419489e-06,
+      "loss": 1.0105,
+      "step": 25360
+    },
+    {
+      "epoch": 0.18364495790715687,
+      "grad_norm": 0.1665814220905304,
+      "learning_rate": 4.816362280758902e-06,
+      "loss": 0.9997,
+      "step": 25370
+    },
+    {
+      "epoch": 0.18371734456774305,
+      "grad_norm": 0.17312000691890717,
+      "learning_rate": 4.816289894098316e-06,
+      "loss": 1.0143,
+      "step": 25380
+    },
+    {
+      "epoch": 0.18378973122832926,
+      "grad_norm": 0.16814294457435608,
+      "learning_rate": 4.8162175074377295e-06,
+      "loss": 0.987,
+      "step": 25390
+    },
+    {
+      "epoch": 0.18386211788891543,
+      "grad_norm": 0.17651145160198212,
+      "learning_rate": 4.816145120777144e-06,
+      "loss": 1.0058,
+      "step": 25400
+    },
+    {
+      "epoch": 0.1839345045495016,
+      "grad_norm": 0.1924390345811844,
+      "learning_rate": 4.816072734116558e-06,
+      "loss": 1.0145,
+      "step": 25410
+    },
+    {
+      "epoch": 0.18400689121008781,
+      "grad_norm": 0.1660076081752777,
+      "learning_rate": 4.816000347455971e-06,
+      "loss": 0.9972,
+      "step": 25420
+    },
+    {
+      "epoch": 0.184079277870674,
+      "grad_norm": 0.16928330063819885,
+      "learning_rate": 4.815927960795385e-06,
+      "loss": 1.0043,
+      "step": 25430
+    },
+    {
+      "epoch": 0.18415166453126017,
+      "grad_norm": 0.2533145546913147,
+      "learning_rate": 4.815855574134799e-06,
+      "loss": 0.9977,
+      "step": 25440
+    },
+    {
+      "epoch": 0.18422405119184637,
+      "grad_norm": 0.17467600107192993,
+      "learning_rate": 4.815783187474213e-06,
+      "loss": 1.0004,
+      "step": 25450
+    },
+    {
+      "epoch": 0.18429643785243255,
+      "grad_norm": 0.16953355073928833,
+      "learning_rate": 4.8157108008136265e-06,
+      "loss": 1.0088,
+      "step": 25460
+    },
+    {
+      "epoch": 0.18436882451301875,
+      "grad_norm": 0.17765529453754425,
+      "learning_rate": 4.81563841415304e-06,
+      "loss": 1.0144,
+      "step": 25470
+    },
+    {
+      "epoch": 0.18444121117360493,
+      "grad_norm": 0.16797038912773132,
+      "learning_rate": 4.815566027492455e-06,
+      "loss": 1.0036,
+      "step": 25480
+    },
+    {
+      "epoch": 0.1845135978341911,
+      "grad_norm": 0.17482724785804749,
+      "learning_rate": 4.815493640831868e-06,
+      "loss": 1.0054,
+      "step": 25490
+    },
+    {
+      "epoch": 0.1845859844947773,
+      "grad_norm": 0.18403145670890808,
+      "learning_rate": 4.815421254171282e-06,
+      "loss": 1.0197,
+      "step": 25500
+    },
+    {
+      "epoch": 0.1846583711553635,
+      "grad_norm": 0.1704588383436203,
+      "learning_rate": 4.8153488675106954e-06,
+      "loss": 1.0074,
+      "step": 25510
+    },
+    {
+      "epoch": 0.18473075781594966,
+      "grad_norm": 0.1732064187526703,
+      "learning_rate": 4.81527648085011e-06,
+      "loss": 1.0107,
+      "step": 25520
+    },
+    {
+      "epoch": 0.18480314447653587,
+      "grad_norm": 0.16858525574207306,
+      "learning_rate": 4.8152040941895235e-06,
+      "loss": 1.0017,
+      "step": 25530
+    },
+    {
+      "epoch": 0.18487553113712205,
+      "grad_norm": 0.17689578235149384,
+      "learning_rate": 4.815131707528937e-06,
+      "loss": 0.995,
+      "step": 25540
+    },
+    {
+      "epoch": 0.18494791779770825,
+      "grad_norm": 0.16487176716327667,
+      "learning_rate": 4.815059320868351e-06,
+      "loss": 1.0149,
+      "step": 25550
+    },
+    {
+      "epoch": 0.18502030445829443,
+      "grad_norm": 0.15995535254478455,
+      "learning_rate": 4.814986934207765e-06,
+      "loss": 1.0098,
+      "step": 25560
+    },
+    {
+      "epoch": 0.1850926911188806,
+      "grad_norm": 0.18028096854686737,
+      "learning_rate": 4.814914547547178e-06,
+      "loss": 0.9971,
+      "step": 25570
+    },
+    {
+      "epoch": 0.1851650777794668,
+      "grad_norm": 0.1847427487373352,
+      "learning_rate": 4.814842160886592e-06,
+      "loss": 0.9958,
+      "step": 25580
+    },
+    {
+      "epoch": 0.18523746444005298,
+      "grad_norm": 0.17257408797740936,
+      "learning_rate": 4.814769774226006e-06,
+      "loss": 1.0213,
+      "step": 25590
+    },
+    {
+      "epoch": 0.18530985110063916,
+      "grad_norm": 0.17929920554161072,
+      "learning_rate": 4.81469738756542e-06,
+      "loss": 1.0089,
+      "step": 25600
+    },
+    {
+      "epoch": 0.18538223776122537,
+      "grad_norm": 0.16596031188964844,
+      "learning_rate": 4.814625000904833e-06,
+      "loss": 1.0069,
+      "step": 25610
+    },
+    {
+      "epoch": 0.18545462442181154,
+      "grad_norm": 0.17687474191188812,
+      "learning_rate": 4.814552614244247e-06,
+      "loss": 1.0072,
+      "step": 25620
+    },
+    {
+      "epoch": 0.18552701108239775,
+      "grad_norm": 0.18298721313476562,
+      "learning_rate": 4.814480227583661e-06,
+      "loss": 1.0038,
+      "step": 25630
+    },
+    {
+      "epoch": 0.18559939774298392,
+      "grad_norm": 0.17622758448123932,
+      "learning_rate": 4.814407840923075e-06,
+      "loss": 1.0052,
+      "step": 25640
+    },
+    {
+      "epoch": 0.1856717844035701,
+      "grad_norm": 0.17824731767177582,
+      "learning_rate": 4.814335454262489e-06,
+      "loss": 1.0,
+      "step": 25650
+    },
+    {
+      "epoch": 0.1857441710641563,
+      "grad_norm": 0.16804032027721405,
+      "learning_rate": 4.814263067601902e-06,
+      "loss": 1.0189,
+      "step": 25660
+    },
+    {
+      "epoch": 0.18581655772474248,
+      "grad_norm": 0.16727906465530396,
+      "learning_rate": 4.814190680941317e-06,
+      "loss": 1.0104,
+      "step": 25670
+    },
+    {
+      "epoch": 0.18588894438532866,
+      "grad_norm": 0.16687215864658356,
+      "learning_rate": 4.81411829428073e-06,
+      "loss": 1.0073,
+      "step": 25680
+    },
+    {
+      "epoch": 0.18596133104591486,
+      "grad_norm": 0.16591374576091766,
+      "learning_rate": 4.814045907620144e-06,
+      "loss": 1.0267,
+      "step": 25690
+    },
+    {
+      "epoch": 0.18603371770650104,
+      "grad_norm": 0.19451576471328735,
+      "learning_rate": 4.8139735209595575e-06,
+      "loss": 0.9904,
+      "step": 25700
+    },
+    {
+      "epoch": 0.18610610436708724,
+      "grad_norm": 0.179254949092865,
+      "learning_rate": 4.813901134298972e-06,
+      "loss": 1.0082,
+      "step": 25710
+    },
+    {
+      "epoch": 0.18617849102767342,
+      "grad_norm": 0.17360366880893707,
+      "learning_rate": 4.813828747638386e-06,
+      "loss": 1.0003,
+      "step": 25720
+    },
+    {
+      "epoch": 0.1862508776882596,
+      "grad_norm": 0.15899336338043213,
+      "learning_rate": 4.813756360977799e-06,
+      "loss": 1.0011,
+      "step": 25730
+    },
+    {
+      "epoch": 0.1863232643488458,
+      "grad_norm": 0.16736361384391785,
+      "learning_rate": 4.813683974317213e-06,
+      "loss": 1.0155,
+      "step": 25740
+    },
+    {
+      "epoch": 0.18639565100943198,
+      "grad_norm": 0.18747742474079132,
+      "learning_rate": 4.813611587656627e-06,
+      "loss": 0.9963,
+      "step": 25750
+    },
+    {
+      "epoch": 0.18646803767001816,
+      "grad_norm": 0.177093505859375,
+      "learning_rate": 4.813539200996041e-06,
+      "loss": 1.0041,
+      "step": 25760
+    },
+    {
+      "epoch": 0.18654042433060436,
+      "grad_norm": 0.16255925595760345,
+      "learning_rate": 4.8134668143354545e-06,
+      "loss": 1.0181,
+      "step": 25770
+    },
+    {
+      "epoch": 0.18661281099119054,
+      "grad_norm": 0.1680876463651657,
+      "learning_rate": 4.813394427674868e-06,
+      "loss": 1.0146,
+      "step": 25780
+    },
+    {
+      "epoch": 0.18668519765177674,
+      "grad_norm": 0.16530559957027435,
+      "learning_rate": 4.813322041014283e-06,
+      "loss": 0.9985,
+      "step": 25790
+    },
+    {
+      "epoch": 0.18675758431236292,
+      "grad_norm": 0.17303499579429626,
+      "learning_rate": 4.813249654353696e-06,
+      "loss": 1.0045,
+      "step": 25800
+    },
+    {
+      "epoch": 0.1868299709729491,
+      "grad_norm": 0.18719731271266937,
+      "learning_rate": 4.81317726769311e-06,
+      "loss": 1.0188,
+      "step": 25810
+    },
+    {
+      "epoch": 0.1869023576335353,
+      "grad_norm": 0.17420874536037445,
+      "learning_rate": 4.8131048810325235e-06,
+      "loss": 1.0054,
+      "step": 25820
+    },
+    {
+      "epoch": 0.18697474429412148,
+      "grad_norm": 0.16283327341079712,
+      "learning_rate": 4.813032494371938e-06,
+      "loss": 1.01,
+      "step": 25830
+    },
+    {
+      "epoch": 0.18704713095470768,
+      "grad_norm": 0.16841505467891693,
+      "learning_rate": 4.8129601077113515e-06,
+      "loss": 1.0008,
+      "step": 25840
+    },
+    {
+      "epoch": 0.18711951761529386,
+      "grad_norm": 0.16842392086982727,
+      "learning_rate": 4.812887721050765e-06,
+      "loss": 0.9914,
+      "step": 25850
+    },
+    {
+      "epoch": 0.18719190427588003,
+      "grad_norm": 0.16453711688518524,
+      "learning_rate": 4.812815334390179e-06,
+      "loss": 1.0025,
+      "step": 25860
+    },
+    {
+      "epoch": 0.18726429093646624,
+      "grad_norm": 0.15623344480991364,
+      "learning_rate": 4.812742947729593e-06,
+      "loss": 1.0011,
+      "step": 25870
+    },
+    {
+      "epoch": 0.18733667759705241,
+      "grad_norm": 0.1698412448167801,
+      "learning_rate": 4.812670561069007e-06,
+      "loss": 1.0064,
+      "step": 25880
+    },
+    {
+      "epoch": 0.1874090642576386,
+      "grad_norm": 0.17024236917495728,
+      "learning_rate": 4.8125981744084205e-06,
+      "loss": 0.9973,
+      "step": 25890
+    },
+    {
+      "epoch": 0.1874814509182248,
+      "grad_norm": 0.1828537881374359,
+      "learning_rate": 4.812525787747834e-06,
+      "loss": 1.0015,
+      "step": 25900
+    },
+    {
+      "epoch": 0.18755383757881097,
+      "grad_norm": 0.25711390376091003,
+      "learning_rate": 4.812453401087248e-06,
+      "loss": 1.0139,
+      "step": 25910
+    },
+    {
+      "epoch": 0.18762622423939718,
+      "grad_norm": 0.17483864724636078,
+      "learning_rate": 4.812381014426662e-06,
+      "loss": 1.0105,
+      "step": 25920
+    },
+    {
+      "epoch": 0.18769861089998335,
+      "grad_norm": 0.18660502135753632,
+      "learning_rate": 4.812308627766076e-06,
+      "loss": 1.0001,
+      "step": 25930
+    },
+    {
+      "epoch": 0.18777099756056953,
+      "grad_norm": 0.16732463240623474,
+      "learning_rate": 4.812236241105489e-06,
+      "loss": 0.9918,
+      "step": 25940
+    },
+    {
+      "epoch": 0.18784338422115573,
+      "grad_norm": 0.16313007473945618,
+      "learning_rate": 4.812163854444903e-06,
+      "loss": 0.9983,
+      "step": 25950
+    },
+    {
+      "epoch": 0.1879157708817419,
+      "grad_norm": 0.17107586562633514,
+      "learning_rate": 4.8120914677843175e-06,
+      "loss": 1.0116,
+      "step": 25960
+    },
+    {
+      "epoch": 0.1879881575423281,
+      "grad_norm": 0.17458048462867737,
+      "learning_rate": 4.812019081123731e-06,
+      "loss": 0.998,
+      "step": 25970
+    },
+    {
+      "epoch": 0.1880605442029143,
+      "grad_norm": 0.16045020520687103,
+      "learning_rate": 4.811946694463145e-06,
+      "loss": 0.9783,
+      "step": 25980
+    },
+    {
+      "epoch": 0.18813293086350047,
+      "grad_norm": 0.1742679625749588,
+      "learning_rate": 4.811874307802558e-06,
+      "loss": 1.0056,
+      "step": 25990
+    },
+    {
+      "epoch": 0.18820531752408667,
+      "grad_norm": 0.21241635084152222,
+      "learning_rate": 4.811801921141973e-06,
+      "loss": 1.0108,
+      "step": 26000
+    },
+    {
+      "epoch": 0.18827770418467285,
+      "grad_norm": 0.2042105346918106,
+      "learning_rate": 4.811729534481386e-06,
+      "loss": 0.9975,
+      "step": 26010
+    },
+    {
+      "epoch": 0.18835009084525903,
+      "grad_norm": 0.18309561908245087,
+      "learning_rate": 4.8116571478208e-06,
+      "loss": 1.0094,
+      "step": 26020
+    },
+    {
+      "epoch": 0.18842247750584523,
+      "grad_norm": 0.17432443797588348,
+      "learning_rate": 4.811584761160214e-06,
+      "loss": 0.9972,
+      "step": 26030
+    },
+    {
+      "epoch": 0.1884948641664314,
+      "grad_norm": 0.17750132083892822,
+      "learning_rate": 4.811512374499628e-06,
+      "loss": 0.9996,
+      "step": 26040
+    },
+    {
+      "epoch": 0.18856725082701759,
+      "grad_norm": 0.17087271809577942,
+      "learning_rate": 4.811439987839042e-06,
+      "loss": 1.0022,
+      "step": 26050
+    },
+    {
+      "epoch": 0.1886396374876038,
+      "grad_norm": 0.1847958117723465,
+      "learning_rate": 4.811367601178455e-06,
+      "loss": 1.0217,
+      "step": 26060
+    },
+    {
+      "epoch": 0.18871202414818997,
+      "grad_norm": 0.20137399435043335,
+      "learning_rate": 4.811295214517869e-06,
+      "loss": 0.9926,
+      "step": 26070
+    },
+    {
+      "epoch": 0.18878441080877617,
+      "grad_norm": 0.18774452805519104,
+      "learning_rate": 4.811222827857283e-06,
+      "loss": 1.0058,
+      "step": 26080
+    },
+    {
+      "epoch": 0.18885679746936235,
+      "grad_norm": 0.18850480020046234,
+      "learning_rate": 4.811150441196697e-06,
+      "loss": 0.9942,
+      "step": 26090
+    },
+    {
+      "epoch": 0.18892918412994852,
+      "grad_norm": 0.18268634378910065,
+      "learning_rate": 4.81107805453611e-06,
+      "loss": 0.9998,
+      "step": 26100
+    },
+    {
+      "epoch": 0.18900157079053473,
+      "grad_norm": 0.16116634011268616,
+      "learning_rate": 4.811005667875524e-06,
+      "loss": 0.9966,
+      "step": 26110
+    },
+    {
+      "epoch": 0.1890739574511209,
+      "grad_norm": 0.18132539093494415,
+      "learning_rate": 4.810933281214938e-06,
+      "loss": 1.006,
+      "step": 26120
+    },
+    {
+      "epoch": 0.18914634411170708,
+      "grad_norm": 0.19577182829380035,
+      "learning_rate": 4.8108608945543515e-06,
+      "loss": 1.0125,
+      "step": 26130
+    },
+    {
+      "epoch": 0.1892187307722933,
+      "grad_norm": 0.2031572461128235,
+      "learning_rate": 4.810788507893765e-06,
+      "loss": 1.0017,
+      "step": 26140
+    },
+    {
+      "epoch": 0.18929111743287946,
+      "grad_norm": 0.1615428924560547,
+      "learning_rate": 4.81071612123318e-06,
+      "loss": 0.9947,
+      "step": 26150
+    },
+    {
+      "epoch": 0.18936350409346567,
+      "grad_norm": 0.18610787391662598,
+      "learning_rate": 4.810643734572593e-06,
+      "loss": 1.0,
+      "step": 26160
+    },
+    {
+      "epoch": 0.18943589075405184,
+      "grad_norm": 0.1651063859462738,
+      "learning_rate": 4.810571347912007e-06,
+      "loss": 0.9999,
+      "step": 26170
+    },
+    {
+      "epoch": 0.18950827741463802,
+      "grad_norm": 0.1639026403427124,
+      "learning_rate": 4.8104989612514204e-06,
+      "loss": 1.0054,
+      "step": 26180
+    },
+    {
+      "epoch": 0.18958066407522423,
+      "grad_norm": 0.17324158549308777,
+      "learning_rate": 4.810426574590835e-06,
+      "loss": 1.0045,
+      "step": 26190
+    },
+    {
+      "epoch": 0.1896530507358104,
+      "grad_norm": 0.1581469178199768,
+      "learning_rate": 4.8103541879302485e-06,
+      "loss": 0.9932,
+      "step": 26200
+    },
+    {
+      "epoch": 0.18972543739639658,
+      "grad_norm": 0.17741291224956512,
+      "learning_rate": 4.810281801269662e-06,
+      "loss": 0.9954,
+      "step": 26210
+    },
+    {
+      "epoch": 0.18979782405698278,
+      "grad_norm": 0.21022439002990723,
+      "learning_rate": 4.810209414609076e-06,
+      "loss": 1.0093,
+      "step": 26220
+    },
+    {
+      "epoch": 0.18987021071756896,
+      "grad_norm": 0.1657036393880844,
+      "learning_rate": 4.81013702794849e-06,
+      "loss": 0.9934,
+      "step": 26230
+    },
+    {
+      "epoch": 0.18994259737815516,
+      "grad_norm": 0.17548425495624542,
+      "learning_rate": 4.810064641287904e-06,
+      "loss": 1.0048,
+      "step": 26240
+    },
+    {
+      "epoch": 0.19001498403874134,
+      "grad_norm": 0.17935971915721893,
+      "learning_rate": 4.8099922546273174e-06,
+      "loss": 1.0143,
+      "step": 26250
+    },
+    {
+      "epoch": 0.19008737069932752,
+      "grad_norm": 0.16744934022426605,
+      "learning_rate": 4.809919867966731e-06,
+      "loss": 0.9974,
+      "step": 26260
+    },
+    {
+      "epoch": 0.19015975735991372,
+      "grad_norm": 0.18624848127365112,
+      "learning_rate": 4.8098474813061455e-06,
+      "loss": 1.0077,
+      "step": 26270
+    },
+    {
+      "epoch": 0.1902321440204999,
+      "grad_norm": 0.21752850711345673,
+      "learning_rate": 4.809775094645559e-06,
+      "loss": 1.0113,
+      "step": 26280
+    },
+    {
+      "epoch": 0.19030453068108608,
+      "grad_norm": 0.16781753301620483,
+      "learning_rate": 4.809702707984973e-06,
+      "loss": 1.0066,
+      "step": 26290
+    },
+    {
+      "epoch": 0.19037691734167228,
+      "grad_norm": 0.16836953163146973,
+      "learning_rate": 4.809630321324386e-06,
+      "loss": 1.0001,
+      "step": 26300
+    },
+    {
+      "epoch": 0.19044930400225846,
+      "grad_norm": 0.2364221215248108,
+      "learning_rate": 4.809557934663801e-06,
+      "loss": 0.9922,
+      "step": 26310
+    },
+    {
+      "epoch": 0.19052169066284466,
+      "grad_norm": 0.18490445613861084,
+      "learning_rate": 4.8094855480032144e-06,
+      "loss": 1.0104,
+      "step": 26320
+    },
+    {
+      "epoch": 0.19059407732343084,
+      "grad_norm": 0.1624281406402588,
+      "learning_rate": 4.809413161342628e-06,
+      "loss": 0.9976,
+      "step": 26330
+    },
+    {
+      "epoch": 0.19066646398401702,
+      "grad_norm": 0.17725110054016113,
+      "learning_rate": 4.809340774682042e-06,
+      "loss": 1.0087,
+      "step": 26340
+    },
+    {
+      "epoch": 0.19073885064460322,
+      "grad_norm": 0.1625022143125534,
+      "learning_rate": 4.809268388021456e-06,
+      "loss": 1.0036,
+      "step": 26350
+    },
+    {
+      "epoch": 0.1908112373051894,
+      "grad_norm": 0.1683979332447052,
+      "learning_rate": 4.80919600136087e-06,
+      "loss": 0.9992,
+      "step": 26360
+    },
+    {
+      "epoch": 0.1908836239657756,
+      "grad_norm": 0.22802671790122986,
+      "learning_rate": 4.809123614700283e-06,
+      "loss": 1.0043,
+      "step": 26370
+    },
+    {
+      "epoch": 0.19095601062636178,
+      "grad_norm": 0.16152718663215637,
+      "learning_rate": 4.809051228039697e-06,
+      "loss": 0.9968,
+      "step": 26380
+    },
+    {
+      "epoch": 0.19102839728694795,
+      "grad_norm": 0.17749905586242676,
+      "learning_rate": 4.8089788413791115e-06,
+      "loss": 1.0124,
+      "step": 26390
+    },
+    {
+      "epoch": 0.19110078394753416,
+      "grad_norm": 0.16626082360744476,
+      "learning_rate": 4.808906454718525e-06,
+      "loss": 0.9949,
+      "step": 26400
+    },
+    {
+      "epoch": 0.19117317060812034,
+      "grad_norm": 0.1978878676891327,
+      "learning_rate": 4.808834068057939e-06,
+      "loss": 1.0022,
+      "step": 26410
+    },
+    {
+      "epoch": 0.1912455572687065,
+      "grad_norm": 0.1693107634782791,
+      "learning_rate": 4.808761681397352e-06,
+      "loss": 0.9924,
+      "step": 26420
+    },
+    {
+      "epoch": 0.19131794392929272,
+      "grad_norm": 0.18437570333480835,
+      "learning_rate": 4.808689294736767e-06,
+      "loss": 1.0065,
+      "step": 26430
+    },
+    {
+      "epoch": 0.1913903305898789,
+      "grad_norm": 0.1534915566444397,
+      "learning_rate": 4.80861690807618e-06,
+      "loss": 1.0144,
+      "step": 26440
+    },
+    {
+      "epoch": 0.1914627172504651,
+      "grad_norm": 0.17550812661647797,
+      "learning_rate": 4.808544521415594e-06,
+      "loss": 0.9951,
+      "step": 26450
+    },
+    {
+      "epoch": 0.19153510391105127,
+      "grad_norm": 0.19502116739749908,
+      "learning_rate": 4.808472134755008e-06,
+      "loss": 0.9999,
+      "step": 26460
+    },
+    {
+      "epoch": 0.19160749057163745,
+      "grad_norm": 0.1792958527803421,
+      "learning_rate": 4.808399748094422e-06,
+      "loss": 0.9968,
+      "step": 26470
+    },
+    {
+      "epoch": 0.19167987723222366,
+      "grad_norm": 0.16823455691337585,
+      "learning_rate": 4.808327361433836e-06,
+      "loss": 1.0029,
+      "step": 26480
+    },
+    {
+      "epoch": 0.19175226389280983,
+      "grad_norm": 0.17597924172878265,
+      "learning_rate": 4.808254974773249e-06,
+      "loss": 1.008,
+      "step": 26490
+    },
+    {
+      "epoch": 0.191824650553396,
+      "grad_norm": 0.1772463321685791,
+      "learning_rate": 4.808182588112663e-06,
+      "loss": 0.9921,
+      "step": 26500
+    },
+    {
+      "epoch": 0.1918970372139822,
+      "grad_norm": 0.1926368921995163,
+      "learning_rate": 4.808110201452077e-06,
+      "loss": 0.9993,
+      "step": 26510
+    },
+    {
+      "epoch": 0.1919694238745684,
+      "grad_norm": 0.16659504175186157,
+      "learning_rate": 4.808037814791491e-06,
+      "loss": 1.0063,
+      "step": 26520
+    },
+    {
+      "epoch": 0.1920418105351546,
+      "grad_norm": 0.1627788245677948,
+      "learning_rate": 4.807965428130905e-06,
+      "loss": 0.9983,
+      "step": 26530
+    },
+    {
+      "epoch": 0.19211419719574077,
+      "grad_norm": 0.30011650919914246,
+      "learning_rate": 4.807893041470318e-06,
+      "loss": 0.9909,
+      "step": 26540
+    },
+    {
+      "epoch": 0.19218658385632695,
+      "grad_norm": 0.17447486519813538,
+      "learning_rate": 4.807820654809732e-06,
+      "loss": 0.9992,
+      "step": 26550
+    },
+    {
+      "epoch": 0.19225897051691315,
+      "grad_norm": 0.17184774577617645,
+      "learning_rate": 4.807748268149146e-06,
+      "loss": 1.0044,
+      "step": 26560
+    },
+    {
+      "epoch": 0.19233135717749933,
+      "grad_norm": 0.17131789028644562,
+      "learning_rate": 4.80767588148856e-06,
+      "loss": 1.0048,
+      "step": 26570
+    },
+    {
+      "epoch": 0.1924037438380855,
+      "grad_norm": 0.1681704968214035,
+      "learning_rate": 4.8076034948279735e-06,
+      "loss": 1.0105,
+      "step": 26580
+    },
+    {
+      "epoch": 0.1924761304986717,
+      "grad_norm": 0.18246489763259888,
+      "learning_rate": 4.807531108167387e-06,
+      "loss": 0.9853,
+      "step": 26590
+    },
+    {
+      "epoch": 0.1925485171592579,
+      "grad_norm": 0.1730792075395584,
+      "learning_rate": 4.807458721506802e-06,
+      "loss": 1.0048,
+      "step": 26600
+    },
+    {
+      "epoch": 0.1926209038198441,
+      "grad_norm": 0.1860353797674179,
+      "learning_rate": 4.807386334846215e-06,
+      "loss": 1.0024,
+      "step": 26610
+    },
+    {
+      "epoch": 0.19269329048043027,
+      "grad_norm": 0.16695286333560944,
+      "learning_rate": 4.807313948185629e-06,
+      "loss": 1.0142,
+      "step": 26620
+    },
+    {
+      "epoch": 0.19276567714101644,
+      "grad_norm": 0.17245034873485565,
+      "learning_rate": 4.8072415615250425e-06,
+      "loss": 1.0003,
+      "step": 26630
+    },
+    {
+      "epoch": 0.19283806380160265,
+      "grad_norm": 0.16842077672481537,
+      "learning_rate": 4.807169174864456e-06,
+      "loss": 1.0141,
+      "step": 26640
+    },
+    {
+      "epoch": 0.19291045046218883,
+      "grad_norm": 0.16834910213947296,
+      "learning_rate": 4.80709678820387e-06,
+      "loss": 0.9903,
+      "step": 26650
+    },
+    {
+      "epoch": 0.192982837122775,
+      "grad_norm": 0.1826198846101761,
+      "learning_rate": 4.807024401543284e-06,
+      "loss": 1.0076,
+      "step": 26660
+    },
+    {
+      "epoch": 0.1930552237833612,
+      "grad_norm": 0.17187774181365967,
+      "learning_rate": 4.806952014882698e-06,
+      "loss": 1.0037,
+      "step": 26670
+    },
+    {
+      "epoch": 0.19312761044394738,
+      "grad_norm": 0.1815541386604309,
+      "learning_rate": 4.806879628222111e-06,
+      "loss": 1.0086,
+      "step": 26680
+    },
+    {
+      "epoch": 0.1931999971045336,
+      "grad_norm": 0.18151985108852386,
+      "learning_rate": 4.806807241561525e-06,
+      "loss": 0.9894,
+      "step": 26690
+    },
+    {
+      "epoch": 0.19327238376511977,
+      "grad_norm": 0.17643725872039795,
+      "learning_rate": 4.806734854900939e-06,
+      "loss": 1.0044,
+      "step": 26700
+    },
+    {
+      "epoch": 0.19334477042570594,
+      "grad_norm": 0.17447534203529358,
+      "learning_rate": 4.806662468240353e-06,
+      "loss": 1.0114,
+      "step": 26710
+    },
+    {
+      "epoch": 0.19341715708629215,
+      "grad_norm": 0.16424672305583954,
+      "learning_rate": 4.806590081579767e-06,
+      "loss": 1.0033,
+      "step": 26720
+    },
+    {
+      "epoch": 0.19348954374687832,
+      "grad_norm": 0.16802680492401123,
+      "learning_rate": 4.80651769491918e-06,
+      "loss": 1.0072,
+      "step": 26730
+    },
+    {
+      "epoch": 0.1935619304074645,
+      "grad_norm": 0.17925448715686798,
+      "learning_rate": 4.806445308258594e-06,
+      "loss": 0.9935,
+      "step": 26740
+    },
+    {
+      "epoch": 0.1936343170680507,
+      "grad_norm": 0.17291328310966492,
+      "learning_rate": 4.806372921598008e-06,
+      "loss": 0.9942,
+      "step": 26750
+    },
+    {
+      "epoch": 0.19370670372863688,
+      "grad_norm": 0.18986187875270844,
+      "learning_rate": 4.806300534937422e-06,
+      "loss": 1.0095,
+      "step": 26760
+    },
+    {
+      "epoch": 0.19377909038922309,
+      "grad_norm": 0.15968811511993408,
+      "learning_rate": 4.806228148276836e-06,
+      "loss": 1.0083,
+      "step": 26770
+    },
+    {
+      "epoch": 0.19385147704980926,
+      "grad_norm": 0.18246155977249146,
+      "learning_rate": 4.806155761616249e-06,
+      "loss": 0.9959,
+      "step": 26780
+    },
+    {
+      "epoch": 0.19392386371039544,
+      "grad_norm": 0.1724570244550705,
+      "learning_rate": 4.806083374955664e-06,
+      "loss": 0.993,
+      "step": 26790
+    },
+    {
+      "epoch": 0.19399625037098164,
+      "grad_norm": 0.16867926716804504,
+      "learning_rate": 4.806010988295077e-06,
+      "loss": 1.0064,
+      "step": 26800
+    },
+    {
+      "epoch": 0.19406863703156782,
+      "grad_norm": 0.16909553110599518,
+      "learning_rate": 4.805938601634491e-06,
+      "loss": 1.0073,
+      "step": 26810
+    },
+    {
+      "epoch": 0.194141023692154,
+      "grad_norm": 0.2134760469198227,
+      "learning_rate": 4.8058662149739046e-06,
+      "loss": 0.9893,
+      "step": 26820
+    },
+    {
+      "epoch": 0.1942134103527402,
+      "grad_norm": 0.16521146893501282,
+      "learning_rate": 4.805793828313319e-06,
+      "loss": 0.9962,
+      "step": 26830
+    },
+    {
+      "epoch": 0.19428579701332638,
+      "grad_norm": 0.16288983821868896,
+      "learning_rate": 4.805721441652733e-06,
+      "loss": 0.9958,
+      "step": 26840
+    },
+    {
+      "epoch": 0.19435818367391258,
+      "grad_norm": 0.17485104501247406,
+      "learning_rate": 4.805649054992146e-06,
+      "loss": 1.0048,
+      "step": 26850
+    },
+    {
+      "epoch": 0.19443057033449876,
+      "grad_norm": 0.15907074511051178,
+      "learning_rate": 4.80557666833156e-06,
+      "loss": 1.0056,
+      "step": 26860
+    },
+    {
+      "epoch": 0.19450295699508494,
+      "grad_norm": 0.17798353731632233,
+      "learning_rate": 4.805504281670974e-06,
+      "loss": 1.0024,
+      "step": 26870
+    },
+    {
+      "epoch": 0.19457534365567114,
+      "grad_norm": 0.18437181413173676,
+      "learning_rate": 4.805431895010388e-06,
+      "loss": 0.9971,
+      "step": 26880
+    },
+    {
+      "epoch": 0.19464773031625732,
+      "grad_norm": 0.15452776849269867,
+      "learning_rate": 4.805359508349802e-06,
+      "loss": 1.0092,
+      "step": 26890
+    },
+    {
+      "epoch": 0.1947201169768435,
+      "grad_norm": 0.16308918595314026,
+      "learning_rate": 4.805287121689215e-06,
+      "loss": 1.0055,
+      "step": 26900
+    },
+    {
+      "epoch": 0.1947925036374297,
+      "grad_norm": 0.1587022840976715,
+      "learning_rate": 4.80521473502863e-06,
+      "loss": 1.0006,
+      "step": 26910
+    },
+    {
+      "epoch": 0.19486489029801587,
+      "grad_norm": 0.1751098334789276,
+      "learning_rate": 4.805142348368043e-06,
+      "loss": 0.9859,
+      "step": 26920
+    },
+    {
+      "epoch": 0.19493727695860208,
+      "grad_norm": 0.15153715014457703,
+      "learning_rate": 4.805069961707457e-06,
+      "loss": 1.0112,
+      "step": 26930
+    },
+    {
+      "epoch": 0.19500966361918826,
+      "grad_norm": 0.16435284912586212,
+      "learning_rate": 4.8049975750468705e-06,
+      "loss": 1.0057,
+      "step": 26940
+    },
+    {
+      "epoch": 0.19508205027977443,
+      "grad_norm": 0.17554335296154022,
+      "learning_rate": 4.804925188386285e-06,
+      "loss": 1.0051,
+      "step": 26950
+    },
+    {
+      "epoch": 0.19515443694036064,
+      "grad_norm": 0.23302575945854187,
+      "learning_rate": 4.804852801725699e-06,
+      "loss": 1.0029,
+      "step": 26960
+    },
+    {
+      "epoch": 0.1952268236009468,
+      "grad_norm": 0.16776177287101746,
+      "learning_rate": 4.804780415065112e-06,
+      "loss": 0.9986,
+      "step": 26970
+    },
+    {
+      "epoch": 0.19529921026153302,
+      "grad_norm": 0.15875661373138428,
+      "learning_rate": 4.804708028404526e-06,
+      "loss": 0.9944,
+      "step": 26980
+    },
+    {
+      "epoch": 0.1953715969221192,
+      "grad_norm": 0.16342279314994812,
+      "learning_rate": 4.80463564174394e-06,
+      "loss": 1.0014,
+      "step": 26990
+    },
+    {
+      "epoch": 0.19544398358270537,
+      "grad_norm": 0.16927549242973328,
+      "learning_rate": 4.804563255083354e-06,
+      "loss": 1.0005,
+      "step": 27000
+    },
+    {
+      "epoch": 0.19551637024329158,
+      "grad_norm": 0.17107799649238586,
+      "learning_rate": 4.8044908684227675e-06,
+      "loss": 0.9865,
+      "step": 27010
+    },
+    {
+      "epoch": 0.19558875690387775,
+      "grad_norm": 0.156640887260437,
+      "learning_rate": 4.804418481762181e-06,
+      "loss": 0.9957,
+      "step": 27020
+    },
+    {
+      "epoch": 0.19566114356446393,
+      "grad_norm": 0.20385904610157013,
+      "learning_rate": 4.804346095101596e-06,
+      "loss": 0.9996,
+      "step": 27030
+    },
+    {
+      "epoch": 0.19573353022505013,
+      "grad_norm": 0.15625648200511932,
+      "learning_rate": 4.804273708441009e-06,
+      "loss": 1.0155,
+      "step": 27040
+    },
+    {
+      "epoch": 0.1958059168856363,
+      "grad_norm": 0.17145240306854248,
+      "learning_rate": 4.804201321780423e-06,
+      "loss": 0.9947,
+      "step": 27050
+    },
+    {
+      "epoch": 0.19587830354622252,
+      "grad_norm": 0.20977072417736053,
+      "learning_rate": 4.8041289351198364e-06,
+      "loss": 1.0036,
+      "step": 27060
+    },
+    {
+      "epoch": 0.1959506902068087,
+      "grad_norm": 0.163489431142807,
+      "learning_rate": 4.804056548459251e-06,
+      "loss": 1.0106,
+      "step": 27070
+    },
+    {
+      "epoch": 0.19602307686739487,
+      "grad_norm": 0.17504650354385376,
+      "learning_rate": 4.8039841617986645e-06,
+      "loss": 0.9896,
+      "step": 27080
+    },
+    {
+      "epoch": 0.19609546352798107,
+      "grad_norm": 0.16525472700595856,
+      "learning_rate": 4.803911775138078e-06,
+      "loss": 1.0065,
+      "step": 27090
+    },
+    {
+      "epoch": 0.19616785018856725,
+      "grad_norm": 0.20987266302108765,
+      "learning_rate": 4.803839388477492e-06,
+      "loss": 1.0003,
+      "step": 27100
+    },
+    {
+      "epoch": 0.19624023684915343,
+      "grad_norm": 0.17910590767860413,
+      "learning_rate": 4.803767001816906e-06,
+      "loss": 0.9886,
+      "step": 27110
+    },
+    {
+      "epoch": 0.19631262350973963,
+      "grad_norm": 0.17571011185646057,
+      "learning_rate": 4.80369461515632e-06,
+      "loss": 0.9865,
+      "step": 27120
+    },
+    {
+      "epoch": 0.1963850101703258,
+      "grad_norm": 0.1787402182817459,
+      "learning_rate": 4.8036222284957334e-06,
+      "loss": 1.0061,
+      "step": 27130
+    },
+    {
+      "epoch": 0.196457396830912,
+      "grad_norm": 0.18772943317890167,
+      "learning_rate": 4.803549841835147e-06,
+      "loss": 0.9935,
+      "step": 27140
+    },
+    {
+      "epoch": 0.1965297834914982,
+      "grad_norm": 0.17668381333351135,
+      "learning_rate": 4.803477455174561e-06,
+      "loss": 1.008,
+      "step": 27150
+    },
+    {
+      "epoch": 0.19660217015208437,
+      "grad_norm": 0.16147243976593018,
+      "learning_rate": 4.803405068513974e-06,
+      "loss": 0.996,
+      "step": 27160
+    },
+    {
+      "epoch": 0.19667455681267057,
+      "grad_norm": 0.1802648901939392,
+      "learning_rate": 4.803332681853388e-06,
+      "loss": 1.0033,
+      "step": 27170
+    },
+    {
+      "epoch": 0.19674694347325675,
+      "grad_norm": 0.1721193492412567,
+      "learning_rate": 4.803260295192802e-06,
+      "loss": 1.0023,
+      "step": 27180
+    },
+    {
+      "epoch": 0.19681933013384292,
+      "grad_norm": 0.1651720553636551,
+      "learning_rate": 4.803187908532216e-06,
+      "loss": 1.0125,
+      "step": 27190
+    },
+    {
+      "epoch": 0.19689171679442913,
+      "grad_norm": 0.2428152859210968,
+      "learning_rate": 4.80311552187163e-06,
+      "loss": 1.0131,
+      "step": 27200
+    },
+    {
+      "epoch": 0.1969641034550153,
+      "grad_norm": 0.17188434302806854,
+      "learning_rate": 4.803043135211043e-06,
+      "loss": 0.9972,
+      "step": 27210
+    },
+    {
+      "epoch": 0.1970364901156015,
+      "grad_norm": 0.1973094493150711,
+      "learning_rate": 4.802970748550458e-06,
+      "loss": 0.9983,
+      "step": 27220
+    },
+    {
+      "epoch": 0.19710887677618769,
+      "grad_norm": 0.16881392896175385,
+      "learning_rate": 4.802898361889871e-06,
+      "loss": 1.0107,
+      "step": 27230
+    },
+    {
+      "epoch": 0.19718126343677386,
+      "grad_norm": 0.17442533373832703,
+      "learning_rate": 4.802825975229285e-06,
+      "loss": 1.0153,
+      "step": 27240
+    },
+    {
+      "epoch": 0.19725365009736007,
+      "grad_norm": 0.18059037625789642,
+      "learning_rate": 4.8027535885686985e-06,
+      "loss": 1.0168,
+      "step": 27250
+    },
+    {
+      "epoch": 0.19732603675794624,
+      "grad_norm": 0.19683484733104706,
+      "learning_rate": 4.802681201908113e-06,
+      "loss": 0.9942,
+      "step": 27260
+    },
+    {
+      "epoch": 0.19739842341853242,
+      "grad_norm": 0.1710311323404312,
+      "learning_rate": 4.802608815247527e-06,
+      "loss": 1.0155,
+      "step": 27270
+    },
+    {
+      "epoch": 0.19747081007911862,
+      "grad_norm": 0.16936126351356506,
+      "learning_rate": 4.80253642858694e-06,
+      "loss": 0.9969,
+      "step": 27280
+    },
+    {
+      "epoch": 0.1975431967397048,
+      "grad_norm": 0.19875222444534302,
+      "learning_rate": 4.802464041926354e-06,
+      "loss": 0.9913,
+      "step": 27290
+    },
+    {
+      "epoch": 0.197615583400291,
+      "grad_norm": 0.16911375522613525,
+      "learning_rate": 4.802391655265768e-06,
+      "loss": 1.0025,
+      "step": 27300
+    },
+    {
+      "epoch": 0.19768797006087718,
+      "grad_norm": 0.16510553658008575,
+      "learning_rate": 4.802319268605182e-06,
+      "loss": 1.0083,
+      "step": 27310
+    },
+    {
+      "epoch": 0.19776035672146336,
+      "grad_norm": 0.17663301527500153,
+      "learning_rate": 4.8022468819445955e-06,
+      "loss": 0.9859,
+      "step": 27320
+    },
+    {
+      "epoch": 0.19783274338204956,
+      "grad_norm": 0.18130449950695038,
+      "learning_rate": 4.802174495284009e-06,
+      "loss": 1.0023,
+      "step": 27330
+    },
+    {
+      "epoch": 0.19790513004263574,
+      "grad_norm": 0.1663258969783783,
+      "learning_rate": 4.802102108623423e-06,
+      "loss": 1.0032,
+      "step": 27340
+    },
+    {
+      "epoch": 0.19797751670322192,
+      "grad_norm": 0.17057499289512634,
+      "learning_rate": 4.802029721962837e-06,
+      "loss": 0.9985,
+      "step": 27350
+    },
+    {
+      "epoch": 0.19804990336380812,
+      "grad_norm": 0.1627529412508011,
+      "learning_rate": 4.801957335302251e-06,
+      "loss": 1.021,
+      "step": 27360
+    },
+    {
+      "epoch": 0.1981222900243943,
+      "grad_norm": 0.1599203497171402,
+      "learning_rate": 4.8018849486416645e-06,
+      "loss": 1.0064,
+      "step": 27370
+    },
+    {
+      "epoch": 0.1981946766849805,
+      "grad_norm": 0.199107825756073,
+      "learning_rate": 4.801812561981078e-06,
+      "loss": 1.0142,
+      "step": 27380
+    },
+    {
+      "epoch": 0.19826706334556668,
+      "grad_norm": 0.19576513767242432,
+      "learning_rate": 4.8017401753204926e-06,
+      "loss": 1.0037,
+      "step": 27390
+    },
+    {
+      "epoch": 0.19833945000615286,
+      "grad_norm": 0.16550497710704803,
+      "learning_rate": 4.801667788659906e-06,
+      "loss": 0.9853,
+      "step": 27400
+    },
+    {
+      "epoch": 0.19841183666673906,
+      "grad_norm": 0.17709434032440186,
+      "learning_rate": 4.80159540199932e-06,
+      "loss": 1.0031,
+      "step": 27410
+    },
+    {
+      "epoch": 0.19848422332732524,
+      "grad_norm": 0.1671217530965805,
+      "learning_rate": 4.801523015338733e-06,
+      "loss": 0.9927,
+      "step": 27420
+    },
+    {
+      "epoch": 0.19855660998791141,
+      "grad_norm": 0.16205404698848724,
+      "learning_rate": 4.801450628678148e-06,
+      "loss": 1.004,
+      "step": 27430
+    },
+    {
+      "epoch": 0.19862899664849762,
+      "grad_norm": 0.21994493901729584,
+      "learning_rate": 4.8013782420175615e-06,
+      "loss": 0.9901,
+      "step": 27440
+    },
+    {
+      "epoch": 0.1987013833090838,
+      "grad_norm": 0.17155443131923676,
+      "learning_rate": 4.801305855356975e-06,
+      "loss": 1.011,
+      "step": 27450
+    },
+    {
+      "epoch": 0.19877376996967,
+      "grad_norm": 0.1630110889673233,
+      "learning_rate": 4.801233468696389e-06,
+      "loss": 0.9974,
+      "step": 27460
+    },
+    {
+      "epoch": 0.19884615663025618,
+      "grad_norm": 0.17589977383613586,
+      "learning_rate": 4.801161082035803e-06,
+      "loss": 0.9816,
+      "step": 27470
+    },
+    {
+      "epoch": 0.19891854329084235,
+      "grad_norm": 0.16798698902130127,
+      "learning_rate": 4.801088695375217e-06,
+      "loss": 1.0086,
+      "step": 27480
+    },
+    {
+      "epoch": 0.19899092995142856,
+      "grad_norm": 0.18393242359161377,
+      "learning_rate": 4.80101630871463e-06,
+      "loss": 1.01,
+      "step": 27490
+    },
+    {
+      "epoch": 0.19906331661201473,
+      "grad_norm": 0.1627999097108841,
+      "learning_rate": 4.800943922054044e-06,
+      "loss": 1.0033,
+      "step": 27500
+    },
+    {
+      "epoch": 0.1991357032726009,
+      "grad_norm": 0.1707964688539505,
+      "learning_rate": 4.8008715353934585e-06,
+      "loss": 0.9966,
+      "step": 27510
+    },
+    {
+      "epoch": 0.19920808993318712,
+      "grad_norm": 0.154254749417305,
+      "learning_rate": 4.800799148732872e-06,
+      "loss": 0.9943,
+      "step": 27520
+    },
+    {
+      "epoch": 0.1992804765937733,
+      "grad_norm": 0.17467810213565826,
+      "learning_rate": 4.800726762072286e-06,
+      "loss": 0.9986,
+      "step": 27530
+    },
+    {
+      "epoch": 0.1993528632543595,
+      "grad_norm": 0.17175132036209106,
+      "learning_rate": 4.800654375411699e-06,
+      "loss": 1.0,
+      "step": 27540
+    },
+    {
+      "epoch": 0.19942524991494567,
+      "grad_norm": 0.17211025953292847,
+      "learning_rate": 4.800581988751114e-06,
+      "loss": 1.011,
+      "step": 27550
+    },
+    {
+      "epoch": 0.19949763657553185,
+      "grad_norm": 0.16861172020435333,
+      "learning_rate": 4.800509602090527e-06,
+      "loss": 1.0082,
+      "step": 27560
+    },
+    {
+      "epoch": 0.19957002323611805,
+      "grad_norm": 0.20472027361392975,
+      "learning_rate": 4.800437215429941e-06,
+      "loss": 0.9906,
+      "step": 27570
+    },
+    {
+      "epoch": 0.19964240989670423,
+      "grad_norm": 0.16416484117507935,
+      "learning_rate": 4.800364828769355e-06,
+      "loss": 0.9934,
+      "step": 27580
+    },
+    {
+      "epoch": 0.19971479655729044,
+      "grad_norm": 0.16409383714199066,
+      "learning_rate": 4.800292442108769e-06,
+      "loss": 0.9885,
+      "step": 27590
+    },
+    {
+      "epoch": 0.1997871832178766,
+      "grad_norm": 0.17067715525627136,
+      "learning_rate": 4.800220055448183e-06,
+      "loss": 0.9954,
+      "step": 27600
+    },
+    {
+      "epoch": 0.1998595698784628,
+      "grad_norm": 0.18361663818359375,
+      "learning_rate": 4.800147668787596e-06,
+      "loss": 1.0033,
+      "step": 27610
+    },
+    {
+      "epoch": 0.199931956539049,
+      "grad_norm": 0.1660345494747162,
+      "learning_rate": 4.80007528212701e-06,
+      "loss": 1.0032,
+      "step": 27620
+    },
+    {
+      "epoch": 0.20000434319963517,
+      "grad_norm": 0.16931919753551483,
+      "learning_rate": 4.8000028954664244e-06,
+      "loss": 1.0006,
+      "step": 27630
+    },
+    {
+      "epoch": 0.20007672986022135,
+      "grad_norm": 0.1660221815109253,
+      "learning_rate": 4.799930508805838e-06,
+      "loss": 1.0126,
+      "step": 27640
+    },
+    {
+      "epoch": 0.20014911652080755,
+      "grad_norm": 0.179294615983963,
+      "learning_rate": 4.799858122145252e-06,
+      "loss": 1.0143,
+      "step": 27650
+    },
+    {
+      "epoch": 0.20022150318139373,
+      "grad_norm": 0.19794364273548126,
+      "learning_rate": 4.799785735484665e-06,
+      "loss": 1.0057,
+      "step": 27660
+    },
+    {
+      "epoch": 0.20029388984197993,
+      "grad_norm": 0.1863810420036316,
+      "learning_rate": 4.79971334882408e-06,
+      "loss": 0.9972,
+      "step": 27670
+    },
+    {
+      "epoch": 0.2003662765025661,
+      "grad_norm": 0.17206966876983643,
+      "learning_rate": 4.799640962163493e-06,
+      "loss": 1.0025,
+      "step": 27680
+    },
+    {
+      "epoch": 0.2004386631631523,
+      "grad_norm": 0.18598827719688416,
+      "learning_rate": 4.799568575502906e-06,
+      "loss": 1.0035,
+      "step": 27690
+    },
+    {
+      "epoch": 0.2005110498237385,
+      "grad_norm": 0.17448210716247559,
+      "learning_rate": 4.799496188842321e-06,
+      "loss": 1.0121,
+      "step": 27700
+    },
+    {
+      "epoch": 0.20058343648432467,
+      "grad_norm": 0.1639571338891983,
+      "learning_rate": 4.799423802181734e-06,
+      "loss": 1.0005,
+      "step": 27710
+    },
+    {
+      "epoch": 0.20065582314491084,
+      "grad_norm": 0.16156673431396484,
+      "learning_rate": 4.799351415521148e-06,
+      "loss": 0.9937,
+      "step": 27720
+    },
+    {
+      "epoch": 0.20072820980549705,
+      "grad_norm": 0.1673559546470642,
+      "learning_rate": 4.7992790288605614e-06,
+      "loss": 0.9872,
+      "step": 27730
+    },
+    {
+      "epoch": 0.20080059646608323,
+      "grad_norm": 0.16240374743938446,
+      "learning_rate": 4.799206642199976e-06,
+      "loss": 1.0194,
+      "step": 27740
+    },
+    {
+      "epoch": 0.20087298312666943,
+      "grad_norm": 0.16212816536426544,
+      "learning_rate": 4.7991342555393895e-06,
+      "loss": 0.9928,
+      "step": 27750
+    },
+    {
+      "epoch": 0.2009453697872556,
+      "grad_norm": 0.15658411383628845,
+      "learning_rate": 4.799061868878803e-06,
+      "loss": 0.9976,
+      "step": 27760
+    },
+    {
+      "epoch": 0.20101775644784178,
+      "grad_norm": 0.16336016356945038,
+      "learning_rate": 4.798989482218217e-06,
+      "loss": 1.0189,
+      "step": 27770
+    },
+    {
+      "epoch": 0.201090143108428,
+      "grad_norm": 0.18533754348754883,
+      "learning_rate": 4.798917095557631e-06,
+      "loss": 0.9867,
+      "step": 27780
+    },
+    {
+      "epoch": 0.20116252976901416,
+      "grad_norm": 0.1857583224773407,
+      "learning_rate": 4.798844708897045e-06,
+      "loss": 0.9994,
+      "step": 27790
+    },
+    {
+      "epoch": 0.20123491642960034,
+      "grad_norm": 0.19050684571266174,
+      "learning_rate": 4.7987723222364584e-06,
+      "loss": 0.997,
+      "step": 27800
+    },
+    {
+      "epoch": 0.20130730309018655,
+      "grad_norm": 0.17281419038772583,
+      "learning_rate": 4.798699935575872e-06,
+      "loss": 1.0018,
+      "step": 27810
+    },
+    {
+      "epoch": 0.20137968975077272,
+      "grad_norm": 0.169466033577919,
+      "learning_rate": 4.7986275489152865e-06,
+      "loss": 0.997,
+      "step": 27820
+    },
+    {
+      "epoch": 0.20145207641135893,
+      "grad_norm": 0.16908937692642212,
+      "learning_rate": 4.7985551622547e-06,
+      "loss": 1.0002,
+      "step": 27830
+    },
+    {
+      "epoch": 0.2015244630719451,
+      "grad_norm": 0.17299243807792664,
+      "learning_rate": 4.798482775594114e-06,
+      "loss": 1.0038,
+      "step": 27840
+    },
+    {
+      "epoch": 0.20159684973253128,
+      "grad_norm": 0.16563472151756287,
+      "learning_rate": 4.798410388933527e-06,
+      "loss": 0.9915,
+      "step": 27850
+    },
+    {
+      "epoch": 0.20166923639311748,
+      "grad_norm": 0.19840186834335327,
+      "learning_rate": 4.798338002272942e-06,
+      "loss": 1.0037,
+      "step": 27860
+    },
+    {
+      "epoch": 0.20174162305370366,
+      "grad_norm": 0.17237645387649536,
+      "learning_rate": 4.7982656156123554e-06,
+      "loss": 1.012,
+      "step": 27870
+    },
+    {
+      "epoch": 0.20181400971428984,
+      "grad_norm": 0.18129993975162506,
+      "learning_rate": 4.798193228951769e-06,
+      "loss": 1.0009,
+      "step": 27880
+    },
+    {
+      "epoch": 0.20188639637487604,
+      "grad_norm": 0.16735847294330597,
+      "learning_rate": 4.798120842291183e-06,
+      "loss": 1.0198,
+      "step": 27890
+    },
+    {
+      "epoch": 0.20195878303546222,
+      "grad_norm": 0.16451966762542725,
+      "learning_rate": 4.798048455630597e-06,
+      "loss": 1.0038,
+      "step": 27900
+    },
+    {
+      "epoch": 0.20203116969604842,
+      "grad_norm": 0.1658269166946411,
+      "learning_rate": 4.797976068970011e-06,
+      "loss": 0.9934,
+      "step": 27910
+    },
+    {
+      "epoch": 0.2021035563566346,
+      "grad_norm": 0.17041842639446259,
+      "learning_rate": 4.797903682309424e-06,
+      "loss": 0.9873,
+      "step": 27920
+    },
+    {
+      "epoch": 0.20217594301722078,
+      "grad_norm": 0.18290521204471588,
+      "learning_rate": 4.797831295648838e-06,
+      "loss": 0.9885,
+      "step": 27930
+    },
+    {
+      "epoch": 0.20224832967780698,
+      "grad_norm": 0.173233300447464,
+      "learning_rate": 4.797758908988252e-06,
+      "loss": 0.9977,
+      "step": 27940
+    },
+    {
+      "epoch": 0.20232071633839316,
+      "grad_norm": 0.1651332825422287,
+      "learning_rate": 4.797686522327666e-06,
+      "loss": 0.9989,
+      "step": 27950
+    },
+    {
+      "epoch": 0.20239310299897934,
+      "grad_norm": 0.18143868446350098,
+      "learning_rate": 4.79761413566708e-06,
+      "loss": 1.0013,
+      "step": 27960
+    },
+    {
+      "epoch": 0.20246548965956554,
+      "grad_norm": 0.18052855134010315,
+      "learning_rate": 4.797541749006493e-06,
+      "loss": 1.0127,
+      "step": 27970
+    },
+    {
+      "epoch": 0.20253787632015172,
+      "grad_norm": 0.1647488921880722,
+      "learning_rate": 4.797469362345907e-06,
+      "loss": 0.9766,
+      "step": 27980
+    },
+    {
+      "epoch": 0.20261026298073792,
+      "grad_norm": 0.18632908165454865,
+      "learning_rate": 4.797396975685321e-06,
+      "loss": 0.9951,
+      "step": 27990
+    },
+    {
+      "epoch": 0.2026826496413241,
+      "grad_norm": 0.1738438606262207,
+      "learning_rate": 4.797324589024735e-06,
+      "loss": 1.0055,
+      "step": 28000
+    },
+    {
+      "epoch": 0.20275503630191027,
+      "grad_norm": 0.17190273106098175,
+      "learning_rate": 4.797252202364149e-06,
+      "loss": 1.003,
+      "step": 28010
+    },
+    {
+      "epoch": 0.20282742296249648,
+      "grad_norm": 0.16988669335842133,
+      "learning_rate": 4.797179815703562e-06,
+      "loss": 0.9932,
+      "step": 28020
+    },
+    {
+      "epoch": 0.20289980962308266,
+      "grad_norm": 0.16658180952072144,
+      "learning_rate": 4.797107429042977e-06,
+      "loss": 1.0075,
+      "step": 28030
+    },
+    {
+      "epoch": 0.20297219628366883,
+      "grad_norm": 0.16553157567977905,
+      "learning_rate": 4.79703504238239e-06,
+      "loss": 1.0056,
+      "step": 28040
+    },
+    {
+      "epoch": 0.20304458294425504,
+      "grad_norm": 0.17701905965805054,
+      "learning_rate": 4.796962655721804e-06,
+      "loss": 1.0115,
+      "step": 28050
+    },
+    {
+      "epoch": 0.2031169696048412,
+      "grad_norm": 0.16136251389980316,
+      "learning_rate": 4.7968902690612175e-06,
+      "loss": 0.9909,
+      "step": 28060
+    },
+    {
+      "epoch": 0.20318935626542742,
+      "grad_norm": 0.17143204808235168,
+      "learning_rate": 4.796817882400632e-06,
+      "loss": 0.9944,
+      "step": 28070
+    },
+    {
+      "epoch": 0.2032617429260136,
+      "grad_norm": 0.16544604301452637,
+      "learning_rate": 4.796745495740046e-06,
+      "loss": 1.0009,
+      "step": 28080
+    },
+    {
+      "epoch": 0.20333412958659977,
+      "grad_norm": 0.1553160548210144,
+      "learning_rate": 4.796673109079459e-06,
+      "loss": 1.0123,
+      "step": 28090
+    },
+    {
+      "epoch": 0.20340651624718598,
+      "grad_norm": 0.16683773696422577,
+      "learning_rate": 4.796600722418873e-06,
+      "loss": 1.0095,
+      "step": 28100
+    },
+    {
+      "epoch": 0.20347890290777215,
+      "grad_norm": 0.16187171638011932,
+      "learning_rate": 4.796528335758287e-06,
+      "loss": 0.9973,
+      "step": 28110
+    },
+    {
+      "epoch": 0.20355128956835836,
+      "grad_norm": 0.1778879463672638,
+      "learning_rate": 4.796455949097701e-06,
+      "loss": 1.012,
+      "step": 28120
+    },
+    {
+      "epoch": 0.20362367622894453,
+      "grad_norm": 0.1824272871017456,
+      "learning_rate": 4.7963835624371146e-06,
+      "loss": 0.9874,
+      "step": 28130
+    },
+    {
+      "epoch": 0.2036960628895307,
+      "grad_norm": 0.15840831398963928,
+      "learning_rate": 4.796311175776528e-06,
+      "loss": 1.0078,
+      "step": 28140
+    },
+    {
+      "epoch": 0.20376844955011691,
+      "grad_norm": 0.1720242202281952,
+      "learning_rate": 4.796238789115943e-06,
+      "loss": 0.9946,
+      "step": 28150
+    },
+    {
+      "epoch": 0.2038408362107031,
+      "grad_norm": 0.18525436520576477,
+      "learning_rate": 4.796166402455356e-06,
+      "loss": 0.9854,
+      "step": 28160
+    },
+    {
+      "epoch": 0.20391322287128927,
+      "grad_norm": 0.16889312863349915,
+      "learning_rate": 4.79609401579477e-06,
+      "loss": 0.9788,
+      "step": 28170
+    },
+    {
+      "epoch": 0.20398560953187547,
+      "grad_norm": 0.1737358719110489,
+      "learning_rate": 4.7960216291341835e-06,
+      "loss": 1.008,
+      "step": 28180
+    },
+    {
+      "epoch": 0.20405799619246165,
+      "grad_norm": 0.1770784854888916,
+      "learning_rate": 4.795949242473598e-06,
+      "loss": 0.9884,
+      "step": 28190
+    },
+    {
+      "epoch": 0.20413038285304785,
+      "grad_norm": 0.17560461163520813,
+      "learning_rate": 4.7958768558130116e-06,
+      "loss": 0.9923,
+      "step": 28200
+    },
+    {
+      "epoch": 0.20420276951363403,
+      "grad_norm": 0.210534930229187,
+      "learning_rate": 4.795804469152425e-06,
+      "loss": 1.0002,
+      "step": 28210
+    },
+    {
+      "epoch": 0.2042751561742202,
+      "grad_norm": 0.1665802001953125,
+      "learning_rate": 4.795732082491839e-06,
+      "loss": 0.9887,
+      "step": 28220
+    },
+    {
+      "epoch": 0.2043475428348064,
+      "grad_norm": 0.17272193729877472,
+      "learning_rate": 4.795659695831252e-06,
+      "loss": 1.007,
+      "step": 28230
+    },
+    {
+      "epoch": 0.2044199294953926,
+      "grad_norm": 0.16487373411655426,
+      "learning_rate": 4.795587309170666e-06,
+      "loss": 1.0067,
+      "step": 28240
+    },
+    {
+      "epoch": 0.20449231615597876,
+      "grad_norm": 0.16630111634731293,
+      "learning_rate": 4.79551492251008e-06,
+      "loss": 0.9795,
+      "step": 28250
+    },
+    {
+      "epoch": 0.20456470281656497,
+      "grad_norm": 0.17112283408641815,
+      "learning_rate": 4.795442535849494e-06,
+      "loss": 1.001,
+      "step": 28260
+    },
+    {
+      "epoch": 0.20463708947715115,
+      "grad_norm": 0.17622940242290497,
+      "learning_rate": 4.795370149188908e-06,
+      "loss": 0.9859,
+      "step": 28270
+    },
+    {
+      "epoch": 0.20470947613773735,
+      "grad_norm": 0.16943074762821198,
+      "learning_rate": 4.795297762528321e-06,
+      "loss": 1.0008,
+      "step": 28280
+    },
+    {
+      "epoch": 0.20478186279832353,
+      "grad_norm": 0.16673767566680908,
+      "learning_rate": 4.795225375867735e-06,
+      "loss": 0.9991,
+      "step": 28290
+    },
+    {
+      "epoch": 0.2048542494589097,
+      "grad_norm": 0.16490839421749115,
+      "learning_rate": 4.795152989207149e-06,
+      "loss": 1.0065,
+      "step": 28300
+    },
+    {
+      "epoch": 0.2049266361194959,
+      "grad_norm": 0.16834768652915955,
+      "learning_rate": 4.795080602546563e-06,
+      "loss": 1.0002,
+      "step": 28310
+    },
+    {
+      "epoch": 0.20499902278008209,
+      "grad_norm": 0.15660522878170013,
+      "learning_rate": 4.795008215885977e-06,
+      "loss": 1.008,
+      "step": 28320
+    },
+    {
+      "epoch": 0.20507140944066826,
+      "grad_norm": 0.1734970211982727,
+      "learning_rate": 4.79493582922539e-06,
+      "loss": 0.9975,
+      "step": 28330
+    },
+    {
+      "epoch": 0.20514379610125447,
+      "grad_norm": 0.17806610465049744,
+      "learning_rate": 4.794863442564805e-06,
+      "loss": 0.9932,
+      "step": 28340
+    },
+    {
+      "epoch": 0.20521618276184064,
+      "grad_norm": 0.1722010225057602,
+      "learning_rate": 4.794791055904218e-06,
+      "loss": 0.9841,
+      "step": 28350
+    },
+    {
+      "epoch": 0.20528856942242685,
+      "grad_norm": 0.1715139001607895,
+      "learning_rate": 4.794718669243632e-06,
+      "loss": 0.9977,
+      "step": 28360
+    },
+    {
+      "epoch": 0.20536095608301302,
+      "grad_norm": 0.17599177360534668,
+      "learning_rate": 4.7946462825830456e-06,
+      "loss": 1.0029,
+      "step": 28370
+    },
+    {
+      "epoch": 0.2054333427435992,
+      "grad_norm": 0.16538375616073608,
+      "learning_rate": 4.79457389592246e-06,
+      "loss": 0.999,
+      "step": 28380
+    },
+    {
+      "epoch": 0.2055057294041854,
+      "grad_norm": 0.1635717749595642,
+      "learning_rate": 4.794501509261874e-06,
+      "loss": 1.0158,
+      "step": 28390
+    },
+    {
+      "epoch": 0.20557811606477158,
+      "grad_norm": 0.17137548327445984,
+      "learning_rate": 4.794429122601287e-06,
+      "loss": 1.0063,
+      "step": 28400
+    },
+    {
+      "epoch": 0.20565050272535776,
+      "grad_norm": 0.26129329204559326,
+      "learning_rate": 4.794356735940701e-06,
+      "loss": 1.0029,
+      "step": 28410
+    },
+    {
+      "epoch": 0.20572288938594396,
+      "grad_norm": 0.19296756386756897,
+      "learning_rate": 4.794284349280115e-06,
+      "loss": 0.9955,
+      "step": 28420
+    },
+    {
+      "epoch": 0.20579527604653014,
+      "grad_norm": 0.17898808419704437,
+      "learning_rate": 4.794211962619529e-06,
+      "loss": 0.9865,
+      "step": 28430
+    },
+    {
+      "epoch": 0.20586766270711634,
+      "grad_norm": 0.16566985845565796,
+      "learning_rate": 4.794139575958943e-06,
+      "loss": 0.9895,
+      "step": 28440
+    },
+    {
+      "epoch": 0.20594004936770252,
+      "grad_norm": 0.1738710254430771,
+      "learning_rate": 4.794067189298356e-06,
+      "loss": 1.0066,
+      "step": 28450
+    },
+    {
+      "epoch": 0.2060124360282887,
+      "grad_norm": 0.18403783440589905,
+      "learning_rate": 4.793994802637771e-06,
+      "loss": 0.9924,
+      "step": 28460
+    },
+    {
+      "epoch": 0.2060848226888749,
+      "grad_norm": 0.1839311420917511,
+      "learning_rate": 4.793922415977184e-06,
+      "loss": 0.9973,
+      "step": 28470
+    },
+    {
+      "epoch": 0.20615720934946108,
+      "grad_norm": 0.17490077018737793,
+      "learning_rate": 4.793850029316598e-06,
+      "loss": 1.0049,
+      "step": 28480
+    },
+    {
+      "epoch": 0.20622959601004726,
+      "grad_norm": 0.17090052366256714,
+      "learning_rate": 4.7937776426560115e-06,
+      "loss": 0.987,
+      "step": 28490
+    },
+    {
+      "epoch": 0.20630198267063346,
+      "grad_norm": 0.1909414380788803,
+      "learning_rate": 4.793705255995426e-06,
+      "loss": 1.0039,
+      "step": 28500
+    },
+    {
+      "epoch": 0.20637436933121964,
+      "grad_norm": 0.18218588829040527,
+      "learning_rate": 4.79363286933484e-06,
+      "loss": 1.0067,
+      "step": 28510
+    },
+    {
+      "epoch": 0.20644675599180584,
+      "grad_norm": 0.2051839381456375,
+      "learning_rate": 4.793560482674253e-06,
+      "loss": 0.9922,
+      "step": 28520
+    },
+    {
+      "epoch": 0.20651914265239202,
+      "grad_norm": 0.17690230906009674,
+      "learning_rate": 4.793488096013667e-06,
+      "loss": 0.9973,
+      "step": 28530
+    },
+    {
+      "epoch": 0.2065915293129782,
+      "grad_norm": 0.32738569378852844,
+      "learning_rate": 4.793415709353081e-06,
+      "loss": 0.998,
+      "step": 28540
+    },
+    {
+      "epoch": 0.2066639159735644,
+      "grad_norm": 0.16709066927433014,
+      "learning_rate": 4.793343322692495e-06,
+      "loss": 0.9968,
+      "step": 28550
+    },
+    {
+      "epoch": 0.20673630263415058,
+      "grad_norm": 0.16835230588912964,
+      "learning_rate": 4.7932709360319085e-06,
+      "loss": 0.9936,
+      "step": 28560
+    },
+    {
+      "epoch": 0.20680868929473675,
+      "grad_norm": 0.18569424748420715,
+      "learning_rate": 4.793198549371322e-06,
+      "loss": 0.9828,
+      "step": 28570
+    },
+    {
+      "epoch": 0.20688107595532296,
+      "grad_norm": 0.16183647513389587,
+      "learning_rate": 4.793126162710736e-06,
+      "loss": 1.0112,
+      "step": 28580
+    },
+    {
+      "epoch": 0.20695346261590913,
+      "grad_norm": 0.1736779808998108,
+      "learning_rate": 4.79305377605015e-06,
+      "loss": 0.9918,
+      "step": 28590
+    },
+    {
+      "epoch": 0.20702584927649534,
+      "grad_norm": 0.6868466138839722,
+      "learning_rate": 4.792981389389564e-06,
+      "loss": 0.9918,
+      "step": 28600
+    },
+    {
+      "epoch": 0.20709823593708152,
+      "grad_norm": 0.1731596291065216,
+      "learning_rate": 4.7929090027289774e-06,
+      "loss": 0.9935,
+      "step": 28610
+    },
+    {
+      "epoch": 0.2071706225976677,
+      "grad_norm": 0.18018291890621185,
+      "learning_rate": 4.792836616068391e-06,
+      "loss": 1.0124,
+      "step": 28620
+    },
+    {
+      "epoch": 0.2072430092582539,
+      "grad_norm": 0.1815289855003357,
+      "learning_rate": 4.7927642294078055e-06,
+      "loss": 1.0116,
+      "step": 28630
+    },
+    {
+      "epoch": 0.20731539591884007,
+      "grad_norm": 0.18109048902988434,
+      "learning_rate": 4.792691842747219e-06,
+      "loss": 0.9968,
+      "step": 28640
+    },
+    {
+      "epoch": 0.20738778257942625,
+      "grad_norm": 0.1723286211490631,
+      "learning_rate": 4.792619456086633e-06,
+      "loss": 1.0088,
+      "step": 28650
+    },
+    {
+      "epoch": 0.20746016924001245,
+      "grad_norm": 0.17731237411499023,
+      "learning_rate": 4.792547069426046e-06,
+      "loss": 1.0042,
+      "step": 28660
+    },
+    {
+      "epoch": 0.20753255590059863,
+      "grad_norm": 0.17550349235534668,
+      "learning_rate": 4.792474682765461e-06,
+      "loss": 1.0046,
+      "step": 28670
+    },
+    {
+      "epoch": 0.20760494256118484,
+      "grad_norm": 0.17998819053173065,
+      "learning_rate": 4.7924022961048745e-06,
+      "loss": 0.9919,
+      "step": 28680
+    },
+    {
+      "epoch": 0.207677329221771,
+      "grad_norm": 0.1708000898361206,
+      "learning_rate": 4.792329909444288e-06,
+      "loss": 1.0019,
+      "step": 28690
+    },
+    {
+      "epoch": 0.2077497158823572,
+      "grad_norm": 0.16459348797798157,
+      "learning_rate": 4.792257522783702e-06,
+      "loss": 1.0008,
+      "step": 28700
+    },
+    {
+      "epoch": 0.2078221025429434,
+      "grad_norm": 0.17152118682861328,
+      "learning_rate": 4.792185136123116e-06,
+      "loss": 0.9919,
+      "step": 28710
+    },
+    {
+      "epoch": 0.20789448920352957,
+      "grad_norm": 0.1598610132932663,
+      "learning_rate": 4.79211274946253e-06,
+      "loss": 1.0093,
+      "step": 28720
+    },
+    {
+      "epoch": 0.20796687586411577,
+      "grad_norm": 0.18231916427612305,
+      "learning_rate": 4.792040362801943e-06,
+      "loss": 1.0098,
+      "step": 28730
+    },
+    {
+      "epoch": 0.20803926252470195,
+      "grad_norm": 0.16905340552330017,
+      "learning_rate": 4.791967976141357e-06,
+      "loss": 0.992,
+      "step": 28740
+    },
+    {
+      "epoch": 0.20811164918528813,
+      "grad_norm": 0.1717272698879242,
+      "learning_rate": 4.791895589480771e-06,
+      "loss": 0.9876,
+      "step": 28750
+    },
+    {
+      "epoch": 0.20818403584587433,
+      "grad_norm": 0.16113172471523285,
+      "learning_rate": 4.791823202820184e-06,
+      "loss": 0.9987,
+      "step": 28760
+    },
+    {
+      "epoch": 0.2082564225064605,
+      "grad_norm": 0.18557120859622955,
+      "learning_rate": 4.791750816159598e-06,
+      "loss": 1.0109,
+      "step": 28770
+    },
+    {
+      "epoch": 0.20832880916704669,
+      "grad_norm": 0.17728063464164734,
+      "learning_rate": 4.791678429499012e-06,
+      "loss": 0.9922,
+      "step": 28780
+    },
+    {
+      "epoch": 0.2084011958276329,
+      "grad_norm": 0.19753523170948029,
+      "learning_rate": 4.791606042838426e-06,
+      "loss": 1.0016,
+      "step": 28790
+    },
+    {
+      "epoch": 0.20847358248821907,
+      "grad_norm": 0.16373999416828156,
+      "learning_rate": 4.7915336561778395e-06,
+      "loss": 0.9975,
+      "step": 28800
+    },
+    {
+      "epoch": 0.20854596914880527,
+      "grad_norm": 0.17161045968532562,
+      "learning_rate": 4.791461269517253e-06,
+      "loss": 0.9926,
+      "step": 28810
+    },
+    {
+      "epoch": 0.20861835580939145,
+      "grad_norm": 0.16987286508083344,
+      "learning_rate": 4.791388882856668e-06,
+      "loss": 0.9974,
+      "step": 28820
+    },
+    {
+      "epoch": 0.20869074246997762,
+      "grad_norm": 0.2174030840396881,
+      "learning_rate": 4.791316496196081e-06,
+      "loss": 1.003,
+      "step": 28830
+    },
+    {
+      "epoch": 0.20876312913056383,
+      "grad_norm": 0.1620844602584839,
+      "learning_rate": 4.791244109535495e-06,
+      "loss": 1.0006,
+      "step": 28840
+    },
+    {
+      "epoch": 0.20883551579115,
+      "grad_norm": 0.18147195875644684,
+      "learning_rate": 4.7911717228749085e-06,
+      "loss": 1.0035,
+      "step": 28850
+    },
+    {
+      "epoch": 0.20890790245173618,
+      "grad_norm": 0.1799267679452896,
+      "learning_rate": 4.791099336214323e-06,
+      "loss": 1.0044,
+      "step": 28860
+    },
+    {
+      "epoch": 0.2089802891123224,
+      "grad_norm": 0.2170770764350891,
+      "learning_rate": 4.7910269495537366e-06,
+      "loss": 0.9663,
+      "step": 28870
+    },
+    {
+      "epoch": 0.20905267577290856,
+      "grad_norm": 0.17900538444519043,
+      "learning_rate": 4.79095456289315e-06,
+      "loss": 0.99,
+      "step": 28880
+    },
+    {
+      "epoch": 0.20912506243349477,
+      "grad_norm": 0.1686086356639862,
+      "learning_rate": 4.790882176232564e-06,
+      "loss": 1.0001,
+      "step": 28890
+    },
+    {
+      "epoch": 0.20919744909408095,
+      "grad_norm": 0.19090880453586578,
+      "learning_rate": 4.790809789571978e-06,
+      "loss": 1.0074,
+      "step": 28900
+    },
+    {
+      "epoch": 0.20926983575466712,
+      "grad_norm": 0.16574802994728088,
+      "learning_rate": 4.790737402911392e-06,
+      "loss": 0.9998,
+      "step": 28910
+    },
+    {
+      "epoch": 0.20934222241525333,
+      "grad_norm": 0.17160232365131378,
+      "learning_rate": 4.7906650162508055e-06,
+      "loss": 1.0036,
+      "step": 28920
+    },
+    {
+      "epoch": 0.2094146090758395,
+      "grad_norm": 0.17954379320144653,
+      "learning_rate": 4.790592629590219e-06,
+      "loss": 0.9932,
+      "step": 28930
+    },
+    {
+      "epoch": 0.20948699573642568,
+      "grad_norm": 0.17119750380516052,
+      "learning_rate": 4.7905202429296336e-06,
+      "loss": 0.9804,
+      "step": 28940
+    },
+    {
+      "epoch": 0.20955938239701188,
+      "grad_norm": 0.16228660941123962,
+      "learning_rate": 4.790447856269047e-06,
+      "loss": 1.0078,
+      "step": 28950
+    },
+    {
+      "epoch": 0.20963176905759806,
+      "grad_norm": 0.16871187090873718,
+      "learning_rate": 4.790375469608461e-06,
+      "loss": 1.0053,
+      "step": 28960
+    },
+    {
+      "epoch": 0.20970415571818427,
+      "grad_norm": 0.1674191802740097,
+      "learning_rate": 4.790303082947874e-06,
+      "loss": 1.0003,
+      "step": 28970
+    },
+    {
+      "epoch": 0.20977654237877044,
+      "grad_norm": 0.16644033789634705,
+      "learning_rate": 4.790230696287289e-06,
+      "loss": 0.9774,
+      "step": 28980
+    },
+    {
+      "epoch": 0.20984892903935662,
+      "grad_norm": 0.17086544632911682,
+      "learning_rate": 4.7901583096267025e-06,
+      "loss": 1.0096,
+      "step": 28990
+    },
+    {
+      "epoch": 0.20992131569994282,
+      "grad_norm": 0.16685600578784943,
+      "learning_rate": 4.790085922966116e-06,
+      "loss": 0.9982,
+      "step": 29000
+    },
+    {
+      "epoch": 0.209993702360529,
+      "grad_norm": 0.16532373428344727,
+      "learning_rate": 4.79001353630553e-06,
+      "loss": 0.976,
+      "step": 29010
+    },
+    {
+      "epoch": 0.21006608902111518,
+      "grad_norm": 0.36561527848243713,
+      "learning_rate": 4.789941149644944e-06,
+      "loss": 0.9778,
+      "step": 29020
+    },
+    {
+      "epoch": 0.21013847568170138,
+      "grad_norm": 0.1807478368282318,
+      "learning_rate": 4.789868762984358e-06,
+      "loss": 1.0065,
+      "step": 29030
+    },
+    {
+      "epoch": 0.21021086234228756,
+      "grad_norm": 0.17527340352535248,
+      "learning_rate": 4.789796376323771e-06,
+      "loss": 0.998,
+      "step": 29040
+    },
+    {
+      "epoch": 0.21028324900287376,
+      "grad_norm": 0.1826309859752655,
+      "learning_rate": 4.789723989663185e-06,
+      "loss": 0.9983,
+      "step": 29050
+    },
+    {
+      "epoch": 0.21035563566345994,
+      "grad_norm": 0.17075999081134796,
+      "learning_rate": 4.7896516030025995e-06,
+      "loss": 1.0071,
+      "step": 29060
+    },
+    {
+      "epoch": 0.21042802232404612,
+      "grad_norm": 0.20827743411064148,
+      "learning_rate": 4.789579216342013e-06,
+      "loss": 0.9932,
+      "step": 29070
+    },
+    {
+      "epoch": 0.21050040898463232,
+      "grad_norm": 0.18631266057491302,
+      "learning_rate": 4.789506829681427e-06,
+      "loss": 1.0037,
+      "step": 29080
+    },
+    {
+      "epoch": 0.2105727956452185,
+      "grad_norm": 0.1681118607521057,
+      "learning_rate": 4.78943444302084e-06,
+      "loss": 0.9982,
+      "step": 29090
+    },
+    {
+      "epoch": 0.21064518230580467,
+      "grad_norm": 0.16737908124923706,
+      "learning_rate": 4.789362056360255e-06,
+      "loss": 0.9948,
+      "step": 29100
+    },
+    {
+      "epoch": 0.21071756896639088,
+      "grad_norm": 0.17825300991535187,
+      "learning_rate": 4.789289669699668e-06,
+      "loss": 1.0048,
+      "step": 29110
+    },
+    {
+      "epoch": 0.21078995562697705,
+      "grad_norm": 0.16629593074321747,
+      "learning_rate": 4.789217283039082e-06,
+      "loss": 1.002,
+      "step": 29120
+    },
+    {
+      "epoch": 0.21086234228756326,
+      "grad_norm": 0.16453807055950165,
+      "learning_rate": 4.789144896378496e-06,
+      "loss": 0.9952,
+      "step": 29130
+    },
+    {
+      "epoch": 0.21093472894814944,
+      "grad_norm": 0.18079112470149994,
+      "learning_rate": 4.78907250971791e-06,
+      "loss": 0.994,
+      "step": 29140
+    },
+    {
+      "epoch": 0.2110071156087356,
+      "grad_norm": 0.16734014451503754,
+      "learning_rate": 4.789000123057324e-06,
+      "loss": 0.9832,
+      "step": 29150
+    },
+    {
+      "epoch": 0.21107950226932182,
+      "grad_norm": 0.16586540639400482,
+      "learning_rate": 4.788927736396737e-06,
+      "loss": 1.0078,
+      "step": 29160
+    },
+    {
+      "epoch": 0.211151888929908,
+      "grad_norm": 0.1610448658466339,
+      "learning_rate": 4.788855349736151e-06,
+      "loss": 0.9818,
+      "step": 29170
+    },
+    {
+      "epoch": 0.21122427559049417,
+      "grad_norm": 0.16332942247390747,
+      "learning_rate": 4.788782963075565e-06,
+      "loss": 0.9989,
+      "step": 29180
+    },
+    {
+      "epoch": 0.21129666225108037,
+      "grad_norm": 0.17163671553134918,
+      "learning_rate": 4.788710576414979e-06,
+      "loss": 1.0085,
+      "step": 29190
+    },
+    {
+      "epoch": 0.21136904891166655,
+      "grad_norm": 0.17151540517807007,
+      "learning_rate": 4.788638189754393e-06,
+      "loss": 0.9963,
+      "step": 29200
+    },
+    {
+      "epoch": 0.21144143557225276,
+      "grad_norm": 0.16174449026584625,
+      "learning_rate": 4.788565803093806e-06,
+      "loss": 1.0069,
+      "step": 29210
+    },
+    {
+      "epoch": 0.21151382223283893,
+      "grad_norm": 0.1772051900625229,
+      "learning_rate": 4.78849341643322e-06,
+      "loss": 1.0059,
+      "step": 29220
+    },
+    {
+      "epoch": 0.2115862088934251,
+      "grad_norm": 0.1666724532842636,
+      "learning_rate": 4.788421029772634e-06,
+      "loss": 0.9863,
+      "step": 29230
+    },
+    {
+      "epoch": 0.21165859555401131,
+      "grad_norm": 0.1606568545103073,
+      "learning_rate": 4.788348643112048e-06,
+      "loss": 0.9924,
+      "step": 29240
+    },
+    {
+      "epoch": 0.2117309822145975,
+      "grad_norm": 0.1638329178094864,
+      "learning_rate": 4.788276256451462e-06,
+      "loss": 0.9987,
+      "step": 29250
+    },
+    {
+      "epoch": 0.21180336887518367,
+      "grad_norm": 0.17658962309360504,
+      "learning_rate": 4.788203869790875e-06,
+      "loss": 0.9978,
+      "step": 29260
+    },
+    {
+      "epoch": 0.21187575553576987,
+      "grad_norm": 0.17163963615894318,
+      "learning_rate": 4.78813148313029e-06,
+      "loss": 0.9986,
+      "step": 29270
+    },
+    {
+      "epoch": 0.21194814219635605,
+      "grad_norm": 0.17665480077266693,
+      "learning_rate": 4.7880590964697024e-06,
+      "loss": 0.9921,
+      "step": 29280
+    },
+    {
+      "epoch": 0.21202052885694225,
+      "grad_norm": 0.17104417085647583,
+      "learning_rate": 4.787986709809117e-06,
+      "loss": 0.9853,
+      "step": 29290
+    },
+    {
+      "epoch": 0.21209291551752843,
+      "grad_norm": 0.17565962672233582,
+      "learning_rate": 4.7879143231485305e-06,
+      "loss": 1.0039,
+      "step": 29300
+    },
+    {
+      "epoch": 0.2121653021781146,
+      "grad_norm": 0.2072620391845703,
+      "learning_rate": 4.787841936487944e-06,
+      "loss": 0.9881,
+      "step": 29310
+    },
+    {
+      "epoch": 0.2122376888387008,
+      "grad_norm": 0.16268736124038696,
+      "learning_rate": 4.787769549827358e-06,
+      "loss": 0.9767,
+      "step": 29320
+    },
+    {
+      "epoch": 0.212310075499287,
+      "grad_norm": 0.19353044033050537,
+      "learning_rate": 4.787697163166772e-06,
+      "loss": 1.002,
+      "step": 29330
+    },
+    {
+      "epoch": 0.2123824621598732,
+      "grad_norm": 0.1635683923959732,
+      "learning_rate": 4.787624776506186e-06,
+      "loss": 0.9846,
+      "step": 29340
+    },
+    {
+      "epoch": 0.21245484882045937,
+      "grad_norm": 0.16228455305099487,
+      "learning_rate": 4.7875523898455994e-06,
+      "loss": 0.996,
+      "step": 29350
+    },
+    {
+      "epoch": 0.21252723548104555,
+      "grad_norm": 0.18255174160003662,
+      "learning_rate": 4.787480003185013e-06,
+      "loss": 0.9805,
+      "step": 29360
+    },
+    {
+      "epoch": 0.21259962214163175,
+      "grad_norm": 0.17667743563652039,
+      "learning_rate": 4.787407616524427e-06,
+      "loss": 0.994,
+      "step": 29370
+    },
+    {
+      "epoch": 0.21267200880221793,
+      "grad_norm": 0.15924212336540222,
+      "learning_rate": 4.787335229863841e-06,
+      "loss": 0.9877,
+      "step": 29380
+    },
+    {
+      "epoch": 0.2127443954628041,
+      "grad_norm": 0.27366387844085693,
+      "learning_rate": 4.787262843203255e-06,
+      "loss": 0.9891,
+      "step": 29390
+    },
+    {
+      "epoch": 0.2128167821233903,
+      "grad_norm": 0.17497357726097107,
+      "learning_rate": 4.787190456542668e-06,
+      "loss": 0.9717,
+      "step": 29400
+    },
+    {
+      "epoch": 0.21288916878397648,
+      "grad_norm": 0.17088007926940918,
+      "learning_rate": 4.787118069882082e-06,
+      "loss": 0.9952,
+      "step": 29410
+    },
+    {
+      "epoch": 0.2129615554445627,
+      "grad_norm": 0.16863124072551727,
+      "learning_rate": 4.7870456832214965e-06,
+      "loss": 0.9988,
+      "step": 29420
+    },
+    {
+      "epoch": 0.21303394210514887,
+      "grad_norm": 0.16611674427986145,
+      "learning_rate": 4.78697329656091e-06,
+      "loss": 0.989,
+      "step": 29430
+    },
+    {
+      "epoch": 0.21310632876573504,
+      "grad_norm": 0.2923133373260498,
+      "learning_rate": 4.786900909900324e-06,
+      "loss": 1.0054,
+      "step": 29440
+    },
+    {
+      "epoch": 0.21317871542632125,
+      "grad_norm": 0.18671530485153198,
+      "learning_rate": 4.786828523239737e-06,
+      "loss": 1.003,
+      "step": 29450
+    },
+    {
+      "epoch": 0.21325110208690742,
+      "grad_norm": 0.17407409846782684,
+      "learning_rate": 4.786756136579152e-06,
+      "loss": 0.9949,
+      "step": 29460
+    },
+    {
+      "epoch": 0.2133234887474936,
+      "grad_norm": 0.27453675866127014,
+      "learning_rate": 4.786683749918565e-06,
+      "loss": 0.9984,
+      "step": 29470
+    },
+    {
+      "epoch": 0.2133958754080798,
+      "grad_norm": 0.16514623165130615,
+      "learning_rate": 4.786611363257979e-06,
+      "loss": 0.9985,
+      "step": 29480
+    },
+    {
+      "epoch": 0.21346826206866598,
+      "grad_norm": 0.17301639914512634,
+      "learning_rate": 4.786538976597393e-06,
+      "loss": 1.0006,
+      "step": 29490
+    },
+    {
+      "epoch": 0.21354064872925219,
+      "grad_norm": 0.1778867542743683,
+      "learning_rate": 4.786466589936807e-06,
+      "loss": 0.9948,
+      "step": 29500
+    },
+    {
+      "epoch": 0.21361303538983836,
+      "grad_norm": 0.1603461503982544,
+      "learning_rate": 4.786394203276221e-06,
+      "loss": 0.9911,
+      "step": 29510
+    },
+    {
+      "epoch": 0.21368542205042454,
+      "grad_norm": 0.234296053647995,
+      "learning_rate": 4.786321816615634e-06,
+      "loss": 0.9971,
+      "step": 29520
+    },
+    {
+      "epoch": 0.21375780871101074,
+      "grad_norm": 0.158965066075325,
+      "learning_rate": 4.786249429955048e-06,
+      "loss": 0.9987,
+      "step": 29530
+    },
+    {
+      "epoch": 0.21383019537159692,
+      "grad_norm": 0.1825021207332611,
+      "learning_rate": 4.786177043294462e-06,
+      "loss": 0.9916,
+      "step": 29540
+    },
+    {
+      "epoch": 0.2139025820321831,
+      "grad_norm": 0.18081267178058624,
+      "learning_rate": 4.786104656633876e-06,
+      "loss": 0.9879,
+      "step": 29550
+    },
+    {
+      "epoch": 0.2139749686927693,
+      "grad_norm": 0.1616426259279251,
+      "learning_rate": 4.78603226997329e-06,
+      "loss": 0.9958,
+      "step": 29560
+    },
+    {
+      "epoch": 0.21404735535335548,
+      "grad_norm": 0.17399275302886963,
+      "learning_rate": 4.785959883312703e-06,
+      "loss": 1.0119,
+      "step": 29570
+    },
+    {
+      "epoch": 0.21411974201394168,
+      "grad_norm": 0.22257418930530548,
+      "learning_rate": 4.785887496652118e-06,
+      "loss": 0.9888,
+      "step": 29580
+    },
+    {
+      "epoch": 0.21419212867452786,
+      "grad_norm": 0.162775918841362,
+      "learning_rate": 4.785815109991531e-06,
+      "loss": 0.9989,
+      "step": 29590
+    },
+    {
+      "epoch": 0.21426451533511404,
+      "grad_norm": 0.16926290094852448,
+      "learning_rate": 4.785742723330945e-06,
+      "loss": 0.9965,
+      "step": 29600
+    },
+    {
+      "epoch": 0.21433690199570024,
+      "grad_norm": 0.18034987151622772,
+      "learning_rate": 4.7856703366703585e-06,
+      "loss": 0.995,
+      "step": 29610
+    },
+    {
+      "epoch": 0.21440928865628642,
+      "grad_norm": 0.17207445204257965,
+      "learning_rate": 4.785597950009773e-06,
+      "loss": 1.0046,
+      "step": 29620
+    },
+    {
+      "epoch": 0.2144816753168726,
+      "grad_norm": 0.21093519032001495,
+      "learning_rate": 4.785525563349187e-06,
+      "loss": 0.9917,
+      "step": 29630
+    },
+    {
+      "epoch": 0.2145540619774588,
+      "grad_norm": 0.16876211762428284,
+      "learning_rate": 4.7854531766886e-06,
+      "loss": 0.9903,
+      "step": 29640
+    },
+    {
+      "epoch": 0.21462644863804498,
+      "grad_norm": 0.24008895456790924,
+      "learning_rate": 4.785380790028014e-06,
+      "loss": 0.9946,
+      "step": 29650
+    },
+    {
+      "epoch": 0.21469883529863118,
+      "grad_norm": 0.16969846189022064,
+      "learning_rate": 4.785308403367428e-06,
+      "loss": 0.9987,
+      "step": 29660
+    },
+    {
+      "epoch": 0.21477122195921736,
+      "grad_norm": 0.1585167944431305,
+      "learning_rate": 4.785236016706842e-06,
+      "loss": 0.9916,
+      "step": 29670
+    },
+    {
+      "epoch": 0.21484360861980353,
+      "grad_norm": 0.16442425549030304,
+      "learning_rate": 4.7851636300462556e-06,
+      "loss": 0.9765,
+      "step": 29680
+    },
+    {
+      "epoch": 0.21491599528038974,
+      "grad_norm": 0.19266870617866516,
+      "learning_rate": 4.785091243385669e-06,
+      "loss": 0.9818,
+      "step": 29690
+    },
+    {
+      "epoch": 0.21498838194097591,
+      "grad_norm": 0.16447675228118896,
+      "learning_rate": 4.785018856725084e-06,
+      "loss": 1.0,
+      "step": 29700
+    },
+    {
+      "epoch": 0.2150607686015621,
+      "grad_norm": 0.1561741828918457,
+      "learning_rate": 4.784946470064497e-06,
+      "loss": 0.9833,
+      "step": 29710
+    },
+    {
+      "epoch": 0.2151331552621483,
+      "grad_norm": 0.1617199033498764,
+      "learning_rate": 4.784874083403911e-06,
+      "loss": 0.997,
+      "step": 29720
+    },
+    {
+      "epoch": 0.21520554192273447,
+      "grad_norm": 0.16869542002677917,
+      "learning_rate": 4.7848016967433245e-06,
+      "loss": 0.9877,
+      "step": 29730
+    },
+    {
+      "epoch": 0.21527792858332068,
+      "grad_norm": 0.1657578945159912,
+      "learning_rate": 4.784729310082739e-06,
+      "loss": 0.984,
+      "step": 29740
+    },
+    {
+      "epoch": 0.21535031524390685,
+      "grad_norm": 0.1783364713191986,
+      "learning_rate": 4.7846569234221526e-06,
+      "loss": 0.9972,
+      "step": 29750
+    },
+    {
+      "epoch": 0.21542270190449303,
+      "grad_norm": 0.16609250009059906,
+      "learning_rate": 4.784584536761566e-06,
+      "loss": 0.9815,
+      "step": 29760
+    },
+    {
+      "epoch": 0.21549508856507923,
+      "grad_norm": 0.16367986798286438,
+      "learning_rate": 4.78451215010098e-06,
+      "loss": 0.9864,
+      "step": 29770
+    },
+    {
+      "epoch": 0.2155674752256654,
+      "grad_norm": 0.1632063388824463,
+      "learning_rate": 4.784439763440394e-06,
+      "loss": 0.9857,
+      "step": 29780
+    },
+    {
+      "epoch": 0.2156398618862516,
+      "grad_norm": 0.2034599334001541,
+      "learning_rate": 4.784367376779808e-06,
+      "loss": 0.9864,
+      "step": 29790
+    },
+    {
+      "epoch": 0.2157122485468378,
+      "grad_norm": 0.15968771278858185,
+      "learning_rate": 4.7842949901192215e-06,
+      "loss": 0.982,
+      "step": 29800
+    },
+    {
+      "epoch": 0.21578463520742397,
+      "grad_norm": 0.16494300961494446,
+      "learning_rate": 4.784222603458635e-06,
+      "loss": 0.9871,
+      "step": 29810
+    },
+    {
+      "epoch": 0.21585702186801017,
+      "grad_norm": 0.1890561729669571,
+      "learning_rate": 4.784150216798049e-06,
+      "loss": 0.9845,
+      "step": 29820
+    },
+    {
+      "epoch": 0.21592940852859635,
+      "grad_norm": 0.18601875007152557,
+      "learning_rate": 4.784077830137462e-06,
+      "loss": 0.9968,
+      "step": 29830
+    },
+    {
+      "epoch": 0.21600179518918253,
+      "grad_norm": 0.18021321296691895,
+      "learning_rate": 4.784005443476876e-06,
+      "loss": 0.9904,
+      "step": 29840
+    },
+    {
+      "epoch": 0.21607418184976873,
+      "grad_norm": 0.17945921421051025,
+      "learning_rate": 4.78393305681629e-06,
+      "loss": 0.9995,
+      "step": 29850
+    },
+    {
+      "epoch": 0.2161465685103549,
+      "grad_norm": 0.16598504781723022,
+      "learning_rate": 4.783860670155704e-06,
+      "loss": 1.0009,
+      "step": 29860
+    },
+    {
+      "epoch": 0.2162189551709411,
+      "grad_norm": 0.1580289602279663,
+      "learning_rate": 4.783788283495118e-06,
+      "loss": 0.9999,
+      "step": 29870
+    },
+    {
+      "epoch": 0.2162913418315273,
+      "grad_norm": 0.16476024687290192,
+      "learning_rate": 4.783715896834531e-06,
+      "loss": 0.9882,
+      "step": 29880
+    },
+    {
+      "epoch": 0.21636372849211347,
+      "grad_norm": 0.17372757196426392,
+      "learning_rate": 4.783643510173946e-06,
+      "loss": 0.9873,
+      "step": 29890
+    },
+    {
+      "epoch": 0.21643611515269967,
+      "grad_norm": 0.17424210906028748,
+      "learning_rate": 4.783571123513359e-06,
+      "loss": 0.9848,
+      "step": 29900
+    },
+    {
+      "epoch": 0.21650850181328585,
+      "grad_norm": 0.16753096878528595,
+      "learning_rate": 4.783498736852773e-06,
+      "loss": 0.9753,
+      "step": 29910
+    },
+    {
+      "epoch": 0.21658088847387202,
+      "grad_norm": 0.16922076046466827,
+      "learning_rate": 4.783426350192187e-06,
+      "loss": 0.9874,
+      "step": 29920
+    },
+    {
+      "epoch": 0.21665327513445823,
+      "grad_norm": 0.18576353788375854,
+      "learning_rate": 4.783353963531601e-06,
+      "loss": 0.9952,
+      "step": 29930
+    },
+    {
+      "epoch": 0.2167256617950444,
+      "grad_norm": 0.15824037790298462,
+      "learning_rate": 4.783281576871015e-06,
+      "loss": 0.9921,
+      "step": 29940
+    },
+    {
+      "epoch": 0.2167980484556306,
+      "grad_norm": 0.18869732320308685,
+      "learning_rate": 4.783209190210428e-06,
+      "loss": 1.0058,
+      "step": 29950
+    },
+    {
+      "epoch": 0.2168704351162168,
+      "grad_norm": 0.17720650136470795,
+      "learning_rate": 4.783136803549842e-06,
+      "loss": 0.9959,
+      "step": 29960
+    },
+    {
+      "epoch": 0.21694282177680296,
+      "grad_norm": 0.17192226648330688,
+      "learning_rate": 4.783064416889256e-06,
+      "loss": 0.9915,
+      "step": 29970
+    },
+    {
+      "epoch": 0.21701520843738917,
+      "grad_norm": 0.17724481225013733,
+      "learning_rate": 4.78299203022867e-06,
+      "loss": 0.9957,
+      "step": 29980
+    },
+    {
+      "epoch": 0.21708759509797534,
+      "grad_norm": 0.16848604381084442,
+      "learning_rate": 4.782919643568084e-06,
+      "loss": 0.995,
+      "step": 29990
+    },
+    {
+      "epoch": 0.21715998175856152,
+      "grad_norm": 0.16832269728183746,
+      "learning_rate": 4.782847256907497e-06,
+      "loss": 1.0102,
+      "step": 30000
+    },
+    {
+      "epoch": 0.21723236841914773,
+      "grad_norm": 0.16782933473587036,
+      "learning_rate": 4.782774870246911e-06,
+      "loss": 0.9875,
+      "step": 30010
+    },
+    {
+      "epoch": 0.2173047550797339,
+      "grad_norm": 0.17606335878372192,
+      "learning_rate": 4.782702483586325e-06,
+      "loss": 0.9907,
+      "step": 30020
+    },
+    {
+      "epoch": 0.2173771417403201,
+      "grad_norm": 0.16667865216732025,
+      "learning_rate": 4.782630096925739e-06,
+      "loss": 0.9906,
+      "step": 30030
+    },
+    {
+      "epoch": 0.21744952840090628,
+      "grad_norm": 0.17801982164382935,
+      "learning_rate": 4.7825577102651525e-06,
+      "loss": 0.995,
+      "step": 30040
+    },
+    {
+      "epoch": 0.21752191506149246,
+      "grad_norm": 0.19045329093933105,
+      "learning_rate": 4.782485323604566e-06,
+      "loss": 0.9944,
+      "step": 30050
+    },
+    {
+      "epoch": 0.21759430172207866,
+      "grad_norm": 0.19208842515945435,
+      "learning_rate": 4.782412936943981e-06,
+      "loss": 0.9972,
+      "step": 30060
+    },
+    {
+      "epoch": 0.21766668838266484,
+      "grad_norm": 0.19080910086631775,
+      "learning_rate": 4.782340550283394e-06,
+      "loss": 0.9907,
+      "step": 30070
+    },
+    {
+      "epoch": 0.21773907504325102,
+      "grad_norm": 0.16448453068733215,
+      "learning_rate": 4.782268163622808e-06,
+      "loss": 0.9954,
+      "step": 30080
+    },
+    {
+      "epoch": 0.21781146170383722,
+      "grad_norm": 0.16412755846977234,
+      "learning_rate": 4.7821957769622214e-06,
+      "loss": 0.9929,
+      "step": 30090
+    },
+    {
+      "epoch": 0.2178838483644234,
+      "grad_norm": 0.17973798513412476,
+      "learning_rate": 4.782123390301636e-06,
+      "loss": 0.9898,
+      "step": 30100
+    },
+    {
+      "epoch": 0.2179562350250096,
+      "grad_norm": 0.16456930339336395,
+      "learning_rate": 4.7820510036410495e-06,
+      "loss": 0.9943,
+      "step": 30110
+    },
+    {
+      "epoch": 0.21802862168559578,
+      "grad_norm": 0.1790180802345276,
+      "learning_rate": 4.781978616980463e-06,
+      "loss": 1.0058,
+      "step": 30120
+    },
+    {
+      "epoch": 0.21810100834618196,
+      "grad_norm": 0.15905308723449707,
+      "learning_rate": 4.781906230319877e-06,
+      "loss": 0.985,
+      "step": 30130
+    },
+    {
+      "epoch": 0.21817339500676816,
+      "grad_norm": 0.17000477015972137,
+      "learning_rate": 4.781833843659291e-06,
+      "loss": 0.9802,
+      "step": 30140
+    },
+    {
+      "epoch": 0.21824578166735434,
+      "grad_norm": 0.19705595076084137,
+      "learning_rate": 4.781761456998705e-06,
+      "loss": 1.002,
+      "step": 30150
+    },
+    {
+      "epoch": 0.21831816832794051,
+      "grad_norm": 0.16319391131401062,
+      "learning_rate": 4.7816890703381185e-06,
+      "loss": 0.997,
+      "step": 30160
+    },
+    {
+      "epoch": 0.21839055498852672,
+      "grad_norm": 0.1748110055923462,
+      "learning_rate": 4.781616683677532e-06,
+      "loss": 0.9865,
+      "step": 30170
+    },
+    {
+      "epoch": 0.2184629416491129,
+      "grad_norm": 0.16149266064167023,
+      "learning_rate": 4.7815442970169465e-06,
+      "loss": 0.9952,
+      "step": 30180
+    },
+    {
+      "epoch": 0.2185353283096991,
+      "grad_norm": 0.22713536024093628,
+      "learning_rate": 4.78147191035636e-06,
+      "loss": 1.0025,
+      "step": 30190
+    },
+    {
+      "epoch": 0.21860771497028528,
+      "grad_norm": 0.1786499321460724,
+      "learning_rate": 4.781399523695774e-06,
+      "loss": 0.9896,
+      "step": 30200
+    },
+    {
+      "epoch": 0.21868010163087145,
+      "grad_norm": 0.19902971386909485,
+      "learning_rate": 4.781327137035187e-06,
+      "loss": 0.9976,
+      "step": 30210
+    },
+    {
+      "epoch": 0.21875248829145766,
+      "grad_norm": 0.17950847744941711,
+      "learning_rate": 4.781254750374602e-06,
+      "loss": 1.0033,
+      "step": 30220
+    },
+    {
+      "epoch": 0.21882487495204384,
+      "grad_norm": 0.1599850207567215,
+      "learning_rate": 4.7811823637140155e-06,
+      "loss": 0.9735,
+      "step": 30230
+    },
+    {
+      "epoch": 0.21889726161263,
+      "grad_norm": 0.15841835737228394,
+      "learning_rate": 4.781109977053429e-06,
+      "loss": 1.0022,
+      "step": 30240
+    },
+    {
+      "epoch": 0.21896964827321622,
+      "grad_norm": 0.17060723900794983,
+      "learning_rate": 4.781037590392843e-06,
+      "loss": 0.9909,
+      "step": 30250
+    },
+    {
+      "epoch": 0.2190420349338024,
+      "grad_norm": 0.16815410554409027,
+      "learning_rate": 4.780965203732257e-06,
+      "loss": 0.9865,
+      "step": 30260
+    },
+    {
+      "epoch": 0.2191144215943886,
+      "grad_norm": 0.2049429714679718,
+      "learning_rate": 4.780892817071671e-06,
+      "loss": 0.978,
+      "step": 30270
+    },
+    {
+      "epoch": 0.21918680825497477,
+      "grad_norm": 0.1744859516620636,
+      "learning_rate": 4.780820430411084e-06,
+      "loss": 0.9946,
+      "step": 30280
+    },
+    {
+      "epoch": 0.21925919491556095,
+      "grad_norm": 0.17535239458084106,
+      "learning_rate": 4.780748043750498e-06,
+      "loss": 0.9893,
+      "step": 30290
+    },
+    {
+      "epoch": 0.21933158157614716,
+      "grad_norm": 0.16993948817253113,
+      "learning_rate": 4.7806756570899125e-06,
+      "loss": 0.995,
+      "step": 30300
+    },
+    {
+      "epoch": 0.21940396823673333,
+      "grad_norm": 0.1737825572490692,
+      "learning_rate": 4.780603270429326e-06,
+      "loss": 0.979,
+      "step": 30310
+    },
+    {
+      "epoch": 0.2194763548973195,
+      "grad_norm": 0.19781017303466797,
+      "learning_rate": 4.78053088376874e-06,
+      "loss": 0.9894,
+      "step": 30320
+    },
+    {
+      "epoch": 0.2195487415579057,
+      "grad_norm": 0.15426485240459442,
+      "learning_rate": 4.780458497108153e-06,
+      "loss": 0.9906,
+      "step": 30330
+    },
+    {
+      "epoch": 0.2196211282184919,
+      "grad_norm": 0.17458288371562958,
+      "learning_rate": 4.780386110447567e-06,
+      "loss": 0.9847,
+      "step": 30340
+    },
+    {
+      "epoch": 0.2196935148790781,
+      "grad_norm": 0.17659978568553925,
+      "learning_rate": 4.7803137237869805e-06,
+      "loss": 0.9937,
+      "step": 30350
+    },
+    {
+      "epoch": 0.21976590153966427,
+      "grad_norm": 0.16080795228481293,
+      "learning_rate": 4.780241337126394e-06,
+      "loss": 0.975,
+      "step": 30360
+    },
+    {
+      "epoch": 0.21983828820025045,
+      "grad_norm": 0.1639893800020218,
+      "learning_rate": 4.780168950465809e-06,
+      "loss": 1.0061,
+      "step": 30370
+    },
+    {
+      "epoch": 0.21991067486083665,
+      "grad_norm": 0.17804419994354248,
+      "learning_rate": 4.780096563805222e-06,
+      "loss": 0.9952,
+      "step": 30380
+    },
+    {
+      "epoch": 0.21998306152142283,
+      "grad_norm": 0.17205168306827545,
+      "learning_rate": 4.780024177144636e-06,
+      "loss": 0.9953,
+      "step": 30390
+    },
+    {
+      "epoch": 0.220055448182009,
+      "grad_norm": 0.2006746232509613,
+      "learning_rate": 4.7799517904840495e-06,
+      "loss": 0.9882,
+      "step": 30400
+    },
+    {
+      "epoch": 0.2201278348425952,
+      "grad_norm": 0.18707719445228577,
+      "learning_rate": 4.779879403823464e-06,
+      "loss": 1.0003,
+      "step": 30410
+    },
+    {
+      "epoch": 0.2202002215031814,
+      "grad_norm": 0.178603395819664,
+      "learning_rate": 4.7798070171628776e-06,
+      "loss": 0.9928,
+      "step": 30420
+    },
+    {
+      "epoch": 0.2202726081637676,
+      "grad_norm": 0.19374102354049683,
+      "learning_rate": 4.779734630502291e-06,
+      "loss": 0.9865,
+      "step": 30430
+    },
+    {
+      "epoch": 0.22034499482435377,
+      "grad_norm": 0.1638360470533371,
+      "learning_rate": 4.779662243841705e-06,
+      "loss": 0.9922,
+      "step": 30440
+    },
+    {
+      "epoch": 0.22041738148493994,
+      "grad_norm": 0.16339033842086792,
+      "learning_rate": 4.779589857181119e-06,
+      "loss": 0.995,
+      "step": 30450
+    },
+    {
+      "epoch": 0.22048976814552615,
+      "grad_norm": 0.17286980152130127,
+      "learning_rate": 4.779517470520533e-06,
+      "loss": 0.9862,
+      "step": 30460
+    },
+    {
+      "epoch": 0.22056215480611233,
+      "grad_norm": 0.17059637606143951,
+      "learning_rate": 4.7794450838599465e-06,
+      "loss": 0.988,
+      "step": 30470
+    },
+    {
+      "epoch": 0.22063454146669853,
+      "grad_norm": 0.15443108975887299,
+      "learning_rate": 4.77937269719936e-06,
+      "loss": 0.9902,
+      "step": 30480
+    },
+    {
+      "epoch": 0.2207069281272847,
+      "grad_norm": 0.16582870483398438,
+      "learning_rate": 4.7793003105387746e-06,
+      "loss": 0.9947,
+      "step": 30490
+    },
+    {
+      "epoch": 0.22077931478787088,
+      "grad_norm": 0.16786135733127594,
+      "learning_rate": 4.779227923878188e-06,
+      "loss": 0.9932,
+      "step": 30500
+    },
+    {
+      "epoch": 0.2208517014484571,
+      "grad_norm": 0.18378998339176178,
+      "learning_rate": 4.779155537217602e-06,
+      "loss": 0.9923,
+      "step": 30510
+    },
+    {
+      "epoch": 0.22092408810904327,
+      "grad_norm": 0.1661590337753296,
+      "learning_rate": 4.779083150557015e-06,
+      "loss": 0.9963,
+      "step": 30520
+    },
+    {
+      "epoch": 0.22099647476962944,
+      "grad_norm": 0.16045540571212769,
+      "learning_rate": 4.77901076389643e-06,
+      "loss": 0.9888,
+      "step": 30530
+    },
+    {
+      "epoch": 0.22106886143021565,
+      "grad_norm": 0.17073200643062592,
+      "learning_rate": 4.7789383772358435e-06,
+      "loss": 0.9863,
+      "step": 30540
+    },
+    {
+      "epoch": 0.22114124809080182,
+      "grad_norm": 0.2749057710170746,
+      "learning_rate": 4.778865990575257e-06,
+      "loss": 0.9993,
+      "step": 30550
+    },
+    {
+      "epoch": 0.22121363475138803,
+      "grad_norm": 0.17265433073043823,
+      "learning_rate": 4.778793603914671e-06,
+      "loss": 0.9898,
+      "step": 30560
+    },
+    {
+      "epoch": 0.2212860214119742,
+      "grad_norm": 0.1878434419631958,
+      "learning_rate": 4.778721217254085e-06,
+      "loss": 0.998,
+      "step": 30570
+    },
+    {
+      "epoch": 0.22135840807256038,
+      "grad_norm": 0.16679958999156952,
+      "learning_rate": 4.778648830593499e-06,
+      "loss": 0.9897,
+      "step": 30580
+    },
+    {
+      "epoch": 0.22143079473314659,
+      "grad_norm": 0.18159635365009308,
+      "learning_rate": 4.778576443932912e-06,
+      "loss": 0.9953,
+      "step": 30590
+    },
+    {
+      "epoch": 0.22150318139373276,
+      "grad_norm": 0.18782582879066467,
+      "learning_rate": 4.778504057272326e-06,
+      "loss": 0.9949,
+      "step": 30600
+    },
+    {
+      "epoch": 0.22157556805431894,
+      "grad_norm": 0.16997790336608887,
+      "learning_rate": 4.77843167061174e-06,
+      "loss": 0.9956,
+      "step": 30610
+    },
+    {
+      "epoch": 0.22164795471490514,
+      "grad_norm": 0.19015435874462128,
+      "learning_rate": 4.778359283951154e-06,
+      "loss": 0.9788,
+      "step": 30620
+    },
+    {
+      "epoch": 0.22172034137549132,
+      "grad_norm": 0.15905068814754486,
+      "learning_rate": 4.778286897290568e-06,
+      "loss": 0.9862,
+      "step": 30630
+    },
+    {
+      "epoch": 0.22179272803607752,
+      "grad_norm": 0.21960864961147308,
+      "learning_rate": 4.778214510629981e-06,
+      "loss": 0.9855,
+      "step": 30640
+    },
+    {
+      "epoch": 0.2218651146966637,
+      "grad_norm": 0.1731971949338913,
+      "learning_rate": 4.778142123969395e-06,
+      "loss": 0.9855,
+      "step": 30650
+    },
+    {
+      "epoch": 0.22193750135724988,
+      "grad_norm": 0.1737217754125595,
+      "learning_rate": 4.7780697373088094e-06,
+      "loss": 0.9929,
+      "step": 30660
+    },
+    {
+      "epoch": 0.22200988801783608,
+      "grad_norm": 0.16913598775863647,
+      "learning_rate": 4.777997350648223e-06,
+      "loss": 0.9939,
+      "step": 30670
+    },
+    {
+      "epoch": 0.22208227467842226,
+      "grad_norm": 0.19872261583805084,
+      "learning_rate": 4.777924963987637e-06,
+      "loss": 1.0059,
+      "step": 30680
+    },
+    {
+      "epoch": 0.22215466133900844,
+      "grad_norm": 0.17020894587039948,
+      "learning_rate": 4.77785257732705e-06,
+      "loss": 0.9992,
+      "step": 30690
+    },
+    {
+      "epoch": 0.22222704799959464,
+      "grad_norm": 0.19177848100662231,
+      "learning_rate": 4.777780190666465e-06,
+      "loss": 0.9826,
+      "step": 30700
+    },
+    {
+      "epoch": 0.22229943466018082,
+      "grad_norm": 0.17280960083007812,
+      "learning_rate": 4.777707804005878e-06,
+      "loss": 0.9868,
+      "step": 30710
+    },
+    {
+      "epoch": 0.22237182132076702,
+      "grad_norm": 0.1845826506614685,
+      "learning_rate": 4.777635417345292e-06,
+      "loss": 0.994,
+      "step": 30720
+    },
+    {
+      "epoch": 0.2224442079813532,
+      "grad_norm": 0.20441372692584991,
+      "learning_rate": 4.777563030684706e-06,
+      "loss": 0.9852,
+      "step": 30730
+    },
+    {
+      "epoch": 0.22251659464193937,
+      "grad_norm": 0.17806535959243774,
+      "learning_rate": 4.77749064402412e-06,
+      "loss": 0.9922,
+      "step": 30740
+    },
+    {
+      "epoch": 0.22258898130252558,
+      "grad_norm": 0.1843547821044922,
+      "learning_rate": 4.777418257363534e-06,
+      "loss": 0.9931,
+      "step": 30750
+    },
+    {
+      "epoch": 0.22266136796311176,
+      "grad_norm": 0.1745116412639618,
+      "learning_rate": 4.777345870702947e-06,
+      "loss": 0.9939,
+      "step": 30760
+    },
+    {
+      "epoch": 0.22273375462369793,
+      "grad_norm": 0.18230880796909332,
+      "learning_rate": 4.777273484042361e-06,
+      "loss": 0.9821,
+      "step": 30770
+    },
+    {
+      "epoch": 0.22280614128428414,
+      "grad_norm": 0.17460425198078156,
+      "learning_rate": 4.777201097381775e-06,
+      "loss": 0.9932,
+      "step": 30780
+    },
+    {
+      "epoch": 0.2228785279448703,
+      "grad_norm": 0.16138465702533722,
+      "learning_rate": 4.777128710721189e-06,
+      "loss": 0.9921,
+      "step": 30790
+    },
+    {
+      "epoch": 0.22295091460545652,
+      "grad_norm": 0.1702491044998169,
+      "learning_rate": 4.777056324060603e-06,
+      "loss": 1.0047,
+      "step": 30800
+    },
+    {
+      "epoch": 0.2230233012660427,
+      "grad_norm": 0.1869451254606247,
+      "learning_rate": 4.776983937400016e-06,
+      "loss": 0.9977,
+      "step": 30810
+    },
+    {
+      "epoch": 0.22309568792662887,
+      "grad_norm": 0.17127086222171783,
+      "learning_rate": 4.776911550739431e-06,
+      "loss": 1.0077,
+      "step": 30820
+    },
+    {
+      "epoch": 0.22316807458721508,
+      "grad_norm": 0.17618906497955322,
+      "learning_rate": 4.776839164078844e-06,
+      "loss": 0.9919,
+      "step": 30830
+    },
+    {
+      "epoch": 0.22324046124780125,
+      "grad_norm": 0.19127856194972992,
+      "learning_rate": 4.776766777418258e-06,
+      "loss": 0.9839,
+      "step": 30840
+    },
+    {
+      "epoch": 0.22331284790838743,
+      "grad_norm": 0.16776026785373688,
+      "learning_rate": 4.7766943907576715e-06,
+      "loss": 0.9903,
+      "step": 30850
+    },
+    {
+      "epoch": 0.22338523456897363,
+      "grad_norm": 0.17897354066371918,
+      "learning_rate": 4.776622004097086e-06,
+      "loss": 0.9847,
+      "step": 30860
+    },
+    {
+      "epoch": 0.2234576212295598,
+      "grad_norm": 0.16106632351875305,
+      "learning_rate": 4.776549617436499e-06,
+      "loss": 1.0027,
+      "step": 30870
+    },
+    {
+      "epoch": 0.22353000789014602,
+      "grad_norm": 0.15779045224189758,
+      "learning_rate": 4.776477230775912e-06,
+      "loss": 0.9942,
+      "step": 30880
+    },
+    {
+      "epoch": 0.2236023945507322,
+      "grad_norm": 0.19655205309391022,
+      "learning_rate": 4.776404844115327e-06,
+      "loss": 1.0076,
+      "step": 30890
+    },
+    {
+      "epoch": 0.22367478121131837,
+      "grad_norm": 0.1667492836713791,
+      "learning_rate": 4.7763324574547405e-06,
+      "loss": 0.9975,
+      "step": 30900
+    },
+    {
+      "epoch": 0.22374716787190457,
+      "grad_norm": 0.18029555678367615,
+      "learning_rate": 4.776260070794154e-06,
+      "loss": 0.9857,
+      "step": 30910
+    },
+    {
+      "epoch": 0.22381955453249075,
+      "grad_norm": 0.1663108766078949,
+      "learning_rate": 4.776187684133568e-06,
+      "loss": 0.9952,
+      "step": 30920
+    },
+    {
+      "epoch": 0.22389194119307693,
+      "grad_norm": 0.18549077212810516,
+      "learning_rate": 4.776115297472982e-06,
+      "loss": 0.9825,
+      "step": 30930
+    },
+    {
+      "epoch": 0.22396432785366313,
+      "grad_norm": 0.17091688513755798,
+      "learning_rate": 4.776042910812396e-06,
+      "loss": 1.0017,
+      "step": 30940
+    },
+    {
+      "epoch": 0.2240367145142493,
+      "grad_norm": 0.18328119814395905,
+      "learning_rate": 4.775970524151809e-06,
+      "loss": 0.9957,
+      "step": 30950
+    },
+    {
+      "epoch": 0.2241091011748355,
+      "grad_norm": 0.18230809271335602,
+      "learning_rate": 4.775898137491223e-06,
+      "loss": 1.0105,
+      "step": 30960
+    },
+    {
+      "epoch": 0.2241814878354217,
+      "grad_norm": 0.17327918112277985,
+      "learning_rate": 4.7758257508306375e-06,
+      "loss": 0.9785,
+      "step": 30970
+    },
+    {
+      "epoch": 0.22425387449600787,
+      "grad_norm": 0.1592261642217636,
+      "learning_rate": 4.775753364170051e-06,
+      "loss": 0.9965,
+      "step": 30980
+    },
+    {
+      "epoch": 0.22432626115659407,
+      "grad_norm": 0.16673846542835236,
+      "learning_rate": 4.775680977509465e-06,
+      "loss": 0.986,
+      "step": 30990
+    },
+    {
+      "epoch": 0.22439864781718025,
+      "grad_norm": 0.1908463090658188,
+      "learning_rate": 4.775608590848878e-06,
+      "loss": 1.0013,
+      "step": 31000
+    },
+    {
+      "epoch": 0.22447103447776642,
+      "grad_norm": 0.16515257954597473,
+      "learning_rate": 4.775536204188293e-06,
+      "loss": 1.0092,
+      "step": 31010
+    },
+    {
+      "epoch": 0.22454342113835263,
+      "grad_norm": 0.1733829379081726,
+      "learning_rate": 4.775463817527706e-06,
+      "loss": 0.9834,
+      "step": 31020
+    },
+    {
+      "epoch": 0.2246158077989388,
+      "grad_norm": 0.17411421239376068,
+      "learning_rate": 4.77539143086712e-06,
+      "loss": 0.9866,
+      "step": 31030
+    },
+    {
+      "epoch": 0.224688194459525,
+      "grad_norm": 0.16972704231739044,
+      "learning_rate": 4.775319044206534e-06,
+      "loss": 0.986,
+      "step": 31040
+    },
+    {
+      "epoch": 0.22476058112011119,
+      "grad_norm": 0.1979493349790573,
+      "learning_rate": 4.775246657545948e-06,
+      "loss": 0.9816,
+      "step": 31050
+    },
+    {
+      "epoch": 0.22483296778069736,
+      "grad_norm": 0.18027926981449127,
+      "learning_rate": 4.775174270885362e-06,
+      "loss": 0.9853,
+      "step": 31060
+    },
+    {
+      "epoch": 0.22490535444128357,
+      "grad_norm": 0.16481108963489532,
+      "learning_rate": 4.775101884224775e-06,
+      "loss": 0.9876,
+      "step": 31070
+    },
+    {
+      "epoch": 0.22497774110186974,
+      "grad_norm": 0.1805502474308014,
+      "learning_rate": 4.775029497564189e-06,
+      "loss": 0.9897,
+      "step": 31080
+    },
+    {
+      "epoch": 0.22505012776245595,
+      "grad_norm": 0.1621004194021225,
+      "learning_rate": 4.774957110903603e-06,
+      "loss": 0.9909,
+      "step": 31090
+    },
+    {
+      "epoch": 0.22512251442304212,
+      "grad_norm": 0.1652892827987671,
+      "learning_rate": 4.774884724243017e-06,
+      "loss": 0.9781,
+      "step": 31100
+    },
+    {
+      "epoch": 0.2251949010836283,
+      "grad_norm": 0.16680368781089783,
+      "learning_rate": 4.774812337582431e-06,
+      "loss": 0.9995,
+      "step": 31110
+    },
+    {
+      "epoch": 0.2252672877442145,
+      "grad_norm": 0.17814873158931732,
+      "learning_rate": 4.774739950921844e-06,
+      "loss": 0.9923,
+      "step": 31120
+    },
+    {
+      "epoch": 0.22533967440480068,
+      "grad_norm": 0.15377861261367798,
+      "learning_rate": 4.774667564261259e-06,
+      "loss": 0.9837,
+      "step": 31130
+    },
+    {
+      "epoch": 0.22541206106538686,
+      "grad_norm": 0.18391862511634827,
+      "learning_rate": 4.774595177600672e-06,
+      "loss": 0.9995,
+      "step": 31140
+    },
+    {
+      "epoch": 0.22548444772597306,
+      "grad_norm": 0.17734524607658386,
+      "learning_rate": 4.774522790940086e-06,
+      "loss": 1.0083,
+      "step": 31150
+    },
+    {
+      "epoch": 0.22555683438655924,
+      "grad_norm": 0.17555080354213715,
+      "learning_rate": 4.7744504042794996e-06,
+      "loss": 1.0057,
+      "step": 31160
+    },
+    {
+      "epoch": 0.22562922104714545,
+      "grad_norm": 0.16660809516906738,
+      "learning_rate": 4.774378017618914e-06,
+      "loss": 0.9793,
+      "step": 31170
+    },
+    {
+      "epoch": 0.22570160770773162,
+      "grad_norm": 0.19567494094371796,
+      "learning_rate": 4.774305630958328e-06,
+      "loss": 0.9865,
+      "step": 31180
+    },
+    {
+      "epoch": 0.2257739943683178,
+      "grad_norm": 0.15271180868148804,
+      "learning_rate": 4.774233244297741e-06,
+      "loss": 0.9831,
+      "step": 31190
+    },
+    {
+      "epoch": 0.225846381028904,
+      "grad_norm": 0.16450783610343933,
+      "learning_rate": 4.774160857637155e-06,
+      "loss": 0.996,
+      "step": 31200
+    },
+    {
+      "epoch": 0.22591876768949018,
+      "grad_norm": 0.2669644057750702,
+      "learning_rate": 4.774088470976569e-06,
+      "loss": 0.9876,
+      "step": 31210
+    },
+    {
+      "epoch": 0.22599115435007636,
+      "grad_norm": 0.1570923775434494,
+      "learning_rate": 4.774016084315983e-06,
+      "loss": 0.9838,
+      "step": 31220
+    },
+    {
+      "epoch": 0.22606354101066256,
+      "grad_norm": 0.1822049915790558,
+      "learning_rate": 4.7739436976553966e-06,
+      "loss": 0.976,
+      "step": 31230
+    },
+    {
+      "epoch": 0.22613592767124874,
+      "grad_norm": 0.17041492462158203,
+      "learning_rate": 4.77387131099481e-06,
+      "loss": 1.0074,
+      "step": 31240
+    },
+    {
+      "epoch": 0.22620831433183494,
+      "grad_norm": 0.1686597615480423,
+      "learning_rate": 4.773798924334224e-06,
+      "loss": 1.0004,
+      "step": 31250
+    },
+    {
+      "epoch": 0.22628070099242112,
+      "grad_norm": 0.16727478802204132,
+      "learning_rate": 4.773726537673638e-06,
+      "loss": 1.0044,
+      "step": 31260
+    },
+    {
+      "epoch": 0.2263530876530073,
+      "grad_norm": 0.16306477785110474,
+      "learning_rate": 4.773654151013052e-06,
+      "loss": 0.9966,
+      "step": 31270
+    },
+    {
+      "epoch": 0.2264254743135935,
+      "grad_norm": 0.17092132568359375,
+      "learning_rate": 4.7735817643524655e-06,
+      "loss": 1.0063,
+      "step": 31280
+    },
+    {
+      "epoch": 0.22649786097417968,
+      "grad_norm": 0.16931623220443726,
+      "learning_rate": 4.773509377691879e-06,
+      "loss": 0.9868,
+      "step": 31290
+    },
+    {
+      "epoch": 0.22657024763476585,
+      "grad_norm": 0.1603298783302307,
+      "learning_rate": 4.7734369910312936e-06,
+      "loss": 0.9837,
+      "step": 31300
+    },
+    {
+      "epoch": 0.22664263429535206,
+      "grad_norm": 0.19515341520309448,
+      "learning_rate": 4.773364604370707e-06,
+      "loss": 0.9844,
+      "step": 31310
+    },
+    {
+      "epoch": 0.22671502095593823,
+      "grad_norm": 0.16910241544246674,
+      "learning_rate": 4.773292217710121e-06,
+      "loss": 0.9948,
+      "step": 31320
+    },
+    {
+      "epoch": 0.22678740761652444,
+      "grad_norm": 0.19384412467479706,
+      "learning_rate": 4.773219831049534e-06,
+      "loss": 0.9617,
+      "step": 31330
+    },
+    {
+      "epoch": 0.22685979427711062,
+      "grad_norm": 0.17638161778450012,
+      "learning_rate": 4.773147444388949e-06,
+      "loss": 0.9865,
+      "step": 31340
+    },
+    {
+      "epoch": 0.2269321809376968,
+      "grad_norm": 0.17400164902210236,
+      "learning_rate": 4.7730750577283625e-06,
+      "loss": 0.997,
+      "step": 31350
+    },
+    {
+      "epoch": 0.227004567598283,
+      "grad_norm": 0.18208061158657074,
+      "learning_rate": 4.773002671067776e-06,
+      "loss": 1.0006,
+      "step": 31360
+    },
+    {
+      "epoch": 0.22707695425886917,
+      "grad_norm": 0.17312349379062653,
+      "learning_rate": 4.77293028440719e-06,
+      "loss": 0.973,
+      "step": 31370
+    },
+    {
+      "epoch": 0.22714934091945535,
+      "grad_norm": 0.18779698014259338,
+      "learning_rate": 4.772857897746604e-06,
+      "loss": 0.9899,
+      "step": 31380
+    },
+    {
+      "epoch": 0.22722172758004155,
+      "grad_norm": 0.1715383231639862,
+      "learning_rate": 4.772785511086018e-06,
+      "loss": 0.9891,
+      "step": 31390
+    },
+    {
+      "epoch": 0.22729411424062773,
+      "grad_norm": 0.18335622549057007,
+      "learning_rate": 4.772713124425431e-06,
+      "loss": 0.9925,
+      "step": 31400
+    },
+    {
+      "epoch": 0.22736650090121394,
+      "grad_norm": 0.1920541524887085,
+      "learning_rate": 4.772640737764845e-06,
+      "loss": 0.9875,
+      "step": 31410
+    },
+    {
+      "epoch": 0.2274388875618001,
+      "grad_norm": 0.16819268465042114,
+      "learning_rate": 4.772568351104259e-06,
+      "loss": 0.9909,
+      "step": 31420
+    },
+    {
+      "epoch": 0.2275112742223863,
+      "grad_norm": 0.1579744815826416,
+      "learning_rate": 4.772495964443672e-06,
+      "loss": 0.9903,
+      "step": 31430
+    },
+    {
+      "epoch": 0.2275836608829725,
+      "grad_norm": 0.1756497621536255,
+      "learning_rate": 4.772423577783086e-06,
+      "loss": 0.9948,
+      "step": 31440
+    },
+    {
+      "epoch": 0.22765604754355867,
+      "grad_norm": 0.1611769050359726,
+      "learning_rate": 4.7723511911225e-06,
+      "loss": 0.9925,
+      "step": 31450
+    },
+    {
+      "epoch": 0.22772843420414485,
+      "grad_norm": 0.19173139333724976,
+      "learning_rate": 4.772278804461914e-06,
+      "loss": 0.9761,
+      "step": 31460
+    },
+    {
+      "epoch": 0.22780082086473105,
+      "grad_norm": 0.20270651578903198,
+      "learning_rate": 4.772206417801328e-06,
+      "loss": 0.9898,
+      "step": 31470
+    },
+    {
+      "epoch": 0.22787320752531723,
+      "grad_norm": 0.17298614978790283,
+      "learning_rate": 4.772134031140741e-06,
+      "loss": 0.9871,
+      "step": 31480
+    },
+    {
+      "epoch": 0.22794559418590343,
+      "grad_norm": 0.15793482959270477,
+      "learning_rate": 4.772061644480156e-06,
+      "loss": 0.9874,
+      "step": 31490
+    },
+    {
+      "epoch": 0.2280179808464896,
+      "grad_norm": 0.1665399670600891,
+      "learning_rate": 4.771989257819569e-06,
+      "loss": 0.9963,
+      "step": 31500
+    },
+    {
+      "epoch": 0.2280903675070758,
+      "grad_norm": 0.1729333996772766,
+      "learning_rate": 4.771916871158983e-06,
+      "loss": 0.9995,
+      "step": 31510
+    },
+    {
+      "epoch": 0.228162754167662,
+      "grad_norm": 0.1829392910003662,
+      "learning_rate": 4.7718444844983965e-06,
+      "loss": 0.9848,
+      "step": 31520
+    },
+    {
+      "epoch": 0.22823514082824817,
+      "grad_norm": 0.17102506756782532,
+      "learning_rate": 4.771772097837811e-06,
+      "loss": 0.9867,
+      "step": 31530
+    },
+    {
+      "epoch": 0.22830752748883434,
+      "grad_norm": 0.1742861568927765,
+      "learning_rate": 4.771699711177225e-06,
+      "loss": 0.9888,
+      "step": 31540
+    },
+    {
+      "epoch": 0.22837991414942055,
+      "grad_norm": 0.17398032546043396,
+      "learning_rate": 4.771627324516638e-06,
+      "loss": 0.9985,
+      "step": 31550
+    },
+    {
+      "epoch": 0.22845230081000673,
+      "grad_norm": 0.168184295296669,
+      "learning_rate": 4.771554937856052e-06,
+      "loss": 0.9877,
+      "step": 31560
+    },
+    {
+      "epoch": 0.22852468747059293,
+      "grad_norm": 0.1651362031698227,
+      "learning_rate": 4.771482551195466e-06,
+      "loss": 0.996,
+      "step": 31570
+    },
+    {
+      "epoch": 0.2285970741311791,
+      "grad_norm": 0.17777453362941742,
+      "learning_rate": 4.77141016453488e-06,
+      "loss": 0.9945,
+      "step": 31580
+    },
+    {
+      "epoch": 0.22866946079176528,
+      "grad_norm": 0.17504815757274628,
+      "learning_rate": 4.7713377778742935e-06,
+      "loss": 1.0095,
+      "step": 31590
+    },
+    {
+      "epoch": 0.2287418474523515,
+      "grad_norm": 0.16557878255844116,
+      "learning_rate": 4.771265391213707e-06,
+      "loss": 0.9934,
+      "step": 31600
+    },
+    {
+      "epoch": 0.22881423411293766,
+      "grad_norm": 0.1588304191827774,
+      "learning_rate": 4.771193004553122e-06,
+      "loss": 0.9674,
+      "step": 31610
+    },
+    {
+      "epoch": 0.22888662077352384,
+      "grad_norm": 0.17581112682819366,
+      "learning_rate": 4.771120617892535e-06,
+      "loss": 0.9903,
+      "step": 31620
+    },
+    {
+      "epoch": 0.22895900743411005,
+      "grad_norm": 0.15916821360588074,
+      "learning_rate": 4.771048231231949e-06,
+      "loss": 0.9846,
+      "step": 31630
+    },
+    {
+      "epoch": 0.22903139409469622,
+      "grad_norm": 0.16646379232406616,
+      "learning_rate": 4.7709758445713625e-06,
+      "loss": 0.9747,
+      "step": 31640
+    },
+    {
+      "epoch": 0.22910378075528243,
+      "grad_norm": 0.18210920691490173,
+      "learning_rate": 4.770903457910777e-06,
+      "loss": 0.9999,
+      "step": 31650
+    },
+    {
+      "epoch": 0.2291761674158686,
+      "grad_norm": 0.1829901784658432,
+      "learning_rate": 4.7708310712501905e-06,
+      "loss": 0.9917,
+      "step": 31660
+    },
+    {
+      "epoch": 0.22924855407645478,
+      "grad_norm": 0.16289442777633667,
+      "learning_rate": 4.770758684589604e-06,
+      "loss": 0.9858,
+      "step": 31670
+    },
+    {
+      "epoch": 0.22932094073704098,
+      "grad_norm": 0.1644997000694275,
+      "learning_rate": 4.770686297929018e-06,
+      "loss": 0.9918,
+      "step": 31680
+    },
+    {
+      "epoch": 0.22939332739762716,
+      "grad_norm": 0.1583040952682495,
+      "learning_rate": 4.770613911268432e-06,
+      "loss": 0.9852,
+      "step": 31690
+    },
+    {
+      "epoch": 0.22946571405821337,
+      "grad_norm": 0.15572409331798553,
+      "learning_rate": 4.770541524607846e-06,
+      "loss": 1.0034,
+      "step": 31700
+    },
+    {
+      "epoch": 0.22953810071879954,
+      "grad_norm": 0.185507133603096,
+      "learning_rate": 4.7704691379472595e-06,
+      "loss": 0.996,
+      "step": 31710
+    },
+    {
+      "epoch": 0.22961048737938572,
+      "grad_norm": 0.16780561208724976,
+      "learning_rate": 4.770396751286673e-06,
+      "loss": 0.9902,
+      "step": 31720
+    },
+    {
+      "epoch": 0.22968287403997192,
+      "grad_norm": 0.1608486771583557,
+      "learning_rate": 4.7703243646260875e-06,
+      "loss": 0.9956,
+      "step": 31730
+    },
+    {
+      "epoch": 0.2297552607005581,
+      "grad_norm": 0.16754546761512756,
+      "learning_rate": 4.770251977965501e-06,
+      "loss": 0.9933,
+      "step": 31740
+    },
+    {
+      "epoch": 0.22982764736114428,
+      "grad_norm": 0.16332760453224182,
+      "learning_rate": 4.770179591304915e-06,
+      "loss": 0.9819,
+      "step": 31750
+    },
+    {
+      "epoch": 0.22990003402173048,
+      "grad_norm": 0.23022449016571045,
+      "learning_rate": 4.770107204644328e-06,
+      "loss": 0.9909,
+      "step": 31760
+    },
+    {
+      "epoch": 0.22997242068231666,
+      "grad_norm": 0.18331989645957947,
+      "learning_rate": 4.770034817983743e-06,
+      "loss": 1.0003,
+      "step": 31770
+    },
+    {
+      "epoch": 0.23004480734290286,
+      "grad_norm": 0.1962156593799591,
+      "learning_rate": 4.7699624313231565e-06,
+      "loss": 0.9884,
+      "step": 31780
+    },
+    {
+      "epoch": 0.23011719400348904,
+      "grad_norm": 0.16017916798591614,
+      "learning_rate": 4.76989004466257e-06,
+      "loss": 0.9935,
+      "step": 31790
+    },
+    {
+      "epoch": 0.23018958066407522,
+      "grad_norm": 0.16973306238651276,
+      "learning_rate": 4.769817658001984e-06,
+      "loss": 1.0018,
+      "step": 31800
+    },
+    {
+      "epoch": 0.23026196732466142,
+      "grad_norm": 0.1658056229352951,
+      "learning_rate": 4.769745271341398e-06,
+      "loss": 0.9564,
+      "step": 31810
+    },
+    {
+      "epoch": 0.2303343539852476,
+      "grad_norm": 0.1671288162469864,
+      "learning_rate": 4.769672884680812e-06,
+      "loss": 0.9972,
+      "step": 31820
+    },
+    {
+      "epoch": 0.23040674064583377,
+      "grad_norm": 0.17118290066719055,
+      "learning_rate": 4.769600498020225e-06,
+      "loss": 0.9972,
+      "step": 31830
+    },
+    {
+      "epoch": 0.23047912730641998,
+      "grad_norm": 0.1895531266927719,
+      "learning_rate": 4.769528111359639e-06,
+      "loss": 0.9919,
+      "step": 31840
+    },
+    {
+      "epoch": 0.23055151396700616,
+      "grad_norm": 0.1677425503730774,
+      "learning_rate": 4.769455724699053e-06,
+      "loss": 0.9709,
+      "step": 31850
+    },
+    {
+      "epoch": 0.23062390062759236,
+      "grad_norm": 0.15633396804332733,
+      "learning_rate": 4.769383338038467e-06,
+      "loss": 0.9762,
+      "step": 31860
+    },
+    {
+      "epoch": 0.23069628728817854,
+      "grad_norm": 0.15433230996131897,
+      "learning_rate": 4.769310951377881e-06,
+      "loss": 0.9902,
+      "step": 31870
+    },
+    {
+      "epoch": 0.2307686739487647,
+      "grad_norm": 0.17825239896774292,
+      "learning_rate": 4.769238564717294e-06,
+      "loss": 0.9933,
+      "step": 31880
+    },
+    {
+      "epoch": 0.23084106060935092,
+      "grad_norm": 0.1626371145248413,
+      "learning_rate": 4.769166178056708e-06,
+      "loss": 0.9893,
+      "step": 31890
+    },
+    {
+      "epoch": 0.2309134472699371,
+      "grad_norm": 0.17792868614196777,
+      "learning_rate": 4.769093791396122e-06,
+      "loss": 0.9825,
+      "step": 31900
+    },
+    {
+      "epoch": 0.23098583393052327,
+      "grad_norm": 0.1791241466999054,
+      "learning_rate": 4.769021404735536e-06,
+      "loss": 0.9916,
+      "step": 31910
+    },
+    {
+      "epoch": 0.23105822059110948,
+      "grad_norm": 0.1636131852865219,
+      "learning_rate": 4.76894901807495e-06,
+      "loss": 1.0103,
+      "step": 31920
+    },
+    {
+      "epoch": 0.23113060725169565,
+      "grad_norm": 0.1715201586484909,
+      "learning_rate": 4.768876631414363e-06,
+      "loss": 0.988,
+      "step": 31930
+    },
+    {
+      "epoch": 0.23120299391228186,
+      "grad_norm": 0.19205105304718018,
+      "learning_rate": 4.768804244753777e-06,
+      "loss": 0.9843,
+      "step": 31940
+    },
+    {
+      "epoch": 0.23127538057286803,
+      "grad_norm": 0.177320197224617,
+      "learning_rate": 4.7687318580931905e-06,
+      "loss": 1.0003,
+      "step": 31950
+    },
+    {
+      "epoch": 0.2313477672334542,
+      "grad_norm": 0.16277745366096497,
+      "learning_rate": 4.768659471432605e-06,
+      "loss": 0.9858,
+      "step": 31960
+    },
+    {
+      "epoch": 0.23142015389404041,
+      "grad_norm": 0.16934184730052948,
+      "learning_rate": 4.7685870847720186e-06,
+      "loss": 1.0059,
+      "step": 31970
+    },
+    {
+      "epoch": 0.2314925405546266,
+      "grad_norm": 0.17568084597587585,
+      "learning_rate": 4.768514698111432e-06,
+      "loss": 0.9886,
+      "step": 31980
+    },
+    {
+      "epoch": 0.23156492721521277,
+      "grad_norm": 0.16883017122745514,
+      "learning_rate": 4.768442311450846e-06,
+      "loss": 1.0016,
+      "step": 31990
+    },
+    {
+      "epoch": 0.23163731387579897,
+      "grad_norm": 0.170290008187294,
+      "learning_rate": 4.76836992479026e-06,
+      "loss": 0.9923,
+      "step": 32000
+    },
+    {
+      "epoch": 0.23170970053638515,
+      "grad_norm": 0.17787618935108185,
+      "learning_rate": 4.768297538129674e-06,
+      "loss": 0.974,
+      "step": 32010
+    },
+    {
+      "epoch": 0.23178208719697135,
+      "grad_norm": 0.16933457553386688,
+      "learning_rate": 4.7682251514690875e-06,
+      "loss": 0.9861,
+      "step": 32020
+    },
+    {
+      "epoch": 0.23185447385755753,
+      "grad_norm": 0.1682525873184204,
+      "learning_rate": 4.768152764808501e-06,
+      "loss": 0.9785,
+      "step": 32030
+    },
+    {
+      "epoch": 0.2319268605181437,
+      "grad_norm": 0.17858469486236572,
+      "learning_rate": 4.768080378147915e-06,
+      "loss": 0.9788,
+      "step": 32040
+    },
+    {
+      "epoch": 0.2319992471787299,
+      "grad_norm": 0.16218888759613037,
+      "learning_rate": 4.768007991487329e-06,
+      "loss": 0.9851,
+      "step": 32050
+    },
+    {
+      "epoch": 0.2320716338393161,
+      "grad_norm": 0.1646839827299118,
+      "learning_rate": 4.767935604826743e-06,
+      "loss": 0.9798,
+      "step": 32060
+    },
+    {
+      "epoch": 0.23214402049990226,
+      "grad_norm": 0.20934203267097473,
+      "learning_rate": 4.767863218166156e-06,
+      "loss": 0.9786,
+      "step": 32070
+    },
+    {
+      "epoch": 0.23221640716048847,
+      "grad_norm": 0.1663549393415451,
+      "learning_rate": 4.76779083150557e-06,
+      "loss": 0.9759,
+      "step": 32080
+    },
+    {
+      "epoch": 0.23228879382107465,
+      "grad_norm": 0.16773077845573425,
+      "learning_rate": 4.7677184448449845e-06,
+      "loss": 0.997,
+      "step": 32090
+    },
+    {
+      "epoch": 0.23236118048166085,
+      "grad_norm": 0.18589721620082855,
+      "learning_rate": 4.767646058184398e-06,
+      "loss": 0.991,
+      "step": 32100
+    },
+    {
+      "epoch": 0.23243356714224703,
+      "grad_norm": 0.17491789162158966,
+      "learning_rate": 4.767573671523812e-06,
+      "loss": 0.9837,
+      "step": 32110
+    },
+    {
+      "epoch": 0.2325059538028332,
+      "grad_norm": 0.18429701030254364,
+      "learning_rate": 4.767501284863225e-06,
+      "loss": 0.9795,
+      "step": 32120
+    },
+    {
+      "epoch": 0.2325783404634194,
+      "grad_norm": 0.16141755878925323,
+      "learning_rate": 4.76742889820264e-06,
+      "loss": 1.0045,
+      "step": 32130
+    },
+    {
+      "epoch": 0.23265072712400559,
+      "grad_norm": 0.18277022242546082,
+      "learning_rate": 4.7673565115420534e-06,
+      "loss": 0.9937,
+      "step": 32140
+    },
+    {
+      "epoch": 0.23272311378459176,
+      "grad_norm": 0.17076079547405243,
+      "learning_rate": 4.767284124881467e-06,
+      "loss": 0.9845,
+      "step": 32150
+    },
+    {
+      "epoch": 0.23279550044517797,
+      "grad_norm": 0.20235513150691986,
+      "learning_rate": 4.767211738220881e-06,
+      "loss": 0.9765,
+      "step": 32160
+    },
+    {
+      "epoch": 0.23286788710576414,
+      "grad_norm": 0.2140243798494339,
+      "learning_rate": 4.767139351560295e-06,
+      "loss": 0.9939,
+      "step": 32170
+    },
+    {
+      "epoch": 0.23294027376635035,
+      "grad_norm": 0.17782087624073029,
+      "learning_rate": 4.767066964899709e-06,
+      "loss": 0.9904,
+      "step": 32180
+    },
+    {
+      "epoch": 0.23301266042693652,
+      "grad_norm": 0.15998302400112152,
+      "learning_rate": 4.766994578239122e-06,
+      "loss": 0.9819,
+      "step": 32190
+    },
+    {
+      "epoch": 0.2330850470875227,
+      "grad_norm": 0.17034156620502472,
+      "learning_rate": 4.766922191578536e-06,
+      "loss": 0.9825,
+      "step": 32200
+    },
+    {
+      "epoch": 0.2331574337481089,
+      "grad_norm": 0.1846659928560257,
+      "learning_rate": 4.7668498049179504e-06,
+      "loss": 0.9788,
+      "step": 32210
+    },
+    {
+      "epoch": 0.23322982040869508,
+      "grad_norm": 0.1705406755208969,
+      "learning_rate": 4.766777418257364e-06,
+      "loss": 0.9898,
+      "step": 32220
+    },
+    {
+      "epoch": 0.2333022070692813,
+      "grad_norm": 0.16298960149288177,
+      "learning_rate": 4.766705031596778e-06,
+      "loss": 1.008,
+      "step": 32230
+    },
+    {
+      "epoch": 0.23337459372986746,
+      "grad_norm": 0.1563328206539154,
+      "learning_rate": 4.766632644936191e-06,
+      "loss": 0.9991,
+      "step": 32240
+    },
+    {
+      "epoch": 0.23344698039045364,
+      "grad_norm": 0.19101522862911224,
+      "learning_rate": 4.766560258275606e-06,
+      "loss": 0.9866,
+      "step": 32250
+    },
+    {
+      "epoch": 0.23351936705103984,
+      "grad_norm": 0.16439121961593628,
+      "learning_rate": 4.766487871615019e-06,
+      "loss": 0.9864,
+      "step": 32260
+    },
+    {
+      "epoch": 0.23359175371162602,
+      "grad_norm": 0.16966257989406586,
+      "learning_rate": 4.766415484954433e-06,
+      "loss": 0.99,
+      "step": 32270
+    },
+    {
+      "epoch": 0.2336641403722122,
+      "grad_norm": 0.17119558155536652,
+      "learning_rate": 4.766343098293847e-06,
+      "loss": 0.985,
+      "step": 32280
+    },
+    {
+      "epoch": 0.2337365270327984,
+      "grad_norm": 0.16533613204956055,
+      "learning_rate": 4.766270711633261e-06,
+      "loss": 0.9725,
+      "step": 32290
+    },
+    {
+      "epoch": 0.23380891369338458,
+      "grad_norm": 0.21587282419204712,
+      "learning_rate": 4.766198324972675e-06,
+      "loss": 0.9851,
+      "step": 32300
+    },
+    {
+      "epoch": 0.23388130035397078,
+      "grad_norm": 0.1782499998807907,
+      "learning_rate": 4.766125938312088e-06,
+      "loss": 0.9908,
+      "step": 32310
+    },
+    {
+      "epoch": 0.23395368701455696,
+      "grad_norm": 0.17237718403339386,
+      "learning_rate": 4.766053551651502e-06,
+      "loss": 0.9953,
+      "step": 32320
+    },
+    {
+      "epoch": 0.23402607367514314,
+      "grad_norm": 0.16807182133197784,
+      "learning_rate": 4.765981164990916e-06,
+      "loss": 0.9937,
+      "step": 32330
+    },
+    {
+      "epoch": 0.23409846033572934,
+      "grad_norm": 0.17400771379470825,
+      "learning_rate": 4.76590877833033e-06,
+      "loss": 0.9917,
+      "step": 32340
+    },
+    {
+      "epoch": 0.23417084699631552,
+      "grad_norm": 0.16965161263942719,
+      "learning_rate": 4.765836391669744e-06,
+      "loss": 0.9828,
+      "step": 32350
+    },
+    {
+      "epoch": 0.2342432336569017,
+      "grad_norm": 0.17362885177135468,
+      "learning_rate": 4.765764005009157e-06,
+      "loss": 0.9778,
+      "step": 32360
+    },
+    {
+      "epoch": 0.2343156203174879,
+      "grad_norm": 0.16530746221542358,
+      "learning_rate": 4.765691618348572e-06,
+      "loss": 1.0049,
+      "step": 32370
+    },
+    {
+      "epoch": 0.23438800697807408,
+      "grad_norm": 0.19061051309108734,
+      "learning_rate": 4.765619231687985e-06,
+      "loss": 1.0037,
+      "step": 32380
+    },
+    {
+      "epoch": 0.23446039363866028,
+      "grad_norm": 0.19029873609542847,
+      "learning_rate": 4.765546845027399e-06,
+      "loss": 0.9754,
+      "step": 32390
+    },
+    {
+      "epoch": 0.23453278029924646,
+      "grad_norm": 0.17064446210861206,
+      "learning_rate": 4.7654744583668125e-06,
+      "loss": 0.996,
+      "step": 32400
+    },
+    {
+      "epoch": 0.23460516695983263,
+      "grad_norm": 0.17918898165225983,
+      "learning_rate": 4.765402071706227e-06,
+      "loss": 1.0014,
+      "step": 32410
+    },
+    {
+      "epoch": 0.23467755362041884,
+      "grad_norm": 0.16727639734745026,
+      "learning_rate": 4.765329685045641e-06,
+      "loss": 0.9911,
+      "step": 32420
+    },
+    {
+      "epoch": 0.23474994028100502,
+      "grad_norm": 0.16551131010055542,
+      "learning_rate": 4.765257298385054e-06,
+      "loss": 1.0024,
+      "step": 32430
+    },
+    {
+      "epoch": 0.2348223269415912,
+      "grad_norm": 0.18284402787685394,
+      "learning_rate": 4.765184911724468e-06,
+      "loss": 0.9876,
+      "step": 32440
+    },
+    {
+      "epoch": 0.2348947136021774,
+      "grad_norm": 0.1902831643819809,
+      "learning_rate": 4.765112525063882e-06,
+      "loss": 0.9884,
+      "step": 32450
+    },
+    {
+      "epoch": 0.23496710026276357,
+      "grad_norm": 0.19926203787326813,
+      "learning_rate": 4.765040138403295e-06,
+      "loss": 0.9889,
+      "step": 32460
+    },
+    {
+      "epoch": 0.23503948692334978,
+      "grad_norm": 0.1620095819234848,
+      "learning_rate": 4.764967751742709e-06,
+      "loss": 0.9882,
+      "step": 32470
+    },
+    {
+      "epoch": 0.23511187358393595,
+      "grad_norm": 0.19026455283164978,
+      "learning_rate": 4.764895365082123e-06,
+      "loss": 0.9652,
+      "step": 32480
+    },
+    {
+      "epoch": 0.23518426024452213,
+      "grad_norm": 0.18305736780166626,
+      "learning_rate": 4.764822978421537e-06,
+      "loss": 0.984,
+      "step": 32490
+    },
+    {
+      "epoch": 0.23525664690510834,
+      "grad_norm": 0.16484969854354858,
+      "learning_rate": 4.76475059176095e-06,
+      "loss": 0.9798,
+      "step": 32500
+    },
+    {
+      "epoch": 0.2353290335656945,
+      "grad_norm": 0.1810574233531952,
+      "learning_rate": 4.764678205100364e-06,
+      "loss": 0.9762,
+      "step": 32510
+    },
+    {
+      "epoch": 0.2354014202262807,
+      "grad_norm": 0.16701748967170715,
+      "learning_rate": 4.7646058184397785e-06,
+      "loss": 0.9907,
+      "step": 32520
+    },
+    {
+      "epoch": 0.2354738068868669,
+      "grad_norm": 0.17856666445732117,
+      "learning_rate": 4.764533431779192e-06,
+      "loss": 0.9772,
+      "step": 32530
+    },
+    {
+      "epoch": 0.23554619354745307,
+      "grad_norm": 0.19198378920555115,
+      "learning_rate": 4.764461045118606e-06,
+      "loss": 0.9729,
+      "step": 32540
+    },
+    {
+      "epoch": 0.23561858020803927,
+      "grad_norm": 0.18789614737033844,
+      "learning_rate": 4.764388658458019e-06,
+      "loss": 0.986,
+      "step": 32550
+    },
+    {
+      "epoch": 0.23569096686862545,
+      "grad_norm": 0.1816016137599945,
+      "learning_rate": 4.764316271797434e-06,
+      "loss": 0.9902,
+      "step": 32560
+    },
+    {
+      "epoch": 0.23576335352921163,
+      "grad_norm": 0.17151084542274475,
+      "learning_rate": 4.764243885136847e-06,
+      "loss": 1.0051,
+      "step": 32570
+    },
+    {
+      "epoch": 0.23583574018979783,
+      "grad_norm": 0.16930073499679565,
+      "learning_rate": 4.764171498476261e-06,
+      "loss": 0.9954,
+      "step": 32580
+    },
+    {
+      "epoch": 0.235908126850384,
+      "grad_norm": 0.16622294485569,
+      "learning_rate": 4.764099111815675e-06,
+      "loss": 0.9795,
+      "step": 32590
+    },
+    {
+      "epoch": 0.23598051351097019,
+      "grad_norm": 0.1712275892496109,
+      "learning_rate": 4.764026725155089e-06,
+      "loss": 0.978,
+      "step": 32600
+    },
+    {
+      "epoch": 0.2360529001715564,
+      "grad_norm": 0.17688055336475372,
+      "learning_rate": 4.763954338494503e-06,
+      "loss": 0.9731,
+      "step": 32610
+    },
+    {
+      "epoch": 0.23612528683214257,
+      "grad_norm": 0.1737777143716812,
+      "learning_rate": 4.763881951833916e-06,
+      "loss": 0.9967,
+      "step": 32620
+    },
+    {
+      "epoch": 0.23619767349272877,
+      "grad_norm": 0.2481534481048584,
+      "learning_rate": 4.76380956517333e-06,
+      "loss": 0.9943,
+      "step": 32630
+    },
+    {
+      "epoch": 0.23627006015331495,
+      "grad_norm": 0.1767176389694214,
+      "learning_rate": 4.7637371785127436e-06,
+      "loss": 0.9884,
+      "step": 32640
+    },
+    {
+      "epoch": 0.23634244681390112,
+      "grad_norm": 0.17765220999717712,
+      "learning_rate": 4.763664791852158e-06,
+      "loss": 0.9768,
+      "step": 32650
+    },
+    {
+      "epoch": 0.23641483347448733,
+      "grad_norm": 0.17526815831661224,
+      "learning_rate": 4.763592405191572e-06,
+      "loss": 0.9778,
+      "step": 32660
+    },
+    {
+      "epoch": 0.2364872201350735,
+      "grad_norm": 0.22117316722869873,
+      "learning_rate": 4.763520018530985e-06,
+      "loss": 0.9754,
+      "step": 32670
+    },
+    {
+      "epoch": 0.23655960679565968,
+      "grad_norm": 0.17404435575008392,
+      "learning_rate": 4.763447631870399e-06,
+      "loss": 0.9928,
+      "step": 32680
+    },
+    {
+      "epoch": 0.2366319934562459,
+      "grad_norm": 0.17994847893714905,
+      "learning_rate": 4.763375245209813e-06,
+      "loss": 0.9899,
+      "step": 32690
+    },
+    {
+      "epoch": 0.23670438011683206,
+      "grad_norm": 0.2047555297613144,
+      "learning_rate": 4.763302858549227e-06,
+      "loss": 0.986,
+      "step": 32700
+    },
+    {
+      "epoch": 0.23677676677741827,
+      "grad_norm": 0.19384850561618805,
+      "learning_rate": 4.7632304718886406e-06,
+      "loss": 0.9829,
+      "step": 32710
+    },
+    {
+      "epoch": 0.23684915343800444,
+      "grad_norm": 0.18670782446861267,
+      "learning_rate": 4.763158085228054e-06,
+      "loss": 0.979,
+      "step": 32720
+    },
+    {
+      "epoch": 0.23692154009859062,
+      "grad_norm": 0.1731204390525818,
+      "learning_rate": 4.763085698567469e-06,
+      "loss": 0.9934,
+      "step": 32730
+    },
+    {
+      "epoch": 0.23699392675917683,
+      "grad_norm": 0.1660483479499817,
+      "learning_rate": 4.763013311906882e-06,
+      "loss": 0.9775,
+      "step": 32740
+    },
+    {
+      "epoch": 0.237066313419763,
+      "grad_norm": 0.1656530648469925,
+      "learning_rate": 4.762940925246296e-06,
+      "loss": 0.9899,
+      "step": 32750
+    },
+    {
+      "epoch": 0.23713870008034918,
+      "grad_norm": 0.16526754200458527,
+      "learning_rate": 4.7628685385857095e-06,
+      "loss": 0.9766,
+      "step": 32760
+    },
+    {
+      "epoch": 0.23721108674093538,
+      "grad_norm": 0.16604174673557281,
+      "learning_rate": 4.762796151925124e-06,
+      "loss": 0.9883,
+      "step": 32770
+    },
+    {
+      "epoch": 0.23728347340152156,
+      "grad_norm": 0.1578882932662964,
+      "learning_rate": 4.7627237652645376e-06,
+      "loss": 0.9888,
+      "step": 32780
+    },
+    {
+      "epoch": 0.23735586006210777,
+      "grad_norm": 0.1657019555568695,
+      "learning_rate": 4.762651378603951e-06,
+      "loss": 0.9808,
+      "step": 32790
+    },
+    {
+      "epoch": 0.23742824672269394,
+      "grad_norm": 0.2883431017398834,
+      "learning_rate": 4.762578991943365e-06,
+      "loss": 0.9934,
+      "step": 32800
+    },
+    {
+      "epoch": 0.23750063338328012,
+      "grad_norm": 0.17559395730495453,
+      "learning_rate": 4.762506605282779e-06,
+      "loss": 0.9899,
+      "step": 32810
+    },
+    {
+      "epoch": 0.23757302004386632,
+      "grad_norm": 0.16590864956378937,
+      "learning_rate": 4.762434218622193e-06,
+      "loss": 0.9745,
+      "step": 32820
+    },
+    {
+      "epoch": 0.2376454067044525,
+      "grad_norm": 0.17122603952884674,
+      "learning_rate": 4.7623618319616065e-06,
+      "loss": 0.9744,
+      "step": 32830
+    },
+    {
+      "epoch": 0.2377177933650387,
+      "grad_norm": 0.16767603158950806,
+      "learning_rate": 4.76228944530102e-06,
+      "loss": 0.9875,
+      "step": 32840
+    },
+    {
+      "epoch": 0.23779018002562488,
+      "grad_norm": 0.16585096716880798,
+      "learning_rate": 4.7622170586404346e-06,
+      "loss": 0.9873,
+      "step": 32850
+    },
+    {
+      "epoch": 0.23786256668621106,
+      "grad_norm": 0.16618821024894714,
+      "learning_rate": 4.762144671979848e-06,
+      "loss": 0.9841,
+      "step": 32860
+    },
+    {
+      "epoch": 0.23793495334679726,
+      "grad_norm": 0.18130016326904297,
+      "learning_rate": 4.762072285319262e-06,
+      "loss": 0.9948,
+      "step": 32870
+    },
+    {
+      "epoch": 0.23800734000738344,
+      "grad_norm": 0.16200359165668488,
+      "learning_rate": 4.7619998986586754e-06,
+      "loss": 0.9913,
+      "step": 32880
+    },
+    {
+      "epoch": 0.23807972666796962,
+      "grad_norm": 0.16042807698249817,
+      "learning_rate": 4.76192751199809e-06,
+      "loss": 0.9835,
+      "step": 32890
+    },
+    {
+      "epoch": 0.23815211332855582,
+      "grad_norm": 0.1726037710905075,
+      "learning_rate": 4.7618551253375035e-06,
+      "loss": 0.9665,
+      "step": 32900
+    },
+    {
+      "epoch": 0.238224499989142,
+      "grad_norm": 0.16508442163467407,
+      "learning_rate": 4.761782738676917e-06,
+      "loss": 1.0037,
+      "step": 32910
+    },
+    {
+      "epoch": 0.2382968866497282,
+      "grad_norm": 0.16878628730773926,
+      "learning_rate": 4.761710352016331e-06,
+      "loss": 0.9897,
+      "step": 32920
+    },
+    {
+      "epoch": 0.23836927331031438,
+      "grad_norm": 0.17071186006069183,
+      "learning_rate": 4.761637965355745e-06,
+      "loss": 1.0066,
+      "step": 32930
+    },
+    {
+      "epoch": 0.23844165997090055,
+      "grad_norm": 0.16093328595161438,
+      "learning_rate": 4.761565578695159e-06,
+      "loss": 0.9938,
+      "step": 32940
+    },
+    {
+      "epoch": 0.23851404663148676,
+      "grad_norm": 0.181466743350029,
+      "learning_rate": 4.7614931920345724e-06,
+      "loss": 0.9863,
+      "step": 32950
+    },
+    {
+      "epoch": 0.23858643329207294,
+      "grad_norm": 0.19711147248744965,
+      "learning_rate": 4.761420805373986e-06,
+      "loss": 0.9867,
+      "step": 32960
+    },
+    {
+      "epoch": 0.2386588199526591,
+      "grad_norm": 0.17717312276363373,
+      "learning_rate": 4.7613484187134005e-06,
+      "loss": 0.9923,
+      "step": 32970
+    },
+    {
+      "epoch": 0.23873120661324532,
+      "grad_norm": 0.17629402875900269,
+      "learning_rate": 4.761276032052814e-06,
+      "loss": 1.0001,
+      "step": 32980
+    },
+    {
+      "epoch": 0.2388035932738315,
+      "grad_norm": 0.16870364546775818,
+      "learning_rate": 4.761203645392227e-06,
+      "loss": 0.9904,
+      "step": 32990
+    },
+    {
+      "epoch": 0.2388759799344177,
+      "grad_norm": 0.17565900087356567,
+      "learning_rate": 4.761131258731641e-06,
+      "loss": 0.9871,
+      "step": 33000
+    },
+    {
+      "epoch": 0.23894836659500387,
+      "grad_norm": 0.20653510093688965,
+      "learning_rate": 4.761058872071055e-06,
+      "loss": 0.9983,
+      "step": 33010
+    },
+    {
+      "epoch": 0.23902075325559005,
+      "grad_norm": 0.165169820189476,
+      "learning_rate": 4.760986485410469e-06,
+      "loss": 0.9812,
+      "step": 33020
+    },
+    {
+      "epoch": 0.23909313991617626,
+      "grad_norm": 0.29448917508125305,
+      "learning_rate": 4.760914098749882e-06,
+      "loss": 0.9681,
+      "step": 33030
+    },
+    {
+      "epoch": 0.23916552657676243,
+      "grad_norm": 0.1737428605556488,
+      "learning_rate": 4.760841712089297e-06,
+      "loss": 0.9895,
+      "step": 33040
+    },
+    {
+      "epoch": 0.2392379132373486,
+      "grad_norm": 0.15930554270744324,
+      "learning_rate": 4.76076932542871e-06,
+      "loss": 0.9923,
+      "step": 33050
+    },
+    {
+      "epoch": 0.23931029989793481,
+      "grad_norm": 0.15467441082000732,
+      "learning_rate": 4.760696938768124e-06,
+      "loss": 0.9791,
+      "step": 33060
+    },
+    {
+      "epoch": 0.239382686558521,
+      "grad_norm": 0.16547755897045135,
+      "learning_rate": 4.7606245521075375e-06,
+      "loss": 0.9836,
+      "step": 33070
+    },
+    {
+      "epoch": 0.2394550732191072,
+      "grad_norm": 0.1736575812101364,
+      "learning_rate": 4.760552165446952e-06,
+      "loss": 0.9818,
+      "step": 33080
+    },
+    {
+      "epoch": 0.23952745987969337,
+      "grad_norm": 0.17099529504776,
+      "learning_rate": 4.760479778786366e-06,
+      "loss": 0.9936,
+      "step": 33090
+    },
+    {
+      "epoch": 0.23959984654027955,
+      "grad_norm": 0.16323433816432953,
+      "learning_rate": 4.760407392125779e-06,
+      "loss": 0.9748,
+      "step": 33100
+    },
+    {
+      "epoch": 0.23967223320086575,
+      "grad_norm": 0.174141988158226,
+      "learning_rate": 4.760335005465193e-06,
+      "loss": 0.9935,
+      "step": 33110
+    },
+    {
+      "epoch": 0.23974461986145193,
+      "grad_norm": 0.17322736978530884,
+      "learning_rate": 4.760262618804607e-06,
+      "loss": 0.9852,
+      "step": 33120
+    },
+    {
+      "epoch": 0.2398170065220381,
+      "grad_norm": 0.1788867712020874,
+      "learning_rate": 4.760190232144021e-06,
+      "loss": 0.9934,
+      "step": 33130
+    },
+    {
+      "epoch": 0.2398893931826243,
+      "grad_norm": 0.1551644206047058,
+      "learning_rate": 4.7601178454834345e-06,
+      "loss": 0.9858,
+      "step": 33140
+    },
+    {
+      "epoch": 0.2399617798432105,
+      "grad_norm": 0.16504356265068054,
+      "learning_rate": 4.760045458822848e-06,
+      "loss": 0.974,
+      "step": 33150
+    },
+    {
+      "epoch": 0.2400341665037967,
+      "grad_norm": 0.16401821374893188,
+      "learning_rate": 4.759973072162263e-06,
+      "loss": 0.9849,
+      "step": 33160
+    },
+    {
+      "epoch": 0.24010655316438287,
+      "grad_norm": 0.1985578089952469,
+      "learning_rate": 4.759900685501676e-06,
+      "loss": 0.9823,
+      "step": 33170
+    },
+    {
+      "epoch": 0.24017893982496905,
+      "grad_norm": 0.1741316169500351,
+      "learning_rate": 4.75982829884109e-06,
+      "loss": 0.9926,
+      "step": 33180
+    },
+    {
+      "epoch": 0.24025132648555525,
+      "grad_norm": 0.17418281733989716,
+      "learning_rate": 4.7597559121805035e-06,
+      "loss": 0.9871,
+      "step": 33190
+    },
+    {
+      "epoch": 0.24032371314614143,
+      "grad_norm": 0.16395236551761627,
+      "learning_rate": 4.759683525519918e-06,
+      "loss": 0.9792,
+      "step": 33200
+    },
+    {
+      "epoch": 0.2403960998067276,
+      "grad_norm": 0.1797424554824829,
+      "learning_rate": 4.7596111388593315e-06,
+      "loss": 0.9794,
+      "step": 33210
+    },
+    {
+      "epoch": 0.2404684864673138,
+      "grad_norm": 0.15799680352210999,
+      "learning_rate": 4.759538752198745e-06,
+      "loss": 0.9801,
+      "step": 33220
+    },
+    {
+      "epoch": 0.24054087312789998,
+      "grad_norm": 0.1746169626712799,
+      "learning_rate": 4.759466365538159e-06,
+      "loss": 0.9923,
+      "step": 33230
+    },
+    {
+      "epoch": 0.2406132597884862,
+      "grad_norm": 0.16156165301799774,
+      "learning_rate": 4.759393978877573e-06,
+      "loss": 0.9762,
+      "step": 33240
+    },
+    {
+      "epoch": 0.24068564644907237,
+      "grad_norm": 0.1618238091468811,
+      "learning_rate": 4.759321592216987e-06,
+      "loss": 0.9839,
+      "step": 33250
+    },
+    {
+      "epoch": 0.24075803310965854,
+      "grad_norm": 0.17595741152763367,
+      "learning_rate": 4.7592492055564005e-06,
+      "loss": 0.9902,
+      "step": 33260
+    },
+    {
+      "epoch": 0.24083041977024475,
+      "grad_norm": 0.16890239715576172,
+      "learning_rate": 4.759176818895814e-06,
+      "loss": 0.9763,
+      "step": 33270
+    },
+    {
+      "epoch": 0.24090280643083092,
+      "grad_norm": 0.17769628763198853,
+      "learning_rate": 4.759104432235228e-06,
+      "loss": 0.9644,
+      "step": 33280
+    },
+    {
+      "epoch": 0.2409751930914171,
+      "grad_norm": 0.17169831693172455,
+      "learning_rate": 4.759032045574642e-06,
+      "loss": 0.9999,
+      "step": 33290
+    },
+    {
+      "epoch": 0.2410475797520033,
+      "grad_norm": 0.16996458172798157,
+      "learning_rate": 4.758959658914056e-06,
+      "loss": 0.9833,
+      "step": 33300
+    },
+    {
+      "epoch": 0.24111996641258948,
+      "grad_norm": 0.1796058863401413,
+      "learning_rate": 4.758887272253469e-06,
+      "loss": 0.9757,
+      "step": 33310
+    },
+    {
+      "epoch": 0.24119235307317569,
+      "grad_norm": 0.17819707095623016,
+      "learning_rate": 4.758814885592883e-06,
+      "loss": 0.9874,
+      "step": 33320
+    },
+    {
+      "epoch": 0.24126473973376186,
+      "grad_norm": 0.16091227531433105,
+      "learning_rate": 4.7587424989322975e-06,
+      "loss": 0.9809,
+      "step": 33330
+    },
+    {
+      "epoch": 0.24133712639434804,
+      "grad_norm": 0.17495249211788177,
+      "learning_rate": 4.758670112271711e-06,
+      "loss": 0.9992,
+      "step": 33340
+    },
+    {
+      "epoch": 0.24140951305493424,
+      "grad_norm": 0.16773755848407745,
+      "learning_rate": 4.758597725611125e-06,
+      "loss": 0.9884,
+      "step": 33350
+    },
+    {
+      "epoch": 0.24148189971552042,
+      "grad_norm": 0.23348310589790344,
+      "learning_rate": 4.758525338950538e-06,
+      "loss": 0.9905,
+      "step": 33360
+    },
+    {
+      "epoch": 0.2415542863761066,
+      "grad_norm": 0.1686268001794815,
+      "learning_rate": 4.758452952289953e-06,
+      "loss": 1.0038,
+      "step": 33370
+    },
+    {
+      "epoch": 0.2416266730366928,
+      "grad_norm": 0.15924613177776337,
+      "learning_rate": 4.758380565629366e-06,
+      "loss": 0.9911,
+      "step": 33380
+    },
+    {
+      "epoch": 0.24169905969727898,
+      "grad_norm": 0.19010992348194122,
+      "learning_rate": 4.75830817896878e-06,
+      "loss": 0.9824,
+      "step": 33390
+    },
+    {
+      "epoch": 0.24177144635786518,
+      "grad_norm": 0.18180611729621887,
+      "learning_rate": 4.758235792308194e-06,
+      "loss": 0.9813,
+      "step": 33400
+    },
+    {
+      "epoch": 0.24184383301845136,
+      "grad_norm": 0.1697949916124344,
+      "learning_rate": 4.758163405647608e-06,
+      "loss": 0.9783,
+      "step": 33410
+    },
+    {
+      "epoch": 0.24191621967903754,
+      "grad_norm": 0.1699085682630539,
+      "learning_rate": 4.758091018987022e-06,
+      "loss": 0.9754,
+      "step": 33420
+    },
+    {
+      "epoch": 0.24198860633962374,
+      "grad_norm": 0.18635696172714233,
+      "learning_rate": 4.758018632326435e-06,
+      "loss": 0.9779,
+      "step": 33430
+    },
+    {
+      "epoch": 0.24206099300020992,
+      "grad_norm": 0.17529450356960297,
+      "learning_rate": 4.757946245665849e-06,
+      "loss": 0.9768,
+      "step": 33440
+    },
+    {
+      "epoch": 0.24213337966079612,
+      "grad_norm": 0.1667739450931549,
+      "learning_rate": 4.757873859005263e-06,
+      "loss": 0.9699,
+      "step": 33450
+    },
+    {
+      "epoch": 0.2422057663213823,
+      "grad_norm": 0.17115871608257294,
+      "learning_rate": 4.757801472344677e-06,
+      "loss": 0.9964,
+      "step": 33460
+    },
+    {
+      "epoch": 0.24227815298196848,
+      "grad_norm": 0.19105277955532074,
+      "learning_rate": 4.757729085684091e-06,
+      "loss": 0.9917,
+      "step": 33470
+    },
+    {
+      "epoch": 0.24235053964255468,
+      "grad_norm": 0.16427116096019745,
+      "learning_rate": 4.757656699023504e-06,
+      "loss": 0.9769,
+      "step": 33480
+    },
+    {
+      "epoch": 0.24242292630314086,
+      "grad_norm": 0.17592819035053253,
+      "learning_rate": 4.757584312362919e-06,
+      "loss": 0.9819,
+      "step": 33490
+    },
+    {
+      "epoch": 0.24249531296372703,
+      "grad_norm": 0.22349461913108826,
+      "learning_rate": 4.757511925702332e-06,
+      "loss": 0.9884,
+      "step": 33500
+    },
+    {
+      "epoch": 0.24256769962431324,
+      "grad_norm": 0.15861837565898895,
+      "learning_rate": 4.757439539041746e-06,
+      "loss": 0.9811,
+      "step": 33510
+    },
+    {
+      "epoch": 0.24264008628489941,
+      "grad_norm": 0.1849350929260254,
+      "learning_rate": 4.7573671523811596e-06,
+      "loss": 0.9861,
+      "step": 33520
+    },
+    {
+      "epoch": 0.24271247294548562,
+      "grad_norm": 0.18118155002593994,
+      "learning_rate": 4.757294765720573e-06,
+      "loss": 0.9915,
+      "step": 33530
+    },
+    {
+      "epoch": 0.2427848596060718,
+      "grad_norm": 0.1626245677471161,
+      "learning_rate": 4.757222379059987e-06,
+      "loss": 1.0021,
+      "step": 33540
+    },
+    {
+      "epoch": 0.24285724626665797,
+      "grad_norm": 0.1634633094072342,
+      "learning_rate": 4.7571499923994e-06,
+      "loss": 0.982,
+      "step": 33550
+    },
+    {
+      "epoch": 0.24292963292724418,
+      "grad_norm": 0.23890525102615356,
+      "learning_rate": 4.757077605738815e-06,
+      "loss": 0.9864,
+      "step": 33560
+    },
+    {
+      "epoch": 0.24300201958783035,
+      "grad_norm": 0.1567389816045761,
+      "learning_rate": 4.7570052190782285e-06,
+      "loss": 0.9626,
+      "step": 33570
+    },
+    {
+      "epoch": 0.24307440624841653,
+      "grad_norm": 0.16346383094787598,
+      "learning_rate": 4.756932832417642e-06,
+      "loss": 0.9792,
+      "step": 33580
+    },
+    {
+      "epoch": 0.24314679290900273,
+      "grad_norm": 0.1675373762845993,
+      "learning_rate": 4.756860445757056e-06,
+      "loss": 0.9689,
+      "step": 33590
+    },
+    {
+      "epoch": 0.2432191795695889,
+      "grad_norm": 0.1947060525417328,
+      "learning_rate": 4.75678805909647e-06,
+      "loss": 0.9719,
+      "step": 33600
+    },
+    {
+      "epoch": 0.24329156623017512,
+      "grad_norm": 0.17084960639476776,
+      "learning_rate": 4.756715672435884e-06,
+      "loss": 0.9709,
+      "step": 33610
+    },
+    {
+      "epoch": 0.2433639528907613,
+      "grad_norm": 0.16235877573490143,
+      "learning_rate": 4.756643285775297e-06,
+      "loss": 0.9808,
+      "step": 33620
+    },
+    {
+      "epoch": 0.24343633955134747,
+      "grad_norm": 0.16720472276210785,
+      "learning_rate": 4.756570899114711e-06,
+      "loss": 0.9817,
+      "step": 33630
+    },
+    {
+      "epoch": 0.24350872621193367,
+      "grad_norm": 0.16974382102489471,
+      "learning_rate": 4.7564985124541255e-06,
+      "loss": 0.9774,
+      "step": 33640
+    },
+    {
+      "epoch": 0.24358111287251985,
+      "grad_norm": 0.16924746334552765,
+      "learning_rate": 4.756426125793539e-06,
+      "loss": 0.9672,
+      "step": 33650
+    },
+    {
+      "epoch": 0.24365349953310603,
+      "grad_norm": 0.19946692883968353,
+      "learning_rate": 4.756353739132953e-06,
+      "loss": 1.0038,
+      "step": 33660
+    },
+    {
+      "epoch": 0.24372588619369223,
+      "grad_norm": 0.17359711229801178,
+      "learning_rate": 4.756281352472366e-06,
+      "loss": 0.9763,
+      "step": 33670
+    },
+    {
+      "epoch": 0.2437982728542784,
+      "grad_norm": 0.1743346005678177,
+      "learning_rate": 4.756208965811781e-06,
+      "loss": 0.9859,
+      "step": 33680
+    },
+    {
+      "epoch": 0.2438706595148646,
+      "grad_norm": 0.1658143252134323,
+      "learning_rate": 4.7561365791511944e-06,
+      "loss": 1.0053,
+      "step": 33690
+    },
+    {
+      "epoch": 0.2439430461754508,
+      "grad_norm": 0.17681175470352173,
+      "learning_rate": 4.756064192490608e-06,
+      "loss": 0.9842,
+      "step": 33700
+    },
+    {
+      "epoch": 0.24401543283603697,
+      "grad_norm": 0.17499324679374695,
+      "learning_rate": 4.755991805830022e-06,
+      "loss": 0.9822,
+      "step": 33710
+    },
+    {
+      "epoch": 0.24408781949662317,
+      "grad_norm": 0.15076082944869995,
+      "learning_rate": 4.755919419169436e-06,
+      "loss": 0.97,
+      "step": 33720
+    },
+    {
+      "epoch": 0.24416020615720935,
+      "grad_norm": 0.17643196880817413,
+      "learning_rate": 4.75584703250885e-06,
+      "loss": 0.9888,
+      "step": 33730
+    },
+    {
+      "epoch": 0.24423259281779552,
+      "grad_norm": 0.24239002168178558,
+      "learning_rate": 4.755774645848263e-06,
+      "loss": 0.9886,
+      "step": 33740
+    },
+    {
+      "epoch": 0.24430497947838173,
+      "grad_norm": 0.16595035791397095,
+      "learning_rate": 4.755702259187677e-06,
+      "loss": 0.991,
+      "step": 33750
+    },
+    {
+      "epoch": 0.2443773661389679,
+      "grad_norm": 0.1724718064069748,
+      "learning_rate": 4.7556298725270914e-06,
+      "loss": 0.9744,
+      "step": 33760
+    },
+    {
+      "epoch": 0.2444497527995541,
+      "grad_norm": 0.17158448696136475,
+      "learning_rate": 4.755557485866505e-06,
+      "loss": 0.9853,
+      "step": 33770
+    },
+    {
+      "epoch": 0.2445221394601403,
+      "grad_norm": 0.23816993832588196,
+      "learning_rate": 4.755485099205919e-06,
+      "loss": 1.0021,
+      "step": 33780
+    },
+    {
+      "epoch": 0.24459452612072646,
+      "grad_norm": 0.16831299662590027,
+      "learning_rate": 4.755412712545332e-06,
+      "loss": 0.9886,
+      "step": 33790
+    },
+    {
+      "epoch": 0.24466691278131267,
+      "grad_norm": 0.1661224067211151,
+      "learning_rate": 4.755340325884747e-06,
+      "loss": 1.002,
+      "step": 33800
+    },
+    {
+      "epoch": 0.24473929944189884,
+      "grad_norm": 0.16861529648303986,
+      "learning_rate": 4.75526793922416e-06,
+      "loss": 0.9858,
+      "step": 33810
+    },
+    {
+      "epoch": 0.24481168610248502,
+      "grad_norm": 0.17031748592853546,
+      "learning_rate": 4.755195552563574e-06,
+      "loss": 0.9806,
+      "step": 33820
+    },
+    {
+      "epoch": 0.24488407276307123,
+      "grad_norm": 0.17578351497650146,
+      "learning_rate": 4.755123165902988e-06,
+      "loss": 0.9785,
+      "step": 33830
+    },
+    {
+      "epoch": 0.2449564594236574,
+      "grad_norm": 0.18001072108745575,
+      "learning_rate": 4.755050779242402e-06,
+      "loss": 0.9939,
+      "step": 33840
+    },
+    {
+      "epoch": 0.2450288460842436,
+      "grad_norm": 0.18239019811153412,
+      "learning_rate": 4.754978392581816e-06,
+      "loss": 0.9748,
+      "step": 33850
+    },
+    {
+      "epoch": 0.24510123274482978,
+      "grad_norm": 0.17754653096199036,
+      "learning_rate": 4.754906005921229e-06,
+      "loss": 0.989,
+      "step": 33860
+    },
+    {
+      "epoch": 0.24517361940541596,
+      "grad_norm": 0.1725796014070511,
+      "learning_rate": 4.754833619260643e-06,
+      "loss": 0.9888,
+      "step": 33870
+    },
+    {
+      "epoch": 0.24524600606600216,
+      "grad_norm": 0.17696940898895264,
+      "learning_rate": 4.754761232600057e-06,
+      "loss": 0.9773,
+      "step": 33880
+    },
+    {
+      "epoch": 0.24531839272658834,
+      "grad_norm": 0.18559856712818146,
+      "learning_rate": 4.754688845939471e-06,
+      "loss": 0.9736,
+      "step": 33890
+    },
+    {
+      "epoch": 0.24539077938717452,
+      "grad_norm": 0.17658168077468872,
+      "learning_rate": 4.754616459278885e-06,
+      "loss": 1.0012,
+      "step": 33900
+    },
+    {
+      "epoch": 0.24546316604776072,
+      "grad_norm": 0.15067683160305023,
+      "learning_rate": 4.754544072618298e-06,
+      "loss": 0.982,
+      "step": 33910
+    },
+    {
+      "epoch": 0.2455355527083469,
+      "grad_norm": 0.18849797546863556,
+      "learning_rate": 4.754471685957712e-06,
+      "loss": 0.9798,
+      "step": 33920
+    },
+    {
+      "epoch": 0.2456079393689331,
+      "grad_norm": 0.16818571090698242,
+      "learning_rate": 4.754399299297126e-06,
+      "loss": 0.9772,
+      "step": 33930
+    },
+    {
+      "epoch": 0.24568032602951928,
+      "grad_norm": 0.17547225952148438,
+      "learning_rate": 4.75432691263654e-06,
+      "loss": 0.9829,
+      "step": 33940
+    },
+    {
+      "epoch": 0.24575271269010546,
+      "grad_norm": 0.17536889016628265,
+      "learning_rate": 4.7542545259759535e-06,
+      "loss": 0.9678,
+      "step": 33950
+    },
+    {
+      "epoch": 0.24582509935069166,
+      "grad_norm": 0.17033427953720093,
+      "learning_rate": 4.754182139315367e-06,
+      "loss": 0.9967,
+      "step": 33960
+    },
+    {
+      "epoch": 0.24589748601127784,
+      "grad_norm": 0.18375976383686066,
+      "learning_rate": 4.754109752654782e-06,
+      "loss": 0.9946,
+      "step": 33970
+    },
+    {
+      "epoch": 0.24596987267186404,
+      "grad_norm": 0.15287001430988312,
+      "learning_rate": 4.754037365994195e-06,
+      "loss": 0.9848,
+      "step": 33980
+    },
+    {
+      "epoch": 0.24604225933245022,
+      "grad_norm": 0.17237313091754913,
+      "learning_rate": 4.753964979333609e-06,
+      "loss": 0.9876,
+      "step": 33990
+    },
+    {
+      "epoch": 0.2461146459930364,
+      "grad_norm": 0.16331906616687775,
+      "learning_rate": 4.7538925926730225e-06,
+      "loss": 0.9934,
+      "step": 34000
+    },
+    {
+      "epoch": 0.2461870326536226,
+      "grad_norm": 0.17287370562553406,
+      "learning_rate": 4.753820206012437e-06,
+      "loss": 0.9888,
+      "step": 34010
+    },
+    {
+      "epoch": 0.24625941931420878,
+      "grad_norm": 0.16767475008964539,
+      "learning_rate": 4.7537478193518505e-06,
+      "loss": 0.9969,
+      "step": 34020
+    },
+    {
+      "epoch": 0.24633180597479495,
+      "grad_norm": 0.1597083956003189,
+      "learning_rate": 4.753675432691264e-06,
+      "loss": 0.9748,
+      "step": 34030
+    },
+    {
+      "epoch": 0.24640419263538116,
+      "grad_norm": 0.17577607929706573,
+      "learning_rate": 4.753603046030678e-06,
+      "loss": 0.9894,
+      "step": 34040
+    },
+    {
+      "epoch": 0.24647657929596734,
+      "grad_norm": 0.1721176654100418,
+      "learning_rate": 4.753530659370091e-06,
+      "loss": 0.9785,
+      "step": 34050
+    },
+    {
+      "epoch": 0.24654896595655354,
+      "grad_norm": 0.17841075360774994,
+      "learning_rate": 4.753458272709505e-06,
+      "loss": 0.9734,
+      "step": 34060
+    },
+    {
+      "epoch": 0.24662135261713972,
+      "grad_norm": 0.18525052070617676,
+      "learning_rate": 4.753385886048919e-06,
+      "loss": 0.9842,
+      "step": 34070
+    },
+    {
+      "epoch": 0.2466937392777259,
+      "grad_norm": 0.18916094303131104,
+      "learning_rate": 4.753313499388333e-06,
+      "loss": 0.9919,
+      "step": 34080
+    },
+    {
+      "epoch": 0.2467661259383121,
+      "grad_norm": 0.21833638846874237,
+      "learning_rate": 4.753241112727747e-06,
+      "loss": 0.9952,
+      "step": 34090
+    },
+    {
+      "epoch": 0.24683851259889827,
+      "grad_norm": 0.1600702404975891,
+      "learning_rate": 4.75316872606716e-06,
+      "loss": 0.9731,
+      "step": 34100
+    },
+    {
+      "epoch": 0.24691089925948445,
+      "grad_norm": 0.16110806167125702,
+      "learning_rate": 4.753096339406574e-06,
+      "loss": 0.9743,
+      "step": 34110
+    },
+    {
+      "epoch": 0.24698328592007066,
+      "grad_norm": 0.16948141157627106,
+      "learning_rate": 4.753023952745988e-06,
+      "loss": 0.9904,
+      "step": 34120
+    },
+    {
+      "epoch": 0.24705567258065683,
+      "grad_norm": 0.16437900066375732,
+      "learning_rate": 4.752951566085402e-06,
+      "loss": 0.9714,
+      "step": 34130
+    },
+    {
+      "epoch": 0.24712805924124304,
+      "grad_norm": 0.1707761436700821,
+      "learning_rate": 4.752879179424816e-06,
+      "loss": 0.9948,
+      "step": 34140
+    },
+    {
+      "epoch": 0.2472004459018292,
+      "grad_norm": 0.17766614258289337,
+      "learning_rate": 4.752806792764229e-06,
+      "loss": 0.9811,
+      "step": 34150
+    },
+    {
+      "epoch": 0.2472728325624154,
+      "grad_norm": 0.16982759535312653,
+      "learning_rate": 4.752734406103644e-06,
+      "loss": 0.988,
+      "step": 34160
+    },
+    {
+      "epoch": 0.2473452192230016,
+      "grad_norm": 0.1820429265499115,
+      "learning_rate": 4.752662019443057e-06,
+      "loss": 0.9801,
+      "step": 34170
+    },
+    {
+      "epoch": 0.24741760588358777,
+      "grad_norm": 0.17136429250240326,
+      "learning_rate": 4.752589632782471e-06,
+      "loss": 0.975,
+      "step": 34180
+    },
+    {
+      "epoch": 0.24748999254417395,
+      "grad_norm": 0.17014792561531067,
+      "learning_rate": 4.7525172461218846e-06,
+      "loss": 0.985,
+      "step": 34190
+    },
+    {
+      "epoch": 0.24756237920476015,
+      "grad_norm": 0.17695526778697968,
+      "learning_rate": 4.752444859461299e-06,
+      "loss": 0.9866,
+      "step": 34200
+    },
+    {
+      "epoch": 0.24763476586534633,
+      "grad_norm": 0.18461161851882935,
+      "learning_rate": 4.752372472800713e-06,
+      "loss": 0.9922,
+      "step": 34210
+    },
+    {
+      "epoch": 0.24770715252593253,
+      "grad_norm": 0.23707006871700287,
+      "learning_rate": 4.752300086140126e-06,
+      "loss": 0.9964,
+      "step": 34220
+    },
+    {
+      "epoch": 0.2477795391865187,
+      "grad_norm": 0.16921405494213104,
+      "learning_rate": 4.75222769947954e-06,
+      "loss": 0.9831,
+      "step": 34230
+    },
+    {
+      "epoch": 0.2478519258471049,
+      "grad_norm": 0.18006683886051178,
+      "learning_rate": 4.752155312818954e-06,
+      "loss": 0.9915,
+      "step": 34240
+    },
+    {
+      "epoch": 0.2479243125076911,
+      "grad_norm": 0.1586627960205078,
+      "learning_rate": 4.752082926158368e-06,
+      "loss": 0.9765,
+      "step": 34250
+    },
+    {
+      "epoch": 0.24799669916827727,
+      "grad_norm": 0.17502956092357635,
+      "learning_rate": 4.7520105394977816e-06,
+      "loss": 0.9833,
+      "step": 34260
+    },
+    {
+      "epoch": 0.24806908582886344,
+      "grad_norm": 0.17137262225151062,
+      "learning_rate": 4.751938152837195e-06,
+      "loss": 0.9813,
+      "step": 34270
+    },
+    {
+      "epoch": 0.24814147248944965,
+      "grad_norm": 0.1806812733411789,
+      "learning_rate": 4.75186576617661e-06,
+      "loss": 0.9949,
+      "step": 34280
+    },
+    {
+      "epoch": 0.24821385915003583,
+      "grad_norm": 0.17505182325839996,
+      "learning_rate": 4.751793379516023e-06,
+      "loss": 0.9949,
+      "step": 34290
+    },
+    {
+      "epoch": 0.24828624581062203,
+      "grad_norm": 0.17659908533096313,
+      "learning_rate": 4.751720992855437e-06,
+      "loss": 0.9938,
+      "step": 34300
+    },
+    {
+      "epoch": 0.2483586324712082,
+      "grad_norm": 0.18105976283550262,
+      "learning_rate": 4.7516486061948505e-06,
+      "loss": 0.9778,
+      "step": 34310
+    },
+    {
+      "epoch": 0.24843101913179438,
+      "grad_norm": 0.18048040568828583,
+      "learning_rate": 4.751576219534265e-06,
+      "loss": 0.9918,
+      "step": 34320
+    },
+    {
+      "epoch": 0.2485034057923806,
+      "grad_norm": 0.17734596133232117,
+      "learning_rate": 4.7515038328736786e-06,
+      "loss": 0.9889,
+      "step": 34330
+    },
+    {
+      "epoch": 0.24857579245296677,
+      "grad_norm": 0.16741421818733215,
+      "learning_rate": 4.751431446213092e-06,
+      "loss": 0.9854,
+      "step": 34340
+    },
+    {
+      "epoch": 0.24864817911355294,
+      "grad_norm": 0.17737902700901031,
+      "learning_rate": 4.751359059552506e-06,
+      "loss": 0.9793,
+      "step": 34350
+    },
+    {
+      "epoch": 0.24872056577413915,
+      "grad_norm": 0.1893605887889862,
+      "learning_rate": 4.75128667289192e-06,
+      "loss": 0.9778,
+      "step": 34360
+    },
+    {
+      "epoch": 0.24879295243472532,
+      "grad_norm": 0.16508178412914276,
+      "learning_rate": 4.751214286231334e-06,
+      "loss": 0.9873,
+      "step": 34370
+    },
+    {
+      "epoch": 0.24886533909531153,
+      "grad_norm": 0.1740712970495224,
+      "learning_rate": 4.7511418995707475e-06,
+      "loss": 0.9862,
+      "step": 34380
+    },
+    {
+      "epoch": 0.2489377257558977,
+      "grad_norm": 0.1603364646434784,
+      "learning_rate": 4.751069512910161e-06,
+      "loss": 0.978,
+      "step": 34390
+    },
+    {
+      "epoch": 0.24901011241648388,
+      "grad_norm": 0.18249155580997467,
+      "learning_rate": 4.750997126249576e-06,
+      "loss": 0.9909,
+      "step": 34400
+    },
+    {
+      "epoch": 0.24908249907707009,
+      "grad_norm": 0.1699957698583603,
+      "learning_rate": 4.750924739588989e-06,
+      "loss": 0.9915,
+      "step": 34410
+    },
+    {
+      "epoch": 0.24915488573765626,
+      "grad_norm": 0.17712600529193878,
+      "learning_rate": 4.750852352928403e-06,
+      "loss": 0.976,
+      "step": 34420
+    },
+    {
+      "epoch": 0.24922727239824244,
+      "grad_norm": 0.17744386196136475,
+      "learning_rate": 4.7507799662678164e-06,
+      "loss": 0.9728,
+      "step": 34430
+    },
+    {
+      "epoch": 0.24929965905882864,
+      "grad_norm": 0.15788348019123077,
+      "learning_rate": 4.750707579607231e-06,
+      "loss": 0.9756,
+      "step": 34440
+    },
+    {
+      "epoch": 0.24937204571941482,
+      "grad_norm": 0.1607343703508377,
+      "learning_rate": 4.7506351929466445e-06,
+      "loss": 0.9801,
+      "step": 34450
+    },
+    {
+      "epoch": 0.24944443238000102,
+      "grad_norm": 0.18192967772483826,
+      "learning_rate": 4.750562806286058e-06,
+      "loss": 0.992,
+      "step": 34460
+    },
+    {
+      "epoch": 0.2495168190405872,
+      "grad_norm": 0.17542362213134766,
+      "learning_rate": 4.750490419625472e-06,
+      "loss": 0.9864,
+      "step": 34470
+    },
+    {
+      "epoch": 0.24958920570117338,
+      "grad_norm": 0.17715929448604584,
+      "learning_rate": 4.750418032964886e-06,
+      "loss": 0.9606,
+      "step": 34480
+    },
+    {
+      "epoch": 0.24966159236175958,
+      "grad_norm": 0.16454724967479706,
+      "learning_rate": 4.7503456463043e-06,
+      "loss": 0.9867,
+      "step": 34490
+    },
+    {
+      "epoch": 0.24973397902234576,
+      "grad_norm": 0.1616222858428955,
+      "learning_rate": 4.7502732596437134e-06,
+      "loss": 0.9823,
+      "step": 34500
+    },
+    {
+      "epoch": 0.24980636568293194,
+      "grad_norm": 0.21941696107387543,
+      "learning_rate": 4.750200872983127e-06,
+      "loss": 0.9895,
+      "step": 34510
+    },
+    {
+      "epoch": 0.24987875234351814,
+      "grad_norm": 0.20772811770439148,
+      "learning_rate": 4.750128486322541e-06,
+      "loss": 0.9953,
+      "step": 34520
+    },
+    {
+      "epoch": 0.24995113900410432,
+      "grad_norm": 0.172512024641037,
+      "learning_rate": 4.750056099661955e-06,
+      "loss": 0.9808,
+      "step": 34530
+    },
+    {
+      "epoch": 0.2500235256646905,
+      "grad_norm": 0.1570970118045807,
+      "learning_rate": 4.749983713001369e-06,
+      "loss": 0.9893,
+      "step": 34540
+    },
+    {
+      "epoch": 0.25009591232527667,
+      "grad_norm": 0.1649353802204132,
+      "learning_rate": 4.749911326340782e-06,
+      "loss": 0.9848,
+      "step": 34550
+    },
+    {
+      "epoch": 0.2501682989858629,
+      "grad_norm": 0.18175838887691498,
+      "learning_rate": 4.749838939680196e-06,
+      "loss": 0.9766,
+      "step": 34560
+    },
+    {
+      "epoch": 0.2502406856464491,
+      "grad_norm": 0.16872644424438477,
+      "learning_rate": 4.7497665530196104e-06,
+      "loss": 0.9747,
+      "step": 34570
+    },
+    {
+      "epoch": 0.25031307230703526,
+      "grad_norm": 0.16489218175411224,
+      "learning_rate": 4.749694166359023e-06,
+      "loss": 0.9861,
+      "step": 34580
+    },
+    {
+      "epoch": 0.25038545896762143,
+      "grad_norm": 0.15996752679347992,
+      "learning_rate": 4.749621779698438e-06,
+      "loss": 0.9866,
+      "step": 34590
+    },
+    {
+      "epoch": 0.2504578456282076,
+      "grad_norm": 0.1725120097398758,
+      "learning_rate": 4.749549393037851e-06,
+      "loss": 0.9784,
+      "step": 34600
+    },
+    {
+      "epoch": 0.25053023228879384,
+      "grad_norm": 0.167087122797966,
+      "learning_rate": 4.749477006377265e-06,
+      "loss": 0.9864,
+      "step": 34610
+    },
+    {
+      "epoch": 0.25060261894938,
+      "grad_norm": 0.1733073741197586,
+      "learning_rate": 4.7494046197166785e-06,
+      "loss": 0.964,
+      "step": 34620
+    },
+    {
+      "epoch": 0.2506750056099662,
+      "grad_norm": 0.1666453331708908,
+      "learning_rate": 4.749332233056093e-06,
+      "loss": 0.9815,
+      "step": 34630
+    },
+    {
+      "epoch": 0.25074739227055237,
+      "grad_norm": 0.2046760767698288,
+      "learning_rate": 4.749259846395507e-06,
+      "loss": 0.974,
+      "step": 34640
+    },
+    {
+      "epoch": 0.25081977893113855,
+      "grad_norm": 0.1714048534631729,
+      "learning_rate": 4.74918745973492e-06,
+      "loss": 0.9842,
+      "step": 34650
+    },
+    {
+      "epoch": 0.2508921655917248,
+      "grad_norm": 0.17349305748939514,
+      "learning_rate": 4.749115073074334e-06,
+      "loss": 0.9822,
+      "step": 34660
+    },
+    {
+      "epoch": 0.25096455225231096,
+      "grad_norm": 0.1701555848121643,
+      "learning_rate": 4.749042686413748e-06,
+      "loss": 0.9751,
+      "step": 34670
+    },
+    {
+      "epoch": 0.25103693891289713,
+      "grad_norm": 0.1766299158334732,
+      "learning_rate": 4.748970299753162e-06,
+      "loss": 0.9783,
+      "step": 34680
+    },
+    {
+      "epoch": 0.2511093255734833,
+      "grad_norm": 0.16215893626213074,
+      "learning_rate": 4.7488979130925755e-06,
+      "loss": 0.9798,
+      "step": 34690
+    },
+    {
+      "epoch": 0.2511817122340695,
+      "grad_norm": 0.16153815388679504,
+      "learning_rate": 4.748825526431989e-06,
+      "loss": 0.9638,
+      "step": 34700
+    },
+    {
+      "epoch": 0.2512540988946557,
+      "grad_norm": 0.17262734472751617,
+      "learning_rate": 4.748753139771403e-06,
+      "loss": 0.9585,
+      "step": 34710
+    },
+    {
+      "epoch": 0.2513264855552419,
+      "grad_norm": 0.1748388707637787,
+      "learning_rate": 4.748680753110817e-06,
+      "loss": 0.9755,
+      "step": 34720
+    },
+    {
+      "epoch": 0.2513988722158281,
+      "grad_norm": 0.16861870884895325,
+      "learning_rate": 4.748608366450231e-06,
+      "loss": 0.9901,
+      "step": 34730
+    },
+    {
+      "epoch": 0.25147125887641425,
+      "grad_norm": 0.1585874855518341,
+      "learning_rate": 4.7485359797896445e-06,
+      "loss": 0.9944,
+      "step": 34740
+    },
+    {
+      "epoch": 0.2515436455370004,
+      "grad_norm": 0.16749711334705353,
+      "learning_rate": 4.748463593129058e-06,
+      "loss": 0.967,
+      "step": 34750
+    },
+    {
+      "epoch": 0.2516160321975866,
+      "grad_norm": 0.19604967534542084,
+      "learning_rate": 4.7483912064684725e-06,
+      "loss": 0.991,
+      "step": 34760
+    },
+    {
+      "epoch": 0.25168841885817284,
+      "grad_norm": 0.16052144765853882,
+      "learning_rate": 4.748318819807886e-06,
+      "loss": 0.9695,
+      "step": 34770
+    },
+    {
+      "epoch": 0.251760805518759,
+      "grad_norm": 0.18518640100955963,
+      "learning_rate": 4.7482464331473e-06,
+      "loss": 0.9893,
+      "step": 34780
+    },
+    {
+      "epoch": 0.2518331921793452,
+      "grad_norm": 0.18303367495536804,
+      "learning_rate": 4.748174046486713e-06,
+      "loss": 0.9683,
+      "step": 34790
+    },
+    {
+      "epoch": 0.25190557883993137,
+      "grad_norm": 0.178907573223114,
+      "learning_rate": 4.748101659826128e-06,
+      "loss": 0.9822,
+      "step": 34800
+    },
+    {
+      "epoch": 0.25197796550051754,
+      "grad_norm": 0.1701265573501587,
+      "learning_rate": 4.7480292731655415e-06,
+      "loss": 0.9788,
+      "step": 34810
+    },
+    {
+      "epoch": 0.2520503521611038,
+      "grad_norm": 0.15824520587921143,
+      "learning_rate": 4.747956886504955e-06,
+      "loss": 0.9705,
+      "step": 34820
+    },
+    {
+      "epoch": 0.25212273882168995,
+      "grad_norm": 0.16767218708992004,
+      "learning_rate": 4.747884499844369e-06,
+      "loss": 0.9993,
+      "step": 34830
+    },
+    {
+      "epoch": 0.25219512548227613,
+      "grad_norm": 0.16409531235694885,
+      "learning_rate": 4.747812113183783e-06,
+      "loss": 0.986,
+      "step": 34840
+    },
+    {
+      "epoch": 0.2522675121428623,
+      "grad_norm": 0.17260675132274628,
+      "learning_rate": 4.747739726523197e-06,
+      "loss": 0.9677,
+      "step": 34850
+    },
+    {
+      "epoch": 0.2523398988034485,
+      "grad_norm": 0.15789243578910828,
+      "learning_rate": 4.74766733986261e-06,
+      "loss": 0.9795,
+      "step": 34860
+    },
+    {
+      "epoch": 0.2524122854640347,
+      "grad_norm": 0.9138031005859375,
+      "learning_rate": 4.747594953202024e-06,
+      "loss": 0.9824,
+      "step": 34870
+    },
+    {
+      "epoch": 0.2524846721246209,
+      "grad_norm": 0.1509862244129181,
+      "learning_rate": 4.7475225665414385e-06,
+      "loss": 0.9766,
+      "step": 34880
+    },
+    {
+      "epoch": 0.25255705878520707,
+      "grad_norm": 0.15754196047782898,
+      "learning_rate": 4.747450179880852e-06,
+      "loss": 0.9781,
+      "step": 34890
+    },
+    {
+      "epoch": 0.25262944544579324,
+      "grad_norm": 0.1629686951637268,
+      "learning_rate": 4.747377793220266e-06,
+      "loss": 0.9731,
+      "step": 34900
+    },
+    {
+      "epoch": 0.2527018321063794,
+      "grad_norm": 0.17362354695796967,
+      "learning_rate": 4.747305406559679e-06,
+      "loss": 0.9777,
+      "step": 34910
+    },
+    {
+      "epoch": 0.2527742187669656,
+      "grad_norm": 0.1758502572774887,
+      "learning_rate": 4.747233019899094e-06,
+      "loss": 0.9851,
+      "step": 34920
+    },
+    {
+      "epoch": 0.25284660542755183,
+      "grad_norm": 0.16180376708507538,
+      "learning_rate": 4.747160633238507e-06,
+      "loss": 0.9728,
+      "step": 34930
+    },
+    {
+      "epoch": 0.252918992088138,
+      "grad_norm": 0.18018090724945068,
+      "learning_rate": 4.747088246577921e-06,
+      "loss": 0.9682,
+      "step": 34940
+    },
+    {
+      "epoch": 0.2529913787487242,
+      "grad_norm": 0.1669149100780487,
+      "learning_rate": 4.747015859917335e-06,
+      "loss": 0.9744,
+      "step": 34950
+    },
+    {
+      "epoch": 0.25306376540931036,
+      "grad_norm": 0.16746865212917328,
+      "learning_rate": 4.746943473256749e-06,
+      "loss": 0.986,
+      "step": 34960
+    },
+    {
+      "epoch": 0.25313615206989654,
+      "grad_norm": 0.1823040395975113,
+      "learning_rate": 4.746871086596163e-06,
+      "loss": 0.9739,
+      "step": 34970
+    },
+    {
+      "epoch": 0.25320853873048277,
+      "grad_norm": 0.17070458829402924,
+      "learning_rate": 4.746798699935576e-06,
+      "loss": 0.9951,
+      "step": 34980
+    },
+    {
+      "epoch": 0.25328092539106895,
+      "grad_norm": 0.16862110793590546,
+      "learning_rate": 4.74672631327499e-06,
+      "loss": 0.9872,
+      "step": 34990
+    },
+    {
+      "epoch": 0.2533533120516551,
+      "grad_norm": 0.17179900407791138,
+      "learning_rate": 4.746653926614404e-06,
+      "loss": 0.9854,
+      "step": 35000
+    },
+    {
+      "epoch": 0.2534256987122413,
+      "grad_norm": 0.17400874197483063,
+      "learning_rate": 4.746581539953818e-06,
+      "loss": 0.9763,
+      "step": 35010
+    },
+    {
+      "epoch": 0.2534980853728275,
+      "grad_norm": 0.20990721881389618,
+      "learning_rate": 4.746509153293232e-06,
+      "loss": 0.9752,
+      "step": 35020
+    },
+    {
+      "epoch": 0.2535704720334137,
+      "grad_norm": 0.17433054745197296,
+      "learning_rate": 4.746436766632645e-06,
+      "loss": 0.9773,
+      "step": 35030
+    },
+    {
+      "epoch": 0.2536428586939999,
+      "grad_norm": 0.17180733382701874,
+      "learning_rate": 4.74636437997206e-06,
+      "loss": 0.9757,
+      "step": 35040
+    },
+    {
+      "epoch": 0.25371524535458606,
+      "grad_norm": 0.19720107316970825,
+      "learning_rate": 4.746291993311473e-06,
+      "loss": 0.9789,
+      "step": 35050
+    },
+    {
+      "epoch": 0.25378763201517224,
+      "grad_norm": 0.18892458081245422,
+      "learning_rate": 4.746219606650887e-06,
+      "loss": 0.9868,
+      "step": 35060
+    },
+    {
+      "epoch": 0.2538600186757584,
+      "grad_norm": 0.18329143524169922,
+      "learning_rate": 4.7461472199903006e-06,
+      "loss": 0.9795,
+      "step": 35070
+    },
+    {
+      "epoch": 0.2539324053363446,
+      "grad_norm": 0.17473115026950836,
+      "learning_rate": 4.746074833329715e-06,
+      "loss": 0.9798,
+      "step": 35080
+    },
+    {
+      "epoch": 0.2540047919969308,
+      "grad_norm": 0.17442865669727325,
+      "learning_rate": 4.746002446669129e-06,
+      "loss": 0.9803,
+      "step": 35090
+    },
+    {
+      "epoch": 0.254077178657517,
+      "grad_norm": 0.17204399406909943,
+      "learning_rate": 4.745930060008542e-06,
+      "loss": 0.9771,
+      "step": 35100
+    },
+    {
+      "epoch": 0.2541495653181032,
+      "grad_norm": 0.1870034784078598,
+      "learning_rate": 4.745857673347956e-06,
+      "loss": 0.9718,
+      "step": 35110
+    },
+    {
+      "epoch": 0.25422195197868935,
+      "grad_norm": 0.1639021635055542,
+      "learning_rate": 4.7457852866873695e-06,
+      "loss": 0.9805,
+      "step": 35120
+    },
+    {
+      "epoch": 0.25429433863927553,
+      "grad_norm": 0.16658686101436615,
+      "learning_rate": 4.745712900026783e-06,
+      "loss": 0.982,
+      "step": 35130
+    },
+    {
+      "epoch": 0.25436672529986176,
+      "grad_norm": 0.164857879281044,
+      "learning_rate": 4.745640513366197e-06,
+      "loss": 0.9885,
+      "step": 35140
+    },
+    {
+      "epoch": 0.25443911196044794,
+      "grad_norm": 0.18031135201454163,
+      "learning_rate": 4.745568126705611e-06,
+      "loss": 0.9763,
+      "step": 35150
+    },
+    {
+      "epoch": 0.2545114986210341,
+      "grad_norm": 0.1674937754869461,
+      "learning_rate": 4.745495740045025e-06,
+      "loss": 0.9891,
+      "step": 35160
+    },
+    {
+      "epoch": 0.2545838852816203,
+      "grad_norm": 0.16859376430511475,
+      "learning_rate": 4.7454233533844384e-06,
+      "loss": 0.9738,
+      "step": 35170
+    },
+    {
+      "epoch": 0.25465627194220647,
+      "grad_norm": 0.1623847633600235,
+      "learning_rate": 4.745350966723852e-06,
+      "loss": 0.9772,
+      "step": 35180
+    },
+    {
+      "epoch": 0.2547286586027927,
+      "grad_norm": 0.25162413716316223,
+      "learning_rate": 4.7452785800632665e-06,
+      "loss": 0.9767,
+      "step": 35190
+    },
+    {
+      "epoch": 0.2548010452633789,
+      "grad_norm": 0.16929900646209717,
+      "learning_rate": 4.74520619340268e-06,
+      "loss": 0.9745,
+      "step": 35200
+    },
+    {
+      "epoch": 0.25487343192396505,
+      "grad_norm": 0.18202784657478333,
+      "learning_rate": 4.745133806742094e-06,
+      "loss": 0.9777,
+      "step": 35210
+    },
+    {
+      "epoch": 0.25494581858455123,
+      "grad_norm": 0.16864082217216492,
+      "learning_rate": 4.745061420081507e-06,
+      "loss": 0.9695,
+      "step": 35220
+    },
+    {
+      "epoch": 0.2550182052451374,
+      "grad_norm": 0.18244844675064087,
+      "learning_rate": 4.744989033420922e-06,
+      "loss": 0.9793,
+      "step": 35230
+    },
+    {
+      "epoch": 0.25509059190572364,
+      "grad_norm": 0.15838764607906342,
+      "learning_rate": 4.7449166467603354e-06,
+      "loss": 0.9749,
+      "step": 35240
+    },
+    {
+      "epoch": 0.2551629785663098,
+      "grad_norm": 0.1680920273065567,
+      "learning_rate": 4.744844260099749e-06,
+      "loss": 0.9788,
+      "step": 35250
+    },
+    {
+      "epoch": 0.255235365226896,
+      "grad_norm": 0.1684524267911911,
+      "learning_rate": 4.744771873439163e-06,
+      "loss": 0.9859,
+      "step": 35260
+    },
+    {
+      "epoch": 0.25530775188748217,
+      "grad_norm": 0.16353952884674072,
+      "learning_rate": 4.744699486778577e-06,
+      "loss": 0.9842,
+      "step": 35270
+    },
+    {
+      "epoch": 0.25538013854806835,
+      "grad_norm": 0.17483973503112793,
+      "learning_rate": 4.744627100117991e-06,
+      "loss": 0.9818,
+      "step": 35280
+    },
+    {
+      "epoch": 0.2554525252086545,
+      "grad_norm": 0.16171082854270935,
+      "learning_rate": 4.744554713457404e-06,
+      "loss": 0.9814,
+      "step": 35290
+    },
+    {
+      "epoch": 0.25552491186924076,
+      "grad_norm": 0.16629914939403534,
+      "learning_rate": 4.744482326796818e-06,
+      "loss": 0.9932,
+      "step": 35300
+    },
+    {
+      "epoch": 0.25559729852982693,
+      "grad_norm": 0.1688779890537262,
+      "learning_rate": 4.744409940136232e-06,
+      "loss": 0.9748,
+      "step": 35310
+    },
+    {
+      "epoch": 0.2556696851904131,
+      "grad_norm": 0.18332324922084808,
+      "learning_rate": 4.744337553475646e-06,
+      "loss": 0.983,
+      "step": 35320
+    },
+    {
+      "epoch": 0.2557420718509993,
+      "grad_norm": 3.5396459102630615,
+      "learning_rate": 4.74426516681506e-06,
+      "loss": 0.9881,
+      "step": 35330
+    },
+    {
+      "epoch": 0.25581445851158546,
+      "grad_norm": 0.16247138381004333,
+      "learning_rate": 4.744192780154473e-06,
+      "loss": 0.9893,
+      "step": 35340
+    },
+    {
+      "epoch": 0.2558868451721717,
+      "grad_norm": 0.17081671953201294,
+      "learning_rate": 4.744120393493887e-06,
+      "loss": 0.9899,
+      "step": 35350
+    },
+    {
+      "epoch": 0.25595923183275787,
+      "grad_norm": 0.1555139124393463,
+      "learning_rate": 4.744048006833301e-06,
+      "loss": 0.9697,
+      "step": 35360
+    },
+    {
+      "epoch": 0.25603161849334405,
+      "grad_norm": 0.19195972383022308,
+      "learning_rate": 4.743975620172715e-06,
+      "loss": 0.9819,
+      "step": 35370
+    },
+    {
+      "epoch": 0.2561040051539302,
+      "grad_norm": 0.2612147927284241,
+      "learning_rate": 4.743903233512129e-06,
+      "loss": 0.9856,
+      "step": 35380
+    },
+    {
+      "epoch": 0.2561763918145164,
+      "grad_norm": 0.17265565693378448,
+      "learning_rate": 4.743830846851542e-06,
+      "loss": 0.973,
+      "step": 35390
+    },
+    {
+      "epoch": 0.25624877847510263,
+      "grad_norm": 0.17444762587547302,
+      "learning_rate": 4.743758460190957e-06,
+      "loss": 0.9811,
+      "step": 35400
+    },
+    {
+      "epoch": 0.2563211651356888,
+      "grad_norm": 0.17848478257656097,
+      "learning_rate": 4.74368607353037e-06,
+      "loss": 0.9701,
+      "step": 35410
+    },
+    {
+      "epoch": 0.256393551796275,
+      "grad_norm": 0.16675600409507751,
+      "learning_rate": 4.743613686869784e-06,
+      "loss": 0.9836,
+      "step": 35420
+    },
+    {
+      "epoch": 0.25646593845686116,
+      "grad_norm": 0.16786222159862518,
+      "learning_rate": 4.7435413002091975e-06,
+      "loss": 0.9816,
+      "step": 35430
+    },
+    {
+      "epoch": 0.25653832511744734,
+      "grad_norm": 0.18785636126995087,
+      "learning_rate": 4.743468913548612e-06,
+      "loss": 0.9693,
+      "step": 35440
+    },
+    {
+      "epoch": 0.2566107117780335,
+      "grad_norm": 0.16622446477413177,
+      "learning_rate": 4.743396526888026e-06,
+      "loss": 0.9818,
+      "step": 35450
+    },
+    {
+      "epoch": 0.25668309843861975,
+      "grad_norm": 0.1868736445903778,
+      "learning_rate": 4.743324140227439e-06,
+      "loss": 0.9806,
+      "step": 35460
+    },
+    {
+      "epoch": 0.2567554850992059,
+      "grad_norm": 0.17753900587558746,
+      "learning_rate": 4.743251753566853e-06,
+      "loss": 0.9739,
+      "step": 35470
+    },
+    {
+      "epoch": 0.2568278717597921,
+      "grad_norm": 0.18363986909389496,
+      "learning_rate": 4.743179366906267e-06,
+      "loss": 0.9767,
+      "step": 35480
+    },
+    {
+      "epoch": 0.2569002584203783,
+      "grad_norm": 0.1745624840259552,
+      "learning_rate": 4.743106980245681e-06,
+      "loss": 0.9889,
+      "step": 35490
+    },
+    {
+      "epoch": 0.25697264508096446,
+      "grad_norm": 0.16066277027130127,
+      "learning_rate": 4.7430345935850945e-06,
+      "loss": 0.9692,
+      "step": 35500
+    },
+    {
+      "epoch": 0.2570450317415507,
+      "grad_norm": 0.16065356135368347,
+      "learning_rate": 4.742962206924508e-06,
+      "loss": 0.9864,
+      "step": 35510
+    },
+    {
+      "epoch": 0.25711741840213687,
+      "grad_norm": 0.1665170043706894,
+      "learning_rate": 4.742889820263923e-06,
+      "loss": 0.979,
+      "step": 35520
+    },
+    {
+      "epoch": 0.25718980506272304,
+      "grad_norm": 0.15713395178318024,
+      "learning_rate": 4.742817433603336e-06,
+      "loss": 0.9603,
+      "step": 35530
+    },
+    {
+      "epoch": 0.2572621917233092,
+      "grad_norm": 0.18510933220386505,
+      "learning_rate": 4.74274504694275e-06,
+      "loss": 0.9726,
+      "step": 35540
+    },
+    {
+      "epoch": 0.2573345783838954,
+      "grad_norm": 0.16515351831912994,
+      "learning_rate": 4.7426726602821635e-06,
+      "loss": 0.9782,
+      "step": 35550
+    },
+    {
+      "epoch": 0.25740696504448163,
+      "grad_norm": 0.17086593806743622,
+      "learning_rate": 4.742600273621578e-06,
+      "loss": 0.9634,
+      "step": 35560
+    },
+    {
+      "epoch": 0.2574793517050678,
+      "grad_norm": 0.17308638989925385,
+      "learning_rate": 4.7425278869609915e-06,
+      "loss": 0.978,
+      "step": 35570
+    },
+    {
+      "epoch": 0.257551738365654,
+      "grad_norm": 0.18150749802589417,
+      "learning_rate": 4.742455500300405e-06,
+      "loss": 0.976,
+      "step": 35580
+    },
+    {
+      "epoch": 0.25762412502624016,
+      "grad_norm": 0.1714148074388504,
+      "learning_rate": 4.742383113639819e-06,
+      "loss": 0.9786,
+      "step": 35590
+    },
+    {
+      "epoch": 0.25769651168682634,
+      "grad_norm": 0.16008371114730835,
+      "learning_rate": 4.742310726979233e-06,
+      "loss": 0.9789,
+      "step": 35600
+    },
+    {
+      "epoch": 0.2577688983474125,
+      "grad_norm": 0.17920182645320892,
+      "learning_rate": 4.742238340318647e-06,
+      "loss": 0.9875,
+      "step": 35610
+    },
+    {
+      "epoch": 0.25784128500799874,
+      "grad_norm": 0.1833036094903946,
+      "learning_rate": 4.7421659536580605e-06,
+      "loss": 1.0031,
+      "step": 35620
+    },
+    {
+      "epoch": 0.2579136716685849,
+      "grad_norm": 0.17967666685581207,
+      "learning_rate": 4.742093566997474e-06,
+      "loss": 0.9975,
+      "step": 35630
+    },
+    {
+      "epoch": 0.2579860583291711,
+      "grad_norm": 0.29100435972213745,
+      "learning_rate": 4.742021180336888e-06,
+      "loss": 0.974,
+      "step": 35640
+    },
+    {
+      "epoch": 0.2580584449897573,
+      "grad_norm": 0.1635904461145401,
+      "learning_rate": 4.741948793676301e-06,
+      "loss": 0.9854,
+      "step": 35650
+    },
+    {
+      "epoch": 0.25813083165034345,
+      "grad_norm": 0.20104598999023438,
+      "learning_rate": 4.741876407015715e-06,
+      "loss": 0.9847,
+      "step": 35660
+    },
+    {
+      "epoch": 0.2582032183109297,
+      "grad_norm": 0.1948116570711136,
+      "learning_rate": 4.741804020355129e-06,
+      "loss": 0.9766,
+      "step": 35670
+    },
+    {
+      "epoch": 0.25827560497151586,
+      "grad_norm": 0.18129467964172363,
+      "learning_rate": 4.741731633694543e-06,
+      "loss": 0.9788,
+      "step": 35680
+    },
+    {
+      "epoch": 0.25834799163210204,
+      "grad_norm": 0.19102171063423157,
+      "learning_rate": 4.741659247033957e-06,
+      "loss": 0.9826,
+      "step": 35690
+    },
+    {
+      "epoch": 0.2584203782926882,
+      "grad_norm": 0.15986351668834686,
+      "learning_rate": 4.74158686037337e-06,
+      "loss": 0.9818,
+      "step": 35700
+    },
+    {
+      "epoch": 0.2584927649532744,
+      "grad_norm": 0.16768378019332886,
+      "learning_rate": 4.741514473712785e-06,
+      "loss": 0.9833,
+      "step": 35710
+    },
+    {
+      "epoch": 0.2585651516138606,
+      "grad_norm": 0.16951607167720795,
+      "learning_rate": 4.741442087052198e-06,
+      "loss": 0.9796,
+      "step": 35720
+    },
+    {
+      "epoch": 0.2586375382744468,
+      "grad_norm": 0.18605226278305054,
+      "learning_rate": 4.741369700391612e-06,
+      "loss": 0.9878,
+      "step": 35730
+    },
+    {
+      "epoch": 0.258709924935033,
+      "grad_norm": 0.15698987245559692,
+      "learning_rate": 4.7412973137310256e-06,
+      "loss": 0.9664,
+      "step": 35740
+    },
+    {
+      "epoch": 0.25878231159561915,
+      "grad_norm": 0.17848019301891327,
+      "learning_rate": 4.74122492707044e-06,
+      "loss": 0.968,
+      "step": 35750
+    },
+    {
+      "epoch": 0.25885469825620533,
+      "grad_norm": 0.18096424639225006,
+      "learning_rate": 4.741152540409854e-06,
+      "loss": 0.9781,
+      "step": 35760
+    },
+    {
+      "epoch": 0.25892708491679156,
+      "grad_norm": 0.16998852789402008,
+      "learning_rate": 4.741080153749267e-06,
+      "loss": 0.9852,
+      "step": 35770
+    },
+    {
+      "epoch": 0.25899947157737774,
+      "grad_norm": 0.16071717441082,
+      "learning_rate": 4.741007767088681e-06,
+      "loss": 0.9852,
+      "step": 35780
+    },
+    {
+      "epoch": 0.2590718582379639,
+      "grad_norm": 0.16710597276687622,
+      "learning_rate": 4.740935380428095e-06,
+      "loss": 0.9728,
+      "step": 35790
+    },
+    {
+      "epoch": 0.2591442448985501,
+      "grad_norm": 0.15692850947380066,
+      "learning_rate": 4.740862993767509e-06,
+      "loss": 0.983,
+      "step": 35800
+    },
+    {
+      "epoch": 0.25921663155913627,
+      "grad_norm": 0.16748061776161194,
+      "learning_rate": 4.7407906071069226e-06,
+      "loss": 0.984,
+      "step": 35810
+    },
+    {
+      "epoch": 0.25928901821972244,
+      "grad_norm": 0.16068707406520844,
+      "learning_rate": 4.740718220446336e-06,
+      "loss": 0.9745,
+      "step": 35820
+    },
+    {
+      "epoch": 0.2593614048803087,
+      "grad_norm": 0.17079469561576843,
+      "learning_rate": 4.740645833785751e-06,
+      "loss": 0.9656,
+      "step": 35830
+    },
+    {
+      "epoch": 0.25943379154089485,
+      "grad_norm": 0.18786334991455078,
+      "learning_rate": 4.740573447125164e-06,
+      "loss": 0.9619,
+      "step": 35840
+    },
+    {
+      "epoch": 0.25950617820148103,
+      "grad_norm": 0.1623920202255249,
+      "learning_rate": 4.740501060464578e-06,
+      "loss": 0.9768,
+      "step": 35850
+    },
+    {
+      "epoch": 0.2595785648620672,
+      "grad_norm": 0.16342291235923767,
+      "learning_rate": 4.7404286738039915e-06,
+      "loss": 0.9867,
+      "step": 35860
+    },
+    {
+      "epoch": 0.2596509515226534,
+      "grad_norm": 0.16781893372535706,
+      "learning_rate": 4.740356287143406e-06,
+      "loss": 0.9758,
+      "step": 35870
+    },
+    {
+      "epoch": 0.2597233381832396,
+      "grad_norm": 0.1712564378976822,
+      "learning_rate": 4.74028390048282e-06,
+      "loss": 0.9558,
+      "step": 35880
+    },
+    {
+      "epoch": 0.2597957248438258,
+      "grad_norm": 0.16046634316444397,
+      "learning_rate": 4.740211513822233e-06,
+      "loss": 0.9723,
+      "step": 35890
+    },
+    {
+      "epoch": 0.25986811150441197,
+      "grad_norm": 0.1630384773015976,
+      "learning_rate": 4.740139127161647e-06,
+      "loss": 0.9766,
+      "step": 35900
+    },
+    {
+      "epoch": 0.25994049816499815,
+      "grad_norm": 0.16825400292873383,
+      "learning_rate": 4.740066740501061e-06,
+      "loss": 0.9824,
+      "step": 35910
+    },
+    {
+      "epoch": 0.2600128848255843,
+      "grad_norm": 0.17138569056987762,
+      "learning_rate": 4.739994353840475e-06,
+      "loss": 0.9749,
+      "step": 35920
+    },
+    {
+      "epoch": 0.26008527148617056,
+      "grad_norm": 0.1562861055135727,
+      "learning_rate": 4.7399219671798885e-06,
+      "loss": 0.9786,
+      "step": 35930
+    },
+    {
+      "epoch": 0.26015765814675673,
+      "grad_norm": 0.1865861713886261,
+      "learning_rate": 4.739849580519302e-06,
+      "loss": 0.9761,
+      "step": 35940
+    },
+    {
+      "epoch": 0.2602300448073429,
+      "grad_norm": 0.168324813246727,
+      "learning_rate": 4.739777193858716e-06,
+      "loss": 0.9813,
+      "step": 35950
+    },
+    {
+      "epoch": 0.2603024314679291,
+      "grad_norm": 0.1690807342529297,
+      "learning_rate": 4.73970480719813e-06,
+      "loss": 0.9779,
+      "step": 35960
+    },
+    {
+      "epoch": 0.26037481812851526,
+      "grad_norm": 0.17235450446605682,
+      "learning_rate": 4.739632420537544e-06,
+      "loss": 0.9858,
+      "step": 35970
+    },
+    {
+      "epoch": 0.26044720478910144,
+      "grad_norm": 0.188417449593544,
+      "learning_rate": 4.7395600338769574e-06,
+      "loss": 0.9765,
+      "step": 35980
+    },
+    {
+      "epoch": 0.26051959144968767,
+      "grad_norm": 0.17113593220710754,
+      "learning_rate": 4.739487647216371e-06,
+      "loss": 0.9687,
+      "step": 35990
+    },
+    {
+      "epoch": 0.26059197811027385,
+      "grad_norm": 0.1908288598060608,
+      "learning_rate": 4.7394152605557855e-06,
+      "loss": 0.9831,
+      "step": 36000
+    },
+    {
+      "epoch": 0.26066436477086,
+      "grad_norm": 0.17590002715587616,
+      "learning_rate": 4.739342873895199e-06,
+      "loss": 0.9856,
+      "step": 36010
+    },
+    {
+      "epoch": 0.2607367514314462,
+      "grad_norm": 0.16687145829200745,
+      "learning_rate": 4.739270487234613e-06,
+      "loss": 0.9793,
+      "step": 36020
+    },
+    {
+      "epoch": 0.2608091380920324,
+      "grad_norm": 0.1600182056427002,
+      "learning_rate": 4.739198100574026e-06,
+      "loss": 0.9738,
+      "step": 36030
+    },
+    {
+      "epoch": 0.2608815247526186,
+      "grad_norm": 0.16335642337799072,
+      "learning_rate": 4.739125713913441e-06,
+      "loss": 0.9686,
+      "step": 36040
+    },
+    {
+      "epoch": 0.2609539114132048,
+      "grad_norm": 0.16597791016101837,
+      "learning_rate": 4.7390533272528544e-06,
+      "loss": 0.9796,
+      "step": 36050
+    },
+    {
+      "epoch": 0.26102629807379096,
+      "grad_norm": 0.1797637939453125,
+      "learning_rate": 4.738980940592268e-06,
+      "loss": 0.9771,
+      "step": 36060
+    },
+    {
+      "epoch": 0.26109868473437714,
+      "grad_norm": 0.17320837080478668,
+      "learning_rate": 4.738908553931682e-06,
+      "loss": 0.9889,
+      "step": 36070
+    },
+    {
+      "epoch": 0.2611710713949633,
+      "grad_norm": 0.16971297562122345,
+      "learning_rate": 4.738836167271096e-06,
+      "loss": 0.9947,
+      "step": 36080
+    },
+    {
+      "epoch": 0.26124345805554955,
+      "grad_norm": 0.17000246047973633,
+      "learning_rate": 4.73876378061051e-06,
+      "loss": 0.9631,
+      "step": 36090
+    },
+    {
+      "epoch": 0.2613158447161357,
+      "grad_norm": 0.16442660987377167,
+      "learning_rate": 4.738691393949923e-06,
+      "loss": 0.9566,
+      "step": 36100
+    },
+    {
+      "epoch": 0.2613882313767219,
+      "grad_norm": 0.15749797224998474,
+      "learning_rate": 4.738619007289337e-06,
+      "loss": 0.9857,
+      "step": 36110
+    },
+    {
+      "epoch": 0.2614606180373081,
+      "grad_norm": 0.1590905487537384,
+      "learning_rate": 4.7385466206287514e-06,
+      "loss": 0.9741,
+      "step": 36120
+    },
+    {
+      "epoch": 0.26153300469789426,
+      "grad_norm": 0.20736895501613617,
+      "learning_rate": 4.738474233968165e-06,
+      "loss": 0.9741,
+      "step": 36130
+    },
+    {
+      "epoch": 0.26160539135848043,
+      "grad_norm": 0.1713234931230545,
+      "learning_rate": 4.738401847307579e-06,
+      "loss": 0.9652,
+      "step": 36140
+    },
+    {
+      "epoch": 0.26167777801906666,
+      "grad_norm": 0.17479397356510162,
+      "learning_rate": 4.738329460646992e-06,
+      "loss": 0.962,
+      "step": 36150
+    },
+    {
+      "epoch": 0.26175016467965284,
+      "grad_norm": 0.15319569408893585,
+      "learning_rate": 4.738257073986407e-06,
+      "loss": 0.9914,
+      "step": 36160
+    },
+    {
+      "epoch": 0.261822551340239,
+      "grad_norm": 0.17199110984802246,
+      "learning_rate": 4.7381846873258195e-06,
+      "loss": 0.9701,
+      "step": 36170
+    },
+    {
+      "epoch": 0.2618949380008252,
+      "grad_norm": 0.17923112213611603,
+      "learning_rate": 4.738112300665233e-06,
+      "loss": 0.9866,
+      "step": 36180
+    },
+    {
+      "epoch": 0.26196732466141137,
+      "grad_norm": 0.17701056599617004,
+      "learning_rate": 4.738039914004648e-06,
+      "loss": 0.9905,
+      "step": 36190
+    },
+    {
+      "epoch": 0.2620397113219976,
+      "grad_norm": 0.18318752944469452,
+      "learning_rate": 4.737967527344061e-06,
+      "loss": 0.9769,
+      "step": 36200
+    },
+    {
+      "epoch": 0.2621120979825838,
+      "grad_norm": 0.16948123276233673,
+      "learning_rate": 4.737895140683475e-06,
+      "loss": 0.9701,
+      "step": 36210
+    },
+    {
+      "epoch": 0.26218448464316996,
+      "grad_norm": 0.16804496943950653,
+      "learning_rate": 4.7378227540228885e-06,
+      "loss": 0.9758,
+      "step": 36220
+    },
+    {
+      "epoch": 0.26225687130375613,
+      "grad_norm": 0.16858185827732086,
+      "learning_rate": 4.737750367362303e-06,
+      "loss": 0.9778,
+      "step": 36230
+    },
+    {
+      "epoch": 0.2623292579643423,
+      "grad_norm": 0.16800491511821747,
+      "learning_rate": 4.7376779807017165e-06,
+      "loss": 0.96,
+      "step": 36240
+    },
+    {
+      "epoch": 0.26240164462492854,
+      "grad_norm": 0.16772626340389252,
+      "learning_rate": 4.73760559404113e-06,
+      "loss": 0.958,
+      "step": 36250
+    },
+    {
+      "epoch": 0.2624740312855147,
+      "grad_norm": 0.16602137684822083,
+      "learning_rate": 4.737533207380544e-06,
+      "loss": 0.9761,
+      "step": 36260
+    },
+    {
+      "epoch": 0.2625464179461009,
+      "grad_norm": 0.17461955547332764,
+      "learning_rate": 4.737460820719958e-06,
+      "loss": 0.9546,
+      "step": 36270
+    },
+    {
+      "epoch": 0.2626188046066871,
+      "grad_norm": 0.1623477339744568,
+      "learning_rate": 4.737388434059372e-06,
+      "loss": 0.981,
+      "step": 36280
+    },
+    {
+      "epoch": 0.26269119126727325,
+      "grad_norm": 0.17813843488693237,
+      "learning_rate": 4.7373160473987855e-06,
+      "loss": 0.9779,
+      "step": 36290
+    },
+    {
+      "epoch": 0.2627635779278594,
+      "grad_norm": 0.18442265689373016,
+      "learning_rate": 4.737243660738199e-06,
+      "loss": 0.9541,
+      "step": 36300
+    },
+    {
+      "epoch": 0.26283596458844566,
+      "grad_norm": 0.16500839591026306,
+      "learning_rate": 4.7371712740776135e-06,
+      "loss": 0.9601,
+      "step": 36310
+    },
+    {
+      "epoch": 0.26290835124903184,
+      "grad_norm": 0.17317786812782288,
+      "learning_rate": 4.737098887417027e-06,
+      "loss": 0.9704,
+      "step": 36320
+    },
+    {
+      "epoch": 0.262980737909618,
+      "grad_norm": 0.1605556458234787,
+      "learning_rate": 4.737026500756441e-06,
+      "loss": 0.9793,
+      "step": 36330
+    },
+    {
+      "epoch": 0.2630531245702042,
+      "grad_norm": 0.16761939227581024,
+      "learning_rate": 4.736954114095854e-06,
+      "loss": 0.9669,
+      "step": 36340
+    },
+    {
+      "epoch": 0.26312551123079037,
+      "grad_norm": 0.19816842675209045,
+      "learning_rate": 4.736881727435269e-06,
+      "loss": 0.9749,
+      "step": 36350
+    },
+    {
+      "epoch": 0.2631978978913766,
+      "grad_norm": 0.15934444963932037,
+      "learning_rate": 4.7368093407746825e-06,
+      "loss": 0.9762,
+      "step": 36360
+    },
+    {
+      "epoch": 0.2632702845519628,
+      "grad_norm": 0.16128577291965485,
+      "learning_rate": 4.736736954114096e-06,
+      "loss": 0.9648,
+      "step": 36370
+    },
+    {
+      "epoch": 0.26334267121254895,
+      "grad_norm": 0.1702321320772171,
+      "learning_rate": 4.73666456745351e-06,
+      "loss": 0.9762,
+      "step": 36380
+    },
+    {
+      "epoch": 0.26341505787313513,
+      "grad_norm": 0.16586679220199585,
+      "learning_rate": 4.736592180792924e-06,
+      "loss": 0.9676,
+      "step": 36390
+    },
+    {
+      "epoch": 0.2634874445337213,
+      "grad_norm": 0.18088088929653168,
+      "learning_rate": 4.736519794132338e-06,
+      "loss": 0.974,
+      "step": 36400
+    },
+    {
+      "epoch": 0.26355983119430754,
+      "grad_norm": 0.2222166806459427,
+      "learning_rate": 4.736447407471751e-06,
+      "loss": 0.9776,
+      "step": 36410
+    },
+    {
+      "epoch": 0.2636322178548937,
+      "grad_norm": 0.15918581187725067,
+      "learning_rate": 4.736375020811165e-06,
+      "loss": 0.9787,
+      "step": 36420
+    },
+    {
+      "epoch": 0.2637046045154799,
+      "grad_norm": 0.1603013426065445,
+      "learning_rate": 4.7363026341505795e-06,
+      "loss": 0.9617,
+      "step": 36430
+    },
+    {
+      "epoch": 0.26377699117606607,
+      "grad_norm": 0.1752447783946991,
+      "learning_rate": 4.736230247489993e-06,
+      "loss": 0.9724,
+      "step": 36440
+    },
+    {
+      "epoch": 0.26384937783665224,
+      "grad_norm": 0.27862676978111267,
+      "learning_rate": 4.736157860829407e-06,
+      "loss": 0.9675,
+      "step": 36450
+    },
+    {
+      "epoch": 0.2639217644972385,
+      "grad_norm": 0.1704714149236679,
+      "learning_rate": 4.73608547416882e-06,
+      "loss": 0.962,
+      "step": 36460
+    },
+    {
+      "epoch": 0.26399415115782465,
+      "grad_norm": 0.2055920511484146,
+      "learning_rate": 4.736013087508235e-06,
+      "loss": 0.9712,
+      "step": 36470
+    },
+    {
+      "epoch": 0.26406653781841083,
+      "grad_norm": 0.168257936835289,
+      "learning_rate": 4.735940700847648e-06,
+      "loss": 0.9722,
+      "step": 36480
+    },
+    {
+      "epoch": 0.264138924478997,
+      "grad_norm": 0.1615269035100937,
+      "learning_rate": 4.735868314187062e-06,
+      "loss": 0.9708,
+      "step": 36490
+    },
+    {
+      "epoch": 0.2642113111395832,
+      "grad_norm": 0.16784435510635376,
+      "learning_rate": 4.735795927526476e-06,
+      "loss": 0.9741,
+      "step": 36500
+    },
+    {
+      "epoch": 0.26428369780016936,
+      "grad_norm": 0.16948504745960236,
+      "learning_rate": 4.73572354086589e-06,
+      "loss": 0.9692,
+      "step": 36510
+    },
+    {
+      "epoch": 0.2643560844607556,
+      "grad_norm": 0.16476517915725708,
+      "learning_rate": 4.735651154205304e-06,
+      "loss": 0.9673,
+      "step": 36520
+    },
+    {
+      "epoch": 0.26442847112134177,
+      "grad_norm": 0.18049409985542297,
+      "learning_rate": 4.735578767544717e-06,
+      "loss": 0.9676,
+      "step": 36530
+    },
+    {
+      "epoch": 0.26450085778192794,
+      "grad_norm": 0.1603485643863678,
+      "learning_rate": 4.735506380884131e-06,
+      "loss": 0.9789,
+      "step": 36540
+    },
+    {
+      "epoch": 0.2645732444425141,
+      "grad_norm": 0.16971318423748016,
+      "learning_rate": 4.735433994223545e-06,
+      "loss": 0.9847,
+      "step": 36550
+    },
+    {
+      "epoch": 0.2646456311031003,
+      "grad_norm": 0.19369632005691528,
+      "learning_rate": 4.735361607562959e-06,
+      "loss": 0.9731,
+      "step": 36560
+    },
+    {
+      "epoch": 0.26471801776368653,
+      "grad_norm": 0.1709279865026474,
+      "learning_rate": 4.735289220902373e-06,
+      "loss": 0.9596,
+      "step": 36570
+    },
+    {
+      "epoch": 0.2647904044242727,
+      "grad_norm": 0.1615041047334671,
+      "learning_rate": 4.735216834241786e-06,
+      "loss": 0.9585,
+      "step": 36580
+    },
+    {
+      "epoch": 0.2648627910848589,
+      "grad_norm": 0.17136956751346588,
+      "learning_rate": 4.7351444475812e-06,
+      "loss": 0.9857,
+      "step": 36590
+    },
+    {
+      "epoch": 0.26493517774544506,
+      "grad_norm": 0.1596122533082962,
+      "learning_rate": 4.735072060920614e-06,
+      "loss": 0.9646,
+      "step": 36600
+    },
+    {
+      "epoch": 0.26500756440603124,
+      "grad_norm": 0.16344700753688812,
+      "learning_rate": 4.734999674260028e-06,
+      "loss": 0.9773,
+      "step": 36610
+    },
+    {
+      "epoch": 0.26507995106661747,
+      "grad_norm": 0.17619621753692627,
+      "learning_rate": 4.734927287599442e-06,
+      "loss": 0.97,
+      "step": 36620
+    },
+    {
+      "epoch": 0.26515233772720365,
+      "grad_norm": 0.16314248740673065,
+      "learning_rate": 4.734854900938855e-06,
+      "loss": 0.9712,
+      "step": 36630
+    },
+    {
+      "epoch": 0.2652247243877898,
+      "grad_norm": 0.1628977507352829,
+      "learning_rate": 4.73478251427827e-06,
+      "loss": 0.9813,
+      "step": 36640
+    },
+    {
+      "epoch": 0.265297111048376,
+      "grad_norm": 0.18795722723007202,
+      "learning_rate": 4.734710127617683e-06,
+      "loss": 0.9799,
+      "step": 36650
+    },
+    {
+      "epoch": 0.2653694977089622,
+      "grad_norm": 0.16708257794380188,
+      "learning_rate": 4.734637740957097e-06,
+      "loss": 0.964,
+      "step": 36660
+    },
+    {
+      "epoch": 0.26544188436954835,
+      "grad_norm": 0.15541145205497742,
+      "learning_rate": 4.7345653542965105e-06,
+      "loss": 0.9709,
+      "step": 36670
+    },
+    {
+      "epoch": 0.2655142710301346,
+      "grad_norm": 0.6976906657218933,
+      "learning_rate": 4.734492967635925e-06,
+      "loss": 0.9633,
+      "step": 36680
+    },
+    {
+      "epoch": 0.26558665769072076,
+      "grad_norm": 0.18083471059799194,
+      "learning_rate": 4.734420580975339e-06,
+      "loss": 0.9868,
+      "step": 36690
+    },
+    {
+      "epoch": 0.26565904435130694,
+      "grad_norm": 0.16377471387386322,
+      "learning_rate": 4.734348194314752e-06,
+      "loss": 0.9749,
+      "step": 36700
+    },
+    {
+      "epoch": 0.2657314310118931,
+      "grad_norm": 0.17875248193740845,
+      "learning_rate": 4.734275807654166e-06,
+      "loss": 0.9759,
+      "step": 36710
+    },
+    {
+      "epoch": 0.2658038176724793,
+      "grad_norm": 0.1629532426595688,
+      "learning_rate": 4.7342034209935794e-06,
+      "loss": 0.9764,
+      "step": 36720
+    },
+    {
+      "epoch": 0.2658762043330655,
+      "grad_norm": 0.18493175506591797,
+      "learning_rate": 4.734131034332993e-06,
+      "loss": 0.9693,
+      "step": 36730
+    },
+    {
+      "epoch": 0.2659485909936517,
+      "grad_norm": 0.1734851896762848,
+      "learning_rate": 4.734058647672407e-06,
+      "loss": 0.9604,
+      "step": 36740
+    },
+    {
+      "epoch": 0.2660209776542379,
+      "grad_norm": 0.15318694710731506,
+      "learning_rate": 4.733986261011821e-06,
+      "loss": 0.9816,
+      "step": 36750
+    },
+    {
+      "epoch": 0.26609336431482405,
+      "grad_norm": 0.1794171929359436,
+      "learning_rate": 4.733913874351235e-06,
+      "loss": 0.977,
+      "step": 36760
+    },
+    {
+      "epoch": 0.26616575097541023,
+      "grad_norm": 0.18346603214740753,
+      "learning_rate": 4.733841487690648e-06,
+      "loss": 0.9839,
+      "step": 36770
+    },
+    {
+      "epoch": 0.26623813763599646,
+      "grad_norm": 0.16709034144878387,
+      "learning_rate": 4.733769101030062e-06,
+      "loss": 0.9943,
+      "step": 36780
+    },
+    {
+      "epoch": 0.26631052429658264,
+      "grad_norm": 0.16048607230186462,
+      "learning_rate": 4.7336967143694764e-06,
+      "loss": 0.963,
+      "step": 36790
+    },
+    {
+      "epoch": 0.2663829109571688,
+      "grad_norm": 0.1640567034482956,
+      "learning_rate": 4.73362432770889e-06,
+      "loss": 0.9744,
+      "step": 36800
+    },
+    {
+      "epoch": 0.266455297617755,
+      "grad_norm": 0.18880528211593628,
+      "learning_rate": 4.733551941048304e-06,
+      "loss": 0.9689,
+      "step": 36810
+    },
+    {
+      "epoch": 0.26652768427834117,
+      "grad_norm": 0.17127233743667603,
+      "learning_rate": 4.733479554387717e-06,
+      "loss": 0.9747,
+      "step": 36820
+    },
+    {
+      "epoch": 0.26660007093892735,
+      "grad_norm": 0.17141376435756683,
+      "learning_rate": 4.733407167727132e-06,
+      "loss": 0.9732,
+      "step": 36830
+    },
+    {
+      "epoch": 0.2666724575995136,
+      "grad_norm": 0.1688418984413147,
+      "learning_rate": 4.733334781066545e-06,
+      "loss": 0.9744,
+      "step": 36840
+    },
+    {
+      "epoch": 0.26674484426009976,
+      "grad_norm": 0.18066221475601196,
+      "learning_rate": 4.733262394405959e-06,
+      "loss": 0.9735,
+      "step": 36850
+    },
+    {
+      "epoch": 0.26681723092068593,
+      "grad_norm": 0.15719619393348694,
+      "learning_rate": 4.733190007745373e-06,
+      "loss": 0.9797,
+      "step": 36860
+    },
+    {
+      "epoch": 0.2668896175812721,
+      "grad_norm": 0.1722886562347412,
+      "learning_rate": 4.733117621084787e-06,
+      "loss": 0.9744,
+      "step": 36870
+    },
+    {
+      "epoch": 0.2669620042418583,
+      "grad_norm": 0.19963102042675018,
+      "learning_rate": 4.733045234424201e-06,
+      "loss": 0.9828,
+      "step": 36880
+    },
+    {
+      "epoch": 0.2670343909024445,
+      "grad_norm": 0.16132299602031708,
+      "learning_rate": 4.732972847763614e-06,
+      "loss": 0.9712,
+      "step": 36890
+    },
+    {
+      "epoch": 0.2671067775630307,
+      "grad_norm": 0.16427114605903625,
+      "learning_rate": 4.732900461103028e-06,
+      "loss": 0.9779,
+      "step": 36900
+    },
+    {
+      "epoch": 0.26717916422361687,
+      "grad_norm": 0.18113258481025696,
+      "learning_rate": 4.732828074442442e-06,
+      "loss": 0.9688,
+      "step": 36910
+    },
+    {
+      "epoch": 0.26725155088420305,
+      "grad_norm": 0.1653916984796524,
+      "learning_rate": 4.732755687781856e-06,
+      "loss": 0.9717,
+      "step": 36920
+    },
+    {
+      "epoch": 0.2673239375447892,
+      "grad_norm": 0.179592102766037,
+      "learning_rate": 4.73268330112127e-06,
+      "loss": 0.9656,
+      "step": 36930
+    },
+    {
+      "epoch": 0.26739632420537546,
+      "grad_norm": 0.19769518077373505,
+      "learning_rate": 4.732610914460683e-06,
+      "loss": 0.9637,
+      "step": 36940
+    },
+    {
+      "epoch": 0.26746871086596163,
+      "grad_norm": 0.17146340012550354,
+      "learning_rate": 4.732538527800098e-06,
+      "loss": 0.9619,
+      "step": 36950
+    },
+    {
+      "epoch": 0.2675410975265478,
+      "grad_norm": 0.1564941555261612,
+      "learning_rate": 4.732466141139511e-06,
+      "loss": 0.9688,
+      "step": 36960
+    },
+    {
+      "epoch": 0.267613484187134,
+      "grad_norm": 0.1622597575187683,
+      "learning_rate": 4.732393754478925e-06,
+      "loss": 0.9848,
+      "step": 36970
+    },
+    {
+      "epoch": 0.26768587084772016,
+      "grad_norm": 0.17522403597831726,
+      "learning_rate": 4.7323213678183385e-06,
+      "loss": 0.9857,
+      "step": 36980
+    },
+    {
+      "epoch": 0.2677582575083064,
+      "grad_norm": 0.18009519577026367,
+      "learning_rate": 4.732248981157753e-06,
+      "loss": 0.9746,
+      "step": 36990
+    },
+    {
+      "epoch": 0.2678306441688926,
+      "grad_norm": 0.16934752464294434,
+      "learning_rate": 4.732176594497167e-06,
+      "loss": 0.9715,
+      "step": 37000
+    },
+    {
+      "epoch": 0.26790303082947875,
+      "grad_norm": 0.17400366067886353,
+      "learning_rate": 4.73210420783658e-06,
+      "loss": 0.985,
+      "step": 37010
+    },
+    {
+      "epoch": 0.2679754174900649,
+      "grad_norm": 0.15376758575439453,
+      "learning_rate": 4.732031821175994e-06,
+      "loss": 0.976,
+      "step": 37020
+    },
+    {
+      "epoch": 0.2680478041506511,
+      "grad_norm": 0.17052818834781647,
+      "learning_rate": 4.731959434515408e-06,
+      "loss": 0.9854,
+      "step": 37030
+    },
+    {
+      "epoch": 0.2681201908112373,
+      "grad_norm": 0.16806358098983765,
+      "learning_rate": 4.731887047854822e-06,
+      "loss": 0.9739,
+      "step": 37040
+    },
+    {
+      "epoch": 0.2681925774718235,
+      "grad_norm": 0.16140809655189514,
+      "learning_rate": 4.7318146611942355e-06,
+      "loss": 0.9864,
+      "step": 37050
+    },
+    {
+      "epoch": 0.2682649641324097,
+      "grad_norm": 0.16904619336128235,
+      "learning_rate": 4.731742274533649e-06,
+      "loss": 0.9594,
+      "step": 37060
+    },
+    {
+      "epoch": 0.26833735079299587,
+      "grad_norm": 0.1671932488679886,
+      "learning_rate": 4.731669887873064e-06,
+      "loss": 0.9694,
+      "step": 37070
+    },
+    {
+      "epoch": 0.26840973745358204,
+      "grad_norm": 0.17153862118721008,
+      "learning_rate": 4.731597501212477e-06,
+      "loss": 0.9811,
+      "step": 37080
+    },
+    {
+      "epoch": 0.2684821241141682,
+      "grad_norm": 0.16389037668704987,
+      "learning_rate": 4.731525114551891e-06,
+      "loss": 0.9758,
+      "step": 37090
+    },
+    {
+      "epoch": 0.26855451077475445,
+      "grad_norm": 0.16687390208244324,
+      "learning_rate": 4.7314527278913045e-06,
+      "loss": 0.9687,
+      "step": 37100
+    },
+    {
+      "epoch": 0.26862689743534063,
+      "grad_norm": 0.2017953097820282,
+      "learning_rate": 4.731380341230719e-06,
+      "loss": 0.9653,
+      "step": 37110
+    },
+    {
+      "epoch": 0.2686992840959268,
+      "grad_norm": 0.15654923021793365,
+      "learning_rate": 4.7313079545701326e-06,
+      "loss": 0.98,
+      "step": 37120
+    },
+    {
+      "epoch": 0.268771670756513,
+      "grad_norm": 0.18685537576675415,
+      "learning_rate": 4.731235567909546e-06,
+      "loss": 0.9759,
+      "step": 37130
+    },
+    {
+      "epoch": 0.26884405741709916,
+      "grad_norm": 0.1896306425333023,
+      "learning_rate": 4.73116318124896e-06,
+      "loss": 0.975,
+      "step": 37140
+    },
+    {
+      "epoch": 0.2689164440776854,
+      "grad_norm": 0.16155771911144257,
+      "learning_rate": 4.731090794588374e-06,
+      "loss": 0.9734,
+      "step": 37150
+    },
+    {
+      "epoch": 0.26898883073827157,
+      "grad_norm": 0.18064841628074646,
+      "learning_rate": 4.731018407927788e-06,
+      "loss": 0.9724,
+      "step": 37160
+    },
+    {
+      "epoch": 0.26906121739885774,
+      "grad_norm": 0.1620803028345108,
+      "learning_rate": 4.7309460212672015e-06,
+      "loss": 0.9608,
+      "step": 37170
+    },
+    {
+      "epoch": 0.2691336040594439,
+      "grad_norm": 0.15905041992664337,
+      "learning_rate": 4.730873634606615e-06,
+      "loss": 0.974,
+      "step": 37180
+    },
+    {
+      "epoch": 0.2692059907200301,
+      "grad_norm": 0.16241392493247986,
+      "learning_rate": 4.730801247946029e-06,
+      "loss": 0.971,
+      "step": 37190
+    },
+    {
+      "epoch": 0.2692783773806163,
+      "grad_norm": 0.1652408093214035,
+      "learning_rate": 4.730728861285443e-06,
+      "loss": 0.9561,
+      "step": 37200
+    },
+    {
+      "epoch": 0.2693507640412025,
+      "grad_norm": 0.1611439734697342,
+      "learning_rate": 4.730656474624857e-06,
+      "loss": 0.9889,
+      "step": 37210
+    },
+    {
+      "epoch": 0.2694231507017887,
+      "grad_norm": 0.16467610001564026,
+      "learning_rate": 4.73058408796427e-06,
+      "loss": 0.9639,
+      "step": 37220
+    },
+    {
+      "epoch": 0.26949553736237486,
+      "grad_norm": 0.17998045682907104,
+      "learning_rate": 4.730511701303684e-06,
+      "loss": 0.9692,
+      "step": 37230
+    },
+    {
+      "epoch": 0.26956792402296104,
+      "grad_norm": 0.16246424615383148,
+      "learning_rate": 4.730439314643098e-06,
+      "loss": 0.981,
+      "step": 37240
+    },
+    {
+      "epoch": 0.2696403106835472,
+      "grad_norm": 0.15857288241386414,
+      "learning_rate": 4.730366927982511e-06,
+      "loss": 0.9596,
+      "step": 37250
+    },
+    {
+      "epoch": 0.26971269734413345,
+      "grad_norm": 0.17110489308834076,
+      "learning_rate": 4.730294541321926e-06,
+      "loss": 0.9907,
+      "step": 37260
+    },
+    {
+      "epoch": 0.2697850840047196,
+      "grad_norm": 0.16102251410484314,
+      "learning_rate": 4.730222154661339e-06,
+      "loss": 0.9728,
+      "step": 37270
+    },
+    {
+      "epoch": 0.2698574706653058,
+      "grad_norm": 0.15859219431877136,
+      "learning_rate": 4.730149768000753e-06,
+      "loss": 0.9693,
+      "step": 37280
+    },
+    {
+      "epoch": 0.269929857325892,
+      "grad_norm": 0.16720248758792877,
+      "learning_rate": 4.7300773813401666e-06,
+      "loss": 0.975,
+      "step": 37290
+    },
+    {
+      "epoch": 0.27000224398647815,
+      "grad_norm": 0.1726258546113968,
+      "learning_rate": 4.730004994679581e-06,
+      "loss": 0.9823,
+      "step": 37300
+    },
+    {
+      "epoch": 0.2700746306470644,
+      "grad_norm": 0.17747798562049866,
+      "learning_rate": 4.729932608018995e-06,
+      "loss": 0.9637,
+      "step": 37310
+    },
+    {
+      "epoch": 0.27014701730765056,
+      "grad_norm": 0.17301778495311737,
+      "learning_rate": 4.729860221358408e-06,
+      "loss": 0.9755,
+      "step": 37320
+    },
+    {
+      "epoch": 0.27021940396823674,
+      "grad_norm": 0.1715429127216339,
+      "learning_rate": 4.729787834697822e-06,
+      "loss": 0.9838,
+      "step": 37330
+    },
+    {
+      "epoch": 0.2702917906288229,
+      "grad_norm": 0.17954185605049133,
+      "learning_rate": 4.729715448037236e-06,
+      "loss": 0.981,
+      "step": 37340
+    },
+    {
+      "epoch": 0.2703641772894091,
+      "grad_norm": 0.1605345904827118,
+      "learning_rate": 4.72964306137665e-06,
+      "loss": 0.9726,
+      "step": 37350
+    },
+    {
+      "epoch": 0.27043656394999527,
+      "grad_norm": 0.1734982430934906,
+      "learning_rate": 4.7295706747160636e-06,
+      "loss": 0.9621,
+      "step": 37360
+    },
+    {
+      "epoch": 0.2705089506105815,
+      "grad_norm": 0.16955071687698364,
+      "learning_rate": 4.729498288055477e-06,
+      "loss": 0.9831,
+      "step": 37370
+    },
+    {
+      "epoch": 0.2705813372711677,
+      "grad_norm": 0.18525801599025726,
+      "learning_rate": 4.729425901394891e-06,
+      "loss": 0.9803,
+      "step": 37380
+    },
+    {
+      "epoch": 0.27065372393175385,
+      "grad_norm": 0.1983799785375595,
+      "learning_rate": 4.729353514734305e-06,
+      "loss": 0.9768,
+      "step": 37390
+    },
+    {
+      "epoch": 0.27072611059234003,
+      "grad_norm": 0.169651597738266,
+      "learning_rate": 4.729281128073719e-06,
+      "loss": 0.9703,
+      "step": 37400
+    },
+    {
+      "epoch": 0.2707984972529262,
+      "grad_norm": 0.20270761847496033,
+      "learning_rate": 4.7292087414131325e-06,
+      "loss": 0.9909,
+      "step": 37410
+    },
+    {
+      "epoch": 0.27087088391351244,
+      "grad_norm": 0.19444452226161957,
+      "learning_rate": 4.729136354752546e-06,
+      "loss": 0.9811,
+      "step": 37420
+    },
+    {
+      "epoch": 0.2709432705740986,
+      "grad_norm": 0.17284293472766876,
+      "learning_rate": 4.729063968091961e-06,
+      "loss": 0.9821,
+      "step": 37430
+    },
+    {
+      "epoch": 0.2710156572346848,
+      "grad_norm": 0.1620352417230606,
+      "learning_rate": 4.728991581431374e-06,
+      "loss": 0.9725,
+      "step": 37440
+    },
+    {
+      "epoch": 0.27108804389527097,
+      "grad_norm": 0.17737331986427307,
+      "learning_rate": 4.728919194770788e-06,
+      "loss": 0.9739,
+      "step": 37450
+    },
+    {
+      "epoch": 0.27116043055585715,
+      "grad_norm": 0.15988625586032867,
+      "learning_rate": 4.7288468081102014e-06,
+      "loss": 0.9737,
+      "step": 37460
+    },
+    {
+      "epoch": 0.2712328172164434,
+      "grad_norm": 0.1713685542345047,
+      "learning_rate": 4.728774421449616e-06,
+      "loss": 0.98,
+      "step": 37470
+    },
+    {
+      "epoch": 0.27130520387702955,
+      "grad_norm": 0.17601056396961212,
+      "learning_rate": 4.7287020347890295e-06,
+      "loss": 0.9833,
+      "step": 37480
+    },
+    {
+      "epoch": 0.27137759053761573,
+      "grad_norm": 0.16959619522094727,
+      "learning_rate": 4.728629648128443e-06,
+      "loss": 0.9658,
+      "step": 37490
+    },
+    {
+      "epoch": 0.2714499771982019,
+      "grad_norm": 0.174661785364151,
+      "learning_rate": 4.728557261467857e-06,
+      "loss": 0.9721,
+      "step": 37500
+    },
+    {
+      "epoch": 0.2715223638587881,
+      "grad_norm": 0.16935496032238007,
+      "learning_rate": 4.728484874807271e-06,
+      "loss": 0.963,
+      "step": 37510
+    },
+    {
+      "epoch": 0.2715947505193743,
+      "grad_norm": 0.1676948070526123,
+      "learning_rate": 4.728412488146685e-06,
+      "loss": 0.969,
+      "step": 37520
+    },
+    {
+      "epoch": 0.2716671371799605,
+      "grad_norm": 0.17413829267024994,
+      "learning_rate": 4.7283401014860984e-06,
+      "loss": 0.9767,
+      "step": 37530
+    },
+    {
+      "epoch": 0.27173952384054667,
+      "grad_norm": 0.1575450748205185,
+      "learning_rate": 4.728267714825512e-06,
+      "loss": 0.9855,
+      "step": 37540
+    },
+    {
+      "epoch": 0.27181191050113285,
+      "grad_norm": 0.15776337683200836,
+      "learning_rate": 4.7281953281649265e-06,
+      "loss": 0.9669,
+      "step": 37550
+    },
+    {
+      "epoch": 0.271884297161719,
+      "grad_norm": 0.16901899874210358,
+      "learning_rate": 4.72812294150434e-06,
+      "loss": 0.9696,
+      "step": 37560
+    },
+    {
+      "epoch": 0.2719566838223052,
+      "grad_norm": 0.18395544588565826,
+      "learning_rate": 4.728050554843754e-06,
+      "loss": 0.9743,
+      "step": 37570
+    },
+    {
+      "epoch": 0.27202907048289143,
+      "grad_norm": 0.15982641279697418,
+      "learning_rate": 4.727978168183167e-06,
+      "loss": 0.9691,
+      "step": 37580
+    },
+    {
+      "epoch": 0.2721014571434776,
+      "grad_norm": 0.16578301787376404,
+      "learning_rate": 4.727905781522582e-06,
+      "loss": 0.9807,
+      "step": 37590
+    },
+    {
+      "epoch": 0.2721738438040638,
+      "grad_norm": 0.1732582002878189,
+      "learning_rate": 4.7278333948619954e-06,
+      "loss": 0.9776,
+      "step": 37600
+    },
+    {
+      "epoch": 0.27224623046464996,
+      "grad_norm": 0.1836758553981781,
+      "learning_rate": 4.727761008201409e-06,
+      "loss": 0.9741,
+      "step": 37610
+    },
+    {
+      "epoch": 0.27231861712523614,
+      "grad_norm": 0.1627260446548462,
+      "learning_rate": 4.727688621540823e-06,
+      "loss": 0.9747,
+      "step": 37620
+    },
+    {
+      "epoch": 0.27239100378582237,
+      "grad_norm": 0.16746656596660614,
+      "learning_rate": 4.727616234880237e-06,
+      "loss": 0.9808,
+      "step": 37630
+    },
+    {
+      "epoch": 0.27246339044640855,
+      "grad_norm": 0.18597693741321564,
+      "learning_rate": 4.727543848219651e-06,
+      "loss": 0.9778,
+      "step": 37640
+    },
+    {
+      "epoch": 0.2725357771069947,
+      "grad_norm": 0.1623985916376114,
+      "learning_rate": 4.727471461559064e-06,
+      "loss": 0.9741,
+      "step": 37650
+    },
+    {
+      "epoch": 0.2726081637675809,
+      "grad_norm": 0.1913863867521286,
+      "learning_rate": 4.727399074898478e-06,
+      "loss": 0.9838,
+      "step": 37660
+    },
+    {
+      "epoch": 0.2726805504281671,
+      "grad_norm": 0.1891731321811676,
+      "learning_rate": 4.7273266882378925e-06,
+      "loss": 0.9743,
+      "step": 37670
+    },
+    {
+      "epoch": 0.2727529370887533,
+      "grad_norm": 0.1674305498600006,
+      "learning_rate": 4.727254301577306e-06,
+      "loss": 0.9798,
+      "step": 37680
+    },
+    {
+      "epoch": 0.2728253237493395,
+      "grad_norm": 0.1666889786720276,
+      "learning_rate": 4.72718191491672e-06,
+      "loss": 0.9566,
+      "step": 37690
+    },
+    {
+      "epoch": 0.27289771040992566,
+      "grad_norm": 0.17124493420124054,
+      "learning_rate": 4.727109528256133e-06,
+      "loss": 0.9701,
+      "step": 37700
+    },
+    {
+      "epoch": 0.27297009707051184,
+      "grad_norm": 0.17866984009742737,
+      "learning_rate": 4.727037141595548e-06,
+      "loss": 0.9893,
+      "step": 37710
+    },
+    {
+      "epoch": 0.273042483731098,
+      "grad_norm": 0.16738882660865784,
+      "learning_rate": 4.726964754934961e-06,
+      "loss": 0.9757,
+      "step": 37720
+    },
+    {
+      "epoch": 0.2731148703916842,
+      "grad_norm": 0.1580776870250702,
+      "learning_rate": 4.726892368274375e-06,
+      "loss": 0.9758,
+      "step": 37730
+    },
+    {
+      "epoch": 0.2731872570522704,
+      "grad_norm": 0.1834014505147934,
+      "learning_rate": 4.726819981613789e-06,
+      "loss": 0.9687,
+      "step": 37740
+    },
+    {
+      "epoch": 0.2732596437128566,
+      "grad_norm": 0.15350131690502167,
+      "learning_rate": 4.726747594953203e-06,
+      "loss": 0.9674,
+      "step": 37750
+    },
+    {
+      "epoch": 0.2733320303734428,
+      "grad_norm": 0.1681162267923355,
+      "learning_rate": 4.726675208292616e-06,
+      "loss": 0.9744,
+      "step": 37760
+    },
+    {
+      "epoch": 0.27340441703402896,
+      "grad_norm": 0.17525209486484528,
+      "learning_rate": 4.7266028216320295e-06,
+      "loss": 0.9663,
+      "step": 37770
+    },
+    {
+      "epoch": 0.27347680369461513,
+      "grad_norm": 0.16645973920822144,
+      "learning_rate": 4.726530434971444e-06,
+      "loss": 0.9631,
+      "step": 37780
+    },
+    {
+      "epoch": 0.27354919035520137,
+      "grad_norm": 0.15926945209503174,
+      "learning_rate": 4.7264580483108575e-06,
+      "loss": 0.9706,
+      "step": 37790
+    },
+    {
+      "epoch": 0.27362157701578754,
+      "grad_norm": 0.18003691732883453,
+      "learning_rate": 4.726385661650271e-06,
+      "loss": 0.961,
+      "step": 37800
+    },
+    {
+      "epoch": 0.2736939636763737,
+      "grad_norm": 0.1554681807756424,
+      "learning_rate": 4.726313274989685e-06,
+      "loss": 0.9708,
+      "step": 37810
+    },
+    {
+      "epoch": 0.2737663503369599,
+      "grad_norm": 0.165873184800148,
+      "learning_rate": 4.726240888329099e-06,
+      "loss": 0.9873,
+      "step": 37820
+    },
+    {
+      "epoch": 0.2738387369975461,
+      "grad_norm": 0.16069376468658447,
+      "learning_rate": 4.726168501668513e-06,
+      "loss": 0.9704,
+      "step": 37830
+    },
+    {
+      "epoch": 0.2739111236581323,
+      "grad_norm": 0.16646680235862732,
+      "learning_rate": 4.7260961150079265e-06,
+      "loss": 0.9794,
+      "step": 37840
+    },
+    {
+      "epoch": 0.2739835103187185,
+      "grad_norm": 0.1622561812400818,
+      "learning_rate": 4.72602372834734e-06,
+      "loss": 0.9655,
+      "step": 37850
+    },
+    {
+      "epoch": 0.27405589697930466,
+      "grad_norm": 0.17769238352775574,
+      "learning_rate": 4.7259513416867546e-06,
+      "loss": 0.9845,
+      "step": 37860
+    },
+    {
+      "epoch": 0.27412828363989084,
+      "grad_norm": 0.1598799079656601,
+      "learning_rate": 4.725878955026168e-06,
+      "loss": 0.9684,
+      "step": 37870
+    },
+    {
+      "epoch": 0.274200670300477,
+      "grad_norm": 0.15777985751628876,
+      "learning_rate": 4.725806568365582e-06,
+      "loss": 0.9754,
+      "step": 37880
+    },
+    {
+      "epoch": 0.2742730569610632,
+      "grad_norm": 0.16836735606193542,
+      "learning_rate": 4.725734181704995e-06,
+      "loss": 0.9621,
+      "step": 37890
+    },
+    {
+      "epoch": 0.2743454436216494,
+      "grad_norm": 0.1695423424243927,
+      "learning_rate": 4.72566179504441e-06,
+      "loss": 0.9573,
+      "step": 37900
+    },
+    {
+      "epoch": 0.2744178302822356,
+      "grad_norm": 0.17105446755886078,
+      "learning_rate": 4.7255894083838235e-06,
+      "loss": 0.9736,
+      "step": 37910
+    },
+    {
+      "epoch": 0.2744902169428218,
+      "grad_norm": 0.17636360228061676,
+      "learning_rate": 4.725517021723237e-06,
+      "loss": 0.9589,
+      "step": 37920
+    },
+    {
+      "epoch": 0.27456260360340795,
+      "grad_norm": 0.2081209272146225,
+      "learning_rate": 4.725444635062651e-06,
+      "loss": 0.9717,
+      "step": 37930
+    },
+    {
+      "epoch": 0.2746349902639941,
+      "grad_norm": 0.1700514554977417,
+      "learning_rate": 4.725372248402065e-06,
+      "loss": 0.9788,
+      "step": 37940
+    },
+    {
+      "epoch": 0.27470737692458036,
+      "grad_norm": 0.18585331737995148,
+      "learning_rate": 4.725299861741479e-06,
+      "loss": 0.9673,
+      "step": 37950
+    },
+    {
+      "epoch": 0.27477976358516654,
+      "grad_norm": 0.1638772189617157,
+      "learning_rate": 4.725227475080892e-06,
+      "loss": 0.977,
+      "step": 37960
+    },
+    {
+      "epoch": 0.2748521502457527,
+      "grad_norm": 0.16300378739833832,
+      "learning_rate": 4.725155088420306e-06,
+      "loss": 0.9904,
+      "step": 37970
+    },
+    {
+      "epoch": 0.2749245369063389,
+      "grad_norm": 0.16224254667758942,
+      "learning_rate": 4.72508270175972e-06,
+      "loss": 0.9626,
+      "step": 37980
+    },
+    {
+      "epoch": 0.27499692356692507,
+      "grad_norm": 0.1576361060142517,
+      "learning_rate": 4.725010315099134e-06,
+      "loss": 0.9815,
+      "step": 37990
+    },
+    {
+      "epoch": 0.2750693102275113,
+      "grad_norm": 0.265109121799469,
+      "learning_rate": 4.724937928438548e-06,
+      "loss": 0.9741,
+      "step": 38000
+    },
+    {
+      "epoch": 0.2751416968880975,
+      "grad_norm": 0.16923700273036957,
+      "learning_rate": 4.724865541777961e-06,
+      "loss": 0.9641,
+      "step": 38010
+    },
+    {
+      "epoch": 0.27521408354868365,
+      "grad_norm": 0.15976910293102264,
+      "learning_rate": 4.724793155117375e-06,
+      "loss": 0.9833,
+      "step": 38020
+    },
+    {
+      "epoch": 0.27528647020926983,
+      "grad_norm": 0.1535775065422058,
+      "learning_rate": 4.724720768456789e-06,
+      "loss": 0.985,
+      "step": 38030
+    },
+    {
+      "epoch": 0.275358856869856,
+      "grad_norm": 0.1646791249513626,
+      "learning_rate": 4.724648381796203e-06,
+      "loss": 0.9652,
+      "step": 38040
+    },
+    {
+      "epoch": 0.2754312435304422,
+      "grad_norm": 0.17050626873970032,
+      "learning_rate": 4.724575995135617e-06,
+      "loss": 0.9634,
+      "step": 38050
+    },
+    {
+      "epoch": 0.2755036301910284,
+      "grad_norm": 0.16307681798934937,
+      "learning_rate": 4.72450360847503e-06,
+      "loss": 0.9639,
+      "step": 38060
+    },
+    {
+      "epoch": 0.2755760168516146,
+      "grad_norm": 0.17553727328777313,
+      "learning_rate": 4.724431221814445e-06,
+      "loss": 0.9771,
+      "step": 38070
+    },
+    {
+      "epoch": 0.27564840351220077,
+      "grad_norm": 0.1655888557434082,
+      "learning_rate": 4.724358835153858e-06,
+      "loss": 0.9717,
+      "step": 38080
+    },
+    {
+      "epoch": 0.27572079017278694,
+      "grad_norm": 0.1691659390926361,
+      "learning_rate": 4.724286448493272e-06,
+      "loss": 0.9491,
+      "step": 38090
+    },
+    {
+      "epoch": 0.2757931768333731,
+      "grad_norm": 0.15853264927864075,
+      "learning_rate": 4.7242140618326856e-06,
+      "loss": 0.9666,
+      "step": 38100
+    },
+    {
+      "epoch": 0.27586556349395935,
+      "grad_norm": 0.15642456710338593,
+      "learning_rate": 4.7241416751721e-06,
+      "loss": 0.9665,
+      "step": 38110
+    },
+    {
+      "epoch": 0.27593795015454553,
+      "grad_norm": 0.16785617172718048,
+      "learning_rate": 4.724069288511514e-06,
+      "loss": 0.9697,
+      "step": 38120
+    },
+    {
+      "epoch": 0.2760103368151317,
+      "grad_norm": 0.2900194823741913,
+      "learning_rate": 4.723996901850927e-06,
+      "loss": 0.9717,
+      "step": 38130
+    },
+    {
+      "epoch": 0.2760827234757179,
+      "grad_norm": 0.23790918290615082,
+      "learning_rate": 4.723924515190341e-06,
+      "loss": 0.9698,
+      "step": 38140
+    },
+    {
+      "epoch": 0.27615511013630406,
+      "grad_norm": 0.15893487632274628,
+      "learning_rate": 4.723852128529755e-06,
+      "loss": 0.975,
+      "step": 38150
+    },
+    {
+      "epoch": 0.2762274967968903,
+      "grad_norm": 0.15974752604961395,
+      "learning_rate": 4.723779741869169e-06,
+      "loss": 0.9819,
+      "step": 38160
+    },
+    {
+      "epoch": 0.27629988345747647,
+      "grad_norm": 0.16646121442317963,
+      "learning_rate": 4.723707355208583e-06,
+      "loss": 0.9664,
+      "step": 38170
+    },
+    {
+      "epoch": 0.27637227011806265,
+      "grad_norm": 0.15674103796482086,
+      "learning_rate": 4.723634968547996e-06,
+      "loss": 0.9783,
+      "step": 38180
+    },
+    {
+      "epoch": 0.2764446567786488,
+      "grad_norm": 0.1640247255563736,
+      "learning_rate": 4.723562581887411e-06,
+      "loss": 0.9746,
+      "step": 38190
+    },
+    {
+      "epoch": 0.276517043439235,
+      "grad_norm": 0.18196041882038116,
+      "learning_rate": 4.723490195226824e-06,
+      "loss": 0.9697,
+      "step": 38200
+    },
+    {
+      "epoch": 0.27658943009982123,
+      "grad_norm": 0.16162289679050446,
+      "learning_rate": 4.723417808566238e-06,
+      "loss": 0.9823,
+      "step": 38210
+    },
+    {
+      "epoch": 0.2766618167604074,
+      "grad_norm": 0.16687557101249695,
+      "learning_rate": 4.7233454219056515e-06,
+      "loss": 0.9688,
+      "step": 38220
+    },
+    {
+      "epoch": 0.2767342034209936,
+      "grad_norm": 0.16218598186969757,
+      "learning_rate": 4.723273035245066e-06,
+      "loss": 0.9888,
+      "step": 38230
+    },
+    {
+      "epoch": 0.27680659008157976,
+      "grad_norm": 0.15919815003871918,
+      "learning_rate": 4.72320064858448e-06,
+      "loss": 0.9664,
+      "step": 38240
+    },
+    {
+      "epoch": 0.27687897674216594,
+      "grad_norm": 0.15851202607154846,
+      "learning_rate": 4.723128261923893e-06,
+      "loss": 0.9657,
+      "step": 38250
+    },
+    {
+      "epoch": 0.2769513634027521,
+      "grad_norm": 0.17171426117420197,
+      "learning_rate": 4.723055875263307e-06,
+      "loss": 0.9671,
+      "step": 38260
+    },
+    {
+      "epoch": 0.27702375006333835,
+      "grad_norm": 0.1656324565410614,
+      "learning_rate": 4.722983488602721e-06,
+      "loss": 0.9644,
+      "step": 38270
+    },
+    {
+      "epoch": 0.2770961367239245,
+      "grad_norm": 0.16705715656280518,
+      "learning_rate": 4.722911101942135e-06,
+      "loss": 0.969,
+      "step": 38280
+    },
+    {
+      "epoch": 0.2771685233845107,
+      "grad_norm": 0.1705339401960373,
+      "learning_rate": 4.722838715281548e-06,
+      "loss": 0.9657,
+      "step": 38290
+    },
+    {
+      "epoch": 0.2772409100450969,
+      "grad_norm": 0.18638475239276886,
+      "learning_rate": 4.722766328620962e-06,
+      "loss": 0.9715,
+      "step": 38300
+    },
+    {
+      "epoch": 0.27731329670568305,
+      "grad_norm": 0.16748540103435516,
+      "learning_rate": 4.722693941960376e-06,
+      "loss": 0.9616,
+      "step": 38310
+    },
+    {
+      "epoch": 0.2773856833662693,
+      "grad_norm": 0.15937179327011108,
+      "learning_rate": 4.722621555299789e-06,
+      "loss": 0.9764,
+      "step": 38320
+    },
+    {
+      "epoch": 0.27745807002685546,
+      "grad_norm": 0.19340196251869202,
+      "learning_rate": 4.722549168639203e-06,
+      "loss": 0.9746,
+      "step": 38330
+    },
+    {
+      "epoch": 0.27753045668744164,
+      "grad_norm": 0.17282438278198242,
+      "learning_rate": 4.7224767819786174e-06,
+      "loss": 0.9691,
+      "step": 38340
+    },
+    {
+      "epoch": 0.2776028433480278,
+      "grad_norm": 0.17909948527812958,
+      "learning_rate": 4.722404395318031e-06,
+      "loss": 0.9802,
+      "step": 38350
+    },
+    {
+      "epoch": 0.277675230008614,
+      "grad_norm": 0.1845092922449112,
+      "learning_rate": 4.722332008657445e-06,
+      "loss": 0.9723,
+      "step": 38360
+    },
+    {
+      "epoch": 0.2777476166692002,
+      "grad_norm": 0.18336127698421478,
+      "learning_rate": 4.722259621996858e-06,
+      "loss": 0.9782,
+      "step": 38370
+    },
+    {
+      "epoch": 0.2778200033297864,
+      "grad_norm": 0.1571909338235855,
+      "learning_rate": 4.722187235336273e-06,
+      "loss": 0.9594,
+      "step": 38380
+    },
+    {
+      "epoch": 0.2778923899903726,
+      "grad_norm": 0.18232674896717072,
+      "learning_rate": 4.722114848675686e-06,
+      "loss": 0.9559,
+      "step": 38390
+    },
+    {
+      "epoch": 0.27796477665095876,
+      "grad_norm": 0.20004482567310333,
+      "learning_rate": 4.7220424620151e-06,
+      "loss": 0.9725,
+      "step": 38400
+    },
+    {
+      "epoch": 0.27803716331154493,
+      "grad_norm": 0.18284298479557037,
+      "learning_rate": 4.721970075354514e-06,
+      "loss": 0.9719,
+      "step": 38410
+    },
+    {
+      "epoch": 0.2781095499721311,
+      "grad_norm": 0.16327325999736786,
+      "learning_rate": 4.721897688693928e-06,
+      "loss": 0.9698,
+      "step": 38420
+    },
+    {
+      "epoch": 0.27818193663271734,
+      "grad_norm": 0.16780653595924377,
+      "learning_rate": 4.721825302033342e-06,
+      "loss": 0.9678,
+      "step": 38430
+    },
+    {
+      "epoch": 0.2782543232933035,
+      "grad_norm": 0.16471821069717407,
+      "learning_rate": 4.721752915372755e-06,
+      "loss": 0.9706,
+      "step": 38440
+    },
+    {
+      "epoch": 0.2783267099538897,
+      "grad_norm": 0.1728796362876892,
+      "learning_rate": 4.721680528712169e-06,
+      "loss": 0.9657,
+      "step": 38450
+    },
+    {
+      "epoch": 0.27839909661447587,
+      "grad_norm": 0.15473975241184235,
+      "learning_rate": 4.721608142051583e-06,
+      "loss": 0.9683,
+      "step": 38460
+    },
+    {
+      "epoch": 0.27847148327506205,
+      "grad_norm": 0.15741673111915588,
+      "learning_rate": 4.721535755390997e-06,
+      "loss": 0.9775,
+      "step": 38470
+    },
+    {
+      "epoch": 0.2785438699356483,
+      "grad_norm": 0.1729719191789627,
+      "learning_rate": 4.721463368730411e-06,
+      "loss": 0.973,
+      "step": 38480
+    },
+    {
+      "epoch": 0.27861625659623446,
+      "grad_norm": 0.16401894390583038,
+      "learning_rate": 4.721390982069824e-06,
+      "loss": 0.9589,
+      "step": 38490
+    },
+    {
+      "epoch": 0.27868864325682063,
+      "grad_norm": 0.16134223341941833,
+      "learning_rate": 4.721318595409239e-06,
+      "loss": 0.9797,
+      "step": 38500
+    },
+    {
+      "epoch": 0.2787610299174068,
+      "grad_norm": 0.1627780795097351,
+      "learning_rate": 4.721246208748652e-06,
+      "loss": 0.976,
+      "step": 38510
+    },
+    {
+      "epoch": 0.278833416577993,
+      "grad_norm": 0.15430185198783875,
+      "learning_rate": 4.721173822088066e-06,
+      "loss": 0.9683,
+      "step": 38520
+    },
+    {
+      "epoch": 0.2789058032385792,
+      "grad_norm": 0.18733610212802887,
+      "learning_rate": 4.7211014354274795e-06,
+      "loss": 0.9688,
+      "step": 38530
+    },
+    {
+      "epoch": 0.2789781898991654,
+      "grad_norm": 0.16492965817451477,
+      "learning_rate": 4.721029048766894e-06,
+      "loss": 0.9683,
+      "step": 38540
+    },
+    {
+      "epoch": 0.2790505765597516,
+      "grad_norm": 0.1765565425157547,
+      "learning_rate": 4.720956662106308e-06,
+      "loss": 0.9645,
+      "step": 38550
+    },
+    {
+      "epoch": 0.27912296322033775,
+      "grad_norm": 0.18229708075523376,
+      "learning_rate": 4.720884275445721e-06,
+      "loss": 0.9562,
+      "step": 38560
+    },
+    {
+      "epoch": 0.2791953498809239,
+      "grad_norm": 0.175114244222641,
+      "learning_rate": 4.720811888785135e-06,
+      "loss": 0.959,
+      "step": 38570
+    },
+    {
+      "epoch": 0.2792677365415101,
+      "grad_norm": 0.18212807178497314,
+      "learning_rate": 4.720739502124549e-06,
+      "loss": 0.9662,
+      "step": 38580
+    },
+    {
+      "epoch": 0.27934012320209634,
+      "grad_norm": 0.20221389830112457,
+      "learning_rate": 4.720667115463963e-06,
+      "loss": 0.9644,
+      "step": 38590
+    },
+    {
+      "epoch": 0.2794125098626825,
+      "grad_norm": 0.16933338344097137,
+      "learning_rate": 4.7205947288033766e-06,
+      "loss": 0.9783,
+      "step": 38600
+    },
+    {
+      "epoch": 0.2794848965232687,
+      "grad_norm": 0.16471229493618011,
+      "learning_rate": 4.72052234214279e-06,
+      "loss": 0.9816,
+      "step": 38610
+    },
+    {
+      "epoch": 0.27955728318385487,
+      "grad_norm": 0.16443496942520142,
+      "learning_rate": 4.720449955482204e-06,
+      "loss": 0.9757,
+      "step": 38620
+    },
+    {
+      "epoch": 0.27962966984444104,
+      "grad_norm": 0.17601965367794037,
+      "learning_rate": 4.720377568821618e-06,
+      "loss": 0.9697,
+      "step": 38630
+    },
+    {
+      "epoch": 0.2797020565050273,
+      "grad_norm": 0.1578618586063385,
+      "learning_rate": 4.720305182161032e-06,
+      "loss": 0.9539,
+      "step": 38640
+    },
+    {
+      "epoch": 0.27977444316561345,
+      "grad_norm": 0.1648857295513153,
+      "learning_rate": 4.7202327955004455e-06,
+      "loss": 0.9604,
+      "step": 38650
+    },
+    {
+      "epoch": 0.27984682982619963,
+      "grad_norm": 0.1769489347934723,
+      "learning_rate": 4.720160408839859e-06,
+      "loss": 0.9612,
+      "step": 38660
+    },
+    {
+      "epoch": 0.2799192164867858,
+      "grad_norm": 0.17651750147342682,
+      "learning_rate": 4.7200880221792736e-06,
+      "loss": 0.976,
+      "step": 38670
+    },
+    {
+      "epoch": 0.279991603147372,
+      "grad_norm": 0.1547250896692276,
+      "learning_rate": 4.720015635518687e-06,
+      "loss": 0.9679,
+      "step": 38680
+    },
+    {
+      "epoch": 0.2800639898079582,
+      "grad_norm": 0.15348443388938904,
+      "learning_rate": 4.719943248858101e-06,
+      "loss": 0.9766,
+      "step": 38690
+    },
+    {
+      "epoch": 0.2801363764685444,
+      "grad_norm": 0.1610197126865387,
+      "learning_rate": 4.719870862197514e-06,
+      "loss": 0.9837,
+      "step": 38700
+    },
+    {
+      "epoch": 0.28020876312913057,
+      "grad_norm": 0.21418476104736328,
+      "learning_rate": 4.719798475536929e-06,
+      "loss": 0.9665,
+      "step": 38710
+    },
+    {
+      "epoch": 0.28028114978971674,
+      "grad_norm": 0.17171064019203186,
+      "learning_rate": 4.7197260888763425e-06,
+      "loss": 0.9687,
+      "step": 38720
+    },
+    {
+      "epoch": 0.2803535364503029,
+      "grad_norm": 0.1554517149925232,
+      "learning_rate": 4.719653702215756e-06,
+      "loss": 0.9636,
+      "step": 38730
+    },
+    {
+      "epoch": 0.28042592311088915,
+      "grad_norm": 0.16150425374507904,
+      "learning_rate": 4.71958131555517e-06,
+      "loss": 0.978,
+      "step": 38740
+    },
+    {
+      "epoch": 0.28049830977147533,
+      "grad_norm": 0.16200552880764008,
+      "learning_rate": 4.719508928894584e-06,
+      "loss": 0.9618,
+      "step": 38750
+    },
+    {
+      "epoch": 0.2805706964320615,
+      "grad_norm": 0.16305282711982727,
+      "learning_rate": 4.719436542233998e-06,
+      "loss": 0.9762,
+      "step": 38760
+    },
+    {
+      "epoch": 0.2806430830926477,
+      "grad_norm": 0.16002973914146423,
+      "learning_rate": 4.719364155573411e-06,
+      "loss": 0.9829,
+      "step": 38770
+    },
+    {
+      "epoch": 0.28071546975323386,
+      "grad_norm": 0.1760619580745697,
+      "learning_rate": 4.719291768912825e-06,
+      "loss": 0.9737,
+      "step": 38780
+    },
+    {
+      "epoch": 0.28078785641382004,
+      "grad_norm": 0.15010814368724823,
+      "learning_rate": 4.7192193822522395e-06,
+      "loss": 0.9658,
+      "step": 38790
+    },
+    {
+      "epoch": 0.28086024307440627,
+      "grad_norm": 0.1825767457485199,
+      "learning_rate": 4.719146995591653e-06,
+      "loss": 0.975,
+      "step": 38800
+    },
+    {
+      "epoch": 0.28093262973499245,
+      "grad_norm": 0.17386701703071594,
+      "learning_rate": 4.719074608931067e-06,
+      "loss": 0.9684,
+      "step": 38810
+    },
+    {
+      "epoch": 0.2810050163955786,
+      "grad_norm": 0.15754207968711853,
+      "learning_rate": 4.71900222227048e-06,
+      "loss": 0.9836,
+      "step": 38820
+    },
+    {
+      "epoch": 0.2810774030561648,
+      "grad_norm": 0.1641159951686859,
+      "learning_rate": 4.718929835609894e-06,
+      "loss": 0.9741,
+      "step": 38830
+    },
+    {
+      "epoch": 0.281149789716751,
+      "grad_norm": 0.16686658561229706,
+      "learning_rate": 4.7188574489493076e-06,
+      "loss": 0.9835,
+      "step": 38840
+    },
+    {
+      "epoch": 0.2812221763773372,
+      "grad_norm": 0.17113393545150757,
+      "learning_rate": 4.718785062288721e-06,
+      "loss": 0.9675,
+      "step": 38850
+    },
+    {
+      "epoch": 0.2812945630379234,
+      "grad_norm": 0.19541935622692108,
+      "learning_rate": 4.718712675628136e-06,
+      "loss": 0.971,
+      "step": 38860
+    },
+    {
+      "epoch": 0.28136694969850956,
+      "grad_norm": 0.171500563621521,
+      "learning_rate": 4.718640288967549e-06,
+      "loss": 0.9693,
+      "step": 38870
+    },
+    {
+      "epoch": 0.28143933635909574,
+      "grad_norm": 0.1516040712594986,
+      "learning_rate": 4.718567902306963e-06,
+      "loss": 0.9637,
+      "step": 38880
+    },
+    {
+      "epoch": 0.2815117230196819,
+      "grad_norm": 0.1661519557237625,
+      "learning_rate": 4.7184955156463765e-06,
+      "loss": 0.9658,
+      "step": 38890
+    },
+    {
+      "epoch": 0.28158410968026815,
+      "grad_norm": 0.16510553658008575,
+      "learning_rate": 4.718423128985791e-06,
+      "loss": 0.9776,
+      "step": 38900
+    },
+    {
+      "epoch": 0.2816564963408543,
+      "grad_norm": 0.386662095785141,
+      "learning_rate": 4.718350742325205e-06,
+      "loss": 0.964,
+      "step": 38910
+    },
+    {
+      "epoch": 0.2817288830014405,
+      "grad_norm": 0.1649596393108368,
+      "learning_rate": 4.718278355664618e-06,
+      "loss": 0.9717,
+      "step": 38920
+    },
+    {
+      "epoch": 0.2818012696620267,
+      "grad_norm": 0.17319610714912415,
+      "learning_rate": 4.718205969004032e-06,
+      "loss": 0.9752,
+      "step": 38930
+    },
+    {
+      "epoch": 0.28187365632261285,
+      "grad_norm": 0.16159221529960632,
+      "learning_rate": 4.718133582343446e-06,
+      "loss": 0.9815,
+      "step": 38940
+    },
+    {
+      "epoch": 0.28194604298319903,
+      "grad_norm": 0.17223593592643738,
+      "learning_rate": 4.71806119568286e-06,
+      "loss": 0.9752,
+      "step": 38950
+    },
+    {
+      "epoch": 0.28201842964378526,
+      "grad_norm": 0.15941330790519714,
+      "learning_rate": 4.7179888090222735e-06,
+      "loss": 0.97,
+      "step": 38960
+    },
+    {
+      "epoch": 0.28209081630437144,
+      "grad_norm": 0.17768092453479767,
+      "learning_rate": 4.717916422361687e-06,
+      "loss": 0.9608,
+      "step": 38970
+    },
+    {
+      "epoch": 0.2821632029649576,
+      "grad_norm": 0.18235769867897034,
+      "learning_rate": 4.717844035701102e-06,
+      "loss": 0.9795,
+      "step": 38980
+    },
+    {
+      "epoch": 0.2822355896255438,
+      "grad_norm": 0.1654520183801651,
+      "learning_rate": 4.717771649040515e-06,
+      "loss": 0.9743,
+      "step": 38990
+    },
+    {
+      "epoch": 0.28230797628612997,
+      "grad_norm": 0.17783403396606445,
+      "learning_rate": 4.717699262379929e-06,
+      "loss": 0.977,
+      "step": 39000
+    },
+    {
+      "epoch": 0.2823803629467162,
+      "grad_norm": 0.15796136856079102,
+      "learning_rate": 4.7176268757193424e-06,
+      "loss": 0.9702,
+      "step": 39010
+    },
+    {
+      "epoch": 0.2824527496073024,
+      "grad_norm": 0.16155339777469635,
+      "learning_rate": 4.717554489058757e-06,
+      "loss": 0.9734,
+      "step": 39020
+    },
+    {
+      "epoch": 0.28252513626788855,
+      "grad_norm": 0.16858787834644318,
+      "learning_rate": 4.7174821023981705e-06,
+      "loss": 0.9644,
+      "step": 39030
+    },
+    {
+      "epoch": 0.28259752292847473,
+      "grad_norm": 0.17033177614212036,
+      "learning_rate": 4.717409715737584e-06,
+      "loss": 0.981,
+      "step": 39040
+    },
+    {
+      "epoch": 0.2826699095890609,
+      "grad_norm": 0.17069222033023834,
+      "learning_rate": 4.717337329076998e-06,
+      "loss": 0.974,
+      "step": 39050
+    },
+    {
+      "epoch": 0.28274229624964714,
+      "grad_norm": 0.17629200220108032,
+      "learning_rate": 4.717264942416412e-06,
+      "loss": 0.9717,
+      "step": 39060
+    },
+    {
+      "epoch": 0.2828146829102333,
+      "grad_norm": 0.1703384816646576,
+      "learning_rate": 4.717192555755826e-06,
+      "loss": 0.9781,
+      "step": 39070
+    },
+    {
+      "epoch": 0.2828870695708195,
+      "grad_norm": 0.16142185032367706,
+      "learning_rate": 4.7171201690952394e-06,
+      "loss": 0.9576,
+      "step": 39080
+    },
+    {
+      "epoch": 0.28295945623140567,
+      "grad_norm": 0.1647561937570572,
+      "learning_rate": 4.717047782434653e-06,
+      "loss": 0.9738,
+      "step": 39090
+    },
+    {
+      "epoch": 0.28303184289199185,
+      "grad_norm": 0.16790105402469635,
+      "learning_rate": 4.7169753957740675e-06,
+      "loss": 0.9632,
+      "step": 39100
+    },
+    {
+      "epoch": 0.283104229552578,
+      "grad_norm": 0.16049116849899292,
+      "learning_rate": 4.716903009113481e-06,
+      "loss": 0.9809,
+      "step": 39110
+    },
+    {
+      "epoch": 0.28317661621316426,
+      "grad_norm": 0.16163350641727448,
+      "learning_rate": 4.716830622452895e-06,
+      "loss": 0.9568,
+      "step": 39120
+    },
+    {
+      "epoch": 0.28324900287375043,
+      "grad_norm": 0.16700412333011627,
+      "learning_rate": 4.716758235792308e-06,
+      "loss": 0.9724,
+      "step": 39130
+    },
+    {
+      "epoch": 0.2833213895343366,
+      "grad_norm": 0.17913348972797394,
+      "learning_rate": 4.716685849131723e-06,
+      "loss": 0.9634,
+      "step": 39140
+    },
+    {
+      "epoch": 0.2833937761949228,
+      "grad_norm": 0.1574385166168213,
+      "learning_rate": 4.7166134624711365e-06,
+      "loss": 0.98,
+      "step": 39150
+    },
+    {
+      "epoch": 0.28346616285550896,
+      "grad_norm": 0.15847839415073395,
+      "learning_rate": 4.71654107581055e-06,
+      "loss": 0.9686,
+      "step": 39160
+    },
+    {
+      "epoch": 0.2835385495160952,
+      "grad_norm": 0.18322902917861938,
+      "learning_rate": 4.716468689149964e-06,
+      "loss": 0.9731,
+      "step": 39170
+    },
+    {
+      "epoch": 0.28361093617668137,
+      "grad_norm": 0.155546173453331,
+      "learning_rate": 4.716396302489378e-06,
+      "loss": 0.9539,
+      "step": 39180
+    },
+    {
+      "epoch": 0.28368332283726755,
+      "grad_norm": 0.17779536545276642,
+      "learning_rate": 4.716323915828792e-06,
+      "loss": 0.9655,
+      "step": 39190
+    },
+    {
+      "epoch": 0.2837557094978537,
+      "grad_norm": 0.1730605959892273,
+      "learning_rate": 4.716251529168205e-06,
+      "loss": 0.9726,
+      "step": 39200
+    },
+    {
+      "epoch": 0.2838280961584399,
+      "grad_norm": 0.19149181246757507,
+      "learning_rate": 4.716179142507619e-06,
+      "loss": 0.9723,
+      "step": 39210
+    },
+    {
+      "epoch": 0.28390048281902613,
+      "grad_norm": 0.15683695673942566,
+      "learning_rate": 4.716106755847033e-06,
+      "loss": 0.974,
+      "step": 39220
+    },
+    {
+      "epoch": 0.2839728694796123,
+      "grad_norm": 0.1871178299188614,
+      "learning_rate": 4.716034369186447e-06,
+      "loss": 0.9814,
+      "step": 39230
+    },
+    {
+      "epoch": 0.2840452561401985,
+      "grad_norm": 0.25293681025505066,
+      "learning_rate": 4.715961982525861e-06,
+      "loss": 0.9697,
+      "step": 39240
+    },
+    {
+      "epoch": 0.28411764280078466,
+      "grad_norm": 0.18877732753753662,
+      "learning_rate": 4.715889595865274e-06,
+      "loss": 0.9743,
+      "step": 39250
+    },
+    {
+      "epoch": 0.28419002946137084,
+      "grad_norm": 0.17080391943454742,
+      "learning_rate": 4.715817209204688e-06,
+      "loss": 0.9627,
+      "step": 39260
+    },
+    {
+      "epoch": 0.2842624161219571,
+      "grad_norm": 0.1600443720817566,
+      "learning_rate": 4.715744822544102e-06,
+      "loss": 0.9624,
+      "step": 39270
+    },
+    {
+      "epoch": 0.28433480278254325,
+      "grad_norm": 0.18293151259422302,
+      "learning_rate": 4.715672435883516e-06,
+      "loss": 0.9775,
+      "step": 39280
+    },
+    {
+      "epoch": 0.2844071894431294,
+      "grad_norm": 0.17442253232002258,
+      "learning_rate": 4.71560004922293e-06,
+      "loss": 0.9586,
+      "step": 39290
+    },
+    {
+      "epoch": 0.2844795761037156,
+      "grad_norm": 0.1556355059146881,
+      "learning_rate": 4.715527662562343e-06,
+      "loss": 0.9798,
+      "step": 39300
+    },
+    {
+      "epoch": 0.2845519627643018,
+      "grad_norm": 0.18220852315425873,
+      "learning_rate": 4.715455275901758e-06,
+      "loss": 0.97,
+      "step": 39310
+    },
+    {
+      "epoch": 0.28462434942488796,
+      "grad_norm": 0.1559944748878479,
+      "learning_rate": 4.715382889241171e-06,
+      "loss": 0.9807,
+      "step": 39320
+    },
+    {
+      "epoch": 0.2846967360854742,
+      "grad_norm": 0.1595684438943863,
+      "learning_rate": 4.715310502580585e-06,
+      "loss": 0.9567,
+      "step": 39330
+    },
+    {
+      "epoch": 0.28476912274606037,
+      "grad_norm": 0.18207839131355286,
+      "learning_rate": 4.7152381159199985e-06,
+      "loss": 0.9801,
+      "step": 39340
+    },
+    {
+      "epoch": 0.28484150940664654,
+      "grad_norm": 0.1691935509443283,
+      "learning_rate": 4.715165729259413e-06,
+      "loss": 0.9671,
+      "step": 39350
+    },
+    {
+      "epoch": 0.2849138960672327,
+      "grad_norm": 0.16009429097175598,
+      "learning_rate": 4.715093342598826e-06,
+      "loss": 0.9612,
+      "step": 39360
+    },
+    {
+      "epoch": 0.2849862827278189,
+      "grad_norm": 0.16721150279045105,
+      "learning_rate": 4.71502095593824e-06,
+      "loss": 0.9633,
+      "step": 39370
+    },
+    {
+      "epoch": 0.28505866938840513,
+      "grad_norm": 0.16945244371891022,
+      "learning_rate": 4.714948569277654e-06,
+      "loss": 0.9679,
+      "step": 39380
+    },
+    {
+      "epoch": 0.2851310560489913,
+      "grad_norm": 0.16579945385456085,
+      "learning_rate": 4.7148761826170675e-06,
+      "loss": 0.9525,
+      "step": 39390
+    },
+    {
+      "epoch": 0.2852034427095775,
+      "grad_norm": 0.17413263022899628,
+      "learning_rate": 4.714803795956481e-06,
+      "loss": 0.9631,
+      "step": 39400
+    },
+    {
+      "epoch": 0.28527582937016366,
+      "grad_norm": 0.16317112743854523,
+      "learning_rate": 4.714731409295895e-06,
+      "loss": 0.9699,
+      "step": 39410
+    },
+    {
+      "epoch": 0.28534821603074983,
+      "grad_norm": 0.16122910380363464,
+      "learning_rate": 4.714659022635309e-06,
+      "loss": 0.9658,
+      "step": 39420
+    },
+    {
+      "epoch": 0.28542060269133607,
+      "grad_norm": 0.17023932933807373,
+      "learning_rate": 4.714586635974723e-06,
+      "loss": 0.9697,
+      "step": 39430
+    },
+    {
+      "epoch": 0.28549298935192224,
+      "grad_norm": 0.17641928791999817,
+      "learning_rate": 4.714514249314136e-06,
+      "loss": 0.9823,
+      "step": 39440
+    },
+    {
+      "epoch": 0.2855653760125084,
+      "grad_norm": 0.16606158018112183,
+      "learning_rate": 4.71444186265355e-06,
+      "loss": 0.9721,
+      "step": 39450
+    },
+    {
+      "epoch": 0.2856377626730946,
+      "grad_norm": 0.1534787118434906,
+      "learning_rate": 4.7143694759929645e-06,
+      "loss": 0.976,
+      "step": 39460
+    },
+    {
+      "epoch": 0.2857101493336808,
+      "grad_norm": 0.15776877105236053,
+      "learning_rate": 4.714297089332378e-06,
+      "loss": 0.9708,
+      "step": 39470
+    },
+    {
+      "epoch": 0.28578253599426695,
+      "grad_norm": 0.1588585376739502,
+      "learning_rate": 4.714224702671792e-06,
+      "loss": 0.9766,
+      "step": 39480
+    },
+    {
+      "epoch": 0.2858549226548532,
+      "grad_norm": 0.16405107080936432,
+      "learning_rate": 4.714152316011205e-06,
+      "loss": 0.9766,
+      "step": 39490
+    },
+    {
+      "epoch": 0.28592730931543936,
+      "grad_norm": 0.15826919674873352,
+      "learning_rate": 4.71407992935062e-06,
+      "loss": 0.987,
+      "step": 39500
+    },
+    {
+      "epoch": 0.28599969597602554,
+      "grad_norm": 0.166813924908638,
+      "learning_rate": 4.714007542690033e-06,
+      "loss": 0.9616,
+      "step": 39510
+    },
+    {
+      "epoch": 0.2860720826366117,
+      "grad_norm": 0.15735673904418945,
+      "learning_rate": 4.713935156029447e-06,
+      "loss": 0.9719,
+      "step": 39520
+    },
+    {
+      "epoch": 0.2861444692971979,
+      "grad_norm": 0.16577894985675812,
+      "learning_rate": 4.713862769368861e-06,
+      "loss": 0.9661,
+      "step": 39530
+    },
+    {
+      "epoch": 0.2862168559577841,
+      "grad_norm": 0.1636783629655838,
+      "learning_rate": 4.713790382708275e-06,
+      "loss": 0.966,
+      "step": 39540
+    },
+    {
+      "epoch": 0.2862892426183703,
+      "grad_norm": 0.16773077845573425,
+      "learning_rate": 4.713717996047689e-06,
+      "loss": 0.9675,
+      "step": 39550
+    },
+    {
+      "epoch": 0.2863616292789565,
+      "grad_norm": 0.16528858244419098,
+      "learning_rate": 4.713645609387102e-06,
+      "loss": 0.953,
+      "step": 39560
+    },
+    {
+      "epoch": 0.28643401593954265,
+      "grad_norm": 0.178873211145401,
+      "learning_rate": 4.713573222726516e-06,
+      "loss": 0.9768,
+      "step": 39570
+    },
+    {
+      "epoch": 0.28650640260012883,
+      "grad_norm": 0.18183675408363342,
+      "learning_rate": 4.71350083606593e-06,
+      "loss": 0.9704,
+      "step": 39580
+    },
+    {
+      "epoch": 0.28657878926071506,
+      "grad_norm": 0.16120202839374542,
+      "learning_rate": 4.713428449405344e-06,
+      "loss": 0.9816,
+      "step": 39590
+    },
+    {
+      "epoch": 0.28665117592130124,
+      "grad_norm": 0.20635010302066803,
+      "learning_rate": 4.713356062744758e-06,
+      "loss": 0.9583,
+      "step": 39600
+    },
+    {
+      "epoch": 0.2867235625818874,
+      "grad_norm": 0.1598854809999466,
+      "learning_rate": 4.713283676084171e-06,
+      "loss": 0.9664,
+      "step": 39610
+    },
+    {
+      "epoch": 0.2867959492424736,
+      "grad_norm": 0.16559354960918427,
+      "learning_rate": 4.713211289423586e-06,
+      "loss": 0.9564,
+      "step": 39620
+    },
+    {
+      "epoch": 0.28686833590305977,
+      "grad_norm": 0.18220894038677216,
+      "learning_rate": 4.713138902762999e-06,
+      "loss": 0.9688,
+      "step": 39630
+    },
+    {
+      "epoch": 0.28694072256364594,
+      "grad_norm": 0.16989223659038544,
+      "learning_rate": 4.713066516102413e-06,
+      "loss": 0.9781,
+      "step": 39640
+    },
+    {
+      "epoch": 0.2870131092242322,
+      "grad_norm": 0.16962113976478577,
+      "learning_rate": 4.712994129441827e-06,
+      "loss": 0.9771,
+      "step": 39650
+    },
+    {
+      "epoch": 0.28708549588481835,
+      "grad_norm": 0.16623146831989288,
+      "learning_rate": 4.712921742781241e-06,
+      "loss": 0.9551,
+      "step": 39660
+    },
+    {
+      "epoch": 0.28715788254540453,
+      "grad_norm": 0.1545698344707489,
+      "learning_rate": 4.712849356120655e-06,
+      "loss": 0.9565,
+      "step": 39670
+    },
+    {
+      "epoch": 0.2872302692059907,
+      "grad_norm": 0.19971857964992523,
+      "learning_rate": 4.712776969460068e-06,
+      "loss": 0.9566,
+      "step": 39680
+    },
+    {
+      "epoch": 0.2873026558665769,
+      "grad_norm": 0.1566108763217926,
+      "learning_rate": 4.712704582799482e-06,
+      "loss": 0.9696,
+      "step": 39690
+    },
+    {
+      "epoch": 0.2873750425271631,
+      "grad_norm": 0.16341611742973328,
+      "learning_rate": 4.712632196138896e-06,
+      "loss": 0.9674,
+      "step": 39700
+    },
+    {
+      "epoch": 0.2874474291877493,
+      "grad_norm": 0.18505293130874634,
+      "learning_rate": 4.71255980947831e-06,
+      "loss": 0.9693,
+      "step": 39710
+    },
+    {
+      "epoch": 0.28751981584833547,
+      "grad_norm": 0.167031392455101,
+      "learning_rate": 4.712487422817724e-06,
+      "loss": 0.9645,
+      "step": 39720
+    },
+    {
+      "epoch": 0.28759220250892165,
+      "grad_norm": 0.17869937419891357,
+      "learning_rate": 4.712415036157137e-06,
+      "loss": 0.9805,
+      "step": 39730
+    },
+    {
+      "epoch": 0.2876645891695078,
+      "grad_norm": 0.16606245934963226,
+      "learning_rate": 4.712342649496552e-06,
+      "loss": 0.9612,
+      "step": 39740
+    },
+    {
+      "epoch": 0.28773697583009405,
+      "grad_norm": 0.19421431422233582,
+      "learning_rate": 4.712270262835965e-06,
+      "loss": 0.9731,
+      "step": 39750
+    },
+    {
+      "epoch": 0.28780936249068023,
+      "grad_norm": 0.17621836066246033,
+      "learning_rate": 4.712197876175379e-06,
+      "loss": 0.9697,
+      "step": 39760
+    },
+    {
+      "epoch": 0.2878817491512664,
+      "grad_norm": 0.1548440009355545,
+      "learning_rate": 4.7121254895147925e-06,
+      "loss": 0.9666,
+      "step": 39770
+    },
+    {
+      "epoch": 0.2879541358118526,
+      "grad_norm": 0.18306025862693787,
+      "learning_rate": 4.712053102854207e-06,
+      "loss": 0.9628,
+      "step": 39780
+    },
+    {
+      "epoch": 0.28802652247243876,
+      "grad_norm": 0.16512498259544373,
+      "learning_rate": 4.711980716193621e-06,
+      "loss": 0.968,
+      "step": 39790
+    },
+    {
+      "epoch": 0.28809890913302494,
+      "grad_norm": 0.17831869423389435,
+      "learning_rate": 4.711908329533034e-06,
+      "loss": 0.9739,
+      "step": 39800
+    },
+    {
+      "epoch": 0.28817129579361117,
+      "grad_norm": 0.15997859835624695,
+      "learning_rate": 4.711835942872448e-06,
+      "loss": 0.9684,
+      "step": 39810
+    },
+    {
+      "epoch": 0.28824368245419735,
+      "grad_norm": 0.1619550734758377,
+      "learning_rate": 4.711763556211862e-06,
+      "loss": 0.9792,
+      "step": 39820
+    },
+    {
+      "epoch": 0.2883160691147835,
+      "grad_norm": 0.1704218089580536,
+      "learning_rate": 4.711691169551276e-06,
+      "loss": 0.9676,
+      "step": 39830
+    },
+    {
+      "epoch": 0.2883884557753697,
+      "grad_norm": 0.17263592779636383,
+      "learning_rate": 4.7116187828906895e-06,
+      "loss": 0.9774,
+      "step": 39840
+    },
+    {
+      "epoch": 0.2884608424359559,
+      "grad_norm": 0.1720539629459381,
+      "learning_rate": 4.711546396230103e-06,
+      "loss": 0.9691,
+      "step": 39850
+    },
+    {
+      "epoch": 0.2885332290965421,
+      "grad_norm": 0.2420406937599182,
+      "learning_rate": 4.711474009569517e-06,
+      "loss": 0.9788,
+      "step": 39860
+    },
+    {
+      "epoch": 0.2886056157571283,
+      "grad_norm": 0.17428769171237946,
+      "learning_rate": 4.711401622908931e-06,
+      "loss": 0.9727,
+      "step": 39870
+    },
+    {
+      "epoch": 0.28867800241771446,
+      "grad_norm": 0.17469775676727295,
+      "learning_rate": 4.711329236248345e-06,
+      "loss": 0.9748,
+      "step": 39880
+    },
+    {
+      "epoch": 0.28875038907830064,
+      "grad_norm": 0.17111310362815857,
+      "learning_rate": 4.7112568495877585e-06,
+      "loss": 0.9626,
+      "step": 39890
+    },
+    {
+      "epoch": 0.2888227757388868,
+      "grad_norm": 0.19012178480625153,
+      "learning_rate": 4.711184462927172e-06,
+      "loss": 0.9675,
+      "step": 39900
+    },
+    {
+      "epoch": 0.28889516239947305,
+      "grad_norm": 0.16587407886981964,
+      "learning_rate": 4.711112076266586e-06,
+      "loss": 0.9665,
+      "step": 39910
+    },
+    {
+      "epoch": 0.2889675490600592,
+      "grad_norm": 0.17747098207473755,
+      "learning_rate": 4.711039689605999e-06,
+      "loss": 0.975,
+      "step": 39920
+    },
+    {
+      "epoch": 0.2890399357206454,
+      "grad_norm": 0.15955297648906708,
+      "learning_rate": 4.710967302945414e-06,
+      "loss": 0.9633,
+      "step": 39930
+    },
+    {
+      "epoch": 0.2891123223812316,
+      "grad_norm": 0.1651538610458374,
+      "learning_rate": 4.710894916284827e-06,
+      "loss": 0.9611,
+      "step": 39940
+    },
+    {
+      "epoch": 0.28918470904181776,
+      "grad_norm": 0.15846842527389526,
+      "learning_rate": 4.710822529624241e-06,
+      "loss": 0.9699,
+      "step": 39950
+    },
+    {
+      "epoch": 0.289257095702404,
+      "grad_norm": 0.17200906574726105,
+      "learning_rate": 4.710750142963655e-06,
+      "loss": 0.9759,
+      "step": 39960
+    },
+    {
+      "epoch": 0.28932948236299016,
+      "grad_norm": 0.18310053646564484,
+      "learning_rate": 4.710677756303069e-06,
+      "loss": 0.9855,
+      "step": 39970
+    },
+    {
+      "epoch": 0.28940186902357634,
+      "grad_norm": 0.17805758118629456,
+      "learning_rate": 4.710605369642483e-06,
+      "loss": 0.9754,
+      "step": 39980
+    },
+    {
+      "epoch": 0.2894742556841625,
+      "grad_norm": 0.164411723613739,
+      "learning_rate": 4.710532982981896e-06,
+      "loss": 0.9643,
+      "step": 39990
+    },
+    {
+      "epoch": 0.2895466423447487,
+      "grad_norm": 0.17421521246433258,
+      "learning_rate": 4.71046059632131e-06,
+      "loss": 0.9709,
+      "step": 40000
+    },
+    {
+      "epoch": 0.28961902900533487,
+      "grad_norm": 0.17790335416793823,
+      "learning_rate": 4.710388209660724e-06,
+      "loss": 0.9615,
+      "step": 40010
+    },
+    {
+      "epoch": 0.2896914156659211,
+      "grad_norm": 0.17614027857780457,
+      "learning_rate": 4.710315823000138e-06,
+      "loss": 0.972,
+      "step": 40020
+    },
+    {
+      "epoch": 0.2897638023265073,
+      "grad_norm": 0.17423102259635925,
+      "learning_rate": 4.710243436339552e-06,
+      "loss": 0.9799,
+      "step": 40030
+    },
+    {
+      "epoch": 0.28983618898709346,
+      "grad_norm": 0.16734719276428223,
+      "learning_rate": 4.710171049678965e-06,
+      "loss": 0.9802,
+      "step": 40040
+    },
+    {
+      "epoch": 0.28990857564767963,
+      "grad_norm": 0.18581393361091614,
+      "learning_rate": 4.710098663018379e-06,
+      "loss": 0.9651,
+      "step": 40050
+    },
+    {
+      "epoch": 0.2899809623082658,
+      "grad_norm": 0.16907867789268494,
+      "learning_rate": 4.710026276357793e-06,
+      "loss": 0.981,
+      "step": 40060
+    },
+    {
+      "epoch": 0.29005334896885204,
+      "grad_norm": 0.164125457406044,
+      "learning_rate": 4.709953889697207e-06,
+      "loss": 0.9739,
+      "step": 40070
+    },
+    {
+      "epoch": 0.2901257356294382,
+      "grad_norm": 0.1936989575624466,
+      "learning_rate": 4.7098815030366205e-06,
+      "loss": 0.9565,
+      "step": 40080
+    },
+    {
+      "epoch": 0.2901981222900244,
+      "grad_norm": 0.16564348340034485,
+      "learning_rate": 4.709809116376034e-06,
+      "loss": 0.9671,
+      "step": 40090
+    },
+    {
+      "epoch": 0.2902705089506106,
+      "grad_norm": 0.17586882412433624,
+      "learning_rate": 4.709736729715449e-06,
+      "loss": 0.9709,
+      "step": 40100
+    },
+    {
+      "epoch": 0.29034289561119675,
+      "grad_norm": 0.18018198013305664,
+      "learning_rate": 4.709664343054862e-06,
+      "loss": 0.9725,
+      "step": 40110
+    },
+    {
+      "epoch": 0.290415282271783,
+      "grad_norm": 0.1636233627796173,
+      "learning_rate": 4.709591956394276e-06,
+      "loss": 0.9762,
+      "step": 40120
+    },
+    {
+      "epoch": 0.29048766893236916,
+      "grad_norm": 0.1633533090353012,
+      "learning_rate": 4.7095195697336895e-06,
+      "loss": 0.972,
+      "step": 40130
+    },
+    {
+      "epoch": 0.29056005559295534,
+      "grad_norm": 0.18192791938781738,
+      "learning_rate": 4.709447183073104e-06,
+      "loss": 0.9612,
+      "step": 40140
+    },
+    {
+      "epoch": 0.2906324422535415,
+      "grad_norm": 0.17899948358535767,
+      "learning_rate": 4.7093747964125176e-06,
+      "loss": 0.9708,
+      "step": 40150
+    },
+    {
+      "epoch": 0.2907048289141277,
+      "grad_norm": 0.17250950634479523,
+      "learning_rate": 4.709302409751931e-06,
+      "loss": 0.9764,
+      "step": 40160
+    },
+    {
+      "epoch": 0.29077721557471387,
+      "grad_norm": 0.1687665432691574,
+      "learning_rate": 4.709230023091345e-06,
+      "loss": 0.9702,
+      "step": 40170
+    },
+    {
+      "epoch": 0.2908496022353001,
+      "grad_norm": 0.1669207513332367,
+      "learning_rate": 4.709157636430759e-06,
+      "loss": 0.9824,
+      "step": 40180
+    },
+    {
+      "epoch": 0.2909219888958863,
+      "grad_norm": 0.16087938845157623,
+      "learning_rate": 4.709085249770173e-06,
+      "loss": 0.9607,
+      "step": 40190
+    },
+    {
+      "epoch": 0.29099437555647245,
+      "grad_norm": 0.18034091591835022,
+      "learning_rate": 4.7090128631095865e-06,
+      "loss": 0.9671,
+      "step": 40200
+    },
+    {
+      "epoch": 0.2910667622170586,
+      "grad_norm": 0.1801455020904541,
+      "learning_rate": 4.708940476449e-06,
+      "loss": 0.975,
+      "step": 40210
+    },
+    {
+      "epoch": 0.2911391488776448,
+      "grad_norm": 0.16992124915122986,
+      "learning_rate": 4.7088680897884146e-06,
+      "loss": 0.9597,
+      "step": 40220
+    },
+    {
+      "epoch": 0.29121153553823104,
+      "grad_norm": 0.1644294410943985,
+      "learning_rate": 4.708795703127828e-06,
+      "loss": 0.97,
+      "step": 40230
+    },
+    {
+      "epoch": 0.2912839221988172,
+      "grad_norm": 0.16996628046035767,
+      "learning_rate": 4.708723316467242e-06,
+      "loss": 0.9661,
+      "step": 40240
+    },
+    {
+      "epoch": 0.2913563088594034,
+      "grad_norm": 0.1804681122303009,
+      "learning_rate": 4.708650929806655e-06,
+      "loss": 0.9691,
+      "step": 40250
+    },
+    {
+      "epoch": 0.29142869551998957,
+      "grad_norm": 0.1625206172466278,
+      "learning_rate": 4.70857854314607e-06,
+      "loss": 0.9725,
+      "step": 40260
+    },
+    {
+      "epoch": 0.29150108218057574,
+      "grad_norm": 0.15384940803050995,
+      "learning_rate": 4.7085061564854835e-06,
+      "loss": 0.9657,
+      "step": 40270
+    },
+    {
+      "epoch": 0.291573468841162,
+      "grad_norm": 0.18380320072174072,
+      "learning_rate": 4.708433769824897e-06,
+      "loss": 0.9646,
+      "step": 40280
+    },
+    {
+      "epoch": 0.29164585550174815,
+      "grad_norm": 0.17061907052993774,
+      "learning_rate": 4.708361383164311e-06,
+      "loss": 0.9681,
+      "step": 40290
+    },
+    {
+      "epoch": 0.29171824216233433,
+      "grad_norm": 0.18094012141227722,
+      "learning_rate": 4.708288996503725e-06,
+      "loss": 0.9638,
+      "step": 40300
+    },
+    {
+      "epoch": 0.2917906288229205,
+      "grad_norm": 0.1657949686050415,
+      "learning_rate": 4.708216609843139e-06,
+      "loss": 0.9689,
+      "step": 40310
+    },
+    {
+      "epoch": 0.2918630154835067,
+      "grad_norm": 0.28880590200424194,
+      "learning_rate": 4.708144223182552e-06,
+      "loss": 0.9702,
+      "step": 40320
+    },
+    {
+      "epoch": 0.29193540214409286,
+      "grad_norm": 0.16735604405403137,
+      "learning_rate": 4.708071836521966e-06,
+      "loss": 0.9845,
+      "step": 40330
+    },
+    {
+      "epoch": 0.2920077888046791,
+      "grad_norm": 0.17504291236400604,
+      "learning_rate": 4.7079994498613805e-06,
+      "loss": 0.9572,
+      "step": 40340
+    },
+    {
+      "epoch": 0.29208017546526527,
+      "grad_norm": 0.15563970804214478,
+      "learning_rate": 4.707927063200794e-06,
+      "loss": 0.9574,
+      "step": 40350
+    },
+    {
+      "epoch": 0.29215256212585144,
+      "grad_norm": 0.15227772295475006,
+      "learning_rate": 4.707854676540208e-06,
+      "loss": 0.9612,
+      "step": 40360
+    },
+    {
+      "epoch": 0.2922249487864376,
+      "grad_norm": 0.16515760123729706,
+      "learning_rate": 4.707782289879621e-06,
+      "loss": 0.9519,
+      "step": 40370
+    },
+    {
+      "epoch": 0.2922973354470238,
+      "grad_norm": 0.17312021553516388,
+      "learning_rate": 4.707709903219036e-06,
+      "loss": 0.9707,
+      "step": 40380
+    },
+    {
+      "epoch": 0.29236972210761003,
+      "grad_norm": 0.16830036044120789,
+      "learning_rate": 4.7076375165584494e-06,
+      "loss": 0.9557,
+      "step": 40390
+    },
+    {
+      "epoch": 0.2924421087681962,
+      "grad_norm": 0.15725110471248627,
+      "learning_rate": 4.707565129897863e-06,
+      "loss": 0.9652,
+      "step": 40400
+    },
+    {
+      "epoch": 0.2925144954287824,
+      "grad_norm": 0.16694726049900055,
+      "learning_rate": 4.707492743237277e-06,
+      "loss": 0.9682,
+      "step": 40410
+    },
+    {
+      "epoch": 0.29258688208936856,
+      "grad_norm": 0.16128145158290863,
+      "learning_rate": 4.70742035657669e-06,
+      "loss": 0.9658,
+      "step": 40420
+    },
+    {
+      "epoch": 0.29265926874995474,
+      "grad_norm": 0.16756875813007355,
+      "learning_rate": 4.707347969916104e-06,
+      "loss": 0.9768,
+      "step": 40430
+    },
+    {
+      "epoch": 0.29273165541054097,
+      "grad_norm": 0.1622963845729828,
+      "learning_rate": 4.7072755832555175e-06,
+      "loss": 0.9754,
+      "step": 40440
+    },
+    {
+      "epoch": 0.29280404207112715,
+      "grad_norm": 0.15665726363658905,
+      "learning_rate": 4.707203196594932e-06,
+      "loss": 0.9553,
+      "step": 40450
+    },
+    {
+      "epoch": 0.2928764287317133,
+      "grad_norm": 0.17446359992027283,
+      "learning_rate": 4.707130809934346e-06,
+      "loss": 0.9724,
+      "step": 40460
+    },
+    {
+      "epoch": 0.2929488153922995,
+      "grad_norm": 0.16420406103134155,
+      "learning_rate": 4.707058423273759e-06,
+      "loss": 0.972,
+      "step": 40470
+    },
+    {
+      "epoch": 0.2930212020528857,
+      "grad_norm": 0.1752181053161621,
+      "learning_rate": 4.706986036613173e-06,
+      "loss": 0.9645,
+      "step": 40480
+    },
+    {
+      "epoch": 0.2930935887134719,
+      "grad_norm": 0.16292783617973328,
+      "learning_rate": 4.706913649952587e-06,
+      "loss": 0.9608,
+      "step": 40490
+    },
+    {
+      "epoch": 0.2931659753740581,
+      "grad_norm": 0.15894421935081482,
+      "learning_rate": 4.706841263292001e-06,
+      "loss": 0.955,
+      "step": 40500
+    },
+    {
+      "epoch": 0.29323836203464426,
+      "grad_norm": 0.1527421772480011,
+      "learning_rate": 4.7067688766314145e-06,
+      "loss": 0.9607,
+      "step": 40510
+    },
+    {
+      "epoch": 0.29331074869523044,
+      "grad_norm": 0.1576659232378006,
+      "learning_rate": 4.706696489970828e-06,
+      "loss": 0.9614,
+      "step": 40520
+    },
+    {
+      "epoch": 0.2933831353558166,
+      "grad_norm": 0.1652728021144867,
+      "learning_rate": 4.706624103310243e-06,
+      "loss": 0.9544,
+      "step": 40530
+    },
+    {
+      "epoch": 0.2934555220164028,
+      "grad_norm": 0.16338014602661133,
+      "learning_rate": 4.706551716649656e-06,
+      "loss": 0.9539,
+      "step": 40540
+    },
+    {
+      "epoch": 0.293527908676989,
+      "grad_norm": 0.16020265221595764,
+      "learning_rate": 4.70647932998907e-06,
+      "loss": 0.9732,
+      "step": 40550
+    },
+    {
+      "epoch": 0.2936002953375752,
+      "grad_norm": 0.17139659821987152,
+      "learning_rate": 4.7064069433284834e-06,
+      "loss": 0.9645,
+      "step": 40560
+    },
+    {
+      "epoch": 0.2936726819981614,
+      "grad_norm": 0.1736135184764862,
+      "learning_rate": 4.706334556667898e-06,
+      "loss": 0.9746,
+      "step": 40570
+    },
+    {
+      "epoch": 0.29374506865874755,
+      "grad_norm": 0.16610664129257202,
+      "learning_rate": 4.7062621700073115e-06,
+      "loss": 0.9673,
+      "step": 40580
+    },
+    {
+      "epoch": 0.29381745531933373,
+      "grad_norm": 0.195650115609169,
+      "learning_rate": 4.706189783346725e-06,
+      "loss": 0.9613,
+      "step": 40590
+    },
+    {
+      "epoch": 0.29388984197991996,
+      "grad_norm": 0.1645543873310089,
+      "learning_rate": 4.706117396686139e-06,
+      "loss": 0.9577,
+      "step": 40600
+    },
+    {
+      "epoch": 0.29396222864050614,
+      "grad_norm": 0.17229747772216797,
+      "learning_rate": 4.706045010025553e-06,
+      "loss": 0.9701,
+      "step": 40610
+    },
+    {
+      "epoch": 0.2940346153010923,
+      "grad_norm": 0.18451061844825745,
+      "learning_rate": 4.705972623364967e-06,
+      "loss": 0.9746,
+      "step": 40620
+    },
+    {
+      "epoch": 0.2941070019616785,
+      "grad_norm": 0.16431106626987457,
+      "learning_rate": 4.7059002367043805e-06,
+      "loss": 0.9575,
+      "step": 40630
+    },
+    {
+      "epoch": 0.29417938862226467,
+      "grad_norm": 0.15613338351249695,
+      "learning_rate": 4.705827850043794e-06,
+      "loss": 0.9684,
+      "step": 40640
+    },
+    {
+      "epoch": 0.2942517752828509,
+      "grad_norm": 0.1578802615404129,
+      "learning_rate": 4.705755463383208e-06,
+      "loss": 0.9641,
+      "step": 40650
+    },
+    {
+      "epoch": 0.2943241619434371,
+      "grad_norm": 0.1636672019958496,
+      "learning_rate": 4.705683076722622e-06,
+      "loss": 0.9686,
+      "step": 40660
+    },
+    {
+      "epoch": 0.29439654860402326,
+      "grad_norm": 0.2066519856452942,
+      "learning_rate": 4.705610690062036e-06,
+      "loss": 0.9649,
+      "step": 40670
+    },
+    {
+      "epoch": 0.29446893526460943,
+      "grad_norm": 0.16676735877990723,
+      "learning_rate": 4.705538303401449e-06,
+      "loss": 0.9666,
+      "step": 40680
+    },
+    {
+      "epoch": 0.2945413219251956,
+      "grad_norm": 0.17513099312782288,
+      "learning_rate": 4.705465916740863e-06,
+      "loss": 0.9768,
+      "step": 40690
+    },
+    {
+      "epoch": 0.2946137085857818,
+      "grad_norm": 0.17959053814411163,
+      "learning_rate": 4.7053935300802775e-06,
+      "loss": 0.9748,
+      "step": 40700
+    },
+    {
+      "epoch": 0.294686095246368,
+      "grad_norm": 0.16386069357395172,
+      "learning_rate": 4.705321143419691e-06,
+      "loss": 0.9679,
+      "step": 40710
+    },
+    {
+      "epoch": 0.2947584819069542,
+      "grad_norm": 0.16159318387508392,
+      "learning_rate": 4.705248756759105e-06,
+      "loss": 0.9673,
+      "step": 40720
+    },
+    {
+      "epoch": 0.29483086856754037,
+      "grad_norm": 0.16015946865081787,
+      "learning_rate": 4.705176370098518e-06,
+      "loss": 0.9542,
+      "step": 40730
+    },
+    {
+      "epoch": 0.29490325522812655,
+      "grad_norm": 0.16827359795570374,
+      "learning_rate": 4.705103983437933e-06,
+      "loss": 0.973,
+      "step": 40740
+    },
+    {
+      "epoch": 0.2949756418887127,
+      "grad_norm": 0.16736365854740143,
+      "learning_rate": 4.705031596777346e-06,
+      "loss": 0.9778,
+      "step": 40750
+    },
+    {
+      "epoch": 0.29504802854929896,
+      "grad_norm": 0.17061574757099152,
+      "learning_rate": 4.70495921011676e-06,
+      "loss": 0.9558,
+      "step": 40760
+    },
+    {
+      "epoch": 0.29512041520988513,
+      "grad_norm": 0.16562294960021973,
+      "learning_rate": 4.704886823456174e-06,
+      "loss": 0.9805,
+      "step": 40770
+    },
+    {
+      "epoch": 0.2951928018704713,
+      "grad_norm": 0.17279978096485138,
+      "learning_rate": 4.704814436795588e-06,
+      "loss": 0.9656,
+      "step": 40780
+    },
+    {
+      "epoch": 0.2952651885310575,
+      "grad_norm": 0.20112721621990204,
+      "learning_rate": 4.704742050135002e-06,
+      "loss": 0.96,
+      "step": 40790
+    },
+    {
+      "epoch": 0.29533757519164366,
+      "grad_norm": 0.1690005213022232,
+      "learning_rate": 4.704669663474415e-06,
+      "loss": 0.961,
+      "step": 40800
+    },
+    {
+      "epoch": 0.2954099618522299,
+      "grad_norm": 0.16406698524951935,
+      "learning_rate": 4.704597276813829e-06,
+      "loss": 0.9705,
+      "step": 40810
+    },
+    {
+      "epoch": 0.2954823485128161,
+      "grad_norm": 0.16599249839782715,
+      "learning_rate": 4.704524890153243e-06,
+      "loss": 0.9577,
+      "step": 40820
+    },
+    {
+      "epoch": 0.29555473517340225,
+      "grad_norm": 0.19940651953220367,
+      "learning_rate": 4.704452503492657e-06,
+      "loss": 0.9678,
+      "step": 40830
+    },
+    {
+      "epoch": 0.2956271218339884,
+      "grad_norm": 0.1629924476146698,
+      "learning_rate": 4.704380116832071e-06,
+      "loss": 0.9655,
+      "step": 40840
+    },
+    {
+      "epoch": 0.2956995084945746,
+      "grad_norm": 0.17306756973266602,
+      "learning_rate": 4.704307730171484e-06,
+      "loss": 0.974,
+      "step": 40850
+    },
+    {
+      "epoch": 0.2957718951551608,
+      "grad_norm": 0.15905143320560455,
+      "learning_rate": 4.704235343510899e-06,
+      "loss": 0.9605,
+      "step": 40860
+    },
+    {
+      "epoch": 0.295844281815747,
+      "grad_norm": 0.172456756234169,
+      "learning_rate": 4.704162956850312e-06,
+      "loss": 0.9663,
+      "step": 40870
+    },
+    {
+      "epoch": 0.2959166684763332,
+      "grad_norm": 0.16704027354717255,
+      "learning_rate": 4.704090570189726e-06,
+      "loss": 0.9636,
+      "step": 40880
+    },
+    {
+      "epoch": 0.29598905513691937,
+      "grad_norm": 0.17687085270881653,
+      "learning_rate": 4.7040181835291396e-06,
+      "loss": 0.9696,
+      "step": 40890
+    },
+    {
+      "epoch": 0.29606144179750554,
+      "grad_norm": 0.16065765917301178,
+      "learning_rate": 4.703945796868554e-06,
+      "loss": 0.9753,
+      "step": 40900
+    },
+    {
+      "epoch": 0.2961338284580917,
+      "grad_norm": 0.2607700526714325,
+      "learning_rate": 4.703873410207968e-06,
+      "loss": 0.9674,
+      "step": 40910
+    },
+    {
+      "epoch": 0.29620621511867795,
+      "grad_norm": 0.1637168526649475,
+      "learning_rate": 4.703801023547381e-06,
+      "loss": 0.9579,
+      "step": 40920
+    },
+    {
+      "epoch": 0.29627860177926413,
+      "grad_norm": 0.16693775355815887,
+      "learning_rate": 4.703728636886795e-06,
+      "loss": 0.9653,
+      "step": 40930
+    },
+    {
+      "epoch": 0.2963509884398503,
+      "grad_norm": 0.16066871583461761,
+      "learning_rate": 4.703656250226209e-06,
+      "loss": 0.9694,
+      "step": 40940
+    },
+    {
+      "epoch": 0.2964233751004365,
+      "grad_norm": 0.16873207688331604,
+      "learning_rate": 4.703583863565622e-06,
+      "loss": 0.9804,
+      "step": 40950
+    },
+    {
+      "epoch": 0.29649576176102266,
+      "grad_norm": 0.17132726311683655,
+      "learning_rate": 4.703511476905036e-06,
+      "loss": 0.9662,
+      "step": 40960
+    },
+    {
+      "epoch": 0.2965681484216089,
+      "grad_norm": 0.16967126727104187,
+      "learning_rate": 4.70343909024445e-06,
+      "loss": 0.9615,
+      "step": 40970
+    },
+    {
+      "epoch": 0.29664053508219507,
+      "grad_norm": 0.16376537084579468,
+      "learning_rate": 4.703366703583864e-06,
+      "loss": 0.9654,
+      "step": 40980
+    },
+    {
+      "epoch": 0.29671292174278124,
+      "grad_norm": 0.18798090517520905,
+      "learning_rate": 4.703294316923277e-06,
+      "loss": 0.9677,
+      "step": 40990
+    },
+    {
+      "epoch": 0.2967853084033674,
+      "grad_norm": 0.1704019010066986,
+      "learning_rate": 4.703221930262691e-06,
+      "loss": 0.9697,
+      "step": 41000
+    },
+    {
+      "epoch": 0.2968576950639536,
+      "grad_norm": 0.15598341822624207,
+      "learning_rate": 4.7031495436021055e-06,
+      "loss": 0.9642,
+      "step": 41010
+    },
+    {
+      "epoch": 0.29693008172453983,
+      "grad_norm": 0.17868080735206604,
+      "learning_rate": 4.703077156941519e-06,
+      "loss": 0.9692,
+      "step": 41020
+    },
+    {
+      "epoch": 0.297002468385126,
+      "grad_norm": 0.16686975955963135,
+      "learning_rate": 4.703004770280933e-06,
+      "loss": 0.9695,
+      "step": 41030
+    },
+    {
+      "epoch": 0.2970748550457122,
+      "grad_norm": 0.16504326462745667,
+      "learning_rate": 4.702932383620346e-06,
+      "loss": 0.9697,
+      "step": 41040
+    },
+    {
+      "epoch": 0.29714724170629836,
+      "grad_norm": 0.17151391506195068,
+      "learning_rate": 4.702859996959761e-06,
+      "loss": 0.9686,
+      "step": 41050
+    },
+    {
+      "epoch": 0.29721962836688454,
+      "grad_norm": 0.16817405819892883,
+      "learning_rate": 4.702787610299174e-06,
+      "loss": 0.9692,
+      "step": 41060
+    },
+    {
+      "epoch": 0.2972920150274707,
+      "grad_norm": 0.1832583099603653,
+      "learning_rate": 4.702715223638588e-06,
+      "loss": 0.9846,
+      "step": 41070
+    },
+    {
+      "epoch": 0.29736440168805695,
+      "grad_norm": 0.171138733625412,
+      "learning_rate": 4.702642836978002e-06,
+      "loss": 0.9749,
+      "step": 41080
+    },
+    {
+      "epoch": 0.2974367883486431,
+      "grad_norm": 0.15057332813739777,
+      "learning_rate": 4.702570450317416e-06,
+      "loss": 0.9668,
+      "step": 41090
+    },
+    {
+      "epoch": 0.2975091750092293,
+      "grad_norm": 0.16757220029830933,
+      "learning_rate": 4.70249806365683e-06,
+      "loss": 0.9772,
+      "step": 41100
+    },
+    {
+      "epoch": 0.2975815616698155,
+      "grad_norm": 0.20239600539207458,
+      "learning_rate": 4.702425676996243e-06,
+      "loss": 0.9588,
+      "step": 41110
+    },
+    {
+      "epoch": 0.29765394833040165,
+      "grad_norm": 0.17006611824035645,
+      "learning_rate": 4.702353290335657e-06,
+      "loss": 0.9626,
+      "step": 41120
+    },
+    {
+      "epoch": 0.2977263349909879,
+      "grad_norm": 0.17428216338157654,
+      "learning_rate": 4.7022809036750714e-06,
+      "loss": 0.963,
+      "step": 41130
+    },
+    {
+      "epoch": 0.29779872165157406,
+      "grad_norm": 0.16272498667240143,
+      "learning_rate": 4.702208517014485e-06,
+      "loss": 0.9712,
+      "step": 41140
+    },
+    {
+      "epoch": 0.29787110831216024,
+      "grad_norm": 0.15650291740894318,
+      "learning_rate": 4.702136130353899e-06,
+      "loss": 0.9564,
+      "step": 41150
+    },
+    {
+      "epoch": 0.2979434949727464,
+      "grad_norm": 0.17676356434822083,
+      "learning_rate": 4.702063743693312e-06,
+      "loss": 0.9722,
+      "step": 41160
+    },
+    {
+      "epoch": 0.2980158816333326,
+      "grad_norm": 0.17107968032360077,
+      "learning_rate": 4.701991357032727e-06,
+      "loss": 0.9638,
+      "step": 41170
+    },
+    {
+      "epoch": 0.2980882682939188,
+      "grad_norm": 0.16407465934753418,
+      "learning_rate": 4.70191897037214e-06,
+      "loss": 0.9646,
+      "step": 41180
+    },
+    {
+      "epoch": 0.298160654954505,
+      "grad_norm": 0.1607753485441208,
+      "learning_rate": 4.701846583711554e-06,
+      "loss": 0.9712,
+      "step": 41190
+    },
+    {
+      "epoch": 0.2982330416150912,
+      "grad_norm": 0.1688770204782486,
+      "learning_rate": 4.701774197050968e-06,
+      "loss": 0.9638,
+      "step": 41200
+    },
+    {
+      "epoch": 0.29830542827567735,
+      "grad_norm": 0.16739125549793243,
+      "learning_rate": 4.701701810390382e-06,
+      "loss": 0.9555,
+      "step": 41210
+    },
+    {
+      "epoch": 0.29837781493626353,
+      "grad_norm": 0.17620337009429932,
+      "learning_rate": 4.701629423729796e-06,
+      "loss": 0.9672,
+      "step": 41220
+    },
+    {
+      "epoch": 0.2984502015968497,
+      "grad_norm": 0.16256146132946014,
+      "learning_rate": 4.701557037069209e-06,
+      "loss": 0.9719,
+      "step": 41230
+    },
+    {
+      "epoch": 0.29852258825743594,
+      "grad_norm": 0.17721201479434967,
+      "learning_rate": 4.701484650408623e-06,
+      "loss": 0.9726,
+      "step": 41240
+    },
+    {
+      "epoch": 0.2985949749180221,
+      "grad_norm": 0.1829133778810501,
+      "learning_rate": 4.701412263748037e-06,
+      "loss": 0.9632,
+      "step": 41250
+    },
+    {
+      "epoch": 0.2986673615786083,
+      "grad_norm": 0.15692444145679474,
+      "learning_rate": 4.701339877087451e-06,
+      "loss": 0.9641,
+      "step": 41260
+    },
+    {
+      "epoch": 0.29873974823919447,
+      "grad_norm": 0.1705268919467926,
+      "learning_rate": 4.701267490426865e-06,
+      "loss": 0.9684,
+      "step": 41270
+    },
+    {
+      "epoch": 0.29881213489978065,
+      "grad_norm": 0.16059453785419464,
+      "learning_rate": 4.701195103766278e-06,
+      "loss": 0.9658,
+      "step": 41280
+    },
+    {
+      "epoch": 0.2988845215603669,
+      "grad_norm": 0.16799260675907135,
+      "learning_rate": 4.701122717105692e-06,
+      "loss": 0.9679,
+      "step": 41290
+    },
+    {
+      "epoch": 0.29895690822095305,
+      "grad_norm": 0.15362030267715454,
+      "learning_rate": 4.701050330445106e-06,
+      "loss": 0.9547,
+      "step": 41300
+    },
+    {
+      "epoch": 0.29902929488153923,
+      "grad_norm": 0.15962745249271393,
+      "learning_rate": 4.70097794378452e-06,
+      "loss": 0.9745,
+      "step": 41310
+    },
+    {
+      "epoch": 0.2991016815421254,
+      "grad_norm": 0.17490293085575104,
+      "learning_rate": 4.7009055571239335e-06,
+      "loss": 0.965,
+      "step": 41320
+    },
+    {
+      "epoch": 0.2991740682027116,
+      "grad_norm": 0.1677936166524887,
+      "learning_rate": 4.700833170463347e-06,
+      "loss": 0.9546,
+      "step": 41330
+    },
+    {
+      "epoch": 0.2992464548632978,
+      "grad_norm": 0.16804441809654236,
+      "learning_rate": 4.700760783802762e-06,
+      "loss": 0.9712,
+      "step": 41340
+    },
+    {
+      "epoch": 0.299318841523884,
+      "grad_norm": 0.21531559526920319,
+      "learning_rate": 4.700688397142175e-06,
+      "loss": 0.9689,
+      "step": 41350
+    },
+    {
+      "epoch": 0.29939122818447017,
+      "grad_norm": 0.18383803963661194,
+      "learning_rate": 4.700616010481589e-06,
+      "loss": 0.9736,
+      "step": 41360
+    },
+    {
+      "epoch": 0.29946361484505635,
+      "grad_norm": 0.17313213646411896,
+      "learning_rate": 4.7005436238210024e-06,
+      "loss": 0.9589,
+      "step": 41370
+    },
+    {
+      "epoch": 0.2995360015056425,
+      "grad_norm": 0.18341746926307678,
+      "learning_rate": 4.700471237160417e-06,
+      "loss": 0.9629,
+      "step": 41380
+    },
+    {
+      "epoch": 0.2996083881662287,
+      "grad_norm": 0.1760926991701126,
+      "learning_rate": 4.7003988504998305e-06,
+      "loss": 0.9681,
+      "step": 41390
+    },
+    {
+      "epoch": 0.29968077482681493,
+      "grad_norm": 0.18593202531337738,
+      "learning_rate": 4.700326463839244e-06,
+      "loss": 0.9667,
+      "step": 41400
+    },
+    {
+      "epoch": 0.2997531614874011,
+      "grad_norm": 0.16508474946022034,
+      "learning_rate": 4.700254077178658e-06,
+      "loss": 0.9772,
+      "step": 41410
+    },
+    {
+      "epoch": 0.2998255481479873,
+      "grad_norm": 0.16628380119800568,
+      "learning_rate": 4.700181690518072e-06,
+      "loss": 0.9648,
+      "step": 41420
+    },
+    {
+      "epoch": 0.29989793480857346,
+      "grad_norm": 0.19085149466991425,
+      "learning_rate": 4.700109303857486e-06,
+      "loss": 0.9726,
+      "step": 41430
+    },
+    {
+      "epoch": 0.29997032146915964,
+      "grad_norm": 0.16664601862430573,
+      "learning_rate": 4.7000369171968995e-06,
+      "loss": 0.9628,
+      "step": 41440
+    },
+    {
+      "epoch": 0.30004270812974587,
+      "grad_norm": 0.17423871159553528,
+      "learning_rate": 4.699964530536313e-06,
+      "loss": 0.966,
+      "step": 41450
+    },
+    {
+      "epoch": 0.30011509479033205,
+      "grad_norm": 0.18593478202819824,
+      "learning_rate": 4.6998921438757275e-06,
+      "loss": 0.9664,
+      "step": 41460
+    },
+    {
+      "epoch": 0.3001874814509182,
+      "grad_norm": 0.19175831973552704,
+      "learning_rate": 4.699819757215141e-06,
+      "loss": 0.9789,
+      "step": 41470
+    },
+    {
+      "epoch": 0.3002598681115044,
+      "grad_norm": 0.15855441987514496,
+      "learning_rate": 4.699747370554554e-06,
+      "loss": 0.9516,
+      "step": 41480
+    },
+    {
+      "epoch": 0.3003322547720906,
+      "grad_norm": 0.2066076695919037,
+      "learning_rate": 4.699674983893968e-06,
+      "loss": 0.9594,
+      "step": 41490
+    },
+    {
+      "epoch": 0.3004046414326768,
+      "grad_norm": 0.9006432890892029,
+      "learning_rate": 4.699602597233382e-06,
+      "loss": 0.9761,
+      "step": 41500
+    },
+    {
+      "epoch": 0.300477028093263,
+      "grad_norm": 0.17038564383983612,
+      "learning_rate": 4.699530210572796e-06,
+      "loss": 0.9686,
+      "step": 41510
+    },
+    {
+      "epoch": 0.30054941475384916,
+      "grad_norm": 0.16303139925003052,
+      "learning_rate": 4.699457823912209e-06,
+      "loss": 0.9774,
+      "step": 41520
+    },
+    {
+      "epoch": 0.30062180141443534,
+      "grad_norm": 0.16983075439929962,
+      "learning_rate": 4.699385437251624e-06,
+      "loss": 0.9731,
+      "step": 41530
+    },
+    {
+      "epoch": 0.3006941880750215,
+      "grad_norm": 0.17250801622867584,
+      "learning_rate": 4.699313050591037e-06,
+      "loss": 0.9432,
+      "step": 41540
+    },
+    {
+      "epoch": 0.3007665747356077,
+      "grad_norm": 0.16830387711524963,
+      "learning_rate": 4.699240663930451e-06,
+      "loss": 0.964,
+      "step": 41550
+    },
+    {
+      "epoch": 0.3008389613961939,
+      "grad_norm": 0.16739524900913239,
+      "learning_rate": 4.6991682772698645e-06,
+      "loss": 0.9785,
+      "step": 41560
+    },
+    {
+      "epoch": 0.3009113480567801,
+      "grad_norm": 0.17023219168186188,
+      "learning_rate": 4.699095890609279e-06,
+      "loss": 0.9729,
+      "step": 41570
+    },
+    {
+      "epoch": 0.3009837347173663,
+      "grad_norm": 0.17168667912483215,
+      "learning_rate": 4.699023503948693e-06,
+      "loss": 0.961,
+      "step": 41580
+    },
+    {
+      "epoch": 0.30105612137795246,
+      "grad_norm": 0.16515317559242249,
+      "learning_rate": 4.698951117288106e-06,
+      "loss": 0.9731,
+      "step": 41590
+    },
+    {
+      "epoch": 0.30112850803853863,
+      "grad_norm": 0.16387787461280823,
+      "learning_rate": 4.69887873062752e-06,
+      "loss": 0.9587,
+      "step": 41600
+    },
+    {
+      "epoch": 0.30120089469912487,
+      "grad_norm": 0.18492285907268524,
+      "learning_rate": 4.698806343966934e-06,
+      "loss": 0.9635,
+      "step": 41610
+    },
+    {
+      "epoch": 0.30127328135971104,
+      "grad_norm": 0.15610884130001068,
+      "learning_rate": 4.698733957306348e-06,
+      "loss": 0.9579,
+      "step": 41620
+    },
+    {
+      "epoch": 0.3013456680202972,
+      "grad_norm": 0.1656864881515503,
+      "learning_rate": 4.6986615706457616e-06,
+      "loss": 0.9637,
+      "step": 41630
+    },
+    {
+      "epoch": 0.3014180546808834,
+      "grad_norm": 0.16656151413917542,
+      "learning_rate": 4.698589183985175e-06,
+      "loss": 0.9704,
+      "step": 41640
+    },
+    {
+      "epoch": 0.3014904413414696,
+      "grad_norm": 0.163262277841568,
+      "learning_rate": 4.69851679732459e-06,
+      "loss": 0.954,
+      "step": 41650
+    },
+    {
+      "epoch": 0.3015628280020558,
+      "grad_norm": 0.17654089629650116,
+      "learning_rate": 4.698444410664003e-06,
+      "loss": 0.981,
+      "step": 41660
+    },
+    {
+      "epoch": 0.301635214662642,
+      "grad_norm": 0.16009564697742462,
+      "learning_rate": 4.698372024003417e-06,
+      "loss": 0.964,
+      "step": 41670
+    },
+    {
+      "epoch": 0.30170760132322816,
+      "grad_norm": 0.17271243035793304,
+      "learning_rate": 4.6982996373428305e-06,
+      "loss": 0.9637,
+      "step": 41680
+    },
+    {
+      "epoch": 0.30177998798381434,
+      "grad_norm": 0.16152308881282806,
+      "learning_rate": 4.698227250682245e-06,
+      "loss": 0.9601,
+      "step": 41690
+    },
+    {
+      "epoch": 0.3018523746444005,
+      "grad_norm": 0.16311658918857574,
+      "learning_rate": 4.6981548640216586e-06,
+      "loss": 0.9648,
+      "step": 41700
+    },
+    {
+      "epoch": 0.30192476130498674,
+      "grad_norm": 0.17179886996746063,
+      "learning_rate": 4.698082477361072e-06,
+      "loss": 0.9636,
+      "step": 41710
+    },
+    {
+      "epoch": 0.3019971479655729,
+      "grad_norm": 0.16331078112125397,
+      "learning_rate": 4.698010090700486e-06,
+      "loss": 0.9694,
+      "step": 41720
+    },
+    {
+      "epoch": 0.3020695346261591,
+      "grad_norm": 0.16688616573810577,
+      "learning_rate": 4.6979377040399e-06,
+      "loss": 0.9646,
+      "step": 41730
+    },
+    {
+      "epoch": 0.3021419212867453,
+      "grad_norm": 0.19026783108711243,
+      "learning_rate": 4.697865317379314e-06,
+      "loss": 0.9706,
+      "step": 41740
+    },
+    {
+      "epoch": 0.30221430794733145,
+      "grad_norm": 0.169419065117836,
+      "learning_rate": 4.6977929307187275e-06,
+      "loss": 0.9559,
+      "step": 41750
+    },
+    {
+      "epoch": 0.3022866946079176,
+      "grad_norm": 0.1800675243139267,
+      "learning_rate": 4.697720544058141e-06,
+      "loss": 0.9822,
+      "step": 41760
+    },
+    {
+      "epoch": 0.30235908126850386,
+      "grad_norm": 0.15072910487651825,
+      "learning_rate": 4.6976481573975556e-06,
+      "loss": 0.962,
+      "step": 41770
+    },
+    {
+      "epoch": 0.30243146792909004,
+      "grad_norm": 0.15950879454612732,
+      "learning_rate": 4.697575770736969e-06,
+      "loss": 0.9697,
+      "step": 41780
+    },
+    {
+      "epoch": 0.3025038545896762,
+      "grad_norm": 0.1841602772474289,
+      "learning_rate": 4.697503384076383e-06,
+      "loss": 0.9731,
+      "step": 41790
+    },
+    {
+      "epoch": 0.3025762412502624,
+      "grad_norm": 0.22333543002605438,
+      "learning_rate": 4.697430997415796e-06,
+      "loss": 0.9594,
+      "step": 41800
+    },
+    {
+      "epoch": 0.30264862791084857,
+      "grad_norm": 0.17645373940467834,
+      "learning_rate": 4.697358610755211e-06,
+      "loss": 0.9595,
+      "step": 41810
+    },
+    {
+      "epoch": 0.3027210145714348,
+      "grad_norm": 0.16420337557792664,
+      "learning_rate": 4.6972862240946245e-06,
+      "loss": 0.9829,
+      "step": 41820
+    },
+    {
+      "epoch": 0.302793401232021,
+      "grad_norm": 0.1616903692483902,
+      "learning_rate": 4.697213837434038e-06,
+      "loss": 0.9627,
+      "step": 41830
+    },
+    {
+      "epoch": 0.30286578789260715,
+      "grad_norm": 0.1770700216293335,
+      "learning_rate": 4.697141450773452e-06,
+      "loss": 0.9546,
+      "step": 41840
+    },
+    {
+      "epoch": 0.30293817455319333,
+      "grad_norm": 0.16673099994659424,
+      "learning_rate": 4.697069064112866e-06,
+      "loss": 0.972,
+      "step": 41850
+    },
+    {
+      "epoch": 0.3030105612137795,
+      "grad_norm": 0.15225054323673248,
+      "learning_rate": 4.69699667745228e-06,
+      "loss": 0.9707,
+      "step": 41860
+    },
+    {
+      "epoch": 0.30308294787436574,
+      "grad_norm": 0.1643424779176712,
+      "learning_rate": 4.6969242907916934e-06,
+      "loss": 0.9601,
+      "step": 41870
+    },
+    {
+      "epoch": 0.3031553345349519,
+      "grad_norm": 0.20740066468715668,
+      "learning_rate": 4.696851904131107e-06,
+      "loss": 0.962,
+      "step": 41880
+    },
+    {
+      "epoch": 0.3032277211955381,
+      "grad_norm": 0.17226648330688477,
+      "learning_rate": 4.696779517470521e-06,
+      "loss": 0.9739,
+      "step": 41890
+    },
+    {
+      "epoch": 0.30330010785612427,
+      "grad_norm": 0.17776334285736084,
+      "learning_rate": 4.696707130809935e-06,
+      "loss": 0.979,
+      "step": 41900
+    },
+    {
+      "epoch": 0.30337249451671044,
+      "grad_norm": 0.15923330187797546,
+      "learning_rate": 4.696634744149349e-06,
+      "loss": 0.9613,
+      "step": 41910
+    },
+    {
+      "epoch": 0.3034448811772966,
+      "grad_norm": 0.1711985170841217,
+      "learning_rate": 4.696562357488762e-06,
+      "loss": 0.9588,
+      "step": 41920
+    },
+    {
+      "epoch": 0.30351726783788285,
+      "grad_norm": 0.1879815310239792,
+      "learning_rate": 4.696489970828176e-06,
+      "loss": 0.9679,
+      "step": 41930
+    },
+    {
+      "epoch": 0.30358965449846903,
+      "grad_norm": 0.17883126437664032,
+      "learning_rate": 4.6964175841675904e-06,
+      "loss": 0.9755,
+      "step": 41940
+    },
+    {
+      "epoch": 0.3036620411590552,
+      "grad_norm": 0.16554105281829834,
+      "learning_rate": 4.696345197507004e-06,
+      "loss": 0.9611,
+      "step": 41950
+    },
+    {
+      "epoch": 0.3037344278196414,
+      "grad_norm": 0.1649336963891983,
+      "learning_rate": 4.696272810846418e-06,
+      "loss": 0.9611,
+      "step": 41960
+    },
+    {
+      "epoch": 0.30380681448022756,
+      "grad_norm": 0.15734368562698364,
+      "learning_rate": 4.696200424185831e-06,
+      "loss": 0.952,
+      "step": 41970
+    },
+    {
+      "epoch": 0.3038792011408138,
+      "grad_norm": 0.17791064083576202,
+      "learning_rate": 4.696128037525246e-06,
+      "loss": 0.9507,
+      "step": 41980
+    },
+    {
+      "epoch": 0.30395158780139997,
+      "grad_norm": 0.18303173780441284,
+      "learning_rate": 4.696055650864659e-06,
+      "loss": 0.9595,
+      "step": 41990
+    },
+    {
+      "epoch": 0.30402397446198615,
+      "grad_norm": 0.1724705994129181,
+      "learning_rate": 4.695983264204073e-06,
+      "loss": 0.9704,
+      "step": 42000
+    },
+    {
+      "epoch": 0.3040963611225723,
+      "grad_norm": 0.164495587348938,
+      "learning_rate": 4.695910877543487e-06,
+      "loss": 0.9598,
+      "step": 42010
+    },
+    {
+      "epoch": 0.3041687477831585,
+      "grad_norm": 0.16240546107292175,
+      "learning_rate": 4.6958384908829e-06,
+      "loss": 0.9621,
+      "step": 42020
+    },
+    {
+      "epoch": 0.30424113444374473,
+      "grad_norm": 0.17704802751541138,
+      "learning_rate": 4.695766104222314e-06,
+      "loss": 0.9589,
+      "step": 42030
+    },
+    {
+      "epoch": 0.3043135211043309,
+      "grad_norm": 0.17840299010276794,
+      "learning_rate": 4.695693717561728e-06,
+      "loss": 0.9773,
+      "step": 42040
+    },
+    {
+      "epoch": 0.3043859077649171,
+      "grad_norm": 0.18059472739696503,
+      "learning_rate": 4.695621330901142e-06,
+      "loss": 0.9743,
+      "step": 42050
+    },
+    {
+      "epoch": 0.30445829442550326,
+      "grad_norm": 0.18101860582828522,
+      "learning_rate": 4.6955489442405555e-06,
+      "loss": 0.9618,
+      "step": 42060
+    },
+    {
+      "epoch": 0.30453068108608944,
+      "grad_norm": 0.16693522036075592,
+      "learning_rate": 4.695476557579969e-06,
+      "loss": 0.9533,
+      "step": 42070
+    },
+    {
+      "epoch": 0.3046030677466756,
+      "grad_norm": 0.1905452311038971,
+      "learning_rate": 4.695404170919383e-06,
+      "loss": 0.9576,
+      "step": 42080
+    },
+    {
+      "epoch": 0.30467545440726185,
+      "grad_norm": 0.16793328523635864,
+      "learning_rate": 4.695331784258797e-06,
+      "loss": 0.9592,
+      "step": 42090
+    },
+    {
+      "epoch": 0.304747841067848,
+      "grad_norm": 0.17872589826583862,
+      "learning_rate": 4.695259397598211e-06,
+      "loss": 0.9747,
+      "step": 42100
+    },
+    {
+      "epoch": 0.3048202277284342,
+      "grad_norm": 0.1794547736644745,
+      "learning_rate": 4.6951870109376244e-06,
+      "loss": 0.951,
+      "step": 42110
+    },
+    {
+      "epoch": 0.3048926143890204,
+      "grad_norm": 0.1795600950717926,
+      "learning_rate": 4.695114624277038e-06,
+      "loss": 0.9495,
+      "step": 42120
+    },
+    {
+      "epoch": 0.30496500104960655,
+      "grad_norm": 0.16469037532806396,
+      "learning_rate": 4.6950422376164525e-06,
+      "loss": 0.9677,
+      "step": 42130
+    },
+    {
+      "epoch": 0.3050373877101928,
+      "grad_norm": 0.15928620100021362,
+      "learning_rate": 4.694969850955866e-06,
+      "loss": 0.9597,
+      "step": 42140
+    },
+    {
+      "epoch": 0.30510977437077896,
+      "grad_norm": 0.17253397405147552,
+      "learning_rate": 4.69489746429528e-06,
+      "loss": 0.9603,
+      "step": 42150
+    },
+    {
+      "epoch": 0.30518216103136514,
+      "grad_norm": 0.1894710212945938,
+      "learning_rate": 4.694825077634693e-06,
+      "loss": 0.9537,
+      "step": 42160
+    },
+    {
+      "epoch": 0.3052545476919513,
+      "grad_norm": 0.16059796512126923,
+      "learning_rate": 4.694752690974108e-06,
+      "loss": 0.9595,
+      "step": 42170
+    },
+    {
+      "epoch": 0.3053269343525375,
+      "grad_norm": 0.156858429312706,
+      "learning_rate": 4.6946803043135215e-06,
+      "loss": 0.9604,
+      "step": 42180
+    },
+    {
+      "epoch": 0.3053993210131237,
+      "grad_norm": 0.17593984305858612,
+      "learning_rate": 4.694607917652935e-06,
+      "loss": 0.955,
+      "step": 42190
+    },
+    {
+      "epoch": 0.3054717076737099,
+      "grad_norm": 0.17821641266345978,
+      "learning_rate": 4.694535530992349e-06,
+      "loss": 0.959,
+      "step": 42200
+    },
+    {
+      "epoch": 0.3055440943342961,
+      "grad_norm": 0.25540754199028015,
+      "learning_rate": 4.694463144331763e-06,
+      "loss": 0.9672,
+      "step": 42210
+    },
+    {
+      "epoch": 0.30561648099488226,
+      "grad_norm": 0.19185206294059753,
+      "learning_rate": 4.694390757671177e-06,
+      "loss": 0.9605,
+      "step": 42220
+    },
+    {
+      "epoch": 0.30568886765546843,
+      "grad_norm": 0.21133938431739807,
+      "learning_rate": 4.69431837101059e-06,
+      "loss": 0.9783,
+      "step": 42230
+    },
+    {
+      "epoch": 0.30576125431605466,
+      "grad_norm": 0.17313557863235474,
+      "learning_rate": 4.694245984350004e-06,
+      "loss": 0.964,
+      "step": 42240
+    },
+    {
+      "epoch": 0.30583364097664084,
+      "grad_norm": 0.18542583286762238,
+      "learning_rate": 4.6941735976894185e-06,
+      "loss": 0.9428,
+      "step": 42250
+    },
+    {
+      "epoch": 0.305906027637227,
+      "grad_norm": 0.17269088327884674,
+      "learning_rate": 4.694101211028832e-06,
+      "loss": 0.974,
+      "step": 42260
+    },
+    {
+      "epoch": 0.3059784142978132,
+      "grad_norm": 0.16022367775440216,
+      "learning_rate": 4.694028824368246e-06,
+      "loss": 0.9644,
+      "step": 42270
+    },
+    {
+      "epoch": 0.30605080095839937,
+      "grad_norm": 0.17748059332370758,
+      "learning_rate": 4.693956437707659e-06,
+      "loss": 0.9604,
+      "step": 42280
+    },
+    {
+      "epoch": 0.30612318761898555,
+      "grad_norm": 0.16204720735549927,
+      "learning_rate": 4.693884051047074e-06,
+      "loss": 0.9672,
+      "step": 42290
+    },
+    {
+      "epoch": 0.3061955742795718,
+      "grad_norm": 0.16240718960762024,
+      "learning_rate": 4.693811664386487e-06,
+      "loss": 0.9703,
+      "step": 42300
+    },
+    {
+      "epoch": 0.30626796094015796,
+      "grad_norm": 0.1833665817975998,
+      "learning_rate": 4.693739277725901e-06,
+      "loss": 0.9671,
+      "step": 42310
+    },
+    {
+      "epoch": 0.30634034760074413,
+      "grad_norm": 0.16456997394561768,
+      "learning_rate": 4.693666891065315e-06,
+      "loss": 0.9566,
+      "step": 42320
+    },
+    {
+      "epoch": 0.3064127342613303,
+      "grad_norm": 0.1582067459821701,
+      "learning_rate": 4.693594504404729e-06,
+      "loss": 0.9696,
+      "step": 42330
+    },
+    {
+      "epoch": 0.3064851209219165,
+      "grad_norm": 0.16254618763923645,
+      "learning_rate": 4.693522117744143e-06,
+      "loss": 0.9624,
+      "step": 42340
+    },
+    {
+      "epoch": 0.3065575075825027,
+      "grad_norm": 0.22306321561336517,
+      "learning_rate": 4.693449731083556e-06,
+      "loss": 0.9663,
+      "step": 42350
+    },
+    {
+      "epoch": 0.3066298942430889,
+      "grad_norm": 0.1735800951719284,
+      "learning_rate": 4.69337734442297e-06,
+      "loss": 0.9727,
+      "step": 42360
+    },
+    {
+      "epoch": 0.3067022809036751,
+      "grad_norm": 0.16660434007644653,
+      "learning_rate": 4.693304957762384e-06,
+      "loss": 0.9558,
+      "step": 42370
+    },
+    {
+      "epoch": 0.30677466756426125,
+      "grad_norm": 0.1673038899898529,
+      "learning_rate": 4.693232571101798e-06,
+      "loss": 0.9772,
+      "step": 42380
+    },
+    {
+      "epoch": 0.3068470542248474,
+      "grad_norm": 0.1805497705936432,
+      "learning_rate": 4.693160184441212e-06,
+      "loss": 0.9679,
+      "step": 42390
+    },
+    {
+      "epoch": 0.30691944088543366,
+      "grad_norm": 0.16256369650363922,
+      "learning_rate": 4.693087797780625e-06,
+      "loss": 0.9636,
+      "step": 42400
+    },
+    {
+      "epoch": 0.30699182754601984,
+      "grad_norm": 0.16821902990341187,
+      "learning_rate": 4.69301541112004e-06,
+      "loss": 0.9604,
+      "step": 42410
+    },
+    {
+      "epoch": 0.307064214206606,
+      "grad_norm": 0.170756995677948,
+      "learning_rate": 4.692943024459453e-06,
+      "loss": 0.9493,
+      "step": 42420
+    },
+    {
+      "epoch": 0.3071366008671922,
+      "grad_norm": 0.1680414229631424,
+      "learning_rate": 4.692870637798867e-06,
+      "loss": 0.9669,
+      "step": 42430
+    },
+    {
+      "epoch": 0.30720898752777837,
+      "grad_norm": 0.17271284759044647,
+      "learning_rate": 4.6927982511382806e-06,
+      "loss": 0.9572,
+      "step": 42440
+    },
+    {
+      "epoch": 0.30728137418836454,
+      "grad_norm": 0.20150049030780792,
+      "learning_rate": 4.692725864477695e-06,
+      "loss": 0.976,
+      "step": 42450
+    },
+    {
+      "epoch": 0.3073537608489508,
+      "grad_norm": 0.16273252665996552,
+      "learning_rate": 4.692653477817109e-06,
+      "loss": 0.9646,
+      "step": 42460
+    },
+    {
+      "epoch": 0.30742614750953695,
+      "grad_norm": 0.17150864005088806,
+      "learning_rate": 4.692581091156522e-06,
+      "loss": 0.9734,
+      "step": 42470
+    },
+    {
+      "epoch": 0.30749853417012313,
+      "grad_norm": 0.16339828073978424,
+      "learning_rate": 4.692508704495936e-06,
+      "loss": 0.959,
+      "step": 42480
+    },
+    {
+      "epoch": 0.3075709208307093,
+      "grad_norm": 0.17035742104053497,
+      "learning_rate": 4.69243631783535e-06,
+      "loss": 0.9637,
+      "step": 42490
+    },
+    {
+      "epoch": 0.3076433074912955,
+      "grad_norm": 0.17820675671100616,
+      "learning_rate": 4.692363931174764e-06,
+      "loss": 0.9647,
+      "step": 42500
+    },
+    {
+      "epoch": 0.3077156941518817,
+      "grad_norm": 0.15689398348331451,
+      "learning_rate": 4.6922915445141776e-06,
+      "loss": 0.966,
+      "step": 42510
+    },
+    {
+      "epoch": 0.3077880808124679,
+      "grad_norm": 0.17147964239120483,
+      "learning_rate": 4.692219157853591e-06,
+      "loss": 0.9748,
+      "step": 42520
+    },
+    {
+      "epoch": 0.30786046747305407,
+      "grad_norm": 0.17200268805027008,
+      "learning_rate": 4.692146771193005e-06,
+      "loss": 0.9656,
+      "step": 42530
+    },
+    {
+      "epoch": 0.30793285413364024,
+      "grad_norm": 0.18745240569114685,
+      "learning_rate": 4.692074384532418e-06,
+      "loss": 0.9666,
+      "step": 42540
+    },
+    {
+      "epoch": 0.3080052407942264,
+      "grad_norm": 0.16291543841362,
+      "learning_rate": 4.692001997871832e-06,
+      "loss": 0.9636,
+      "step": 42550
+    },
+    {
+      "epoch": 0.30807762745481265,
+      "grad_norm": 0.16443657875061035,
+      "learning_rate": 4.6919296112112465e-06,
+      "loss": 0.9673,
+      "step": 42560
+    },
+    {
+      "epoch": 0.30815001411539883,
+      "grad_norm": 0.1615900695323944,
+      "learning_rate": 4.69185722455066e-06,
+      "loss": 0.9578,
+      "step": 42570
+    },
+    {
+      "epoch": 0.308222400775985,
+      "grad_norm": 0.17126181721687317,
+      "learning_rate": 4.691784837890074e-06,
+      "loss": 0.9695,
+      "step": 42580
+    },
+    {
+      "epoch": 0.3082947874365712,
+      "grad_norm": 0.15183715522289276,
+      "learning_rate": 4.691712451229487e-06,
+      "loss": 0.9695,
+      "step": 42590
+    },
+    {
+      "epoch": 0.30836717409715736,
+      "grad_norm": 0.15649786591529846,
+      "learning_rate": 4.691640064568902e-06,
+      "loss": 0.9661,
+      "step": 42600
+    },
+    {
+      "epoch": 0.30843956075774354,
+      "grad_norm": 0.15690255165100098,
+      "learning_rate": 4.691567677908315e-06,
+      "loss": 0.9723,
+      "step": 42610
+    },
+    {
+      "epoch": 0.30851194741832977,
+      "grad_norm": 0.16935759782791138,
+      "learning_rate": 4.691495291247729e-06,
+      "loss": 0.9786,
+      "step": 42620
+    },
+    {
+      "epoch": 0.30858433407891595,
+      "grad_norm": 0.19255200028419495,
+      "learning_rate": 4.691422904587143e-06,
+      "loss": 0.9625,
+      "step": 42630
+    },
+    {
+      "epoch": 0.3086567207395021,
+      "grad_norm": 0.16277821362018585,
+      "learning_rate": 4.691350517926557e-06,
+      "loss": 0.9501,
+      "step": 42640
+    },
+    {
+      "epoch": 0.3087291074000883,
+      "grad_norm": 0.15837381780147552,
+      "learning_rate": 4.691278131265971e-06,
+      "loss": 0.9564,
+      "step": 42650
+    },
+    {
+      "epoch": 0.3088014940606745,
+      "grad_norm": 0.16267119348049164,
+      "learning_rate": 4.691205744605384e-06,
+      "loss": 0.961,
+      "step": 42660
+    },
+    {
+      "epoch": 0.3088738807212607,
+      "grad_norm": 0.17577193677425385,
+      "learning_rate": 4.691133357944798e-06,
+      "loss": 0.9807,
+      "step": 42670
+    },
+    {
+      "epoch": 0.3089462673818469,
+      "grad_norm": 0.16068010032176971,
+      "learning_rate": 4.691060971284212e-06,
+      "loss": 0.9623,
+      "step": 42680
+    },
+    {
+      "epoch": 0.30901865404243306,
+      "grad_norm": 0.16693639755249023,
+      "learning_rate": 4.690988584623626e-06,
+      "loss": 0.9602,
+      "step": 42690
+    },
+    {
+      "epoch": 0.30909104070301924,
+      "grad_norm": 0.17629244923591614,
+      "learning_rate": 4.69091619796304e-06,
+      "loss": 0.9785,
+      "step": 42700
+    },
+    {
+      "epoch": 0.3091634273636054,
+      "grad_norm": 0.15694406628608704,
+      "learning_rate": 4.690843811302453e-06,
+      "loss": 0.9756,
+      "step": 42710
+    },
+    {
+      "epoch": 0.30923581402419165,
+      "grad_norm": 0.16043122112751007,
+      "learning_rate": 4.690771424641867e-06,
+      "loss": 0.9697,
+      "step": 42720
+    },
+    {
+      "epoch": 0.3093082006847778,
+      "grad_norm": 0.18830914795398712,
+      "learning_rate": 4.690699037981281e-06,
+      "loss": 0.9642,
+      "step": 42730
+    },
+    {
+      "epoch": 0.309380587345364,
+      "grad_norm": 0.1912577599287033,
+      "learning_rate": 4.690626651320695e-06,
+      "loss": 0.962,
+      "step": 42740
+    },
+    {
+      "epoch": 0.3094529740059502,
+      "grad_norm": 0.16711075603961945,
+      "learning_rate": 4.690554264660109e-06,
+      "loss": 0.959,
+      "step": 42750
+    },
+    {
+      "epoch": 0.30952536066653635,
+      "grad_norm": 0.17442958056926727,
+      "learning_rate": 4.690481877999522e-06,
+      "loss": 0.9752,
+      "step": 42760
+    },
+    {
+      "epoch": 0.3095977473271226,
+      "grad_norm": 0.17477945983409882,
+      "learning_rate": 4.690409491338937e-06,
+      "loss": 0.9708,
+      "step": 42770
+    },
+    {
+      "epoch": 0.30967013398770876,
+      "grad_norm": 0.1687520295381546,
+      "learning_rate": 4.69033710467835e-06,
+      "loss": 0.9602,
+      "step": 42780
+    },
+    {
+      "epoch": 0.30974252064829494,
+      "grad_norm": 0.1918717324733734,
+      "learning_rate": 4.690264718017764e-06,
+      "loss": 0.9622,
+      "step": 42790
+    },
+    {
+      "epoch": 0.3098149073088811,
+      "grad_norm": 0.17581704258918762,
+      "learning_rate": 4.6901923313571775e-06,
+      "loss": 0.9502,
+      "step": 42800
+    },
+    {
+      "epoch": 0.3098872939694673,
+      "grad_norm": 0.1803891360759735,
+      "learning_rate": 4.690119944696592e-06,
+      "loss": 0.9743,
+      "step": 42810
+    },
+    {
+      "epoch": 0.30995968063005347,
+      "grad_norm": 0.16164544224739075,
+      "learning_rate": 4.690047558036006e-06,
+      "loss": 0.9597,
+      "step": 42820
+    },
+    {
+      "epoch": 0.3100320672906397,
+      "grad_norm": 0.15992704033851624,
+      "learning_rate": 4.689975171375419e-06,
+      "loss": 0.9745,
+      "step": 42830
+    },
+    {
+      "epoch": 0.3101044539512259,
+      "grad_norm": 0.16522592306137085,
+      "learning_rate": 4.689902784714833e-06,
+      "loss": 0.9392,
+      "step": 42840
+    },
+    {
+      "epoch": 0.31017684061181205,
+      "grad_norm": 0.17075496912002563,
+      "learning_rate": 4.689830398054247e-06,
+      "loss": 0.9706,
+      "step": 42850
+    },
+    {
+      "epoch": 0.31024922727239823,
+      "grad_norm": 0.16021926701068878,
+      "learning_rate": 4.689758011393661e-06,
+      "loss": 0.9651,
+      "step": 42860
+    },
+    {
+      "epoch": 0.3103216139329844,
+      "grad_norm": 0.16339780390262604,
+      "learning_rate": 4.6896856247330745e-06,
+      "loss": 0.9626,
+      "step": 42870
+    },
+    {
+      "epoch": 0.31039400059357064,
+      "grad_norm": 0.15929320454597473,
+      "learning_rate": 4.689613238072488e-06,
+      "loss": 0.9484,
+      "step": 42880
+    },
+    {
+      "epoch": 0.3104663872541568,
+      "grad_norm": 0.1761467158794403,
+      "learning_rate": 4.689540851411903e-06,
+      "loss": 0.9623,
+      "step": 42890
+    },
+    {
+      "epoch": 0.310538773914743,
+      "grad_norm": 0.16710932552814484,
+      "learning_rate": 4.689468464751316e-06,
+      "loss": 0.959,
+      "step": 42900
+    },
+    {
+      "epoch": 0.31061116057532917,
+      "grad_norm": 0.16465498507022858,
+      "learning_rate": 4.68939607809073e-06,
+      "loss": 0.9584,
+      "step": 42910
+    },
+    {
+      "epoch": 0.31068354723591535,
+      "grad_norm": 0.1967269629240036,
+      "learning_rate": 4.6893236914301435e-06,
+      "loss": 0.9583,
+      "step": 42920
+    },
+    {
+      "epoch": 0.3107559338965016,
+      "grad_norm": 0.16868357360363007,
+      "learning_rate": 4.689251304769558e-06,
+      "loss": 0.9559,
+      "step": 42930
+    },
+    {
+      "epoch": 0.31082832055708776,
+      "grad_norm": 0.15778405964374542,
+      "learning_rate": 4.6891789181089715e-06,
+      "loss": 0.9707,
+      "step": 42940
+    },
+    {
+      "epoch": 0.31090070721767393,
+      "grad_norm": 0.20243608951568604,
+      "learning_rate": 4.689106531448385e-06,
+      "loss": 0.9471,
+      "step": 42950
+    },
+    {
+      "epoch": 0.3109730938782601,
+      "grad_norm": 0.17772892117500305,
+      "learning_rate": 4.689034144787799e-06,
+      "loss": 0.9589,
+      "step": 42960
+    },
+    {
+      "epoch": 0.3110454805388463,
+      "grad_norm": 0.17439591884613037,
+      "learning_rate": 4.688961758127213e-06,
+      "loss": 0.9604,
+      "step": 42970
+    },
+    {
+      "epoch": 0.31111786719943246,
+      "grad_norm": 0.15958340466022491,
+      "learning_rate": 4.688889371466627e-06,
+      "loss": 0.967,
+      "step": 42980
+    },
+    {
+      "epoch": 0.3111902538600187,
+      "grad_norm": 0.16158387064933777,
+      "learning_rate": 4.6888169848060405e-06,
+      "loss": 0.9569,
+      "step": 42990
+    },
+    {
+      "epoch": 0.31126264052060487,
+      "grad_norm": 0.17039985954761505,
+      "learning_rate": 4.688744598145454e-06,
+      "loss": 0.9571,
+      "step": 43000
+    },
+    {
+      "epoch": 0.31133502718119105,
+      "grad_norm": 0.1581193506717682,
+      "learning_rate": 4.6886722114848685e-06,
+      "loss": 0.9568,
+      "step": 43010
+    },
+    {
+      "epoch": 0.3114074138417772,
+      "grad_norm": 0.16639426350593567,
+      "learning_rate": 4.688599824824282e-06,
+      "loss": 0.9495,
+      "step": 43020
+    },
+    {
+      "epoch": 0.3114798005023634,
+      "grad_norm": 0.1791532039642334,
+      "learning_rate": 4.688527438163696e-06,
+      "loss": 0.973,
+      "step": 43030
+    },
+    {
+      "epoch": 0.31155218716294963,
+      "grad_norm": 0.16932880878448486,
+      "learning_rate": 4.688455051503109e-06,
+      "loss": 0.966,
+      "step": 43040
+    },
+    {
+      "epoch": 0.3116245738235358,
+      "grad_norm": 0.17780663073062897,
+      "learning_rate": 4.688382664842524e-06,
+      "loss": 0.9554,
+      "step": 43050
+    },
+    {
+      "epoch": 0.311696960484122,
+      "grad_norm": 0.1656964272260666,
+      "learning_rate": 4.6883102781819375e-06,
+      "loss": 0.9728,
+      "step": 43060
+    },
+    {
+      "epoch": 0.31176934714470816,
+      "grad_norm": 0.15379230678081512,
+      "learning_rate": 4.68823789152135e-06,
+      "loss": 0.9739,
+      "step": 43070
+    },
+    {
+      "epoch": 0.31184173380529434,
+      "grad_norm": 0.17420606315135956,
+      "learning_rate": 4.688165504860765e-06,
+      "loss": 0.9573,
+      "step": 43080
+    },
+    {
+      "epoch": 0.3119141204658806,
+      "grad_norm": 0.17237995564937592,
+      "learning_rate": 4.688093118200178e-06,
+      "loss": 0.9633,
+      "step": 43090
+    },
+    {
+      "epoch": 0.31198650712646675,
+      "grad_norm": 0.254451185464859,
+      "learning_rate": 4.688020731539592e-06,
+      "loss": 0.9597,
+      "step": 43100
+    },
+    {
+      "epoch": 0.3120588937870529,
+      "grad_norm": 0.17835448682308197,
+      "learning_rate": 4.6879483448790056e-06,
+      "loss": 0.9491,
+      "step": 43110
+    },
+    {
+      "epoch": 0.3121312804476391,
+      "grad_norm": 0.1625124216079712,
+      "learning_rate": 4.68787595821842e-06,
+      "loss": 0.9623,
+      "step": 43120
+    },
+    {
+      "epoch": 0.3122036671082253,
+      "grad_norm": 0.1648617535829544,
+      "learning_rate": 4.687803571557834e-06,
+      "loss": 0.9626,
+      "step": 43130
+    },
+    {
+      "epoch": 0.31227605376881146,
+      "grad_norm": 0.15769249200820923,
+      "learning_rate": 4.687731184897247e-06,
+      "loss": 0.9585,
+      "step": 43140
+    },
+    {
+      "epoch": 0.3123484404293977,
+      "grad_norm": 0.16843098402023315,
+      "learning_rate": 4.687658798236661e-06,
+      "loss": 0.9519,
+      "step": 43150
+    },
+    {
+      "epoch": 0.31242082708998387,
+      "grad_norm": 0.17640751600265503,
+      "learning_rate": 4.687586411576075e-06,
+      "loss": 0.9519,
+      "step": 43160
+    },
+    {
+      "epoch": 0.31249321375057004,
+      "grad_norm": 0.2160252183675766,
+      "learning_rate": 4.687514024915489e-06,
+      "loss": 0.9505,
+      "step": 43170
+    },
+    {
+      "epoch": 0.3125656004111562,
+      "grad_norm": 0.1599891185760498,
+      "learning_rate": 4.6874416382549026e-06,
+      "loss": 0.9592,
+      "step": 43180
+    },
+    {
+      "epoch": 0.3126379870717424,
+      "grad_norm": 0.17435148358345032,
+      "learning_rate": 4.687369251594316e-06,
+      "loss": 0.9647,
+      "step": 43190
+    },
+    {
+      "epoch": 0.31271037373232863,
+      "grad_norm": 0.18090033531188965,
+      "learning_rate": 4.687296864933731e-06,
+      "loss": 0.9587,
+      "step": 43200
+    },
+    {
+      "epoch": 0.3127827603929148,
+      "grad_norm": 0.15829293429851532,
+      "learning_rate": 4.687224478273144e-06,
+      "loss": 0.9601,
+      "step": 43210
+    },
+    {
+      "epoch": 0.312855147053501,
+      "grad_norm": 0.17692400515079498,
+      "learning_rate": 4.687152091612558e-06,
+      "loss": 0.964,
+      "step": 43220
+    },
+    {
+      "epoch": 0.31292753371408716,
+      "grad_norm": 0.18780209124088287,
+      "learning_rate": 4.6870797049519715e-06,
+      "loss": 0.9679,
+      "step": 43230
+    },
+    {
+      "epoch": 0.31299992037467333,
+      "grad_norm": 0.20248176157474518,
+      "learning_rate": 4.687007318291386e-06,
+      "loss": 0.9639,
+      "step": 43240
+    },
+    {
+      "epoch": 0.31307230703525957,
+      "grad_norm": 0.1714249551296234,
+      "learning_rate": 4.6869349316307996e-06,
+      "loss": 0.9524,
+      "step": 43250
+    },
+    {
+      "epoch": 0.31314469369584574,
+      "grad_norm": 0.1626296043395996,
+      "learning_rate": 4.686862544970213e-06,
+      "loss": 0.9643,
+      "step": 43260
+    },
+    {
+      "epoch": 0.3132170803564319,
+      "grad_norm": 0.21206989884376526,
+      "learning_rate": 4.686790158309627e-06,
+      "loss": 0.9681,
+      "step": 43270
+    },
+    {
+      "epoch": 0.3132894670170181,
+      "grad_norm": 0.16204078495502472,
+      "learning_rate": 4.686717771649041e-06,
+      "loss": 0.9671,
+      "step": 43280
+    },
+    {
+      "epoch": 0.3133618536776043,
+      "grad_norm": 0.1752130091190338,
+      "learning_rate": 4.686645384988455e-06,
+      "loss": 0.969,
+      "step": 43290
+    },
+    {
+      "epoch": 0.31343424033819045,
+      "grad_norm": 0.16563290357589722,
+      "learning_rate": 4.6865729983278685e-06,
+      "loss": 0.9686,
+      "step": 43300
+    },
+    {
+      "epoch": 0.3135066269987767,
+      "grad_norm": 0.16079412400722504,
+      "learning_rate": 4.686500611667282e-06,
+      "loss": 0.9783,
+      "step": 43310
+    },
+    {
+      "epoch": 0.31357901365936286,
+      "grad_norm": 0.19905321300029755,
+      "learning_rate": 4.686428225006696e-06,
+      "loss": 0.9657,
+      "step": 43320
+    },
+    {
+      "epoch": 0.31365140031994904,
+      "grad_norm": 0.1775355190038681,
+      "learning_rate": 4.68635583834611e-06,
+      "loss": 0.9536,
+      "step": 43330
+    },
+    {
+      "epoch": 0.3137237869805352,
+      "grad_norm": 0.16202837228775024,
+      "learning_rate": 4.686283451685524e-06,
+      "loss": 0.9783,
+      "step": 43340
+    },
+    {
+      "epoch": 0.3137961736411214,
+      "grad_norm": 0.17238853871822357,
+      "learning_rate": 4.686211065024937e-06,
+      "loss": 0.9829,
+      "step": 43350
+    },
+    {
+      "epoch": 0.3138685603017076,
+      "grad_norm": 0.16124895215034485,
+      "learning_rate": 4.686138678364351e-06,
+      "loss": 0.9646,
+      "step": 43360
+    },
+    {
+      "epoch": 0.3139409469622938,
+      "grad_norm": 0.3990856111049652,
+      "learning_rate": 4.6860662917037655e-06,
+      "loss": 0.9583,
+      "step": 43370
+    },
+    {
+      "epoch": 0.31401333362288,
+      "grad_norm": 0.168771892786026,
+      "learning_rate": 4.685993905043179e-06,
+      "loss": 0.9749,
+      "step": 43380
+    },
+    {
+      "epoch": 0.31408572028346615,
+      "grad_norm": 0.16756808757781982,
+      "learning_rate": 4.685921518382593e-06,
+      "loss": 0.9551,
+      "step": 43390
+    },
+    {
+      "epoch": 0.31415810694405233,
+      "grad_norm": 0.16926801204681396,
+      "learning_rate": 4.685849131722006e-06,
+      "loss": 0.9635,
+      "step": 43400
+    },
+    {
+      "epoch": 0.31423049360463856,
+      "grad_norm": 0.17521944642066956,
+      "learning_rate": 4.685776745061421e-06,
+      "loss": 0.967,
+      "step": 43410
+    },
+    {
+      "epoch": 0.31430288026522474,
+      "grad_norm": 0.16481991112232208,
+      "learning_rate": 4.6857043584008344e-06,
+      "loss": 0.9695,
+      "step": 43420
+    },
+    {
+      "epoch": 0.3143752669258109,
+      "grad_norm": 0.15918724238872528,
+      "learning_rate": 4.685631971740248e-06,
+      "loss": 0.9679,
+      "step": 43430
+    },
+    {
+      "epoch": 0.3144476535863971,
+      "grad_norm": 0.1576610803604126,
+      "learning_rate": 4.685559585079662e-06,
+      "loss": 0.9572,
+      "step": 43440
+    },
+    {
+      "epoch": 0.31452004024698327,
+      "grad_norm": 0.16008280217647552,
+      "learning_rate": 4.685487198419076e-06,
+      "loss": 0.9626,
+      "step": 43450
+    },
+    {
+      "epoch": 0.3145924269075695,
+      "grad_norm": 0.16831348836421967,
+      "learning_rate": 4.68541481175849e-06,
+      "loss": 0.9513,
+      "step": 43460
+    },
+    {
+      "epoch": 0.3146648135681557,
+      "grad_norm": 0.2367899864912033,
+      "learning_rate": 4.685342425097903e-06,
+      "loss": 0.9394,
+      "step": 43470
+    },
+    {
+      "epoch": 0.31473720022874185,
+      "grad_norm": 0.16397932171821594,
+      "learning_rate": 4.685270038437317e-06,
+      "loss": 0.9559,
+      "step": 43480
+    },
+    {
+      "epoch": 0.31480958688932803,
+      "grad_norm": 0.16159677505493164,
+      "learning_rate": 4.6851976517767314e-06,
+      "loss": 0.9623,
+      "step": 43490
+    },
+    {
+      "epoch": 0.3148819735499142,
+      "grad_norm": 0.16152982413768768,
+      "learning_rate": 4.685125265116145e-06,
+      "loss": 0.9662,
+      "step": 43500
+    },
+    {
+      "epoch": 0.3149543602105004,
+      "grad_norm": 0.19313648343086243,
+      "learning_rate": 4.685052878455559e-06,
+      "loss": 0.9582,
+      "step": 43510
+    },
+    {
+      "epoch": 0.3150267468710866,
+      "grad_norm": 0.16740714013576508,
+      "learning_rate": 4.684980491794972e-06,
+      "loss": 0.959,
+      "step": 43520
+    },
+    {
+      "epoch": 0.3150991335316728,
+      "grad_norm": 0.1715911626815796,
+      "learning_rate": 4.684908105134387e-06,
+      "loss": 0.9513,
+      "step": 43530
+    },
+    {
+      "epoch": 0.31517152019225897,
+      "grad_norm": 0.1693735122680664,
+      "learning_rate": 4.6848357184738e-06,
+      "loss": 0.9747,
+      "step": 43540
+    },
+    {
+      "epoch": 0.31524390685284515,
+      "grad_norm": 0.16398315131664276,
+      "learning_rate": 4.684763331813214e-06,
+      "loss": 0.9639,
+      "step": 43550
+    },
+    {
+      "epoch": 0.3153162935134313,
+      "grad_norm": 0.16864167153835297,
+      "learning_rate": 4.684690945152628e-06,
+      "loss": 0.9509,
+      "step": 43560
+    },
+    {
+      "epoch": 0.31538868017401755,
+      "grad_norm": 0.15336643159389496,
+      "learning_rate": 4.684618558492042e-06,
+      "loss": 0.9669,
+      "step": 43570
+    },
+    {
+      "epoch": 0.31546106683460373,
+      "grad_norm": 0.17718195915222168,
+      "learning_rate": 4.684546171831456e-06,
+      "loss": 0.9624,
+      "step": 43580
+    },
+    {
+      "epoch": 0.3155334534951899,
+      "grad_norm": 0.17874468863010406,
+      "learning_rate": 4.684473785170869e-06,
+      "loss": 0.9659,
+      "step": 43590
+    },
+    {
+      "epoch": 0.3156058401557761,
+      "grad_norm": 0.1714981645345688,
+      "learning_rate": 4.684401398510283e-06,
+      "loss": 0.9621,
+      "step": 43600
+    },
+    {
+      "epoch": 0.31567822681636226,
+      "grad_norm": 0.15359127521514893,
+      "learning_rate": 4.6843290118496965e-06,
+      "loss": 0.9701,
+      "step": 43610
+    },
+    {
+      "epoch": 0.3157506134769485,
+      "grad_norm": 0.2659284472465515,
+      "learning_rate": 4.68425662518911e-06,
+      "loss": 0.9573,
+      "step": 43620
+    },
+    {
+      "epoch": 0.31582300013753467,
+      "grad_norm": 0.15532797574996948,
+      "learning_rate": 4.684184238528524e-06,
+      "loss": 0.96,
+      "step": 43630
+    },
+    {
+      "epoch": 0.31589538679812085,
+      "grad_norm": 0.17984911799430847,
+      "learning_rate": 4.684111851867938e-06,
+      "loss": 0.9591,
+      "step": 43640
+    },
+    {
+      "epoch": 0.315967773458707,
+      "grad_norm": 0.16166174411773682,
+      "learning_rate": 4.684039465207352e-06,
+      "loss": 0.9545,
+      "step": 43650
+    },
+    {
+      "epoch": 0.3160401601192932,
+      "grad_norm": 0.16693688929080963,
+      "learning_rate": 4.6839670785467655e-06,
+      "loss": 0.972,
+      "step": 43660
+    },
+    {
+      "epoch": 0.3161125467798794,
+      "grad_norm": 0.1600932478904724,
+      "learning_rate": 4.683894691886179e-06,
+      "loss": 0.9576,
+      "step": 43670
+    },
+    {
+      "epoch": 0.3161849334404656,
+      "grad_norm": 0.1789911538362503,
+      "learning_rate": 4.6838223052255935e-06,
+      "loss": 0.9519,
+      "step": 43680
+    },
+    {
+      "epoch": 0.3162573201010518,
+      "grad_norm": 0.16878649592399597,
+      "learning_rate": 4.683749918565007e-06,
+      "loss": 0.9537,
+      "step": 43690
+    },
+    {
+      "epoch": 0.31632970676163796,
+      "grad_norm": 0.18558557331562042,
+      "learning_rate": 4.683677531904421e-06,
+      "loss": 0.9598,
+      "step": 43700
+    },
+    {
+      "epoch": 0.31640209342222414,
+      "grad_norm": 0.1579427868127823,
+      "learning_rate": 4.683605145243834e-06,
+      "loss": 0.9635,
+      "step": 43710
+    },
+    {
+      "epoch": 0.3164744800828103,
+      "grad_norm": 0.16161714494228363,
+      "learning_rate": 4.683532758583249e-06,
+      "loss": 0.9515,
+      "step": 43720
+    },
+    {
+      "epoch": 0.31654686674339655,
+      "grad_norm": 0.15619002282619476,
+      "learning_rate": 4.6834603719226625e-06,
+      "loss": 0.9658,
+      "step": 43730
+    },
+    {
+      "epoch": 0.3166192534039827,
+      "grad_norm": 0.17198404669761658,
+      "learning_rate": 4.683387985262076e-06,
+      "loss": 0.9492,
+      "step": 43740
+    },
+    {
+      "epoch": 0.3166916400645689,
+      "grad_norm": 0.18472731113433838,
+      "learning_rate": 4.68331559860149e-06,
+      "loss": 0.9596,
+      "step": 43750
+    },
+    {
+      "epoch": 0.3167640267251551,
+      "grad_norm": 0.17256297171115875,
+      "learning_rate": 4.683243211940904e-06,
+      "loss": 0.9668,
+      "step": 43760
+    },
+    {
+      "epoch": 0.31683641338574126,
+      "grad_norm": 0.17887750267982483,
+      "learning_rate": 4.683170825280318e-06,
+      "loss": 0.9639,
+      "step": 43770
+    },
+    {
+      "epoch": 0.3169088000463275,
+      "grad_norm": 0.17158237099647522,
+      "learning_rate": 4.683098438619731e-06,
+      "loss": 0.9678,
+      "step": 43780
+    },
+    {
+      "epoch": 0.31698118670691366,
+      "grad_norm": 0.1695108711719513,
+      "learning_rate": 4.683026051959145e-06,
+      "loss": 0.9517,
+      "step": 43790
+    },
+    {
+      "epoch": 0.31705357336749984,
+      "grad_norm": 0.1559198945760727,
+      "learning_rate": 4.6829536652985595e-06,
+      "loss": 0.9627,
+      "step": 43800
+    },
+    {
+      "epoch": 0.317125960028086,
+      "grad_norm": 0.15818607807159424,
+      "learning_rate": 4.682881278637973e-06,
+      "loss": 0.9667,
+      "step": 43810
+    },
+    {
+      "epoch": 0.3171983466886722,
+      "grad_norm": 0.1614452749490738,
+      "learning_rate": 4.682808891977387e-06,
+      "loss": 0.9597,
+      "step": 43820
+    },
+    {
+      "epoch": 0.31727073334925837,
+      "grad_norm": 0.17955391108989716,
+      "learning_rate": 4.6827365053168e-06,
+      "loss": 0.9531,
+      "step": 43830
+    },
+    {
+      "epoch": 0.3173431200098446,
+      "grad_norm": 0.1729903519153595,
+      "learning_rate": 4.682664118656215e-06,
+      "loss": 0.9437,
+      "step": 43840
+    },
+    {
+      "epoch": 0.3174155066704308,
+      "grad_norm": 0.17108261585235596,
+      "learning_rate": 4.682591731995628e-06,
+      "loss": 0.9588,
+      "step": 43850
+    },
+    {
+      "epoch": 0.31748789333101696,
+      "grad_norm": 0.16403260827064514,
+      "learning_rate": 4.682519345335042e-06,
+      "loss": 0.9561,
+      "step": 43860
+    },
+    {
+      "epoch": 0.31756027999160313,
+      "grad_norm": 0.17436562478542328,
+      "learning_rate": 4.682446958674456e-06,
+      "loss": 0.9576,
+      "step": 43870
+    },
+    {
+      "epoch": 0.3176326666521893,
+      "grad_norm": 0.16612768173217773,
+      "learning_rate": 4.68237457201387e-06,
+      "loss": 0.9558,
+      "step": 43880
+    },
+    {
+      "epoch": 0.31770505331277554,
+      "grad_norm": 0.16263121366500854,
+      "learning_rate": 4.682302185353284e-06,
+      "loss": 0.9564,
+      "step": 43890
+    },
+    {
+      "epoch": 0.3177774399733617,
+      "grad_norm": 0.24618317186832428,
+      "learning_rate": 4.682229798692697e-06,
+      "loss": 0.9667,
+      "step": 43900
+    },
+    {
+      "epoch": 0.3178498266339479,
+      "grad_norm": 0.16008852422237396,
+      "learning_rate": 4.682157412032111e-06,
+      "loss": 0.973,
+      "step": 43910
+    },
+    {
+      "epoch": 0.3179222132945341,
+      "grad_norm": 0.16652582585811615,
+      "learning_rate": 4.682085025371525e-06,
+      "loss": 0.9534,
+      "step": 43920
+    },
+    {
+      "epoch": 0.31799459995512025,
+      "grad_norm": 0.19293102622032166,
+      "learning_rate": 4.682012638710939e-06,
+      "loss": 0.9546,
+      "step": 43930
+    },
+    {
+      "epoch": 0.3180669866157065,
+      "grad_norm": 0.16652163863182068,
+      "learning_rate": 4.681940252050353e-06,
+      "loss": 0.9599,
+      "step": 43940
+    },
+    {
+      "epoch": 0.31813937327629266,
+      "grad_norm": 0.16719068586826324,
+      "learning_rate": 4.681867865389766e-06,
+      "loss": 0.9807,
+      "step": 43950
+    },
+    {
+      "epoch": 0.31821175993687884,
+      "grad_norm": 0.16578418016433716,
+      "learning_rate": 4.68179547872918e-06,
+      "loss": 0.9627,
+      "step": 43960
+    },
+    {
+      "epoch": 0.318284146597465,
+      "grad_norm": 0.20773844420909882,
+      "learning_rate": 4.681723092068594e-06,
+      "loss": 0.9506,
+      "step": 43970
+    },
+    {
+      "epoch": 0.3183565332580512,
+      "grad_norm": 0.20871974527835846,
+      "learning_rate": 4.681650705408008e-06,
+      "loss": 0.9609,
+      "step": 43980
+    },
+    {
+      "epoch": 0.3184289199186374,
+      "grad_norm": 0.19961602985858917,
+      "learning_rate": 4.6815783187474216e-06,
+      "loss": 0.9696,
+      "step": 43990
+    },
+    {
+      "epoch": 0.3185013065792236,
+      "grad_norm": 0.1632424145936966,
+      "learning_rate": 4.681505932086835e-06,
+      "loss": 0.9569,
+      "step": 44000
+    },
+    {
+      "epoch": 0.3185736932398098,
+      "grad_norm": 0.15213897824287415,
+      "learning_rate": 4.68143354542625e-06,
+      "loss": 0.9503,
+      "step": 44010
+    },
+    {
+      "epoch": 0.31864607990039595,
+      "grad_norm": 0.16649749875068665,
+      "learning_rate": 4.681361158765663e-06,
+      "loss": 0.9645,
+      "step": 44020
+    },
+    {
+      "epoch": 0.3187184665609821,
+      "grad_norm": 0.16990630328655243,
+      "learning_rate": 4.681288772105077e-06,
+      "loss": 0.9532,
+      "step": 44030
+    },
+    {
+      "epoch": 0.3187908532215683,
+      "grad_norm": 0.15943020582199097,
+      "learning_rate": 4.6812163854444905e-06,
+      "loss": 0.9601,
+      "step": 44040
+    },
+    {
+      "epoch": 0.31886323988215454,
+      "grad_norm": 0.1748354434967041,
+      "learning_rate": 4.681143998783905e-06,
+      "loss": 0.9482,
+      "step": 44050
+    },
+    {
+      "epoch": 0.3189356265427407,
+      "grad_norm": 0.16795013844966888,
+      "learning_rate": 4.6810716121233186e-06,
+      "loss": 0.9617,
+      "step": 44060
+    },
+    {
+      "epoch": 0.3190080132033269,
+      "grad_norm": 0.17851009964942932,
+      "learning_rate": 4.680999225462732e-06,
+      "loss": 0.9599,
+      "step": 44070
+    },
+    {
+      "epoch": 0.31908039986391307,
+      "grad_norm": 0.16591772437095642,
+      "learning_rate": 4.680926838802146e-06,
+      "loss": 0.9617,
+      "step": 44080
+    },
+    {
+      "epoch": 0.31915278652449924,
+      "grad_norm": 0.16842985153198242,
+      "learning_rate": 4.68085445214156e-06,
+      "loss": 0.9649,
+      "step": 44090
+    },
+    {
+      "epoch": 0.3192251731850855,
+      "grad_norm": 0.1664043366909027,
+      "learning_rate": 4.680782065480974e-06,
+      "loss": 0.9515,
+      "step": 44100
+    },
+    {
+      "epoch": 0.31929755984567165,
+      "grad_norm": 0.1651202142238617,
+      "learning_rate": 4.6807096788203875e-06,
+      "loss": 0.9699,
+      "step": 44110
+    },
+    {
+      "epoch": 0.31936994650625783,
+      "grad_norm": 0.18120470643043518,
+      "learning_rate": 4.680637292159801e-06,
+      "loss": 0.96,
+      "step": 44120
+    },
+    {
+      "epoch": 0.319442333166844,
+      "grad_norm": 0.1606922298669815,
+      "learning_rate": 4.680564905499215e-06,
+      "loss": 0.9478,
+      "step": 44130
+    },
+    {
+      "epoch": 0.3195147198274302,
+      "grad_norm": 0.16963617503643036,
+      "learning_rate": 4.680492518838628e-06,
+      "loss": 0.9625,
+      "step": 44140
+    },
+    {
+      "epoch": 0.3195871064880164,
+      "grad_norm": 0.17569056153297424,
+      "learning_rate": 4.680420132178042e-06,
+      "loss": 0.9659,
+      "step": 44150
+    },
+    {
+      "epoch": 0.3196594931486026,
+      "grad_norm": 0.16258999705314636,
+      "learning_rate": 4.6803477455174564e-06,
+      "loss": 0.9563,
+      "step": 44160
+    },
+    {
+      "epoch": 0.31973187980918877,
+      "grad_norm": 0.22039775550365448,
+      "learning_rate": 4.68027535885687e-06,
+      "loss": 0.9474,
+      "step": 44170
+    },
+    {
+      "epoch": 0.31980426646977494,
+      "grad_norm": 0.1699448972940445,
+      "learning_rate": 4.680202972196284e-06,
+      "loss": 0.9796,
+      "step": 44180
+    },
+    {
+      "epoch": 0.3198766531303611,
+      "grad_norm": 0.15425299108028412,
+      "learning_rate": 4.680130585535697e-06,
+      "loss": 0.9552,
+      "step": 44190
+    },
+    {
+      "epoch": 0.3199490397909473,
+      "grad_norm": 0.17444351315498352,
+      "learning_rate": 4.680058198875112e-06,
+      "loss": 0.9538,
+      "step": 44200
+    },
+    {
+      "epoch": 0.32002142645153353,
+      "grad_norm": 0.17369182407855988,
+      "learning_rate": 4.679985812214525e-06,
+      "loss": 0.9556,
+      "step": 44210
+    },
+    {
+      "epoch": 0.3200938131121197,
+      "grad_norm": 0.17337919771671295,
+      "learning_rate": 4.679913425553939e-06,
+      "loss": 0.9626,
+      "step": 44220
+    },
+    {
+      "epoch": 0.3201661997727059,
+      "grad_norm": 0.1717376410961151,
+      "learning_rate": 4.679841038893353e-06,
+      "loss": 0.9431,
+      "step": 44230
+    },
+    {
+      "epoch": 0.32023858643329206,
+      "grad_norm": 0.16457335650920868,
+      "learning_rate": 4.679768652232767e-06,
+      "loss": 0.9493,
+      "step": 44240
+    },
+    {
+      "epoch": 0.32031097309387824,
+      "grad_norm": 0.17030712962150574,
+      "learning_rate": 4.679696265572181e-06,
+      "loss": 0.9625,
+      "step": 44250
+    },
+    {
+      "epoch": 0.32038335975446447,
+      "grad_norm": 0.1790560930967331,
+      "learning_rate": 4.679623878911594e-06,
+      "loss": 0.9629,
+      "step": 44260
+    },
+    {
+      "epoch": 0.32045574641505065,
+      "grad_norm": 0.15533442795276642,
+      "learning_rate": 4.679551492251008e-06,
+      "loss": 0.965,
+      "step": 44270
+    },
+    {
+      "epoch": 0.3205281330756368,
+      "grad_norm": 0.15475206077098846,
+      "learning_rate": 4.679479105590422e-06,
+      "loss": 0.9614,
+      "step": 44280
+    },
+    {
+      "epoch": 0.320600519736223,
+      "grad_norm": 0.15645426511764526,
+      "learning_rate": 4.679406718929836e-06,
+      "loss": 0.9481,
+      "step": 44290
+    },
+    {
+      "epoch": 0.3206729063968092,
+      "grad_norm": 0.17138896882534027,
+      "learning_rate": 4.67933433226925e-06,
+      "loss": 0.9605,
+      "step": 44300
+    },
+    {
+      "epoch": 0.3207452930573954,
+      "grad_norm": 0.17752832174301147,
+      "learning_rate": 4.679261945608663e-06,
+      "loss": 0.9538,
+      "step": 44310
+    },
+    {
+      "epoch": 0.3208176797179816,
+      "grad_norm": 0.17925305664539337,
+      "learning_rate": 4.679189558948078e-06,
+      "loss": 0.9536,
+      "step": 44320
+    },
+    {
+      "epoch": 0.32089006637856776,
+      "grad_norm": 0.16660036146640778,
+      "learning_rate": 4.679117172287491e-06,
+      "loss": 0.9566,
+      "step": 44330
+    },
+    {
+      "epoch": 0.32096245303915394,
+      "grad_norm": 0.16511604189872742,
+      "learning_rate": 4.679044785626905e-06,
+      "loss": 0.9473,
+      "step": 44340
+    },
+    {
+      "epoch": 0.3210348396997401,
+      "grad_norm": 0.17227579653263092,
+      "learning_rate": 4.6789723989663185e-06,
+      "loss": 0.9642,
+      "step": 44350
+    },
+    {
+      "epoch": 0.3211072263603263,
+      "grad_norm": 0.17164403200149536,
+      "learning_rate": 4.678900012305733e-06,
+      "loss": 0.9601,
+      "step": 44360
+    },
+    {
+      "epoch": 0.3211796130209125,
+      "grad_norm": 0.1575680375099182,
+      "learning_rate": 4.678827625645147e-06,
+      "loss": 0.9672,
+      "step": 44370
+    },
+    {
+      "epoch": 0.3212519996814987,
+      "grad_norm": 0.15979178249835968,
+      "learning_rate": 4.67875523898456e-06,
+      "loss": 0.9615,
+      "step": 44380
+    },
+    {
+      "epoch": 0.3213243863420849,
+      "grad_norm": 0.19064490497112274,
+      "learning_rate": 4.678682852323974e-06,
+      "loss": 0.9672,
+      "step": 44390
+    },
+    {
+      "epoch": 0.32139677300267105,
+      "grad_norm": 0.16969703137874603,
+      "learning_rate": 4.678610465663388e-06,
+      "loss": 0.9418,
+      "step": 44400
+    },
+    {
+      "epoch": 0.32146915966325723,
+      "grad_norm": 0.1616748571395874,
+      "learning_rate": 4.678538079002802e-06,
+      "loss": 0.9613,
+      "step": 44410
+    },
+    {
+      "epoch": 0.32154154632384346,
+      "grad_norm": 0.15733925998210907,
+      "learning_rate": 4.6784656923422155e-06,
+      "loss": 0.9679,
+      "step": 44420
+    },
+    {
+      "epoch": 0.32161393298442964,
+      "grad_norm": 0.1610860526561737,
+      "learning_rate": 4.678393305681629e-06,
+      "loss": 0.9692,
+      "step": 44430
+    },
+    {
+      "epoch": 0.3216863196450158,
+      "grad_norm": 0.16648943722248077,
+      "learning_rate": 4.678320919021044e-06,
+      "loss": 0.9565,
+      "step": 44440
+    },
+    {
+      "epoch": 0.321758706305602,
+      "grad_norm": 0.186976358294487,
+      "learning_rate": 4.678248532360457e-06,
+      "loss": 0.9474,
+      "step": 44450
+    },
+    {
+      "epoch": 0.32183109296618817,
+      "grad_norm": 0.1633586436510086,
+      "learning_rate": 4.678176145699871e-06,
+      "loss": 0.9468,
+      "step": 44460
+    },
+    {
+      "epoch": 0.3219034796267744,
+      "grad_norm": 0.197808638215065,
+      "learning_rate": 4.6781037590392845e-06,
+      "loss": 0.9601,
+      "step": 44470
+    },
+    {
+      "epoch": 0.3219758662873606,
+      "grad_norm": 0.15820321440696716,
+      "learning_rate": 4.678031372378699e-06,
+      "loss": 0.9537,
+      "step": 44480
+    },
+    {
+      "epoch": 0.32204825294794676,
+      "grad_norm": 0.1628575474023819,
+      "learning_rate": 4.6779589857181125e-06,
+      "loss": 0.9503,
+      "step": 44490
+    },
+    {
+      "epoch": 0.32212063960853293,
+      "grad_norm": 0.1750411093235016,
+      "learning_rate": 4.677886599057526e-06,
+      "loss": 0.9594,
+      "step": 44500
+    },
+    {
+      "epoch": 0.3221930262691191,
+      "grad_norm": 0.17343439161777496,
+      "learning_rate": 4.67781421239694e-06,
+      "loss": 0.9584,
+      "step": 44510
+    },
+    {
+      "epoch": 0.32226541292970534,
+      "grad_norm": 0.16742408275604248,
+      "learning_rate": 4.677741825736354e-06,
+      "loss": 0.9638,
+      "step": 44520
+    },
+    {
+      "epoch": 0.3223377995902915,
+      "grad_norm": 0.1644875556230545,
+      "learning_rate": 4.677669439075768e-06,
+      "loss": 0.9536,
+      "step": 44530
+    },
+    {
+      "epoch": 0.3224101862508777,
+      "grad_norm": 0.16791434586048126,
+      "learning_rate": 4.6775970524151815e-06,
+      "loss": 0.9669,
+      "step": 44540
+    },
+    {
+      "epoch": 0.32248257291146387,
+      "grad_norm": 0.16774770617485046,
+      "learning_rate": 4.677524665754595e-06,
+      "loss": 0.9569,
+      "step": 44550
+    },
+    {
+      "epoch": 0.32255495957205005,
+      "grad_norm": 0.15673328936100006,
+      "learning_rate": 4.677452279094009e-06,
+      "loss": 0.9394,
+      "step": 44560
+    },
+    {
+      "epoch": 0.3226273462326362,
+      "grad_norm": 0.17572014033794403,
+      "learning_rate": 4.677379892433423e-06,
+      "loss": 0.957,
+      "step": 44570
+    },
+    {
+      "epoch": 0.32269973289322246,
+      "grad_norm": 0.18698441982269287,
+      "learning_rate": 4.677307505772837e-06,
+      "loss": 0.9637,
+      "step": 44580
+    },
+    {
+      "epoch": 0.32277211955380863,
+      "grad_norm": 0.17538724839687347,
+      "learning_rate": 4.67723511911225e-06,
+      "loss": 0.968,
+      "step": 44590
+    },
+    {
+      "epoch": 0.3228445062143948,
+      "grad_norm": 0.16755321621894836,
+      "learning_rate": 4.677162732451664e-06,
+      "loss": 0.9542,
+      "step": 44600
+    },
+    {
+      "epoch": 0.322916892874981,
+      "grad_norm": 0.1889658123254776,
+      "learning_rate": 4.6770903457910785e-06,
+      "loss": 0.9668,
+      "step": 44610
+    },
+    {
+      "epoch": 0.32298927953556716,
+      "grad_norm": 0.15155941247940063,
+      "learning_rate": 4.677017959130492e-06,
+      "loss": 0.963,
+      "step": 44620
+    },
+    {
+      "epoch": 0.3230616661961534,
+      "grad_norm": 0.17296727001667023,
+      "learning_rate": 4.676945572469906e-06,
+      "loss": 0.9658,
+      "step": 44630
+    },
+    {
+      "epoch": 0.3231340528567396,
+      "grad_norm": 0.15777067840099335,
+      "learning_rate": 4.676873185809319e-06,
+      "loss": 0.956,
+      "step": 44640
+    },
+    {
+      "epoch": 0.32320643951732575,
+      "grad_norm": 0.1767292618751526,
+      "learning_rate": 4.676800799148734e-06,
+      "loss": 0.9635,
+      "step": 44650
+    },
+    {
+      "epoch": 0.3232788261779119,
+      "grad_norm": 0.1630258858203888,
+      "learning_rate": 4.6767284124881466e-06,
+      "loss": 0.9621,
+      "step": 44660
+    },
+    {
+      "epoch": 0.3233512128384981,
+      "grad_norm": 0.16284741461277008,
+      "learning_rate": 4.676656025827561e-06,
+      "loss": 0.9644,
+      "step": 44670
+    },
+    {
+      "epoch": 0.32342359949908434,
+      "grad_norm": 0.16810114681720734,
+      "learning_rate": 4.676583639166975e-06,
+      "loss": 0.9712,
+      "step": 44680
+    },
+    {
+      "epoch": 0.3234959861596705,
+      "grad_norm": 0.16822989284992218,
+      "learning_rate": 4.676511252506388e-06,
+      "loss": 0.9616,
+      "step": 44690
+    },
+    {
+      "epoch": 0.3235683728202567,
+      "grad_norm": 0.1614549160003662,
+      "learning_rate": 4.676438865845802e-06,
+      "loss": 0.9493,
+      "step": 44700
+    },
+    {
+      "epoch": 0.32364075948084287,
+      "grad_norm": 0.1698823720216751,
+      "learning_rate": 4.676366479185216e-06,
+      "loss": 0.9479,
+      "step": 44710
+    },
+    {
+      "epoch": 0.32371314614142904,
+      "grad_norm": 0.16118445992469788,
+      "learning_rate": 4.67629409252463e-06,
+      "loss": 0.9754,
+      "step": 44720
+    },
+    {
+      "epoch": 0.3237855328020152,
+      "grad_norm": 0.16466830670833588,
+      "learning_rate": 4.6762217058640436e-06,
+      "loss": 0.978,
+      "step": 44730
+    },
+    {
+      "epoch": 0.32385791946260145,
+      "grad_norm": 0.1972661316394806,
+      "learning_rate": 4.676149319203457e-06,
+      "loss": 0.9743,
+      "step": 44740
+    },
+    {
+      "epoch": 0.32393030612318763,
+      "grad_norm": 0.17572824656963348,
+      "learning_rate": 4.676076932542871e-06,
+      "loss": 0.9744,
+      "step": 44750
+    },
+    {
+      "epoch": 0.3240026927837738,
+      "grad_norm": 0.18128469586372375,
+      "learning_rate": 4.676004545882285e-06,
+      "loss": 0.9504,
+      "step": 44760
+    },
+    {
+      "epoch": 0.32407507944436,
+      "grad_norm": 0.2079382985830307,
+      "learning_rate": 4.675932159221699e-06,
+      "loss": 0.9561,
+      "step": 44770
+    },
+    {
+      "epoch": 0.32414746610494616,
+      "grad_norm": 0.15759167075157166,
+      "learning_rate": 4.6758597725611125e-06,
+      "loss": 0.9527,
+      "step": 44780
+    },
+    {
+      "epoch": 0.3242198527655324,
+      "grad_norm": 0.1726982742547989,
+      "learning_rate": 4.675787385900526e-06,
+      "loss": 0.9738,
+      "step": 44790
+    },
+    {
+      "epoch": 0.32429223942611857,
+      "grad_norm": 0.2257843315601349,
+      "learning_rate": 4.6757149992399406e-06,
+      "loss": 0.9718,
+      "step": 44800
+    },
+    {
+      "epoch": 0.32436462608670474,
+      "grad_norm": 0.16954675316810608,
+      "learning_rate": 4.675642612579354e-06,
+      "loss": 0.9619,
+      "step": 44810
+    },
+    {
+      "epoch": 0.3244370127472909,
+      "grad_norm": 0.17804615199565887,
+      "learning_rate": 4.675570225918768e-06,
+      "loss": 0.9657,
+      "step": 44820
+    },
+    {
+      "epoch": 0.3245093994078771,
+      "grad_norm": 0.18081621825695038,
+      "learning_rate": 4.675497839258181e-06,
+      "loss": 0.9528,
+      "step": 44830
+    },
+    {
+      "epoch": 0.32458178606846333,
+      "grad_norm": 0.19913703203201294,
+      "learning_rate": 4.675425452597596e-06,
+      "loss": 0.9552,
+      "step": 44840
+    },
+    {
+      "epoch": 0.3246541727290495,
+      "grad_norm": 0.15979525446891785,
+      "learning_rate": 4.6753530659370095e-06,
+      "loss": 0.9465,
+      "step": 44850
+    },
+    {
+      "epoch": 0.3247265593896357,
+      "grad_norm": 0.1660212129354477,
+      "learning_rate": 4.675280679276423e-06,
+      "loss": 0.9517,
+      "step": 44860
+    },
+    {
+      "epoch": 0.32479894605022186,
+      "grad_norm": 0.1735403835773468,
+      "learning_rate": 4.675208292615837e-06,
+      "loss": 0.9641,
+      "step": 44870
+    },
+    {
+      "epoch": 0.32487133271080804,
+      "grad_norm": 0.15234126150608063,
+      "learning_rate": 4.675135905955251e-06,
+      "loss": 0.9705,
+      "step": 44880
+    },
+    {
+      "epoch": 0.3249437193713942,
+      "grad_norm": 0.16520267724990845,
+      "learning_rate": 4.675063519294665e-06,
+      "loss": 0.9587,
+      "step": 44890
+    },
+    {
+      "epoch": 0.32501610603198045,
+      "grad_norm": 0.1735251098871231,
+      "learning_rate": 4.6749911326340784e-06,
+      "loss": 0.9668,
+      "step": 44900
+    },
+    {
+      "epoch": 0.3250884926925666,
+      "grad_norm": 0.17319133877754211,
+      "learning_rate": 4.674918745973492e-06,
+      "loss": 0.9523,
+      "step": 44910
+    },
+    {
+      "epoch": 0.3251608793531528,
+      "grad_norm": 0.17401674389839172,
+      "learning_rate": 4.6748463593129065e-06,
+      "loss": 0.9708,
+      "step": 44920
+    },
+    {
+      "epoch": 0.325233266013739,
+      "grad_norm": 0.16565395891666412,
+      "learning_rate": 4.67477397265232e-06,
+      "loss": 0.9531,
+      "step": 44930
+    },
+    {
+      "epoch": 0.32530565267432515,
+      "grad_norm": 0.18349653482437134,
+      "learning_rate": 4.674701585991734e-06,
+      "loss": 0.9586,
+      "step": 44940
+    },
+    {
+      "epoch": 0.3253780393349114,
+      "grad_norm": 0.15808264911174774,
+      "learning_rate": 4.674629199331147e-06,
+      "loss": 0.9453,
+      "step": 44950
+    },
+    {
+      "epoch": 0.32545042599549756,
+      "grad_norm": 0.16079016029834747,
+      "learning_rate": 4.674556812670562e-06,
+      "loss": 0.9568,
+      "step": 44960
+    },
+    {
+      "epoch": 0.32552281265608374,
+      "grad_norm": 0.1582205444574356,
+      "learning_rate": 4.6744844260099754e-06,
+      "loss": 0.9383,
+      "step": 44970
+    },
+    {
+      "epoch": 0.3255951993166699,
+      "grad_norm": 0.1639067679643631,
+      "learning_rate": 4.674412039349389e-06,
+      "loss": 0.9554,
+      "step": 44980
+    },
+    {
+      "epoch": 0.3256675859772561,
+      "grad_norm": 0.17060793936252594,
+      "learning_rate": 4.674339652688803e-06,
+      "loss": 0.9534,
+      "step": 44990
+    },
+    {
+      "epoch": 0.3257399726378423,
+      "grad_norm": 0.18423733115196228,
+      "learning_rate": 4.674267266028217e-06,
+      "loss": 0.9478,
+      "step": 45000
+    },
+    {
+      "epoch": 0.3258123592984285,
+      "grad_norm": 0.18846924602985382,
+      "learning_rate": 4.674194879367631e-06,
+      "loss": 0.9734,
+      "step": 45010
+    },
+    {
+      "epoch": 0.3258847459590147,
+      "grad_norm": 0.16957645118236542,
+      "learning_rate": 4.674122492707044e-06,
+      "loss": 0.9489,
+      "step": 45020
+    },
+    {
+      "epoch": 0.32595713261960085,
+      "grad_norm": 0.15767230093479156,
+      "learning_rate": 4.674050106046458e-06,
+      "loss": 0.9565,
+      "step": 45030
+    },
+    {
+      "epoch": 0.32602951928018703,
+      "grad_norm": 0.15942777693271637,
+      "learning_rate": 4.6739777193858724e-06,
+      "loss": 0.9586,
+      "step": 45040
+    },
+    {
+      "epoch": 0.3261019059407732,
+      "grad_norm": 0.16869819164276123,
+      "learning_rate": 4.673905332725286e-06,
+      "loss": 0.9599,
+      "step": 45050
+    },
+    {
+      "epoch": 0.32617429260135944,
+      "grad_norm": 0.17203491926193237,
+      "learning_rate": 4.6738329460647e-06,
+      "loss": 0.9671,
+      "step": 45060
+    },
+    {
+      "epoch": 0.3262466792619456,
+      "grad_norm": 0.15850797295570374,
+      "learning_rate": 4.673760559404113e-06,
+      "loss": 0.9588,
+      "step": 45070
+    },
+    {
+      "epoch": 0.3263190659225318,
+      "grad_norm": 0.16606780886650085,
+      "learning_rate": 4.673688172743528e-06,
+      "loss": 0.9549,
+      "step": 45080
+    },
+    {
+      "epoch": 0.32639145258311797,
+      "grad_norm": 0.1549416184425354,
+      "learning_rate": 4.673615786082941e-06,
+      "loss": 0.9558,
+      "step": 45090
+    },
+    {
+      "epoch": 0.32646383924370415,
+      "grad_norm": 0.16568376123905182,
+      "learning_rate": 4.673543399422355e-06,
+      "loss": 0.9557,
+      "step": 45100
+    },
+    {
+      "epoch": 0.3265362259042904,
+      "grad_norm": 0.17167599499225616,
+      "learning_rate": 4.673471012761769e-06,
+      "loss": 0.9581,
+      "step": 45110
+    },
+    {
+      "epoch": 0.32660861256487655,
+      "grad_norm": 0.1696111410856247,
+      "learning_rate": 4.673398626101183e-06,
+      "loss": 0.9523,
+      "step": 45120
+    },
+    {
+      "epoch": 0.32668099922546273,
+      "grad_norm": 0.15967723727226257,
+      "learning_rate": 4.673326239440597e-06,
+      "loss": 0.9456,
+      "step": 45130
+    },
+    {
+      "epoch": 0.3267533858860489,
+      "grad_norm": 0.1622992306947708,
+      "learning_rate": 4.67325385278001e-06,
+      "loss": 0.9549,
+      "step": 45140
+    },
+    {
+      "epoch": 0.3268257725466351,
+      "grad_norm": 0.17789630591869354,
+      "learning_rate": 4.673181466119424e-06,
+      "loss": 0.9494,
+      "step": 45150
+    },
+    {
+      "epoch": 0.3268981592072213,
+      "grad_norm": 0.16522136330604553,
+      "learning_rate": 4.673109079458838e-06,
+      "loss": 0.974,
+      "step": 45160
+    },
+    {
+      "epoch": 0.3269705458678075,
+      "grad_norm": 0.17646393179893494,
+      "learning_rate": 4.673036692798252e-06,
+      "loss": 0.9629,
+      "step": 45170
+    },
+    {
+      "epoch": 0.32704293252839367,
+      "grad_norm": 0.16956348717212677,
+      "learning_rate": 4.672964306137666e-06,
+      "loss": 0.96,
+      "step": 45180
+    },
+    {
+      "epoch": 0.32711531918897985,
+      "grad_norm": 0.166047602891922,
+      "learning_rate": 4.672891919477079e-06,
+      "loss": 0.9432,
+      "step": 45190
+    },
+    {
+      "epoch": 0.327187705849566,
+      "grad_norm": 0.1728239208459854,
+      "learning_rate": 4.672819532816493e-06,
+      "loss": 0.9631,
+      "step": 45200
+    },
+    {
+      "epoch": 0.32726009251015226,
+      "grad_norm": 0.16295363008975983,
+      "learning_rate": 4.6727471461559065e-06,
+      "loss": 0.9609,
+      "step": 45210
+    },
+    {
+      "epoch": 0.32733247917073843,
+      "grad_norm": 0.16612671315670013,
+      "learning_rate": 4.67267475949532e-06,
+      "loss": 0.9499,
+      "step": 45220
+    },
+    {
+      "epoch": 0.3274048658313246,
+      "grad_norm": 0.16281329095363617,
+      "learning_rate": 4.6726023728347345e-06,
+      "loss": 0.9565,
+      "step": 45230
+    },
+    {
+      "epoch": 0.3274772524919108,
+      "grad_norm": 0.15651023387908936,
+      "learning_rate": 4.672529986174148e-06,
+      "loss": 0.9623,
+      "step": 45240
+    },
+    {
+      "epoch": 0.32754963915249696,
+      "grad_norm": 0.21686317026615143,
+      "learning_rate": 4.672457599513562e-06,
+      "loss": 0.9478,
+      "step": 45250
+    },
+    {
+      "epoch": 0.32762202581308314,
+      "grad_norm": 0.16553688049316406,
+      "learning_rate": 4.672385212852975e-06,
+      "loss": 0.9527,
+      "step": 45260
+    },
+    {
+      "epoch": 0.32769441247366937,
+      "grad_norm": 0.16378672420978546,
+      "learning_rate": 4.67231282619239e-06,
+      "loss": 0.9625,
+      "step": 45270
+    },
+    {
+      "epoch": 0.32776679913425555,
+      "grad_norm": 0.16440148651599884,
+      "learning_rate": 4.6722404395318035e-06,
+      "loss": 0.9716,
+      "step": 45280
+    },
+    {
+      "epoch": 0.3278391857948417,
+      "grad_norm": 0.17290063202381134,
+      "learning_rate": 4.672168052871217e-06,
+      "loss": 0.9489,
+      "step": 45290
+    },
+    {
+      "epoch": 0.3279115724554279,
+      "grad_norm": 0.16323043406009674,
+      "learning_rate": 4.672095666210631e-06,
+      "loss": 0.9543,
+      "step": 45300
+    },
+    {
+      "epoch": 0.3279839591160141,
+      "grad_norm": 0.3352501690387726,
+      "learning_rate": 4.672023279550045e-06,
+      "loss": 0.964,
+      "step": 45310
+    },
+    {
+      "epoch": 0.3280563457766003,
+      "grad_norm": 0.16609349846839905,
+      "learning_rate": 4.671950892889459e-06,
+      "loss": 0.9649,
+      "step": 45320
+    },
+    {
+      "epoch": 0.3281287324371865,
+      "grad_norm": 0.17989186942577362,
+      "learning_rate": 4.671878506228872e-06,
+      "loss": 0.9558,
+      "step": 45330
+    },
+    {
+      "epoch": 0.32820111909777266,
+      "grad_norm": 0.1710348278284073,
+      "learning_rate": 4.671806119568286e-06,
+      "loss": 0.9666,
+      "step": 45340
+    },
+    {
+      "epoch": 0.32827350575835884,
+      "grad_norm": 0.16703477501869202,
+      "learning_rate": 4.6717337329077e-06,
+      "loss": 0.9709,
+      "step": 45350
+    },
+    {
+      "epoch": 0.328345892418945,
+      "grad_norm": 0.16510546207427979,
+      "learning_rate": 4.671661346247114e-06,
+      "loss": 0.9561,
+      "step": 45360
+    },
+    {
+      "epoch": 0.32841827907953125,
+      "grad_norm": 0.17808891832828522,
+      "learning_rate": 4.671588959586528e-06,
+      "loss": 0.9671,
+      "step": 45370
+    },
+    {
+      "epoch": 0.3284906657401174,
+      "grad_norm": 0.15966980159282684,
+      "learning_rate": 4.671516572925941e-06,
+      "loss": 0.9631,
+      "step": 45380
+    },
+    {
+      "epoch": 0.3285630524007036,
+      "grad_norm": 0.18822209537029266,
+      "learning_rate": 4.671444186265355e-06,
+      "loss": 0.9539,
+      "step": 45390
+    },
+    {
+      "epoch": 0.3286354390612898,
+      "grad_norm": 0.16146880388259888,
+      "learning_rate": 4.671371799604769e-06,
+      "loss": 0.9603,
+      "step": 45400
+    },
+    {
+      "epoch": 0.32870782572187596,
+      "grad_norm": 0.16964489221572876,
+      "learning_rate": 4.671299412944183e-06,
+      "loss": 0.9706,
+      "step": 45410
+    },
+    {
+      "epoch": 0.32878021238246213,
+      "grad_norm": 0.17912639677524567,
+      "learning_rate": 4.671227026283597e-06,
+      "loss": 0.9671,
+      "step": 45420
+    },
+    {
+      "epoch": 0.32885259904304837,
+      "grad_norm": 0.15870171785354614,
+      "learning_rate": 4.67115463962301e-06,
+      "loss": 0.9596,
+      "step": 45430
+    },
+    {
+      "epoch": 0.32892498570363454,
+      "grad_norm": 0.16293965280056,
+      "learning_rate": 4.671082252962425e-06,
+      "loss": 0.9536,
+      "step": 45440
+    },
+    {
+      "epoch": 0.3289973723642207,
+      "grad_norm": 0.1799435019493103,
+      "learning_rate": 4.671009866301838e-06,
+      "loss": 0.953,
+      "step": 45450
+    },
+    {
+      "epoch": 0.3290697590248069,
+      "grad_norm": 0.16505300998687744,
+      "learning_rate": 4.670937479641252e-06,
+      "loss": 0.9579,
+      "step": 45460
+    },
+    {
+      "epoch": 0.3291421456853931,
+      "grad_norm": 0.1620611697435379,
+      "learning_rate": 4.6708650929806656e-06,
+      "loss": 0.9364,
+      "step": 45470
+    },
+    {
+      "epoch": 0.3292145323459793,
+      "grad_norm": 0.16515015065670013,
+      "learning_rate": 4.67079270632008e-06,
+      "loss": 0.9711,
+      "step": 45480
+    },
+    {
+      "epoch": 0.3292869190065655,
+      "grad_norm": 0.16195544600486755,
+      "learning_rate": 4.670720319659494e-06,
+      "loss": 0.9485,
+      "step": 45490
+    },
+    {
+      "epoch": 0.32935930566715166,
+      "grad_norm": 0.1693519651889801,
+      "learning_rate": 4.670647932998907e-06,
+      "loss": 0.9392,
+      "step": 45500
+    },
+    {
+      "epoch": 0.32943169232773784,
+      "grad_norm": 0.16994911432266235,
+      "learning_rate": 4.670575546338321e-06,
+      "loss": 0.9574,
+      "step": 45510
+    },
+    {
+      "epoch": 0.329504078988324,
+      "grad_norm": 0.16897796094417572,
+      "learning_rate": 4.670503159677735e-06,
+      "loss": 0.9588,
+      "step": 45520
+    },
+    {
+      "epoch": 0.32957646564891024,
+      "grad_norm": 0.1709592044353485,
+      "learning_rate": 4.670430773017149e-06,
+      "loss": 0.9397,
+      "step": 45530
+    },
+    {
+      "epoch": 0.3296488523094964,
+      "grad_norm": 0.177882120013237,
+      "learning_rate": 4.6703583863565626e-06,
+      "loss": 0.9669,
+      "step": 45540
+    },
+    {
+      "epoch": 0.3297212389700826,
+      "grad_norm": 0.18497739732265472,
+      "learning_rate": 4.670285999695976e-06,
+      "loss": 0.9451,
+      "step": 45550
+    },
+    {
+      "epoch": 0.3297936256306688,
+      "grad_norm": 0.16033156216144562,
+      "learning_rate": 4.670213613035391e-06,
+      "loss": 0.9574,
+      "step": 45560
+    },
+    {
+      "epoch": 0.32986601229125495,
+      "grad_norm": 0.1641259789466858,
+      "learning_rate": 4.670141226374804e-06,
+      "loss": 0.94,
+      "step": 45570
+    },
+    {
+      "epoch": 0.3299383989518411,
+      "grad_norm": 0.1763799786567688,
+      "learning_rate": 4.670068839714218e-06,
+      "loss": 0.9508,
+      "step": 45580
+    },
+    {
+      "epoch": 0.33001078561242736,
+      "grad_norm": 0.17956474423408508,
+      "learning_rate": 4.6699964530536315e-06,
+      "loss": 0.9598,
+      "step": 45590
+    },
+    {
+      "epoch": 0.33008317227301354,
+      "grad_norm": 0.21450495719909668,
+      "learning_rate": 4.669924066393046e-06,
+      "loss": 0.9572,
+      "step": 45600
+    },
+    {
+      "epoch": 0.3301555589335997,
+      "grad_norm": 0.1846141368150711,
+      "learning_rate": 4.66985167973246e-06,
+      "loss": 0.968,
+      "step": 45610
+    },
+    {
+      "epoch": 0.3302279455941859,
+      "grad_norm": 0.1598512977361679,
+      "learning_rate": 4.669779293071873e-06,
+      "loss": 0.9591,
+      "step": 45620
+    },
+    {
+      "epoch": 0.33030033225477207,
+      "grad_norm": 0.15021023154258728,
+      "learning_rate": 4.669706906411287e-06,
+      "loss": 0.9389,
+      "step": 45630
+    },
+    {
+      "epoch": 0.3303727189153583,
+      "grad_norm": 0.16507980227470398,
+      "learning_rate": 4.669634519750701e-06,
+      "loss": 0.9638,
+      "step": 45640
+    },
+    {
+      "epoch": 0.3304451055759445,
+      "grad_norm": 0.16119886934757233,
+      "learning_rate": 4.669562133090115e-06,
+      "loss": 0.9521,
+      "step": 45650
+    },
+    {
+      "epoch": 0.33051749223653065,
+      "grad_norm": 0.19781506061553955,
+      "learning_rate": 4.6694897464295285e-06,
+      "loss": 0.9473,
+      "step": 45660
+    },
+    {
+      "epoch": 0.33058987889711683,
+      "grad_norm": 0.1651494801044464,
+      "learning_rate": 4.669417359768942e-06,
+      "loss": 0.9482,
+      "step": 45670
+    },
+    {
+      "epoch": 0.330662265557703,
+      "grad_norm": 0.16081689298152924,
+      "learning_rate": 4.669344973108357e-06,
+      "loss": 0.9678,
+      "step": 45680
+    },
+    {
+      "epoch": 0.33073465221828924,
+      "grad_norm": 0.1630435436964035,
+      "learning_rate": 4.66927258644777e-06,
+      "loss": 0.9627,
+      "step": 45690
+    },
+    {
+      "epoch": 0.3308070388788754,
+      "grad_norm": 0.2068457007408142,
+      "learning_rate": 4.669200199787184e-06,
+      "loss": 0.9521,
+      "step": 45700
+    },
+    {
+      "epoch": 0.3308794255394616,
+      "grad_norm": 0.168662890791893,
+      "learning_rate": 4.6691278131265974e-06,
+      "loss": 0.9453,
+      "step": 45710
+    },
+    {
+      "epoch": 0.33095181220004777,
+      "grad_norm": 0.18764568865299225,
+      "learning_rate": 4.669055426466011e-06,
+      "loss": 0.9483,
+      "step": 45720
+    },
+    {
+      "epoch": 0.33102419886063394,
+      "grad_norm": 0.1561020016670227,
+      "learning_rate": 4.668983039805425e-06,
+      "loss": 0.9565,
+      "step": 45730
+    },
+    {
+      "epoch": 0.3310965855212202,
+      "grad_norm": 0.22187693417072296,
+      "learning_rate": 4.668910653144838e-06,
+      "loss": 0.9594,
+      "step": 45740
+    },
+    {
+      "epoch": 0.33116897218180635,
+      "grad_norm": 0.16155502200126648,
+      "learning_rate": 4.668838266484253e-06,
+      "loss": 0.9664,
+      "step": 45750
+    },
+    {
+      "epoch": 0.33124135884239253,
+      "grad_norm": 0.1851801574230194,
+      "learning_rate": 4.668765879823666e-06,
+      "loss": 0.9517,
+      "step": 45760
+    },
+    {
+      "epoch": 0.3313137455029787,
+      "grad_norm": 0.16832536458969116,
+      "learning_rate": 4.66869349316308e-06,
+      "loss": 0.9532,
+      "step": 45770
+    },
+    {
+      "epoch": 0.3313861321635649,
+      "grad_norm": 0.17694327235221863,
+      "learning_rate": 4.668621106502494e-06,
+      "loss": 0.9466,
+      "step": 45780
+    },
+    {
+      "epoch": 0.33145851882415106,
+      "grad_norm": 0.16291755437850952,
+      "learning_rate": 4.668548719841908e-06,
+      "loss": 0.9544,
+      "step": 45790
+    },
+    {
+      "epoch": 0.3315309054847373,
+      "grad_norm": 0.1645163744688034,
+      "learning_rate": 4.668476333181322e-06,
+      "loss": 0.9519,
+      "step": 45800
+    },
+    {
+      "epoch": 0.33160329214532347,
+      "grad_norm": 0.1698862910270691,
+      "learning_rate": 4.668403946520735e-06,
+      "loss": 0.9583,
+      "step": 45810
+    },
+    {
+      "epoch": 0.33167567880590965,
+      "grad_norm": 0.16201475262641907,
+      "learning_rate": 4.668331559860149e-06,
+      "loss": 0.9518,
+      "step": 45820
+    },
+    {
+      "epoch": 0.3317480654664958,
+      "grad_norm": 0.2548742890357971,
+      "learning_rate": 4.668259173199563e-06,
+      "loss": 0.9571,
+      "step": 45830
+    },
+    {
+      "epoch": 0.331820452127082,
+      "grad_norm": 0.17558850347995758,
+      "learning_rate": 4.668186786538977e-06,
+      "loss": 0.9502,
+      "step": 45840
+    },
+    {
+      "epoch": 0.33189283878766823,
+      "grad_norm": 0.2049540877342224,
+      "learning_rate": 4.668114399878391e-06,
+      "loss": 0.9686,
+      "step": 45850
+    },
+    {
+      "epoch": 0.3319652254482544,
+      "grad_norm": 0.15438438951969147,
+      "learning_rate": 4.668042013217804e-06,
+      "loss": 0.9603,
+      "step": 45860
+    },
+    {
+      "epoch": 0.3320376121088406,
+      "grad_norm": 0.15590500831604004,
+      "learning_rate": 4.667969626557219e-06,
+      "loss": 0.9566,
+      "step": 45870
+    },
+    {
+      "epoch": 0.33210999876942676,
+      "grad_norm": 0.15624962747097015,
+      "learning_rate": 4.667897239896632e-06,
+      "loss": 0.9586,
+      "step": 45880
+    },
+    {
+      "epoch": 0.33218238543001294,
+      "grad_norm": 0.15918295085430145,
+      "learning_rate": 4.667824853236046e-06,
+      "loss": 0.951,
+      "step": 45890
+    },
+    {
+      "epoch": 0.33225477209059917,
+      "grad_norm": 0.16058406233787537,
+      "learning_rate": 4.6677524665754595e-06,
+      "loss": 0.958,
+      "step": 45900
+    },
+    {
+      "epoch": 0.33232715875118535,
+      "grad_norm": 0.1862640231847763,
+      "learning_rate": 4.667680079914874e-06,
+      "loss": 0.9443,
+      "step": 45910
+    },
+    {
+      "epoch": 0.3323995454117715,
+      "grad_norm": 0.1649313122034073,
+      "learning_rate": 4.667607693254288e-06,
+      "loss": 0.9523,
+      "step": 45920
+    },
+    {
+      "epoch": 0.3324719320723577,
+      "grad_norm": 0.17208698391914368,
+      "learning_rate": 4.667535306593701e-06,
+      "loss": 0.9776,
+      "step": 45930
+    },
+    {
+      "epoch": 0.3325443187329439,
+      "grad_norm": 0.19187243282794952,
+      "learning_rate": 4.667462919933115e-06,
+      "loss": 0.9472,
+      "step": 45940
+    },
+    {
+      "epoch": 0.33261670539353005,
+      "grad_norm": 0.17830708622932434,
+      "learning_rate": 4.667390533272529e-06,
+      "loss": 0.9662,
+      "step": 45950
+    },
+    {
+      "epoch": 0.3326890920541163,
+      "grad_norm": 0.16068105399608612,
+      "learning_rate": 4.667318146611943e-06,
+      "loss": 0.9581,
+      "step": 45960
+    },
+    {
+      "epoch": 0.33276147871470246,
+      "grad_norm": 0.15991735458374023,
+      "learning_rate": 4.6672457599513565e-06,
+      "loss": 0.9495,
+      "step": 45970
+    },
+    {
+      "epoch": 0.33283386537528864,
+      "grad_norm": 0.20799915492534637,
+      "learning_rate": 4.66717337329077e-06,
+      "loss": 0.9464,
+      "step": 45980
+    },
+    {
+      "epoch": 0.3329062520358748,
+      "grad_norm": 0.16113980114459991,
+      "learning_rate": 4.667100986630184e-06,
+      "loss": 0.9635,
+      "step": 45990
+    },
+    {
+      "epoch": 0.332978638696461,
+      "grad_norm": 0.20556610822677612,
+      "learning_rate": 4.667028599969598e-06,
+      "loss": 0.9714,
+      "step": 46000
+    },
+    {
+      "epoch": 0.3330510253570472,
+      "grad_norm": 0.17026638984680176,
+      "learning_rate": 4.666956213309012e-06,
+      "loss": 0.9572,
+      "step": 46010
+    },
+    {
+      "epoch": 0.3331234120176334,
+      "grad_norm": 0.1647808998823166,
+      "learning_rate": 4.6668838266484255e-06,
+      "loss": 0.9645,
+      "step": 46020
+    },
+    {
+      "epoch": 0.3331957986782196,
+      "grad_norm": 0.17708559334278107,
+      "learning_rate": 4.666811439987839e-06,
+      "loss": 0.9605,
+      "step": 46030
+    },
+    {
+      "epoch": 0.33326818533880576,
+      "grad_norm": 0.17343778908252716,
+      "learning_rate": 4.6667390533272535e-06,
+      "loss": 0.9694,
+      "step": 46040
+    },
+    {
+      "epoch": 0.33334057199939193,
+      "grad_norm": 0.18600593507289886,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.9542,
+      "step": 46050
+    },
+    {
+      "epoch": 0.33341295865997816,
+      "grad_norm": 0.1633295863866806,
+      "learning_rate": 4.666594280006081e-06,
+      "loss": 0.9564,
+      "step": 46060
+    },
+    {
+      "epoch": 0.33348534532056434,
+      "grad_norm": 0.16525378823280334,
+      "learning_rate": 4.666521893345494e-06,
+      "loss": 0.9482,
+      "step": 46070
+    },
+    {
+      "epoch": 0.3335577319811505,
+      "grad_norm": 0.16414770483970642,
+      "learning_rate": 4.666449506684909e-06,
+      "loss": 0.9537,
+      "step": 46080
+    },
+    {
+      "epoch": 0.3336301186417367,
+      "grad_norm": 0.1624254733324051,
+      "learning_rate": 4.6663771200243225e-06,
+      "loss": 0.9482,
+      "step": 46090
+    },
+    {
+      "epoch": 0.33370250530232287,
+      "grad_norm": 0.18863092362880707,
+      "learning_rate": 4.666304733363736e-06,
+      "loss": 0.9574,
+      "step": 46100
+    },
+    {
+      "epoch": 0.33377489196290905,
+      "grad_norm": 0.16010300815105438,
+      "learning_rate": 4.66623234670315e-06,
+      "loss": 0.9503,
+      "step": 46110
+    },
+    {
+      "epoch": 0.3338472786234953,
+      "grad_norm": 0.16204185783863068,
+      "learning_rate": 4.666159960042564e-06,
+      "loss": 0.9577,
+      "step": 46120
+    },
+    {
+      "epoch": 0.33391966528408146,
+      "grad_norm": 0.16618123650550842,
+      "learning_rate": 4.666087573381978e-06,
+      "loss": 0.9709,
+      "step": 46130
+    },
+    {
+      "epoch": 0.33399205194466763,
+      "grad_norm": 0.1760234236717224,
+      "learning_rate": 4.666015186721391e-06,
+      "loss": 0.9694,
+      "step": 46140
+    },
+    {
+      "epoch": 0.3340644386052538,
+      "grad_norm": 0.26092368364334106,
+      "learning_rate": 4.665942800060805e-06,
+      "loss": 0.9568,
+      "step": 46150
+    },
+    {
+      "epoch": 0.33413682526584,
+      "grad_norm": 0.17464442551136017,
+      "learning_rate": 4.6658704134002195e-06,
+      "loss": 0.9586,
+      "step": 46160
+    },
+    {
+      "epoch": 0.3342092119264262,
+      "grad_norm": 0.15855874121189117,
+      "learning_rate": 4.665798026739633e-06,
+      "loss": 0.9455,
+      "step": 46170
+    },
+    {
+      "epoch": 0.3342815985870124,
+      "grad_norm": 0.1912544220685959,
+      "learning_rate": 4.665725640079047e-06,
+      "loss": 0.9646,
+      "step": 46180
+    },
+    {
+      "epoch": 0.3343539852475986,
+      "grad_norm": 0.1546599417924881,
+      "learning_rate": 4.66565325341846e-06,
+      "loss": 0.9577,
+      "step": 46190
+    },
+    {
+      "epoch": 0.33442637190818475,
+      "grad_norm": 0.17187544703483582,
+      "learning_rate": 4.665580866757875e-06,
+      "loss": 0.9353,
+      "step": 46200
+    },
+    {
+      "epoch": 0.3344987585687709,
+      "grad_norm": 0.17121565341949463,
+      "learning_rate": 4.665508480097288e-06,
+      "loss": 0.9612,
+      "step": 46210
+    },
+    {
+      "epoch": 0.33457114522935716,
+      "grad_norm": 0.15924374759197235,
+      "learning_rate": 4.665436093436702e-06,
+      "loss": 0.9596,
+      "step": 46220
+    },
+    {
+      "epoch": 0.33464353188994334,
+      "grad_norm": 0.19250038266181946,
+      "learning_rate": 4.665363706776116e-06,
+      "loss": 0.9633,
+      "step": 46230
+    },
+    {
+      "epoch": 0.3347159185505295,
+      "grad_norm": 0.17079396545886993,
+      "learning_rate": 4.66529132011553e-06,
+      "loss": 0.9491,
+      "step": 46240
+    },
+    {
+      "epoch": 0.3347883052111157,
+      "grad_norm": 0.19403786957263947,
+      "learning_rate": 4.665218933454943e-06,
+      "loss": 0.9462,
+      "step": 46250
+    },
+    {
+      "epoch": 0.33486069187170187,
+      "grad_norm": 0.18236984312534332,
+      "learning_rate": 4.6651465467943565e-06,
+      "loss": 0.9459,
+      "step": 46260
+    },
+    {
+      "epoch": 0.3349330785322881,
+      "grad_norm": 0.14901652932167053,
+      "learning_rate": 4.665074160133771e-06,
+      "loss": 0.9659,
+      "step": 46270
+    },
+    {
+      "epoch": 0.3350054651928743,
+      "grad_norm": 0.17317718267440796,
+      "learning_rate": 4.6650017734731846e-06,
+      "loss": 0.9512,
+      "step": 46280
+    },
+    {
+      "epoch": 0.33507785185346045,
+      "grad_norm": 0.16543887555599213,
+      "learning_rate": 4.664929386812598e-06,
+      "loss": 0.9405,
+      "step": 46290
+    },
+    {
+      "epoch": 0.33515023851404663,
+      "grad_norm": 0.16926343739032745,
+      "learning_rate": 4.664857000152012e-06,
+      "loss": 0.9575,
+      "step": 46300
+    },
+    {
+      "epoch": 0.3352226251746328,
+      "grad_norm": 0.16885744035243988,
+      "learning_rate": 4.664784613491426e-06,
+      "loss": 0.9496,
+      "step": 46310
+    },
+    {
+      "epoch": 0.335295011835219,
+      "grad_norm": 0.15920351445674896,
+      "learning_rate": 4.66471222683084e-06,
+      "loss": 0.9579,
+      "step": 46320
+    },
+    {
+      "epoch": 0.3353673984958052,
+      "grad_norm": 0.15638776123523712,
+      "learning_rate": 4.6646398401702535e-06,
+      "loss": 0.9599,
+      "step": 46330
+    },
+    {
+      "epoch": 0.3354397851563914,
+      "grad_norm": 0.15919549763202667,
+      "learning_rate": 4.664567453509667e-06,
+      "loss": 0.953,
+      "step": 46340
+    },
+    {
+      "epoch": 0.33551217181697757,
+      "grad_norm": 0.1604597270488739,
+      "learning_rate": 4.664495066849082e-06,
+      "loss": 0.9597,
+      "step": 46350
+    },
+    {
+      "epoch": 0.33558455847756374,
+      "grad_norm": 0.19076739251613617,
+      "learning_rate": 4.664422680188495e-06,
+      "loss": 0.9795,
+      "step": 46360
+    },
+    {
+      "epoch": 0.3356569451381499,
+      "grad_norm": 0.15637139976024628,
+      "learning_rate": 4.664350293527909e-06,
+      "loss": 0.9549,
+      "step": 46370
+    },
+    {
+      "epoch": 0.33572933179873615,
+      "grad_norm": 0.17368392646312714,
+      "learning_rate": 4.6642779068673224e-06,
+      "loss": 0.9644,
+      "step": 46380
+    },
+    {
+      "epoch": 0.33580171845932233,
+      "grad_norm": 0.1643470972776413,
+      "learning_rate": 4.664205520206737e-06,
+      "loss": 0.9517,
+      "step": 46390
+    },
+    {
+      "epoch": 0.3358741051199085,
+      "grad_norm": 0.17304439842700958,
+      "learning_rate": 4.6641331335461505e-06,
+      "loss": 0.9686,
+      "step": 46400
+    },
+    {
+      "epoch": 0.3359464917804947,
+      "grad_norm": 0.15978851914405823,
+      "learning_rate": 4.664060746885564e-06,
+      "loss": 0.9658,
+      "step": 46410
+    },
+    {
+      "epoch": 0.33601887844108086,
+      "grad_norm": 0.16269192099571228,
+      "learning_rate": 4.663988360224978e-06,
+      "loss": 0.9419,
+      "step": 46420
+    },
+    {
+      "epoch": 0.3360912651016671,
+      "grad_norm": 0.16838060319423676,
+      "learning_rate": 4.663915973564392e-06,
+      "loss": 0.9444,
+      "step": 46430
+    },
+    {
+      "epoch": 0.33616365176225327,
+      "grad_norm": 0.1612749993801117,
+      "learning_rate": 4.663843586903806e-06,
+      "loss": 0.9599,
+      "step": 46440
+    },
+    {
+      "epoch": 0.33623603842283944,
+      "grad_norm": 0.1748184859752655,
+      "learning_rate": 4.6637712002432194e-06,
+      "loss": 0.9482,
+      "step": 46450
+    },
+    {
+      "epoch": 0.3363084250834256,
+      "grad_norm": 0.1573769748210907,
+      "learning_rate": 4.663698813582633e-06,
+      "loss": 0.9552,
+      "step": 46460
+    },
+    {
+      "epoch": 0.3363808117440118,
+      "grad_norm": 0.16278138756752014,
+      "learning_rate": 4.6636264269220475e-06,
+      "loss": 0.9531,
+      "step": 46470
+    },
+    {
+      "epoch": 0.336453198404598,
+      "grad_norm": 0.19514703750610352,
+      "learning_rate": 4.663554040261461e-06,
+      "loss": 0.9648,
+      "step": 46480
+    },
+    {
+      "epoch": 0.3365255850651842,
+      "grad_norm": 0.15883156657218933,
+      "learning_rate": 4.663481653600875e-06,
+      "loss": 0.9489,
+      "step": 46490
+    },
+    {
+      "epoch": 0.3365979717257704,
+      "grad_norm": 0.20735056698322296,
+      "learning_rate": 4.663409266940288e-06,
+      "loss": 0.9552,
+      "step": 46500
+    },
+    {
+      "epoch": 0.33667035838635656,
+      "grad_norm": 0.15435609221458435,
+      "learning_rate": 4.663336880279703e-06,
+      "loss": 0.9573,
+      "step": 46510
+    },
+    {
+      "epoch": 0.33674274504694274,
+      "grad_norm": 0.15631069242954254,
+      "learning_rate": 4.6632644936191164e-06,
+      "loss": 0.9501,
+      "step": 46520
+    },
+    {
+      "epoch": 0.3368151317075289,
+      "grad_norm": 0.17219269275665283,
+      "learning_rate": 4.66319210695853e-06,
+      "loss": 0.9483,
+      "step": 46530
+    },
+    {
+      "epoch": 0.33688751836811515,
+      "grad_norm": 0.19126524031162262,
+      "learning_rate": 4.663119720297944e-06,
+      "loss": 0.9449,
+      "step": 46540
+    },
+    {
+      "epoch": 0.3369599050287013,
+      "grad_norm": 0.16317123174667358,
+      "learning_rate": 4.663047333637358e-06,
+      "loss": 0.9521,
+      "step": 46550
+    },
+    {
+      "epoch": 0.3370322916892875,
+      "grad_norm": 0.15335845947265625,
+      "learning_rate": 4.662974946976772e-06,
+      "loss": 0.9538,
+      "step": 46560
+    },
+    {
+      "epoch": 0.3371046783498737,
+      "grad_norm": 0.1824522316455841,
+      "learning_rate": 4.662902560316185e-06,
+      "loss": 0.9511,
+      "step": 46570
+    },
+    {
+      "epoch": 0.33717706501045985,
+      "grad_norm": 0.16539111733436584,
+      "learning_rate": 4.662830173655599e-06,
+      "loss": 0.9501,
+      "step": 46580
+    },
+    {
+      "epoch": 0.3372494516710461,
+      "grad_norm": 0.16969068348407745,
+      "learning_rate": 4.6627577869950134e-06,
+      "loss": 0.9363,
+      "step": 46590
+    },
+    {
+      "epoch": 0.33732183833163226,
+      "grad_norm": 0.15810436010360718,
+      "learning_rate": 4.662685400334427e-06,
+      "loss": 0.9563,
+      "step": 46600
+    },
+    {
+      "epoch": 0.33739422499221844,
+      "grad_norm": 0.1647157222032547,
+      "learning_rate": 4.662613013673841e-06,
+      "loss": 0.9552,
+      "step": 46610
+    },
+    {
+      "epoch": 0.3374666116528046,
+      "grad_norm": 0.1537630707025528,
+      "learning_rate": 4.662540627013254e-06,
+      "loss": 0.956,
+      "step": 46620
+    },
+    {
+      "epoch": 0.3375389983133908,
+      "grad_norm": 0.16150793433189392,
+      "learning_rate": 4.662468240352668e-06,
+      "loss": 0.9567,
+      "step": 46630
+    },
+    {
+      "epoch": 0.33761138497397697,
+      "grad_norm": 0.17372149229049683,
+      "learning_rate": 4.662395853692082e-06,
+      "loss": 0.9595,
+      "step": 46640
+    },
+    {
+      "epoch": 0.3376837716345632,
+      "grad_norm": 0.18082787096500397,
+      "learning_rate": 4.662323467031496e-06,
+      "loss": 0.9577,
+      "step": 46650
+    },
+    {
+      "epoch": 0.3377561582951494,
+      "grad_norm": 0.17019541561603546,
+      "learning_rate": 4.66225108037091e-06,
+      "loss": 0.9532,
+      "step": 46660
+    },
+    {
+      "epoch": 0.33782854495573555,
+      "grad_norm": 0.15364877879619598,
+      "learning_rate": 4.662178693710323e-06,
+      "loss": 0.9584,
+      "step": 46670
+    },
+    {
+      "epoch": 0.33790093161632173,
+      "grad_norm": 0.15855498611927032,
+      "learning_rate": 4.662106307049738e-06,
+      "loss": 0.9702,
+      "step": 46680
+    },
+    {
+      "epoch": 0.3379733182769079,
+      "grad_norm": 0.23772907257080078,
+      "learning_rate": 4.662033920389151e-06,
+      "loss": 0.9503,
+      "step": 46690
+    },
+    {
+      "epoch": 0.33804570493749414,
+      "grad_norm": 0.16535037755966187,
+      "learning_rate": 4.661961533728565e-06,
+      "loss": 0.9469,
+      "step": 46700
+    },
+    {
+      "epoch": 0.3381180915980803,
+      "grad_norm": 0.17086680233478546,
+      "learning_rate": 4.6618891470679785e-06,
+      "loss": 0.9591,
+      "step": 46710
+    },
+    {
+      "epoch": 0.3381904782586665,
+      "grad_norm": 0.16868987679481506,
+      "learning_rate": 4.661816760407393e-06,
+      "loss": 0.9501,
+      "step": 46720
+    },
+    {
+      "epoch": 0.33826286491925267,
+      "grad_norm": 0.16112284362316132,
+      "learning_rate": 4.661744373746807e-06,
+      "loss": 0.9581,
+      "step": 46730
+    },
+    {
+      "epoch": 0.33833525157983885,
+      "grad_norm": 0.17157232761383057,
+      "learning_rate": 4.66167198708622e-06,
+      "loss": 0.9574,
+      "step": 46740
+    },
+    {
+      "epoch": 0.3384076382404251,
+      "grad_norm": 0.16588850319385529,
+      "learning_rate": 4.661599600425634e-06,
+      "loss": 0.9397,
+      "step": 46750
+    },
+    {
+      "epoch": 0.33848002490101126,
+      "grad_norm": 0.16449251770973206,
+      "learning_rate": 4.661527213765048e-06,
+      "loss": 0.9504,
+      "step": 46760
+    },
+    {
+      "epoch": 0.33855241156159743,
+      "grad_norm": 0.17265821993350983,
+      "learning_rate": 4.661454827104462e-06,
+      "loss": 0.9542,
+      "step": 46770
+    },
+    {
+      "epoch": 0.3386247982221836,
+      "grad_norm": 0.1667519360780716,
+      "learning_rate": 4.661382440443875e-06,
+      "loss": 0.9454,
+      "step": 46780
+    },
+    {
+      "epoch": 0.3386971848827698,
+      "grad_norm": 0.16387122869491577,
+      "learning_rate": 4.661310053783289e-06,
+      "loss": 0.9588,
+      "step": 46790
+    },
+    {
+      "epoch": 0.33876957154335596,
+      "grad_norm": 0.15918706357479095,
+      "learning_rate": 4.661237667122703e-06,
+      "loss": 0.9506,
+      "step": 46800
+    },
+    {
+      "epoch": 0.3388419582039422,
+      "grad_norm": 0.18269628286361694,
+      "learning_rate": 4.661165280462116e-06,
+      "loss": 0.9541,
+      "step": 46810
+    },
+    {
+      "epoch": 0.33891434486452837,
+      "grad_norm": 0.16732244193553925,
+      "learning_rate": 4.66109289380153e-06,
+      "loss": 0.9683,
+      "step": 46820
+    },
+    {
+      "epoch": 0.33898673152511455,
+      "grad_norm": 0.17757059633731842,
+      "learning_rate": 4.6610205071409445e-06,
+      "loss": 0.9507,
+      "step": 46830
+    },
+    {
+      "epoch": 0.3390591181857007,
+      "grad_norm": 0.16032300889492035,
+      "learning_rate": 4.660948120480358e-06,
+      "loss": 0.9469,
+      "step": 46840
+    },
+    {
+      "epoch": 0.3391315048462869,
+      "grad_norm": 0.1686878800392151,
+      "learning_rate": 4.660875733819772e-06,
+      "loss": 0.9517,
+      "step": 46850
+    },
+    {
+      "epoch": 0.33920389150687313,
+      "grad_norm": 0.24848999083042145,
+      "learning_rate": 4.660803347159185e-06,
+      "loss": 0.9647,
+      "step": 46860
+    },
+    {
+      "epoch": 0.3392762781674593,
+      "grad_norm": 0.16146300733089447,
+      "learning_rate": 4.6607309604986e-06,
+      "loss": 0.969,
+      "step": 46870
+    },
+    {
+      "epoch": 0.3393486648280455,
+      "grad_norm": 0.17109398543834686,
+      "learning_rate": 4.660658573838013e-06,
+      "loss": 0.9475,
+      "step": 46880
+    },
+    {
+      "epoch": 0.33942105148863166,
+      "grad_norm": 0.16845859587192535,
+      "learning_rate": 4.660586187177427e-06,
+      "loss": 0.9659,
+      "step": 46890
+    },
+    {
+      "epoch": 0.33949343814921784,
+      "grad_norm": 0.17198343575000763,
+      "learning_rate": 4.660513800516841e-06,
+      "loss": 0.951,
+      "step": 46900
+    },
+    {
+      "epoch": 0.3395658248098041,
+      "grad_norm": 0.1685718148946762,
+      "learning_rate": 4.660441413856255e-06,
+      "loss": 0.9511,
+      "step": 46910
+    },
+    {
+      "epoch": 0.33963821147039025,
+      "grad_norm": 0.17462746798992157,
+      "learning_rate": 4.660369027195669e-06,
+      "loss": 0.9518,
+      "step": 46920
+    },
+    {
+      "epoch": 0.3397105981309764,
+      "grad_norm": 0.15987993776798248,
+      "learning_rate": 4.660296640535082e-06,
+      "loss": 0.9521,
+      "step": 46930
+    },
+    {
+      "epoch": 0.3397829847915626,
+      "grad_norm": 0.16598616540431976,
+      "learning_rate": 4.660224253874496e-06,
+      "loss": 0.9543,
+      "step": 46940
+    },
+    {
+      "epoch": 0.3398553714521488,
+      "grad_norm": 0.16702087223529816,
+      "learning_rate": 4.66015186721391e-06,
+      "loss": 0.9475,
+      "step": 46950
+    },
+    {
+      "epoch": 0.339927758112735,
+      "grad_norm": 0.20819000899791718,
+      "learning_rate": 4.660079480553324e-06,
+      "loss": 0.9534,
+      "step": 46960
+    },
+    {
+      "epoch": 0.3400001447733212,
+      "grad_norm": 0.1900683492422104,
+      "learning_rate": 4.660007093892738e-06,
+      "loss": 0.9521,
+      "step": 46970
+    },
+    {
+      "epoch": 0.34007253143390737,
+      "grad_norm": 0.15491485595703125,
+      "learning_rate": 4.659934707232151e-06,
+      "loss": 0.9561,
+      "step": 46980
+    },
+    {
+      "epoch": 0.34014491809449354,
+      "grad_norm": 0.17071206867694855,
+      "learning_rate": 4.659862320571566e-06,
+      "loss": 0.9486,
+      "step": 46990
+    },
+    {
+      "epoch": 0.3402173047550797,
+      "grad_norm": 0.17024581134319305,
+      "learning_rate": 4.659789933910979e-06,
+      "loss": 0.9481,
+      "step": 47000
+    },
+    {
+      "epoch": 0.3402896914156659,
+      "grad_norm": 0.1717842072248459,
+      "learning_rate": 4.659717547250393e-06,
+      "loss": 0.9476,
+      "step": 47010
+    },
+    {
+      "epoch": 0.34036207807625213,
+      "grad_norm": 0.16945697367191315,
+      "learning_rate": 4.6596451605898066e-06,
+      "loss": 0.9598,
+      "step": 47020
+    },
+    {
+      "epoch": 0.3404344647368383,
+      "grad_norm": 0.1646365374326706,
+      "learning_rate": 4.659572773929221e-06,
+      "loss": 0.9447,
+      "step": 47030
+    },
+    {
+      "epoch": 0.3405068513974245,
+      "grad_norm": 0.1725275218486786,
+      "learning_rate": 4.659500387268635e-06,
+      "loss": 0.9625,
+      "step": 47040
+    },
+    {
+      "epoch": 0.34057923805801066,
+      "grad_norm": 0.15354198217391968,
+      "learning_rate": 4.659428000608048e-06,
+      "loss": 0.9535,
+      "step": 47050
+    },
+    {
+      "epoch": 0.34065162471859683,
+      "grad_norm": 0.15781398117542267,
+      "learning_rate": 4.659355613947462e-06,
+      "loss": 0.9584,
+      "step": 47060
+    },
+    {
+      "epoch": 0.34072401137918307,
+      "grad_norm": 0.22969284653663635,
+      "learning_rate": 4.659283227286876e-06,
+      "loss": 0.9555,
+      "step": 47070
+    },
+    {
+      "epoch": 0.34079639803976924,
+      "grad_norm": 0.16445045173168182,
+      "learning_rate": 4.65921084062629e-06,
+      "loss": 0.9676,
+      "step": 47080
+    },
+    {
+      "epoch": 0.3408687847003554,
+      "grad_norm": 0.16403111815452576,
+      "learning_rate": 4.6591384539657036e-06,
+      "loss": 0.9633,
+      "step": 47090
+    },
+    {
+      "epoch": 0.3409411713609416,
+      "grad_norm": 0.17574258148670197,
+      "learning_rate": 4.659066067305117e-06,
+      "loss": 0.94,
+      "step": 47100
+    },
+    {
+      "epoch": 0.3410135580215278,
+      "grad_norm": 0.16448186337947845,
+      "learning_rate": 4.658993680644532e-06,
+      "loss": 0.9693,
+      "step": 47110
+    },
+    {
+      "epoch": 0.341085944682114,
+      "grad_norm": 0.15972661972045898,
+      "learning_rate": 4.658921293983945e-06,
+      "loss": 0.9592,
+      "step": 47120
+    },
+    {
+      "epoch": 0.3411583313427002,
+      "grad_norm": 0.15439611673355103,
+      "learning_rate": 4.658848907323359e-06,
+      "loss": 0.9685,
+      "step": 47130
+    },
+    {
+      "epoch": 0.34123071800328636,
+      "grad_norm": 0.1713487207889557,
+      "learning_rate": 4.6587765206627725e-06,
+      "loss": 0.9415,
+      "step": 47140
+    },
+    {
+      "epoch": 0.34130310466387254,
+      "grad_norm": 0.15895883738994598,
+      "learning_rate": 4.658704134002187e-06,
+      "loss": 0.9427,
+      "step": 47150
+    },
+    {
+      "epoch": 0.3413754913244587,
+      "grad_norm": 0.15406803786754608,
+      "learning_rate": 4.658631747341601e-06,
+      "loss": 0.9464,
+      "step": 47160
+    },
+    {
+      "epoch": 0.3414478779850449,
+      "grad_norm": 0.1744023710489273,
+      "learning_rate": 4.658559360681014e-06,
+      "loss": 0.9538,
+      "step": 47170
+    },
+    {
+      "epoch": 0.3415202646456311,
+      "grad_norm": 0.16929666697978973,
+      "learning_rate": 4.658486974020428e-06,
+      "loss": 0.9457,
+      "step": 47180
+    },
+    {
+      "epoch": 0.3415926513062173,
+      "grad_norm": 0.1510300189256668,
+      "learning_rate": 4.658414587359842e-06,
+      "loss": 0.9548,
+      "step": 47190
+    },
+    {
+      "epoch": 0.3416650379668035,
+      "grad_norm": 0.1694670170545578,
+      "learning_rate": 4.658342200699256e-06,
+      "loss": 0.9565,
+      "step": 47200
+    },
+    {
+      "epoch": 0.34173742462738965,
+      "grad_norm": 0.1610107421875,
+      "learning_rate": 4.6582698140386695e-06,
+      "loss": 0.9569,
+      "step": 47210
+    },
+    {
+      "epoch": 0.34180981128797583,
+      "grad_norm": 0.1698133498430252,
+      "learning_rate": 4.658197427378083e-06,
+      "loss": 0.9563,
+      "step": 47220
+    },
+    {
+      "epoch": 0.34188219794856206,
+      "grad_norm": 0.17170600593090057,
+      "learning_rate": 4.658125040717497e-06,
+      "loss": 0.9611,
+      "step": 47230
+    },
+    {
+      "epoch": 0.34195458460914824,
+      "grad_norm": 0.1630273014307022,
+      "learning_rate": 4.658052654056911e-06,
+      "loss": 0.9517,
+      "step": 47240
+    },
+    {
+      "epoch": 0.3420269712697344,
+      "grad_norm": 0.1633155643939972,
+      "learning_rate": 4.657980267396325e-06,
+      "loss": 0.9635,
+      "step": 47250
+    },
+    {
+      "epoch": 0.3420993579303206,
+      "grad_norm": 0.1674155741930008,
+      "learning_rate": 4.6579078807357384e-06,
+      "loss": 0.944,
+      "step": 47260
+    },
+    {
+      "epoch": 0.34217174459090677,
+      "grad_norm": 0.18419358134269714,
+      "learning_rate": 4.657835494075152e-06,
+      "loss": 0.9554,
+      "step": 47270
+    },
+    {
+      "epoch": 0.342244131251493,
+      "grad_norm": 0.17149895429611206,
+      "learning_rate": 4.6577631074145665e-06,
+      "loss": 0.9533,
+      "step": 47280
+    },
+    {
+      "epoch": 0.3423165179120792,
+      "grad_norm": 0.15570290386676788,
+      "learning_rate": 4.65769072075398e-06,
+      "loss": 0.9394,
+      "step": 47290
+    },
+    {
+      "epoch": 0.34238890457266535,
+      "grad_norm": 0.1557348072528839,
+      "learning_rate": 4.657618334093394e-06,
+      "loss": 0.9467,
+      "step": 47300
+    },
+    {
+      "epoch": 0.34246129123325153,
+      "grad_norm": 0.16203594207763672,
+      "learning_rate": 4.657545947432807e-06,
+      "loss": 0.9445,
+      "step": 47310
+    },
+    {
+      "epoch": 0.3425336778938377,
+      "grad_norm": 0.20134076476097107,
+      "learning_rate": 4.657473560772221e-06,
+      "loss": 0.948,
+      "step": 47320
+    },
+    {
+      "epoch": 0.3426060645544239,
+      "grad_norm": 0.1600342094898224,
+      "learning_rate": 4.657401174111635e-06,
+      "loss": 0.9633,
+      "step": 47330
+    },
+    {
+      "epoch": 0.3426784512150101,
+      "grad_norm": 0.17118972539901733,
+      "learning_rate": 4.657328787451049e-06,
+      "loss": 0.9599,
+      "step": 47340
+    },
+    {
+      "epoch": 0.3427508378755963,
+      "grad_norm": 0.15433147549629211,
+      "learning_rate": 4.657256400790463e-06,
+      "loss": 0.9496,
+      "step": 47350
+    },
+    {
+      "epoch": 0.34282322453618247,
+      "grad_norm": 0.1639435738325119,
+      "learning_rate": 4.657184014129876e-06,
+      "loss": 0.9455,
+      "step": 47360
+    },
+    {
+      "epoch": 0.34289561119676865,
+      "grad_norm": 0.1590513288974762,
+      "learning_rate": 4.65711162746929e-06,
+      "loss": 0.968,
+      "step": 47370
+    },
+    {
+      "epoch": 0.3429679978573548,
+      "grad_norm": 0.15320120751857758,
+      "learning_rate": 4.657039240808704e-06,
+      "loss": 0.9613,
+      "step": 47380
+    },
+    {
+      "epoch": 0.34304038451794105,
+      "grad_norm": 0.1623704880475998,
+      "learning_rate": 4.656966854148118e-06,
+      "loss": 0.949,
+      "step": 47390
+    },
+    {
+      "epoch": 0.34311277117852723,
+      "grad_norm": 0.28263241052627563,
+      "learning_rate": 4.656894467487532e-06,
+      "loss": 0.9426,
+      "step": 47400
+    },
+    {
+      "epoch": 0.3431851578391134,
+      "grad_norm": 0.1681540608406067,
+      "learning_rate": 4.656822080826945e-06,
+      "loss": 0.9484,
+      "step": 47410
+    },
+    {
+      "epoch": 0.3432575444996996,
+      "grad_norm": 0.17275528609752655,
+      "learning_rate": 4.656749694166359e-06,
+      "loss": 0.9456,
+      "step": 47420
+    },
+    {
+      "epoch": 0.34332993116028576,
+      "grad_norm": 0.1764938235282898,
+      "learning_rate": 4.656677307505773e-06,
+      "loss": 0.9555,
+      "step": 47430
+    },
+    {
+      "epoch": 0.343402317820872,
+      "grad_norm": 0.18242251873016357,
+      "learning_rate": 4.656604920845187e-06,
+      "loss": 0.9464,
+      "step": 47440
+    },
+    {
+      "epoch": 0.34347470448145817,
+      "grad_norm": 0.1684602051973343,
+      "learning_rate": 4.6565325341846005e-06,
+      "loss": 0.9521,
+      "step": 47450
+    },
+    {
+      "epoch": 0.34354709114204435,
+      "grad_norm": 0.448024719953537,
+      "learning_rate": 4.656460147524014e-06,
+      "loss": 0.9466,
+      "step": 47460
+    },
+    {
+      "epoch": 0.3436194778026305,
+      "grad_norm": 0.20128124952316284,
+      "learning_rate": 4.656387760863429e-06,
+      "loss": 0.9388,
+      "step": 47470
+    },
+    {
+      "epoch": 0.3436918644632167,
+      "grad_norm": 0.18526539206504822,
+      "learning_rate": 4.656315374202842e-06,
+      "loss": 0.9565,
+      "step": 47480
+    },
+    {
+      "epoch": 0.34376425112380293,
+      "grad_norm": 0.1590823531150818,
+      "learning_rate": 4.656242987542256e-06,
+      "loss": 0.9447,
+      "step": 47490
+    },
+    {
+      "epoch": 0.3438366377843891,
+      "grad_norm": 0.21153424680233002,
+      "learning_rate": 4.6561706008816695e-06,
+      "loss": 0.9503,
+      "step": 47500
+    },
+    {
+      "epoch": 0.3439090244449753,
+      "grad_norm": 0.18983012437820435,
+      "learning_rate": 4.656098214221084e-06,
+      "loss": 0.9614,
+      "step": 47510
+    },
+    {
+      "epoch": 0.34398141110556146,
+      "grad_norm": 0.1883876472711563,
+      "learning_rate": 4.6560258275604975e-06,
+      "loss": 0.9572,
+      "step": 47520
+    },
+    {
+      "epoch": 0.34405379776614764,
+      "grad_norm": 0.15937159955501556,
+      "learning_rate": 4.655953440899911e-06,
+      "loss": 0.9579,
+      "step": 47530
+    },
+    {
+      "epoch": 0.3441261844267338,
+      "grad_norm": 0.17712438106536865,
+      "learning_rate": 4.655881054239325e-06,
+      "loss": 0.9693,
+      "step": 47540
+    },
+    {
+      "epoch": 0.34419857108732005,
+      "grad_norm": 0.17279085516929626,
+      "learning_rate": 4.655808667578739e-06,
+      "loss": 0.9572,
+      "step": 47550
+    },
+    {
+      "epoch": 0.3442709577479062,
+      "grad_norm": 0.16314037144184113,
+      "learning_rate": 4.655736280918153e-06,
+      "loss": 0.9495,
+      "step": 47560
+    },
+    {
+      "epoch": 0.3443433444084924,
+      "grad_norm": 0.1699339896440506,
+      "learning_rate": 4.6556638942575665e-06,
+      "loss": 0.9369,
+      "step": 47570
+    },
+    {
+      "epoch": 0.3444157310690786,
+      "grad_norm": 0.16625383496284485,
+      "learning_rate": 4.65559150759698e-06,
+      "loss": 0.9402,
+      "step": 47580
+    },
+    {
+      "epoch": 0.34448811772966476,
+      "grad_norm": 0.15905995666980743,
+      "learning_rate": 4.6555191209363946e-06,
+      "loss": 0.96,
+      "step": 47590
+    },
+    {
+      "epoch": 0.344560504390251,
+      "grad_norm": 0.1678093671798706,
+      "learning_rate": 4.655446734275808e-06,
+      "loss": 0.9517,
+      "step": 47600
+    },
+    {
+      "epoch": 0.34463289105083716,
+      "grad_norm": 0.1573289930820465,
+      "learning_rate": 4.655374347615222e-06,
+      "loss": 0.9514,
+      "step": 47610
+    },
+    {
+      "epoch": 0.34470527771142334,
+      "grad_norm": 0.1935252696275711,
+      "learning_rate": 4.655301960954635e-06,
+      "loss": 0.9646,
+      "step": 47620
+    },
+    {
+      "epoch": 0.3447776643720095,
+      "grad_norm": 0.16098780930042267,
+      "learning_rate": 4.65522957429405e-06,
+      "loss": 0.9546,
+      "step": 47630
+    },
+    {
+      "epoch": 0.3448500510325957,
+      "grad_norm": 0.1690308153629303,
+      "learning_rate": 4.6551571876334635e-06,
+      "loss": 0.9697,
+      "step": 47640
+    },
+    {
+      "epoch": 0.3449224376931819,
+      "grad_norm": 0.159522145986557,
+      "learning_rate": 4.655084800972877e-06,
+      "loss": 0.9472,
+      "step": 47650
+    },
+    {
+      "epoch": 0.3449948243537681,
+      "grad_norm": 0.2347215861082077,
+      "learning_rate": 4.655012414312291e-06,
+      "loss": 0.9475,
+      "step": 47660
+    },
+    {
+      "epoch": 0.3450672110143543,
+      "grad_norm": 0.16307930648326874,
+      "learning_rate": 4.654940027651705e-06,
+      "loss": 0.9627,
+      "step": 47670
+    },
+    {
+      "epoch": 0.34513959767494046,
+      "grad_norm": 0.4456617832183838,
+      "learning_rate": 4.654867640991119e-06,
+      "loss": 0.9653,
+      "step": 47680
+    },
+    {
+      "epoch": 0.34521198433552663,
+      "grad_norm": 0.19592955708503723,
+      "learning_rate": 4.654795254330532e-06,
+      "loss": 0.958,
+      "step": 47690
+    },
+    {
+      "epoch": 0.3452843709961128,
+      "grad_norm": 0.1693570464849472,
+      "learning_rate": 4.654722867669946e-06,
+      "loss": 0.9591,
+      "step": 47700
+    },
+    {
+      "epoch": 0.34535675765669904,
+      "grad_norm": 0.16405141353607178,
+      "learning_rate": 4.6546504810093605e-06,
+      "loss": 0.9568,
+      "step": 47710
+    },
+    {
+      "epoch": 0.3454291443172852,
+      "grad_norm": 0.15746448934078217,
+      "learning_rate": 4.654578094348774e-06,
+      "loss": 0.9587,
+      "step": 47720
+    },
+    {
+      "epoch": 0.3455015309778714,
+      "grad_norm": 0.17681455612182617,
+      "learning_rate": 4.654505707688188e-06,
+      "loss": 0.9533,
+      "step": 47730
+    },
+    {
+      "epoch": 0.3455739176384576,
+      "grad_norm": 0.14864014089107513,
+      "learning_rate": 4.654433321027601e-06,
+      "loss": 0.9444,
+      "step": 47740
+    },
+    {
+      "epoch": 0.34564630429904375,
+      "grad_norm": 0.17312194406986237,
+      "learning_rate": 4.654360934367016e-06,
+      "loss": 0.9546,
+      "step": 47750
+    },
+    {
+      "epoch": 0.34571869095963,
+      "grad_norm": 0.17767411470413208,
+      "learning_rate": 4.654288547706429e-06,
+      "loss": 0.9547,
+      "step": 47760
+    },
+    {
+      "epoch": 0.34579107762021616,
+      "grad_norm": 0.16189977526664734,
+      "learning_rate": 4.654216161045843e-06,
+      "loss": 0.959,
+      "step": 47770
+    },
+    {
+      "epoch": 0.34586346428080234,
+      "grad_norm": 0.15381406247615814,
+      "learning_rate": 4.654143774385257e-06,
+      "loss": 0.9532,
+      "step": 47780
+    },
+    {
+      "epoch": 0.3459358509413885,
+      "grad_norm": 0.1887199878692627,
+      "learning_rate": 4.654071387724671e-06,
+      "loss": 0.9523,
+      "step": 47790
+    },
+    {
+      "epoch": 0.3460082376019747,
+      "grad_norm": 0.1715165674686432,
+      "learning_rate": 4.653999001064085e-06,
+      "loss": 0.9387,
+      "step": 47800
+    },
+    {
+      "epoch": 0.3460806242625609,
+      "grad_norm": 0.17425018548965454,
+      "learning_rate": 4.653926614403498e-06,
+      "loss": 0.9659,
+      "step": 47810
+    },
+    {
+      "epoch": 0.3461530109231471,
+      "grad_norm": 0.16004815697669983,
+      "learning_rate": 4.653854227742912e-06,
+      "loss": 0.9524,
+      "step": 47820
+    },
+    {
+      "epoch": 0.3462253975837333,
+      "grad_norm": 0.16618654131889343,
+      "learning_rate": 4.653781841082326e-06,
+      "loss": 0.9573,
+      "step": 47830
+    },
+    {
+      "epoch": 0.34629778424431945,
+      "grad_norm": 0.18237411975860596,
+      "learning_rate": 4.653709454421739e-06,
+      "loss": 0.9503,
+      "step": 47840
+    },
+    {
+      "epoch": 0.3463701709049056,
+      "grad_norm": 0.17747117578983307,
+      "learning_rate": 4.653637067761153e-06,
+      "loss": 0.951,
+      "step": 47850
+    },
+    {
+      "epoch": 0.3464425575654918,
+      "grad_norm": 0.1547970473766327,
+      "learning_rate": 4.653564681100567e-06,
+      "loss": 0.9493,
+      "step": 47860
+    },
+    {
+      "epoch": 0.34651494422607804,
+      "grad_norm": 0.20299063622951508,
+      "learning_rate": 4.653492294439981e-06,
+      "loss": 0.9549,
+      "step": 47870
+    },
+    {
+      "epoch": 0.3465873308866642,
+      "grad_norm": 0.16217590868473053,
+      "learning_rate": 4.6534199077793945e-06,
+      "loss": 0.951,
+      "step": 47880
+    },
+    {
+      "epoch": 0.3466597175472504,
+      "grad_norm": 0.21849872171878815,
+      "learning_rate": 4.653347521118808e-06,
+      "loss": 0.9457,
+      "step": 47890
+    },
+    {
+      "epoch": 0.34673210420783657,
+      "grad_norm": 0.22776605188846588,
+      "learning_rate": 4.653275134458223e-06,
+      "loss": 0.9705,
+      "step": 47900
+    },
+    {
+      "epoch": 0.34680449086842274,
+      "grad_norm": 0.1702580451965332,
+      "learning_rate": 4.653202747797636e-06,
+      "loss": 0.9504,
+      "step": 47910
+    },
+    {
+      "epoch": 0.346876877529009,
+      "grad_norm": 0.16190893948078156,
+      "learning_rate": 4.65313036113705e-06,
+      "loss": 0.9618,
+      "step": 47920
+    },
+    {
+      "epoch": 0.34694926418959515,
+      "grad_norm": 0.18085525929927826,
+      "learning_rate": 4.6530579744764634e-06,
+      "loss": 0.9564,
+      "step": 47930
+    },
+    {
+      "epoch": 0.34702165085018133,
+      "grad_norm": 0.16075456142425537,
+      "learning_rate": 4.652985587815878e-06,
+      "loss": 0.9576,
+      "step": 47940
+    },
+    {
+      "epoch": 0.3470940375107675,
+      "grad_norm": 0.1670777052640915,
+      "learning_rate": 4.6529132011552915e-06,
+      "loss": 0.9598,
+      "step": 47950
+    },
+    {
+      "epoch": 0.3471664241713537,
+      "grad_norm": 0.1586279422044754,
+      "learning_rate": 4.652840814494705e-06,
+      "loss": 0.9469,
+      "step": 47960
+    },
+    {
+      "epoch": 0.3472388108319399,
+      "grad_norm": 0.163706973195076,
+      "learning_rate": 4.652768427834119e-06,
+      "loss": 0.9456,
+      "step": 47970
+    },
+    {
+      "epoch": 0.3473111974925261,
+      "grad_norm": 0.16007377207279205,
+      "learning_rate": 4.652696041173533e-06,
+      "loss": 0.9445,
+      "step": 47980
+    },
+    {
+      "epoch": 0.34738358415311227,
+      "grad_norm": 0.18147924542427063,
+      "learning_rate": 4.652623654512947e-06,
+      "loss": 0.9527,
+      "step": 47990
+    },
+    {
+      "epoch": 0.34745597081369844,
+      "grad_norm": 0.1525728702545166,
+      "learning_rate": 4.6525512678523604e-06,
+      "loss": 0.9419,
+      "step": 48000
+    },
+    {
+      "epoch": 0.3475283574742846,
+      "grad_norm": 0.168410062789917,
+      "learning_rate": 4.652478881191774e-06,
+      "loss": 0.9464,
+      "step": 48010
+    },
+    {
+      "epoch": 0.3476007441348708,
+      "grad_norm": 0.16513432562351227,
+      "learning_rate": 4.652406494531188e-06,
+      "loss": 0.9467,
+      "step": 48020
+    },
+    {
+      "epoch": 0.34767313079545703,
+      "grad_norm": 0.14979685842990875,
+      "learning_rate": 4.652334107870602e-06,
+      "loss": 0.9547,
+      "step": 48030
+    },
+    {
+      "epoch": 0.3477455174560432,
+      "grad_norm": 0.17704786360263824,
+      "learning_rate": 4.652261721210016e-06,
+      "loss": 0.9491,
+      "step": 48040
+    },
+    {
+      "epoch": 0.3478179041166294,
+      "grad_norm": 0.1605810523033142,
+      "learning_rate": 4.652189334549429e-06,
+      "loss": 0.9566,
+      "step": 48050
+    },
+    {
+      "epoch": 0.34789029077721556,
+      "grad_norm": 0.1907358467578888,
+      "learning_rate": 4.652116947888843e-06,
+      "loss": 0.9568,
+      "step": 48060
+    },
+    {
+      "epoch": 0.34796267743780174,
+      "grad_norm": 0.16195763647556305,
+      "learning_rate": 4.6520445612282574e-06,
+      "loss": 0.9532,
+      "step": 48070
+    },
+    {
+      "epoch": 0.34803506409838797,
+      "grad_norm": 0.1610960215330124,
+      "learning_rate": 4.651972174567671e-06,
+      "loss": 0.9452,
+      "step": 48080
+    },
+    {
+      "epoch": 0.34810745075897415,
+      "grad_norm": 0.17993375658988953,
+      "learning_rate": 4.651899787907085e-06,
+      "loss": 0.9497,
+      "step": 48090
+    },
+    {
+      "epoch": 0.3481798374195603,
+      "grad_norm": 0.16861571371555328,
+      "learning_rate": 4.651827401246498e-06,
+      "loss": 0.9503,
+      "step": 48100
+    },
+    {
+      "epoch": 0.3482522240801465,
+      "grad_norm": 0.16078142821788788,
+      "learning_rate": 4.651755014585913e-06,
+      "loss": 0.9387,
+      "step": 48110
+    },
+    {
+      "epoch": 0.3483246107407327,
+      "grad_norm": 0.1687428504228592,
+      "learning_rate": 4.651682627925326e-06,
+      "loss": 0.9459,
+      "step": 48120
+    },
+    {
+      "epoch": 0.3483969974013189,
+      "grad_norm": 0.17936325073242188,
+      "learning_rate": 4.65161024126474e-06,
+      "loss": 0.944,
+      "step": 48130
+    },
+    {
+      "epoch": 0.3484693840619051,
+      "grad_norm": 0.16002655029296875,
+      "learning_rate": 4.651537854604154e-06,
+      "loss": 0.9528,
+      "step": 48140
+    },
+    {
+      "epoch": 0.34854177072249126,
+      "grad_norm": 0.16870175302028656,
+      "learning_rate": 4.651465467943568e-06,
+      "loss": 0.9461,
+      "step": 48150
+    },
+    {
+      "epoch": 0.34861415738307744,
+      "grad_norm": 0.37288644909858704,
+      "learning_rate": 4.651393081282982e-06,
+      "loss": 0.9438,
+      "step": 48160
+    },
+    {
+      "epoch": 0.3486865440436636,
+      "grad_norm": 0.1594185084104538,
+      "learning_rate": 4.651320694622395e-06,
+      "loss": 0.9631,
+      "step": 48170
+    },
+    {
+      "epoch": 0.34875893070424985,
+      "grad_norm": 0.16958339512348175,
+      "learning_rate": 4.651248307961809e-06,
+      "loss": 0.9672,
+      "step": 48180
+    },
+    {
+      "epoch": 0.348831317364836,
+      "grad_norm": 0.2117815464735031,
+      "learning_rate": 4.651175921301223e-06,
+      "loss": 0.9624,
+      "step": 48190
+    },
+    {
+      "epoch": 0.3489037040254222,
+      "grad_norm": 0.16270878911018372,
+      "learning_rate": 4.651103534640637e-06,
+      "loss": 0.9621,
+      "step": 48200
+    },
+    {
+      "epoch": 0.3489760906860084,
+      "grad_norm": 0.1788318157196045,
+      "learning_rate": 4.651031147980051e-06,
+      "loss": 0.9604,
+      "step": 48210
+    },
+    {
+      "epoch": 0.34904847734659455,
+      "grad_norm": 0.20249681174755096,
+      "learning_rate": 4.650958761319464e-06,
+      "loss": 0.9452,
+      "step": 48220
+    },
+    {
+      "epoch": 0.34912086400718073,
+      "grad_norm": 0.16005674004554749,
+      "learning_rate": 4.650886374658879e-06,
+      "loss": 0.9375,
+      "step": 48230
+    },
+    {
+      "epoch": 0.34919325066776696,
+      "grad_norm": 0.182306170463562,
+      "learning_rate": 4.650813987998292e-06,
+      "loss": 0.9549,
+      "step": 48240
+    },
+    {
+      "epoch": 0.34926563732835314,
+      "grad_norm": 0.1851535588502884,
+      "learning_rate": 4.650741601337706e-06,
+      "loss": 0.9459,
+      "step": 48250
+    },
+    {
+      "epoch": 0.3493380239889393,
+      "grad_norm": 0.1800885647535324,
+      "learning_rate": 4.6506692146771195e-06,
+      "loss": 0.9594,
+      "step": 48260
+    },
+    {
+      "epoch": 0.3494104106495255,
+      "grad_norm": 0.1562681943178177,
+      "learning_rate": 4.650596828016534e-06,
+      "loss": 0.9548,
+      "step": 48270
+    },
+    {
+      "epoch": 0.34948279731011167,
+      "grad_norm": 0.164944589138031,
+      "learning_rate": 4.650524441355948e-06,
+      "loss": 0.9595,
+      "step": 48280
+    },
+    {
+      "epoch": 0.3495551839706979,
+      "grad_norm": 0.15770290791988373,
+      "learning_rate": 4.650452054695361e-06,
+      "loss": 0.9573,
+      "step": 48290
+    },
+    {
+      "epoch": 0.3496275706312841,
+      "grad_norm": 0.19235900044441223,
+      "learning_rate": 4.650379668034775e-06,
+      "loss": 0.9381,
+      "step": 48300
+    },
+    {
+      "epoch": 0.34969995729187026,
+      "grad_norm": 0.17154589295387268,
+      "learning_rate": 4.650307281374189e-06,
+      "loss": 0.9434,
+      "step": 48310
+    },
+    {
+      "epoch": 0.34977234395245643,
+      "grad_norm": 0.19060273468494415,
+      "learning_rate": 4.650234894713603e-06,
+      "loss": 0.9666,
+      "step": 48320
+    },
+    {
+      "epoch": 0.3498447306130426,
+      "grad_norm": 0.16510871052742004,
+      "learning_rate": 4.6501625080530165e-06,
+      "loss": 0.9498,
+      "step": 48330
+    },
+    {
+      "epoch": 0.34991711727362884,
+      "grad_norm": 0.1761731505393982,
+      "learning_rate": 4.65009012139243e-06,
+      "loss": 0.9522,
+      "step": 48340
+    },
+    {
+      "epoch": 0.349989503934215,
+      "grad_norm": 0.17731162905693054,
+      "learning_rate": 4.650017734731845e-06,
+      "loss": 0.9507,
+      "step": 48350
+    },
+    {
+      "epoch": 0.3500618905948012,
+      "grad_norm": 0.17625099420547485,
+      "learning_rate": 4.649945348071258e-06,
+      "loss": 0.9528,
+      "step": 48360
+    },
+    {
+      "epoch": 0.35013427725538737,
+      "grad_norm": 0.16574423015117645,
+      "learning_rate": 4.649872961410671e-06,
+      "loss": 0.969,
+      "step": 48370
+    },
+    {
+      "epoch": 0.35020666391597355,
+      "grad_norm": 0.16655471920967102,
+      "learning_rate": 4.6498005747500855e-06,
+      "loss": 0.9462,
+      "step": 48380
+    },
+    {
+      "epoch": 0.3502790505765597,
+      "grad_norm": 0.17919887602329254,
+      "learning_rate": 4.649728188089499e-06,
+      "loss": 0.9669,
+      "step": 48390
+    },
+    {
+      "epoch": 0.35035143723714596,
+      "grad_norm": 0.1737910956144333,
+      "learning_rate": 4.649655801428913e-06,
+      "loss": 0.9544,
+      "step": 48400
+    },
+    {
+      "epoch": 0.35042382389773213,
+      "grad_norm": 0.1664748638868332,
+      "learning_rate": 4.649583414768326e-06,
+      "loss": 0.9394,
+      "step": 48410
+    },
+    {
+      "epoch": 0.3504962105583183,
+      "grad_norm": 0.1567506492137909,
+      "learning_rate": 4.649511028107741e-06,
+      "loss": 0.9453,
+      "step": 48420
+    },
+    {
+      "epoch": 0.3505685972189045,
+      "grad_norm": 0.17070920765399933,
+      "learning_rate": 4.649438641447154e-06,
+      "loss": 0.9549,
+      "step": 48430
+    },
+    {
+      "epoch": 0.35064098387949066,
+      "grad_norm": 0.16167403757572174,
+      "learning_rate": 4.649366254786568e-06,
+      "loss": 0.9453,
+      "step": 48440
+    },
+    {
+      "epoch": 0.3507133705400769,
+      "grad_norm": 0.15973427891731262,
+      "learning_rate": 4.649293868125982e-06,
+      "loss": 0.9561,
+      "step": 48450
+    },
+    {
+      "epoch": 0.3507857572006631,
+      "grad_norm": 0.16272811591625214,
+      "learning_rate": 4.649221481465396e-06,
+      "loss": 0.9554,
+      "step": 48460
+    },
+    {
+      "epoch": 0.35085814386124925,
+      "grad_norm": 0.1659621149301529,
+      "learning_rate": 4.64914909480481e-06,
+      "loss": 0.9474,
+      "step": 48470
+    },
+    {
+      "epoch": 0.3509305305218354,
+      "grad_norm": 0.16306394338607788,
+      "learning_rate": 4.649076708144223e-06,
+      "loss": 0.9511,
+      "step": 48480
+    },
+    {
+      "epoch": 0.3510029171824216,
+      "grad_norm": 0.1689397245645523,
+      "learning_rate": 4.649004321483637e-06,
+      "loss": 0.9528,
+      "step": 48490
+    },
+    {
+      "epoch": 0.35107530384300784,
+      "grad_norm": 0.16257858276367188,
+      "learning_rate": 4.648931934823051e-06,
+      "loss": 0.9544,
+      "step": 48500
+    },
+    {
+      "epoch": 0.351147690503594,
+      "grad_norm": 0.1643674671649933,
+      "learning_rate": 4.648859548162465e-06,
+      "loss": 0.9441,
+      "step": 48510
+    },
+    {
+      "epoch": 0.3512200771641802,
+      "grad_norm": 0.18621480464935303,
+      "learning_rate": 4.648787161501879e-06,
+      "loss": 0.9677,
+      "step": 48520
+    },
+    {
+      "epoch": 0.35129246382476637,
+      "grad_norm": 0.15909579396247864,
+      "learning_rate": 4.648714774841292e-06,
+      "loss": 0.9404,
+      "step": 48530
+    },
+    {
+      "epoch": 0.35136485048535254,
+      "grad_norm": 0.16073808073997498,
+      "learning_rate": 4.648642388180707e-06,
+      "loss": 0.9649,
+      "step": 48540
+    },
+    {
+      "epoch": 0.3514372371459387,
+      "grad_norm": 0.15749648213386536,
+      "learning_rate": 4.64857000152012e-06,
+      "loss": 0.9665,
+      "step": 48550
+    },
+    {
+      "epoch": 0.35150962380652495,
+      "grad_norm": 0.18623562157154083,
+      "learning_rate": 4.648497614859534e-06,
+      "loss": 0.9529,
+      "step": 48560
+    },
+    {
+      "epoch": 0.35158201046711113,
+      "grad_norm": 0.16134946048259735,
+      "learning_rate": 4.6484252281989476e-06,
+      "loss": 0.9499,
+      "step": 48570
+    },
+    {
+      "epoch": 0.3516543971276973,
+      "grad_norm": 0.15732122957706451,
+      "learning_rate": 4.648352841538362e-06,
+      "loss": 0.9572,
+      "step": 48580
+    },
+    {
+      "epoch": 0.3517267837882835,
+      "grad_norm": 0.1690632700920105,
+      "learning_rate": 4.648280454877776e-06,
+      "loss": 0.9506,
+      "step": 48590
+    },
+    {
+      "epoch": 0.35179917044886966,
+      "grad_norm": 0.16241760551929474,
+      "learning_rate": 4.648208068217189e-06,
+      "loss": 0.9518,
+      "step": 48600
+    },
+    {
+      "epoch": 0.3518715571094559,
+      "grad_norm": 0.1884649395942688,
+      "learning_rate": 4.648135681556603e-06,
+      "loss": 0.9577,
+      "step": 48610
+    },
+    {
+      "epoch": 0.35194394377004207,
+      "grad_norm": 0.16294220089912415,
+      "learning_rate": 4.648063294896017e-06,
+      "loss": 0.9432,
+      "step": 48620
+    },
+    {
+      "epoch": 0.35201633043062824,
+      "grad_norm": 0.18577656149864197,
+      "learning_rate": 4.647990908235431e-06,
+      "loss": 0.9513,
+      "step": 48630
+    },
+    {
+      "epoch": 0.3520887170912144,
+      "grad_norm": 0.16119709610939026,
+      "learning_rate": 4.647918521574845e-06,
+      "loss": 0.9478,
+      "step": 48640
+    },
+    {
+      "epoch": 0.3521611037518006,
+      "grad_norm": 0.1651376187801361,
+      "learning_rate": 4.647846134914258e-06,
+      "loss": 0.9487,
+      "step": 48650
+    },
+    {
+      "epoch": 0.35223349041238683,
+      "grad_norm": 0.18918465077877045,
+      "learning_rate": 4.647773748253672e-06,
+      "loss": 0.9562,
+      "step": 48660
+    },
+    {
+      "epoch": 0.352305877072973,
+      "grad_norm": 0.1643255203962326,
+      "learning_rate": 4.647701361593086e-06,
+      "loss": 0.9502,
+      "step": 48670
+    },
+    {
+      "epoch": 0.3523782637335592,
+      "grad_norm": 0.16544142365455627,
+      "learning_rate": 4.6476289749325e-06,
+      "loss": 0.9399,
+      "step": 48680
+    },
+    {
+      "epoch": 0.35245065039414536,
+      "grad_norm": 0.1629703789949417,
+      "learning_rate": 4.6475565882719135e-06,
+      "loss": 0.9514,
+      "step": 48690
+    },
+    {
+      "epoch": 0.35252303705473154,
+      "grad_norm": 0.15862901508808136,
+      "learning_rate": 4.647484201611327e-06,
+      "loss": 0.949,
+      "step": 48700
+    },
+    {
+      "epoch": 0.35259542371531777,
+      "grad_norm": 0.1527785062789917,
+      "learning_rate": 4.647411814950742e-06,
+      "loss": 0.9548,
+      "step": 48710
+    },
+    {
+      "epoch": 0.35266781037590395,
+      "grad_norm": 0.1874808520078659,
+      "learning_rate": 4.647339428290155e-06,
+      "loss": 0.9653,
+      "step": 48720
+    },
+    {
+      "epoch": 0.3527401970364901,
+      "grad_norm": 0.18184229731559753,
+      "learning_rate": 4.647267041629569e-06,
+      "loss": 0.9433,
+      "step": 48730
+    },
+    {
+      "epoch": 0.3528125836970763,
+      "grad_norm": 0.16013579070568085,
+      "learning_rate": 4.6471946549689824e-06,
+      "loss": 0.9545,
+      "step": 48740
+    },
+    {
+      "epoch": 0.3528849703576625,
+      "grad_norm": 0.1711476594209671,
+      "learning_rate": 4.647122268308397e-06,
+      "loss": 0.9541,
+      "step": 48750
+    },
+    {
+      "epoch": 0.35295735701824865,
+      "grad_norm": 0.16671577095985413,
+      "learning_rate": 4.6470498816478105e-06,
+      "loss": 0.9568,
+      "step": 48760
+    },
+    {
+      "epoch": 0.3530297436788349,
+      "grad_norm": 0.2846975326538086,
+      "learning_rate": 4.646977494987224e-06,
+      "loss": 0.9496,
+      "step": 48770
+    },
+    {
+      "epoch": 0.35310213033942106,
+      "grad_norm": 0.15886595845222473,
+      "learning_rate": 4.646905108326638e-06,
+      "loss": 0.9667,
+      "step": 48780
+    },
+    {
+      "epoch": 0.35317451700000724,
+      "grad_norm": 0.1654849648475647,
+      "learning_rate": 4.646832721666052e-06,
+      "loss": 0.9483,
+      "step": 48790
+    },
+    {
+      "epoch": 0.3532469036605934,
+      "grad_norm": 0.1595975011587143,
+      "learning_rate": 4.646760335005466e-06,
+      "loss": 0.952,
+      "step": 48800
+    },
+    {
+      "epoch": 0.3533192903211796,
+      "grad_norm": 0.19034674763679504,
+      "learning_rate": 4.6466879483448794e-06,
+      "loss": 0.9593,
+      "step": 48810
+    },
+    {
+      "epoch": 0.3533916769817658,
+      "grad_norm": 0.16194772720336914,
+      "learning_rate": 4.646615561684293e-06,
+      "loss": 0.9536,
+      "step": 48820
+    },
+    {
+      "epoch": 0.353464063642352,
+      "grad_norm": 0.19409111142158508,
+      "learning_rate": 4.6465431750237075e-06,
+      "loss": 0.9578,
+      "step": 48830
+    },
+    {
+      "epoch": 0.3535364503029382,
+      "grad_norm": 0.15368357300758362,
+      "learning_rate": 4.646470788363121e-06,
+      "loss": 0.9635,
+      "step": 48840
+    },
+    {
+      "epoch": 0.35360883696352435,
+      "grad_norm": 0.1651766002178192,
+      "learning_rate": 4.646398401702535e-06,
+      "loss": 0.9542,
+      "step": 48850
+    },
+    {
+      "epoch": 0.35368122362411053,
+      "grad_norm": 0.16302619874477386,
+      "learning_rate": 4.646326015041948e-06,
+      "loss": 0.9451,
+      "step": 48860
+    },
+    {
+      "epoch": 0.35375361028469676,
+      "grad_norm": 0.16752517223358154,
+      "learning_rate": 4.646253628381363e-06,
+      "loss": 0.956,
+      "step": 48870
+    },
+    {
+      "epoch": 0.35382599694528294,
+      "grad_norm": 0.16661721467971802,
+      "learning_rate": 4.6461812417207765e-06,
+      "loss": 0.9501,
+      "step": 48880
+    },
+    {
+      "epoch": 0.3538983836058691,
+      "grad_norm": 0.16403451561927795,
+      "learning_rate": 4.64610885506019e-06,
+      "loss": 0.945,
+      "step": 48890
+    },
+    {
+      "epoch": 0.3539707702664553,
+      "grad_norm": 0.16170132160186768,
+      "learning_rate": 4.646036468399604e-06,
+      "loss": 0.9528,
+      "step": 48900
+    },
+    {
+      "epoch": 0.35404315692704147,
+      "grad_norm": 0.15352827310562134,
+      "learning_rate": 4.645964081739017e-06,
+      "loss": 0.964,
+      "step": 48910
+    },
+    {
+      "epoch": 0.35411554358762765,
+      "grad_norm": 0.2216578722000122,
+      "learning_rate": 4.645891695078431e-06,
+      "loss": 0.9571,
+      "step": 48920
+    },
+    {
+      "epoch": 0.3541879302482139,
+      "grad_norm": 0.1692575216293335,
+      "learning_rate": 4.6458193084178445e-06,
+      "loss": 0.9552,
+      "step": 48930
+    },
+    {
+      "epoch": 0.35426031690880005,
+      "grad_norm": 0.16259534657001495,
+      "learning_rate": 4.645746921757259e-06,
+      "loss": 0.9497,
+      "step": 48940
+    },
+    {
+      "epoch": 0.35433270356938623,
+      "grad_norm": 0.2892276644706726,
+      "learning_rate": 4.645674535096673e-06,
+      "loss": 0.9429,
+      "step": 48950
+    },
+    {
+      "epoch": 0.3544050902299724,
+      "grad_norm": 0.1655515879392624,
+      "learning_rate": 4.645602148436086e-06,
+      "loss": 0.9362,
+      "step": 48960
+    },
+    {
+      "epoch": 0.3544774768905586,
+      "grad_norm": 0.1588042974472046,
+      "learning_rate": 4.6455297617755e-06,
+      "loss": 0.9581,
+      "step": 48970
+    },
+    {
+      "epoch": 0.3545498635511448,
+      "grad_norm": 0.16203071177005768,
+      "learning_rate": 4.645457375114914e-06,
+      "loss": 0.9525,
+      "step": 48980
+    },
+    {
+      "epoch": 0.354622250211731,
+      "grad_norm": 0.1807693988084793,
+      "learning_rate": 4.645384988454328e-06,
+      "loss": 0.9412,
+      "step": 48990
+    },
+    {
+      "epoch": 0.35469463687231717,
+      "grad_norm": 0.14859171211719513,
+      "learning_rate": 4.6453126017937415e-06,
+      "loss": 0.955,
+      "step": 49000
+    },
+    {
+      "epoch": 0.35476702353290335,
+      "grad_norm": 0.17705701291561127,
+      "learning_rate": 4.645240215133155e-06,
+      "loss": 0.9459,
+      "step": 49010
+    },
+    {
+      "epoch": 0.3548394101934895,
+      "grad_norm": 0.1593550145626068,
+      "learning_rate": 4.64516782847257e-06,
+      "loss": 0.9333,
+      "step": 49020
+    },
+    {
+      "epoch": 0.35491179685407576,
+      "grad_norm": 0.16969063878059387,
+      "learning_rate": 4.645095441811983e-06,
+      "loss": 0.9572,
+      "step": 49030
+    },
+    {
+      "epoch": 0.35498418351466193,
+      "grad_norm": 0.16061443090438843,
+      "learning_rate": 4.645023055151397e-06,
+      "loss": 0.9582,
+      "step": 49040
+    },
+    {
+      "epoch": 0.3550565701752481,
+      "grad_norm": 0.17726626992225647,
+      "learning_rate": 4.6449506684908105e-06,
+      "loss": 0.9508,
+      "step": 49050
+    },
+    {
+      "epoch": 0.3551289568358343,
+      "grad_norm": 0.1747400313615799,
+      "learning_rate": 4.644878281830225e-06,
+      "loss": 0.9436,
+      "step": 49060
+    },
+    {
+      "epoch": 0.35520134349642046,
+      "grad_norm": 0.1950230598449707,
+      "learning_rate": 4.6448058951696385e-06,
+      "loss": 0.9571,
+      "step": 49070
+    },
+    {
+      "epoch": 0.35527373015700664,
+      "grad_norm": 0.1573755443096161,
+      "learning_rate": 4.644733508509052e-06,
+      "loss": 0.9574,
+      "step": 49080
+    },
+    {
+      "epoch": 0.35534611681759287,
+      "grad_norm": 0.17132724821567535,
+      "learning_rate": 4.644661121848466e-06,
+      "loss": 0.9586,
+      "step": 49090
+    },
+    {
+      "epoch": 0.35541850347817905,
+      "grad_norm": 0.18015651404857635,
+      "learning_rate": 4.64458873518788e-06,
+      "loss": 0.9312,
+      "step": 49100
+    },
+    {
+      "epoch": 0.3554908901387652,
+      "grad_norm": 0.17248773574829102,
+      "learning_rate": 4.644516348527294e-06,
+      "loss": 0.9444,
+      "step": 49110
+    },
+    {
+      "epoch": 0.3555632767993514,
+      "grad_norm": 0.16664332151412964,
+      "learning_rate": 4.6444439618667075e-06,
+      "loss": 0.9524,
+      "step": 49120
+    },
+    {
+      "epoch": 0.3556356634599376,
+      "grad_norm": 0.16630926728248596,
+      "learning_rate": 4.644371575206121e-06,
+      "loss": 0.9397,
+      "step": 49130
+    },
+    {
+      "epoch": 0.3557080501205238,
+      "grad_norm": 0.17303983867168427,
+      "learning_rate": 4.6442991885455356e-06,
+      "loss": 0.9622,
+      "step": 49140
+    },
+    {
+      "epoch": 0.35578043678111,
+      "grad_norm": 0.15973907709121704,
+      "learning_rate": 4.644226801884949e-06,
+      "loss": 0.9364,
+      "step": 49150
+    },
+    {
+      "epoch": 0.35585282344169616,
+      "grad_norm": 0.1687871366739273,
+      "learning_rate": 4.644154415224363e-06,
+      "loss": 0.9577,
+      "step": 49160
+    },
+    {
+      "epoch": 0.35592521010228234,
+      "grad_norm": 0.16706955432891846,
+      "learning_rate": 4.644082028563776e-06,
+      "loss": 0.9619,
+      "step": 49170
+    },
+    {
+      "epoch": 0.3559975967628685,
+      "grad_norm": 0.19494742155075073,
+      "learning_rate": 4.644009641903191e-06,
+      "loss": 0.947,
+      "step": 49180
+    },
+    {
+      "epoch": 0.35606998342345475,
+      "grad_norm": 0.1793723702430725,
+      "learning_rate": 4.6439372552426045e-06,
+      "loss": 0.9618,
+      "step": 49190
+    },
+    {
+      "epoch": 0.3561423700840409,
+      "grad_norm": 0.16906292736530304,
+      "learning_rate": 4.643864868582018e-06,
+      "loss": 0.9433,
+      "step": 49200
+    },
+    {
+      "epoch": 0.3562147567446271,
+      "grad_norm": 0.1776910424232483,
+      "learning_rate": 4.643792481921432e-06,
+      "loss": 0.9434,
+      "step": 49210
+    },
+    {
+      "epoch": 0.3562871434052133,
+      "grad_norm": 0.17651747167110443,
+      "learning_rate": 4.643720095260846e-06,
+      "loss": 0.9524,
+      "step": 49220
+    },
+    {
+      "epoch": 0.35635953006579946,
+      "grad_norm": 0.1679781824350357,
+      "learning_rate": 4.64364770860026e-06,
+      "loss": 0.9429,
+      "step": 49230
+    },
+    {
+      "epoch": 0.3564319167263857,
+      "grad_norm": 0.17428244650363922,
+      "learning_rate": 4.643575321939673e-06,
+      "loss": 0.9497,
+      "step": 49240
+    },
+    {
+      "epoch": 0.35650430338697187,
+      "grad_norm": 0.16317783296108246,
+      "learning_rate": 4.643502935279087e-06,
+      "loss": 0.9579,
+      "step": 49250
+    },
+    {
+      "epoch": 0.35657669004755804,
+      "grad_norm": 0.17340287566184998,
+      "learning_rate": 4.643430548618501e-06,
+      "loss": 0.9584,
+      "step": 49260
+    },
+    {
+      "epoch": 0.3566490767081442,
+      "grad_norm": 0.17054349184036255,
+      "learning_rate": 4.643358161957915e-06,
+      "loss": 0.9482,
+      "step": 49270
+    },
+    {
+      "epoch": 0.3567214633687304,
+      "grad_norm": 0.15321429073810577,
+      "learning_rate": 4.643285775297329e-06,
+      "loss": 0.9532,
+      "step": 49280
+    },
+    {
+      "epoch": 0.3567938500293166,
+      "grad_norm": 0.16249078512191772,
+      "learning_rate": 4.643213388636742e-06,
+      "loss": 0.9366,
+      "step": 49290
+    },
+    {
+      "epoch": 0.3568662366899028,
+      "grad_norm": 0.23813359439373016,
+      "learning_rate": 4.643141001976156e-06,
+      "loss": 0.9628,
+      "step": 49300
+    },
+    {
+      "epoch": 0.356938623350489,
+      "grad_norm": 0.18279701471328735,
+      "learning_rate": 4.64306861531557e-06,
+      "loss": 0.962,
+      "step": 49310
+    },
+    {
+      "epoch": 0.35701101001107516,
+      "grad_norm": 0.1736493855714798,
+      "learning_rate": 4.642996228654984e-06,
+      "loss": 0.9578,
+      "step": 49320
+    },
+    {
+      "epoch": 0.35708339667166134,
+      "grad_norm": 0.16122330725193024,
+      "learning_rate": 4.642923841994398e-06,
+      "loss": 0.9503,
+      "step": 49330
+    },
+    {
+      "epoch": 0.3571557833322475,
+      "grad_norm": 0.15669801831245422,
+      "learning_rate": 4.642851455333811e-06,
+      "loss": 0.9562,
+      "step": 49340
+    },
+    {
+      "epoch": 0.35722816999283374,
+      "grad_norm": 0.16338752210140228,
+      "learning_rate": 4.642779068673226e-06,
+      "loss": 0.9554,
+      "step": 49350
+    },
+    {
+      "epoch": 0.3573005566534199,
+      "grad_norm": 0.16918496787548065,
+      "learning_rate": 4.642706682012639e-06,
+      "loss": 0.946,
+      "step": 49360
+    },
+    {
+      "epoch": 0.3573729433140061,
+      "grad_norm": 0.17653869092464447,
+      "learning_rate": 4.642634295352053e-06,
+      "loss": 0.9528,
+      "step": 49370
+    },
+    {
+      "epoch": 0.3574453299745923,
+      "grad_norm": 0.1767107993364334,
+      "learning_rate": 4.642561908691467e-06,
+      "loss": 0.9456,
+      "step": 49380
+    },
+    {
+      "epoch": 0.35751771663517845,
+      "grad_norm": 0.16276022791862488,
+      "learning_rate": 4.642489522030881e-06,
+      "loss": 0.9418,
+      "step": 49390
+    },
+    {
+      "epoch": 0.3575901032957647,
+      "grad_norm": 0.15793178975582123,
+      "learning_rate": 4.642417135370295e-06,
+      "loss": 0.9514,
+      "step": 49400
+    },
+    {
+      "epoch": 0.35766248995635086,
+      "grad_norm": 0.16991020739078522,
+      "learning_rate": 4.642344748709708e-06,
+      "loss": 0.939,
+      "step": 49410
+    },
+    {
+      "epoch": 0.35773487661693704,
+      "grad_norm": 0.1681062877178192,
+      "learning_rate": 4.642272362049122e-06,
+      "loss": 0.9462,
+      "step": 49420
+    },
+    {
+      "epoch": 0.3578072632775232,
+      "grad_norm": 0.18085841834545135,
+      "learning_rate": 4.6421999753885355e-06,
+      "loss": 0.945,
+      "step": 49430
+    },
+    {
+      "epoch": 0.3578796499381094,
+      "grad_norm": 0.1598934829235077,
+      "learning_rate": 4.642127588727949e-06,
+      "loss": 0.9579,
+      "step": 49440
+    },
+    {
+      "epoch": 0.35795203659869557,
+      "grad_norm": 0.19113032519817352,
+      "learning_rate": 4.642055202067363e-06,
+      "loss": 0.9484,
+      "step": 49450
+    },
+    {
+      "epoch": 0.3580244232592818,
+      "grad_norm": 0.16052620112895966,
+      "learning_rate": 4.641982815406777e-06,
+      "loss": 0.9557,
+      "step": 49460
+    },
+    {
+      "epoch": 0.358096809919868,
+      "grad_norm": 0.16908268630504608,
+      "learning_rate": 4.641910428746191e-06,
+      "loss": 0.9453,
+      "step": 49470
+    },
+    {
+      "epoch": 0.35816919658045415,
+      "grad_norm": 0.19930988550186157,
+      "learning_rate": 4.6418380420856044e-06,
+      "loss": 0.9453,
+      "step": 49480
+    },
+    {
+      "epoch": 0.35824158324104033,
+      "grad_norm": 0.16531233489513397,
+      "learning_rate": 4.641765655425018e-06,
+      "loss": 0.9414,
+      "step": 49490
+    },
+    {
+      "epoch": 0.3583139699016265,
+      "grad_norm": 0.16358982026576996,
+      "learning_rate": 4.6416932687644325e-06,
+      "loss": 0.9555,
+      "step": 49500
+    },
+    {
+      "epoch": 0.35838635656221274,
+      "grad_norm": 0.16903631389141083,
+      "learning_rate": 4.641620882103846e-06,
+      "loss": 0.9489,
+      "step": 49510
+    },
+    {
+      "epoch": 0.3584587432227989,
+      "grad_norm": 0.1676577776670456,
+      "learning_rate": 4.64154849544326e-06,
+      "loss": 0.956,
+      "step": 49520
+    },
+    {
+      "epoch": 0.3585311298833851,
+      "grad_norm": 0.1515856236219406,
+      "learning_rate": 4.641476108782673e-06,
+      "loss": 0.9528,
+      "step": 49530
+    },
+    {
+      "epoch": 0.35860351654397127,
+      "grad_norm": 0.16165132820606232,
+      "learning_rate": 4.641403722122088e-06,
+      "loss": 0.9508,
+      "step": 49540
+    },
+    {
+      "epoch": 0.35867590320455744,
+      "grad_norm": 0.1639813631772995,
+      "learning_rate": 4.6413313354615014e-06,
+      "loss": 0.9484,
+      "step": 49550
+    },
+    {
+      "epoch": 0.3587482898651437,
+      "grad_norm": 0.1556709110736847,
+      "learning_rate": 4.641258948800915e-06,
+      "loss": 0.9368,
+      "step": 49560
+    },
+    {
+      "epoch": 0.35882067652572985,
+      "grad_norm": 0.16210544109344482,
+      "learning_rate": 4.641186562140329e-06,
+      "loss": 0.9479,
+      "step": 49570
+    },
+    {
+      "epoch": 0.35889306318631603,
+      "grad_norm": 0.16633464395999908,
+      "learning_rate": 4.641114175479743e-06,
+      "loss": 0.9517,
+      "step": 49580
+    },
+    {
+      "epoch": 0.3589654498469022,
+      "grad_norm": 0.16160385310649872,
+      "learning_rate": 4.641041788819157e-06,
+      "loss": 0.9523,
+      "step": 49590
+    },
+    {
+      "epoch": 0.3590378365074884,
+      "grad_norm": 0.16910237073898315,
+      "learning_rate": 4.64096940215857e-06,
+      "loss": 0.9453,
+      "step": 49600
+    },
+    {
+      "epoch": 0.35911022316807456,
+      "grad_norm": 0.20249851047992706,
+      "learning_rate": 4.640897015497984e-06,
+      "loss": 0.9446,
+      "step": 49610
+    },
+    {
+      "epoch": 0.3591826098286608,
+      "grad_norm": 0.16274423897266388,
+      "learning_rate": 4.6408246288373985e-06,
+      "loss": 0.951,
+      "step": 49620
+    },
+    {
+      "epoch": 0.35925499648924697,
+      "grad_norm": 0.16763456165790558,
+      "learning_rate": 4.640752242176812e-06,
+      "loss": 0.9379,
+      "step": 49630
+    },
+    {
+      "epoch": 0.35932738314983315,
+      "grad_norm": 0.15942569077014923,
+      "learning_rate": 4.640679855516226e-06,
+      "loss": 0.9472,
+      "step": 49640
+    },
+    {
+      "epoch": 0.3593997698104193,
+      "grad_norm": 0.17928923666477203,
+      "learning_rate": 4.640607468855639e-06,
+      "loss": 0.9577,
+      "step": 49650
+    },
+    {
+      "epoch": 0.3594721564710055,
+      "grad_norm": 0.16758985817432404,
+      "learning_rate": 4.640535082195054e-06,
+      "loss": 0.9494,
+      "step": 49660
+    },
+    {
+      "epoch": 0.35954454313159173,
+      "grad_norm": 0.18791483342647552,
+      "learning_rate": 4.640462695534467e-06,
+      "loss": 0.9423,
+      "step": 49670
+    },
+    {
+      "epoch": 0.3596169297921779,
+      "grad_norm": 0.19120506942272186,
+      "learning_rate": 4.640390308873881e-06,
+      "loss": 0.9602,
+      "step": 49680
+    },
+    {
+      "epoch": 0.3596893164527641,
+      "grad_norm": 0.17304813861846924,
+      "learning_rate": 4.640317922213295e-06,
+      "loss": 0.9485,
+      "step": 49690
+    },
+    {
+      "epoch": 0.35976170311335026,
+      "grad_norm": 0.16456131637096405,
+      "learning_rate": 4.640245535552709e-06,
+      "loss": 0.9379,
+      "step": 49700
+    },
+    {
+      "epoch": 0.35983408977393644,
+      "grad_norm": 0.1641353815793991,
+      "learning_rate": 4.640173148892123e-06,
+      "loss": 0.9485,
+      "step": 49710
+    },
+    {
+      "epoch": 0.35990647643452267,
+      "grad_norm": 0.17807172238826752,
+      "learning_rate": 4.640100762231536e-06,
+      "loss": 0.939,
+      "step": 49720
+    },
+    {
+      "epoch": 0.35997886309510885,
+      "grad_norm": 0.17257729172706604,
+      "learning_rate": 4.64002837557095e-06,
+      "loss": 0.9506,
+      "step": 49730
+    },
+    {
+      "epoch": 0.360051249755695,
+      "grad_norm": 0.16776679456233978,
+      "learning_rate": 4.639955988910364e-06,
+      "loss": 0.9543,
+      "step": 49740
+    },
+    {
+      "epoch": 0.3601236364162812,
+      "grad_norm": 0.16268086433410645,
+      "learning_rate": 4.639883602249778e-06,
+      "loss": 0.9506,
+      "step": 49750
+    },
+    {
+      "epoch": 0.3601960230768674,
+      "grad_norm": 0.18560020625591278,
+      "learning_rate": 4.639811215589192e-06,
+      "loss": 0.952,
+      "step": 49760
+    },
+    {
+      "epoch": 0.36026840973745355,
+      "grad_norm": 0.1764240264892578,
+      "learning_rate": 4.639738828928605e-06,
+      "loss": 0.9481,
+      "step": 49770
+    },
+    {
+      "epoch": 0.3603407963980398,
+      "grad_norm": 0.18193010985851288,
+      "learning_rate": 4.63966644226802e-06,
+      "loss": 0.9451,
+      "step": 49780
+    },
+    {
+      "epoch": 0.36041318305862596,
+      "grad_norm": 0.15828032791614532,
+      "learning_rate": 4.639594055607433e-06,
+      "loss": 0.9549,
+      "step": 49790
+    },
+    {
+      "epoch": 0.36048556971921214,
+      "grad_norm": 0.20637375116348267,
+      "learning_rate": 4.639521668946847e-06,
+      "loss": 0.9618,
+      "step": 49800
+    },
+    {
+      "epoch": 0.3605579563797983,
+      "grad_norm": 0.15375225245952606,
+      "learning_rate": 4.6394492822862605e-06,
+      "loss": 0.9475,
+      "step": 49810
+    },
+    {
+      "epoch": 0.3606303430403845,
+      "grad_norm": 0.16718937456607819,
+      "learning_rate": 4.639376895625675e-06,
+      "loss": 0.9517,
+      "step": 49820
+    },
+    {
+      "epoch": 0.3607027297009707,
+      "grad_norm": 0.1495947241783142,
+      "learning_rate": 4.639304508965089e-06,
+      "loss": 0.9442,
+      "step": 49830
+    },
+    {
+      "epoch": 0.3607751163615569,
+      "grad_norm": 0.16354574263095856,
+      "learning_rate": 4.639232122304502e-06,
+      "loss": 0.9516,
+      "step": 49840
+    },
+    {
+      "epoch": 0.3608475030221431,
+      "grad_norm": 0.16006174683570862,
+      "learning_rate": 4.639159735643916e-06,
+      "loss": 0.9571,
+      "step": 49850
+    },
+    {
+      "epoch": 0.36091988968272926,
+      "grad_norm": 0.1511203497648239,
+      "learning_rate": 4.63908734898333e-06,
+      "loss": 0.9498,
+      "step": 49860
+    },
+    {
+      "epoch": 0.36099227634331543,
+      "grad_norm": 0.16783535480499268,
+      "learning_rate": 4.639014962322744e-06,
+      "loss": 0.9507,
+      "step": 49870
+    },
+    {
+      "epoch": 0.36106466300390166,
+      "grad_norm": 0.1579425036907196,
+      "learning_rate": 4.6389425756621576e-06,
+      "loss": 0.945,
+      "step": 49880
+    },
+    {
+      "epoch": 0.36113704966448784,
+      "grad_norm": 0.16360943019390106,
+      "learning_rate": 4.638870189001571e-06,
+      "loss": 0.9373,
+      "step": 49890
+    },
+    {
+      "epoch": 0.361209436325074,
+      "grad_norm": 0.15648721158504486,
+      "learning_rate": 4.638797802340985e-06,
+      "loss": 0.9672,
+      "step": 49900
+    },
+    {
+      "epoch": 0.3612818229856602,
+      "grad_norm": 0.16514001786708832,
+      "learning_rate": 4.638725415680399e-06,
+      "loss": 0.9446,
+      "step": 49910
+    },
+    {
+      "epoch": 0.36135420964624637,
+      "grad_norm": 0.20961642265319824,
+      "learning_rate": 4.638653029019813e-06,
+      "loss": 0.9411,
+      "step": 49920
+    },
+    {
+      "epoch": 0.3614265963068326,
+      "grad_norm": 0.1592184156179428,
+      "learning_rate": 4.6385806423592265e-06,
+      "loss": 0.9524,
+      "step": 49930
+    },
+    {
+      "epoch": 0.3614989829674188,
+      "grad_norm": 0.16908025741577148,
+      "learning_rate": 4.63850825569864e-06,
+      "loss": 0.9332,
+      "step": 49940
+    },
+    {
+      "epoch": 0.36157136962800496,
+      "grad_norm": 0.15697824954986572,
+      "learning_rate": 4.6384358690380546e-06,
+      "loss": 0.9351,
+      "step": 49950
+    },
+    {
+      "epoch": 0.36164375628859113,
+      "grad_norm": 0.16380582749843597,
+      "learning_rate": 4.638363482377467e-06,
+      "loss": 0.9462,
+      "step": 49960
+    },
+    {
+      "epoch": 0.3617161429491773,
+      "grad_norm": 0.15993693470954895,
+      "learning_rate": 4.638291095716882e-06,
+      "loss": 0.9584,
+      "step": 49970
+    },
+    {
+      "epoch": 0.3617885296097635,
+      "grad_norm": 0.17109087109565735,
+      "learning_rate": 4.638218709056295e-06,
+      "loss": 0.9411,
+      "step": 49980
+    },
+    {
+      "epoch": 0.3618609162703497,
+      "grad_norm": 0.16840317845344543,
+      "learning_rate": 4.638146322395709e-06,
+      "loss": 0.9362,
+      "step": 49990
+    },
+    {
+      "epoch": 0.3619333029309359,
+      "grad_norm": 0.16615648567676544,
+      "learning_rate": 4.638073935735123e-06,
+      "loss": 0.9501,
+      "step": 50000
+    },
+    {
+      "epoch": 0.3620056895915221,
+      "grad_norm": 0.16905289888381958,
+      "learning_rate": 4.638001549074537e-06,
+      "loss": 0.9524,
+      "step": 50010
+    },
+    {
+      "epoch": 0.36207807625210825,
+      "grad_norm": 0.17198826372623444,
+      "learning_rate": 4.637929162413951e-06,
+      "loss": 0.9435,
+      "step": 50020
+    },
+    {
+      "epoch": 0.3621504629126944,
+      "grad_norm": 0.17047207057476044,
+      "learning_rate": 4.637856775753364e-06,
+      "loss": 0.9492,
+      "step": 50030
+    },
+    {
+      "epoch": 0.36222284957328066,
+      "grad_norm": 0.18584772944450378,
+      "learning_rate": 4.637784389092778e-06,
+      "loss": 0.9482,
+      "step": 50040
+    },
+    {
+      "epoch": 0.36229523623386684,
+      "grad_norm": 0.16773472726345062,
+      "learning_rate": 4.637712002432192e-06,
+      "loss": 0.9572,
+      "step": 50050
+    },
+    {
+      "epoch": 0.362367622894453,
+      "grad_norm": 0.16679717600345612,
+      "learning_rate": 4.637639615771606e-06,
+      "loss": 0.9529,
+      "step": 50060
+    },
+    {
+      "epoch": 0.3624400095550392,
+      "grad_norm": 0.15634585916996002,
+      "learning_rate": 4.63756722911102e-06,
+      "loss": 0.9401,
+      "step": 50070
+    },
+    {
+      "epoch": 0.36251239621562537,
+      "grad_norm": 0.15569652616977692,
+      "learning_rate": 4.637494842450433e-06,
+      "loss": 0.9416,
+      "step": 50080
+    },
+    {
+      "epoch": 0.3625847828762116,
+      "grad_norm": 0.16276825964450836,
+      "learning_rate": 4.637422455789847e-06,
+      "loss": 0.9422,
+      "step": 50090
+    },
+    {
+      "epoch": 0.3626571695367978,
+      "grad_norm": 0.16607090830802917,
+      "learning_rate": 4.637350069129261e-06,
+      "loss": 0.9587,
+      "step": 50100
+    },
+    {
+      "epoch": 0.36272955619738395,
+      "grad_norm": 0.17144577205181122,
+      "learning_rate": 4.637277682468675e-06,
+      "loss": 0.9356,
+      "step": 50110
+    },
+    {
+      "epoch": 0.36280194285797013,
+      "grad_norm": 0.16335241496562958,
+      "learning_rate": 4.637205295808089e-06,
+      "loss": 0.9581,
+      "step": 50120
+    },
+    {
+      "epoch": 0.3628743295185563,
+      "grad_norm": 0.17196884751319885,
+      "learning_rate": 4.637132909147502e-06,
+      "loss": 0.9345,
+      "step": 50130
+    },
+    {
+      "epoch": 0.3629467161791425,
+      "grad_norm": 0.1646929383277893,
+      "learning_rate": 4.637060522486917e-06,
+      "loss": 0.9388,
+      "step": 50140
+    },
+    {
+      "epoch": 0.3630191028397287,
+      "grad_norm": 0.16134804487228394,
+      "learning_rate": 4.63698813582633e-06,
+      "loss": 0.9546,
+      "step": 50150
+    },
+    {
+      "epoch": 0.3630914895003149,
+      "grad_norm": 0.16641758382320404,
+      "learning_rate": 4.636915749165744e-06,
+      "loss": 0.9681,
+      "step": 50160
+    },
+    {
+      "epoch": 0.36316387616090107,
+      "grad_norm": 0.1564699113368988,
+      "learning_rate": 4.6368433625051575e-06,
+      "loss": 0.9527,
+      "step": 50170
+    },
+    {
+      "epoch": 0.36323626282148724,
+      "grad_norm": 0.16455498337745667,
+      "learning_rate": 4.636770975844572e-06,
+      "loss": 0.9407,
+      "step": 50180
+    },
+    {
+      "epoch": 0.3633086494820734,
+      "grad_norm": 0.16031110286712646,
+      "learning_rate": 4.636698589183986e-06,
+      "loss": 0.9418,
+      "step": 50190
+    },
+    {
+      "epoch": 0.36338103614265965,
+      "grad_norm": 0.1539280265569687,
+      "learning_rate": 4.636626202523399e-06,
+      "loss": 0.9496,
+      "step": 50200
+    },
+    {
+      "epoch": 0.36345342280324583,
+      "grad_norm": 0.20575940608978271,
+      "learning_rate": 4.636553815862813e-06,
+      "loss": 0.9332,
+      "step": 50210
+    },
+    {
+      "epoch": 0.363525809463832,
+      "grad_norm": 0.17900900542736053,
+      "learning_rate": 4.636481429202227e-06,
+      "loss": 0.9533,
+      "step": 50220
+    },
+    {
+      "epoch": 0.3635981961244182,
+      "grad_norm": 0.1708286702632904,
+      "learning_rate": 4.636409042541641e-06,
+      "loss": 0.9504,
+      "step": 50230
+    },
+    {
+      "epoch": 0.36367058278500436,
+      "grad_norm": 0.16686797142028809,
+      "learning_rate": 4.6363366558810545e-06,
+      "loss": 0.951,
+      "step": 50240
+    },
+    {
+      "epoch": 0.3637429694455906,
+      "grad_norm": 0.1779949814081192,
+      "learning_rate": 4.636264269220468e-06,
+      "loss": 0.954,
+      "step": 50250
+    },
+    {
+      "epoch": 0.36381535610617677,
+      "grad_norm": 0.16261711716651917,
+      "learning_rate": 4.636191882559883e-06,
+      "loss": 0.9552,
+      "step": 50260
+    },
+    {
+      "epoch": 0.36388774276676294,
+      "grad_norm": 0.15804466605186462,
+      "learning_rate": 4.636119495899296e-06,
+      "loss": 0.952,
+      "step": 50270
+    },
+    {
+      "epoch": 0.3639601294273491,
+      "grad_norm": 0.18854446709156036,
+      "learning_rate": 4.63604710923871e-06,
+      "loss": 0.937,
+      "step": 50280
+    },
+    {
+      "epoch": 0.3640325160879353,
+      "grad_norm": 0.16362988948822021,
+      "learning_rate": 4.6359747225781234e-06,
+      "loss": 0.9439,
+      "step": 50290
+    },
+    {
+      "epoch": 0.3641049027485215,
+      "grad_norm": 0.1906929761171341,
+      "learning_rate": 4.635902335917538e-06,
+      "loss": 0.9522,
+      "step": 50300
+    },
+    {
+      "epoch": 0.3641772894091077,
+      "grad_norm": 0.1719408482313156,
+      "learning_rate": 4.6358299492569515e-06,
+      "loss": 0.9409,
+      "step": 50310
+    },
+    {
+      "epoch": 0.3642496760696939,
+      "grad_norm": 0.17714013159275055,
+      "learning_rate": 4.635757562596365e-06,
+      "loss": 0.9425,
+      "step": 50320
+    },
+    {
+      "epoch": 0.36432206273028006,
+      "grad_norm": 0.1718074232339859,
+      "learning_rate": 4.635685175935779e-06,
+      "loss": 0.9431,
+      "step": 50330
+    },
+    {
+      "epoch": 0.36439444939086624,
+      "grad_norm": 0.16074173152446747,
+      "learning_rate": 4.635612789275193e-06,
+      "loss": 0.9598,
+      "step": 50340
+    },
+    {
+      "epoch": 0.3644668360514524,
+      "grad_norm": 0.1573699712753296,
+      "learning_rate": 4.635540402614607e-06,
+      "loss": 0.9416,
+      "step": 50350
+    },
+    {
+      "epoch": 0.36453922271203865,
+      "grad_norm": 0.15923991799354553,
+      "learning_rate": 4.6354680159540205e-06,
+      "loss": 0.9591,
+      "step": 50360
+    },
+    {
+      "epoch": 0.3646116093726248,
+      "grad_norm": 0.17145873606204987,
+      "learning_rate": 4.635395629293434e-06,
+      "loss": 0.9399,
+      "step": 50370
+    },
+    {
+      "epoch": 0.364683996033211,
+      "grad_norm": 0.16001589596271515,
+      "learning_rate": 4.6353232426328485e-06,
+      "loss": 0.9558,
+      "step": 50380
+    },
+    {
+      "epoch": 0.3647563826937972,
+      "grad_norm": 0.16999055445194244,
+      "learning_rate": 4.635250855972262e-06,
+      "loss": 0.9505,
+      "step": 50390
+    },
+    {
+      "epoch": 0.36482876935438335,
+      "grad_norm": 0.1643807590007782,
+      "learning_rate": 4.635178469311676e-06,
+      "loss": 0.9385,
+      "step": 50400
+    },
+    {
+      "epoch": 0.3649011560149696,
+      "grad_norm": 0.1788293868303299,
+      "learning_rate": 4.635106082651089e-06,
+      "loss": 0.9312,
+      "step": 50410
+    },
+    {
+      "epoch": 0.36497354267555576,
+      "grad_norm": 0.17690715193748474,
+      "learning_rate": 4.635033695990504e-06,
+      "loss": 0.9524,
+      "step": 50420
+    },
+    {
+      "epoch": 0.36504592933614194,
+      "grad_norm": 0.17854566872119904,
+      "learning_rate": 4.6349613093299175e-06,
+      "loss": 0.952,
+      "step": 50430
+    },
+    {
+      "epoch": 0.3651183159967281,
+      "grad_norm": 0.4458518326282501,
+      "learning_rate": 4.634888922669331e-06,
+      "loss": 0.9479,
+      "step": 50440
+    },
+    {
+      "epoch": 0.3651907026573143,
+      "grad_norm": 0.15616470575332642,
+      "learning_rate": 4.634816536008745e-06,
+      "loss": 0.9488,
+      "step": 50450
+    },
+    {
+      "epoch": 0.3652630893179005,
+      "grad_norm": 0.16164493560791016,
+      "learning_rate": 4.634744149348159e-06,
+      "loss": 0.9507,
+      "step": 50460
+    },
+    {
+      "epoch": 0.3653354759784867,
+      "grad_norm": 0.16681145131587982,
+      "learning_rate": 4.634671762687573e-06,
+      "loss": 0.9473,
+      "step": 50470
+    },
+    {
+      "epoch": 0.3654078626390729,
+      "grad_norm": 0.1800214797258377,
+      "learning_rate": 4.634599376026986e-06,
+      "loss": 0.9655,
+      "step": 50480
+    },
+    {
+      "epoch": 0.36548024929965905,
+      "grad_norm": 0.1790829747915268,
+      "learning_rate": 4.6345269893664e-06,
+      "loss": 0.9581,
+      "step": 50490
+    },
+    {
+      "epoch": 0.36555263596024523,
+      "grad_norm": 0.1491861194372177,
+      "learning_rate": 4.634454602705814e-06,
+      "loss": 0.9409,
+      "step": 50500
+    },
+    {
+      "epoch": 0.3656250226208314,
+      "grad_norm": 0.15858380496501923,
+      "learning_rate": 4.634382216045227e-06,
+      "loss": 0.9392,
+      "step": 50510
+    },
+    {
+      "epoch": 0.36569740928141764,
+      "grad_norm": 0.17849485576152802,
+      "learning_rate": 4.634309829384641e-06,
+      "loss": 0.9595,
+      "step": 50520
+    },
+    {
+      "epoch": 0.3657697959420038,
+      "grad_norm": 0.16567130386829376,
+      "learning_rate": 4.634237442724055e-06,
+      "loss": 0.9496,
+      "step": 50530
+    },
+    {
+      "epoch": 0.36584218260259,
+      "grad_norm": 0.22058269381523132,
+      "learning_rate": 4.634165056063469e-06,
+      "loss": 0.9507,
+      "step": 50540
+    },
+    {
+      "epoch": 0.36591456926317617,
+      "grad_norm": 0.16320207715034485,
+      "learning_rate": 4.6340926694028825e-06,
+      "loss": 0.9408,
+      "step": 50550
+    },
+    {
+      "epoch": 0.36598695592376235,
+      "grad_norm": 0.19167020916938782,
+      "learning_rate": 4.634020282742296e-06,
+      "loss": 0.9467,
+      "step": 50560
+    },
+    {
+      "epoch": 0.3660593425843486,
+      "grad_norm": 0.17240388691425323,
+      "learning_rate": 4.633947896081711e-06,
+      "loss": 0.9623,
+      "step": 50570
+    },
+    {
+      "epoch": 0.36613172924493476,
+      "grad_norm": 0.20315563678741455,
+      "learning_rate": 4.633875509421124e-06,
+      "loss": 0.9401,
+      "step": 50580
+    },
+    {
+      "epoch": 0.36620411590552093,
+      "grad_norm": 0.1570211499929428,
+      "learning_rate": 4.633803122760538e-06,
+      "loss": 0.9451,
+      "step": 50590
+    },
+    {
+      "epoch": 0.3662765025661071,
+      "grad_norm": 0.18627873063087463,
+      "learning_rate": 4.6337307360999515e-06,
+      "loss": 0.9548,
+      "step": 50600
+    },
+    {
+      "epoch": 0.3663488892266933,
+      "grad_norm": 0.19180041551589966,
+      "learning_rate": 4.633658349439366e-06,
+      "loss": 0.9445,
+      "step": 50610
+    },
+    {
+      "epoch": 0.3664212758872795,
+      "grad_norm": 0.19508033990859985,
+      "learning_rate": 4.6335859627787796e-06,
+      "loss": 0.9493,
+      "step": 50620
+    },
+    {
+      "epoch": 0.3664936625478657,
+      "grad_norm": 0.18519370257854462,
+      "learning_rate": 4.633513576118193e-06,
+      "loss": 0.9308,
+      "step": 50630
+    },
+    {
+      "epoch": 0.36656604920845187,
+      "grad_norm": 0.1680838167667389,
+      "learning_rate": 4.633441189457607e-06,
+      "loss": 0.9335,
+      "step": 50640
+    },
+    {
+      "epoch": 0.36663843586903805,
+      "grad_norm": 0.16373924911022186,
+      "learning_rate": 4.633368802797021e-06,
+      "loss": 0.9413,
+      "step": 50650
+    },
+    {
+      "epoch": 0.3667108225296242,
+      "grad_norm": 0.16160275042057037,
+      "learning_rate": 4.633296416136435e-06,
+      "loss": 0.9487,
+      "step": 50660
+    },
+    {
+      "epoch": 0.3667832091902104,
+      "grad_norm": 0.1721445471048355,
+      "learning_rate": 4.6332240294758485e-06,
+      "loss": 0.9351,
+      "step": 50670
+    },
+    {
+      "epoch": 0.36685559585079663,
+      "grad_norm": 0.1686573028564453,
+      "learning_rate": 4.633151642815262e-06,
+      "loss": 0.9441,
+      "step": 50680
+    },
+    {
+      "epoch": 0.3669279825113828,
+      "grad_norm": 0.15622854232788086,
+      "learning_rate": 4.633079256154676e-06,
+      "loss": 0.9586,
+      "step": 50690
+    },
+    {
+      "epoch": 0.367000369171969,
+      "grad_norm": 0.14963869750499725,
+      "learning_rate": 4.63300686949409e-06,
+      "loss": 0.9496,
+      "step": 50700
+    },
+    {
+      "epoch": 0.36707275583255516,
+      "grad_norm": 0.17733648419380188,
+      "learning_rate": 4.632934482833504e-06,
+      "loss": 0.9453,
+      "step": 50710
+    },
+    {
+      "epoch": 0.36714514249314134,
+      "grad_norm": 0.15873117744922638,
+      "learning_rate": 4.632862096172917e-06,
+      "loss": 0.9551,
+      "step": 50720
+    },
+    {
+      "epoch": 0.3672175291537276,
+      "grad_norm": 0.15206924080848694,
+      "learning_rate": 4.632789709512331e-06,
+      "loss": 0.9318,
+      "step": 50730
+    },
+    {
+      "epoch": 0.36728991581431375,
+      "grad_norm": 0.16713659465312958,
+      "learning_rate": 4.6327173228517455e-06,
+      "loss": 0.9512,
+      "step": 50740
+    },
+    {
+      "epoch": 0.3673623024748999,
+      "grad_norm": 0.1614990532398224,
+      "learning_rate": 4.632644936191159e-06,
+      "loss": 0.9528,
+      "step": 50750
+    },
+    {
+      "epoch": 0.3674346891354861,
+      "grad_norm": 0.16960258781909943,
+      "learning_rate": 4.632572549530573e-06,
+      "loss": 0.9465,
+      "step": 50760
+    },
+    {
+      "epoch": 0.3675070757960723,
+      "grad_norm": 0.19546057283878326,
+      "learning_rate": 4.632500162869986e-06,
+      "loss": 0.9463,
+      "step": 50770
+    },
+    {
+      "epoch": 0.3675794624566585,
+      "grad_norm": 0.1588112711906433,
+      "learning_rate": 4.632427776209401e-06,
+      "loss": 0.9295,
+      "step": 50780
+    },
+    {
+      "epoch": 0.3676518491172447,
+      "grad_norm": 0.17281554639339447,
+      "learning_rate": 4.632355389548814e-06,
+      "loss": 0.956,
+      "step": 50790
+    },
+    {
+      "epoch": 0.36772423577783087,
+      "grad_norm": 0.15607163310050964,
+      "learning_rate": 4.632283002888228e-06,
+      "loss": 0.9452,
+      "step": 50800
+    },
+    {
+      "epoch": 0.36779662243841704,
+      "grad_norm": 0.16770051419734955,
+      "learning_rate": 4.632210616227642e-06,
+      "loss": 0.9582,
+      "step": 50810
+    },
+    {
+      "epoch": 0.3678690090990032,
+      "grad_norm": 0.15941555798053741,
+      "learning_rate": 4.632138229567056e-06,
+      "loss": 0.9394,
+      "step": 50820
+    },
+    {
+      "epoch": 0.3679413957595894,
+      "grad_norm": 0.1637001931667328,
+      "learning_rate": 4.63206584290647e-06,
+      "loss": 0.9377,
+      "step": 50830
+    },
+    {
+      "epoch": 0.36801378242017563,
+      "grad_norm": 0.17539703845977783,
+      "learning_rate": 4.631993456245883e-06,
+      "loss": 0.9532,
+      "step": 50840
+    },
+    {
+      "epoch": 0.3680861690807618,
+      "grad_norm": 0.15488503873348236,
+      "learning_rate": 4.631921069585297e-06,
+      "loss": 0.9572,
+      "step": 50850
+    },
+    {
+      "epoch": 0.368158555741348,
+      "grad_norm": 0.17145054042339325,
+      "learning_rate": 4.6318486829247114e-06,
+      "loss": 0.9558,
+      "step": 50860
+    },
+    {
+      "epoch": 0.36823094240193416,
+      "grad_norm": 0.16223368048667908,
+      "learning_rate": 4.631776296264125e-06,
+      "loss": 0.9489,
+      "step": 50870
+    },
+    {
+      "epoch": 0.36830332906252033,
+      "grad_norm": 0.17038589715957642,
+      "learning_rate": 4.631703909603539e-06,
+      "loss": 0.9607,
+      "step": 50880
+    },
+    {
+      "epoch": 0.36837571572310657,
+      "grad_norm": 0.17411848902702332,
+      "learning_rate": 4.631631522942952e-06,
+      "loss": 0.9467,
+      "step": 50890
+    },
+    {
+      "epoch": 0.36844810238369274,
+      "grad_norm": 0.15501435101032257,
+      "learning_rate": 4.631559136282367e-06,
+      "loss": 0.946,
+      "step": 50900
+    },
+    {
+      "epoch": 0.3685204890442789,
+      "grad_norm": 0.15863391757011414,
+      "learning_rate": 4.63148674962178e-06,
+      "loss": 0.9441,
+      "step": 50910
+    },
+    {
+      "epoch": 0.3685928757048651,
+      "grad_norm": 0.17681151628494263,
+      "learning_rate": 4.631414362961194e-06,
+      "loss": 0.9548,
+      "step": 50920
+    },
+    {
+      "epoch": 0.3686652623654513,
+      "grad_norm": 0.1619083285331726,
+      "learning_rate": 4.631341976300608e-06,
+      "loss": 0.9488,
+      "step": 50930
+    },
+    {
+      "epoch": 0.3687376490260375,
+      "grad_norm": 0.16921238601207733,
+      "learning_rate": 4.631269589640022e-06,
+      "loss": 0.9414,
+      "step": 50940
+    },
+    {
+      "epoch": 0.3688100356866237,
+      "grad_norm": 0.16636481881141663,
+      "learning_rate": 4.631197202979436e-06,
+      "loss": 0.941,
+      "step": 50950
+    },
+    {
+      "epoch": 0.36888242234720986,
+      "grad_norm": 0.16682793200016022,
+      "learning_rate": 4.631124816318849e-06,
+      "loss": 0.9447,
+      "step": 50960
+    },
+    {
+      "epoch": 0.36895480900779604,
+      "grad_norm": 0.1645735502243042,
+      "learning_rate": 4.631052429658263e-06,
+      "loss": 0.9495,
+      "step": 50970
+    },
+    {
+      "epoch": 0.3690271956683822,
+      "grad_norm": 0.16850468516349792,
+      "learning_rate": 4.630980042997677e-06,
+      "loss": 0.9512,
+      "step": 50980
+    },
+    {
+      "epoch": 0.36909958232896845,
+      "grad_norm": 0.16100184619426727,
+      "learning_rate": 4.630907656337091e-06,
+      "loss": 0.9483,
+      "step": 50990
+    },
+    {
+      "epoch": 0.3691719689895546,
+      "grad_norm": 0.1832258701324463,
+      "learning_rate": 4.630835269676505e-06,
+      "loss": 0.9398,
+      "step": 51000
+    },
+    {
+      "epoch": 0.3692443556501408,
+      "grad_norm": 0.1679748147726059,
+      "learning_rate": 4.630762883015918e-06,
+      "loss": 0.9545,
+      "step": 51010
+    },
+    {
+      "epoch": 0.369316742310727,
+      "grad_norm": 0.5284795165061951,
+      "learning_rate": 4.630690496355332e-06,
+      "loss": 0.9481,
+      "step": 51020
+    },
+    {
+      "epoch": 0.36938912897131315,
+      "grad_norm": 0.16089926660060883,
+      "learning_rate": 4.6306181096947454e-06,
+      "loss": 0.9499,
+      "step": 51030
+    },
+    {
+      "epoch": 0.36946151563189933,
+      "grad_norm": 0.16643081605434418,
+      "learning_rate": 4.630545723034159e-06,
+      "loss": 0.948,
+      "step": 51040
+    },
+    {
+      "epoch": 0.36953390229248556,
+      "grad_norm": 0.24914591014385223,
+      "learning_rate": 4.6304733363735735e-06,
+      "loss": 0.9461,
+      "step": 51050
+    },
+    {
+      "epoch": 0.36960628895307174,
+      "grad_norm": 0.16817229986190796,
+      "learning_rate": 4.630400949712987e-06,
+      "loss": 0.9431,
+      "step": 51060
+    },
+    {
+      "epoch": 0.3696786756136579,
+      "grad_norm": 0.15578976273536682,
+      "learning_rate": 4.630328563052401e-06,
+      "loss": 0.9523,
+      "step": 51070
+    },
+    {
+      "epoch": 0.3697510622742441,
+      "grad_norm": 0.1646818220615387,
+      "learning_rate": 4.630256176391814e-06,
+      "loss": 0.9489,
+      "step": 51080
+    },
+    {
+      "epoch": 0.36982344893483027,
+      "grad_norm": 0.15397506952285767,
+      "learning_rate": 4.630183789731229e-06,
+      "loss": 0.939,
+      "step": 51090
+    },
+    {
+      "epoch": 0.3698958355954165,
+      "grad_norm": 0.1653456836938858,
+      "learning_rate": 4.6301114030706424e-06,
+      "loss": 0.9552,
+      "step": 51100
+    },
+    {
+      "epoch": 0.3699682222560027,
+      "grad_norm": 0.1986481249332428,
+      "learning_rate": 4.630039016410056e-06,
+      "loss": 0.9519,
+      "step": 51110
+    },
+    {
+      "epoch": 0.37004060891658885,
+      "grad_norm": 0.17512935400009155,
+      "learning_rate": 4.62996662974947e-06,
+      "loss": 0.9582,
+      "step": 51120
+    },
+    {
+      "epoch": 0.37011299557717503,
+      "grad_norm": 0.17302803695201874,
+      "learning_rate": 4.629894243088884e-06,
+      "loss": 0.9275,
+      "step": 51130
+    },
+    {
+      "epoch": 0.3701853822377612,
+      "grad_norm": 0.16957373917102814,
+      "learning_rate": 4.629821856428298e-06,
+      "loss": 0.9455,
+      "step": 51140
+    },
+    {
+      "epoch": 0.37025776889834744,
+      "grad_norm": 0.1609341949224472,
+      "learning_rate": 4.629749469767711e-06,
+      "loss": 0.9449,
+      "step": 51150
+    },
+    {
+      "epoch": 0.3703301555589336,
+      "grad_norm": 0.2060326784849167,
+      "learning_rate": 4.629677083107125e-06,
+      "loss": 0.9487,
+      "step": 51160
+    },
+    {
+      "epoch": 0.3704025422195198,
+      "grad_norm": 0.15667761862277985,
+      "learning_rate": 4.6296046964465395e-06,
+      "loss": 0.9421,
+      "step": 51170
+    },
+    {
+      "epoch": 0.37047492888010597,
+      "grad_norm": 0.15692569315433502,
+      "learning_rate": 4.629532309785953e-06,
+      "loss": 0.9383,
+      "step": 51180
+    },
+    {
+      "epoch": 0.37054731554069215,
+      "grad_norm": 0.15926668047904968,
+      "learning_rate": 4.629459923125367e-06,
+      "loss": 0.9483,
+      "step": 51190
+    },
+    {
+      "epoch": 0.3706197022012783,
+      "grad_norm": 0.16595801711082458,
+      "learning_rate": 4.62938753646478e-06,
+      "loss": 0.9489,
+      "step": 51200
+    },
+    {
+      "epoch": 0.37069208886186455,
+      "grad_norm": 0.1648326963186264,
+      "learning_rate": 4.629315149804195e-06,
+      "loss": 0.9634,
+      "step": 51210
+    },
+    {
+      "epoch": 0.37076447552245073,
+      "grad_norm": 0.15535977482795715,
+      "learning_rate": 4.629242763143608e-06,
+      "loss": 0.9404,
+      "step": 51220
+    },
+    {
+      "epoch": 0.3708368621830369,
+      "grad_norm": 0.1696339249610901,
+      "learning_rate": 4.629170376483022e-06,
+      "loss": 0.9539,
+      "step": 51230
+    },
+    {
+      "epoch": 0.3709092488436231,
+      "grad_norm": 0.15960168838500977,
+      "learning_rate": 4.629097989822436e-06,
+      "loss": 0.9377,
+      "step": 51240
+    },
+    {
+      "epoch": 0.37098163550420926,
+      "grad_norm": 0.20475220680236816,
+      "learning_rate": 4.62902560316185e-06,
+      "loss": 0.9529,
+      "step": 51250
+    },
+    {
+      "epoch": 0.3710540221647955,
+      "grad_norm": 0.16864748299121857,
+      "learning_rate": 4.628953216501264e-06,
+      "loss": 0.9446,
+      "step": 51260
+    },
+    {
+      "epoch": 0.37112640882538167,
+      "grad_norm": 0.1571142077445984,
+      "learning_rate": 4.628880829840677e-06,
+      "loss": 0.9558,
+      "step": 51270
+    },
+    {
+      "epoch": 0.37119879548596785,
+      "grad_norm": 0.19135212898254395,
+      "learning_rate": 4.628808443180091e-06,
+      "loss": 0.9484,
+      "step": 51280
+    },
+    {
+      "epoch": 0.371271182146554,
+      "grad_norm": 0.15567857027053833,
+      "learning_rate": 4.628736056519505e-06,
+      "loss": 0.9503,
+      "step": 51290
+    },
+    {
+      "epoch": 0.3713435688071402,
+      "grad_norm": 0.16534672677516937,
+      "learning_rate": 4.628663669858919e-06,
+      "loss": 0.9329,
+      "step": 51300
+    },
+    {
+      "epoch": 0.37141595546772643,
+      "grad_norm": 0.16259850561618805,
+      "learning_rate": 4.628591283198333e-06,
+      "loss": 0.9419,
+      "step": 51310
+    },
+    {
+      "epoch": 0.3714883421283126,
+      "grad_norm": 0.20146603882312775,
+      "learning_rate": 4.628518896537746e-06,
+      "loss": 0.956,
+      "step": 51320
+    },
+    {
+      "epoch": 0.3715607287888988,
+      "grad_norm": 0.18213625252246857,
+      "learning_rate": 4.62844650987716e-06,
+      "loss": 0.9639,
+      "step": 51330
+    },
+    {
+      "epoch": 0.37163311544948496,
+      "grad_norm": 0.15779289603233337,
+      "learning_rate": 4.628374123216574e-06,
+      "loss": 0.9425,
+      "step": 51340
+    },
+    {
+      "epoch": 0.37170550211007114,
+      "grad_norm": 0.16919735074043274,
+      "learning_rate": 4.628301736555988e-06,
+      "loss": 0.9501,
+      "step": 51350
+    },
+    {
+      "epoch": 0.3717778887706573,
+      "grad_norm": 0.1676865667104721,
+      "learning_rate": 4.6282293498954016e-06,
+      "loss": 0.9453,
+      "step": 51360
+    },
+    {
+      "epoch": 0.37185027543124355,
+      "grad_norm": 0.15803247690200806,
+      "learning_rate": 4.628156963234815e-06,
+      "loss": 0.9435,
+      "step": 51370
+    },
+    {
+      "epoch": 0.3719226620918297,
+      "grad_norm": 0.1553627997636795,
+      "learning_rate": 4.62808457657423e-06,
+      "loss": 0.9638,
+      "step": 51380
+    },
+    {
+      "epoch": 0.3719950487524159,
+      "grad_norm": 0.16494029760360718,
+      "learning_rate": 4.628012189913643e-06,
+      "loss": 0.9366,
+      "step": 51390
+    },
+    {
+      "epoch": 0.3720674354130021,
+      "grad_norm": 0.16437461972236633,
+      "learning_rate": 4.627939803253057e-06,
+      "loss": 0.94,
+      "step": 51400
+    },
+    {
+      "epoch": 0.37213982207358826,
+      "grad_norm": 0.15786989033222198,
+      "learning_rate": 4.6278674165924705e-06,
+      "loss": 0.9574,
+      "step": 51410
+    },
+    {
+      "epoch": 0.3722122087341745,
+      "grad_norm": 0.16579070687294006,
+      "learning_rate": 4.627795029931885e-06,
+      "loss": 0.949,
+      "step": 51420
+    },
+    {
+      "epoch": 0.37228459539476066,
+      "grad_norm": 0.16631585359573364,
+      "learning_rate": 4.6277226432712986e-06,
+      "loss": 0.9393,
+      "step": 51430
+    },
+    {
+      "epoch": 0.37235698205534684,
+      "grad_norm": 0.21473245322704315,
+      "learning_rate": 4.627650256610712e-06,
+      "loss": 0.9482,
+      "step": 51440
+    },
+    {
+      "epoch": 0.372429368715933,
+      "grad_norm": 0.17821331322193146,
+      "learning_rate": 4.627577869950126e-06,
+      "loss": 0.9511,
+      "step": 51450
+    },
+    {
+      "epoch": 0.3725017553765192,
+      "grad_norm": 0.16437208652496338,
+      "learning_rate": 4.62750548328954e-06,
+      "loss": 0.9522,
+      "step": 51460
+    },
+    {
+      "epoch": 0.3725741420371054,
+      "grad_norm": 0.1748124063014984,
+      "learning_rate": 4.627433096628954e-06,
+      "loss": 0.951,
+      "step": 51470
+    },
+    {
+      "epoch": 0.3726465286976916,
+      "grad_norm": 0.1691410392522812,
+      "learning_rate": 4.6273607099683675e-06,
+      "loss": 0.9452,
+      "step": 51480
+    },
+    {
+      "epoch": 0.3727189153582778,
+      "grad_norm": 0.15897352993488312,
+      "learning_rate": 4.627288323307781e-06,
+      "loss": 0.9503,
+      "step": 51490
+    },
+    {
+      "epoch": 0.37279130201886396,
+      "grad_norm": 0.17863501608371735,
+      "learning_rate": 4.6272159366471956e-06,
+      "loss": 0.9485,
+      "step": 51500
+    },
+    {
+      "epoch": 0.37286368867945013,
+      "grad_norm": 0.1577998846769333,
+      "learning_rate": 4.627143549986609e-06,
+      "loss": 0.9468,
+      "step": 51510
+    },
+    {
+      "epoch": 0.3729360753400363,
+      "grad_norm": 0.17091219127178192,
+      "learning_rate": 4.627071163326023e-06,
+      "loss": 0.949,
+      "step": 51520
+    },
+    {
+      "epoch": 0.37300846200062254,
+      "grad_norm": 0.16450020670890808,
+      "learning_rate": 4.626998776665436e-06,
+      "loss": 0.9375,
+      "step": 51530
+    },
+    {
+      "epoch": 0.3730808486612087,
+      "grad_norm": 0.20446856319904327,
+      "learning_rate": 4.626926390004851e-06,
+      "loss": 0.945,
+      "step": 51540
+    },
+    {
+      "epoch": 0.3731532353217949,
+      "grad_norm": 0.18482069671154022,
+      "learning_rate": 4.626854003344264e-06,
+      "loss": 0.9365,
+      "step": 51550
+    },
+    {
+      "epoch": 0.3732256219823811,
+      "grad_norm": 0.2597537934780121,
+      "learning_rate": 4.626781616683677e-06,
+      "loss": 0.9474,
+      "step": 51560
+    },
+    {
+      "epoch": 0.37329800864296725,
+      "grad_norm": 0.16503006219863892,
+      "learning_rate": 4.626709230023092e-06,
+      "loss": 0.9384,
+      "step": 51570
+    },
+    {
+      "epoch": 0.3733703953035535,
+      "grad_norm": 0.14666348695755005,
+      "learning_rate": 4.626636843362505e-06,
+      "loss": 0.9408,
+      "step": 51580
+    },
+    {
+      "epoch": 0.37344278196413966,
+      "grad_norm": 0.1652412712574005,
+      "learning_rate": 4.626564456701919e-06,
+      "loss": 0.9559,
+      "step": 51590
+    },
+    {
+      "epoch": 0.37351516862472584,
+      "grad_norm": 0.15990935266017914,
+      "learning_rate": 4.626492070041333e-06,
+      "loss": 0.9363,
+      "step": 51600
+    },
+    {
+      "epoch": 0.373587555285312,
+      "grad_norm": 0.16256429255008698,
+      "learning_rate": 4.626419683380747e-06,
+      "loss": 0.947,
+      "step": 51610
+    },
+    {
+      "epoch": 0.3736599419458982,
+      "grad_norm": 0.166569322347641,
+      "learning_rate": 4.626347296720161e-06,
+      "loss": 0.9548,
+      "step": 51620
+    },
+    {
+      "epoch": 0.3737323286064844,
+      "grad_norm": 0.18789708614349365,
+      "learning_rate": 4.626274910059574e-06,
+      "loss": 0.9502,
+      "step": 51630
+    },
+    {
+      "epoch": 0.3738047152670706,
+      "grad_norm": 0.16426165401935577,
+      "learning_rate": 4.626202523398988e-06,
+      "loss": 0.942,
+      "step": 51640
+    },
+    {
+      "epoch": 0.3738771019276568,
+      "grad_norm": 0.15513013303279877,
+      "learning_rate": 4.626130136738402e-06,
+      "loss": 0.9541,
+      "step": 51650
+    },
+    {
+      "epoch": 0.37394948858824295,
+      "grad_norm": 0.1891675591468811,
+      "learning_rate": 4.626057750077816e-06,
+      "loss": 0.956,
+      "step": 51660
+    },
+    {
+      "epoch": 0.3740218752488291,
+      "grad_norm": 0.17240507900714874,
+      "learning_rate": 4.62598536341723e-06,
+      "loss": 0.9564,
+      "step": 51670
+    },
+    {
+      "epoch": 0.37409426190941536,
+      "grad_norm": 0.16830044984817505,
+      "learning_rate": 4.625912976756643e-06,
+      "loss": 0.9326,
+      "step": 51680
+    },
+    {
+      "epoch": 0.37416664857000154,
+      "grad_norm": 0.16331635415554047,
+      "learning_rate": 4.625840590096058e-06,
+      "loss": 0.945,
+      "step": 51690
+    },
+    {
+      "epoch": 0.3742390352305877,
+      "grad_norm": 0.15753015875816345,
+      "learning_rate": 4.625768203435471e-06,
+      "loss": 0.9527,
+      "step": 51700
+    },
+    {
+      "epoch": 0.3743114218911739,
+      "grad_norm": 0.16955150663852692,
+      "learning_rate": 4.625695816774885e-06,
+      "loss": 0.9453,
+      "step": 51710
+    },
+    {
+      "epoch": 0.37438380855176007,
+      "grad_norm": 0.17448243498802185,
+      "learning_rate": 4.6256234301142985e-06,
+      "loss": 0.9526,
+      "step": 51720
+    },
+    {
+      "epoch": 0.37445619521234624,
+      "grad_norm": 0.17462722957134247,
+      "learning_rate": 4.625551043453713e-06,
+      "loss": 0.9496,
+      "step": 51730
+    },
+    {
+      "epoch": 0.3745285818729325,
+      "grad_norm": 0.15947075188159943,
+      "learning_rate": 4.625478656793127e-06,
+      "loss": 0.9412,
+      "step": 51740
+    },
+    {
+      "epoch": 0.37460096853351865,
+      "grad_norm": 0.16033557057380676,
+      "learning_rate": 4.62540627013254e-06,
+      "loss": 0.9442,
+      "step": 51750
+    },
+    {
+      "epoch": 0.37467335519410483,
+      "grad_norm": 0.15499283373355865,
+      "learning_rate": 4.625333883471954e-06,
+      "loss": 0.9436,
+      "step": 51760
+    },
+    {
+      "epoch": 0.374745741854691,
+      "grad_norm": 0.16719605028629303,
+      "learning_rate": 4.625261496811368e-06,
+      "loss": 0.9456,
+      "step": 51770
+    },
+    {
+      "epoch": 0.3748181285152772,
+      "grad_norm": 0.16747929155826569,
+      "learning_rate": 4.625189110150782e-06,
+      "loss": 0.9396,
+      "step": 51780
+    },
+    {
+      "epoch": 0.3748905151758634,
+      "grad_norm": 0.16449685394763947,
+      "learning_rate": 4.6251167234901955e-06,
+      "loss": 0.9478,
+      "step": 51790
+    },
+    {
+      "epoch": 0.3749629018364496,
+      "grad_norm": 0.16664811968803406,
+      "learning_rate": 4.625044336829609e-06,
+      "loss": 0.9462,
+      "step": 51800
+    },
+    {
+      "epoch": 0.37503528849703577,
+      "grad_norm": 0.16544151306152344,
+      "learning_rate": 4.624971950169024e-06,
+      "loss": 0.9416,
+      "step": 51810
+    },
+    {
+      "epoch": 0.37510767515762194,
+      "grad_norm": 0.1873922049999237,
+      "learning_rate": 4.624899563508437e-06,
+      "loss": 0.9471,
+      "step": 51820
+    },
+    {
+      "epoch": 0.3751800618182081,
+      "grad_norm": 0.1557958424091339,
+      "learning_rate": 4.624827176847851e-06,
+      "loss": 0.9429,
+      "step": 51830
+    },
+    {
+      "epoch": 0.37525244847879435,
+      "grad_norm": 0.1720801442861557,
+      "learning_rate": 4.6247547901872644e-06,
+      "loss": 0.9529,
+      "step": 51840
+    },
+    {
+      "epoch": 0.37532483513938053,
+      "grad_norm": 0.2492019236087799,
+      "learning_rate": 4.624682403526679e-06,
+      "loss": 0.9338,
+      "step": 51850
+    },
+    {
+      "epoch": 0.3753972217999667,
+      "grad_norm": 0.15926505625247955,
+      "learning_rate": 4.6246100168660925e-06,
+      "loss": 0.9455,
+      "step": 51860
+    },
+    {
+      "epoch": 0.3754696084605529,
+      "grad_norm": 0.16041946411132812,
+      "learning_rate": 4.624537630205506e-06,
+      "loss": 0.9507,
+      "step": 51870
+    },
+    {
+      "epoch": 0.37554199512113906,
+      "grad_norm": 0.1602320671081543,
+      "learning_rate": 4.62446524354492e-06,
+      "loss": 0.9493,
+      "step": 51880
+    },
+    {
+      "epoch": 0.37561438178172524,
+      "grad_norm": 0.18607132136821747,
+      "learning_rate": 4.624392856884334e-06,
+      "loss": 0.9591,
+      "step": 51890
+    },
+    {
+      "epoch": 0.37568676844231147,
+      "grad_norm": 0.1767975091934204,
+      "learning_rate": 4.624320470223748e-06,
+      "loss": 0.9571,
+      "step": 51900
+    },
+    {
+      "epoch": 0.37575915510289765,
+      "grad_norm": 0.15797516703605652,
+      "learning_rate": 4.6242480835631615e-06,
+      "loss": 0.9301,
+      "step": 51910
+    },
+    {
+      "epoch": 0.3758315417634838,
+      "grad_norm": 0.16194427013397217,
+      "learning_rate": 4.624175696902575e-06,
+      "loss": 0.9337,
+      "step": 51920
+    },
+    {
+      "epoch": 0.37590392842407,
+      "grad_norm": 0.15718939900398254,
+      "learning_rate": 4.624103310241989e-06,
+      "loss": 0.9361,
+      "step": 51930
+    },
+    {
+      "epoch": 0.3759763150846562,
+      "grad_norm": 0.15875135362148285,
+      "learning_rate": 4.624030923581403e-06,
+      "loss": 0.9309,
+      "step": 51940
+    },
+    {
+      "epoch": 0.3760487017452424,
+      "grad_norm": 0.1533413827419281,
+      "learning_rate": 4.623958536920817e-06,
+      "loss": 0.9316,
+      "step": 51950
+    },
+    {
+      "epoch": 0.3761210884058286,
+      "grad_norm": 0.21360956132411957,
+      "learning_rate": 4.62388615026023e-06,
+      "loss": 0.9535,
+      "step": 51960
+    },
+    {
+      "epoch": 0.37619347506641476,
+      "grad_norm": 0.16147691011428833,
+      "learning_rate": 4.623813763599644e-06,
+      "loss": 0.9492,
+      "step": 51970
+    },
+    {
+      "epoch": 0.37626586172700094,
+      "grad_norm": 0.1707833856344223,
+      "learning_rate": 4.6237413769390585e-06,
+      "loss": 0.9319,
+      "step": 51980
+    },
+    {
+      "epoch": 0.3763382483875871,
+      "grad_norm": 0.17225024104118347,
+      "learning_rate": 4.623668990278472e-06,
+      "loss": 0.9323,
+      "step": 51990
+    },
+    {
+      "epoch": 0.37641063504817335,
+      "grad_norm": 0.16834914684295654,
+      "learning_rate": 4.623596603617886e-06,
+      "loss": 0.9474,
+      "step": 52000
+    },
+    {
+      "epoch": 0.3764830217087595,
+      "grad_norm": 0.1597452461719513,
+      "learning_rate": 4.623524216957299e-06,
+      "loss": 0.938,
+      "step": 52010
+    },
+    {
+      "epoch": 0.3765554083693457,
+      "grad_norm": 0.15453481674194336,
+      "learning_rate": 4.623451830296714e-06,
+      "loss": 0.9591,
+      "step": 52020
+    },
+    {
+      "epoch": 0.3766277950299319,
+      "grad_norm": 0.16118237376213074,
+      "learning_rate": 4.623379443636127e-06,
+      "loss": 0.9533,
+      "step": 52030
+    },
+    {
+      "epoch": 0.37670018169051805,
+      "grad_norm": 0.1625969558954239,
+      "learning_rate": 4.623307056975541e-06,
+      "loss": 0.9415,
+      "step": 52040
+    },
+    {
+      "epoch": 0.37677256835110423,
+      "grad_norm": 0.1555289477109909,
+      "learning_rate": 4.623234670314955e-06,
+      "loss": 0.9301,
+      "step": 52050
+    },
+    {
+      "epoch": 0.37684495501169046,
+      "grad_norm": 0.16258127987384796,
+      "learning_rate": 4.623162283654369e-06,
+      "loss": 0.9373,
+      "step": 52060
+    },
+    {
+      "epoch": 0.37691734167227664,
+      "grad_norm": 0.15739430487155914,
+      "learning_rate": 4.623089896993783e-06,
+      "loss": 0.9482,
+      "step": 52070
+    },
+    {
+      "epoch": 0.3769897283328628,
+      "grad_norm": 0.1597059667110443,
+      "learning_rate": 4.623017510333196e-06,
+      "loss": 0.9505,
+      "step": 52080
+    },
+    {
+      "epoch": 0.377062114993449,
+      "grad_norm": 0.16315515339374542,
+      "learning_rate": 4.62294512367261e-06,
+      "loss": 0.9478,
+      "step": 52090
+    },
+    {
+      "epoch": 0.37713450165403517,
+      "grad_norm": 0.17426295578479767,
+      "learning_rate": 4.6228727370120236e-06,
+      "loss": 0.9359,
+      "step": 52100
+    },
+    {
+      "epoch": 0.3772068883146214,
+      "grad_norm": 0.16051162779331207,
+      "learning_rate": 4.622800350351437e-06,
+      "loss": 0.9403,
+      "step": 52110
+    },
+    {
+      "epoch": 0.3772792749752076,
+      "grad_norm": 0.17289088666439056,
+      "learning_rate": 4.622727963690851e-06,
+      "loss": 0.9381,
+      "step": 52120
+    },
+    {
+      "epoch": 0.37735166163579376,
+      "grad_norm": 0.17087596654891968,
+      "learning_rate": 4.622655577030265e-06,
+      "loss": 0.9479,
+      "step": 52130
+    },
+    {
+      "epoch": 0.37742404829637993,
+      "grad_norm": 0.16720347106456757,
+      "learning_rate": 4.622583190369679e-06,
+      "loss": 0.936,
+      "step": 52140
+    },
+    {
+      "epoch": 0.3774964349569661,
+      "grad_norm": 0.18241596221923828,
+      "learning_rate": 4.6225108037090925e-06,
+      "loss": 0.943,
+      "step": 52150
+    },
+    {
+      "epoch": 0.37756882161755234,
+      "grad_norm": 0.17475546896457672,
+      "learning_rate": 4.622438417048506e-06,
+      "loss": 0.9408,
+      "step": 52160
+    },
+    {
+      "epoch": 0.3776412082781385,
+      "grad_norm": 0.15939339995384216,
+      "learning_rate": 4.6223660303879206e-06,
+      "loss": 0.948,
+      "step": 52170
+    },
+    {
+      "epoch": 0.3777135949387247,
+      "grad_norm": 0.1768542230129242,
+      "learning_rate": 4.622293643727334e-06,
+      "loss": 0.9561,
+      "step": 52180
+    },
+    {
+      "epoch": 0.37778598159931087,
+      "grad_norm": 0.15933936834335327,
+      "learning_rate": 4.622221257066748e-06,
+      "loss": 0.9486,
+      "step": 52190
+    },
+    {
+      "epoch": 0.37785836825989705,
+      "grad_norm": 0.16799773275852203,
+      "learning_rate": 4.622148870406161e-06,
+      "loss": 0.9352,
+      "step": 52200
+    },
+    {
+      "epoch": 0.3779307549204833,
+      "grad_norm": 0.2672235369682312,
+      "learning_rate": 4.622076483745576e-06,
+      "loss": 0.9393,
+      "step": 52210
+    },
+    {
+      "epoch": 0.37800314158106946,
+      "grad_norm": 0.1704738289117813,
+      "learning_rate": 4.6220040970849895e-06,
+      "loss": 0.941,
+      "step": 52220
+    },
+    {
+      "epoch": 0.37807552824165563,
+      "grad_norm": 0.16287149488925934,
+      "learning_rate": 4.621931710424403e-06,
+      "loss": 0.9248,
+      "step": 52230
+    },
+    {
+      "epoch": 0.3781479149022418,
+      "grad_norm": 0.1523078829050064,
+      "learning_rate": 4.621859323763817e-06,
+      "loss": 0.9593,
+      "step": 52240
+    },
+    {
+      "epoch": 0.378220301562828,
+      "grad_norm": 0.16731417179107666,
+      "learning_rate": 4.621786937103231e-06,
+      "loss": 0.9534,
+      "step": 52250
+    },
+    {
+      "epoch": 0.37829268822341416,
+      "grad_norm": 0.2127581089735031,
+      "learning_rate": 4.621714550442645e-06,
+      "loss": 0.9454,
+      "step": 52260
+    },
+    {
+      "epoch": 0.3783650748840004,
+      "grad_norm": 0.17544521391391754,
+      "learning_rate": 4.621642163782058e-06,
+      "loss": 0.9495,
+      "step": 52270
+    },
+    {
+      "epoch": 0.3784374615445866,
+      "grad_norm": 0.1671367883682251,
+      "learning_rate": 4.621569777121472e-06,
+      "loss": 0.9431,
+      "step": 52280
+    },
+    {
+      "epoch": 0.37850984820517275,
+      "grad_norm": 0.15478673577308655,
+      "learning_rate": 4.6214973904608865e-06,
+      "loss": 0.9475,
+      "step": 52290
+    },
+    {
+      "epoch": 0.3785822348657589,
+      "grad_norm": 0.1520804613828659,
+      "learning_rate": 4.6214250038003e-06,
+      "loss": 0.9435,
+      "step": 52300
+    },
+    {
+      "epoch": 0.3786546215263451,
+      "grad_norm": 0.1676923781633377,
+      "learning_rate": 4.621352617139714e-06,
+      "loss": 0.9409,
+      "step": 52310
+    },
+    {
+      "epoch": 0.37872700818693134,
+      "grad_norm": 0.15937332808971405,
+      "learning_rate": 4.621280230479127e-06,
+      "loss": 0.9341,
+      "step": 52320
+    },
+    {
+      "epoch": 0.3787993948475175,
+      "grad_norm": 0.16753974556922913,
+      "learning_rate": 4.621207843818542e-06,
+      "loss": 0.9533,
+      "step": 52330
+    },
+    {
+      "epoch": 0.3788717815081037,
+      "grad_norm": 0.17047527432441711,
+      "learning_rate": 4.621135457157955e-06,
+      "loss": 0.948,
+      "step": 52340
+    },
+    {
+      "epoch": 0.37894416816868987,
+      "grad_norm": 0.1727355569601059,
+      "learning_rate": 4.621063070497369e-06,
+      "loss": 0.9421,
+      "step": 52350
+    },
+    {
+      "epoch": 0.37901655482927604,
+      "grad_norm": 0.1628546565771103,
+      "learning_rate": 4.620990683836783e-06,
+      "loss": 0.9464,
+      "step": 52360
+    },
+    {
+      "epoch": 0.3790889414898623,
+      "grad_norm": 0.17123012244701385,
+      "learning_rate": 4.620918297176197e-06,
+      "loss": 0.9472,
+      "step": 52370
+    },
+    {
+      "epoch": 0.37916132815044845,
+      "grad_norm": 0.1762952357530594,
+      "learning_rate": 4.620845910515611e-06,
+      "loss": 0.9495,
+      "step": 52380
+    },
+    {
+      "epoch": 0.37923371481103463,
+      "grad_norm": 0.17170877754688263,
+      "learning_rate": 4.620773523855024e-06,
+      "loss": 0.9504,
+      "step": 52390
+    },
+    {
+      "epoch": 0.3793061014716208,
+      "grad_norm": 0.16942241787910461,
+      "learning_rate": 4.620701137194438e-06,
+      "loss": 0.9413,
+      "step": 52400
+    },
+    {
+      "epoch": 0.379378488132207,
+      "grad_norm": 0.15613293647766113,
+      "learning_rate": 4.6206287505338524e-06,
+      "loss": 0.9331,
+      "step": 52410
+    },
+    {
+      "epoch": 0.37945087479279316,
+      "grad_norm": 0.18218636512756348,
+      "learning_rate": 4.620556363873266e-06,
+      "loss": 0.9552,
+      "step": 52420
+    },
+    {
+      "epoch": 0.3795232614533794,
+      "grad_norm": 0.16046783328056335,
+      "learning_rate": 4.62048397721268e-06,
+      "loss": 0.9505,
+      "step": 52430
+    },
+    {
+      "epoch": 0.37959564811396557,
+      "grad_norm": 0.16807179152965546,
+      "learning_rate": 4.620411590552093e-06,
+      "loss": 0.9554,
+      "step": 52440
+    },
+    {
+      "epoch": 0.37966803477455174,
+      "grad_norm": 0.18919555842876434,
+      "learning_rate": 4.620339203891508e-06,
+      "loss": 0.9436,
+      "step": 52450
+    },
+    {
+      "epoch": 0.3797404214351379,
+      "grad_norm": 0.16013583540916443,
+      "learning_rate": 4.620266817230921e-06,
+      "loss": 0.9366,
+      "step": 52460
+    },
+    {
+      "epoch": 0.3798128080957241,
+      "grad_norm": 0.16914841532707214,
+      "learning_rate": 4.620194430570335e-06,
+      "loss": 0.9469,
+      "step": 52470
+    },
+    {
+      "epoch": 0.37988519475631033,
+      "grad_norm": 0.1607542335987091,
+      "learning_rate": 4.620122043909749e-06,
+      "loss": 0.9475,
+      "step": 52480
+    },
+    {
+      "epoch": 0.3799575814168965,
+      "grad_norm": 0.15814971923828125,
+      "learning_rate": 4.620049657249163e-06,
+      "loss": 0.9308,
+      "step": 52490
+    },
+    {
+      "epoch": 0.3800299680774827,
+      "grad_norm": 0.1508866399526596,
+      "learning_rate": 4.619977270588577e-06,
+      "loss": 0.9608,
+      "step": 52500
+    },
+    {
+      "epoch": 0.38010235473806886,
+      "grad_norm": 0.22690674662590027,
+      "learning_rate": 4.61990488392799e-06,
+      "loss": 0.9331,
+      "step": 52510
+    },
+    {
+      "epoch": 0.38017474139865504,
+      "grad_norm": 0.1720598340034485,
+      "learning_rate": 4.619832497267404e-06,
+      "loss": 0.9616,
+      "step": 52520
+    },
+    {
+      "epoch": 0.38024712805924127,
+      "grad_norm": 0.16492782533168793,
+      "learning_rate": 4.619760110606818e-06,
+      "loss": 0.9386,
+      "step": 52530
+    },
+    {
+      "epoch": 0.38031951471982745,
+      "grad_norm": 0.1547376811504364,
+      "learning_rate": 4.619687723946232e-06,
+      "loss": 0.9459,
+      "step": 52540
+    },
+    {
+      "epoch": 0.3803919013804136,
+      "grad_norm": 0.18093635141849518,
+      "learning_rate": 4.619615337285646e-06,
+      "loss": 0.9616,
+      "step": 52550
+    },
+    {
+      "epoch": 0.3804642880409998,
+      "grad_norm": 0.16667254269123077,
+      "learning_rate": 4.619542950625059e-06,
+      "loss": 0.9431,
+      "step": 52560
+    },
+    {
+      "epoch": 0.380536674701586,
+      "grad_norm": 0.15666356682777405,
+      "learning_rate": 4.619470563964473e-06,
+      "loss": 0.9355,
+      "step": 52570
+    },
+    {
+      "epoch": 0.38060906136217215,
+      "grad_norm": 0.21181073784828186,
+      "learning_rate": 4.619398177303887e-06,
+      "loss": 0.9516,
+      "step": 52580
+    },
+    {
+      "epoch": 0.3806814480227584,
+      "grad_norm": 0.16820743680000305,
+      "learning_rate": 4.619325790643301e-06,
+      "loss": 0.942,
+      "step": 52590
+    },
+    {
+      "epoch": 0.38075383468334456,
+      "grad_norm": 0.19321396946907043,
+      "learning_rate": 4.6192534039827145e-06,
+      "loss": 0.9395,
+      "step": 52600
+    },
+    {
+      "epoch": 0.38082622134393074,
+      "grad_norm": 0.16616934537887573,
+      "learning_rate": 4.619181017322128e-06,
+      "loss": 0.9376,
+      "step": 52610
+    },
+    {
+      "epoch": 0.3808986080045169,
+      "grad_norm": 0.19867192208766937,
+      "learning_rate": 4.619108630661542e-06,
+      "loss": 0.945,
+      "step": 52620
+    },
+    {
+      "epoch": 0.3809709946651031,
+      "grad_norm": 0.15694333612918854,
+      "learning_rate": 4.619036244000955e-06,
+      "loss": 0.9276,
+      "step": 52630
+    },
+    {
+      "epoch": 0.3810433813256893,
+      "grad_norm": 0.16102519631385803,
+      "learning_rate": 4.61896385734037e-06,
+      "loss": 0.9493,
+      "step": 52640
+    },
+    {
+      "epoch": 0.3811157679862755,
+      "grad_norm": 0.15607930719852448,
+      "learning_rate": 4.6188914706797835e-06,
+      "loss": 0.9453,
+      "step": 52650
+    },
+    {
+      "epoch": 0.3811881546468617,
+      "grad_norm": 0.1629965603351593,
+      "learning_rate": 4.618819084019197e-06,
+      "loss": 0.9471,
+      "step": 52660
+    },
+    {
+      "epoch": 0.38126054130744785,
+      "grad_norm": 0.15766113996505737,
+      "learning_rate": 4.618746697358611e-06,
+      "loss": 0.9396,
+      "step": 52670
+    },
+    {
+      "epoch": 0.38133292796803403,
+      "grad_norm": 0.1607961505651474,
+      "learning_rate": 4.618674310698025e-06,
+      "loss": 0.9483,
+      "step": 52680
+    },
+    {
+      "epoch": 0.38140531462862026,
+      "grad_norm": 0.1705811619758606,
+      "learning_rate": 4.618601924037439e-06,
+      "loss": 0.9418,
+      "step": 52690
+    },
+    {
+      "epoch": 0.38147770128920644,
+      "grad_norm": 0.1601417511701584,
+      "learning_rate": 4.618529537376852e-06,
+      "loss": 0.9441,
+      "step": 52700
+    },
+    {
+      "epoch": 0.3815500879497926,
+      "grad_norm": 0.1780814677476883,
+      "learning_rate": 4.618457150716266e-06,
+      "loss": 0.9551,
+      "step": 52710
+    },
+    {
+      "epoch": 0.3816224746103788,
+      "grad_norm": 0.15561501681804657,
+      "learning_rate": 4.61838476405568e-06,
+      "loss": 0.9423,
+      "step": 52720
+    },
+    {
+      "epoch": 0.38169486127096497,
+      "grad_norm": 0.15728692710399628,
+      "learning_rate": 4.618312377395094e-06,
+      "loss": 0.9354,
+      "step": 52730
+    },
+    {
+      "epoch": 0.3817672479315512,
+      "grad_norm": 0.20734450221061707,
+      "learning_rate": 4.618239990734508e-06,
+      "loss": 0.9548,
+      "step": 52740
+    },
+    {
+      "epoch": 0.3818396345921374,
+      "grad_norm": 0.17452536523342133,
+      "learning_rate": 4.618167604073921e-06,
+      "loss": 0.9418,
+      "step": 52750
+    },
+    {
+      "epoch": 0.38191202125272355,
+      "grad_norm": 0.17230959236621857,
+      "learning_rate": 4.618095217413335e-06,
+      "loss": 0.9517,
+      "step": 52760
+    },
+    {
+      "epoch": 0.38198440791330973,
+      "grad_norm": 0.16141583025455475,
+      "learning_rate": 4.618022830752749e-06,
+      "loss": 0.9378,
+      "step": 52770
+    },
+    {
+      "epoch": 0.3820567945738959,
+      "grad_norm": 0.16719284653663635,
+      "learning_rate": 4.617950444092163e-06,
+      "loss": 0.9458,
+      "step": 52780
+    },
+    {
+      "epoch": 0.3821291812344821,
+      "grad_norm": 0.1570596694946289,
+      "learning_rate": 4.617878057431577e-06,
+      "loss": 0.9495,
+      "step": 52790
+    },
+    {
+      "epoch": 0.3822015678950683,
+      "grad_norm": 0.17521663010120392,
+      "learning_rate": 4.61780567077099e-06,
+      "loss": 0.9413,
+      "step": 52800
+    },
+    {
+      "epoch": 0.3822739545556545,
+      "grad_norm": 0.1793406456708908,
+      "learning_rate": 4.617733284110405e-06,
+      "loss": 0.9346,
+      "step": 52810
+    },
+    {
+      "epoch": 0.38234634121624067,
+      "grad_norm": 0.18589749932289124,
+      "learning_rate": 4.617660897449818e-06,
+      "loss": 0.9405,
+      "step": 52820
+    },
+    {
+      "epoch": 0.38241872787682685,
+      "grad_norm": 0.15368783473968506,
+      "learning_rate": 4.617588510789232e-06,
+      "loss": 0.9345,
+      "step": 52830
+    },
+    {
+      "epoch": 0.382491114537413,
+      "grad_norm": 0.1717151403427124,
+      "learning_rate": 4.6175161241286456e-06,
+      "loss": 0.9413,
+      "step": 52840
+    },
+    {
+      "epoch": 0.38256350119799926,
+      "grad_norm": 0.1748562455177307,
+      "learning_rate": 4.61744373746806e-06,
+      "loss": 0.9458,
+      "step": 52850
+    },
+    {
+      "epoch": 0.38263588785858543,
+      "grad_norm": 0.17434179782867432,
+      "learning_rate": 4.617371350807474e-06,
+      "loss": 0.9401,
+      "step": 52860
+    },
+    {
+      "epoch": 0.3827082745191716,
+      "grad_norm": 0.1836162507534027,
+      "learning_rate": 4.617298964146887e-06,
+      "loss": 0.9455,
+      "step": 52870
+    },
+    {
+      "epoch": 0.3827806611797578,
+      "grad_norm": 0.15718111395835876,
+      "learning_rate": 4.617226577486301e-06,
+      "loss": 0.9388,
+      "step": 52880
+    },
+    {
+      "epoch": 0.38285304784034396,
+      "grad_norm": 0.15703000128269196,
+      "learning_rate": 4.617154190825715e-06,
+      "loss": 0.9423,
+      "step": 52890
+    },
+    {
+      "epoch": 0.3829254345009302,
+      "grad_norm": 0.16674445569515228,
+      "learning_rate": 4.617081804165129e-06,
+      "loss": 0.951,
+      "step": 52900
+    },
+    {
+      "epoch": 0.38299782116151637,
+      "grad_norm": 0.157858207821846,
+      "learning_rate": 4.6170094175045426e-06,
+      "loss": 0.9484,
+      "step": 52910
+    },
+    {
+      "epoch": 0.38307020782210255,
+      "grad_norm": 0.17335231602191925,
+      "learning_rate": 4.616937030843956e-06,
+      "loss": 0.9482,
+      "step": 52920
+    },
+    {
+      "epoch": 0.3831425944826887,
+      "grad_norm": 0.16465920209884644,
+      "learning_rate": 4.616864644183371e-06,
+      "loss": 0.9378,
+      "step": 52930
+    },
+    {
+      "epoch": 0.3832149811432749,
+      "grad_norm": 0.17838311195373535,
+      "learning_rate": 4.616792257522784e-06,
+      "loss": 0.956,
+      "step": 52940
+    },
+    {
+      "epoch": 0.3832873678038611,
+      "grad_norm": 0.1656845659017563,
+      "learning_rate": 4.616719870862198e-06,
+      "loss": 0.9409,
+      "step": 52950
+    },
+    {
+      "epoch": 0.3833597544644473,
+      "grad_norm": 0.15346546471118927,
+      "learning_rate": 4.6166474842016115e-06,
+      "loss": 0.9334,
+      "step": 52960
+    },
+    {
+      "epoch": 0.3834321411250335,
+      "grad_norm": 0.17252103984355927,
+      "learning_rate": 4.616575097541026e-06,
+      "loss": 0.9422,
+      "step": 52970
+    },
+    {
+      "epoch": 0.38350452778561966,
+      "grad_norm": 0.16853582859039307,
+      "learning_rate": 4.6165027108804396e-06,
+      "loss": 0.944,
+      "step": 52980
+    },
+    {
+      "epoch": 0.38357691444620584,
+      "grad_norm": 0.17308548092842102,
+      "learning_rate": 4.616430324219853e-06,
+      "loss": 0.9396,
+      "step": 52990
+    },
+    {
+      "epoch": 0.383649301106792,
+      "grad_norm": 0.15582707524299622,
+      "learning_rate": 4.616357937559267e-06,
+      "loss": 0.9445,
+      "step": 53000
+    },
+    {
+      "epoch": 0.38372168776737825,
+      "grad_norm": 0.15688510239124298,
+      "learning_rate": 4.616285550898681e-06,
+      "loss": 0.9359,
+      "step": 53010
+    },
+    {
+      "epoch": 0.3837940744279644,
+      "grad_norm": 0.16674372553825378,
+      "learning_rate": 4.616213164238095e-06,
+      "loss": 0.9378,
+      "step": 53020
+    },
+    {
+      "epoch": 0.3838664610885506,
+      "grad_norm": 0.15617065131664276,
+      "learning_rate": 4.6161407775775085e-06,
+      "loss": 0.9443,
+      "step": 53030
+    },
+    {
+      "epoch": 0.3839388477491368,
+      "grad_norm": 0.18050551414489746,
+      "learning_rate": 4.616068390916922e-06,
+      "loss": 0.9282,
+      "step": 53040
+    },
+    {
+      "epoch": 0.38401123440972296,
+      "grad_norm": 0.18156303465366364,
+      "learning_rate": 4.6159960042563366e-06,
+      "loss": 0.9467,
+      "step": 53050
+    },
+    {
+      "epoch": 0.3840836210703092,
+      "grad_norm": 0.1625535637140274,
+      "learning_rate": 4.61592361759575e-06,
+      "loss": 0.9361,
+      "step": 53060
+    },
+    {
+      "epoch": 0.38415600773089537,
+      "grad_norm": 0.150069460272789,
+      "learning_rate": 4.615851230935164e-06,
+      "loss": 0.9384,
+      "step": 53070
+    },
+    {
+      "epoch": 0.38422839439148154,
+      "grad_norm": 0.17181698977947235,
+      "learning_rate": 4.615778844274577e-06,
+      "loss": 0.9275,
+      "step": 53080
+    },
+    {
+      "epoch": 0.3843007810520677,
+      "grad_norm": 0.16848745942115784,
+      "learning_rate": 4.615706457613992e-06,
+      "loss": 0.9463,
+      "step": 53090
+    },
+    {
+      "epoch": 0.3843731677126539,
+      "grad_norm": 0.15281365811824799,
+      "learning_rate": 4.6156340709534055e-06,
+      "loss": 0.9435,
+      "step": 53100
+    },
+    {
+      "epoch": 0.3844455543732401,
+      "grad_norm": 0.16574618220329285,
+      "learning_rate": 4.615561684292819e-06,
+      "loss": 0.9379,
+      "step": 53110
+    },
+    {
+      "epoch": 0.3845179410338263,
+      "grad_norm": 0.1556422859430313,
+      "learning_rate": 4.615489297632233e-06,
+      "loss": 0.9404,
+      "step": 53120
+    },
+    {
+      "epoch": 0.3845903276944125,
+      "grad_norm": 0.16364730894565582,
+      "learning_rate": 4.615416910971647e-06,
+      "loss": 0.9436,
+      "step": 53130
+    },
+    {
+      "epoch": 0.38466271435499866,
+      "grad_norm": 0.18387825787067413,
+      "learning_rate": 4.61534452431106e-06,
+      "loss": 0.9542,
+      "step": 53140
+    },
+    {
+      "epoch": 0.38473510101558483,
+      "grad_norm": 0.16818496584892273,
+      "learning_rate": 4.615272137650474e-06,
+      "loss": 0.9338,
+      "step": 53150
+    },
+    {
+      "epoch": 0.384807487676171,
+      "grad_norm": 0.17835207283496857,
+      "learning_rate": 4.615199750989888e-06,
+      "loss": 0.949,
+      "step": 53160
+    },
+    {
+      "epoch": 0.38487987433675724,
+      "grad_norm": 0.15151627361774445,
+      "learning_rate": 4.615127364329302e-06,
+      "loss": 0.9357,
+      "step": 53170
+    },
+    {
+      "epoch": 0.3849522609973434,
+      "grad_norm": 0.1659085601568222,
+      "learning_rate": 4.615054977668715e-06,
+      "loss": 0.939,
+      "step": 53180
+    },
+    {
+      "epoch": 0.3850246476579296,
+      "grad_norm": 0.19778911769390106,
+      "learning_rate": 4.614982591008129e-06,
+      "loss": 0.95,
+      "step": 53190
+    },
+    {
+      "epoch": 0.3850970343185158,
+      "grad_norm": 0.1595744490623474,
+      "learning_rate": 4.614910204347543e-06,
+      "loss": 0.9478,
+      "step": 53200
+    },
+    {
+      "epoch": 0.38516942097910195,
+      "grad_norm": 0.16915087401866913,
+      "learning_rate": 4.614837817686957e-06,
+      "loss": 0.9462,
+      "step": 53210
+    },
+    {
+      "epoch": 0.3852418076396882,
+      "grad_norm": 0.19996778666973114,
+      "learning_rate": 4.614765431026371e-06,
+      "loss": 0.9251,
+      "step": 53220
+    },
+    {
+      "epoch": 0.38531419430027436,
+      "grad_norm": 0.17590823769569397,
+      "learning_rate": 4.614693044365784e-06,
+      "loss": 0.9577,
+      "step": 53230
+    },
+    {
+      "epoch": 0.38538658096086054,
+      "grad_norm": 0.16508999466896057,
+      "learning_rate": 4.614620657705199e-06,
+      "loss": 0.9411,
+      "step": 53240
+    },
+    {
+      "epoch": 0.3854589676214467,
+      "grad_norm": 0.15081745386123657,
+      "learning_rate": 4.614548271044612e-06,
+      "loss": 0.9475,
+      "step": 53250
+    },
+    {
+      "epoch": 0.3855313542820329,
+      "grad_norm": 0.17549897730350494,
+      "learning_rate": 4.614475884384026e-06,
+      "loss": 0.9367,
+      "step": 53260
+    },
+    {
+      "epoch": 0.38560374094261907,
+      "grad_norm": 0.1669255644083023,
+      "learning_rate": 4.6144034977234395e-06,
+      "loss": 0.9399,
+      "step": 53270
+    },
+    {
+      "epoch": 0.3856761276032053,
+      "grad_norm": 0.15719322860240936,
+      "learning_rate": 4.614331111062854e-06,
+      "loss": 0.9427,
+      "step": 53280
+    },
+    {
+      "epoch": 0.3857485142637915,
+      "grad_norm": 0.1589440554380417,
+      "learning_rate": 4.614258724402268e-06,
+      "loss": 0.9389,
+      "step": 53290
+    },
+    {
+      "epoch": 0.38582090092437765,
+      "grad_norm": 0.1575065553188324,
+      "learning_rate": 4.614186337741681e-06,
+      "loss": 0.9479,
+      "step": 53300
+    },
+    {
+      "epoch": 0.38589328758496383,
+      "grad_norm": 0.18554028868675232,
+      "learning_rate": 4.614113951081095e-06,
+      "loss": 0.9357,
+      "step": 53310
+    },
+    {
+      "epoch": 0.38596567424555,
+      "grad_norm": 0.1604183316230774,
+      "learning_rate": 4.614041564420509e-06,
+      "loss": 0.9588,
+      "step": 53320
+    },
+    {
+      "epoch": 0.38603806090613624,
+      "grad_norm": 0.18416175246238708,
+      "learning_rate": 4.613969177759923e-06,
+      "loss": 0.9475,
+      "step": 53330
+    },
+    {
+      "epoch": 0.3861104475667224,
+      "grad_norm": 0.17273585498332977,
+      "learning_rate": 4.6138967910993365e-06,
+      "loss": 0.9493,
+      "step": 53340
+    },
+    {
+      "epoch": 0.3861828342273086,
+      "grad_norm": 0.15361787378787994,
+      "learning_rate": 4.61382440443875e-06,
+      "loss": 0.9467,
+      "step": 53350
+    },
+    {
+      "epoch": 0.38625522088789477,
+      "grad_norm": 0.16739359498023987,
+      "learning_rate": 4.613752017778164e-06,
+      "loss": 0.9473,
+      "step": 53360
+    },
+    {
+      "epoch": 0.38632760754848094,
+      "grad_norm": 0.17038732767105103,
+      "learning_rate": 4.613679631117578e-06,
+      "loss": 0.9526,
+      "step": 53370
+    },
+    {
+      "epoch": 0.3863999942090672,
+      "grad_norm": 0.16086862981319427,
+      "learning_rate": 4.613607244456992e-06,
+      "loss": 0.9518,
+      "step": 53380
+    },
+    {
+      "epoch": 0.38647238086965335,
+      "grad_norm": 0.18573077023029327,
+      "learning_rate": 4.6135348577964055e-06,
+      "loss": 0.9483,
+      "step": 53390
+    },
+    {
+      "epoch": 0.38654476753023953,
+      "grad_norm": 0.17494763433933258,
+      "learning_rate": 4.613462471135819e-06,
+      "loss": 0.9391,
+      "step": 53400
+    },
+    {
+      "epoch": 0.3866171541908257,
+      "grad_norm": 0.19013044238090515,
+      "learning_rate": 4.6133900844752335e-06,
+      "loss": 0.9467,
+      "step": 53410
+    },
+    {
+      "epoch": 0.3866895408514119,
+      "grad_norm": 0.16894245147705078,
+      "learning_rate": 4.613317697814647e-06,
+      "loss": 0.9394,
+      "step": 53420
+    },
+    {
+      "epoch": 0.3867619275119981,
+      "grad_norm": 0.15704815089702606,
+      "learning_rate": 4.613245311154061e-06,
+      "loss": 0.9343,
+      "step": 53430
+    },
+    {
+      "epoch": 0.3868343141725843,
+      "grad_norm": 0.15952664613723755,
+      "learning_rate": 4.613172924493474e-06,
+      "loss": 0.9402,
+      "step": 53440
+    },
+    {
+      "epoch": 0.38690670083317047,
+      "grad_norm": 0.15518204867839813,
+      "learning_rate": 4.613100537832889e-06,
+      "loss": 0.9413,
+      "step": 53450
+    },
+    {
+      "epoch": 0.38697908749375665,
+      "grad_norm": 0.16507835686206818,
+      "learning_rate": 4.6130281511723025e-06,
+      "loss": 0.9396,
+      "step": 53460
+    },
+    {
+      "epoch": 0.3870514741543428,
+      "grad_norm": 0.17540588974952698,
+      "learning_rate": 4.612955764511716e-06,
+      "loss": 0.9663,
+      "step": 53470
+    },
+    {
+      "epoch": 0.387123860814929,
+      "grad_norm": 0.16836467385292053,
+      "learning_rate": 4.61288337785113e-06,
+      "loss": 0.9458,
+      "step": 53480
+    },
+    {
+      "epoch": 0.38719624747551523,
+      "grad_norm": 0.17496466636657715,
+      "learning_rate": 4.612810991190544e-06,
+      "loss": 0.9404,
+      "step": 53490
+    },
+    {
+      "epoch": 0.3872686341361014,
+      "grad_norm": 0.17381425201892853,
+      "learning_rate": 4.612738604529958e-06,
+      "loss": 0.9517,
+      "step": 53500
+    },
+    {
+      "epoch": 0.3873410207966876,
+      "grad_norm": 0.15964674949645996,
+      "learning_rate": 4.612666217869371e-06,
+      "loss": 0.9452,
+      "step": 53510
+    },
+    {
+      "epoch": 0.38741340745727376,
+      "grad_norm": 0.16532573103904724,
+      "learning_rate": 4.612593831208785e-06,
+      "loss": 0.9341,
+      "step": 53520
+    },
+    {
+      "epoch": 0.38748579411785994,
+      "grad_norm": 0.17337869107723236,
+      "learning_rate": 4.6125214445481995e-06,
+      "loss": 0.9402,
+      "step": 53530
+    },
+    {
+      "epoch": 0.38755818077844617,
+      "grad_norm": 0.1612296998500824,
+      "learning_rate": 4.612449057887613e-06,
+      "loss": 0.9458,
+      "step": 53540
+    },
+    {
+      "epoch": 0.38763056743903235,
+      "grad_norm": 0.1577647626399994,
+      "learning_rate": 4.612376671227027e-06,
+      "loss": 0.9357,
+      "step": 53550
+    },
+    {
+      "epoch": 0.3877029540996185,
+      "grad_norm": 0.1562313437461853,
+      "learning_rate": 4.61230428456644e-06,
+      "loss": 0.9474,
+      "step": 53560
+    },
+    {
+      "epoch": 0.3877753407602047,
+      "grad_norm": 0.1620018631219864,
+      "learning_rate": 4.612231897905855e-06,
+      "loss": 0.9408,
+      "step": 53570
+    },
+    {
+      "epoch": 0.3878477274207909,
+      "grad_norm": 0.21448646485805511,
+      "learning_rate": 4.612159511245268e-06,
+      "loss": 0.954,
+      "step": 53580
+    },
+    {
+      "epoch": 0.3879201140813771,
+      "grad_norm": 0.16676433384418488,
+      "learning_rate": 4.612087124584682e-06,
+      "loss": 0.9484,
+      "step": 53590
+    },
+    {
+      "epoch": 0.3879925007419633,
+      "grad_norm": 0.16710571944713593,
+      "learning_rate": 4.612014737924096e-06,
+      "loss": 0.9479,
+      "step": 53600
+    },
+    {
+      "epoch": 0.38806488740254946,
+      "grad_norm": 0.1688128262758255,
+      "learning_rate": 4.61194235126351e-06,
+      "loss": 0.941,
+      "step": 53610
+    },
+    {
+      "epoch": 0.38813727406313564,
+      "grad_norm": 0.15826979279518127,
+      "learning_rate": 4.611869964602924e-06,
+      "loss": 0.9407,
+      "step": 53620
+    },
+    {
+      "epoch": 0.3882096607237218,
+      "grad_norm": 0.3778460621833801,
+      "learning_rate": 4.611797577942337e-06,
+      "loss": 0.9465,
+      "step": 53630
+    },
+    {
+      "epoch": 0.388282047384308,
+      "grad_norm": 0.1658174842596054,
+      "learning_rate": 4.611725191281751e-06,
+      "loss": 0.9414,
+      "step": 53640
+    },
+    {
+      "epoch": 0.3883544340448942,
+      "grad_norm": 0.16166210174560547,
+      "learning_rate": 4.611652804621165e-06,
+      "loss": 0.9495,
+      "step": 53650
+    },
+    {
+      "epoch": 0.3884268207054804,
+      "grad_norm": 0.17662234604358673,
+      "learning_rate": 4.611580417960579e-06,
+      "loss": 0.9516,
+      "step": 53660
+    },
+    {
+      "epoch": 0.3884992073660666,
+      "grad_norm": 0.16207055747509003,
+      "learning_rate": 4.611508031299992e-06,
+      "loss": 0.9417,
+      "step": 53670
+    },
+    {
+      "epoch": 0.38857159402665276,
+      "grad_norm": 0.17646333575248718,
+      "learning_rate": 4.611435644639406e-06,
+      "loss": 0.9415,
+      "step": 53680
+    },
+    {
+      "epoch": 0.38864398068723893,
+      "grad_norm": 0.15925532579421997,
+      "learning_rate": 4.61136325797882e-06,
+      "loss": 0.9373,
+      "step": 53690
+    },
+    {
+      "epoch": 0.38871636734782516,
+      "grad_norm": 0.15857084095478058,
+      "learning_rate": 4.6112908713182335e-06,
+      "loss": 0.9461,
+      "step": 53700
+    },
+    {
+      "epoch": 0.38878875400841134,
+      "grad_norm": 0.16620729863643646,
+      "learning_rate": 4.611218484657647e-06,
+      "loss": 0.9419,
+      "step": 53710
+    },
+    {
+      "epoch": 0.3888611406689975,
+      "grad_norm": 0.14982983469963074,
+      "learning_rate": 4.6111460979970616e-06,
+      "loss": 0.9434,
+      "step": 53720
+    },
+    {
+      "epoch": 0.3889335273295837,
+      "grad_norm": 0.16091054677963257,
+      "learning_rate": 4.611073711336475e-06,
+      "loss": 0.939,
+      "step": 53730
+    },
+    {
+      "epoch": 0.38900591399016987,
+      "grad_norm": 0.17934353649616241,
+      "learning_rate": 4.611001324675889e-06,
+      "loss": 0.9422,
+      "step": 53740
+    },
+    {
+      "epoch": 0.3890783006507561,
+      "grad_norm": 0.22706985473632812,
+      "learning_rate": 4.610928938015302e-06,
+      "loss": 0.9458,
+      "step": 53750
+    },
+    {
+      "epoch": 0.3891506873113423,
+      "grad_norm": 0.19281072914600372,
+      "learning_rate": 4.610856551354717e-06,
+      "loss": 0.9466,
+      "step": 53760
+    },
+    {
+      "epoch": 0.38922307397192846,
+      "grad_norm": 0.17095661163330078,
+      "learning_rate": 4.6107841646941305e-06,
+      "loss": 0.9353,
+      "step": 53770
+    },
+    {
+      "epoch": 0.38929546063251463,
+      "grad_norm": 0.1668483316898346,
+      "learning_rate": 4.610711778033544e-06,
+      "loss": 0.948,
+      "step": 53780
+    },
+    {
+      "epoch": 0.3893678472931008,
+      "grad_norm": 0.1767456829547882,
+      "learning_rate": 4.610639391372958e-06,
+      "loss": 0.9435,
+      "step": 53790
+    },
+    {
+      "epoch": 0.389440233953687,
+      "grad_norm": 0.15454381704330444,
+      "learning_rate": 4.610567004712372e-06,
+      "loss": 0.9393,
+      "step": 53800
+    },
+    {
+      "epoch": 0.3895126206142732,
+      "grad_norm": 0.16419751942157745,
+      "learning_rate": 4.610494618051786e-06,
+      "loss": 0.9305,
+      "step": 53810
+    },
+    {
+      "epoch": 0.3895850072748594,
+      "grad_norm": 0.1819341629743576,
+      "learning_rate": 4.610422231391199e-06,
+      "loss": 0.9309,
+      "step": 53820
+    },
+    {
+      "epoch": 0.3896573939354456,
+      "grad_norm": 0.15951742231845856,
+      "learning_rate": 4.610349844730613e-06,
+      "loss": 0.945,
+      "step": 53830
+    },
+    {
+      "epoch": 0.38972978059603175,
+      "grad_norm": 0.15791021287441254,
+      "learning_rate": 4.6102774580700275e-06,
+      "loss": 0.9441,
+      "step": 53840
+    },
+    {
+      "epoch": 0.3898021672566179,
+      "grad_norm": 0.16241905093193054,
+      "learning_rate": 4.610205071409441e-06,
+      "loss": 0.9372,
+      "step": 53850
+    },
+    {
+      "epoch": 0.38987455391720416,
+      "grad_norm": 0.1761135458946228,
+      "learning_rate": 4.610132684748855e-06,
+      "loss": 0.9509,
+      "step": 53860
+    },
+    {
+      "epoch": 0.38994694057779034,
+      "grad_norm": 0.16063787043094635,
+      "learning_rate": 4.610060298088268e-06,
+      "loss": 0.9453,
+      "step": 53870
+    },
+    {
+      "epoch": 0.3900193272383765,
+      "grad_norm": 0.17681707441806793,
+      "learning_rate": 4.609987911427683e-06,
+      "loss": 0.9474,
+      "step": 53880
+    },
+    {
+      "epoch": 0.3900917138989627,
+      "grad_norm": 0.18907004594802856,
+      "learning_rate": 4.6099155247670964e-06,
+      "loss": 0.9496,
+      "step": 53890
+    },
+    {
+      "epoch": 0.39016410055954887,
+      "grad_norm": 0.1664309799671173,
+      "learning_rate": 4.60984313810651e-06,
+      "loss": 0.9371,
+      "step": 53900
+    },
+    {
+      "epoch": 0.3902364872201351,
+      "grad_norm": 0.16614876687526703,
+      "learning_rate": 4.609770751445924e-06,
+      "loss": 0.9383,
+      "step": 53910
+    },
+    {
+      "epoch": 0.3903088738807213,
+      "grad_norm": 0.16756542026996613,
+      "learning_rate": 4.609698364785338e-06,
+      "loss": 0.9444,
+      "step": 53920
+    },
+    {
+      "epoch": 0.39038126054130745,
+      "grad_norm": 0.19262923300266266,
+      "learning_rate": 4.609625978124752e-06,
+      "loss": 0.9412,
+      "step": 53930
+    },
+    {
+      "epoch": 0.3904536472018936,
+      "grad_norm": 0.1710319072008133,
+      "learning_rate": 4.609553591464165e-06,
+      "loss": 0.9517,
+      "step": 53940
+    },
+    {
+      "epoch": 0.3905260338624798,
+      "grad_norm": 0.15746866166591644,
+      "learning_rate": 4.609481204803579e-06,
+      "loss": 0.9364,
+      "step": 53950
+    },
+    {
+      "epoch": 0.39059842052306604,
+      "grad_norm": 0.16860923171043396,
+      "learning_rate": 4.6094088181429934e-06,
+      "loss": 0.9342,
+      "step": 53960
+    },
+    {
+      "epoch": 0.3906708071836522,
+      "grad_norm": 0.20959050953388214,
+      "learning_rate": 4.609336431482407e-06,
+      "loss": 0.9483,
+      "step": 53970
+    },
+    {
+      "epoch": 0.3907431938442384,
+      "grad_norm": 0.1634407639503479,
+      "learning_rate": 4.609264044821821e-06,
+      "loss": 0.9353,
+      "step": 53980
+    },
+    {
+      "epoch": 0.39081558050482457,
+      "grad_norm": 0.16322919726371765,
+      "learning_rate": 4.609191658161234e-06,
+      "loss": 0.9264,
+      "step": 53990
+    },
+    {
+      "epoch": 0.39088796716541074,
+      "grad_norm": 0.14622409641742706,
+      "learning_rate": 4.609119271500648e-06,
+      "loss": 0.9348,
+      "step": 54000
+    },
+    {
+      "epoch": 0.3909603538259969,
+      "grad_norm": 0.1500609666109085,
+      "learning_rate": 4.609046884840062e-06,
+      "loss": 0.9372,
+      "step": 54010
+    },
+    {
+      "epoch": 0.39103274048658315,
+      "grad_norm": 0.18767625093460083,
+      "learning_rate": 4.608974498179476e-06,
+      "loss": 0.9268,
+      "step": 54020
+    },
+    {
+      "epoch": 0.39110512714716933,
+      "grad_norm": 0.16851924359798431,
+      "learning_rate": 4.60890211151889e-06,
+      "loss": 0.9589,
+      "step": 54030
+    },
+    {
+      "epoch": 0.3911775138077555,
+      "grad_norm": 0.1965654194355011,
+      "learning_rate": 4.608829724858303e-06,
+      "loss": 0.9422,
+      "step": 54040
+    },
+    {
+      "epoch": 0.3912499004683417,
+      "grad_norm": 0.15963523089885712,
+      "learning_rate": 4.608757338197718e-06,
+      "loss": 0.9442,
+      "step": 54050
+    },
+    {
+      "epoch": 0.39132228712892786,
+      "grad_norm": 0.16074392199516296,
+      "learning_rate": 4.608684951537131e-06,
+      "loss": 0.9486,
+      "step": 54060
+    },
+    {
+      "epoch": 0.3913946737895141,
+      "grad_norm": 0.1722477674484253,
+      "learning_rate": 4.608612564876545e-06,
+      "loss": 0.9376,
+      "step": 54070
+    },
+    {
+      "epoch": 0.39146706045010027,
+      "grad_norm": 0.16781273484230042,
+      "learning_rate": 4.6085401782159585e-06,
+      "loss": 0.9461,
+      "step": 54080
+    },
+    {
+      "epoch": 0.39153944711068644,
+      "grad_norm": 0.17819204926490784,
+      "learning_rate": 4.608467791555373e-06,
+      "loss": 0.9365,
+      "step": 54090
+    },
+    {
+      "epoch": 0.3916118337712726,
+      "grad_norm": 0.1631801426410675,
+      "learning_rate": 4.608395404894787e-06,
+      "loss": 0.9405,
+      "step": 54100
+    },
+    {
+      "epoch": 0.3916842204318588,
+      "grad_norm": 0.1547449231147766,
+      "learning_rate": 4.6083230182342e-06,
+      "loss": 0.9377,
+      "step": 54110
+    },
+    {
+      "epoch": 0.39175660709244503,
+      "grad_norm": 0.16404055058956146,
+      "learning_rate": 4.608250631573614e-06,
+      "loss": 0.9356,
+      "step": 54120
+    },
+    {
+      "epoch": 0.3918289937530312,
+      "grad_norm": 0.1674279123544693,
+      "learning_rate": 4.608178244913028e-06,
+      "loss": 0.9219,
+      "step": 54130
+    },
+    {
+      "epoch": 0.3919013804136174,
+      "grad_norm": 0.14838223159313202,
+      "learning_rate": 4.608105858252442e-06,
+      "loss": 0.9361,
+      "step": 54140
+    },
+    {
+      "epoch": 0.39197376707420356,
+      "grad_norm": 0.16514058411121368,
+      "learning_rate": 4.6080334715918555e-06,
+      "loss": 0.9414,
+      "step": 54150
+    },
+    {
+      "epoch": 0.39204615373478974,
+      "grad_norm": 0.16169683635234833,
+      "learning_rate": 4.607961084931269e-06,
+      "loss": 0.9349,
+      "step": 54160
+    },
+    {
+      "epoch": 0.3921185403953759,
+      "grad_norm": 0.1660471111536026,
+      "learning_rate": 4.607888698270684e-06,
+      "loss": 0.9509,
+      "step": 54170
+    },
+    {
+      "epoch": 0.39219092705596215,
+      "grad_norm": 0.1746322214603424,
+      "learning_rate": 4.607816311610097e-06,
+      "loss": 0.9371,
+      "step": 54180
+    },
+    {
+      "epoch": 0.3922633137165483,
+      "grad_norm": 0.16147038340568542,
+      "learning_rate": 4.607743924949511e-06,
+      "loss": 0.9442,
+      "step": 54190
+    },
+    {
+      "epoch": 0.3923357003771345,
+      "grad_norm": 0.1733742505311966,
+      "learning_rate": 4.6076715382889245e-06,
+      "loss": 0.9382,
+      "step": 54200
+    },
+    {
+      "epoch": 0.3924080870377207,
+      "grad_norm": 0.18472987413406372,
+      "learning_rate": 4.607599151628338e-06,
+      "loss": 0.9304,
+      "step": 54210
+    },
+    {
+      "epoch": 0.39248047369830685,
+      "grad_norm": 0.16237011551856995,
+      "learning_rate": 4.607526764967752e-06,
+      "loss": 0.9406,
+      "step": 54220
+    },
+    {
+      "epoch": 0.3925528603588931,
+      "grad_norm": 0.2291804850101471,
+      "learning_rate": 4.607454378307165e-06,
+      "loss": 0.9429,
+      "step": 54230
+    },
+    {
+      "epoch": 0.39262524701947926,
+      "grad_norm": 0.15868277847766876,
+      "learning_rate": 4.60738199164658e-06,
+      "loss": 0.9355,
+      "step": 54240
+    },
+    {
+      "epoch": 0.39269763368006544,
+      "grad_norm": 0.17885923385620117,
+      "learning_rate": 4.607309604985993e-06,
+      "loss": 0.9412,
+      "step": 54250
+    },
+    {
+      "epoch": 0.3927700203406516,
+      "grad_norm": 0.1670595407485962,
+      "learning_rate": 4.607237218325407e-06,
+      "loss": 0.9417,
+      "step": 54260
+    },
+    {
+      "epoch": 0.3928424070012378,
+      "grad_norm": 0.15766389667987823,
+      "learning_rate": 4.607164831664821e-06,
+      "loss": 0.9434,
+      "step": 54270
+    },
+    {
+      "epoch": 0.392914793661824,
+      "grad_norm": 0.16400344669818878,
+      "learning_rate": 4.607092445004235e-06,
+      "loss": 0.9412,
+      "step": 54280
+    },
+    {
+      "epoch": 0.3929871803224102,
+      "grad_norm": 0.16506126523017883,
+      "learning_rate": 4.607020058343649e-06,
+      "loss": 0.9409,
+      "step": 54290
+    },
+    {
+      "epoch": 0.3930595669829964,
+      "grad_norm": 0.16669459640979767,
+      "learning_rate": 4.606947671683062e-06,
+      "loss": 0.9329,
+      "step": 54300
+    },
+    {
+      "epoch": 0.39313195364358255,
+      "grad_norm": 0.18172895908355713,
+      "learning_rate": 4.606875285022476e-06,
+      "loss": 0.9446,
+      "step": 54310
+    },
+    {
+      "epoch": 0.39320434030416873,
+      "grad_norm": 0.1678895652294159,
+      "learning_rate": 4.60680289836189e-06,
+      "loss": 0.9421,
+      "step": 54320
+    },
+    {
+      "epoch": 0.3932767269647549,
+      "grad_norm": 0.15718461573123932,
+      "learning_rate": 4.606730511701304e-06,
+      "loss": 0.9324,
+      "step": 54330
+    },
+    {
+      "epoch": 0.39334911362534114,
+      "grad_norm": 0.16904427111148834,
+      "learning_rate": 4.606658125040718e-06,
+      "loss": 0.9321,
+      "step": 54340
+    },
+    {
+      "epoch": 0.3934215002859273,
+      "grad_norm": 0.16376027464866638,
+      "learning_rate": 4.606585738380131e-06,
+      "loss": 0.9483,
+      "step": 54350
+    },
+    {
+      "epoch": 0.3934938869465135,
+      "grad_norm": 0.15465879440307617,
+      "learning_rate": 4.606513351719546e-06,
+      "loss": 0.931,
+      "step": 54360
+    },
+    {
+      "epoch": 0.39356627360709967,
+      "grad_norm": 0.171060249209404,
+      "learning_rate": 4.606440965058959e-06,
+      "loss": 0.9366,
+      "step": 54370
+    },
+    {
+      "epoch": 0.39363866026768585,
+      "grad_norm": 0.1620955467224121,
+      "learning_rate": 4.606368578398373e-06,
+      "loss": 0.9333,
+      "step": 54380
+    },
+    {
+      "epoch": 0.3937110469282721,
+      "grad_norm": 0.1702883541584015,
+      "learning_rate": 4.6062961917377866e-06,
+      "loss": 0.9442,
+      "step": 54390
+    },
+    {
+      "epoch": 0.39378343358885826,
+      "grad_norm": 0.16319170594215393,
+      "learning_rate": 4.606223805077201e-06,
+      "loss": 0.9315,
+      "step": 54400
+    },
+    {
+      "epoch": 0.39385582024944443,
+      "grad_norm": 0.16425073146820068,
+      "learning_rate": 4.606151418416615e-06,
+      "loss": 0.9407,
+      "step": 54410
+    },
+    {
+      "epoch": 0.3939282069100306,
+      "grad_norm": 0.1441926509141922,
+      "learning_rate": 4.606079031756028e-06,
+      "loss": 0.9393,
+      "step": 54420
+    },
+    {
+      "epoch": 0.3940005935706168,
+      "grad_norm": 0.1872769445180893,
+      "learning_rate": 4.606006645095442e-06,
+      "loss": 0.9417,
+      "step": 54430
+    },
+    {
+      "epoch": 0.394072980231203,
+      "grad_norm": 0.15714821219444275,
+      "learning_rate": 4.605934258434856e-06,
+      "loss": 0.9343,
+      "step": 54440
+    },
+    {
+      "epoch": 0.3941453668917892,
+      "grad_norm": 0.16071169078350067,
+      "learning_rate": 4.60586187177427e-06,
+      "loss": 0.9447,
+      "step": 54450
+    },
+    {
+      "epoch": 0.39421775355237537,
+      "grad_norm": 0.15835289657115936,
+      "learning_rate": 4.6057894851136836e-06,
+      "loss": 0.95,
+      "step": 54460
+    },
+    {
+      "epoch": 0.39429014021296155,
+      "grad_norm": 0.15297073125839233,
+      "learning_rate": 4.605717098453097e-06,
+      "loss": 0.9402,
+      "step": 54470
+    },
+    {
+      "epoch": 0.3943625268735477,
+      "grad_norm": 0.1636633574962616,
+      "learning_rate": 4.605644711792512e-06,
+      "loss": 0.9444,
+      "step": 54480
+    },
+    {
+      "epoch": 0.39443491353413396,
+      "grad_norm": 0.15706664323806763,
+      "learning_rate": 4.605572325131925e-06,
+      "loss": 0.9578,
+      "step": 54490
+    },
+    {
+      "epoch": 0.39450730019472013,
+      "grad_norm": 0.15871857106685638,
+      "learning_rate": 4.605499938471339e-06,
+      "loss": 0.9271,
+      "step": 54500
+    },
+    {
+      "epoch": 0.3945796868553063,
+      "grad_norm": 0.16093158721923828,
+      "learning_rate": 4.6054275518107525e-06,
+      "loss": 0.9584,
+      "step": 54510
+    },
+    {
+      "epoch": 0.3946520735158925,
+      "grad_norm": 0.2100452184677124,
+      "learning_rate": 4.605355165150167e-06,
+      "loss": 0.9364,
+      "step": 54520
+    },
+    {
+      "epoch": 0.39472446017647866,
+      "grad_norm": 0.1580246388912201,
+      "learning_rate": 4.6052827784895806e-06,
+      "loss": 0.9394,
+      "step": 54530
+    },
+    {
+      "epoch": 0.39479684683706484,
+      "grad_norm": 0.16030821204185486,
+      "learning_rate": 4.605210391828994e-06,
+      "loss": 0.9389,
+      "step": 54540
+    },
+    {
+      "epoch": 0.3948692334976511,
+      "grad_norm": 0.17322495579719543,
+      "learning_rate": 4.605138005168408e-06,
+      "loss": 0.9363,
+      "step": 54550
+    },
+    {
+      "epoch": 0.39494162015823725,
+      "grad_norm": 0.19092749059200287,
+      "learning_rate": 4.605065618507822e-06,
+      "loss": 0.9458,
+      "step": 54560
+    },
+    {
+      "epoch": 0.3950140068188234,
+      "grad_norm": 0.14710497856140137,
+      "learning_rate": 4.604993231847236e-06,
+      "loss": 0.9424,
+      "step": 54570
+    },
+    {
+      "epoch": 0.3950863934794096,
+      "grad_norm": 0.17393282055854797,
+      "learning_rate": 4.6049208451866495e-06,
+      "loss": 0.9462,
+      "step": 54580
+    },
+    {
+      "epoch": 0.3951587801399958,
+      "grad_norm": 0.1747346818447113,
+      "learning_rate": 4.604848458526063e-06,
+      "loss": 0.9349,
+      "step": 54590
+    },
+    {
+      "epoch": 0.395231166800582,
+      "grad_norm": 0.1654556542634964,
+      "learning_rate": 4.604776071865477e-06,
+      "loss": 0.9427,
+      "step": 54600
+    },
+    {
+      "epoch": 0.3953035534611682,
+      "grad_norm": 0.1744566261768341,
+      "learning_rate": 4.604703685204891e-06,
+      "loss": 0.9456,
+      "step": 54610
+    },
+    {
+      "epoch": 0.39537594012175437,
+      "grad_norm": 0.1695273369550705,
+      "learning_rate": 4.604631298544305e-06,
+      "loss": 0.943,
+      "step": 54620
+    },
+    {
+      "epoch": 0.39544832678234054,
+      "grad_norm": 0.17201577126979828,
+      "learning_rate": 4.6045589118837184e-06,
+      "loss": 0.9452,
+      "step": 54630
+    },
+    {
+      "epoch": 0.3955207134429267,
+      "grad_norm": 0.18179161846637726,
+      "learning_rate": 4.604486525223132e-06,
+      "loss": 0.93,
+      "step": 54640
+    },
+    {
+      "epoch": 0.39559310010351295,
+      "grad_norm": 0.16751891374588013,
+      "learning_rate": 4.6044141385625465e-06,
+      "loss": 0.9283,
+      "step": 54650
+    },
+    {
+      "epoch": 0.39566548676409913,
+      "grad_norm": 0.15544885396957397,
+      "learning_rate": 4.60434175190196e-06,
+      "loss": 0.946,
+      "step": 54660
+    },
+    {
+      "epoch": 0.3957378734246853,
+      "grad_norm": 0.15887774527072906,
+      "learning_rate": 4.604269365241374e-06,
+      "loss": 0.9441,
+      "step": 54670
+    },
+    {
+      "epoch": 0.3958102600852715,
+      "grad_norm": 0.16352955996990204,
+      "learning_rate": 4.604196978580787e-06,
+      "loss": 0.9427,
+      "step": 54680
+    },
+    {
+      "epoch": 0.39588264674585766,
+      "grad_norm": 0.1617666482925415,
+      "learning_rate": 4.604124591920202e-06,
+      "loss": 0.9415,
+      "step": 54690
+    },
+    {
+      "epoch": 0.39595503340644383,
+      "grad_norm": 0.15895050764083862,
+      "learning_rate": 4.6040522052596154e-06,
+      "loss": 0.953,
+      "step": 54700
+    },
+    {
+      "epoch": 0.39602742006703007,
+      "grad_norm": 0.158672496676445,
+      "learning_rate": 4.603979818599029e-06,
+      "loss": 0.9369,
+      "step": 54710
+    },
+    {
+      "epoch": 0.39609980672761624,
+      "grad_norm": 0.17853258550167084,
+      "learning_rate": 4.603907431938443e-06,
+      "loss": 0.9429,
+      "step": 54720
+    },
+    {
+      "epoch": 0.3961721933882024,
+      "grad_norm": 0.1713215857744217,
+      "learning_rate": 4.603835045277856e-06,
+      "loss": 0.935,
+      "step": 54730
+    },
+    {
+      "epoch": 0.3962445800487886,
+      "grad_norm": 0.1543017029762268,
+      "learning_rate": 4.60376265861727e-06,
+      "loss": 0.9297,
+      "step": 54740
+    },
+    {
+      "epoch": 0.3963169667093748,
+      "grad_norm": 0.17130827903747559,
+      "learning_rate": 4.603690271956684e-06,
+      "loss": 0.9472,
+      "step": 54750
+    },
+    {
+      "epoch": 0.396389353369961,
+      "grad_norm": 0.1728035807609558,
+      "learning_rate": 4.603617885296098e-06,
+      "loss": 0.9448,
+      "step": 54760
+    },
+    {
+      "epoch": 0.3964617400305472,
+      "grad_norm": 0.16144831478595734,
+      "learning_rate": 4.603545498635512e-06,
+      "loss": 0.9515,
+      "step": 54770
+    },
+    {
+      "epoch": 0.39653412669113336,
+      "grad_norm": 0.15876467525959015,
+      "learning_rate": 4.603473111974925e-06,
+      "loss": 0.945,
+      "step": 54780
+    },
+    {
+      "epoch": 0.39660651335171954,
+      "grad_norm": 0.1569388210773468,
+      "learning_rate": 4.603400725314339e-06,
+      "loss": 0.9366,
+      "step": 54790
+    },
+    {
+      "epoch": 0.3966789000123057,
+      "grad_norm": 0.1643538773059845,
+      "learning_rate": 4.603328338653753e-06,
+      "loss": 0.9337,
+      "step": 54800
+    },
+    {
+      "epoch": 0.39675128667289195,
+      "grad_norm": 0.15085361897945404,
+      "learning_rate": 4.603255951993167e-06,
+      "loss": 0.9506,
+      "step": 54810
+    },
+    {
+      "epoch": 0.3968236733334781,
+      "grad_norm": 0.15235082805156708,
+      "learning_rate": 4.6031835653325805e-06,
+      "loss": 0.9261,
+      "step": 54820
+    },
+    {
+      "epoch": 0.3968960599940643,
+      "grad_norm": 0.15740101039409637,
+      "learning_rate": 4.603111178671994e-06,
+      "loss": 0.9483,
+      "step": 54830
+    },
+    {
+      "epoch": 0.3969684466546505,
+      "grad_norm": 0.17915703356266022,
+      "learning_rate": 4.603038792011409e-06,
+      "loss": 0.9333,
+      "step": 54840
+    },
+    {
+      "epoch": 0.39704083331523665,
+      "grad_norm": 0.18708273768424988,
+      "learning_rate": 4.602966405350822e-06,
+      "loss": 0.9457,
+      "step": 54850
+    },
+    {
+      "epoch": 0.39711321997582283,
+      "grad_norm": 0.1566181778907776,
+      "learning_rate": 4.602894018690236e-06,
+      "loss": 0.9296,
+      "step": 54860
+    },
+    {
+      "epoch": 0.39718560663640906,
+      "grad_norm": 0.20385728776454926,
+      "learning_rate": 4.6028216320296495e-06,
+      "loss": 0.9426,
+      "step": 54870
+    },
+    {
+      "epoch": 0.39725799329699524,
+      "grad_norm": 0.15828682482242584,
+      "learning_rate": 4.602749245369064e-06,
+      "loss": 0.9304,
+      "step": 54880
+    },
+    {
+      "epoch": 0.3973303799575814,
+      "grad_norm": 0.16250182688236237,
+      "learning_rate": 4.6026768587084775e-06,
+      "loss": 0.9444,
+      "step": 54890
+    },
+    {
+      "epoch": 0.3974027666181676,
+      "grad_norm": 0.16645143926143646,
+      "learning_rate": 4.602604472047891e-06,
+      "loss": 0.9462,
+      "step": 54900
+    },
+    {
+      "epoch": 0.39747515327875377,
+      "grad_norm": 0.17297451198101044,
+      "learning_rate": 4.602532085387305e-06,
+      "loss": 0.9392,
+      "step": 54910
+    },
+    {
+      "epoch": 0.39754753993934,
+      "grad_norm": 0.16319099068641663,
+      "learning_rate": 4.602459698726719e-06,
+      "loss": 0.9384,
+      "step": 54920
+    },
+    {
+      "epoch": 0.3976199265999262,
+      "grad_norm": 0.17599353194236755,
+      "learning_rate": 4.602387312066133e-06,
+      "loss": 0.9392,
+      "step": 54930
+    },
+    {
+      "epoch": 0.39769231326051235,
+      "grad_norm": 0.16673387587070465,
+      "learning_rate": 4.6023149254055465e-06,
+      "loss": 0.9367,
+      "step": 54940
+    },
+    {
+      "epoch": 0.39776469992109853,
+      "grad_norm": 0.17548586428165436,
+      "learning_rate": 4.60224253874496e-06,
+      "loss": 0.9384,
+      "step": 54950
+    },
+    {
+      "epoch": 0.3978370865816847,
+      "grad_norm": 0.15889111161231995,
+      "learning_rate": 4.6021701520843745e-06,
+      "loss": 0.9319,
+      "step": 54960
+    },
+    {
+      "epoch": 0.39790947324227094,
+      "grad_norm": 0.15126605331897736,
+      "learning_rate": 4.602097765423788e-06,
+      "loss": 0.9389,
+      "step": 54970
+    },
+    {
+      "epoch": 0.3979818599028571,
+      "grad_norm": 0.1905619353055954,
+      "learning_rate": 4.602025378763202e-06,
+      "loss": 0.9516,
+      "step": 54980
+    },
+    {
+      "epoch": 0.3980542465634433,
+      "grad_norm": 0.18187619745731354,
+      "learning_rate": 4.601952992102615e-06,
+      "loss": 0.9277,
+      "step": 54990
+    },
+    {
+      "epoch": 0.39812663322402947,
+      "grad_norm": 0.1634269803762436,
+      "learning_rate": 4.60188060544203e-06,
+      "loss": 0.9381,
+      "step": 55000
+    },
+    {
+      "epoch": 0.39819901988461565,
+      "grad_norm": 0.16369114816188812,
+      "learning_rate": 4.6018082187814435e-06,
+      "loss": 0.9403,
+      "step": 55010
+    },
+    {
+      "epoch": 0.3982714065452018,
+      "grad_norm": 0.17202910780906677,
+      "learning_rate": 4.601735832120857e-06,
+      "loss": 0.9411,
+      "step": 55020
+    },
+    {
+      "epoch": 0.39834379320578805,
+      "grad_norm": 0.17395305633544922,
+      "learning_rate": 4.601663445460271e-06,
+      "loss": 0.9236,
+      "step": 55030
+    },
+    {
+      "epoch": 0.39841617986637423,
+      "grad_norm": 0.17423111200332642,
+      "learning_rate": 4.601591058799685e-06,
+      "loss": 0.9404,
+      "step": 55040
+    },
+    {
+      "epoch": 0.3984885665269604,
+      "grad_norm": 0.15829895436763763,
+      "learning_rate": 4.601518672139099e-06,
+      "loss": 0.9254,
+      "step": 55050
+    },
+    {
+      "epoch": 0.3985609531875466,
+      "grad_norm": 0.17353561520576477,
+      "learning_rate": 4.601446285478512e-06,
+      "loss": 0.9326,
+      "step": 55060
+    },
+    {
+      "epoch": 0.39863333984813276,
+      "grad_norm": 0.1960097998380661,
+      "learning_rate": 4.601373898817926e-06,
+      "loss": 0.9582,
+      "step": 55070
+    },
+    {
+      "epoch": 0.398705726508719,
+      "grad_norm": 0.15898558497428894,
+      "learning_rate": 4.6013015121573405e-06,
+      "loss": 0.9513,
+      "step": 55080
+    },
+    {
+      "epoch": 0.39877811316930517,
+      "grad_norm": 0.17398889362812042,
+      "learning_rate": 4.601229125496754e-06,
+      "loss": 0.9299,
+      "step": 55090
+    },
+    {
+      "epoch": 0.39885049982989135,
+      "grad_norm": 0.21108154952526093,
+      "learning_rate": 4.601156738836168e-06,
+      "loss": 0.9514,
+      "step": 55100
+    },
+    {
+      "epoch": 0.3989228864904775,
+      "grad_norm": 0.15072329342365265,
+      "learning_rate": 4.601084352175581e-06,
+      "loss": 0.9276,
+      "step": 55110
+    },
+    {
+      "epoch": 0.3989952731510637,
+      "grad_norm": 0.18398626148700714,
+      "learning_rate": 4.601011965514996e-06,
+      "loss": 0.9373,
+      "step": 55120
+    },
+    {
+      "epoch": 0.39906765981164993,
+      "grad_norm": 0.26923868060112,
+      "learning_rate": 4.600939578854409e-06,
+      "loss": 0.9442,
+      "step": 55130
+    },
+    {
+      "epoch": 0.3991400464722361,
+      "grad_norm": 0.16427594423294067,
+      "learning_rate": 4.600867192193823e-06,
+      "loss": 0.9444,
+      "step": 55140
+    },
+    {
+      "epoch": 0.3992124331328223,
+      "grad_norm": 0.17052388191223145,
+      "learning_rate": 4.600794805533237e-06,
+      "loss": 0.947,
+      "step": 55150
+    },
+    {
+      "epoch": 0.39928481979340846,
+      "grad_norm": 0.18360772728919983,
+      "learning_rate": 4.600722418872651e-06,
+      "loss": 0.9419,
+      "step": 55160
+    },
+    {
+      "epoch": 0.39935720645399464,
+      "grad_norm": 0.16440702974796295,
+      "learning_rate": 4.600650032212065e-06,
+      "loss": 0.9214,
+      "step": 55170
+    },
+    {
+      "epoch": 0.39942959311458087,
+      "grad_norm": 0.15219788253307343,
+      "learning_rate": 4.600577645551478e-06,
+      "loss": 0.957,
+      "step": 55180
+    },
+    {
+      "epoch": 0.39950197977516705,
+      "grad_norm": 0.16597037017345428,
+      "learning_rate": 4.600505258890892e-06,
+      "loss": 0.9424,
+      "step": 55190
+    },
+    {
+      "epoch": 0.3995743664357532,
+      "grad_norm": 0.2117663323879242,
+      "learning_rate": 4.600432872230306e-06,
+      "loss": 0.9397,
+      "step": 55200
+    },
+    {
+      "epoch": 0.3996467530963394,
+      "grad_norm": 0.16794174909591675,
+      "learning_rate": 4.60036048556972e-06,
+      "loss": 0.9553,
+      "step": 55210
+    },
+    {
+      "epoch": 0.3997191397569256,
+      "grad_norm": 0.1689329296350479,
+      "learning_rate": 4.600288098909134e-06,
+      "loss": 0.9436,
+      "step": 55220
+    },
+    {
+      "epoch": 0.39979152641751176,
+      "grad_norm": 0.1604575663805008,
+      "learning_rate": 4.600215712248547e-06,
+      "loss": 0.942,
+      "step": 55230
+    },
+    {
+      "epoch": 0.399863913078098,
+      "grad_norm": 0.16758602857589722,
+      "learning_rate": 4.600143325587961e-06,
+      "loss": 0.9411,
+      "step": 55240
+    },
+    {
+      "epoch": 0.39993629973868416,
+      "grad_norm": 0.167888805270195,
+      "learning_rate": 4.600070938927375e-06,
+      "loss": 0.9467,
+      "step": 55250
+    },
+    {
+      "epoch": 0.40000868639927034,
+      "grad_norm": 0.16341939568519592,
+      "learning_rate": 4.599998552266788e-06,
+      "loss": 0.9469,
+      "step": 55260
+    },
+    {
+      "epoch": 0.4000810730598565,
+      "grad_norm": 0.15733623504638672,
+      "learning_rate": 4.5999261656062026e-06,
+      "loss": 0.9472,
+      "step": 55270
+    },
+    {
+      "epoch": 0.4001534597204427,
+      "grad_norm": 0.15709537267684937,
+      "learning_rate": 4.599853778945616e-06,
+      "loss": 0.9363,
+      "step": 55280
+    },
+    {
+      "epoch": 0.4002258463810289,
+      "grad_norm": 0.1718069463968277,
+      "learning_rate": 4.59978139228503e-06,
+      "loss": 0.9339,
+      "step": 55290
+    },
+    {
+      "epoch": 0.4002982330416151,
+      "grad_norm": 0.15917843580245972,
+      "learning_rate": 4.599709005624443e-06,
+      "loss": 0.9363,
+      "step": 55300
+    },
+    {
+      "epoch": 0.4003706197022013,
+      "grad_norm": 0.15500162541866302,
+      "learning_rate": 4.599636618963858e-06,
+      "loss": 0.9295,
+      "step": 55310
+    },
+    {
+      "epoch": 0.40044300636278746,
+      "grad_norm": 0.1756095588207245,
+      "learning_rate": 4.5995642323032715e-06,
+      "loss": 0.9356,
+      "step": 55320
+    },
+    {
+      "epoch": 0.40051539302337363,
+      "grad_norm": 0.16309203207492828,
+      "learning_rate": 4.599491845642685e-06,
+      "loss": 0.9378,
+      "step": 55330
+    },
+    {
+      "epoch": 0.40058777968395987,
+      "grad_norm": 0.16361068189144135,
+      "learning_rate": 4.599419458982099e-06,
+      "loss": 0.9376,
+      "step": 55340
+    },
+    {
+      "epoch": 0.40066016634454604,
+      "grad_norm": 0.16683772206306458,
+      "learning_rate": 4.599347072321513e-06,
+      "loss": 0.9455,
+      "step": 55350
+    },
+    {
+      "epoch": 0.4007325530051322,
+      "grad_norm": 0.1727033108472824,
+      "learning_rate": 4.599274685660927e-06,
+      "loss": 0.9312,
+      "step": 55360
+    },
+    {
+      "epoch": 0.4008049396657184,
+      "grad_norm": 0.22760114073753357,
+      "learning_rate": 4.5992022990003404e-06,
+      "loss": 0.9331,
+      "step": 55370
+    },
+    {
+      "epoch": 0.4008773263263046,
+      "grad_norm": 0.16040877997875214,
+      "learning_rate": 4.599129912339754e-06,
+      "loss": 0.9554,
+      "step": 55380
+    },
+    {
+      "epoch": 0.40094971298689075,
+      "grad_norm": 0.17181342840194702,
+      "learning_rate": 4.599057525679168e-06,
+      "loss": 0.9522,
+      "step": 55390
+    },
+    {
+      "epoch": 0.401022099647477,
+      "grad_norm": 0.16263987123966217,
+      "learning_rate": 4.598985139018582e-06,
+      "loss": 0.9407,
+      "step": 55400
+    },
+    {
+      "epoch": 0.40109448630806316,
+      "grad_norm": 0.16319935023784637,
+      "learning_rate": 4.598912752357996e-06,
+      "loss": 0.9337,
+      "step": 55410
+    },
+    {
+      "epoch": 0.40116687296864934,
+      "grad_norm": 0.17521025240421295,
+      "learning_rate": 4.598840365697409e-06,
+      "loss": 0.9386,
+      "step": 55420
+    },
+    {
+      "epoch": 0.4012392596292355,
+      "grad_norm": 0.16432663798332214,
+      "learning_rate": 4.598767979036823e-06,
+      "loss": 0.9387,
+      "step": 55430
+    },
+    {
+      "epoch": 0.4013116462898217,
+      "grad_norm": 0.16351726651191711,
+      "learning_rate": 4.5986955923762374e-06,
+      "loss": 0.9424,
+      "step": 55440
+    },
+    {
+      "epoch": 0.4013840329504079,
+      "grad_norm": 0.15134920179843903,
+      "learning_rate": 4.598623205715651e-06,
+      "loss": 0.9225,
+      "step": 55450
+    },
+    {
+      "epoch": 0.4014564196109941,
+      "grad_norm": 0.1916009783744812,
+      "learning_rate": 4.598550819055065e-06,
+      "loss": 0.942,
+      "step": 55460
+    },
+    {
+      "epoch": 0.4015288062715803,
+      "grad_norm": 0.1707172840833664,
+      "learning_rate": 4.598478432394478e-06,
+      "loss": 0.9395,
+      "step": 55470
+    },
+    {
+      "epoch": 0.40160119293216645,
+      "grad_norm": 0.20359551906585693,
+      "learning_rate": 4.598406045733893e-06,
+      "loss": 0.9398,
+      "step": 55480
+    },
+    {
+      "epoch": 0.4016735795927526,
+      "grad_norm": 0.16758215427398682,
+      "learning_rate": 4.598333659073306e-06,
+      "loss": 0.9317,
+      "step": 55490
+    },
+    {
+      "epoch": 0.40174596625333886,
+      "grad_norm": 0.17241047322750092,
+      "learning_rate": 4.59826127241272e-06,
+      "loss": 0.9463,
+      "step": 55500
+    },
+    {
+      "epoch": 0.40181835291392504,
+      "grad_norm": 0.17256489396095276,
+      "learning_rate": 4.598188885752134e-06,
+      "loss": 0.9383,
+      "step": 55510
+    },
+    {
+      "epoch": 0.4018907395745112,
+      "grad_norm": 0.16239801049232483,
+      "learning_rate": 4.598116499091548e-06,
+      "loss": 0.9344,
+      "step": 55520
+    },
+    {
+      "epoch": 0.4019631262350974,
+      "grad_norm": 0.15352807939052582,
+      "learning_rate": 4.598044112430962e-06,
+      "loss": 0.9391,
+      "step": 55530
+    },
+    {
+      "epoch": 0.40203551289568357,
+      "grad_norm": 0.15835584700107574,
+      "learning_rate": 4.597971725770375e-06,
+      "loss": 0.9441,
+      "step": 55540
+    },
+    {
+      "epoch": 0.40210789955626974,
+      "grad_norm": 0.17135757207870483,
+      "learning_rate": 4.597899339109789e-06,
+      "loss": 0.9403,
+      "step": 55550
+    },
+    {
+      "epoch": 0.402180286216856,
+      "grad_norm": 0.18612030148506165,
+      "learning_rate": 4.597826952449203e-06,
+      "loss": 0.954,
+      "step": 55560
+    },
+    {
+      "epoch": 0.40225267287744215,
+      "grad_norm": 0.16078492999076843,
+      "learning_rate": 4.597754565788617e-06,
+      "loss": 0.947,
+      "step": 55570
+    },
+    {
+      "epoch": 0.40232505953802833,
+      "grad_norm": 0.16282187402248383,
+      "learning_rate": 4.597682179128031e-06,
+      "loss": 0.9369,
+      "step": 55580
+    },
+    {
+      "epoch": 0.4023974461986145,
+      "grad_norm": 0.18029029667377472,
+      "learning_rate": 4.597609792467444e-06,
+      "loss": 0.9432,
+      "step": 55590
+    },
+    {
+      "epoch": 0.4024698328592007,
+      "grad_norm": 0.15934710204601288,
+      "learning_rate": 4.597537405806859e-06,
+      "loss": 0.9534,
+      "step": 55600
+    },
+    {
+      "epoch": 0.4025422195197869,
+      "grad_norm": 0.16303454339504242,
+      "learning_rate": 4.597465019146272e-06,
+      "loss": 0.9452,
+      "step": 55610
+    },
+    {
+      "epoch": 0.4026146061803731,
+      "grad_norm": 0.16990971565246582,
+      "learning_rate": 4.597392632485686e-06,
+      "loss": 0.9413,
+      "step": 55620
+    },
+    {
+      "epoch": 0.40268699284095927,
+      "grad_norm": 0.18538545072078705,
+      "learning_rate": 4.5973202458250995e-06,
+      "loss": 0.9341,
+      "step": 55630
+    },
+    {
+      "epoch": 0.40275937950154544,
+      "grad_norm": 0.16913770139217377,
+      "learning_rate": 4.597247859164514e-06,
+      "loss": 0.9415,
+      "step": 55640
+    },
+    {
+      "epoch": 0.4028317661621316,
+      "grad_norm": 0.1886037439107895,
+      "learning_rate": 4.597175472503928e-06,
+      "loss": 0.9407,
+      "step": 55650
+    },
+    {
+      "epoch": 0.40290415282271785,
+      "grad_norm": 0.1722469925880432,
+      "learning_rate": 4.597103085843341e-06,
+      "loss": 0.945,
+      "step": 55660
+    },
+    {
+      "epoch": 0.40297653948330403,
+      "grad_norm": 0.16154325008392334,
+      "learning_rate": 4.597030699182755e-06,
+      "loss": 0.9329,
+      "step": 55670
+    },
+    {
+      "epoch": 0.4030489261438902,
+      "grad_norm": 0.16594886779785156,
+      "learning_rate": 4.596958312522169e-06,
+      "loss": 0.946,
+      "step": 55680
+    },
+    {
+      "epoch": 0.4031213128044764,
+      "grad_norm": 0.15325994789600372,
+      "learning_rate": 4.596885925861583e-06,
+      "loss": 0.9348,
+      "step": 55690
+    },
+    {
+      "epoch": 0.40319369946506256,
+      "grad_norm": 0.1615227907896042,
+      "learning_rate": 4.5968135392009965e-06,
+      "loss": 0.9294,
+      "step": 55700
+    },
+    {
+      "epoch": 0.4032660861256488,
+      "grad_norm": 0.1949525773525238,
+      "learning_rate": 4.59674115254041e-06,
+      "loss": 0.9388,
+      "step": 55710
+    },
+    {
+      "epoch": 0.40333847278623497,
+      "grad_norm": 0.1611870676279068,
+      "learning_rate": 4.596668765879825e-06,
+      "loss": 0.9464,
+      "step": 55720
+    },
+    {
+      "epoch": 0.40341085944682115,
+      "grad_norm": 0.16078144311904907,
+      "learning_rate": 4.596596379219238e-06,
+      "loss": 0.9239,
+      "step": 55730
+    },
+    {
+      "epoch": 0.4034832461074073,
+      "grad_norm": 0.16597573459148407,
+      "learning_rate": 4.596523992558652e-06,
+      "loss": 0.9325,
+      "step": 55740
+    },
+    {
+      "epoch": 0.4035556327679935,
+      "grad_norm": 0.16398844122886658,
+      "learning_rate": 4.5964516058980655e-06,
+      "loss": 0.942,
+      "step": 55750
+    },
+    {
+      "epoch": 0.4036280194285797,
+      "grad_norm": 0.16814014315605164,
+      "learning_rate": 4.59637921923748e-06,
+      "loss": 0.9423,
+      "step": 55760
+    },
+    {
+      "epoch": 0.4037004060891659,
+      "grad_norm": 0.1564667820930481,
+      "learning_rate": 4.5963068325768935e-06,
+      "loss": 0.939,
+      "step": 55770
+    },
+    {
+      "epoch": 0.4037727927497521,
+      "grad_norm": 0.16583354771137238,
+      "learning_rate": 4.596234445916307e-06,
+      "loss": 0.9374,
+      "step": 55780
+    },
+    {
+      "epoch": 0.40384517941033826,
+      "grad_norm": 0.1725219190120697,
+      "learning_rate": 4.596162059255721e-06,
+      "loss": 0.9323,
+      "step": 55790
+    },
+    {
+      "epoch": 0.40391756607092444,
+      "grad_norm": 0.15034452080726624,
+      "learning_rate": 4.596089672595134e-06,
+      "loss": 0.9276,
+      "step": 55800
+    },
+    {
+      "epoch": 0.4039899527315106,
+      "grad_norm": 0.17058463394641876,
+      "learning_rate": 4.596017285934548e-06,
+      "loss": 0.9497,
+      "step": 55810
+    },
+    {
+      "epoch": 0.40406233939209685,
+      "grad_norm": 0.15496298670768738,
+      "learning_rate": 4.595944899273962e-06,
+      "loss": 0.9416,
+      "step": 55820
+    },
+    {
+      "epoch": 0.404134726052683,
+      "grad_norm": 0.16956356167793274,
+      "learning_rate": 4.595872512613376e-06,
+      "loss": 0.9354,
+      "step": 55830
+    },
+    {
+      "epoch": 0.4042071127132692,
+      "grad_norm": 0.15078359842300415,
+      "learning_rate": 4.59580012595279e-06,
+      "loss": 0.9287,
+      "step": 55840
+    },
+    {
+      "epoch": 0.4042794993738554,
+      "grad_norm": 0.1605517864227295,
+      "learning_rate": 4.595727739292203e-06,
+      "loss": 0.9434,
+      "step": 55850
+    },
+    {
+      "epoch": 0.40435188603444155,
+      "grad_norm": 0.16339904069900513,
+      "learning_rate": 4.595655352631617e-06,
+      "loss": 0.9406,
+      "step": 55860
+    },
+    {
+      "epoch": 0.4044242726950278,
+      "grad_norm": 0.18486179411411285,
+      "learning_rate": 4.595582965971031e-06,
+      "loss": 0.9444,
+      "step": 55870
+    },
+    {
+      "epoch": 0.40449665935561396,
+      "grad_norm": 0.16515012085437775,
+      "learning_rate": 4.595510579310445e-06,
+      "loss": 0.9375,
+      "step": 55880
+    },
+    {
+      "epoch": 0.40456904601620014,
+      "grad_norm": 0.16577614843845367,
+      "learning_rate": 4.595438192649859e-06,
+      "loss": 0.9506,
+      "step": 55890
+    },
+    {
+      "epoch": 0.4046414326767863,
+      "grad_norm": 0.16598597168922424,
+      "learning_rate": 4.595365805989272e-06,
+      "loss": 0.9229,
+      "step": 55900
+    },
+    {
+      "epoch": 0.4047138193373725,
+      "grad_norm": 0.1684425175189972,
+      "learning_rate": 4.595293419328687e-06,
+      "loss": 0.9365,
+      "step": 55910
+    },
+    {
+      "epoch": 0.40478620599795867,
+      "grad_norm": 0.15921281278133392,
+      "learning_rate": 4.5952210326681e-06,
+      "loss": 0.9394,
+      "step": 55920
+    },
+    {
+      "epoch": 0.4048585926585449,
+      "grad_norm": 0.16952553391456604,
+      "learning_rate": 4.595148646007514e-06,
+      "loss": 0.9461,
+      "step": 55930
+    },
+    {
+      "epoch": 0.4049309793191311,
+      "grad_norm": 0.1839471012353897,
+      "learning_rate": 4.5950762593469276e-06,
+      "loss": 0.9523,
+      "step": 55940
+    },
+    {
+      "epoch": 0.40500336597971726,
+      "grad_norm": 0.16442671418190002,
+      "learning_rate": 4.595003872686342e-06,
+      "loss": 0.9517,
+      "step": 55950
+    },
+    {
+      "epoch": 0.40507575264030343,
+      "grad_norm": 0.1691816747188568,
+      "learning_rate": 4.594931486025756e-06,
+      "loss": 0.9393,
+      "step": 55960
+    },
+    {
+      "epoch": 0.4051481393008896,
+      "grad_norm": 0.17636831104755402,
+      "learning_rate": 4.594859099365169e-06,
+      "loss": 0.9443,
+      "step": 55970
+    },
+    {
+      "epoch": 0.40522052596147584,
+      "grad_norm": 0.16882002353668213,
+      "learning_rate": 4.594786712704583e-06,
+      "loss": 0.9407,
+      "step": 55980
+    },
+    {
+      "epoch": 0.405292912622062,
+      "grad_norm": 0.16932517290115356,
+      "learning_rate": 4.594714326043997e-06,
+      "loss": 0.9443,
+      "step": 55990
+    },
+    {
+      "epoch": 0.4053652992826482,
+      "grad_norm": 0.17036424577236176,
+      "learning_rate": 4.594641939383411e-06,
+      "loss": 0.9496,
+      "step": 56000
+    },
+    {
+      "epoch": 0.40543768594323437,
+      "grad_norm": 0.16463550925254822,
+      "learning_rate": 4.5945695527228246e-06,
+      "loss": 0.9404,
+      "step": 56010
+    },
+    {
+      "epoch": 0.40551007260382055,
+      "grad_norm": 0.16589607298374176,
+      "learning_rate": 4.594497166062238e-06,
+      "loss": 0.9475,
+      "step": 56020
+    },
+    {
+      "epoch": 0.4055824592644068,
+      "grad_norm": 0.18529045581817627,
+      "learning_rate": 4.594424779401652e-06,
+      "loss": 0.9464,
+      "step": 56030
+    },
+    {
+      "epoch": 0.40565484592499296,
+      "grad_norm": 0.15794023871421814,
+      "learning_rate": 4.594352392741066e-06,
+      "loss": 0.9315,
+      "step": 56040
+    },
+    {
+      "epoch": 0.40572723258557913,
+      "grad_norm": 0.16272369027137756,
+      "learning_rate": 4.59428000608048e-06,
+      "loss": 0.9486,
+      "step": 56050
+    },
+    {
+      "epoch": 0.4057996192461653,
+      "grad_norm": 0.16422978043556213,
+      "learning_rate": 4.5942076194198935e-06,
+      "loss": 0.9569,
+      "step": 56060
+    },
+    {
+      "epoch": 0.4058720059067515,
+      "grad_norm": 0.15604273974895477,
+      "learning_rate": 4.594135232759307e-06,
+      "loss": 0.9499,
+      "step": 56070
+    },
+    {
+      "epoch": 0.40594439256733766,
+      "grad_norm": 0.17885945737361908,
+      "learning_rate": 4.5940628460987216e-06,
+      "loss": 0.9409,
+      "step": 56080
+    },
+    {
+      "epoch": 0.4060167792279239,
+      "grad_norm": 0.19414786994457245,
+      "learning_rate": 4.593990459438135e-06,
+      "loss": 0.9399,
+      "step": 56090
+    },
+    {
+      "epoch": 0.4060891658885101,
+      "grad_norm": 0.1650743931531906,
+      "learning_rate": 4.593918072777549e-06,
+      "loss": 0.9367,
+      "step": 56100
+    },
+    {
+      "epoch": 0.40616155254909625,
+      "grad_norm": 0.1537282019853592,
+      "learning_rate": 4.5938456861169624e-06,
+      "loss": 0.9325,
+      "step": 56110
+    },
+    {
+      "epoch": 0.4062339392096824,
+      "grad_norm": 0.16103175282478333,
+      "learning_rate": 4.593773299456377e-06,
+      "loss": 0.9262,
+      "step": 56120
+    },
+    {
+      "epoch": 0.4063063258702686,
+      "grad_norm": 0.1802387833595276,
+      "learning_rate": 4.5937009127957905e-06,
+      "loss": 0.9475,
+      "step": 56130
+    },
+    {
+      "epoch": 0.40637871253085484,
+      "grad_norm": 0.16064372658729553,
+      "learning_rate": 4.593628526135204e-06,
+      "loss": 0.9182,
+      "step": 56140
+    },
+    {
+      "epoch": 0.406451099191441,
+      "grad_norm": 0.15981996059417725,
+      "learning_rate": 4.593556139474618e-06,
+      "loss": 0.934,
+      "step": 56150
+    },
+    {
+      "epoch": 0.4065234858520272,
+      "grad_norm": 0.153685063123703,
+      "learning_rate": 4.593483752814032e-06,
+      "loss": 0.9424,
+      "step": 56160
+    },
+    {
+      "epoch": 0.40659587251261337,
+      "grad_norm": 0.1737099438905716,
+      "learning_rate": 4.593411366153446e-06,
+      "loss": 0.9264,
+      "step": 56170
+    },
+    {
+      "epoch": 0.40666825917319954,
+      "grad_norm": 0.15085779130458832,
+      "learning_rate": 4.5933389794928594e-06,
+      "loss": 0.9393,
+      "step": 56180
+    },
+    {
+      "epoch": 0.4067406458337858,
+      "grad_norm": 0.16470050811767578,
+      "learning_rate": 4.593266592832273e-06,
+      "loss": 0.9372,
+      "step": 56190
+    },
+    {
+      "epoch": 0.40681303249437195,
+      "grad_norm": 0.17371892929077148,
+      "learning_rate": 4.5931942061716875e-06,
+      "loss": 0.9426,
+      "step": 56200
+    },
+    {
+      "epoch": 0.40688541915495813,
+      "grad_norm": 0.1519443243741989,
+      "learning_rate": 4.593121819511101e-06,
+      "loss": 0.9331,
+      "step": 56210
+    },
+    {
+      "epoch": 0.4069578058155443,
+      "grad_norm": 0.1687236726284027,
+      "learning_rate": 4.593049432850515e-06,
+      "loss": 0.9469,
+      "step": 56220
+    },
+    {
+      "epoch": 0.4070301924761305,
+      "grad_norm": 0.1575406938791275,
+      "learning_rate": 4.592977046189928e-06,
+      "loss": 0.938,
+      "step": 56230
+    },
+    {
+      "epoch": 0.4071025791367167,
+      "grad_norm": 0.15486252307891846,
+      "learning_rate": 4.592904659529343e-06,
+      "loss": 0.9325,
+      "step": 56240
+    },
+    {
+      "epoch": 0.4071749657973029,
+      "grad_norm": 0.16617028415203094,
+      "learning_rate": 4.5928322728687564e-06,
+      "loss": 0.9382,
+      "step": 56250
+    },
+    {
+      "epoch": 0.40724735245788907,
+      "grad_norm": 0.16430149972438812,
+      "learning_rate": 4.59275988620817e-06,
+      "loss": 0.9316,
+      "step": 56260
+    },
+    {
+      "epoch": 0.40731973911847524,
+      "grad_norm": 0.17749901115894318,
+      "learning_rate": 4.592687499547584e-06,
+      "loss": 0.9332,
+      "step": 56270
+    },
+    {
+      "epoch": 0.4073921257790614,
+      "grad_norm": 0.21616603434085846,
+      "learning_rate": 4.592615112886998e-06,
+      "loss": 0.9409,
+      "step": 56280
+    },
+    {
+      "epoch": 0.4074645124396476,
+      "grad_norm": 0.1658083200454712,
+      "learning_rate": 4.592542726226412e-06,
+      "loss": 0.9445,
+      "step": 56290
+    },
+    {
+      "epoch": 0.40753689910023383,
+      "grad_norm": 0.15681838989257812,
+      "learning_rate": 4.592470339565825e-06,
+      "loss": 0.9343,
+      "step": 56300
+    },
+    {
+      "epoch": 0.40760928576082,
+      "grad_norm": 0.15368938446044922,
+      "learning_rate": 4.592397952905239e-06,
+      "loss": 0.9424,
+      "step": 56310
+    },
+    {
+      "epoch": 0.4076816724214062,
+      "grad_norm": 0.24562257528305054,
+      "learning_rate": 4.5923255662446534e-06,
+      "loss": 0.9345,
+      "step": 56320
+    },
+    {
+      "epoch": 0.40775405908199236,
+      "grad_norm": 0.1594623327255249,
+      "learning_rate": 4.592253179584066e-06,
+      "loss": 0.9462,
+      "step": 56330
+    },
+    {
+      "epoch": 0.40782644574257854,
+      "grad_norm": 0.1747690588235855,
+      "learning_rate": 4.59218079292348e-06,
+      "loss": 0.9347,
+      "step": 56340
+    },
+    {
+      "epoch": 0.40789883240316477,
+      "grad_norm": 0.1642814725637436,
+      "learning_rate": 4.592108406262894e-06,
+      "loss": 0.9402,
+      "step": 56350
+    },
+    {
+      "epoch": 0.40797121906375095,
+      "grad_norm": 0.15779854357242584,
+      "learning_rate": 4.592036019602308e-06,
+      "loss": 0.9327,
+      "step": 56360
+    },
+    {
+      "epoch": 0.4080436057243371,
+      "grad_norm": 0.15080593526363373,
+      "learning_rate": 4.5919636329417215e-06,
+      "loss": 0.9267,
+      "step": 56370
+    },
+    {
+      "epoch": 0.4081159923849233,
+      "grad_norm": 0.1907733529806137,
+      "learning_rate": 4.591891246281135e-06,
+      "loss": 0.9358,
+      "step": 56380
+    },
+    {
+      "epoch": 0.4081883790455095,
+      "grad_norm": 0.18039512634277344,
+      "learning_rate": 4.59181885962055e-06,
+      "loss": 0.9398,
+      "step": 56390
+    },
+    {
+      "epoch": 0.4082607657060957,
+      "grad_norm": 0.17534703016281128,
+      "learning_rate": 4.591746472959963e-06,
+      "loss": 0.94,
+      "step": 56400
+    },
+    {
+      "epoch": 0.4083331523666819,
+      "grad_norm": 0.16364185512065887,
+      "learning_rate": 4.591674086299377e-06,
+      "loss": 0.9374,
+      "step": 56410
+    },
+    {
+      "epoch": 0.40840553902726806,
+      "grad_norm": 0.16373589634895325,
+      "learning_rate": 4.5916016996387905e-06,
+      "loss": 0.9334,
+      "step": 56420
+    },
+    {
+      "epoch": 0.40847792568785424,
+      "grad_norm": 0.249722421169281,
+      "learning_rate": 4.591529312978205e-06,
+      "loss": 0.9358,
+      "step": 56430
+    },
+    {
+      "epoch": 0.4085503123484404,
+      "grad_norm": 0.16310162842273712,
+      "learning_rate": 4.5914569263176185e-06,
+      "loss": 0.9516,
+      "step": 56440
+    },
+    {
+      "epoch": 0.4086226990090266,
+      "grad_norm": 0.15711535513401031,
+      "learning_rate": 4.591384539657032e-06,
+      "loss": 0.9374,
+      "step": 56450
+    },
+    {
+      "epoch": 0.4086950856696128,
+      "grad_norm": 0.16699931025505066,
+      "learning_rate": 4.591312152996446e-06,
+      "loss": 0.9478,
+      "step": 56460
+    },
+    {
+      "epoch": 0.408767472330199,
+      "grad_norm": 0.16310204565525055,
+      "learning_rate": 4.59123976633586e-06,
+      "loss": 0.9335,
+      "step": 56470
+    },
+    {
+      "epoch": 0.4088398589907852,
+      "grad_norm": 0.1538310945034027,
+      "learning_rate": 4.591167379675274e-06,
+      "loss": 0.9424,
+      "step": 56480
+    },
+    {
+      "epoch": 0.40891224565137135,
+      "grad_norm": 0.1658681482076645,
+      "learning_rate": 4.5910949930146875e-06,
+      "loss": 0.9302,
+      "step": 56490
+    },
+    {
+      "epoch": 0.40898463231195753,
+      "grad_norm": 0.18781772255897522,
+      "learning_rate": 4.591022606354101e-06,
+      "loss": 0.9369,
+      "step": 56500
+    },
+    {
+      "epoch": 0.40905701897254376,
+      "grad_norm": 0.17703670263290405,
+      "learning_rate": 4.5909502196935155e-06,
+      "loss": 0.9407,
+      "step": 56510
+    },
+    {
+      "epoch": 0.40912940563312994,
+      "grad_norm": 0.16049979627132416,
+      "learning_rate": 4.590877833032929e-06,
+      "loss": 0.9452,
+      "step": 56520
+    },
+    {
+      "epoch": 0.4092017922937161,
+      "grad_norm": 0.16507183015346527,
+      "learning_rate": 4.590805446372343e-06,
+      "loss": 0.9304,
+      "step": 56530
+    },
+    {
+      "epoch": 0.4092741789543023,
+      "grad_norm": 0.17175514996051788,
+      "learning_rate": 4.590733059711756e-06,
+      "loss": 0.939,
+      "step": 56540
+    },
+    {
+      "epoch": 0.40934656561488847,
+      "grad_norm": 0.15628288686275482,
+      "learning_rate": 4.590660673051171e-06,
+      "loss": 0.9413,
+      "step": 56550
+    },
+    {
+      "epoch": 0.4094189522754747,
+      "grad_norm": 0.17340143024921417,
+      "learning_rate": 4.5905882863905845e-06,
+      "loss": 0.9347,
+      "step": 56560
+    },
+    {
+      "epoch": 0.4094913389360609,
+      "grad_norm": 0.17264612019062042,
+      "learning_rate": 4.590515899729998e-06,
+      "loss": 0.9275,
+      "step": 56570
+    },
+    {
+      "epoch": 0.40956372559664705,
+      "grad_norm": 0.1592729389667511,
+      "learning_rate": 4.590443513069412e-06,
+      "loss": 0.936,
+      "step": 56580
+    },
+    {
+      "epoch": 0.40963611225723323,
+      "grad_norm": 0.17017214000225067,
+      "learning_rate": 4.590371126408826e-06,
+      "loss": 0.9455,
+      "step": 56590
+    },
+    {
+      "epoch": 0.4097084989178194,
+      "grad_norm": 0.25028762221336365,
+      "learning_rate": 4.59029873974824e-06,
+      "loss": 0.9368,
+      "step": 56600
+    },
+    {
+      "epoch": 0.4097808855784056,
+      "grad_norm": 0.1710430234670639,
+      "learning_rate": 4.590226353087653e-06,
+      "loss": 0.932,
+      "step": 56610
+    },
+    {
+      "epoch": 0.4098532722389918,
+      "grad_norm": 0.18898555636405945,
+      "learning_rate": 4.590153966427067e-06,
+      "loss": 0.9284,
+      "step": 56620
+    },
+    {
+      "epoch": 0.409925658899578,
+      "grad_norm": 0.15929999947547913,
+      "learning_rate": 4.590081579766481e-06,
+      "loss": 0.9334,
+      "step": 56630
+    },
+    {
+      "epoch": 0.40999804556016417,
+      "grad_norm": 0.18224512040615082,
+      "learning_rate": 4.590009193105895e-06,
+      "loss": 0.9523,
+      "step": 56640
+    },
+    {
+      "epoch": 0.41007043222075035,
+      "grad_norm": 0.16598254442214966,
+      "learning_rate": 4.589936806445309e-06,
+      "loss": 0.9455,
+      "step": 56650
+    },
+    {
+      "epoch": 0.4101428188813365,
+      "grad_norm": 0.1503967046737671,
+      "learning_rate": 4.589864419784722e-06,
+      "loss": 0.9469,
+      "step": 56660
+    },
+    {
+      "epoch": 0.41021520554192276,
+      "grad_norm": 0.17737853527069092,
+      "learning_rate": 4.589792033124136e-06,
+      "loss": 0.946,
+      "step": 56670
+    },
+    {
+      "epoch": 0.41028759220250893,
+      "grad_norm": 0.16925589740276337,
+      "learning_rate": 4.58971964646355e-06,
+      "loss": 0.9484,
+      "step": 56680
+    },
+    {
+      "epoch": 0.4103599788630951,
+      "grad_norm": 0.15373647212982178,
+      "learning_rate": 4.589647259802964e-06,
+      "loss": 0.9357,
+      "step": 56690
+    },
+    {
+      "epoch": 0.4104323655236813,
+      "grad_norm": 0.1628134548664093,
+      "learning_rate": 4.589574873142378e-06,
+      "loss": 0.9328,
+      "step": 56700
+    },
+    {
+      "epoch": 0.41050475218426746,
+      "grad_norm": 0.1642216444015503,
+      "learning_rate": 4.589502486481791e-06,
+      "loss": 0.9391,
+      "step": 56710
+    },
+    {
+      "epoch": 0.4105771388448537,
+      "grad_norm": 0.18736900389194489,
+      "learning_rate": 4.589430099821206e-06,
+      "loss": 0.9277,
+      "step": 56720
+    },
+    {
+      "epoch": 0.41064952550543987,
+      "grad_norm": 0.16991931200027466,
+      "learning_rate": 4.589357713160619e-06,
+      "loss": 0.9257,
+      "step": 56730
+    },
+    {
+      "epoch": 0.41072191216602605,
+      "grad_norm": 0.1831955909729004,
+      "learning_rate": 4.589285326500033e-06,
+      "loss": 0.9251,
+      "step": 56740
+    },
+    {
+      "epoch": 0.4107942988266122,
+      "grad_norm": 0.16035877168178558,
+      "learning_rate": 4.5892129398394466e-06,
+      "loss": 0.9424,
+      "step": 56750
+    },
+    {
+      "epoch": 0.4108666854871984,
+      "grad_norm": 0.16036030650138855,
+      "learning_rate": 4.589140553178861e-06,
+      "loss": 0.9377,
+      "step": 56760
+    },
+    {
+      "epoch": 0.4109390721477846,
+      "grad_norm": 0.1828073412179947,
+      "learning_rate": 4.589068166518275e-06,
+      "loss": 0.9371,
+      "step": 56770
+    },
+    {
+      "epoch": 0.4110114588083708,
+      "grad_norm": 0.16397011280059814,
+      "learning_rate": 4.588995779857688e-06,
+      "loss": 0.9394,
+      "step": 56780
+    },
+    {
+      "epoch": 0.411083845468957,
+      "grad_norm": 0.14961424469947815,
+      "learning_rate": 4.588923393197102e-06,
+      "loss": 0.9346,
+      "step": 56790
+    },
+    {
+      "epoch": 0.41115623212954316,
+      "grad_norm": 0.16437679529190063,
+      "learning_rate": 4.588851006536516e-06,
+      "loss": 0.9223,
+      "step": 56800
+    },
+    {
+      "epoch": 0.41122861879012934,
+      "grad_norm": 0.15859009325504303,
+      "learning_rate": 4.58877861987593e-06,
+      "loss": 0.9479,
+      "step": 56810
+    },
+    {
+      "epoch": 0.4113010054507155,
+      "grad_norm": 0.15533125400543213,
+      "learning_rate": 4.5887062332153436e-06,
+      "loss": 0.9188,
+      "step": 56820
+    },
+    {
+      "epoch": 0.41137339211130175,
+      "grad_norm": 0.16156156361103058,
+      "learning_rate": 4.588633846554757e-06,
+      "loss": 0.9446,
+      "step": 56830
+    },
+    {
+      "epoch": 0.4114457787718879,
+      "grad_norm": 0.169430211186409,
+      "learning_rate": 4.588561459894172e-06,
+      "loss": 0.9298,
+      "step": 56840
+    },
+    {
+      "epoch": 0.4115181654324741,
+      "grad_norm": 0.18322457373142242,
+      "learning_rate": 4.588489073233585e-06,
+      "loss": 0.9431,
+      "step": 56850
+    },
+    {
+      "epoch": 0.4115905520930603,
+      "grad_norm": 0.17053550481796265,
+      "learning_rate": 4.588416686572998e-06,
+      "loss": 0.9349,
+      "step": 56860
+    },
+    {
+      "epoch": 0.41166293875364646,
+      "grad_norm": 0.18169362843036652,
+      "learning_rate": 4.5883442999124125e-06,
+      "loss": 0.9257,
+      "step": 56870
+    },
+    {
+      "epoch": 0.4117353254142327,
+      "grad_norm": 0.2942195236682892,
+      "learning_rate": 4.588271913251826e-06,
+      "loss": 0.9326,
+      "step": 56880
+    },
+    {
+      "epoch": 0.41180771207481887,
+      "grad_norm": 0.21721670031547546,
+      "learning_rate": 4.58819952659124e-06,
+      "loss": 0.9412,
+      "step": 56890
+    },
+    {
+      "epoch": 0.41188009873540504,
+      "grad_norm": 0.1641770750284195,
+      "learning_rate": 4.588127139930653e-06,
+      "loss": 0.9299,
+      "step": 56900
+    },
+    {
+      "epoch": 0.4119524853959912,
+      "grad_norm": 0.17069405317306519,
+      "learning_rate": 4.588054753270068e-06,
+      "loss": 0.9389,
+      "step": 56910
+    },
+    {
+      "epoch": 0.4120248720565774,
+      "grad_norm": 0.16552869975566864,
+      "learning_rate": 4.5879823666094814e-06,
+      "loss": 0.943,
+      "step": 56920
+    },
+    {
+      "epoch": 0.41209725871716363,
+      "grad_norm": 0.2061159610748291,
+      "learning_rate": 4.587909979948895e-06,
+      "loss": 0.9433,
+      "step": 56930
+    },
+    {
+      "epoch": 0.4121696453777498,
+      "grad_norm": 0.16571536660194397,
+      "learning_rate": 4.587837593288309e-06,
+      "loss": 0.9441,
+      "step": 56940
+    },
+    {
+      "epoch": 0.412242032038336,
+      "grad_norm": 0.15703128278255463,
+      "learning_rate": 4.587765206627723e-06,
+      "loss": 0.94,
+      "step": 56950
+    },
+    {
+      "epoch": 0.41231441869892216,
+      "grad_norm": 0.16326795518398285,
+      "learning_rate": 4.587692819967137e-06,
+      "loss": 0.9363,
+      "step": 56960
+    },
+    {
+      "epoch": 0.41238680535950833,
+      "grad_norm": 0.16321924328804016,
+      "learning_rate": 4.58762043330655e-06,
+      "loss": 0.9219,
+      "step": 56970
+    },
+    {
+      "epoch": 0.4124591920200945,
+      "grad_norm": 0.16104656457901,
+      "learning_rate": 4.587548046645964e-06,
+      "loss": 0.9407,
+      "step": 56980
+    },
+    {
+      "epoch": 0.41253157868068074,
+      "grad_norm": 0.16342920064926147,
+      "learning_rate": 4.5874756599853784e-06,
+      "loss": 0.9396,
+      "step": 56990
+    },
+    {
+      "epoch": 0.4126039653412669,
+      "grad_norm": 0.18352803587913513,
+      "learning_rate": 4.587403273324792e-06,
+      "loss": 0.9289,
+      "step": 57000
+    },
+    {
+      "epoch": 0.4126763520018531,
+      "grad_norm": 0.14929810166358948,
+      "learning_rate": 4.587330886664206e-06,
+      "loss": 0.9403,
+      "step": 57010
+    },
+    {
+      "epoch": 0.4127487386624393,
+      "grad_norm": 0.15996743738651276,
+      "learning_rate": 4.587258500003619e-06,
+      "loss": 0.9487,
+      "step": 57020
+    },
+    {
+      "epoch": 0.41282112532302545,
+      "grad_norm": 0.15918391942977905,
+      "learning_rate": 4.587186113343034e-06,
+      "loss": 0.9334,
+      "step": 57030
+    },
+    {
+      "epoch": 0.4128935119836117,
+      "grad_norm": 0.16165630519390106,
+      "learning_rate": 4.587113726682447e-06,
+      "loss": 0.9335,
+      "step": 57040
+    },
+    {
+      "epoch": 0.41296589864419786,
+      "grad_norm": 0.16731780767440796,
+      "learning_rate": 4.587041340021861e-06,
+      "loss": 0.9401,
+      "step": 57050
+    },
+    {
+      "epoch": 0.41303828530478404,
+      "grad_norm": 0.21615374088287354,
+      "learning_rate": 4.586968953361275e-06,
+      "loss": 0.9381,
+      "step": 57060
+    },
+    {
+      "epoch": 0.4131106719653702,
+      "grad_norm": 0.17595277726650238,
+      "learning_rate": 4.586896566700689e-06,
+      "loss": 0.9435,
+      "step": 57070
+    },
+    {
+      "epoch": 0.4131830586259564,
+      "grad_norm": 0.2352835237979889,
+      "learning_rate": 4.586824180040103e-06,
+      "loss": 0.9311,
+      "step": 57080
+    },
+    {
+      "epoch": 0.4132554452865426,
+      "grad_norm": 0.15337133407592773,
+      "learning_rate": 4.586751793379516e-06,
+      "loss": 0.9489,
+      "step": 57090
+    },
+    {
+      "epoch": 0.4133278319471288,
+      "grad_norm": 0.18751218914985657,
+      "learning_rate": 4.58667940671893e-06,
+      "loss": 0.9385,
+      "step": 57100
+    },
+    {
+      "epoch": 0.413400218607715,
+      "grad_norm": 0.3108257055282593,
+      "learning_rate": 4.586607020058344e-06,
+      "loss": 0.9399,
+      "step": 57110
+    },
+    {
+      "epoch": 0.41347260526830115,
+      "grad_norm": 0.15695133805274963,
+      "learning_rate": 4.586534633397758e-06,
+      "loss": 0.931,
+      "step": 57120
+    },
+    {
+      "epoch": 0.41354499192888733,
+      "grad_norm": 0.1587478369474411,
+      "learning_rate": 4.586462246737172e-06,
+      "loss": 0.9228,
+      "step": 57130
+    },
+    {
+      "epoch": 0.4136173785894735,
+      "grad_norm": 0.17306017875671387,
+      "learning_rate": 4.586389860076585e-06,
+      "loss": 0.9294,
+      "step": 57140
+    },
+    {
+      "epoch": 0.41368976525005974,
+      "grad_norm": 0.1602945774793625,
+      "learning_rate": 4.586317473416e-06,
+      "loss": 0.9387,
+      "step": 57150
+    },
+    {
+      "epoch": 0.4137621519106459,
+      "grad_norm": 0.17871308326721191,
+      "learning_rate": 4.586245086755413e-06,
+      "loss": 0.9349,
+      "step": 57160
+    },
+    {
+      "epoch": 0.4138345385712321,
+      "grad_norm": 0.18266141414642334,
+      "learning_rate": 4.586172700094827e-06,
+      "loss": 0.9463,
+      "step": 57170
+    },
+    {
+      "epoch": 0.41390692523181827,
+      "grad_norm": 0.15963898599147797,
+      "learning_rate": 4.5861003134342405e-06,
+      "loss": 0.9383,
+      "step": 57180
+    },
+    {
+      "epoch": 0.41397931189240444,
+      "grad_norm": 0.18138666450977325,
+      "learning_rate": 4.586027926773655e-06,
+      "loss": 0.9332,
+      "step": 57190
+    },
+    {
+      "epoch": 0.4140516985529907,
+      "grad_norm": 0.16729888319969177,
+      "learning_rate": 4.585955540113069e-06,
+      "loss": 0.9286,
+      "step": 57200
+    },
+    {
+      "epoch": 0.41412408521357685,
+      "grad_norm": 0.17177051305770874,
+      "learning_rate": 4.585883153452482e-06,
+      "loss": 0.9317,
+      "step": 57210
+    },
+    {
+      "epoch": 0.41419647187416303,
+      "grad_norm": 0.17003445327281952,
+      "learning_rate": 4.585810766791896e-06,
+      "loss": 0.9415,
+      "step": 57220
+    },
+    {
+      "epoch": 0.4142688585347492,
+      "grad_norm": 0.3177216649055481,
+      "learning_rate": 4.58573838013131e-06,
+      "loss": 0.9546,
+      "step": 57230
+    },
+    {
+      "epoch": 0.4143412451953354,
+      "grad_norm": 0.15487100183963776,
+      "learning_rate": 4.585665993470724e-06,
+      "loss": 0.9362,
+      "step": 57240
+    },
+    {
+      "epoch": 0.4144136318559216,
+      "grad_norm": 0.158743754029274,
+      "learning_rate": 4.5855936068101375e-06,
+      "loss": 0.941,
+      "step": 57250
+    },
+    {
+      "epoch": 0.4144860185165078,
+      "grad_norm": 0.16190417110919952,
+      "learning_rate": 4.585521220149551e-06,
+      "loss": 0.9304,
+      "step": 57260
+    },
+    {
+      "epoch": 0.41455840517709397,
+      "grad_norm": 0.16800768673419952,
+      "learning_rate": 4.585448833488965e-06,
+      "loss": 0.9331,
+      "step": 57270
+    },
+    {
+      "epoch": 0.41463079183768015,
+      "grad_norm": 0.15928290784358978,
+      "learning_rate": 4.585376446828379e-06,
+      "loss": 0.9298,
+      "step": 57280
+    },
+    {
+      "epoch": 0.4147031784982663,
+      "grad_norm": 0.17402148246765137,
+      "learning_rate": 4.585304060167793e-06,
+      "loss": 0.9381,
+      "step": 57290
+    },
+    {
+      "epoch": 0.4147755651588525,
+      "grad_norm": 0.16189469397068024,
+      "learning_rate": 4.5852316735072065e-06,
+      "loss": 0.9407,
+      "step": 57300
+    },
+    {
+      "epoch": 0.41484795181943873,
+      "grad_norm": 0.16404691338539124,
+      "learning_rate": 4.58515928684662e-06,
+      "loss": 0.9396,
+      "step": 57310
+    },
+    {
+      "epoch": 0.4149203384800249,
+      "grad_norm": 0.15537384152412415,
+      "learning_rate": 4.5850869001860345e-06,
+      "loss": 0.9291,
+      "step": 57320
+    },
+    {
+      "epoch": 0.4149927251406111,
+      "grad_norm": 0.16837313771247864,
+      "learning_rate": 4.585014513525448e-06,
+      "loss": 0.9345,
+      "step": 57330
+    },
+    {
+      "epoch": 0.41506511180119726,
+      "grad_norm": 0.17618413269519806,
+      "learning_rate": 4.584942126864862e-06,
+      "loss": 0.9325,
+      "step": 57340
+    },
+    {
+      "epoch": 0.41513749846178344,
+      "grad_norm": 0.16522805392742157,
+      "learning_rate": 4.584869740204275e-06,
+      "loss": 0.9328,
+      "step": 57350
+    },
+    {
+      "epoch": 0.41520988512236967,
+      "grad_norm": 0.18333765864372253,
+      "learning_rate": 4.58479735354369e-06,
+      "loss": 0.9246,
+      "step": 57360
+    },
+    {
+      "epoch": 0.41528227178295585,
+      "grad_norm": 0.18241846561431885,
+      "learning_rate": 4.5847249668831035e-06,
+      "loss": 0.9327,
+      "step": 57370
+    },
+    {
+      "epoch": 0.415354658443542,
+      "grad_norm": 0.1651061475276947,
+      "learning_rate": 4.584652580222517e-06,
+      "loss": 0.9293,
+      "step": 57380
+    },
+    {
+      "epoch": 0.4154270451041282,
+      "grad_norm": 0.14222125709056854,
+      "learning_rate": 4.584580193561931e-06,
+      "loss": 0.9349,
+      "step": 57390
+    },
+    {
+      "epoch": 0.4154994317647144,
+      "grad_norm": 0.17054183781147003,
+      "learning_rate": 4.584507806901344e-06,
+      "loss": 0.9352,
+      "step": 57400
+    },
+    {
+      "epoch": 0.4155718184253006,
+      "grad_norm": 0.16434767842292786,
+      "learning_rate": 4.584435420240758e-06,
+      "loss": 0.9448,
+      "step": 57410
+    },
+    {
+      "epoch": 0.4156442050858868,
+      "grad_norm": 0.15329429507255554,
+      "learning_rate": 4.584363033580172e-06,
+      "loss": 0.9352,
+      "step": 57420
+    },
+    {
+      "epoch": 0.41571659174647296,
+      "grad_norm": 0.1638394594192505,
+      "learning_rate": 4.584290646919586e-06,
+      "loss": 0.927,
+      "step": 57430
+    },
+    {
+      "epoch": 0.41578897840705914,
+      "grad_norm": 0.1653471738100052,
+      "learning_rate": 4.584218260259e-06,
+      "loss": 0.9451,
+      "step": 57440
+    },
+    {
+      "epoch": 0.4158613650676453,
+      "grad_norm": 0.1822325438261032,
+      "learning_rate": 4.584145873598413e-06,
+      "loss": 0.9372,
+      "step": 57450
+    },
+    {
+      "epoch": 0.41593375172823155,
+      "grad_norm": 0.16440334916114807,
+      "learning_rate": 4.584073486937827e-06,
+      "loss": 0.9434,
+      "step": 57460
+    },
+    {
+      "epoch": 0.4160061383888177,
+      "grad_norm": 0.1566876322031021,
+      "learning_rate": 4.584001100277241e-06,
+      "loss": 0.9454,
+      "step": 57470
+    },
+    {
+      "epoch": 0.4160785250494039,
+      "grad_norm": 0.15633165836334229,
+      "learning_rate": 4.583928713616655e-06,
+      "loss": 0.9438,
+      "step": 57480
+    },
+    {
+      "epoch": 0.4161509117099901,
+      "grad_norm": 0.1904531866312027,
+      "learning_rate": 4.5838563269560686e-06,
+      "loss": 0.9433,
+      "step": 57490
+    },
+    {
+      "epoch": 0.41622329837057626,
+      "grad_norm": 0.15552623569965363,
+      "learning_rate": 4.583783940295482e-06,
+      "loss": 0.9351,
+      "step": 57500
+    },
+    {
+      "epoch": 0.41629568503116243,
+      "grad_norm": 0.2083786427974701,
+      "learning_rate": 4.583711553634897e-06,
+      "loss": 0.9324,
+      "step": 57510
+    },
+    {
+      "epoch": 0.41636807169174866,
+      "grad_norm": 0.16016191244125366,
+      "learning_rate": 4.58363916697431e-06,
+      "loss": 0.9311,
+      "step": 57520
+    },
+    {
+      "epoch": 0.41644045835233484,
+      "grad_norm": 0.1558094024658203,
+      "learning_rate": 4.583566780313724e-06,
+      "loss": 0.9262,
+      "step": 57530
+    },
+    {
+      "epoch": 0.416512845012921,
+      "grad_norm": 0.16131691634655,
+      "learning_rate": 4.5834943936531375e-06,
+      "loss": 0.9343,
+      "step": 57540
+    },
+    {
+      "epoch": 0.4165852316735072,
+      "grad_norm": 0.1501682698726654,
+      "learning_rate": 4.583422006992552e-06,
+      "loss": 0.944,
+      "step": 57550
+    },
+    {
+      "epoch": 0.41665761833409337,
+      "grad_norm": 0.18568944931030273,
+      "learning_rate": 4.5833496203319656e-06,
+      "loss": 0.9248,
+      "step": 57560
+    },
+    {
+      "epoch": 0.4167300049946796,
+      "grad_norm": 0.16975927352905273,
+      "learning_rate": 4.583277233671379e-06,
+      "loss": 0.9348,
+      "step": 57570
+    },
+    {
+      "epoch": 0.4168023916552658,
+      "grad_norm": 0.15231750905513763,
+      "learning_rate": 4.583204847010793e-06,
+      "loss": 0.9445,
+      "step": 57580
+    },
+    {
+      "epoch": 0.41687477831585196,
+      "grad_norm": 0.2114260494709015,
+      "learning_rate": 4.583132460350207e-06,
+      "loss": 0.9306,
+      "step": 57590
+    },
+    {
+      "epoch": 0.41694716497643813,
+      "grad_norm": 0.17215648293495178,
+      "learning_rate": 4.583060073689621e-06,
+      "loss": 0.9206,
+      "step": 57600
+    },
+    {
+      "epoch": 0.4170195516370243,
+      "grad_norm": 0.16319842636585236,
+      "learning_rate": 4.5829876870290345e-06,
+      "loss": 0.9321,
+      "step": 57610
+    },
+    {
+      "epoch": 0.41709193829761054,
+      "grad_norm": 0.1795549988746643,
+      "learning_rate": 4.582915300368448e-06,
+      "loss": 0.9346,
+      "step": 57620
+    },
+    {
+      "epoch": 0.4171643249581967,
+      "grad_norm": 0.18302929401397705,
+      "learning_rate": 4.582842913707863e-06,
+      "loss": 0.9413,
+      "step": 57630
+    },
+    {
+      "epoch": 0.4172367116187829,
+      "grad_norm": 0.16946756839752197,
+      "learning_rate": 4.582770527047276e-06,
+      "loss": 0.9397,
+      "step": 57640
+    },
+    {
+      "epoch": 0.4173090982793691,
+      "grad_norm": 0.16745789349079132,
+      "learning_rate": 4.58269814038669e-06,
+      "loss": 0.9381,
+      "step": 57650
+    },
+    {
+      "epoch": 0.41738148493995525,
+      "grad_norm": 0.15687045454978943,
+      "learning_rate": 4.5826257537261034e-06,
+      "loss": 0.9341,
+      "step": 57660
+    },
+    {
+      "epoch": 0.4174538716005414,
+      "grad_norm": 0.1755257248878479,
+      "learning_rate": 4.582553367065518e-06,
+      "loss": 0.9442,
+      "step": 57670
+    },
+    {
+      "epoch": 0.41752625826112766,
+      "grad_norm": 0.1668943166732788,
+      "learning_rate": 4.5824809804049315e-06,
+      "loss": 0.9397,
+      "step": 57680
+    },
+    {
+      "epoch": 0.41759864492171384,
+      "grad_norm": 0.16578532755374908,
+      "learning_rate": 4.582408593744345e-06,
+      "loss": 0.9375,
+      "step": 57690
+    },
+    {
+      "epoch": 0.4176710315823,
+      "grad_norm": 0.15865838527679443,
+      "learning_rate": 4.582336207083759e-06,
+      "loss": 0.9431,
+      "step": 57700
+    },
+    {
+      "epoch": 0.4177434182428862,
+      "grad_norm": 0.1591697633266449,
+      "learning_rate": 4.582263820423173e-06,
+      "loss": 0.9313,
+      "step": 57710
+    },
+    {
+      "epoch": 0.41781580490347237,
+      "grad_norm": 0.23013950884342194,
+      "learning_rate": 4.582191433762587e-06,
+      "loss": 0.9357,
+      "step": 57720
+    },
+    {
+      "epoch": 0.4178881915640586,
+      "grad_norm": 0.1570298671722412,
+      "learning_rate": 4.5821190471020004e-06,
+      "loss": 0.9206,
+      "step": 57730
+    },
+    {
+      "epoch": 0.4179605782246448,
+      "grad_norm": 0.16404032707214355,
+      "learning_rate": 4.582046660441414e-06,
+      "loss": 0.9465,
+      "step": 57740
+    },
+    {
+      "epoch": 0.41803296488523095,
+      "grad_norm": 0.16130004823207855,
+      "learning_rate": 4.5819742737808285e-06,
+      "loss": 0.9389,
+      "step": 57750
+    },
+    {
+      "epoch": 0.4181053515458171,
+      "grad_norm": 0.1546400487422943,
+      "learning_rate": 4.581901887120242e-06,
+      "loss": 0.9361,
+      "step": 57760
+    },
+    {
+      "epoch": 0.4181777382064033,
+      "grad_norm": 0.2353212833404541,
+      "learning_rate": 4.581829500459656e-06,
+      "loss": 0.9321,
+      "step": 57770
+    },
+    {
+      "epoch": 0.41825012486698954,
+      "grad_norm": 0.16489247977733612,
+      "learning_rate": 4.581757113799069e-06,
+      "loss": 0.9414,
+      "step": 57780
+    },
+    {
+      "epoch": 0.4183225115275757,
+      "grad_norm": 0.17833788692951202,
+      "learning_rate": 4.581684727138484e-06,
+      "loss": 0.9375,
+      "step": 57790
+    },
+    {
+      "epoch": 0.4183948981881619,
+      "grad_norm": 0.16194747388362885,
+      "learning_rate": 4.5816123404778974e-06,
+      "loss": 0.9359,
+      "step": 57800
+    },
+    {
+      "epoch": 0.41846728484874807,
+      "grad_norm": 0.16397050023078918,
+      "learning_rate": 4.581539953817311e-06,
+      "loss": 0.9314,
+      "step": 57810
+    },
+    {
+      "epoch": 0.41853967150933424,
+      "grad_norm": 0.18386350572109222,
+      "learning_rate": 4.581467567156725e-06,
+      "loss": 0.9458,
+      "step": 57820
+    },
+    {
+      "epoch": 0.4186120581699204,
+      "grad_norm": 0.17196989059448242,
+      "learning_rate": 4.581395180496139e-06,
+      "loss": 0.9213,
+      "step": 57830
+    },
+    {
+      "epoch": 0.41868444483050665,
+      "grad_norm": 0.15747104585170746,
+      "learning_rate": 4.581322793835553e-06,
+      "loss": 0.9346,
+      "step": 57840
+    },
+    {
+      "epoch": 0.41875683149109283,
+      "grad_norm": 0.19495005905628204,
+      "learning_rate": 4.581250407174966e-06,
+      "loss": 0.9384,
+      "step": 57850
+    },
+    {
+      "epoch": 0.418829218151679,
+      "grad_norm": 0.15831686556339264,
+      "learning_rate": 4.58117802051438e-06,
+      "loss": 0.9382,
+      "step": 57860
+    },
+    {
+      "epoch": 0.4189016048122652,
+      "grad_norm": 0.16458268463611603,
+      "learning_rate": 4.5811056338537945e-06,
+      "loss": 0.92,
+      "step": 57870
+    },
+    {
+      "epoch": 0.41897399147285136,
+      "grad_norm": 0.1708781123161316,
+      "learning_rate": 4.581033247193208e-06,
+      "loss": 0.9377,
+      "step": 57880
+    },
+    {
+      "epoch": 0.4190463781334376,
+      "grad_norm": 0.18208129703998566,
+      "learning_rate": 4.580960860532622e-06,
+      "loss": 0.9399,
+      "step": 57890
+    },
+    {
+      "epoch": 0.41911876479402377,
+      "grad_norm": 0.15881425142288208,
+      "learning_rate": 4.580888473872035e-06,
+      "loss": 0.9298,
+      "step": 57900
+    },
+    {
+      "epoch": 0.41919115145460994,
+      "grad_norm": 0.1868477463722229,
+      "learning_rate": 4.580816087211449e-06,
+      "loss": 0.9473,
+      "step": 57910
+    },
+    {
+      "epoch": 0.4192635381151961,
+      "grad_norm": 0.17207710444927216,
+      "learning_rate": 4.5807437005508625e-06,
+      "loss": 0.9367,
+      "step": 57920
+    },
+    {
+      "epoch": 0.4193359247757823,
+      "grad_norm": 0.19365693628787994,
+      "learning_rate": 4.580671313890276e-06,
+      "loss": 0.9333,
+      "step": 57930
+    },
+    {
+      "epoch": 0.41940831143636853,
+      "grad_norm": 0.1523778736591339,
+      "learning_rate": 4.580598927229691e-06,
+      "loss": 0.9295,
+      "step": 57940
+    },
+    {
+      "epoch": 0.4194806980969547,
+      "grad_norm": 0.15326263010501862,
+      "learning_rate": 4.580526540569104e-06,
+      "loss": 0.9508,
+      "step": 57950
+    },
+    {
+      "epoch": 0.4195530847575409,
+      "grad_norm": 0.1532769799232483,
+      "learning_rate": 4.580454153908518e-06,
+      "loss": 0.9285,
+      "step": 57960
+    },
+    {
+      "epoch": 0.41962547141812706,
+      "grad_norm": 0.15870678424835205,
+      "learning_rate": 4.5803817672479315e-06,
+      "loss": 0.9472,
+      "step": 57970
+    },
+    {
+      "epoch": 0.41969785807871324,
+      "grad_norm": 0.1667070984840393,
+      "learning_rate": 4.580309380587346e-06,
+      "loss": 0.939,
+      "step": 57980
+    },
+    {
+      "epoch": 0.41977024473929947,
+      "grad_norm": 0.4463184177875519,
+      "learning_rate": 4.5802369939267595e-06,
+      "loss": 0.9313,
+      "step": 57990
+    },
+    {
+      "epoch": 0.41984263139988565,
+      "grad_norm": 0.15991568565368652,
+      "learning_rate": 4.580164607266173e-06,
+      "loss": 0.9425,
+      "step": 58000
+    },
+    {
+      "epoch": 0.4199150180604718,
+      "grad_norm": 0.16935986280441284,
+      "learning_rate": 4.580092220605587e-06,
+      "loss": 0.9291,
+      "step": 58010
+    },
+    {
+      "epoch": 0.419987404721058,
+      "grad_norm": 0.17099948227405548,
+      "learning_rate": 4.580019833945001e-06,
+      "loss": 0.9335,
+      "step": 58020
+    },
+    {
+      "epoch": 0.4200597913816442,
+      "grad_norm": 0.15163956582546234,
+      "learning_rate": 4.579947447284415e-06,
+      "loss": 0.9305,
+      "step": 58030
+    },
+    {
+      "epoch": 0.42013217804223035,
+      "grad_norm": 0.1671072095632553,
+      "learning_rate": 4.5798750606238285e-06,
+      "loss": 0.9379,
+      "step": 58040
+    },
+    {
+      "epoch": 0.4202045647028166,
+      "grad_norm": 0.16619133949279785,
+      "learning_rate": 4.579802673963242e-06,
+      "loss": 0.9309,
+      "step": 58050
+    },
+    {
+      "epoch": 0.42027695136340276,
+      "grad_norm": 0.16173526644706726,
+      "learning_rate": 4.579730287302656e-06,
+      "loss": 0.9369,
+      "step": 58060
+    },
+    {
+      "epoch": 0.42034933802398894,
+      "grad_norm": 0.16461287438869476,
+      "learning_rate": 4.57965790064207e-06,
+      "loss": 0.9386,
+      "step": 58070
+    },
+    {
+      "epoch": 0.4204217246845751,
+      "grad_norm": 0.19412462413311005,
+      "learning_rate": 4.579585513981484e-06,
+      "loss": 0.9284,
+      "step": 58080
+    },
+    {
+      "epoch": 0.4204941113451613,
+      "grad_norm": 0.18617956340312958,
+      "learning_rate": 4.579513127320897e-06,
+      "loss": 0.9284,
+      "step": 58090
+    },
+    {
+      "epoch": 0.4205664980057475,
+      "grad_norm": 0.1594972461462021,
+      "learning_rate": 4.579440740660311e-06,
+      "loss": 0.9367,
+      "step": 58100
+    },
+    {
+      "epoch": 0.4206388846663337,
+      "grad_norm": 0.1762949526309967,
+      "learning_rate": 4.5793683539997255e-06,
+      "loss": 0.9305,
+      "step": 58110
+    },
+    {
+      "epoch": 0.4207112713269199,
+      "grad_norm": 0.20204119384288788,
+      "learning_rate": 4.579295967339139e-06,
+      "loss": 0.939,
+      "step": 58120
+    },
+    {
+      "epoch": 0.42078365798750605,
+      "grad_norm": 0.16462816298007965,
+      "learning_rate": 4.579223580678553e-06,
+      "loss": 0.9302,
+      "step": 58130
+    },
+    {
+      "epoch": 0.42085604464809223,
+      "grad_norm": 0.15783581137657166,
+      "learning_rate": 4.579151194017966e-06,
+      "loss": 0.9372,
+      "step": 58140
+    },
+    {
+      "epoch": 0.42092843130867846,
+      "grad_norm": 0.1847236305475235,
+      "learning_rate": 4.579078807357381e-06,
+      "loss": 0.9338,
+      "step": 58150
+    },
+    {
+      "epoch": 0.42100081796926464,
+      "grad_norm": 0.16038931906223297,
+      "learning_rate": 4.579006420696794e-06,
+      "loss": 0.9347,
+      "step": 58160
+    },
+    {
+      "epoch": 0.4210732046298508,
+      "grad_norm": 0.17341868579387665,
+      "learning_rate": 4.578934034036208e-06,
+      "loss": 0.9468,
+      "step": 58170
+    },
+    {
+      "epoch": 0.421145591290437,
+      "grad_norm": 0.17107468843460083,
+      "learning_rate": 4.578861647375622e-06,
+      "loss": 0.9385,
+      "step": 58180
+    },
+    {
+      "epoch": 0.42121797795102317,
+      "grad_norm": 0.16987687349319458,
+      "learning_rate": 4.578789260715036e-06,
+      "loss": 0.9256,
+      "step": 58190
+    },
+    {
+      "epoch": 0.42129036461160935,
+      "grad_norm": 0.15816541016101837,
+      "learning_rate": 4.57871687405445e-06,
+      "loss": 0.9309,
+      "step": 58200
+    },
+    {
+      "epoch": 0.4213627512721956,
+      "grad_norm": 0.17220225930213928,
+      "learning_rate": 4.578644487393863e-06,
+      "loss": 0.931,
+      "step": 58210
+    },
+    {
+      "epoch": 0.42143513793278176,
+      "grad_norm": 0.15358738601207733,
+      "learning_rate": 4.578572100733277e-06,
+      "loss": 0.9363,
+      "step": 58220
+    },
+    {
+      "epoch": 0.42150752459336793,
+      "grad_norm": 0.1880992203950882,
+      "learning_rate": 4.578499714072691e-06,
+      "loss": 0.9336,
+      "step": 58230
+    },
+    {
+      "epoch": 0.4215799112539541,
+      "grad_norm": 0.1707928329706192,
+      "learning_rate": 4.578427327412105e-06,
+      "loss": 0.9444,
+      "step": 58240
+    },
+    {
+      "epoch": 0.4216522979145403,
+      "grad_norm": 0.16345572471618652,
+      "learning_rate": 4.578354940751519e-06,
+      "loss": 0.9401,
+      "step": 58250
+    },
+    {
+      "epoch": 0.4217246845751265,
+      "grad_norm": 0.15612851083278656,
+      "learning_rate": 4.578282554090932e-06,
+      "loss": 0.9325,
+      "step": 58260
+    },
+    {
+      "epoch": 0.4217970712357127,
+      "grad_norm": 0.15962456166744232,
+      "learning_rate": 4.578210167430347e-06,
+      "loss": 0.933,
+      "step": 58270
+    },
+    {
+      "epoch": 0.42186945789629887,
+      "grad_norm": 0.16563986241817474,
+      "learning_rate": 4.57813778076976e-06,
+      "loss": 0.9284,
+      "step": 58280
+    },
+    {
+      "epoch": 0.42194184455688505,
+      "grad_norm": 0.18625997006893158,
+      "learning_rate": 4.578065394109174e-06,
+      "loss": 0.927,
+      "step": 58290
+    },
+    {
+      "epoch": 0.4220142312174712,
+      "grad_norm": 0.16373422741889954,
+      "learning_rate": 4.5779930074485876e-06,
+      "loss": 0.9354,
+      "step": 58300
+    },
+    {
+      "epoch": 0.42208661787805746,
+      "grad_norm": 0.16232380270957947,
+      "learning_rate": 4.577920620788002e-06,
+      "loss": 0.9257,
+      "step": 58310
+    },
+    {
+      "epoch": 0.42215900453864363,
+      "grad_norm": 0.1764630377292633,
+      "learning_rate": 4.577848234127416e-06,
+      "loss": 0.9374,
+      "step": 58320
+    },
+    {
+      "epoch": 0.4222313911992298,
+      "grad_norm": 0.1576317995786667,
+      "learning_rate": 4.577775847466829e-06,
+      "loss": 0.9403,
+      "step": 58330
+    },
+    {
+      "epoch": 0.422303777859816,
+      "grad_norm": 0.18781952559947968,
+      "learning_rate": 4.577703460806243e-06,
+      "loss": 0.9278,
+      "step": 58340
+    },
+    {
+      "epoch": 0.42237616452040216,
+      "grad_norm": 0.1682749092578888,
+      "learning_rate": 4.577631074145657e-06,
+      "loss": 0.9339,
+      "step": 58350
+    },
+    {
+      "epoch": 0.42244855118098834,
+      "grad_norm": 0.17663206160068512,
+      "learning_rate": 4.577558687485071e-06,
+      "loss": 0.937,
+      "step": 58360
+    },
+    {
+      "epoch": 0.4225209378415746,
+      "grad_norm": 0.16293081641197205,
+      "learning_rate": 4.577486300824485e-06,
+      "loss": 0.924,
+      "step": 58370
+    },
+    {
+      "epoch": 0.42259332450216075,
+      "grad_norm": 0.17560496926307678,
+      "learning_rate": 4.577413914163898e-06,
+      "loss": 0.9319,
+      "step": 58380
+    },
+    {
+      "epoch": 0.4226657111627469,
+      "grad_norm": 0.1667516529560089,
+      "learning_rate": 4.577341527503313e-06,
+      "loss": 0.9406,
+      "step": 58390
+    },
+    {
+      "epoch": 0.4227380978233331,
+      "grad_norm": 0.15829908847808838,
+      "learning_rate": 4.577269140842726e-06,
+      "loss": 0.9388,
+      "step": 58400
+    },
+    {
+      "epoch": 0.4228104844839193,
+      "grad_norm": 0.16750116646289825,
+      "learning_rate": 4.57719675418214e-06,
+      "loss": 0.9415,
+      "step": 58410
+    },
+    {
+      "epoch": 0.4228828711445055,
+      "grad_norm": 0.21571747958660126,
+      "learning_rate": 4.5771243675215535e-06,
+      "loss": 0.9387,
+      "step": 58420
+    },
+    {
+      "epoch": 0.4229552578050917,
+      "grad_norm": 0.16604144871234894,
+      "learning_rate": 4.577051980860968e-06,
+      "loss": 0.9345,
+      "step": 58430
+    },
+    {
+      "epoch": 0.42302764446567787,
+      "grad_norm": 0.16571690142154694,
+      "learning_rate": 4.576979594200382e-06,
+      "loss": 0.9398,
+      "step": 58440
+    },
+    {
+      "epoch": 0.42310003112626404,
+      "grad_norm": 0.1598815768957138,
+      "learning_rate": 4.576907207539794e-06,
+      "loss": 0.938,
+      "step": 58450
+    },
+    {
+      "epoch": 0.4231724177868502,
+      "grad_norm": 0.16254471242427826,
+      "learning_rate": 4.576834820879209e-06,
+      "loss": 0.9153,
+      "step": 58460
+    },
+    {
+      "epoch": 0.42324480444743645,
+      "grad_norm": 0.17115645110607147,
+      "learning_rate": 4.5767624342186224e-06,
+      "loss": 0.936,
+      "step": 58470
+    },
+    {
+      "epoch": 0.42331719110802263,
+      "grad_norm": 0.1529657244682312,
+      "learning_rate": 4.576690047558036e-06,
+      "loss": 0.9381,
+      "step": 58480
+    },
+    {
+      "epoch": 0.4233895777686088,
+      "grad_norm": 0.15278260409832,
+      "learning_rate": 4.57661766089745e-06,
+      "loss": 0.9446,
+      "step": 58490
+    },
+    {
+      "epoch": 0.423461964429195,
+      "grad_norm": 0.15874941647052765,
+      "learning_rate": 4.576545274236864e-06,
+      "loss": 0.943,
+      "step": 58500
+    },
+    {
+      "epoch": 0.42353435108978116,
+      "grad_norm": 0.1559789478778839,
+      "learning_rate": 4.576472887576278e-06,
+      "loss": 0.9342,
+      "step": 58510
+    },
+    {
+      "epoch": 0.42360673775036733,
+      "grad_norm": 0.1483520269393921,
+      "learning_rate": 4.576400500915691e-06,
+      "loss": 0.927,
+      "step": 58520
+    },
+    {
+      "epoch": 0.42367912441095357,
+      "grad_norm": 0.16274036467075348,
+      "learning_rate": 4.576328114255105e-06,
+      "loss": 0.9228,
+      "step": 58530
+    },
+    {
+      "epoch": 0.42375151107153974,
+      "grad_norm": 0.16001389920711517,
+      "learning_rate": 4.5762557275945194e-06,
+      "loss": 0.9284,
+      "step": 58540
+    },
+    {
+      "epoch": 0.4238238977321259,
+      "grad_norm": 0.16106966137886047,
+      "learning_rate": 4.576183340933933e-06,
+      "loss": 0.9275,
+      "step": 58550
+    },
+    {
+      "epoch": 0.4238962843927121,
+      "grad_norm": 0.17756831645965576,
+      "learning_rate": 4.576110954273347e-06,
+      "loss": 0.9324,
+      "step": 58560
+    },
+    {
+      "epoch": 0.4239686710532983,
+      "grad_norm": 0.18322084844112396,
+      "learning_rate": 4.57603856761276e-06,
+      "loss": 0.9276,
+      "step": 58570
+    },
+    {
+      "epoch": 0.4240410577138845,
+      "grad_norm": 0.17278200387954712,
+      "learning_rate": 4.575966180952175e-06,
+      "loss": 0.9419,
+      "step": 58580
+    },
+    {
+      "epoch": 0.4241134443744707,
+      "grad_norm": 0.16556476056575775,
+      "learning_rate": 4.575893794291588e-06,
+      "loss": 0.9291,
+      "step": 58590
+    },
+    {
+      "epoch": 0.42418583103505686,
+      "grad_norm": 0.15898211300373077,
+      "learning_rate": 4.575821407631002e-06,
+      "loss": 0.9342,
+      "step": 58600
+    },
+    {
+      "epoch": 0.42425821769564304,
+      "grad_norm": 0.16985763609409332,
+      "learning_rate": 4.575749020970416e-06,
+      "loss": 0.9345,
+      "step": 58610
+    },
+    {
+      "epoch": 0.4243306043562292,
+      "grad_norm": 0.16115114092826843,
+      "learning_rate": 4.57567663430983e-06,
+      "loss": 0.9404,
+      "step": 58620
+    },
+    {
+      "epoch": 0.42440299101681545,
+      "grad_norm": 0.16031764447689056,
+      "learning_rate": 4.575604247649244e-06,
+      "loss": 0.9347,
+      "step": 58630
+    },
+    {
+      "epoch": 0.4244753776774016,
+      "grad_norm": 0.15465863049030304,
+      "learning_rate": 4.575531860988657e-06,
+      "loss": 0.9354,
+      "step": 58640
+    },
+    {
+      "epoch": 0.4245477643379878,
+      "grad_norm": 0.17583829164505005,
+      "learning_rate": 4.575459474328071e-06,
+      "loss": 0.9308,
+      "step": 58650
+    },
+    {
+      "epoch": 0.424620150998574,
+      "grad_norm": 0.16195513308048248,
+      "learning_rate": 4.575387087667485e-06,
+      "loss": 0.9312,
+      "step": 58660
+    },
+    {
+      "epoch": 0.42469253765916015,
+      "grad_norm": 0.24764384329319,
+      "learning_rate": 4.575314701006899e-06,
+      "loss": 0.9256,
+      "step": 58670
+    },
+    {
+      "epoch": 0.4247649243197464,
+      "grad_norm": 0.15488171577453613,
+      "learning_rate": 4.575242314346313e-06,
+      "loss": 0.9204,
+      "step": 58680
+    },
+    {
+      "epoch": 0.42483731098033256,
+      "grad_norm": 0.16304126381874084,
+      "learning_rate": 4.575169927685726e-06,
+      "loss": 0.9366,
+      "step": 58690
+    },
+    {
+      "epoch": 0.42490969764091874,
+      "grad_norm": 0.15404012799263,
+      "learning_rate": 4.57509754102514e-06,
+      "loss": 0.9253,
+      "step": 58700
+    },
+    {
+      "epoch": 0.4249820843015049,
+      "grad_norm": 0.17125609517097473,
+      "learning_rate": 4.575025154364554e-06,
+      "loss": 0.9325,
+      "step": 58710
+    },
+    {
+      "epoch": 0.4250544709620911,
+      "grad_norm": 0.17699052393436432,
+      "learning_rate": 4.574952767703968e-06,
+      "loss": 0.9248,
+      "step": 58720
+    },
+    {
+      "epoch": 0.42512685762267727,
+      "grad_norm": 0.1671576052904129,
+      "learning_rate": 4.5748803810433815e-06,
+      "loss": 0.9407,
+      "step": 58730
+    },
+    {
+      "epoch": 0.4251992442832635,
+      "grad_norm": 0.48583319783210754,
+      "learning_rate": 4.574807994382795e-06,
+      "loss": 0.9151,
+      "step": 58740
+    },
+    {
+      "epoch": 0.4252716309438497,
+      "grad_norm": 0.15986119210720062,
+      "learning_rate": 4.57473560772221e-06,
+      "loss": 0.9351,
+      "step": 58750
+    },
+    {
+      "epoch": 0.42534401760443585,
+      "grad_norm": 0.14684616029262543,
+      "learning_rate": 4.574663221061623e-06,
+      "loss": 0.9456,
+      "step": 58760
+    },
+    {
+      "epoch": 0.42541640426502203,
+      "grad_norm": 0.2594200074672699,
+      "learning_rate": 4.574590834401037e-06,
+      "loss": 0.9368,
+      "step": 58770
+    },
+    {
+      "epoch": 0.4254887909256082,
+      "grad_norm": 0.15739043056964874,
+      "learning_rate": 4.5745184477404505e-06,
+      "loss": 0.9336,
+      "step": 58780
+    },
+    {
+      "epoch": 0.42556117758619444,
+      "grad_norm": 0.1712053120136261,
+      "learning_rate": 4.574446061079865e-06,
+      "loss": 0.9341,
+      "step": 58790
+    },
+    {
+      "epoch": 0.4256335642467806,
+      "grad_norm": 0.1579931527376175,
+      "learning_rate": 4.5743736744192785e-06,
+      "loss": 0.9315,
+      "step": 58800
+    },
+    {
+      "epoch": 0.4257059509073668,
+      "grad_norm": 0.16759207844734192,
+      "learning_rate": 4.574301287758692e-06,
+      "loss": 0.9334,
+      "step": 58810
+    },
+    {
+      "epoch": 0.42577833756795297,
+      "grad_norm": 0.15357771515846252,
+      "learning_rate": 4.574228901098106e-06,
+      "loss": 0.9341,
+      "step": 58820
+    },
+    {
+      "epoch": 0.42585072422853915,
+      "grad_norm": 0.16257642209529877,
+      "learning_rate": 4.57415651443752e-06,
+      "loss": 0.9319,
+      "step": 58830
+    },
+    {
+      "epoch": 0.4259231108891254,
+      "grad_norm": 0.1543009728193283,
+      "learning_rate": 4.574084127776934e-06,
+      "loss": 0.9544,
+      "step": 58840
+    },
+    {
+      "epoch": 0.42599549754971155,
+      "grad_norm": 0.18277451395988464,
+      "learning_rate": 4.5740117411163475e-06,
+      "loss": 0.9316,
+      "step": 58850
+    },
+    {
+      "epoch": 0.42606788421029773,
+      "grad_norm": 0.1689336895942688,
+      "learning_rate": 4.573939354455761e-06,
+      "loss": 0.9406,
+      "step": 58860
+    },
+    {
+      "epoch": 0.4261402708708839,
+      "grad_norm": 0.16068103909492493,
+      "learning_rate": 4.5738669677951756e-06,
+      "loss": 0.9412,
+      "step": 58870
+    },
+    {
+      "epoch": 0.4262126575314701,
+      "grad_norm": 0.1947869062423706,
+      "learning_rate": 4.573794581134589e-06,
+      "loss": 0.9303,
+      "step": 58880
+    },
+    {
+      "epoch": 0.42628504419205626,
+      "grad_norm": 0.16800038516521454,
+      "learning_rate": 4.573722194474003e-06,
+      "loss": 0.9349,
+      "step": 58890
+    },
+    {
+      "epoch": 0.4263574308526425,
+      "grad_norm": 0.16128326952457428,
+      "learning_rate": 4.573649807813416e-06,
+      "loss": 0.9433,
+      "step": 58900
+    },
+    {
+      "epoch": 0.42642981751322867,
+      "grad_norm": 0.16078263521194458,
+      "learning_rate": 4.573577421152831e-06,
+      "loss": 0.9305,
+      "step": 58910
+    },
+    {
+      "epoch": 0.42650220417381485,
+      "grad_norm": 0.20979563891887665,
+      "learning_rate": 4.5735050344922445e-06,
+      "loss": 0.9481,
+      "step": 58920
+    },
+    {
+      "epoch": 0.426574590834401,
+      "grad_norm": 0.16073863208293915,
+      "learning_rate": 4.573432647831658e-06,
+      "loss": 0.9395,
+      "step": 58930
+    },
+    {
+      "epoch": 0.4266469774949872,
+      "grad_norm": 0.17131203413009644,
+      "learning_rate": 4.573360261171072e-06,
+      "loss": 0.9441,
+      "step": 58940
+    },
+    {
+      "epoch": 0.42671936415557343,
+      "grad_norm": 0.150382399559021,
+      "learning_rate": 4.573287874510486e-06,
+      "loss": 0.926,
+      "step": 58950
+    },
+    {
+      "epoch": 0.4267917508161596,
+      "grad_norm": 0.15955784916877747,
+      "learning_rate": 4.5732154878499e-06,
+      "loss": 0.9287,
+      "step": 58960
+    },
+    {
+      "epoch": 0.4268641374767458,
+      "grad_norm": 0.16103602945804596,
+      "learning_rate": 4.573143101189313e-06,
+      "loss": 0.9269,
+      "step": 58970
+    },
+    {
+      "epoch": 0.42693652413733196,
+      "grad_norm": 0.17586109042167664,
+      "learning_rate": 4.573070714528727e-06,
+      "loss": 0.9279,
+      "step": 58980
+    },
+    {
+      "epoch": 0.42700891079791814,
+      "grad_norm": 0.1569269299507141,
+      "learning_rate": 4.572998327868141e-06,
+      "loss": 0.932,
+      "step": 58990
+    },
+    {
+      "epoch": 0.42708129745850437,
+      "grad_norm": 0.1663513034582138,
+      "learning_rate": 4.572925941207554e-06,
+      "loss": 0.9276,
+      "step": 59000
+    },
+    {
+      "epoch": 0.42715368411909055,
+      "grad_norm": 0.16358311474323273,
+      "learning_rate": 4.572853554546968e-06,
+      "loss": 0.936,
+      "step": 59010
+    },
+    {
+      "epoch": 0.4272260707796767,
+      "grad_norm": 0.15506871044635773,
+      "learning_rate": 4.572781167886382e-06,
+      "loss": 0.9271,
+      "step": 59020
+    },
+    {
+      "epoch": 0.4272984574402629,
+      "grad_norm": 0.156698539853096,
+      "learning_rate": 4.572708781225796e-06,
+      "loss": 0.9327,
+      "step": 59030
+    },
+    {
+      "epoch": 0.4273708441008491,
+      "grad_norm": 0.1768382340669632,
+      "learning_rate": 4.5726363945652096e-06,
+      "loss": 0.9408,
+      "step": 59040
+    },
+    {
+      "epoch": 0.42744323076143526,
+      "grad_norm": 0.15888147056102753,
+      "learning_rate": 4.572564007904623e-06,
+      "loss": 0.9294,
+      "step": 59050
+    },
+    {
+      "epoch": 0.4275156174220215,
+      "grad_norm": 0.1897684782743454,
+      "learning_rate": 4.572491621244038e-06,
+      "loss": 0.9478,
+      "step": 59060
+    },
+    {
+      "epoch": 0.42758800408260766,
+      "grad_norm": 0.18450258672237396,
+      "learning_rate": 4.572419234583451e-06,
+      "loss": 0.9202,
+      "step": 59070
+    },
+    {
+      "epoch": 0.42766039074319384,
+      "grad_norm": 0.16853807866573334,
+      "learning_rate": 4.572346847922865e-06,
+      "loss": 0.9362,
+      "step": 59080
+    },
+    {
+      "epoch": 0.42773277740378,
+      "grad_norm": 0.16247627139091492,
+      "learning_rate": 4.5722744612622785e-06,
+      "loss": 0.9351,
+      "step": 59090
+    },
+    {
+      "epoch": 0.4278051640643662,
+      "grad_norm": 0.17450258135795593,
+      "learning_rate": 4.572202074601693e-06,
+      "loss": 0.9304,
+      "step": 59100
+    },
+    {
+      "epoch": 0.4278775507249524,
+      "grad_norm": 0.22442488372325897,
+      "learning_rate": 4.572129687941107e-06,
+      "loss": 0.9176,
+      "step": 59110
+    },
+    {
+      "epoch": 0.4279499373855386,
+      "grad_norm": 0.1596069484949112,
+      "learning_rate": 4.57205730128052e-06,
+      "loss": 0.9467,
+      "step": 59120
+    },
+    {
+      "epoch": 0.4280223240461248,
+      "grad_norm": 0.16665290296077728,
+      "learning_rate": 4.571984914619934e-06,
+      "loss": 0.9333,
+      "step": 59130
+    },
+    {
+      "epoch": 0.42809471070671096,
+      "grad_norm": 0.15915903449058533,
+      "learning_rate": 4.571912527959348e-06,
+      "loss": 0.9386,
+      "step": 59140
+    },
+    {
+      "epoch": 0.42816709736729713,
+      "grad_norm": 0.16277122497558594,
+      "learning_rate": 4.571840141298762e-06,
+      "loss": 0.9298,
+      "step": 59150
+    },
+    {
+      "epoch": 0.42823948402788337,
+      "grad_norm": 0.14925530552864075,
+      "learning_rate": 4.5717677546381755e-06,
+      "loss": 0.9454,
+      "step": 59160
+    },
+    {
+      "epoch": 0.42831187068846954,
+      "grad_norm": 0.1798754781484604,
+      "learning_rate": 4.571695367977589e-06,
+      "loss": 0.9397,
+      "step": 59170
+    },
+    {
+      "epoch": 0.4283842573490557,
+      "grad_norm": 0.16380873322486877,
+      "learning_rate": 4.571622981317004e-06,
+      "loss": 0.9515,
+      "step": 59180
+    },
+    {
+      "epoch": 0.4284566440096419,
+      "grad_norm": 0.16432473063468933,
+      "learning_rate": 4.571550594656417e-06,
+      "loss": 0.9351,
+      "step": 59190
+    },
+    {
+      "epoch": 0.4285290306702281,
+      "grad_norm": 0.14648674428462982,
+      "learning_rate": 4.571478207995831e-06,
+      "loss": 0.9283,
+      "step": 59200
+    },
+    {
+      "epoch": 0.4286014173308143,
+      "grad_norm": 0.1510661244392395,
+      "learning_rate": 4.5714058213352444e-06,
+      "loss": 0.9295,
+      "step": 59210
+    },
+    {
+      "epoch": 0.4286738039914005,
+      "grad_norm": 0.18739067018032074,
+      "learning_rate": 4.571333434674659e-06,
+      "loss": 0.937,
+      "step": 59220
+    },
+    {
+      "epoch": 0.42874619065198666,
+      "grad_norm": 0.16375543177127838,
+      "learning_rate": 4.5712610480140725e-06,
+      "loss": 0.9354,
+      "step": 59230
+    },
+    {
+      "epoch": 0.42881857731257284,
+      "grad_norm": 0.16390521824359894,
+      "learning_rate": 4.571188661353486e-06,
+      "loss": 0.9353,
+      "step": 59240
+    },
+    {
+      "epoch": 0.428890963973159,
+      "grad_norm": 0.14936009049415588,
+      "learning_rate": 4.5711162746929e-06,
+      "loss": 0.9253,
+      "step": 59250
+    },
+    {
+      "epoch": 0.4289633506337452,
+      "grad_norm": 0.16098490357398987,
+      "learning_rate": 4.571043888032314e-06,
+      "loss": 0.9187,
+      "step": 59260
+    },
+    {
+      "epoch": 0.4290357372943314,
+      "grad_norm": 0.16449196636676788,
+      "learning_rate": 4.570971501371728e-06,
+      "loss": 0.93,
+      "step": 59270
+    },
+    {
+      "epoch": 0.4291081239549176,
+      "grad_norm": 0.44586077332496643,
+      "learning_rate": 4.5708991147111414e-06,
+      "loss": 0.9421,
+      "step": 59280
+    },
+    {
+      "epoch": 0.4291805106155038,
+      "grad_norm": 0.15456175804138184,
+      "learning_rate": 4.570826728050555e-06,
+      "loss": 0.9223,
+      "step": 59290
+    },
+    {
+      "epoch": 0.42925289727608995,
+      "grad_norm": 0.17588649690151215,
+      "learning_rate": 4.570754341389969e-06,
+      "loss": 0.9438,
+      "step": 59300
+    },
+    {
+      "epoch": 0.4293252839366761,
+      "grad_norm": 0.15864711999893188,
+      "learning_rate": 4.570681954729383e-06,
+      "loss": 0.9313,
+      "step": 59310
+    },
+    {
+      "epoch": 0.42939767059726236,
+      "grad_norm": 0.16494864225387573,
+      "learning_rate": 4.570609568068797e-06,
+      "loss": 0.9457,
+      "step": 59320
+    },
+    {
+      "epoch": 0.42947005725784854,
+      "grad_norm": 0.15359218418598175,
+      "learning_rate": 4.57053718140821e-06,
+      "loss": 0.9295,
+      "step": 59330
+    },
+    {
+      "epoch": 0.4295424439184347,
+      "grad_norm": 0.16325341165065765,
+      "learning_rate": 4.570464794747624e-06,
+      "loss": 0.9434,
+      "step": 59340
+    },
+    {
+      "epoch": 0.4296148305790209,
+      "grad_norm": 0.15335196256637573,
+      "learning_rate": 4.5703924080870385e-06,
+      "loss": 0.9255,
+      "step": 59350
+    },
+    {
+      "epoch": 0.42968721723960707,
+      "grad_norm": 0.15363959968090057,
+      "learning_rate": 4.570320021426452e-06,
+      "loss": 0.9349,
+      "step": 59360
+    },
+    {
+      "epoch": 0.4297596039001933,
+      "grad_norm": 0.16724181175231934,
+      "learning_rate": 4.570247634765866e-06,
+      "loss": 0.9281,
+      "step": 59370
+    },
+    {
+      "epoch": 0.4298319905607795,
+      "grad_norm": 0.1659107208251953,
+      "learning_rate": 4.570175248105279e-06,
+      "loss": 0.9412,
+      "step": 59380
+    },
+    {
+      "epoch": 0.42990437722136565,
+      "grad_norm": 0.17168617248535156,
+      "learning_rate": 4.570102861444694e-06,
+      "loss": 0.9401,
+      "step": 59390
+    },
+    {
+      "epoch": 0.42997676388195183,
+      "grad_norm": 0.15352968871593475,
+      "learning_rate": 4.570030474784107e-06,
+      "loss": 0.9271,
+      "step": 59400
+    },
+    {
+      "epoch": 0.430049150542538,
+      "grad_norm": 0.17253489792346954,
+      "learning_rate": 4.569958088123521e-06,
+      "loss": 0.9324,
+      "step": 59410
+    },
+    {
+      "epoch": 0.4301215372031242,
+      "grad_norm": 0.15288107097148895,
+      "learning_rate": 4.569885701462935e-06,
+      "loss": 0.9274,
+      "step": 59420
+    },
+    {
+      "epoch": 0.4301939238637104,
+      "grad_norm": 0.17181651294231415,
+      "learning_rate": 4.569813314802349e-06,
+      "loss": 0.9399,
+      "step": 59430
+    },
+    {
+      "epoch": 0.4302663105242966,
+      "grad_norm": 0.16761989891529083,
+      "learning_rate": 4.569740928141763e-06,
+      "loss": 0.938,
+      "step": 59440
+    },
+    {
+      "epoch": 0.43033869718488277,
+      "grad_norm": 0.16204999387264252,
+      "learning_rate": 4.569668541481176e-06,
+      "loss": 0.9332,
+      "step": 59450
+    },
+    {
+      "epoch": 0.43041108384546894,
+      "grad_norm": 0.1525745391845703,
+      "learning_rate": 4.56959615482059e-06,
+      "loss": 0.93,
+      "step": 59460
+    },
+    {
+      "epoch": 0.4304834705060551,
+      "grad_norm": 0.17561571300029755,
+      "learning_rate": 4.569523768160004e-06,
+      "loss": 0.9375,
+      "step": 59470
+    },
+    {
+      "epoch": 0.43055585716664135,
+      "grad_norm": 0.1534716933965683,
+      "learning_rate": 4.569451381499418e-06,
+      "loss": 0.9513,
+      "step": 59480
+    },
+    {
+      "epoch": 0.43062824382722753,
+      "grad_norm": 0.16287165880203247,
+      "learning_rate": 4.569378994838832e-06,
+      "loss": 0.9542,
+      "step": 59490
+    },
+    {
+      "epoch": 0.4307006304878137,
+      "grad_norm": 0.1597660928964615,
+      "learning_rate": 4.569306608178245e-06,
+      "loss": 0.926,
+      "step": 59500
+    },
+    {
+      "epoch": 0.4307730171483999,
+      "grad_norm": 0.18294881284236908,
+      "learning_rate": 4.569234221517659e-06,
+      "loss": 0.9382,
+      "step": 59510
+    },
+    {
+      "epoch": 0.43084540380898606,
+      "grad_norm": 0.1525103747844696,
+      "learning_rate": 4.5691618348570725e-06,
+      "loss": 0.937,
+      "step": 59520
+    },
+    {
+      "epoch": 0.4309177904695723,
+      "grad_norm": 0.15872421860694885,
+      "learning_rate": 4.569089448196486e-06,
+      "loss": 0.9225,
+      "step": 59530
+    },
+    {
+      "epoch": 0.43099017713015847,
+      "grad_norm": 0.15989533066749573,
+      "learning_rate": 4.5690170615359005e-06,
+      "loss": 0.926,
+      "step": 59540
+    },
+    {
+      "epoch": 0.43106256379074465,
+      "grad_norm": 0.16288037598133087,
+      "learning_rate": 4.568944674875314e-06,
+      "loss": 0.9269,
+      "step": 59550
+    },
+    {
+      "epoch": 0.4311349504513308,
+      "grad_norm": 0.16788716614246368,
+      "learning_rate": 4.568872288214728e-06,
+      "loss": 0.9176,
+      "step": 59560
+    },
+    {
+      "epoch": 0.431207337111917,
+      "grad_norm": 0.15858720242977142,
+      "learning_rate": 4.568799901554141e-06,
+      "loss": 0.9246,
+      "step": 59570
+    },
+    {
+      "epoch": 0.4312797237725032,
+      "grad_norm": 0.17023035883903503,
+      "learning_rate": 4.568727514893556e-06,
+      "loss": 0.9416,
+      "step": 59580
+    },
+    {
+      "epoch": 0.4313521104330894,
+      "grad_norm": 0.16211800277233124,
+      "learning_rate": 4.5686551282329695e-06,
+      "loss": 0.9295,
+      "step": 59590
+    },
+    {
+      "epoch": 0.4314244970936756,
+      "grad_norm": 0.16199994087219238,
+      "learning_rate": 4.568582741572383e-06,
+      "loss": 0.9308,
+      "step": 59600
+    },
+    {
+      "epoch": 0.43149688375426176,
+      "grad_norm": 0.1606810986995697,
+      "learning_rate": 4.568510354911797e-06,
+      "loss": 0.9309,
+      "step": 59610
+    },
+    {
+      "epoch": 0.43156927041484794,
+      "grad_norm": 0.20290134847164154,
+      "learning_rate": 4.568437968251211e-06,
+      "loss": 0.9311,
+      "step": 59620
+    },
+    {
+      "epoch": 0.4316416570754341,
+      "grad_norm": 0.18084190785884857,
+      "learning_rate": 4.568365581590625e-06,
+      "loss": 0.9283,
+      "step": 59630
+    },
+    {
+      "epoch": 0.43171404373602035,
+      "grad_norm": 0.15636208653450012,
+      "learning_rate": 4.568293194930038e-06,
+      "loss": 0.939,
+      "step": 59640
+    },
+    {
+      "epoch": 0.4317864303966065,
+      "grad_norm": 0.16248819231987,
+      "learning_rate": 4.568220808269452e-06,
+      "loss": 0.9457,
+      "step": 59650
+    },
+    {
+      "epoch": 0.4318588170571927,
+      "grad_norm": 0.16725608706474304,
+      "learning_rate": 4.5681484216088665e-06,
+      "loss": 0.9239,
+      "step": 59660
+    },
+    {
+      "epoch": 0.4319312037177789,
+      "grad_norm": 0.15592262148857117,
+      "learning_rate": 4.56807603494828e-06,
+      "loss": 0.9274,
+      "step": 59670
+    },
+    {
+      "epoch": 0.43200359037836505,
+      "grad_norm": 0.16762572526931763,
+      "learning_rate": 4.568003648287694e-06,
+      "loss": 0.9387,
+      "step": 59680
+    },
+    {
+      "epoch": 0.4320759770389513,
+      "grad_norm": 0.16156117618083954,
+      "learning_rate": 4.567931261627107e-06,
+      "loss": 0.9486,
+      "step": 59690
+    },
+    {
+      "epoch": 0.43214836369953746,
+      "grad_norm": 0.1718205362558365,
+      "learning_rate": 4.567858874966522e-06,
+      "loss": 0.928,
+      "step": 59700
+    },
+    {
+      "epoch": 0.43222075036012364,
+      "grad_norm": 0.15854863822460175,
+      "learning_rate": 4.567786488305935e-06,
+      "loss": 0.9373,
+      "step": 59710
+    },
+    {
+      "epoch": 0.4322931370207098,
+      "grad_norm": 0.1781233698129654,
+      "learning_rate": 4.567714101645349e-06,
+      "loss": 0.9324,
+      "step": 59720
+    },
+    {
+      "epoch": 0.432365523681296,
+      "grad_norm": 0.15047183632850647,
+      "learning_rate": 4.567641714984763e-06,
+      "loss": 0.9246,
+      "step": 59730
+    },
+    {
+      "epoch": 0.4324379103418822,
+      "grad_norm": 0.1613796353340149,
+      "learning_rate": 4.567569328324177e-06,
+      "loss": 0.9407,
+      "step": 59740
+    },
+    {
+      "epoch": 0.4325102970024684,
+      "grad_norm": 0.15587370097637177,
+      "learning_rate": 4.567496941663591e-06,
+      "loss": 0.9229,
+      "step": 59750
+    },
+    {
+      "epoch": 0.4325826836630546,
+      "grad_norm": 0.15507788956165314,
+      "learning_rate": 4.567424555003004e-06,
+      "loss": 0.9281,
+      "step": 59760
+    },
+    {
+      "epoch": 0.43265507032364076,
+      "grad_norm": 0.15560147166252136,
+      "learning_rate": 4.567352168342418e-06,
+      "loss": 0.9467,
+      "step": 59770
+    },
+    {
+      "epoch": 0.43272745698422693,
+      "grad_norm": 0.1680835783481598,
+      "learning_rate": 4.567279781681832e-06,
+      "loss": 0.9317,
+      "step": 59780
+    },
+    {
+      "epoch": 0.4327998436448131,
+      "grad_norm": 0.18736176192760468,
+      "learning_rate": 4.567207395021246e-06,
+      "loss": 0.9314,
+      "step": 59790
+    },
+    {
+      "epoch": 0.43287223030539934,
+      "grad_norm": 0.1607762575149536,
+      "learning_rate": 4.56713500836066e-06,
+      "loss": 0.9391,
+      "step": 59800
+    },
+    {
+      "epoch": 0.4329446169659855,
+      "grad_norm": 0.16497263312339783,
+      "learning_rate": 4.567062621700073e-06,
+      "loss": 0.9347,
+      "step": 59810
+    },
+    {
+      "epoch": 0.4330170036265717,
+      "grad_norm": 0.14769291877746582,
+      "learning_rate": 4.566990235039488e-06,
+      "loss": 0.9089,
+      "step": 59820
+    },
+    {
+      "epoch": 0.43308939028715787,
+      "grad_norm": 0.1630723625421524,
+      "learning_rate": 4.566917848378901e-06,
+      "loss": 0.9294,
+      "step": 59830
+    },
+    {
+      "epoch": 0.43316177694774405,
+      "grad_norm": 0.1905062049627304,
+      "learning_rate": 4.566845461718315e-06,
+      "loss": 0.9248,
+      "step": 59840
+    },
+    {
+      "epoch": 0.4332341636083303,
+      "grad_norm": 0.16832837462425232,
+      "learning_rate": 4.566773075057729e-06,
+      "loss": 0.9424,
+      "step": 59850
+    },
+    {
+      "epoch": 0.43330655026891646,
+      "grad_norm": 0.16252058744430542,
+      "learning_rate": 4.566700688397143e-06,
+      "loss": 0.9272,
+      "step": 59860
+    },
+    {
+      "epoch": 0.43337893692950263,
+      "grad_norm": 0.18382105231285095,
+      "learning_rate": 4.566628301736557e-06,
+      "loss": 0.9257,
+      "step": 59870
+    },
+    {
+      "epoch": 0.4334513235900888,
+      "grad_norm": 0.16314184665679932,
+      "learning_rate": 4.56655591507597e-06,
+      "loss": 0.9395,
+      "step": 59880
+    },
+    {
+      "epoch": 0.433523710250675,
+      "grad_norm": 0.17523783445358276,
+      "learning_rate": 4.566483528415384e-06,
+      "loss": 0.926,
+      "step": 59890
+    },
+    {
+      "epoch": 0.4335960969112612,
+      "grad_norm": 0.16567225754261017,
+      "learning_rate": 4.566411141754798e-06,
+      "loss": 0.9382,
+      "step": 59900
+    },
+    {
+      "epoch": 0.4336684835718474,
+      "grad_norm": 0.16518938541412354,
+      "learning_rate": 4.566338755094212e-06,
+      "loss": 0.9248,
+      "step": 59910
+    },
+    {
+      "epoch": 0.4337408702324336,
+      "grad_norm": 0.147177055478096,
+      "learning_rate": 4.566266368433626e-06,
+      "loss": 0.9317,
+      "step": 59920
+    },
+    {
+      "epoch": 0.43381325689301975,
+      "grad_norm": 0.1555219441652298,
+      "learning_rate": 4.566193981773039e-06,
+      "loss": 0.9323,
+      "step": 59930
+    },
+    {
+      "epoch": 0.4338856435536059,
+      "grad_norm": 0.1518564671278,
+      "learning_rate": 4.566121595112453e-06,
+      "loss": 0.9413,
+      "step": 59940
+    },
+    {
+      "epoch": 0.4339580302141921,
+      "grad_norm": 0.1824176162481308,
+      "learning_rate": 4.566049208451867e-06,
+      "loss": 0.9254,
+      "step": 59950
+    },
+    {
+      "epoch": 0.43403041687477834,
+      "grad_norm": 0.16142070293426514,
+      "learning_rate": 4.565976821791281e-06,
+      "loss": 0.9313,
+      "step": 59960
+    },
+    {
+      "epoch": 0.4341028035353645,
+      "grad_norm": 0.1917445957660675,
+      "learning_rate": 4.5659044351306945e-06,
+      "loss": 0.9205,
+      "step": 59970
+    },
+    {
+      "epoch": 0.4341751901959507,
+      "grad_norm": 0.1893223375082016,
+      "learning_rate": 4.565832048470108e-06,
+      "loss": 0.932,
+      "step": 59980
+    },
+    {
+      "epoch": 0.43424757685653687,
+      "grad_norm": 0.16585801541805267,
+      "learning_rate": 4.565759661809523e-06,
+      "loss": 0.9321,
+      "step": 59990
+    },
+    {
+      "epoch": 0.43431996351712304,
+      "grad_norm": 0.1668858528137207,
+      "learning_rate": 4.565687275148936e-06,
+      "loss": 0.9322,
+      "step": 60000
+    },
+    {
+      "epoch": 0.4343923501777093,
+      "grad_norm": 0.16714520752429962,
+      "learning_rate": 4.56561488848835e-06,
+      "loss": 0.9363,
+      "step": 60010
+    },
+    {
+      "epoch": 0.43446473683829545,
+      "grad_norm": 0.5191177129745483,
+      "learning_rate": 4.5655425018277634e-06,
+      "loss": 0.9313,
+      "step": 60020
+    },
+    {
+      "epoch": 0.43453712349888163,
+      "grad_norm": 0.16211098432540894,
+      "learning_rate": 4.565470115167178e-06,
+      "loss": 0.9287,
+      "step": 60030
+    },
+    {
+      "epoch": 0.4346095101594678,
+      "grad_norm": 0.15494702756404877,
+      "learning_rate": 4.565397728506591e-06,
+      "loss": 0.9223,
+      "step": 60040
+    },
+    {
+      "epoch": 0.434681896820054,
+      "grad_norm": 0.1619596928358078,
+      "learning_rate": 4.565325341846005e-06,
+      "loss": 0.9234,
+      "step": 60050
+    },
+    {
+      "epoch": 0.4347542834806402,
+      "grad_norm": 0.1678442656993866,
+      "learning_rate": 4.565252955185419e-06,
+      "loss": 0.931,
+      "step": 60060
+    },
+    {
+      "epoch": 0.4348266701412264,
+      "grad_norm": 0.1571994423866272,
+      "learning_rate": 4.565180568524832e-06,
+      "loss": 0.9274,
+      "step": 60070
+    },
+    {
+      "epoch": 0.43489905680181257,
+      "grad_norm": 0.16465076804161072,
+      "learning_rate": 4.565108181864246e-06,
+      "loss": 0.9344,
+      "step": 60080
+    },
+    {
+      "epoch": 0.43497144346239874,
+      "grad_norm": 0.1529812514781952,
+      "learning_rate": 4.56503579520366e-06,
+      "loss": 0.9154,
+      "step": 60090
+    },
+    {
+      "epoch": 0.4350438301229849,
+      "grad_norm": 0.1658780574798584,
+      "learning_rate": 4.564963408543074e-06,
+      "loss": 0.9321,
+      "step": 60100
+    },
+    {
+      "epoch": 0.4351162167835711,
+      "grad_norm": 0.15867140889167786,
+      "learning_rate": 4.564891021882488e-06,
+      "loss": 0.9317,
+      "step": 60110
+    },
+    {
+      "epoch": 0.43518860344415733,
+      "grad_norm": 0.1777503490447998,
+      "learning_rate": 4.564818635221901e-06,
+      "loss": 0.9306,
+      "step": 60120
+    },
+    {
+      "epoch": 0.4352609901047435,
+      "grad_norm": 0.15686187148094177,
+      "learning_rate": 4.564746248561315e-06,
+      "loss": 0.9357,
+      "step": 60130
+    },
+    {
+      "epoch": 0.4353333767653297,
+      "grad_norm": 0.15787990391254425,
+      "learning_rate": 4.564673861900729e-06,
+      "loss": 0.9264,
+      "step": 60140
+    },
+    {
+      "epoch": 0.43540576342591586,
+      "grad_norm": 0.16923680901527405,
+      "learning_rate": 4.564601475240143e-06,
+      "loss": 0.9343,
+      "step": 60150
+    },
+    {
+      "epoch": 0.43547815008650204,
+      "grad_norm": 0.1720929890871048,
+      "learning_rate": 4.564529088579557e-06,
+      "loss": 0.9384,
+      "step": 60160
+    },
+    {
+      "epoch": 0.43555053674708827,
+      "grad_norm": 0.2149965465068817,
+      "learning_rate": 4.56445670191897e-06,
+      "loss": 0.9439,
+      "step": 60170
+    },
+    {
+      "epoch": 0.43562292340767444,
+      "grad_norm": 0.15889641642570496,
+      "learning_rate": 4.564384315258385e-06,
+      "loss": 0.926,
+      "step": 60180
+    },
+    {
+      "epoch": 0.4356953100682606,
+      "grad_norm": 0.203876331448555,
+      "learning_rate": 4.564311928597798e-06,
+      "loss": 0.9298,
+      "step": 60190
+    },
+    {
+      "epoch": 0.4357676967288468,
+      "grad_norm": 0.15438738465309143,
+      "learning_rate": 4.564239541937212e-06,
+      "loss": 0.9273,
+      "step": 60200
+    },
+    {
+      "epoch": 0.435840083389433,
+      "grad_norm": 0.1764325052499771,
+      "learning_rate": 4.5641671552766255e-06,
+      "loss": 0.9513,
+      "step": 60210
+    },
+    {
+      "epoch": 0.4359124700500192,
+      "grad_norm": 0.1527344286441803,
+      "learning_rate": 4.56409476861604e-06,
+      "loss": 0.9259,
+      "step": 60220
+    },
+    {
+      "epoch": 0.4359848567106054,
+      "grad_norm": 0.15969283878803253,
+      "learning_rate": 4.564022381955454e-06,
+      "loss": 0.9343,
+      "step": 60230
+    },
+    {
+      "epoch": 0.43605724337119156,
+      "grad_norm": 0.1652412861585617,
+      "learning_rate": 4.563949995294867e-06,
+      "loss": 0.9318,
+      "step": 60240
+    },
+    {
+      "epoch": 0.43612963003177774,
+      "grad_norm": 0.1574745625257492,
+      "learning_rate": 4.563877608634281e-06,
+      "loss": 0.9204,
+      "step": 60250
+    },
+    {
+      "epoch": 0.4362020166923639,
+      "grad_norm": 0.16842874884605408,
+      "learning_rate": 4.563805221973695e-06,
+      "loss": 0.928,
+      "step": 60260
+    },
+    {
+      "epoch": 0.4362744033529501,
+      "grad_norm": 0.17006808519363403,
+      "learning_rate": 4.563732835313109e-06,
+      "loss": 0.9243,
+      "step": 60270
+    },
+    {
+      "epoch": 0.4363467900135363,
+      "grad_norm": 0.16430647671222687,
+      "learning_rate": 4.5636604486525225e-06,
+      "loss": 0.9332,
+      "step": 60280
+    },
+    {
+      "epoch": 0.4364191766741225,
+      "grad_norm": 0.15201468765735626,
+      "learning_rate": 4.563588061991936e-06,
+      "loss": 0.9301,
+      "step": 60290
+    },
+    {
+      "epoch": 0.4364915633347087,
+      "grad_norm": 0.15162621438503265,
+      "learning_rate": 4.563515675331351e-06,
+      "loss": 0.9387,
+      "step": 60300
+    },
+    {
+      "epoch": 0.43656394999529485,
+      "grad_norm": 0.16102218627929688,
+      "learning_rate": 4.563443288670764e-06,
+      "loss": 0.9418,
+      "step": 60310
+    },
+    {
+      "epoch": 0.43663633665588103,
+      "grad_norm": 0.1847027838230133,
+      "learning_rate": 4.563370902010178e-06,
+      "loss": 0.9279,
+      "step": 60320
+    },
+    {
+      "epoch": 0.43670872331646726,
+      "grad_norm": 0.18758545815944672,
+      "learning_rate": 4.5632985153495915e-06,
+      "loss": 0.9381,
+      "step": 60330
+    },
+    {
+      "epoch": 0.43678110997705344,
+      "grad_norm": 0.16631188988685608,
+      "learning_rate": 4.563226128689006e-06,
+      "loss": 0.933,
+      "step": 60340
+    },
+    {
+      "epoch": 0.4368534966376396,
+      "grad_norm": 0.1626015156507492,
+      "learning_rate": 4.5631537420284196e-06,
+      "loss": 0.9305,
+      "step": 60350
+    },
+    {
+      "epoch": 0.4369258832982258,
+      "grad_norm": 0.16860458254814148,
+      "learning_rate": 4.563081355367833e-06,
+      "loss": 0.9341,
+      "step": 60360
+    },
+    {
+      "epoch": 0.43699826995881197,
+      "grad_norm": 0.15493610501289368,
+      "learning_rate": 4.563008968707247e-06,
+      "loss": 0.9321,
+      "step": 60370
+    },
+    {
+      "epoch": 0.4370706566193982,
+      "grad_norm": 0.16478878259658813,
+      "learning_rate": 4.562936582046661e-06,
+      "loss": 0.9218,
+      "step": 60380
+    },
+    {
+      "epoch": 0.4371430432799844,
+      "grad_norm": 0.16642285883426666,
+      "learning_rate": 4.562864195386075e-06,
+      "loss": 0.9375,
+      "step": 60390
+    },
+    {
+      "epoch": 0.43721542994057055,
+      "grad_norm": 0.1687694787979126,
+      "learning_rate": 4.5627918087254885e-06,
+      "loss": 0.922,
+      "step": 60400
+    },
+    {
+      "epoch": 0.43728781660115673,
+      "grad_norm": 0.15206386148929596,
+      "learning_rate": 4.562719422064902e-06,
+      "loss": 0.9289,
+      "step": 60410
+    },
+    {
+      "epoch": 0.4373602032617429,
+      "grad_norm": 0.1521196812391281,
+      "learning_rate": 4.5626470354043166e-06,
+      "loss": 0.9391,
+      "step": 60420
+    },
+    {
+      "epoch": 0.43743258992232914,
+      "grad_norm": 0.1528043895959854,
+      "learning_rate": 4.56257464874373e-06,
+      "loss": 0.9537,
+      "step": 60430
+    },
+    {
+      "epoch": 0.4375049765829153,
+      "grad_norm": 0.16489632427692413,
+      "learning_rate": 4.562502262083144e-06,
+      "loss": 0.9462,
+      "step": 60440
+    },
+    {
+      "epoch": 0.4375773632435015,
+      "grad_norm": 0.1611616611480713,
+      "learning_rate": 4.562429875422557e-06,
+      "loss": 0.94,
+      "step": 60450
+    },
+    {
+      "epoch": 0.43764974990408767,
+      "grad_norm": 0.1894521713256836,
+      "learning_rate": 4.562357488761972e-06,
+      "loss": 0.937,
+      "step": 60460
+    },
+    {
+      "epoch": 0.43772213656467385,
+      "grad_norm": 0.1543973833322525,
+      "learning_rate": 4.5622851021013855e-06,
+      "loss": 0.9228,
+      "step": 60470
+    },
+    {
+      "epoch": 0.43779452322526,
+      "grad_norm": 0.15751250088214874,
+      "learning_rate": 4.562212715440799e-06,
+      "loss": 0.9269,
+      "step": 60480
+    },
+    {
+      "epoch": 0.43786690988584626,
+      "grad_norm": 0.15188553929328918,
+      "learning_rate": 4.562140328780213e-06,
+      "loss": 0.9333,
+      "step": 60490
+    },
+    {
+      "epoch": 0.43793929654643243,
+      "grad_norm": 0.1614580750465393,
+      "learning_rate": 4.562067942119627e-06,
+      "loss": 0.9432,
+      "step": 60500
+    },
+    {
+      "epoch": 0.4380116832070186,
+      "grad_norm": 0.18022243678569794,
+      "learning_rate": 4.561995555459041e-06,
+      "loss": 0.9232,
+      "step": 60510
+    },
+    {
+      "epoch": 0.4380840698676048,
+      "grad_norm": 0.16363121569156647,
+      "learning_rate": 4.561923168798454e-06,
+      "loss": 0.9374,
+      "step": 60520
+    },
+    {
+      "epoch": 0.43815645652819096,
+      "grad_norm": 0.16248828172683716,
+      "learning_rate": 4.561850782137868e-06,
+      "loss": 0.9284,
+      "step": 60530
+    },
+    {
+      "epoch": 0.4382288431887772,
+      "grad_norm": 0.20947439968585968,
+      "learning_rate": 4.5617783954772825e-06,
+      "loss": 0.9374,
+      "step": 60540
+    },
+    {
+      "epoch": 0.43830122984936337,
+      "grad_norm": 0.1605253666639328,
+      "learning_rate": 4.561706008816696e-06,
+      "loss": 0.9469,
+      "step": 60550
+    },
+    {
+      "epoch": 0.43837361650994955,
+      "grad_norm": 0.1608964055776596,
+      "learning_rate": 4.56163362215611e-06,
+      "loss": 0.9317,
+      "step": 60560
+    },
+    {
+      "epoch": 0.4384460031705357,
+      "grad_norm": 0.18973439931869507,
+      "learning_rate": 4.561561235495523e-06,
+      "loss": 0.9383,
+      "step": 60570
+    },
+    {
+      "epoch": 0.4385183898311219,
+      "grad_norm": 0.16821689903736115,
+      "learning_rate": 4.561488848834937e-06,
+      "loss": 0.9244,
+      "step": 60580
+    },
+    {
+      "epoch": 0.43859077649170813,
+      "grad_norm": 0.16728858649730682,
+      "learning_rate": 4.561416462174351e-06,
+      "loss": 0.9172,
+      "step": 60590
+    },
+    {
+      "epoch": 0.4386631631522943,
+      "grad_norm": 0.15404008328914642,
+      "learning_rate": 4.561344075513764e-06,
+      "loss": 0.9352,
+      "step": 60600
+    },
+    {
+      "epoch": 0.4387355498128805,
+      "grad_norm": 0.20690277218818665,
+      "learning_rate": 4.561271688853179e-06,
+      "loss": 0.9347,
+      "step": 60610
+    },
+    {
+      "epoch": 0.43880793647346666,
+      "grad_norm": 0.16291332244873047,
+      "learning_rate": 4.561199302192592e-06,
+      "loss": 0.9295,
+      "step": 60620
+    },
+    {
+      "epoch": 0.43888032313405284,
+      "grad_norm": 0.18108299374580383,
+      "learning_rate": 4.561126915532006e-06,
+      "loss": 0.9339,
+      "step": 60630
+    },
+    {
+      "epoch": 0.438952709794639,
+      "grad_norm": 0.15784204006195068,
+      "learning_rate": 4.5610545288714195e-06,
+      "loss": 0.9267,
+      "step": 60640
+    },
+    {
+      "epoch": 0.43902509645522525,
+      "grad_norm": 0.15953408181667328,
+      "learning_rate": 4.560982142210834e-06,
+      "loss": 0.9392,
+      "step": 60650
+    },
+    {
+      "epoch": 0.4390974831158114,
+      "grad_norm": 0.14857307076454163,
+      "learning_rate": 4.560909755550248e-06,
+      "loss": 0.9322,
+      "step": 60660
+    },
+    {
+      "epoch": 0.4391698697763976,
+      "grad_norm": 0.16134792566299438,
+      "learning_rate": 4.560837368889661e-06,
+      "loss": 0.9428,
+      "step": 60670
+    },
+    {
+      "epoch": 0.4392422564369838,
+      "grad_norm": 0.15805503726005554,
+      "learning_rate": 4.560764982229075e-06,
+      "loss": 0.9246,
+      "step": 60680
+    },
+    {
+      "epoch": 0.43931464309756996,
+      "grad_norm": 0.16385342180728912,
+      "learning_rate": 4.560692595568489e-06,
+      "loss": 0.9318,
+      "step": 60690
+    },
+    {
+      "epoch": 0.4393870297581562,
+      "grad_norm": 0.16679242253303528,
+      "learning_rate": 4.560620208907903e-06,
+      "loss": 0.915,
+      "step": 60700
+    },
+    {
+      "epoch": 0.43945941641874237,
+      "grad_norm": 0.15610092878341675,
+      "learning_rate": 4.5605478222473165e-06,
+      "loss": 0.926,
+      "step": 60710
+    },
+    {
+      "epoch": 0.43953180307932854,
+      "grad_norm": 0.16888998448848724,
+      "learning_rate": 4.56047543558673e-06,
+      "loss": 0.9093,
+      "step": 60720
+    },
+    {
+      "epoch": 0.4396041897399147,
+      "grad_norm": 0.31106677651405334,
+      "learning_rate": 4.560403048926144e-06,
+      "loss": 0.9279,
+      "step": 60730
+    },
+    {
+      "epoch": 0.4396765764005009,
+      "grad_norm": 0.16773858666419983,
+      "learning_rate": 4.560330662265558e-06,
+      "loss": 0.9385,
+      "step": 60740
+    },
+    {
+      "epoch": 0.43974896306108713,
+      "grad_norm": 0.1473310887813568,
+      "learning_rate": 4.560258275604972e-06,
+      "loss": 0.9347,
+      "step": 60750
+    },
+    {
+      "epoch": 0.4398213497216733,
+      "grad_norm": 0.17835259437561035,
+      "learning_rate": 4.5601858889443854e-06,
+      "loss": 0.9369,
+      "step": 60760
+    },
+    {
+      "epoch": 0.4398937363822595,
+      "grad_norm": 0.15652461349964142,
+      "learning_rate": 4.560113502283799e-06,
+      "loss": 0.9309,
+      "step": 60770
+    },
+    {
+      "epoch": 0.43996612304284566,
+      "grad_norm": 0.15282343327999115,
+      "learning_rate": 4.5600411156232135e-06,
+      "loss": 0.927,
+      "step": 60780
+    },
+    {
+      "epoch": 0.44003850970343183,
+      "grad_norm": 0.16254651546478271,
+      "learning_rate": 4.559968728962627e-06,
+      "loss": 0.9371,
+      "step": 60790
+    },
+    {
+      "epoch": 0.440110896364018,
+      "grad_norm": 0.16506025195121765,
+      "learning_rate": 4.559896342302041e-06,
+      "loss": 0.9312,
+      "step": 60800
+    },
+    {
+      "epoch": 0.44018328302460424,
+      "grad_norm": 0.16597755253314972,
+      "learning_rate": 4.559823955641454e-06,
+      "loss": 0.9336,
+      "step": 60810
+    },
+    {
+      "epoch": 0.4402556696851904,
+      "grad_norm": 0.15858708322048187,
+      "learning_rate": 4.559751568980869e-06,
+      "loss": 0.9463,
+      "step": 60820
+    },
+    {
+      "epoch": 0.4403280563457766,
+      "grad_norm": 0.16279590129852295,
+      "learning_rate": 4.5596791823202824e-06,
+      "loss": 0.9377,
+      "step": 60830
+    },
+    {
+      "epoch": 0.4404004430063628,
+      "grad_norm": 0.16087251901626587,
+      "learning_rate": 4.559606795659696e-06,
+      "loss": 0.9271,
+      "step": 60840
+    },
+    {
+      "epoch": 0.44047282966694895,
+      "grad_norm": 0.17947107553482056,
+      "learning_rate": 4.55953440899911e-06,
+      "loss": 0.9367,
+      "step": 60850
+    },
+    {
+      "epoch": 0.4405452163275352,
+      "grad_norm": 0.17072343826293945,
+      "learning_rate": 4.559462022338524e-06,
+      "loss": 0.9294,
+      "step": 60860
+    },
+    {
+      "epoch": 0.44061760298812136,
+      "grad_norm": 0.16467486321926117,
+      "learning_rate": 4.559389635677938e-06,
+      "loss": 0.9368,
+      "step": 60870
+    },
+    {
+      "epoch": 0.44068998964870754,
+      "grad_norm": 0.1682046353816986,
+      "learning_rate": 4.559317249017351e-06,
+      "loss": 0.9272,
+      "step": 60880
+    },
+    {
+      "epoch": 0.4407623763092937,
+      "grad_norm": 0.16743090748786926,
+      "learning_rate": 4.559244862356765e-06,
+      "loss": 0.9356,
+      "step": 60890
+    },
+    {
+      "epoch": 0.4408347629698799,
+      "grad_norm": 0.16245803236961365,
+      "learning_rate": 4.5591724756961795e-06,
+      "loss": 0.9326,
+      "step": 60900
+    },
+    {
+      "epoch": 0.4409071496304661,
+      "grad_norm": 0.1696007251739502,
+      "learning_rate": 4.559100089035593e-06,
+      "loss": 0.9131,
+      "step": 60910
+    },
+    {
+      "epoch": 0.4409795362910523,
+      "grad_norm": 0.16754589974880219,
+      "learning_rate": 4.559027702375007e-06,
+      "loss": 0.9268,
+      "step": 60920
+    },
+    {
+      "epoch": 0.4410519229516385,
+      "grad_norm": 0.17982327938079834,
+      "learning_rate": 4.55895531571442e-06,
+      "loss": 0.9366,
+      "step": 60930
+    },
+    {
+      "epoch": 0.44112430961222465,
+      "grad_norm": 0.15991713106632233,
+      "learning_rate": 4.558882929053835e-06,
+      "loss": 0.9275,
+      "step": 60940
+    },
+    {
+      "epoch": 0.44119669627281083,
+      "grad_norm": 0.16579997539520264,
+      "learning_rate": 4.558810542393248e-06,
+      "loss": 0.935,
+      "step": 60950
+    },
+    {
+      "epoch": 0.44126908293339706,
+      "grad_norm": 0.1842290610074997,
+      "learning_rate": 4.558738155732662e-06,
+      "loss": 0.9342,
+      "step": 60960
+    },
+    {
+      "epoch": 0.44134146959398324,
+      "grad_norm": 0.1753285974264145,
+      "learning_rate": 4.558665769072076e-06,
+      "loss": 0.9382,
+      "step": 60970
+    },
+    {
+      "epoch": 0.4414138562545694,
+      "grad_norm": 0.1587180495262146,
+      "learning_rate": 4.55859338241149e-06,
+      "loss": 0.9397,
+      "step": 60980
+    },
+    {
+      "epoch": 0.4414862429151556,
+      "grad_norm": 0.2102600336074829,
+      "learning_rate": 4.558520995750904e-06,
+      "loss": 0.9305,
+      "step": 60990
+    },
+    {
+      "epoch": 0.44155862957574177,
+      "grad_norm": 0.1546296626329422,
+      "learning_rate": 4.558448609090317e-06,
+      "loss": 0.9228,
+      "step": 61000
+    },
+    {
+      "epoch": 0.44163101623632794,
+      "grad_norm": 0.16108326613903046,
+      "learning_rate": 4.558376222429731e-06,
+      "loss": 0.94,
+      "step": 61010
+    },
+    {
+      "epoch": 0.4417034028969142,
+      "grad_norm": 0.17151787877082825,
+      "learning_rate": 4.558303835769145e-06,
+      "loss": 0.9354,
+      "step": 61020
+    },
+    {
+      "epoch": 0.44177578955750035,
+      "grad_norm": 0.17950987815856934,
+      "learning_rate": 4.558231449108559e-06,
+      "loss": 0.9282,
+      "step": 61030
+    },
+    {
+      "epoch": 0.44184817621808653,
+      "grad_norm": 0.16002346575260162,
+      "learning_rate": 4.558159062447973e-06,
+      "loss": 0.9411,
+      "step": 61040
+    },
+    {
+      "epoch": 0.4419205628786727,
+      "grad_norm": 0.15132936835289001,
+      "learning_rate": 4.558086675787386e-06,
+      "loss": 0.9273,
+      "step": 61050
+    },
+    {
+      "epoch": 0.4419929495392589,
+      "grad_norm": 0.19634242355823517,
+      "learning_rate": 4.558014289126801e-06,
+      "loss": 0.9273,
+      "step": 61060
+    },
+    {
+      "epoch": 0.4420653361998451,
+      "grad_norm": 0.17866051197052002,
+      "learning_rate": 4.557941902466214e-06,
+      "loss": 0.9362,
+      "step": 61070
+    },
+    {
+      "epoch": 0.4421377228604313,
+      "grad_norm": 0.16125331819057465,
+      "learning_rate": 4.557869515805628e-06,
+      "loss": 0.9297,
+      "step": 61080
+    },
+    {
+      "epoch": 0.44221010952101747,
+      "grad_norm": 0.16221415996551514,
+      "learning_rate": 4.5577971291450416e-06,
+      "loss": 0.925,
+      "step": 61090
+    },
+    {
+      "epoch": 0.44228249618160365,
+      "grad_norm": 0.1527089923620224,
+      "learning_rate": 4.557724742484455e-06,
+      "loss": 0.9194,
+      "step": 61100
+    },
+    {
+      "epoch": 0.4423548828421898,
+      "grad_norm": 0.17006568610668182,
+      "learning_rate": 4.557652355823869e-06,
+      "loss": 0.9305,
+      "step": 61110
+    },
+    {
+      "epoch": 0.44242726950277605,
+      "grad_norm": 0.1684480905532837,
+      "learning_rate": 4.557579969163282e-06,
+      "loss": 0.9277,
+      "step": 61120
+    },
+    {
+      "epoch": 0.44249965616336223,
+      "grad_norm": 0.16048870980739594,
+      "learning_rate": 4.557507582502697e-06,
+      "loss": 0.9333,
+      "step": 61130
+    },
+    {
+      "epoch": 0.4425720428239484,
+      "grad_norm": 0.16578523814678192,
+      "learning_rate": 4.5574351958421105e-06,
+      "loss": 0.9288,
+      "step": 61140
+    },
+    {
+      "epoch": 0.4426444294845346,
+      "grad_norm": 0.16607344150543213,
+      "learning_rate": 4.557362809181524e-06,
+      "loss": 0.9125,
+      "step": 61150
+    },
+    {
+      "epoch": 0.44271681614512076,
+      "grad_norm": 0.1659523993730545,
+      "learning_rate": 4.557290422520938e-06,
+      "loss": 0.9324,
+      "step": 61160
+    },
+    {
+      "epoch": 0.44278920280570694,
+      "grad_norm": 0.17935402691364288,
+      "learning_rate": 4.557218035860352e-06,
+      "loss": 0.9362,
+      "step": 61170
+    },
+    {
+      "epoch": 0.44286158946629317,
+      "grad_norm": 0.16320429742336273,
+      "learning_rate": 4.557145649199766e-06,
+      "loss": 0.9305,
+      "step": 61180
+    },
+    {
+      "epoch": 0.44293397612687935,
+      "grad_norm": 0.15417033433914185,
+      "learning_rate": 4.557073262539179e-06,
+      "loss": 0.9378,
+      "step": 61190
+    },
+    {
+      "epoch": 0.4430063627874655,
+      "grad_norm": 0.14864222705364227,
+      "learning_rate": 4.557000875878593e-06,
+      "loss": 0.9401,
+      "step": 61200
+    },
+    {
+      "epoch": 0.4430787494480517,
+      "grad_norm": 0.189174622297287,
+      "learning_rate": 4.5569284892180075e-06,
+      "loss": 0.9357,
+      "step": 61210
+    },
+    {
+      "epoch": 0.4431511361086379,
+      "grad_norm": 0.15933531522750854,
+      "learning_rate": 4.556856102557421e-06,
+      "loss": 0.9296,
+      "step": 61220
+    },
+    {
+      "epoch": 0.4432235227692241,
+      "grad_norm": 0.15585920214653015,
+      "learning_rate": 4.556783715896835e-06,
+      "loss": 0.9277,
+      "step": 61230
+    },
+    {
+      "epoch": 0.4432959094298103,
+      "grad_norm": 0.17847231030464172,
+      "learning_rate": 4.556711329236248e-06,
+      "loss": 0.9382,
+      "step": 61240
+    },
+    {
+      "epoch": 0.44336829609039646,
+      "grad_norm": 0.15778236091136932,
+      "learning_rate": 4.556638942575663e-06,
+      "loss": 0.9191,
+      "step": 61250
+    },
+    {
+      "epoch": 0.44344068275098264,
+      "grad_norm": 0.15541084110736847,
+      "learning_rate": 4.556566555915076e-06,
+      "loss": 0.9389,
+      "step": 61260
+    },
+    {
+      "epoch": 0.4435130694115688,
+      "grad_norm": 0.17525739967823029,
+      "learning_rate": 4.55649416925449e-06,
+      "loss": 0.9447,
+      "step": 61270
+    },
+    {
+      "epoch": 0.44358545607215505,
+      "grad_norm": 0.15737952291965485,
+      "learning_rate": 4.556421782593904e-06,
+      "loss": 0.9284,
+      "step": 61280
+    },
+    {
+      "epoch": 0.4436578427327412,
+      "grad_norm": 0.16541706025600433,
+      "learning_rate": 4.556349395933318e-06,
+      "loss": 0.9284,
+      "step": 61290
+    },
+    {
+      "epoch": 0.4437302293933274,
+      "grad_norm": 0.1597539633512497,
+      "learning_rate": 4.556277009272732e-06,
+      "loss": 0.9297,
+      "step": 61300
+    },
+    {
+      "epoch": 0.4438026160539136,
+      "grad_norm": 0.15804599225521088,
+      "learning_rate": 4.556204622612145e-06,
+      "loss": 0.9233,
+      "step": 61310
+    },
+    {
+      "epoch": 0.44387500271449976,
+      "grad_norm": 0.1523711085319519,
+      "learning_rate": 4.556132235951559e-06,
+      "loss": 0.9259,
+      "step": 61320
+    },
+    {
+      "epoch": 0.44394738937508593,
+      "grad_norm": 0.18194106221199036,
+      "learning_rate": 4.556059849290973e-06,
+      "loss": 0.937,
+      "step": 61330
+    },
+    {
+      "epoch": 0.44401977603567216,
+      "grad_norm": 0.15899352729320526,
+      "learning_rate": 4.555987462630387e-06,
+      "loss": 0.9282,
+      "step": 61340
+    },
+    {
+      "epoch": 0.44409216269625834,
+      "grad_norm": 0.14628228545188904,
+      "learning_rate": 4.555915075969801e-06,
+      "loss": 0.9209,
+      "step": 61350
+    },
+    {
+      "epoch": 0.4441645493568445,
+      "grad_norm": 0.15761525928974152,
+      "learning_rate": 4.555842689309214e-06,
+      "loss": 0.9334,
+      "step": 61360
+    },
+    {
+      "epoch": 0.4442369360174307,
+      "grad_norm": 0.1836678832769394,
+      "learning_rate": 4.555770302648628e-06,
+      "loss": 0.9142,
+      "step": 61370
+    },
+    {
+      "epoch": 0.44430932267801687,
+      "grad_norm": 0.1548013538122177,
+      "learning_rate": 4.555697915988042e-06,
+      "loss": 0.9247,
+      "step": 61380
+    },
+    {
+      "epoch": 0.4443817093386031,
+      "grad_norm": 0.16135182976722717,
+      "learning_rate": 4.555625529327456e-06,
+      "loss": 0.9311,
+      "step": 61390
+    },
+    {
+      "epoch": 0.4444540959991893,
+      "grad_norm": 0.1575031578540802,
+      "learning_rate": 4.55555314266687e-06,
+      "loss": 0.9356,
+      "step": 61400
+    },
+    {
+      "epoch": 0.44452648265977546,
+      "grad_norm": 0.170990988612175,
+      "learning_rate": 4.555480756006283e-06,
+      "loss": 0.9471,
+      "step": 61410
+    },
+    {
+      "epoch": 0.44459886932036163,
+      "grad_norm": 0.157973051071167,
+      "learning_rate": 4.555408369345698e-06,
+      "loss": 0.9324,
+      "step": 61420
+    },
+    {
+      "epoch": 0.4446712559809478,
+      "grad_norm": 0.15711332857608795,
+      "learning_rate": 4.555335982685111e-06,
+      "loss": 0.9344,
+      "step": 61430
+    },
+    {
+      "epoch": 0.44474364264153404,
+      "grad_norm": 0.16744282841682434,
+      "learning_rate": 4.555263596024525e-06,
+      "loss": 0.9255,
+      "step": 61440
+    },
+    {
+      "epoch": 0.4448160293021202,
+      "grad_norm": 0.16048872470855713,
+      "learning_rate": 4.5551912093639385e-06,
+      "loss": 0.9338,
+      "step": 61450
+    },
+    {
+      "epoch": 0.4448884159627064,
+      "grad_norm": 0.17134064435958862,
+      "learning_rate": 4.555118822703353e-06,
+      "loss": 0.9264,
+      "step": 61460
+    },
+    {
+      "epoch": 0.4449608026232926,
+      "grad_norm": 0.15895238518714905,
+      "learning_rate": 4.555046436042767e-06,
+      "loss": 0.9251,
+      "step": 61470
+    },
+    {
+      "epoch": 0.44503318928387875,
+      "grad_norm": 0.16355475783348083,
+      "learning_rate": 4.55497404938218e-06,
+      "loss": 0.9235,
+      "step": 61480
+    },
+    {
+      "epoch": 0.445105575944465,
+      "grad_norm": 0.15274769067764282,
+      "learning_rate": 4.554901662721594e-06,
+      "loss": 0.9244,
+      "step": 61490
+    },
+    {
+      "epoch": 0.44517796260505116,
+      "grad_norm": 0.24952305853366852,
+      "learning_rate": 4.554829276061008e-06,
+      "loss": 0.9222,
+      "step": 61500
+    },
+    {
+      "epoch": 0.44525034926563734,
+      "grad_norm": 0.17668789625167847,
+      "learning_rate": 4.554756889400422e-06,
+      "loss": 0.9155,
+      "step": 61510
+    },
+    {
+      "epoch": 0.4453227359262235,
+      "grad_norm": 0.22048163414001465,
+      "learning_rate": 4.5546845027398355e-06,
+      "loss": 0.921,
+      "step": 61520
+    },
+    {
+      "epoch": 0.4453951225868097,
+      "grad_norm": 0.1538519263267517,
+      "learning_rate": 4.554612116079249e-06,
+      "loss": 0.9302,
+      "step": 61530
+    },
+    {
+      "epoch": 0.44546750924739587,
+      "grad_norm": 0.1744624376296997,
+      "learning_rate": 4.554539729418664e-06,
+      "loss": 0.9266,
+      "step": 61540
+    },
+    {
+      "epoch": 0.4455398959079821,
+      "grad_norm": 0.17444263398647308,
+      "learning_rate": 4.554467342758077e-06,
+      "loss": 0.9306,
+      "step": 61550
+    },
+    {
+      "epoch": 0.4456122825685683,
+      "grad_norm": 0.16291213035583496,
+      "learning_rate": 4.554394956097491e-06,
+      "loss": 0.9388,
+      "step": 61560
+    },
+    {
+      "epoch": 0.44568466922915445,
+      "grad_norm": 0.16425669193267822,
+      "learning_rate": 4.5543225694369044e-06,
+      "loss": 0.9165,
+      "step": 61570
+    },
+    {
+      "epoch": 0.4457570558897406,
+      "grad_norm": 0.16991227865219116,
+      "learning_rate": 4.554250182776319e-06,
+      "loss": 0.9208,
+      "step": 61580
+    },
+    {
+      "epoch": 0.4458294425503268,
+      "grad_norm": 0.17110396921634674,
+      "learning_rate": 4.5541777961157325e-06,
+      "loss": 0.9288,
+      "step": 61590
+    },
+    {
+      "epoch": 0.44590182921091304,
+      "grad_norm": 0.17774710059165955,
+      "learning_rate": 4.554105409455146e-06,
+      "loss": 0.9334,
+      "step": 61600
+    },
+    {
+      "epoch": 0.4459742158714992,
+      "grad_norm": 0.18292616307735443,
+      "learning_rate": 4.55403302279456e-06,
+      "loss": 0.9189,
+      "step": 61610
+    },
+    {
+      "epoch": 0.4460466025320854,
+      "grad_norm": 0.15850049257278442,
+      "learning_rate": 4.553960636133974e-06,
+      "loss": 0.9296,
+      "step": 61620
+    },
+    {
+      "epoch": 0.44611898919267157,
+      "grad_norm": 0.18073786795139313,
+      "learning_rate": 4.553888249473387e-06,
+      "loss": 0.9159,
+      "step": 61630
+    },
+    {
+      "epoch": 0.44619137585325774,
+      "grad_norm": 0.15304653346538544,
+      "learning_rate": 4.553815862812801e-06,
+      "loss": 0.9312,
+      "step": 61640
+    },
+    {
+      "epoch": 0.446263762513844,
+      "grad_norm": 0.18145766854286194,
+      "learning_rate": 4.553743476152215e-06,
+      "loss": 0.9203,
+      "step": 61650
+    },
+    {
+      "epoch": 0.44633614917443015,
+      "grad_norm": 0.16524021327495575,
+      "learning_rate": 4.553671089491629e-06,
+      "loss": 0.9272,
+      "step": 61660
+    },
+    {
+      "epoch": 0.44640853583501633,
+      "grad_norm": 0.17342203855514526,
+      "learning_rate": 4.553598702831042e-06,
+      "loss": 0.9182,
+      "step": 61670
+    },
+    {
+      "epoch": 0.4464809224956025,
+      "grad_norm": 0.165330708026886,
+      "learning_rate": 4.553526316170456e-06,
+      "loss": 0.9312,
+      "step": 61680
+    },
+    {
+      "epoch": 0.4465533091561887,
+      "grad_norm": 0.166887104511261,
+      "learning_rate": 4.55345392950987e-06,
+      "loss": 0.9259,
+      "step": 61690
+    },
+    {
+      "epoch": 0.44662569581677486,
+      "grad_norm": 0.16618217527866364,
+      "learning_rate": 4.553381542849284e-06,
+      "loss": 0.9182,
+      "step": 61700
+    },
+    {
+      "epoch": 0.4466980824773611,
+      "grad_norm": 0.15841081738471985,
+      "learning_rate": 4.553309156188698e-06,
+      "loss": 0.9348,
+      "step": 61710
+    },
+    {
+      "epoch": 0.44677046913794727,
+      "grad_norm": 0.15774931013584137,
+      "learning_rate": 4.553236769528111e-06,
+      "loss": 0.9285,
+      "step": 61720
+    },
+    {
+      "epoch": 0.44684285579853344,
+      "grad_norm": 0.16773931682109833,
+      "learning_rate": 4.553164382867526e-06,
+      "loss": 0.9249,
+      "step": 61730
+    },
+    {
+      "epoch": 0.4469152424591196,
+      "grad_norm": 0.16781432926654816,
+      "learning_rate": 4.553091996206939e-06,
+      "loss": 0.9372,
+      "step": 61740
+    },
+    {
+      "epoch": 0.4469876291197058,
+      "grad_norm": 0.15466833114624023,
+      "learning_rate": 4.553019609546353e-06,
+      "loss": 0.9349,
+      "step": 61750
+    },
+    {
+      "epoch": 0.44706001578029203,
+      "grad_norm": 0.1791720688343048,
+      "learning_rate": 4.5529472228857665e-06,
+      "loss": 0.9285,
+      "step": 61760
+    },
+    {
+      "epoch": 0.4471324024408782,
+      "grad_norm": 0.15556438267230988,
+      "learning_rate": 4.552874836225181e-06,
+      "loss": 0.9269,
+      "step": 61770
+    },
+    {
+      "epoch": 0.4472047891014644,
+      "grad_norm": 0.1610628068447113,
+      "learning_rate": 4.552802449564595e-06,
+      "loss": 0.919,
+      "step": 61780
+    },
+    {
+      "epoch": 0.44727717576205056,
+      "grad_norm": 0.16492611169815063,
+      "learning_rate": 4.552730062904008e-06,
+      "loss": 0.9167,
+      "step": 61790
+    },
+    {
+      "epoch": 0.44734956242263674,
+      "grad_norm": 0.15720947086811066,
+      "learning_rate": 4.552657676243422e-06,
+      "loss": 0.9226,
+      "step": 61800
+    },
+    {
+      "epoch": 0.44742194908322297,
+      "grad_norm": 0.1607959121465683,
+      "learning_rate": 4.552585289582836e-06,
+      "loss": 0.9386,
+      "step": 61810
+    },
+    {
+      "epoch": 0.44749433574380915,
+      "grad_norm": 0.19096055626869202,
+      "learning_rate": 4.55251290292225e-06,
+      "loss": 0.9261,
+      "step": 61820
+    },
+    {
+      "epoch": 0.4475667224043953,
+      "grad_norm": 0.1570555865764618,
+      "learning_rate": 4.5524405162616636e-06,
+      "loss": 0.9382,
+      "step": 61830
+    },
+    {
+      "epoch": 0.4476391090649815,
+      "grad_norm": 0.16010279953479767,
+      "learning_rate": 4.552368129601077e-06,
+      "loss": 0.9222,
+      "step": 61840
+    },
+    {
+      "epoch": 0.4477114957255677,
+      "grad_norm": 0.1684756577014923,
+      "learning_rate": 4.552295742940492e-06,
+      "loss": 0.9308,
+      "step": 61850
+    },
+    {
+      "epoch": 0.44778388238615385,
+      "grad_norm": 0.15752823650836945,
+      "learning_rate": 4.552223356279905e-06,
+      "loss": 0.9248,
+      "step": 61860
+    },
+    {
+      "epoch": 0.4478562690467401,
+      "grad_norm": 0.15489928424358368,
+      "learning_rate": 4.552150969619319e-06,
+      "loss": 0.9236,
+      "step": 61870
+    },
+    {
+      "epoch": 0.44792865570732626,
+      "grad_norm": 0.1638929694890976,
+      "learning_rate": 4.5520785829587325e-06,
+      "loss": 0.9329,
+      "step": 61880
+    },
+    {
+      "epoch": 0.44800104236791244,
+      "grad_norm": 0.15785187482833862,
+      "learning_rate": 4.552006196298147e-06,
+      "loss": 0.9346,
+      "step": 61890
+    },
+    {
+      "epoch": 0.4480734290284986,
+      "grad_norm": 0.16453564167022705,
+      "learning_rate": 4.5519338096375606e-06,
+      "loss": 0.9417,
+      "step": 61900
+    },
+    {
+      "epoch": 0.4481458156890848,
+      "grad_norm": 0.15828080475330353,
+      "learning_rate": 4.551861422976974e-06,
+      "loss": 0.9302,
+      "step": 61910
+    },
+    {
+      "epoch": 0.448218202349671,
+      "grad_norm": 0.1843886375427246,
+      "learning_rate": 4.551789036316388e-06,
+      "loss": 0.9294,
+      "step": 61920
+    },
+    {
+      "epoch": 0.4482905890102572,
+      "grad_norm": 0.17021195590496063,
+      "learning_rate": 4.551716649655802e-06,
+      "loss": 0.9208,
+      "step": 61930
+    },
+    {
+      "epoch": 0.4483629756708434,
+      "grad_norm": 0.16324369609355927,
+      "learning_rate": 4.551644262995216e-06,
+      "loss": 0.9285,
+      "step": 61940
+    },
+    {
+      "epoch": 0.44843536233142955,
+      "grad_norm": 0.16327841579914093,
+      "learning_rate": 4.5515718763346295e-06,
+      "loss": 0.9336,
+      "step": 61950
+    },
+    {
+      "epoch": 0.44850774899201573,
+      "grad_norm": 0.16559500992298126,
+      "learning_rate": 4.551499489674043e-06,
+      "loss": 0.9416,
+      "step": 61960
+    },
+    {
+      "epoch": 0.44858013565260196,
+      "grad_norm": 0.16372480988502502,
+      "learning_rate": 4.551427103013457e-06,
+      "loss": 0.9315,
+      "step": 61970
+    },
+    {
+      "epoch": 0.44865252231318814,
+      "grad_norm": 0.16200987994670868,
+      "learning_rate": 4.551354716352871e-06,
+      "loss": 0.9153,
+      "step": 61980
+    },
+    {
+      "epoch": 0.4487249089737743,
+      "grad_norm": 0.17236877977848053,
+      "learning_rate": 4.551282329692285e-06,
+      "loss": 0.9357,
+      "step": 61990
+    },
+    {
+      "epoch": 0.4487972956343605,
+      "grad_norm": 0.22826656699180603,
+      "learning_rate": 4.551209943031698e-06,
+      "loss": 0.9265,
+      "step": 62000
+    },
+    {
+      "epoch": 0.44886968229494667,
+      "grad_norm": 0.1597038060426712,
+      "learning_rate": 4.551137556371112e-06,
+      "loss": 0.9292,
+      "step": 62010
+    },
+    {
+      "epoch": 0.44894206895553285,
+      "grad_norm": 0.15479369461536407,
+      "learning_rate": 4.5510651697105265e-06,
+      "loss": 0.9452,
+      "step": 62020
+    },
+    {
+      "epoch": 0.4490144556161191,
+      "grad_norm": 0.1617387980222702,
+      "learning_rate": 4.55099278304994e-06,
+      "loss": 0.9311,
+      "step": 62030
+    },
+    {
+      "epoch": 0.44908684227670526,
+      "grad_norm": 0.17810650169849396,
+      "learning_rate": 4.550920396389354e-06,
+      "loss": 0.9337,
+      "step": 62040
+    },
+    {
+      "epoch": 0.44915922893729143,
+      "grad_norm": 0.15863506495952606,
+      "learning_rate": 4.550848009728767e-06,
+      "loss": 0.9391,
+      "step": 62050
+    },
+    {
+      "epoch": 0.4492316155978776,
+      "grad_norm": 0.18483230471611023,
+      "learning_rate": 4.550775623068182e-06,
+      "loss": 0.924,
+      "step": 62060
+    },
+    {
+      "epoch": 0.4493040022584638,
+      "grad_norm": 0.15691833198070526,
+      "learning_rate": 4.550703236407595e-06,
+      "loss": 0.9286,
+      "step": 62070
+    },
+    {
+      "epoch": 0.44937638891905,
+      "grad_norm": 0.15761037170886993,
+      "learning_rate": 4.550630849747009e-06,
+      "loss": 0.9414,
+      "step": 62080
+    },
+    {
+      "epoch": 0.4494487755796362,
+      "grad_norm": 0.15282666683197021,
+      "learning_rate": 4.550558463086423e-06,
+      "loss": 0.931,
+      "step": 62090
+    },
+    {
+      "epoch": 0.44952116224022237,
+      "grad_norm": 0.1992768943309784,
+      "learning_rate": 4.550486076425837e-06,
+      "loss": 0.938,
+      "step": 62100
+    },
+    {
+      "epoch": 0.44959354890080855,
+      "grad_norm": 0.15177763998508453,
+      "learning_rate": 4.550413689765251e-06,
+      "loss": 0.9436,
+      "step": 62110
+    },
+    {
+      "epoch": 0.4496659355613947,
+      "grad_norm": 0.16348634660243988,
+      "learning_rate": 4.550341303104664e-06,
+      "loss": 0.9394,
+      "step": 62120
+    },
+    {
+      "epoch": 0.44973832222198096,
+      "grad_norm": 0.18338032066822052,
+      "learning_rate": 4.550268916444078e-06,
+      "loss": 0.9319,
+      "step": 62130
+    },
+    {
+      "epoch": 0.44981070888256713,
+      "grad_norm": 0.22670073807239532,
+      "learning_rate": 4.5501965297834924e-06,
+      "loss": 0.936,
+      "step": 62140
+    },
+    {
+      "epoch": 0.4498830955431533,
+      "grad_norm": 0.17168883979320526,
+      "learning_rate": 4.550124143122906e-06,
+      "loss": 0.9335,
+      "step": 62150
+    },
+    {
+      "epoch": 0.4499554822037395,
+      "grad_norm": 0.16021141409873962,
+      "learning_rate": 4.550051756462319e-06,
+      "loss": 0.9373,
+      "step": 62160
+    },
+    {
+      "epoch": 0.45002786886432566,
+      "grad_norm": 0.1671023815870285,
+      "learning_rate": 4.549979369801733e-06,
+      "loss": 0.927,
+      "step": 62170
+    },
+    {
+      "epoch": 0.4501002555249119,
+      "grad_norm": 0.15953001379966736,
+      "learning_rate": 4.549906983141147e-06,
+      "loss": 0.9296,
+      "step": 62180
+    },
+    {
+      "epoch": 0.4501726421854981,
+      "grad_norm": 0.19120389223098755,
+      "learning_rate": 4.5498345964805605e-06,
+      "loss": 0.9239,
+      "step": 62190
+    },
+    {
+      "epoch": 0.45024502884608425,
+      "grad_norm": 0.1692996323108673,
+      "learning_rate": 4.549762209819974e-06,
+      "loss": 0.9201,
+      "step": 62200
+    },
+    {
+      "epoch": 0.4503174155066704,
+      "grad_norm": 0.15161260962486267,
+      "learning_rate": 4.549689823159389e-06,
+      "loss": 0.933,
+      "step": 62210
+    },
+    {
+      "epoch": 0.4503898021672566,
+      "grad_norm": 0.1791296750307083,
+      "learning_rate": 4.549617436498802e-06,
+      "loss": 0.9206,
+      "step": 62220
+    },
+    {
+      "epoch": 0.4504621888278428,
+      "grad_norm": 0.16728129982948303,
+      "learning_rate": 4.549545049838216e-06,
+      "loss": 0.9243,
+      "step": 62230
+    },
+    {
+      "epoch": 0.450534575488429,
+      "grad_norm": 0.15448437631130219,
+      "learning_rate": 4.5494726631776294e-06,
+      "loss": 0.922,
+      "step": 62240
+    },
+    {
+      "epoch": 0.4506069621490152,
+      "grad_norm": 0.17442266643047333,
+      "learning_rate": 4.549400276517044e-06,
+      "loss": 0.9266,
+      "step": 62250
+    },
+    {
+      "epoch": 0.45067934880960137,
+      "grad_norm": 0.16572901606559753,
+      "learning_rate": 4.5493278898564575e-06,
+      "loss": 0.9262,
+      "step": 62260
+    },
+    {
+      "epoch": 0.45075173547018754,
+      "grad_norm": 0.19222629070281982,
+      "learning_rate": 4.549255503195871e-06,
+      "loss": 0.9271,
+      "step": 62270
+    },
+    {
+      "epoch": 0.4508241221307737,
+      "grad_norm": 0.180272176861763,
+      "learning_rate": 4.549183116535285e-06,
+      "loss": 0.9252,
+      "step": 62280
+    },
+    {
+      "epoch": 0.45089650879135995,
+      "grad_norm": 0.1528458446264267,
+      "learning_rate": 4.549110729874699e-06,
+      "loss": 0.916,
+      "step": 62290
+    },
+    {
+      "epoch": 0.45096889545194613,
+      "grad_norm": 0.16604572534561157,
+      "learning_rate": 4.549038343214113e-06,
+      "loss": 0.9209,
+      "step": 62300
+    },
+    {
+      "epoch": 0.4510412821125323,
+      "grad_norm": 0.1774618774652481,
+      "learning_rate": 4.5489659565535264e-06,
+      "loss": 0.9266,
+      "step": 62310
+    },
+    {
+      "epoch": 0.4511136687731185,
+      "grad_norm": 0.16191911697387695,
+      "learning_rate": 4.54889356989294e-06,
+      "loss": 0.9276,
+      "step": 62320
+    },
+    {
+      "epoch": 0.45118605543370466,
+      "grad_norm": 0.18299001455307007,
+      "learning_rate": 4.5488211832323545e-06,
+      "loss": 0.9334,
+      "step": 62330
+    },
+    {
+      "epoch": 0.4512584420942909,
+      "grad_norm": 0.14922909438610077,
+      "learning_rate": 4.548748796571768e-06,
+      "loss": 0.9232,
+      "step": 62340
+    },
+    {
+      "epoch": 0.45133082875487707,
+      "grad_norm": 0.14895036816596985,
+      "learning_rate": 4.548676409911182e-06,
+      "loss": 0.9299,
+      "step": 62350
+    },
+    {
+      "epoch": 0.45140321541546324,
+      "grad_norm": 0.15128493309020996,
+      "learning_rate": 4.548604023250595e-06,
+      "loss": 0.9409,
+      "step": 62360
+    },
+    {
+      "epoch": 0.4514756020760494,
+      "grad_norm": 0.1607513725757599,
+      "learning_rate": 4.54853163659001e-06,
+      "loss": 0.9143,
+      "step": 62370
+    },
+    {
+      "epoch": 0.4515479887366356,
+      "grad_norm": 0.1618734747171402,
+      "learning_rate": 4.5484592499294235e-06,
+      "loss": 0.9249,
+      "step": 62380
+    },
+    {
+      "epoch": 0.4516203753972218,
+      "grad_norm": 0.18562345206737518,
+      "learning_rate": 4.548386863268837e-06,
+      "loss": 0.9292,
+      "step": 62390
+    },
+    {
+      "epoch": 0.451692762057808,
+      "grad_norm": 0.17333509027957916,
+      "learning_rate": 4.548314476608251e-06,
+      "loss": 0.929,
+      "step": 62400
+    },
+    {
+      "epoch": 0.4517651487183942,
+      "grad_norm": 0.1685020625591278,
+      "learning_rate": 4.548242089947665e-06,
+      "loss": 0.9247,
+      "step": 62410
+    },
+    {
+      "epoch": 0.45183753537898036,
+      "grad_norm": 0.16872678697109222,
+      "learning_rate": 4.548169703287079e-06,
+      "loss": 0.9225,
+      "step": 62420
+    },
+    {
+      "epoch": 0.45190992203956654,
+      "grad_norm": 0.15158583223819733,
+      "learning_rate": 4.548097316626492e-06,
+      "loss": 0.9277,
+      "step": 62430
+    },
+    {
+      "epoch": 0.4519823087001527,
+      "grad_norm": 0.18446789681911469,
+      "learning_rate": 4.548024929965906e-06,
+      "loss": 0.9227,
+      "step": 62440
+    },
+    {
+      "epoch": 0.45205469536073895,
+      "grad_norm": 0.17495040595531464,
+      "learning_rate": 4.5479525433053205e-06,
+      "loss": 0.9362,
+      "step": 62450
+    },
+    {
+      "epoch": 0.4521270820213251,
+      "grad_norm": 0.25016602873802185,
+      "learning_rate": 4.547880156644734e-06,
+      "loss": 0.9354,
+      "step": 62460
+    },
+    {
+      "epoch": 0.4521994686819113,
+      "grad_norm": 0.1564486026763916,
+      "learning_rate": 4.547807769984148e-06,
+      "loss": 0.9244,
+      "step": 62470
+    },
+    {
+      "epoch": 0.4522718553424975,
+      "grad_norm": 0.1539490669965744,
+      "learning_rate": 4.547735383323561e-06,
+      "loss": 0.9209,
+      "step": 62480
+    },
+    {
+      "epoch": 0.45234424200308365,
+      "grad_norm": 0.17183518409729004,
+      "learning_rate": 4.547662996662976e-06,
+      "loss": 0.9361,
+      "step": 62490
+    },
+    {
+      "epoch": 0.4524166286636699,
+      "grad_norm": 0.17086255550384521,
+      "learning_rate": 4.547590610002389e-06,
+      "loss": 0.9142,
+      "step": 62500
+    },
+    {
+      "epoch": 0.45248901532425606,
+      "grad_norm": 0.16200555860996246,
+      "learning_rate": 4.547518223341803e-06,
+      "loss": 0.942,
+      "step": 62510
+    },
+    {
+      "epoch": 0.45256140198484224,
+      "grad_norm": 0.1565650850534439,
+      "learning_rate": 4.547445836681217e-06,
+      "loss": 0.938,
+      "step": 62520
+    },
+    {
+      "epoch": 0.4526337886454284,
+      "grad_norm": 0.17040525376796722,
+      "learning_rate": 4.547373450020631e-06,
+      "loss": 0.9279,
+      "step": 62530
+    },
+    {
+      "epoch": 0.4527061753060146,
+      "grad_norm": 0.17260707914829254,
+      "learning_rate": 4.547301063360045e-06,
+      "loss": 0.9246,
+      "step": 62540
+    },
+    {
+      "epoch": 0.45277856196660077,
+      "grad_norm": 0.16091448068618774,
+      "learning_rate": 4.547228676699458e-06,
+      "loss": 0.9301,
+      "step": 62550
+    },
+    {
+      "epoch": 0.452850948627187,
+      "grad_norm": 0.15594513714313507,
+      "learning_rate": 4.547156290038872e-06,
+      "loss": 0.9388,
+      "step": 62560
+    },
+    {
+      "epoch": 0.4529233352877732,
+      "grad_norm": 0.17561206221580505,
+      "learning_rate": 4.547083903378286e-06,
+      "loss": 0.9311,
+      "step": 62570
+    },
+    {
+      "epoch": 0.45299572194835935,
+      "grad_norm": 0.1688334196805954,
+      "learning_rate": 4.5470115167177e-06,
+      "loss": 0.9374,
+      "step": 62580
+    },
+    {
+      "epoch": 0.45306810860894553,
+      "grad_norm": 0.26716697216033936,
+      "learning_rate": 4.546939130057114e-06,
+      "loss": 0.9332,
+      "step": 62590
+    },
+    {
+      "epoch": 0.4531404952695317,
+      "grad_norm": 0.16558301448822021,
+      "learning_rate": 4.546866743396527e-06,
+      "loss": 0.9361,
+      "step": 62600
+    },
+    {
+      "epoch": 0.45321288193011794,
+      "grad_norm": 0.16215580701828003,
+      "learning_rate": 4.546794356735941e-06,
+      "loss": 0.944,
+      "step": 62610
+    },
+    {
+      "epoch": 0.4532852685907041,
+      "grad_norm": 0.17625215649604797,
+      "learning_rate": 4.546721970075355e-06,
+      "loss": 0.9408,
+      "step": 62620
+    },
+    {
+      "epoch": 0.4533576552512903,
+      "grad_norm": 0.1480870544910431,
+      "learning_rate": 4.546649583414769e-06,
+      "loss": 0.929,
+      "step": 62630
+    },
+    {
+      "epoch": 0.45343004191187647,
+      "grad_norm": 0.17280828952789307,
+      "learning_rate": 4.5465771967541826e-06,
+      "loss": 0.9177,
+      "step": 62640
+    },
+    {
+      "epoch": 0.45350242857246265,
+      "grad_norm": 0.1757204234600067,
+      "learning_rate": 4.546504810093596e-06,
+      "loss": 0.9396,
+      "step": 62650
+    },
+    {
+      "epoch": 0.4535748152330489,
+      "grad_norm": 0.16843682527542114,
+      "learning_rate": 4.546432423433011e-06,
+      "loss": 0.9311,
+      "step": 62660
+    },
+    {
+      "epoch": 0.45364720189363505,
+      "grad_norm": 0.14993290603160858,
+      "learning_rate": 4.546360036772424e-06,
+      "loss": 0.9257,
+      "step": 62670
+    },
+    {
+      "epoch": 0.45371958855422123,
+      "grad_norm": 0.1843591332435608,
+      "learning_rate": 4.546287650111838e-06,
+      "loss": 0.9305,
+      "step": 62680
+    },
+    {
+      "epoch": 0.4537919752148074,
+      "grad_norm": 0.1545378565788269,
+      "learning_rate": 4.5462152634512515e-06,
+      "loss": 0.93,
+      "step": 62690
+    },
+    {
+      "epoch": 0.4538643618753936,
+      "grad_norm": 0.1681130826473236,
+      "learning_rate": 4.546142876790665e-06,
+      "loss": 0.9191,
+      "step": 62700
+    },
+    {
+      "epoch": 0.4539367485359798,
+      "grad_norm": 0.15913967788219452,
+      "learning_rate": 4.546070490130079e-06,
+      "loss": 0.9356,
+      "step": 62710
+    },
+    {
+      "epoch": 0.454009135196566,
+      "grad_norm": 0.16639158129692078,
+      "learning_rate": 4.545998103469493e-06,
+      "loss": 0.9405,
+      "step": 62720
+    },
+    {
+      "epoch": 0.45408152185715217,
+      "grad_norm": 0.1884455382823944,
+      "learning_rate": 4.545925716808907e-06,
+      "loss": 0.931,
+      "step": 62730
+    },
+    {
+      "epoch": 0.45415390851773835,
+      "grad_norm": 0.15364967286586761,
+      "learning_rate": 4.54585333014832e-06,
+      "loss": 0.9278,
+      "step": 62740
+    },
+    {
+      "epoch": 0.4542262951783245,
+      "grad_norm": 0.15758123993873596,
+      "learning_rate": 4.545780943487734e-06,
+      "loss": 0.9337,
+      "step": 62750
+    },
+    {
+      "epoch": 0.4542986818389107,
+      "grad_norm": 0.16586291790008545,
+      "learning_rate": 4.545708556827148e-06,
+      "loss": 0.9298,
+      "step": 62760
+    },
+    {
+      "epoch": 0.45437106849949693,
+      "grad_norm": 0.15221761167049408,
+      "learning_rate": 4.545636170166562e-06,
+      "loss": 0.9273,
+      "step": 62770
+    },
+    {
+      "epoch": 0.4544434551600831,
+      "grad_norm": 0.16838103532791138,
+      "learning_rate": 4.545563783505976e-06,
+      "loss": 0.9418,
+      "step": 62780
+    },
+    {
+      "epoch": 0.4545158418206693,
+      "grad_norm": 0.15279430150985718,
+      "learning_rate": 4.545491396845389e-06,
+      "loss": 0.9251,
+      "step": 62790
+    },
+    {
+      "epoch": 0.45458822848125546,
+      "grad_norm": 0.161879301071167,
+      "learning_rate": 4.545419010184803e-06,
+      "loss": 0.9355,
+      "step": 62800
+    },
+    {
+      "epoch": 0.45466061514184164,
+      "grad_norm": 0.15122537314891815,
+      "learning_rate": 4.545346623524217e-06,
+      "loss": 0.9282,
+      "step": 62810
+    },
+    {
+      "epoch": 0.45473300180242787,
+      "grad_norm": 0.1683684140443802,
+      "learning_rate": 4.545274236863631e-06,
+      "loss": 0.9341,
+      "step": 62820
+    },
+    {
+      "epoch": 0.45480538846301405,
+      "grad_norm": 0.16994281113147736,
+      "learning_rate": 4.545201850203045e-06,
+      "loss": 0.9273,
+      "step": 62830
+    },
+    {
+      "epoch": 0.4548777751236002,
+      "grad_norm": 0.15986701846122742,
+      "learning_rate": 4.545129463542458e-06,
+      "loss": 0.9214,
+      "step": 62840
+    },
+    {
+      "epoch": 0.4549501617841864,
+      "grad_norm": 0.1651432365179062,
+      "learning_rate": 4.545057076881873e-06,
+      "loss": 0.932,
+      "step": 62850
+    },
+    {
+      "epoch": 0.4550225484447726,
+      "grad_norm": 0.1712128072977066,
+      "learning_rate": 4.544984690221286e-06,
+      "loss": 0.9326,
+      "step": 62860
+    },
+    {
+      "epoch": 0.4550949351053588,
+      "grad_norm": 0.17609210312366486,
+      "learning_rate": 4.5449123035607e-06,
+      "loss": 0.938,
+      "step": 62870
+    },
+    {
+      "epoch": 0.455167321765945,
+      "grad_norm": 0.17556455731391907,
+      "learning_rate": 4.544839916900114e-06,
+      "loss": 0.9311,
+      "step": 62880
+    },
+    {
+      "epoch": 0.45523970842653116,
+      "grad_norm": 0.1542162448167801,
+      "learning_rate": 4.544767530239528e-06,
+      "loss": 0.9262,
+      "step": 62890
+    },
+    {
+      "epoch": 0.45531209508711734,
+      "grad_norm": 0.21663711965084076,
+      "learning_rate": 4.544695143578942e-06,
+      "loss": 0.9301,
+      "step": 62900
+    },
+    {
+      "epoch": 0.4553844817477035,
+      "grad_norm": 0.1632581651210785,
+      "learning_rate": 4.544622756918355e-06,
+      "loss": 0.942,
+      "step": 62910
+    },
+    {
+      "epoch": 0.4554568684082897,
+      "grad_norm": 0.1683775782585144,
+      "learning_rate": 4.544550370257769e-06,
+      "loss": 0.9313,
+      "step": 62920
+    },
+    {
+      "epoch": 0.4555292550688759,
+      "grad_norm": 0.16265662014484406,
+      "learning_rate": 4.544477983597183e-06,
+      "loss": 0.9255,
+      "step": 62930
+    },
+    {
+      "epoch": 0.4556016417294621,
+      "grad_norm": 0.16888990998268127,
+      "learning_rate": 4.544405596936597e-06,
+      "loss": 0.9327,
+      "step": 62940
+    },
+    {
+      "epoch": 0.4556740283900483,
+      "grad_norm": 0.1617661714553833,
+      "learning_rate": 4.544333210276011e-06,
+      "loss": 0.9365,
+      "step": 62950
+    },
+    {
+      "epoch": 0.45574641505063446,
+      "grad_norm": 0.15065929293632507,
+      "learning_rate": 4.544260823615424e-06,
+      "loss": 0.9411,
+      "step": 62960
+    },
+    {
+      "epoch": 0.45581880171122063,
+      "grad_norm": 0.15729419887065887,
+      "learning_rate": 4.544188436954839e-06,
+      "loss": 0.9305,
+      "step": 62970
+    },
+    {
+      "epoch": 0.45589118837180687,
+      "grad_norm": 0.1598915308713913,
+      "learning_rate": 4.544116050294252e-06,
+      "loss": 0.9236,
+      "step": 62980
+    },
+    {
+      "epoch": 0.45596357503239304,
+      "grad_norm": 0.1573273241519928,
+      "learning_rate": 4.544043663633666e-06,
+      "loss": 0.9221,
+      "step": 62990
+    },
+    {
+      "epoch": 0.4560359616929792,
+      "grad_norm": 0.17503662407398224,
+      "learning_rate": 4.5439712769730795e-06,
+      "loss": 0.9322,
+      "step": 63000
+    },
+    {
+      "epoch": 0.4561083483535654,
+      "grad_norm": 0.16443364322185516,
+      "learning_rate": 4.543898890312494e-06,
+      "loss": 0.9312,
+      "step": 63010
+    },
+    {
+      "epoch": 0.4561807350141516,
+      "grad_norm": 0.15545983612537384,
+      "learning_rate": 4.543826503651908e-06,
+      "loss": 0.9267,
+      "step": 63020
+    },
+    {
+      "epoch": 0.4562531216747378,
+      "grad_norm": 0.25189727544784546,
+      "learning_rate": 4.543754116991321e-06,
+      "loss": 0.919,
+      "step": 63030
+    },
+    {
+      "epoch": 0.456325508335324,
+      "grad_norm": 0.17259828746318817,
+      "learning_rate": 4.543681730330735e-06,
+      "loss": 0.9431,
+      "step": 63040
+    },
+    {
+      "epoch": 0.45639789499591016,
+      "grad_norm": 0.1540062129497528,
+      "learning_rate": 4.543609343670149e-06,
+      "loss": 0.9239,
+      "step": 63050
+    },
+    {
+      "epoch": 0.45647028165649634,
+      "grad_norm": 0.16821375489234924,
+      "learning_rate": 4.543536957009563e-06,
+      "loss": 0.9303,
+      "step": 63060
+    },
+    {
+      "epoch": 0.4565426683170825,
+      "grad_norm": 0.15353234112262726,
+      "learning_rate": 4.5434645703489765e-06,
+      "loss": 0.9242,
+      "step": 63070
+    },
+    {
+      "epoch": 0.4566150549776687,
+      "grad_norm": 0.15900860726833344,
+      "learning_rate": 4.54339218368839e-06,
+      "loss": 0.9232,
+      "step": 63080
+    },
+    {
+      "epoch": 0.4566874416382549,
+      "grad_norm": 0.2517249882221222,
+      "learning_rate": 4.543319797027805e-06,
+      "loss": 0.9415,
+      "step": 63090
+    },
+    {
+      "epoch": 0.4567598282988411,
+      "grad_norm": 0.16736070811748505,
+      "learning_rate": 4.543247410367218e-06,
+      "loss": 0.9241,
+      "step": 63100
+    },
+    {
+      "epoch": 0.4568322149594273,
+      "grad_norm": 0.1665504276752472,
+      "learning_rate": 4.543175023706632e-06,
+      "loss": 0.9348,
+      "step": 63110
+    },
+    {
+      "epoch": 0.45690460162001345,
+      "grad_norm": 0.1781657338142395,
+      "learning_rate": 4.5431026370460455e-06,
+      "loss": 0.9379,
+      "step": 63120
+    },
+    {
+      "epoch": 0.4569769882805996,
+      "grad_norm": 0.15879395604133606,
+      "learning_rate": 4.54303025038546e-06,
+      "loss": 0.9426,
+      "step": 63130
+    },
+    {
+      "epoch": 0.45704937494118586,
+      "grad_norm": 0.1565670371055603,
+      "learning_rate": 4.5429578637248735e-06,
+      "loss": 0.9176,
+      "step": 63140
+    },
+    {
+      "epoch": 0.45712176160177204,
+      "grad_norm": 0.15761685371398926,
+      "learning_rate": 4.542885477064287e-06,
+      "loss": 0.9228,
+      "step": 63150
+    },
+    {
+      "epoch": 0.4571941482623582,
+      "grad_norm": 0.15627005696296692,
+      "learning_rate": 4.542813090403701e-06,
+      "loss": 0.9354,
+      "step": 63160
+    },
+    {
+      "epoch": 0.4572665349229444,
+      "grad_norm": 0.15798023343086243,
+      "learning_rate": 4.542740703743115e-06,
+      "loss": 0.9131,
+      "step": 63170
+    },
+    {
+      "epoch": 0.45733892158353057,
+      "grad_norm": 0.1624073088169098,
+      "learning_rate": 4.542668317082529e-06,
+      "loss": 0.927,
+      "step": 63180
+    },
+    {
+      "epoch": 0.4574113082441168,
+      "grad_norm": 0.15683312714099884,
+      "learning_rate": 4.5425959304219425e-06,
+      "loss": 0.9402,
+      "step": 63190
+    },
+    {
+      "epoch": 0.457483694904703,
+      "grad_norm": 0.18702444434165955,
+      "learning_rate": 4.542523543761356e-06,
+      "loss": 0.9368,
+      "step": 63200
+    },
+    {
+      "epoch": 0.45755608156528915,
+      "grad_norm": 0.15603706240653992,
+      "learning_rate": 4.54245115710077e-06,
+      "loss": 0.9298,
+      "step": 63210
+    },
+    {
+      "epoch": 0.45762846822587533,
+      "grad_norm": 0.15618003904819489,
+      "learning_rate": 4.542378770440183e-06,
+      "loss": 0.9393,
+      "step": 63220
+    },
+    {
+      "epoch": 0.4577008548864615,
+      "grad_norm": 0.1605190932750702,
+      "learning_rate": 4.542306383779597e-06,
+      "loss": 0.9349,
+      "step": 63230
+    },
+    {
+      "epoch": 0.4577732415470477,
+      "grad_norm": 0.16419483721256256,
+      "learning_rate": 4.542233997119011e-06,
+      "loss": 0.9356,
+      "step": 63240
+    },
+    {
+      "epoch": 0.4578456282076339,
+      "grad_norm": 0.15912272036075592,
+      "learning_rate": 4.542161610458425e-06,
+      "loss": 0.932,
+      "step": 63250
+    },
+    {
+      "epoch": 0.4579180148682201,
+      "grad_norm": 0.16210229694843292,
+      "learning_rate": 4.542089223797839e-06,
+      "loss": 0.9353,
+      "step": 63260
+    },
+    {
+      "epoch": 0.45799040152880627,
+      "grad_norm": 0.16142286360263824,
+      "learning_rate": 4.542016837137252e-06,
+      "loss": 0.9464,
+      "step": 63270
+    },
+    {
+      "epoch": 0.45806278818939244,
+      "grad_norm": 0.21197547018527985,
+      "learning_rate": 4.541944450476667e-06,
+      "loss": 0.9221,
+      "step": 63280
+    },
+    {
+      "epoch": 0.4581351748499786,
+      "grad_norm": 0.17035934329032898,
+      "learning_rate": 4.54187206381608e-06,
+      "loss": 0.9299,
+      "step": 63290
+    },
+    {
+      "epoch": 0.45820756151056485,
+      "grad_norm": 0.20072509348392487,
+      "learning_rate": 4.541799677155494e-06,
+      "loss": 0.9152,
+      "step": 63300
+    },
+    {
+      "epoch": 0.45827994817115103,
+      "grad_norm": 0.1753024309873581,
+      "learning_rate": 4.5417272904949075e-06,
+      "loss": 0.9106,
+      "step": 63310
+    },
+    {
+      "epoch": 0.4583523348317372,
+      "grad_norm": 0.16419456899166107,
+      "learning_rate": 4.541654903834322e-06,
+      "loss": 0.9305,
+      "step": 63320
+    },
+    {
+      "epoch": 0.4584247214923234,
+      "grad_norm": 0.15261687338352203,
+      "learning_rate": 4.541582517173736e-06,
+      "loss": 0.9305,
+      "step": 63330
+    },
+    {
+      "epoch": 0.45849710815290956,
+      "grad_norm": 0.16415435075759888,
+      "learning_rate": 4.541510130513149e-06,
+      "loss": 0.9342,
+      "step": 63340
+    },
+    {
+      "epoch": 0.4585694948134958,
+      "grad_norm": 0.15612776577472687,
+      "learning_rate": 4.541437743852563e-06,
+      "loss": 0.9125,
+      "step": 63350
+    },
+    {
+      "epoch": 0.45864188147408197,
+      "grad_norm": 0.1619674116373062,
+      "learning_rate": 4.541365357191977e-06,
+      "loss": 0.9358,
+      "step": 63360
+    },
+    {
+      "epoch": 0.45871426813466815,
+      "grad_norm": 0.1574641913175583,
+      "learning_rate": 4.541292970531391e-06,
+      "loss": 0.9322,
+      "step": 63370
+    },
+    {
+      "epoch": 0.4587866547952543,
+      "grad_norm": 0.15913856029510498,
+      "learning_rate": 4.5412205838708046e-06,
+      "loss": 0.9454,
+      "step": 63380
+    },
+    {
+      "epoch": 0.4588590414558405,
+      "grad_norm": 0.15667736530303955,
+      "learning_rate": 4.541148197210218e-06,
+      "loss": 0.9376,
+      "step": 63390
+    },
+    {
+      "epoch": 0.45893142811642673,
+      "grad_norm": 0.15261350572109222,
+      "learning_rate": 4.541075810549632e-06,
+      "loss": 0.9474,
+      "step": 63400
+    },
+    {
+      "epoch": 0.4590038147770129,
+      "grad_norm": 0.1693621575832367,
+      "learning_rate": 4.541003423889046e-06,
+      "loss": 0.9309,
+      "step": 63410
+    },
+    {
+      "epoch": 0.4590762014375991,
+      "grad_norm": 0.15610769391059875,
+      "learning_rate": 4.54093103722846e-06,
+      "loss": 0.9228,
+      "step": 63420
+    },
+    {
+      "epoch": 0.45914858809818526,
+      "grad_norm": 0.17140141129493713,
+      "learning_rate": 4.5408586505678735e-06,
+      "loss": 0.9351,
+      "step": 63430
+    },
+    {
+      "epoch": 0.45922097475877144,
+      "grad_norm": 0.16948135197162628,
+      "learning_rate": 4.540786263907287e-06,
+      "loss": 0.9241,
+      "step": 63440
+    },
+    {
+      "epoch": 0.4592933614193576,
+      "grad_norm": 0.15839575231075287,
+      "learning_rate": 4.5407138772467016e-06,
+      "loss": 0.9205,
+      "step": 63450
+    },
+    {
+      "epoch": 0.45936574807994385,
+      "grad_norm": 0.16708028316497803,
+      "learning_rate": 4.540641490586115e-06,
+      "loss": 0.9118,
+      "step": 63460
+    },
+    {
+      "epoch": 0.45943813474053,
+      "grad_norm": 0.16213096678256989,
+      "learning_rate": 4.540569103925529e-06,
+      "loss": 0.9338,
+      "step": 63470
+    },
+    {
+      "epoch": 0.4595105214011162,
+      "grad_norm": 0.16923314332962036,
+      "learning_rate": 4.540496717264942e-06,
+      "loss": 0.9167,
+      "step": 63480
+    },
+    {
+      "epoch": 0.4595829080617024,
+      "grad_norm": 0.15057291090488434,
+      "learning_rate": 4.540424330604357e-06,
+      "loss": 0.9275,
+      "step": 63490
+    },
+    {
+      "epoch": 0.45965529472228855,
+      "grad_norm": 0.15825194120407104,
+      "learning_rate": 4.5403519439437705e-06,
+      "loss": 0.9367,
+      "step": 63500
+    },
+    {
+      "epoch": 0.4597276813828748,
+      "grad_norm": 0.16979360580444336,
+      "learning_rate": 4.540279557283184e-06,
+      "loss": 0.9326,
+      "step": 63510
+    },
+    {
+      "epoch": 0.45980006804346096,
+      "grad_norm": 0.16070100665092468,
+      "learning_rate": 4.540207170622598e-06,
+      "loss": 0.9278,
+      "step": 63520
+    },
+    {
+      "epoch": 0.45987245470404714,
+      "grad_norm": 0.15810810029506683,
+      "learning_rate": 4.540134783962012e-06,
+      "loss": 0.9187,
+      "step": 63530
+    },
+    {
+      "epoch": 0.4599448413646333,
+      "grad_norm": 0.15783362090587616,
+      "learning_rate": 4.540062397301426e-06,
+      "loss": 0.9289,
+      "step": 63540
+    },
+    {
+      "epoch": 0.4600172280252195,
+      "grad_norm": 0.15892358124256134,
+      "learning_rate": 4.539990010640839e-06,
+      "loss": 0.9234,
+      "step": 63550
+    },
+    {
+      "epoch": 0.4600896146858057,
+      "grad_norm": 0.15859895944595337,
+      "learning_rate": 4.539917623980253e-06,
+      "loss": 0.9265,
+      "step": 63560
+    },
+    {
+      "epoch": 0.4601620013463919,
+      "grad_norm": 0.16993699967861176,
+      "learning_rate": 4.5398452373196675e-06,
+      "loss": 0.9481,
+      "step": 63570
+    },
+    {
+      "epoch": 0.4602343880069781,
+      "grad_norm": 0.1587960124015808,
+      "learning_rate": 4.539772850659081e-06,
+      "loss": 0.9308,
+      "step": 63580
+    },
+    {
+      "epoch": 0.46030677466756426,
+      "grad_norm": 0.17357853055000305,
+      "learning_rate": 4.539700463998495e-06,
+      "loss": 0.9403,
+      "step": 63590
+    },
+    {
+      "epoch": 0.46037916132815043,
+      "grad_norm": 0.172526553273201,
+      "learning_rate": 4.539628077337908e-06,
+      "loss": 0.9237,
+      "step": 63600
+    },
+    {
+      "epoch": 0.4604515479887366,
+      "grad_norm": 0.1549219787120819,
+      "learning_rate": 4.539555690677323e-06,
+      "loss": 0.9297,
+      "step": 63610
+    },
+    {
+      "epoch": 0.46052393464932284,
+      "grad_norm": 0.1645515263080597,
+      "learning_rate": 4.5394833040167364e-06,
+      "loss": 0.9213,
+      "step": 63620
+    },
+    {
+      "epoch": 0.460596321309909,
+      "grad_norm": 0.1633620709180832,
+      "learning_rate": 4.53941091735615e-06,
+      "loss": 0.9293,
+      "step": 63630
+    },
+    {
+      "epoch": 0.4606687079704952,
+      "grad_norm": 0.17372886836528778,
+      "learning_rate": 4.539338530695564e-06,
+      "loss": 0.9202,
+      "step": 63640
+    },
+    {
+      "epoch": 0.46074109463108137,
+      "grad_norm": 0.37556585669517517,
+      "learning_rate": 4.539266144034978e-06,
+      "loss": 0.9344,
+      "step": 63650
+    },
+    {
+      "epoch": 0.46081348129166755,
+      "grad_norm": 0.161403089761734,
+      "learning_rate": 4.539193757374392e-06,
+      "loss": 0.9468,
+      "step": 63660
+    },
+    {
+      "epoch": 0.4608858679522538,
+      "grad_norm": 0.1543300747871399,
+      "learning_rate": 4.539121370713805e-06,
+      "loss": 0.9162,
+      "step": 63670
+    },
+    {
+      "epoch": 0.46095825461283996,
+      "grad_norm": 0.2087777554988861,
+      "learning_rate": 4.539048984053219e-06,
+      "loss": 0.9278,
+      "step": 63680
+    },
+    {
+      "epoch": 0.46103064127342613,
+      "grad_norm": 0.15856918692588806,
+      "learning_rate": 4.5389765973926334e-06,
+      "loss": 0.9155,
+      "step": 63690
+    },
+    {
+      "epoch": 0.4611030279340123,
+      "grad_norm": 0.156173437833786,
+      "learning_rate": 4.538904210732047e-06,
+      "loss": 0.9355,
+      "step": 63700
+    },
+    {
+      "epoch": 0.4611754145945985,
+      "grad_norm": 0.1753517985343933,
+      "learning_rate": 4.538831824071461e-06,
+      "loss": 0.9267,
+      "step": 63710
+    },
+    {
+      "epoch": 0.4612478012551847,
+      "grad_norm": 0.18503955006599426,
+      "learning_rate": 4.538759437410874e-06,
+      "loss": 0.9342,
+      "step": 63720
+    },
+    {
+      "epoch": 0.4613201879157709,
+      "grad_norm": 0.1558404415845871,
+      "learning_rate": 4.538687050750289e-06,
+      "loss": 0.9214,
+      "step": 63730
+    },
+    {
+      "epoch": 0.4613925745763571,
+      "grad_norm": 0.154067724943161,
+      "learning_rate": 4.538614664089702e-06,
+      "loss": 0.9238,
+      "step": 63740
+    },
+    {
+      "epoch": 0.46146496123694325,
+      "grad_norm": 0.1731044501066208,
+      "learning_rate": 4.538542277429115e-06,
+      "loss": 0.9452,
+      "step": 63750
+    },
+    {
+      "epoch": 0.4615373478975294,
+      "grad_norm": 0.17740470170974731,
+      "learning_rate": 4.53846989076853e-06,
+      "loss": 0.931,
+      "step": 63760
+    },
+    {
+      "epoch": 0.4616097345581156,
+      "grad_norm": 0.15755429863929749,
+      "learning_rate": 4.538397504107943e-06,
+      "loss": 0.918,
+      "step": 63770
+    },
+    {
+      "epoch": 0.46168212121870184,
+      "grad_norm": 0.15609146654605865,
+      "learning_rate": 4.538325117447357e-06,
+      "loss": 0.9306,
+      "step": 63780
+    },
+    {
+      "epoch": 0.461754507879288,
+      "grad_norm": 0.17046941816806793,
+      "learning_rate": 4.5382527307867704e-06,
+      "loss": 0.927,
+      "step": 63790
+    },
+    {
+      "epoch": 0.4618268945398742,
+      "grad_norm": 0.1781318038702011,
+      "learning_rate": 4.538180344126185e-06,
+      "loss": 0.9184,
+      "step": 63800
+    },
+    {
+      "epoch": 0.46189928120046037,
+      "grad_norm": 0.16463392972946167,
+      "learning_rate": 4.5381079574655985e-06,
+      "loss": 0.929,
+      "step": 63810
+    },
+    {
+      "epoch": 0.46197166786104654,
+      "grad_norm": 0.1558876931667328,
+      "learning_rate": 4.538035570805012e-06,
+      "loss": 0.9194,
+      "step": 63820
+    },
+    {
+      "epoch": 0.4620440545216328,
+      "grad_norm": 0.17599675059318542,
+      "learning_rate": 4.537963184144426e-06,
+      "loss": 0.9254,
+      "step": 63830
+    },
+    {
+      "epoch": 0.46211644118221895,
+      "grad_norm": 0.1566462218761444,
+      "learning_rate": 4.53789079748384e-06,
+      "loss": 0.9412,
+      "step": 63840
+    },
+    {
+      "epoch": 0.46218882784280513,
+      "grad_norm": 0.15512123703956604,
+      "learning_rate": 4.537818410823254e-06,
+      "loss": 0.9367,
+      "step": 63850
+    },
+    {
+      "epoch": 0.4622612145033913,
+      "grad_norm": 0.16452373564243317,
+      "learning_rate": 4.5377460241626675e-06,
+      "loss": 0.9274,
+      "step": 63860
+    },
+    {
+      "epoch": 0.4623336011639775,
+      "grad_norm": 0.16838426887989044,
+      "learning_rate": 4.537673637502081e-06,
+      "loss": 0.9143,
+      "step": 63870
+    },
+    {
+      "epoch": 0.4624059878245637,
+      "grad_norm": 0.16255596280097961,
+      "learning_rate": 4.5376012508414955e-06,
+      "loss": 0.9174,
+      "step": 63880
+    },
+    {
+      "epoch": 0.4624783744851499,
+      "grad_norm": 0.18602770566940308,
+      "learning_rate": 4.537528864180909e-06,
+      "loss": 0.9149,
+      "step": 63890
+    },
+    {
+      "epoch": 0.46255076114573607,
+      "grad_norm": 0.19781062006950378,
+      "learning_rate": 4.537456477520323e-06,
+      "loss": 0.936,
+      "step": 63900
+    },
+    {
+      "epoch": 0.46262314780632224,
+      "grad_norm": 0.1478962004184723,
+      "learning_rate": 4.537384090859736e-06,
+      "loss": 0.9191,
+      "step": 63910
+    },
+    {
+      "epoch": 0.4626955344669084,
+      "grad_norm": 0.15507012605667114,
+      "learning_rate": 4.537311704199151e-06,
+      "loss": 0.9242,
+      "step": 63920
+    },
+    {
+      "epoch": 0.46276792112749465,
+      "grad_norm": 0.17668579518795013,
+      "learning_rate": 4.5372393175385645e-06,
+      "loss": 0.9339,
+      "step": 63930
+    },
+    {
+      "epoch": 0.46284030778808083,
+      "grad_norm": 0.14959698915481567,
+      "learning_rate": 4.537166930877978e-06,
+      "loss": 0.9114,
+      "step": 63940
+    },
+    {
+      "epoch": 0.462912694448667,
+      "grad_norm": 0.17925885319709778,
+      "learning_rate": 4.537094544217392e-06,
+      "loss": 0.9206,
+      "step": 63950
+    },
+    {
+      "epoch": 0.4629850811092532,
+      "grad_norm": 0.16243740916252136,
+      "learning_rate": 4.537022157556806e-06,
+      "loss": 0.9196,
+      "step": 63960
+    },
+    {
+      "epoch": 0.46305746776983936,
+      "grad_norm": 0.3166872560977936,
+      "learning_rate": 4.53694977089622e-06,
+      "loss": 0.9352,
+      "step": 63970
+    },
+    {
+      "epoch": 0.46312985443042554,
+      "grad_norm": 0.16058658063411713,
+      "learning_rate": 4.536877384235633e-06,
+      "loss": 0.9097,
+      "step": 63980
+    },
+    {
+      "epoch": 0.46320224109101177,
+      "grad_norm": 0.15642769634723663,
+      "learning_rate": 4.536804997575047e-06,
+      "loss": 0.9307,
+      "step": 63990
+    },
+    {
+      "epoch": 0.46327462775159794,
+      "grad_norm": 0.41400203108787537,
+      "learning_rate": 4.5367326109144615e-06,
+      "loss": 0.9321,
+      "step": 64000
+    },
+    {
+      "epoch": 0.4633470144121841,
+      "grad_norm": 0.17071636021137238,
+      "learning_rate": 4.536660224253875e-06,
+      "loss": 0.9323,
+      "step": 64010
+    },
+    {
+      "epoch": 0.4634194010727703,
+      "grad_norm": 0.16446176171302795,
+      "learning_rate": 4.536587837593289e-06,
+      "loss": 0.9172,
+      "step": 64020
+    },
+    {
+      "epoch": 0.4634917877333565,
+      "grad_norm": 0.17229726910591125,
+      "learning_rate": 4.536515450932702e-06,
+      "loss": 0.9246,
+      "step": 64030
+    },
+    {
+      "epoch": 0.4635641743939427,
+      "grad_norm": 0.14730413258075714,
+      "learning_rate": 4.536443064272116e-06,
+      "loss": 0.9353,
+      "step": 64040
+    },
+    {
+      "epoch": 0.4636365610545289,
+      "grad_norm": 0.1702919751405716,
+      "learning_rate": 4.53637067761153e-06,
+      "loss": 0.9239,
+      "step": 64050
+    },
+    {
+      "epoch": 0.46370894771511506,
+      "grad_norm": 0.15090477466583252,
+      "learning_rate": 4.536298290950944e-06,
+      "loss": 0.9189,
+      "step": 64060
+    },
+    {
+      "epoch": 0.46378133437570124,
+      "grad_norm": 0.15662406384944916,
+      "learning_rate": 4.536225904290358e-06,
+      "loss": 0.9161,
+      "step": 64070
+    },
+    {
+      "epoch": 0.4638537210362874,
+      "grad_norm": 0.15603668987751007,
+      "learning_rate": 4.536153517629771e-06,
+      "loss": 0.9194,
+      "step": 64080
+    },
+    {
+      "epoch": 0.46392610769687365,
+      "grad_norm": 0.16202674806118011,
+      "learning_rate": 4.536081130969186e-06,
+      "loss": 0.9402,
+      "step": 64090
+    },
+    {
+      "epoch": 0.4639984943574598,
+      "grad_norm": 0.2072402536869049,
+      "learning_rate": 4.536008744308599e-06,
+      "loss": 0.9153,
+      "step": 64100
+    },
+    {
+      "epoch": 0.464070881018046,
+      "grad_norm": 0.16444271802902222,
+      "learning_rate": 4.535936357648013e-06,
+      "loss": 0.9216,
+      "step": 64110
+    },
+    {
+      "epoch": 0.4641432676786322,
+      "grad_norm": 0.19051018357276917,
+      "learning_rate": 4.5358639709874266e-06,
+      "loss": 0.9233,
+      "step": 64120
+    },
+    {
+      "epoch": 0.46421565433921835,
+      "grad_norm": 0.1983797252178192,
+      "learning_rate": 4.535791584326841e-06,
+      "loss": 0.9345,
+      "step": 64130
+    },
+    {
+      "epoch": 0.46428804099980453,
+      "grad_norm": 0.17344747483730316,
+      "learning_rate": 4.535719197666255e-06,
+      "loss": 0.9311,
+      "step": 64140
+    },
+    {
+      "epoch": 0.46436042766039076,
+      "grad_norm": 0.16626109182834625,
+      "learning_rate": 4.535646811005668e-06,
+      "loss": 0.9188,
+      "step": 64150
+    },
+    {
+      "epoch": 0.46443281432097694,
+      "grad_norm": 0.162200927734375,
+      "learning_rate": 4.535574424345082e-06,
+      "loss": 0.9248,
+      "step": 64160
+    },
+    {
+      "epoch": 0.4645052009815631,
+      "grad_norm": 0.43572789430618286,
+      "learning_rate": 4.535502037684496e-06,
+      "loss": 0.9225,
+      "step": 64170
+    },
+    {
+      "epoch": 0.4645775876421493,
+      "grad_norm": 0.18060703575611115,
+      "learning_rate": 4.53542965102391e-06,
+      "loss": 0.9357,
+      "step": 64180
+    },
+    {
+      "epoch": 0.46464997430273547,
+      "grad_norm": 0.16320358216762543,
+      "learning_rate": 4.5353572643633236e-06,
+      "loss": 0.9227,
+      "step": 64190
+    },
+    {
+      "epoch": 0.4647223609633217,
+      "grad_norm": 0.1623474657535553,
+      "learning_rate": 4.535284877702737e-06,
+      "loss": 0.9375,
+      "step": 64200
+    },
+    {
+      "epoch": 0.4647947476239079,
+      "grad_norm": 0.1604037880897522,
+      "learning_rate": 4.535212491042152e-06,
+      "loss": 0.9255,
+      "step": 64210
+    },
+    {
+      "epoch": 0.46486713428449405,
+      "grad_norm": 0.2360948622226715,
+      "learning_rate": 4.535140104381565e-06,
+      "loss": 0.9329,
+      "step": 64220
+    },
+    {
+      "epoch": 0.46493952094508023,
+      "grad_norm": 0.17647242546081543,
+      "learning_rate": 4.535067717720979e-06,
+      "loss": 0.9377,
+      "step": 64230
+    },
+    {
+      "epoch": 0.4650119076056664,
+      "grad_norm": 0.1619800180196762,
+      "learning_rate": 4.5349953310603925e-06,
+      "loss": 0.9352,
+      "step": 64240
+    },
+    {
+      "epoch": 0.46508429426625264,
+      "grad_norm": 0.2056133896112442,
+      "learning_rate": 4.534922944399807e-06,
+      "loss": 0.9284,
+      "step": 64250
+    },
+    {
+      "epoch": 0.4651566809268388,
+      "grad_norm": 0.16904328763484955,
+      "learning_rate": 4.5348505577392206e-06,
+      "loss": 0.9397,
+      "step": 64260
+    },
+    {
+      "epoch": 0.465229067587425,
+      "grad_norm": 0.17163747549057007,
+      "learning_rate": 4.534778171078634e-06,
+      "loss": 0.9299,
+      "step": 64270
+    },
+    {
+      "epoch": 0.46530145424801117,
+      "grad_norm": 0.15540345013141632,
+      "learning_rate": 4.534705784418048e-06,
+      "loss": 0.9273,
+      "step": 64280
+    },
+    {
+      "epoch": 0.46537384090859735,
+      "grad_norm": 0.16602273285388947,
+      "learning_rate": 4.534633397757461e-06,
+      "loss": 0.9229,
+      "step": 64290
+    },
+    {
+      "epoch": 0.4654462275691835,
+      "grad_norm": 0.15529848635196686,
+      "learning_rate": 4.534561011096875e-06,
+      "loss": 0.9227,
+      "step": 64300
+    },
+    {
+      "epoch": 0.46551861422976976,
+      "grad_norm": 0.15982522070407867,
+      "learning_rate": 4.534488624436289e-06,
+      "loss": 0.9184,
+      "step": 64310
+    },
+    {
+      "epoch": 0.46559100089035593,
+      "grad_norm": 0.16728608310222626,
+      "learning_rate": 4.534416237775703e-06,
+      "loss": 0.9283,
+      "step": 64320
+    },
+    {
+      "epoch": 0.4656633875509421,
+      "grad_norm": 0.1668313890695572,
+      "learning_rate": 4.534343851115117e-06,
+      "loss": 0.9257,
+      "step": 64330
+    },
+    {
+      "epoch": 0.4657357742115283,
+      "grad_norm": 0.16521133482456207,
+      "learning_rate": 4.53427146445453e-06,
+      "loss": 0.9199,
+      "step": 64340
+    },
+    {
+      "epoch": 0.46580816087211446,
+      "grad_norm": 0.1640012413263321,
+      "learning_rate": 4.534199077793944e-06,
+      "loss": 0.9251,
+      "step": 64350
+    },
+    {
+      "epoch": 0.4658805475327007,
+      "grad_norm": 0.15635322034358978,
+      "learning_rate": 4.5341266911333584e-06,
+      "loss": 0.9295,
+      "step": 64360
+    },
+    {
+      "epoch": 0.46595293419328687,
+      "grad_norm": 0.1682094931602478,
+      "learning_rate": 4.534054304472772e-06,
+      "loss": 0.9379,
+      "step": 64370
+    },
+    {
+      "epoch": 0.46602532085387305,
+      "grad_norm": 0.1606939285993576,
+      "learning_rate": 4.533981917812186e-06,
+      "loss": 0.925,
+      "step": 64380
+    },
+    {
+      "epoch": 0.4660977075144592,
+      "grad_norm": 0.1753852218389511,
+      "learning_rate": 4.533909531151599e-06,
+      "loss": 0.9238,
+      "step": 64390
+    },
+    {
+      "epoch": 0.4661700941750454,
+      "grad_norm": 0.20732562243938446,
+      "learning_rate": 4.533837144491014e-06,
+      "loss": 0.9268,
+      "step": 64400
+    },
+    {
+      "epoch": 0.46624248083563163,
+      "grad_norm": 0.16475224494934082,
+      "learning_rate": 4.533764757830427e-06,
+      "loss": 0.9164,
+      "step": 64410
+    },
+    {
+      "epoch": 0.4663148674962178,
+      "grad_norm": 0.17399507761001587,
+      "learning_rate": 4.533692371169841e-06,
+      "loss": 0.9204,
+      "step": 64420
+    },
+    {
+      "epoch": 0.466387254156804,
+      "grad_norm": 0.2416888028383255,
+      "learning_rate": 4.533619984509255e-06,
+      "loss": 0.9399,
+      "step": 64430
+    },
+    {
+      "epoch": 0.46645964081739016,
+      "grad_norm": 0.1748601794242859,
+      "learning_rate": 4.533547597848669e-06,
+      "loss": 0.9227,
+      "step": 64440
+    },
+    {
+      "epoch": 0.46653202747797634,
+      "grad_norm": 0.15142974257469177,
+      "learning_rate": 4.533475211188083e-06,
+      "loss": 0.9286,
+      "step": 64450
+    },
+    {
+      "epoch": 0.4666044141385626,
+      "grad_norm": 0.14932338893413544,
+      "learning_rate": 4.533402824527496e-06,
+      "loss": 0.9318,
+      "step": 64460
+    },
+    {
+      "epoch": 0.46667680079914875,
+      "grad_norm": 0.16411228477954865,
+      "learning_rate": 4.53333043786691e-06,
+      "loss": 0.9183,
+      "step": 64470
+    },
+    {
+      "epoch": 0.4667491874597349,
+      "grad_norm": 0.16134792566299438,
+      "learning_rate": 4.533258051206324e-06,
+      "loss": 0.9311,
+      "step": 64480
+    },
+    {
+      "epoch": 0.4668215741203211,
+      "grad_norm": 0.15893922746181488,
+      "learning_rate": 4.533185664545738e-06,
+      "loss": 0.9197,
+      "step": 64490
+    },
+    {
+      "epoch": 0.4668939607809073,
+      "grad_norm": 0.17452390491962433,
+      "learning_rate": 4.533113277885152e-06,
+      "loss": 0.9372,
+      "step": 64500
+    },
+    {
+      "epoch": 0.46696634744149346,
+      "grad_norm": 0.16843093931674957,
+      "learning_rate": 4.533040891224565e-06,
+      "loss": 0.9192,
+      "step": 64510
+    },
+    {
+      "epoch": 0.4670387341020797,
+      "grad_norm": 0.15596753358840942,
+      "learning_rate": 4.53296850456398e-06,
+      "loss": 0.9322,
+      "step": 64520
+    },
+    {
+      "epoch": 0.46711112076266587,
+      "grad_norm": 0.1649407148361206,
+      "learning_rate": 4.532896117903393e-06,
+      "loss": 0.9248,
+      "step": 64530
+    },
+    {
+      "epoch": 0.46718350742325204,
+      "grad_norm": 0.19480155408382416,
+      "learning_rate": 4.532823731242807e-06,
+      "loss": 0.9253,
+      "step": 64540
+    },
+    {
+      "epoch": 0.4672558940838382,
+      "grad_norm": 0.1667308658361435,
+      "learning_rate": 4.5327513445822205e-06,
+      "loss": 0.9249,
+      "step": 64550
+    },
+    {
+      "epoch": 0.4673282807444244,
+      "grad_norm": 0.16336029767990112,
+      "learning_rate": 4.532678957921635e-06,
+      "loss": 0.924,
+      "step": 64560
+    },
+    {
+      "epoch": 0.46740066740501063,
+      "grad_norm": 0.1569215953350067,
+      "learning_rate": 4.532606571261049e-06,
+      "loss": 0.9127,
+      "step": 64570
+    },
+    {
+      "epoch": 0.4674730540655968,
+      "grad_norm": 0.16221965849399567,
+      "learning_rate": 4.532534184600462e-06,
+      "loss": 0.9228,
+      "step": 64580
+    },
+    {
+      "epoch": 0.467545440726183,
+      "grad_norm": 0.16710902750492096,
+      "learning_rate": 4.532461797939876e-06,
+      "loss": 0.918,
+      "step": 64590
+    },
+    {
+      "epoch": 0.46761782738676916,
+      "grad_norm": 0.17595240473747253,
+      "learning_rate": 4.53238941127929e-06,
+      "loss": 0.9273,
+      "step": 64600
+    },
+    {
+      "epoch": 0.46769021404735533,
+      "grad_norm": 0.16132394969463348,
+      "learning_rate": 4.532317024618704e-06,
+      "loss": 0.927,
+      "step": 64610
+    },
+    {
+      "epoch": 0.46776260070794157,
+      "grad_norm": 0.14544646441936493,
+      "learning_rate": 4.5322446379581175e-06,
+      "loss": 0.925,
+      "step": 64620
+    },
+    {
+      "epoch": 0.46783498736852774,
+      "grad_norm": 0.1635567545890808,
+      "learning_rate": 4.532172251297531e-06,
+      "loss": 0.9207,
+      "step": 64630
+    },
+    {
+      "epoch": 0.4679073740291139,
+      "grad_norm": 0.15619686245918274,
+      "learning_rate": 4.532099864636945e-06,
+      "loss": 0.914,
+      "step": 64640
+    },
+    {
+      "epoch": 0.4679797606897001,
+      "grad_norm": 0.17399942874908447,
+      "learning_rate": 4.532027477976359e-06,
+      "loss": 0.9301,
+      "step": 64650
+    },
+    {
+      "epoch": 0.4680521473502863,
+      "grad_norm": 0.15665535628795624,
+      "learning_rate": 4.531955091315773e-06,
+      "loss": 0.9231,
+      "step": 64660
+    },
+    {
+      "epoch": 0.46812453401087245,
+      "grad_norm": 0.16816446185112,
+      "learning_rate": 4.5318827046551865e-06,
+      "loss": 0.9334,
+      "step": 64670
+    },
+    {
+      "epoch": 0.4681969206714587,
+      "grad_norm": 0.16004996001720428,
+      "learning_rate": 4.5318103179946e-06,
+      "loss": 0.9201,
+      "step": 64680
+    },
+    {
+      "epoch": 0.46826930733204486,
+      "grad_norm": 0.15259157121181488,
+      "learning_rate": 4.5317379313340145e-06,
+      "loss": 0.9352,
+      "step": 64690
+    },
+    {
+      "epoch": 0.46834169399263104,
+      "grad_norm": 0.1640315055847168,
+      "learning_rate": 4.531665544673428e-06,
+      "loss": 0.9199,
+      "step": 64700
+    },
+    {
+      "epoch": 0.4684140806532172,
+      "grad_norm": 0.15679945051670074,
+      "learning_rate": 4.531593158012842e-06,
+      "loss": 0.9235,
+      "step": 64710
+    },
+    {
+      "epoch": 0.4684864673138034,
+      "grad_norm": 0.16989102959632874,
+      "learning_rate": 4.531520771352255e-06,
+      "loss": 0.9157,
+      "step": 64720
+    },
+    {
+      "epoch": 0.4685588539743896,
+      "grad_norm": 0.16276511549949646,
+      "learning_rate": 4.53144838469167e-06,
+      "loss": 0.9257,
+      "step": 64730
+    },
+    {
+      "epoch": 0.4686312406349758,
+      "grad_norm": 0.1543579399585724,
+      "learning_rate": 4.5313759980310835e-06,
+      "loss": 0.91,
+      "step": 64740
+    },
+    {
+      "epoch": 0.468703627295562,
+      "grad_norm": 0.1687450408935547,
+      "learning_rate": 4.531303611370497e-06,
+      "loss": 0.9423,
+      "step": 64750
+    },
+    {
+      "epoch": 0.46877601395614815,
+      "grad_norm": 0.15913762152194977,
+      "learning_rate": 4.531231224709911e-06,
+      "loss": 0.9296,
+      "step": 64760
+    },
+    {
+      "epoch": 0.46884840061673433,
+      "grad_norm": 0.18352152407169342,
+      "learning_rate": 4.531158838049325e-06,
+      "loss": 0.9307,
+      "step": 64770
+    },
+    {
+      "epoch": 0.46892078727732056,
+      "grad_norm": 0.16911835968494415,
+      "learning_rate": 4.531086451388739e-06,
+      "loss": 0.9296,
+      "step": 64780
+    },
+    {
+      "epoch": 0.46899317393790674,
+      "grad_norm": 0.17307941615581512,
+      "learning_rate": 4.531014064728152e-06,
+      "loss": 0.9198,
+      "step": 64790
+    },
+    {
+      "epoch": 0.4690655605984929,
+      "grad_norm": 0.16192714869976044,
+      "learning_rate": 4.530941678067566e-06,
+      "loss": 0.9313,
+      "step": 64800
+    },
+    {
+      "epoch": 0.4691379472590791,
+      "grad_norm": 0.16948965191841125,
+      "learning_rate": 4.53086929140698e-06,
+      "loss": 0.9317,
+      "step": 64810
+    },
+    {
+      "epoch": 0.46921033391966527,
+      "grad_norm": 0.18143318593502045,
+      "learning_rate": 4.530796904746393e-06,
+      "loss": 0.9153,
+      "step": 64820
+    },
+    {
+      "epoch": 0.46928272058025144,
+      "grad_norm": 0.15184900164604187,
+      "learning_rate": 4.530724518085807e-06,
+      "loss": 0.9259,
+      "step": 64830
+    },
+    {
+      "epoch": 0.4693551072408377,
+      "grad_norm": 0.23423556983470917,
+      "learning_rate": 4.530652131425221e-06,
+      "loss": 0.9186,
+      "step": 64840
+    },
+    {
+      "epoch": 0.46942749390142385,
+      "grad_norm": 0.15671993792057037,
+      "learning_rate": 4.530579744764635e-06,
+      "loss": 0.9295,
+      "step": 64850
+    },
+    {
+      "epoch": 0.46949988056201003,
+      "grad_norm": 0.1595701277256012,
+      "learning_rate": 4.5305073581040486e-06,
+      "loss": 0.917,
+      "step": 64860
+    },
+    {
+      "epoch": 0.4695722672225962,
+      "grad_norm": 0.1630265712738037,
+      "learning_rate": 4.530434971443462e-06,
+      "loss": 0.9234,
+      "step": 64870
+    },
+    {
+      "epoch": 0.4696446538831824,
+      "grad_norm": 0.16177038848400116,
+      "learning_rate": 4.530362584782877e-06,
+      "loss": 0.9365,
+      "step": 64880
+    },
+    {
+      "epoch": 0.4697170405437686,
+      "grad_norm": 0.16307857632637024,
+      "learning_rate": 4.53029019812229e-06,
+      "loss": 0.9355,
+      "step": 64890
+    },
+    {
+      "epoch": 0.4697894272043548,
+      "grad_norm": 0.15251775085926056,
+      "learning_rate": 4.530217811461704e-06,
+      "loss": 0.9335,
+      "step": 64900
+    },
+    {
+      "epoch": 0.46986181386494097,
+      "grad_norm": 0.172450453042984,
+      "learning_rate": 4.5301454248011175e-06,
+      "loss": 0.9269,
+      "step": 64910
+    },
+    {
+      "epoch": 0.46993420052552715,
+      "grad_norm": 0.1565367430448532,
+      "learning_rate": 4.530073038140532e-06,
+      "loss": 0.9327,
+      "step": 64920
+    },
+    {
+      "epoch": 0.4700065871861133,
+      "grad_norm": 0.3131227493286133,
+      "learning_rate": 4.5300006514799456e-06,
+      "loss": 0.9213,
+      "step": 64930
+    },
+    {
+      "epoch": 0.47007897384669955,
+      "grad_norm": 0.1655091792345047,
+      "learning_rate": 4.529928264819359e-06,
+      "loss": 0.9409,
+      "step": 64940
+    },
+    {
+      "epoch": 0.47015136050728573,
+      "grad_norm": 0.1536223590373993,
+      "learning_rate": 4.529855878158773e-06,
+      "loss": 0.92,
+      "step": 64950
+    },
+    {
+      "epoch": 0.4702237471678719,
+      "grad_norm": 0.1615433245897293,
+      "learning_rate": 4.529783491498187e-06,
+      "loss": 0.9194,
+      "step": 64960
+    },
+    {
+      "epoch": 0.4702961338284581,
+      "grad_norm": 0.15238776803016663,
+      "learning_rate": 4.529711104837601e-06,
+      "loss": 0.9202,
+      "step": 64970
+    },
+    {
+      "epoch": 0.47036852048904426,
+      "grad_norm": 0.15168006718158722,
+      "learning_rate": 4.5296387181770145e-06,
+      "loss": 0.9292,
+      "step": 64980
+    },
+    {
+      "epoch": 0.47044090714963044,
+      "grad_norm": 0.15964418649673462,
+      "learning_rate": 4.529566331516428e-06,
+      "loss": 0.9273,
+      "step": 64990
+    },
+    {
+      "epoch": 0.47051329381021667,
+      "grad_norm": 0.1611081212759018,
+      "learning_rate": 4.5294939448558426e-06,
+      "loss": 0.9226,
+      "step": 65000
+    },
+    {
+      "epoch": 0.47058568047080285,
+      "grad_norm": 0.168733611702919,
+      "learning_rate": 4.529421558195256e-06,
+      "loss": 0.9212,
+      "step": 65010
+    },
+    {
+      "epoch": 0.470658067131389,
+      "grad_norm": 0.17109130322933197,
+      "learning_rate": 4.52934917153467e-06,
+      "loss": 0.9261,
+      "step": 65020
+    },
+    {
+      "epoch": 0.4707304537919752,
+      "grad_norm": 0.16328299045562744,
+      "learning_rate": 4.529276784874083e-06,
+      "loss": 0.9293,
+      "step": 65030
+    },
+    {
+      "epoch": 0.4708028404525614,
+      "grad_norm": 0.16657480597496033,
+      "learning_rate": 4.529204398213498e-06,
+      "loss": 0.9165,
+      "step": 65040
+    },
+    {
+      "epoch": 0.4708752271131476,
+      "grad_norm": 0.1572638899087906,
+      "learning_rate": 4.5291320115529115e-06,
+      "loss": 0.9346,
+      "step": 65050
+    },
+    {
+      "epoch": 0.4709476137737338,
+      "grad_norm": 0.16349254548549652,
+      "learning_rate": 4.529059624892325e-06,
+      "loss": 0.9294,
+      "step": 65060
+    },
+    {
+      "epoch": 0.47102000043431996,
+      "grad_norm": 0.16738827526569366,
+      "learning_rate": 4.528987238231739e-06,
+      "loss": 0.9208,
+      "step": 65070
+    },
+    {
+      "epoch": 0.47109238709490614,
+      "grad_norm": 0.1925373077392578,
+      "learning_rate": 4.528914851571153e-06,
+      "loss": 0.9135,
+      "step": 65080
+    },
+    {
+      "epoch": 0.4711647737554923,
+      "grad_norm": 0.1586436629295349,
+      "learning_rate": 4.528842464910567e-06,
+      "loss": 0.929,
+      "step": 65090
+    },
+    {
+      "epoch": 0.47123716041607855,
+      "grad_norm": 0.15072402358055115,
+      "learning_rate": 4.5287700782499804e-06,
+      "loss": 0.9166,
+      "step": 65100
+    },
+    {
+      "epoch": 0.4713095470766647,
+      "grad_norm": 0.1562575101852417,
+      "learning_rate": 4.528697691589394e-06,
+      "loss": 0.9136,
+      "step": 65110
+    },
+    {
+      "epoch": 0.4713819337372509,
+      "grad_norm": 1.8733214139938354,
+      "learning_rate": 4.5286253049288085e-06,
+      "loss": 0.9095,
+      "step": 65120
+    },
+    {
+      "epoch": 0.4714543203978371,
+      "grad_norm": 0.1686108261346817,
+      "learning_rate": 4.528552918268222e-06,
+      "loss": 0.9196,
+      "step": 65130
+    },
+    {
+      "epoch": 0.47152670705842326,
+      "grad_norm": 0.17261675000190735,
+      "learning_rate": 4.528480531607636e-06,
+      "loss": 0.9238,
+      "step": 65140
+    },
+    {
+      "epoch": 0.4715990937190095,
+      "grad_norm": 0.19091512262821198,
+      "learning_rate": 4.528408144947049e-06,
+      "loss": 0.9248,
+      "step": 65150
+    },
+    {
+      "epoch": 0.47167148037959566,
+      "grad_norm": 0.1787658929824829,
+      "learning_rate": 4.528335758286464e-06,
+      "loss": 0.9335,
+      "step": 65160
+    },
+    {
+      "epoch": 0.47174386704018184,
+      "grad_norm": 0.1631409078836441,
+      "learning_rate": 4.5282633716258774e-06,
+      "loss": 0.9183,
+      "step": 65170
+    },
+    {
+      "epoch": 0.471816253700768,
+      "grad_norm": 0.16262884438037872,
+      "learning_rate": 4.528190984965291e-06,
+      "loss": 0.9278,
+      "step": 65180
+    },
+    {
+      "epoch": 0.4718886403613542,
+      "grad_norm": 0.16872116923332214,
+      "learning_rate": 4.528118598304705e-06,
+      "loss": 0.9303,
+      "step": 65190
+    },
+    {
+      "epoch": 0.47196102702194037,
+      "grad_norm": 0.17301858961582184,
+      "learning_rate": 4.528046211644119e-06,
+      "loss": 0.941,
+      "step": 65200
+    },
+    {
+      "epoch": 0.4720334136825266,
+      "grad_norm": 0.17453503608703613,
+      "learning_rate": 4.527973824983533e-06,
+      "loss": 0.9196,
+      "step": 65210
+    },
+    {
+      "epoch": 0.4721058003431128,
+      "grad_norm": 0.1696014106273651,
+      "learning_rate": 4.527901438322946e-06,
+      "loss": 0.9275,
+      "step": 65220
+    },
+    {
+      "epoch": 0.47217818700369896,
+      "grad_norm": 0.18995621800422668,
+      "learning_rate": 4.52782905166236e-06,
+      "loss": 0.9225,
+      "step": 65230
+    },
+    {
+      "epoch": 0.47225057366428513,
+      "grad_norm": 0.1921137422323227,
+      "learning_rate": 4.5277566650017744e-06,
+      "loss": 0.9353,
+      "step": 65240
+    },
+    {
+      "epoch": 0.4723229603248713,
+      "grad_norm": 0.15778936445713043,
+      "learning_rate": 4.527684278341188e-06,
+      "loss": 0.9268,
+      "step": 65250
+    },
+    {
+      "epoch": 0.47239534698545754,
+      "grad_norm": 0.17273864150047302,
+      "learning_rate": 4.527611891680602e-06,
+      "loss": 0.9206,
+      "step": 65260
+    },
+    {
+      "epoch": 0.4724677336460437,
+      "grad_norm": 0.16963228583335876,
+      "learning_rate": 4.527539505020015e-06,
+      "loss": 0.9257,
+      "step": 65270
+    },
+    {
+      "epoch": 0.4725401203066299,
+      "grad_norm": 0.15939907729625702,
+      "learning_rate": 4.527467118359429e-06,
+      "loss": 0.9197,
+      "step": 65280
+    },
+    {
+      "epoch": 0.4726125069672161,
+      "grad_norm": 0.16947107017040253,
+      "learning_rate": 4.527394731698843e-06,
+      "loss": 0.9286,
+      "step": 65290
+    },
+    {
+      "epoch": 0.47268489362780225,
+      "grad_norm": 0.2224172055721283,
+      "learning_rate": 4.527322345038257e-06,
+      "loss": 0.9384,
+      "step": 65300
+    },
+    {
+      "epoch": 0.4727572802883885,
+      "grad_norm": 0.16941924393177032,
+      "learning_rate": 4.527249958377671e-06,
+      "loss": 0.9309,
+      "step": 65310
+    },
+    {
+      "epoch": 0.47282966694897466,
+      "grad_norm": 0.17061692476272583,
+      "learning_rate": 4.527177571717084e-06,
+      "loss": 0.9248,
+      "step": 65320
+    },
+    {
+      "epoch": 0.47290205360956084,
+      "grad_norm": 0.18717791140079498,
+      "learning_rate": 4.527105185056499e-06,
+      "loss": 0.9196,
+      "step": 65330
+    },
+    {
+      "epoch": 0.472974440270147,
+      "grad_norm": 0.163266122341156,
+      "learning_rate": 4.5270327983959114e-06,
+      "loss": 0.9331,
+      "step": 65340
+    },
+    {
+      "epoch": 0.4730468269307332,
+      "grad_norm": 0.16074928641319275,
+      "learning_rate": 4.526960411735326e-06,
+      "loss": 0.9223,
+      "step": 65350
+    },
+    {
+      "epoch": 0.47311921359131937,
+      "grad_norm": 0.15289919078350067,
+      "learning_rate": 4.5268880250747395e-06,
+      "loss": 0.9275,
+      "step": 65360
+    },
+    {
+      "epoch": 0.4731916002519056,
+      "grad_norm": 0.15109410881996155,
+      "learning_rate": 4.526815638414153e-06,
+      "loss": 0.9392,
+      "step": 65370
+    },
+    {
+      "epoch": 0.4732639869124918,
+      "grad_norm": 0.1612526923418045,
+      "learning_rate": 4.526743251753567e-06,
+      "loss": 0.9317,
+      "step": 65380
+    },
+    {
+      "epoch": 0.47333637357307795,
+      "grad_norm": 0.1555621474981308,
+      "learning_rate": 4.526670865092981e-06,
+      "loss": 0.9258,
+      "step": 65390
+    },
+    {
+      "epoch": 0.4734087602336641,
+      "grad_norm": 0.1611545830965042,
+      "learning_rate": 4.526598478432395e-06,
+      "loss": 0.9283,
+      "step": 65400
+    },
+    {
+      "epoch": 0.4734811468942503,
+      "grad_norm": 0.15748530626296997,
+      "learning_rate": 4.5265260917718085e-06,
+      "loss": 0.928,
+      "step": 65410
+    },
+    {
+      "epoch": 0.47355353355483654,
+      "grad_norm": 0.15606287121772766,
+      "learning_rate": 4.526453705111222e-06,
+      "loss": 0.9109,
+      "step": 65420
+    },
+    {
+      "epoch": 0.4736259202154227,
+      "grad_norm": 0.15752071142196655,
+      "learning_rate": 4.526381318450636e-06,
+      "loss": 0.9277,
+      "step": 65430
+    },
+    {
+      "epoch": 0.4736983068760089,
+      "grad_norm": 0.15977409482002258,
+      "learning_rate": 4.52630893179005e-06,
+      "loss": 0.9256,
+      "step": 65440
+    },
+    {
+      "epoch": 0.47377069353659507,
+      "grad_norm": 0.18297438323497772,
+      "learning_rate": 4.526236545129464e-06,
+      "loss": 0.9271,
+      "step": 65450
+    },
+    {
+      "epoch": 0.47384308019718124,
+      "grad_norm": 0.16020241379737854,
+      "learning_rate": 4.526164158468877e-06,
+      "loss": 0.9215,
+      "step": 65460
+    },
+    {
+      "epoch": 0.4739154668577675,
+      "grad_norm": 0.16484937071800232,
+      "learning_rate": 4.526091771808291e-06,
+      "loss": 0.9252,
+      "step": 65470
+    },
+    {
+      "epoch": 0.47398785351835365,
+      "grad_norm": 0.15512751042842865,
+      "learning_rate": 4.5260193851477055e-06,
+      "loss": 0.9287,
+      "step": 65480
+    },
+    {
+      "epoch": 0.47406024017893983,
+      "grad_norm": 0.17977750301361084,
+      "learning_rate": 4.525946998487119e-06,
+      "loss": 0.9197,
+      "step": 65490
+    },
+    {
+      "epoch": 0.474132626839526,
+      "grad_norm": 0.17437465488910675,
+      "learning_rate": 4.525874611826533e-06,
+      "loss": 0.9398,
+      "step": 65500
+    },
+    {
+      "epoch": 0.4742050135001122,
+      "grad_norm": 0.1768895983695984,
+      "learning_rate": 4.525802225165946e-06,
+      "loss": 0.9288,
+      "step": 65510
+    },
+    {
+      "epoch": 0.47427740016069836,
+      "grad_norm": 0.15914186835289001,
+      "learning_rate": 4.525729838505361e-06,
+      "loss": 0.9102,
+      "step": 65520
+    },
+    {
+      "epoch": 0.4743497868212846,
+      "grad_norm": 0.1690351516008377,
+      "learning_rate": 4.525657451844774e-06,
+      "loss": 0.9324,
+      "step": 65530
+    },
+    {
+      "epoch": 0.47442217348187077,
+      "grad_norm": 0.15182551741600037,
+      "learning_rate": 4.525585065184188e-06,
+      "loss": 0.9163,
+      "step": 65540
+    },
+    {
+      "epoch": 0.47449456014245694,
+      "grad_norm": 0.17388121783733368,
+      "learning_rate": 4.525512678523602e-06,
+      "loss": 0.9364,
+      "step": 65550
+    },
+    {
+      "epoch": 0.4745669468030431,
+      "grad_norm": 0.1595803201198578,
+      "learning_rate": 4.525440291863016e-06,
+      "loss": 0.934,
+      "step": 65560
+    },
+    {
+      "epoch": 0.4746393334636293,
+      "grad_norm": 0.15675079822540283,
+      "learning_rate": 4.52536790520243e-06,
+      "loss": 0.9422,
+      "step": 65570
+    },
+    {
+      "epoch": 0.47471172012421553,
+      "grad_norm": 0.37786662578582764,
+      "learning_rate": 4.525295518541843e-06,
+      "loss": 0.9165,
+      "step": 65580
+    },
+    {
+      "epoch": 0.4747841067848017,
+      "grad_norm": 0.16655652225017548,
+      "learning_rate": 4.525223131881257e-06,
+      "loss": 0.9209,
+      "step": 65590
+    },
+    {
+      "epoch": 0.4748564934453879,
+      "grad_norm": 0.18225440382957458,
+      "learning_rate": 4.525150745220671e-06,
+      "loss": 0.9251,
+      "step": 65600
+    },
+    {
+      "epoch": 0.47492888010597406,
+      "grad_norm": 0.17587362229824066,
+      "learning_rate": 4.525078358560085e-06,
+      "loss": 0.9156,
+      "step": 65610
+    },
+    {
+      "epoch": 0.47500126676656024,
+      "grad_norm": 0.1652805507183075,
+      "learning_rate": 4.525005971899499e-06,
+      "loss": 0.9384,
+      "step": 65620
+    },
+    {
+      "epoch": 0.47507365342714647,
+      "grad_norm": 0.19912905991077423,
+      "learning_rate": 4.524933585238912e-06,
+      "loss": 0.9288,
+      "step": 65630
+    },
+    {
+      "epoch": 0.47514604008773265,
+      "grad_norm": 0.16045509278774261,
+      "learning_rate": 4.524861198578327e-06,
+      "loss": 0.9382,
+      "step": 65640
+    },
+    {
+      "epoch": 0.4752184267483188,
+      "grad_norm": 0.15262702107429504,
+      "learning_rate": 4.52478881191774e-06,
+      "loss": 0.9139,
+      "step": 65650
+    },
+    {
+      "epoch": 0.475290813408905,
+      "grad_norm": 0.14551232755184174,
+      "learning_rate": 4.524716425257154e-06,
+      "loss": 0.9337,
+      "step": 65660
+    },
+    {
+      "epoch": 0.4753632000694912,
+      "grad_norm": 0.1558513045310974,
+      "learning_rate": 4.5246440385965676e-06,
+      "loss": 0.923,
+      "step": 65670
+    },
+    {
+      "epoch": 0.4754355867300774,
+      "grad_norm": 0.15776222944259644,
+      "learning_rate": 4.524571651935982e-06,
+      "loss": 0.9251,
+      "step": 65680
+    },
+    {
+      "epoch": 0.4755079733906636,
+      "grad_norm": 0.15665759146213531,
+      "learning_rate": 4.524499265275396e-06,
+      "loss": 0.9342,
+      "step": 65690
+    },
+    {
+      "epoch": 0.47558036005124976,
+      "grad_norm": 0.16208912432193756,
+      "learning_rate": 4.524426878614809e-06,
+      "loss": 0.9177,
+      "step": 65700
+    },
+    {
+      "epoch": 0.47565274671183594,
+      "grad_norm": 0.15631617605686188,
+      "learning_rate": 4.524354491954223e-06,
+      "loss": 0.9125,
+      "step": 65710
+    },
+    {
+      "epoch": 0.4757251333724221,
+      "grad_norm": 0.14883607625961304,
+      "learning_rate": 4.524282105293637e-06,
+      "loss": 0.9268,
+      "step": 65720
+    },
+    {
+      "epoch": 0.4757975200330083,
+      "grad_norm": 0.1741674840450287,
+      "learning_rate": 4.524209718633051e-06,
+      "loss": 0.9362,
+      "step": 65730
+    },
+    {
+      "epoch": 0.4758699066935945,
+      "grad_norm": 0.17127837240695953,
+      "learning_rate": 4.5241373319724646e-06,
+      "loss": 0.9356,
+      "step": 65740
+    },
+    {
+      "epoch": 0.4759422933541807,
+      "grad_norm": 0.15458397567272186,
+      "learning_rate": 4.524064945311878e-06,
+      "loss": 0.9194,
+      "step": 65750
+    },
+    {
+      "epoch": 0.4760146800147669,
+      "grad_norm": 0.17374902963638306,
+      "learning_rate": 4.523992558651293e-06,
+      "loss": 0.9327,
+      "step": 65760
+    },
+    {
+      "epoch": 0.47608706667535305,
+      "grad_norm": 0.16826695203781128,
+      "learning_rate": 4.523920171990706e-06,
+      "loss": 0.9353,
+      "step": 65770
+    },
+    {
+      "epoch": 0.47615945333593923,
+      "grad_norm": 0.17883165180683136,
+      "learning_rate": 4.52384778533012e-06,
+      "loss": 0.9363,
+      "step": 65780
+    },
+    {
+      "epoch": 0.47623183999652546,
+      "grad_norm": 0.15835782885551453,
+      "learning_rate": 4.5237753986695335e-06,
+      "loss": 0.9221,
+      "step": 65790
+    },
+    {
+      "epoch": 0.47630422665711164,
+      "grad_norm": 0.1708236187696457,
+      "learning_rate": 4.523703012008948e-06,
+      "loss": 0.932,
+      "step": 65800
+    },
+    {
+      "epoch": 0.4763766133176978,
+      "grad_norm": 0.15170232951641083,
+      "learning_rate": 4.5236306253483616e-06,
+      "loss": 0.9146,
+      "step": 65810
+    },
+    {
+      "epoch": 0.476448999978284,
+      "grad_norm": 0.23991255462169647,
+      "learning_rate": 4.523558238687775e-06,
+      "loss": 0.9293,
+      "step": 65820
+    },
+    {
+      "epoch": 0.47652138663887017,
+      "grad_norm": 0.15470875799655914,
+      "learning_rate": 4.523485852027189e-06,
+      "loss": 0.9262,
+      "step": 65830
+    },
+    {
+      "epoch": 0.4765937732994564,
+      "grad_norm": 0.1519736647605896,
+      "learning_rate": 4.523413465366603e-06,
+      "loss": 0.9267,
+      "step": 65840
+    },
+    {
+      "epoch": 0.4766661599600426,
+      "grad_norm": 0.15895114839076996,
+      "learning_rate": 4.523341078706017e-06,
+      "loss": 0.9305,
+      "step": 65850
+    },
+    {
+      "epoch": 0.47673854662062876,
+      "grad_norm": 0.21327605843544006,
+      "learning_rate": 4.5232686920454305e-06,
+      "loss": 0.9106,
+      "step": 65860
+    },
+    {
+      "epoch": 0.47681093328121493,
+      "grad_norm": 0.15450114011764526,
+      "learning_rate": 4.523196305384844e-06,
+      "loss": 0.9167,
+      "step": 65870
+    },
+    {
+      "epoch": 0.4768833199418011,
+      "grad_norm": 0.17385606467723846,
+      "learning_rate": 4.523123918724258e-06,
+      "loss": 0.9264,
+      "step": 65880
+    },
+    {
+      "epoch": 0.4769557066023873,
+      "grad_norm": 0.15806423127651215,
+      "learning_rate": 4.523051532063671e-06,
+      "loss": 0.9311,
+      "step": 65890
+    },
+    {
+      "epoch": 0.4770280932629735,
+      "grad_norm": 0.15671569108963013,
+      "learning_rate": 4.522979145403085e-06,
+      "loss": 0.9292,
+      "step": 65900
+    },
+    {
+      "epoch": 0.4771004799235597,
+      "grad_norm": 0.18585889041423798,
+      "learning_rate": 4.5229067587424994e-06,
+      "loss": 0.9216,
+      "step": 65910
+    },
+    {
+      "epoch": 0.47717286658414587,
+      "grad_norm": 0.1866776943206787,
+      "learning_rate": 4.522834372081913e-06,
+      "loss": 0.9221,
+      "step": 65920
+    },
+    {
+      "epoch": 0.47724525324473205,
+      "grad_norm": 0.1546396166086197,
+      "learning_rate": 4.522761985421327e-06,
+      "loss": 0.9335,
+      "step": 65930
+    },
+    {
+      "epoch": 0.4773176399053182,
+      "grad_norm": 0.16696631908416748,
+      "learning_rate": 4.52268959876074e-06,
+      "loss": 0.9369,
+      "step": 65940
+    },
+    {
+      "epoch": 0.47739002656590446,
+      "grad_norm": 0.1489332616329193,
+      "learning_rate": 4.522617212100155e-06,
+      "loss": 0.9173,
+      "step": 65950
+    },
+    {
+      "epoch": 0.47746241322649063,
+      "grad_norm": 0.16513670980930328,
+      "learning_rate": 4.522544825439568e-06,
+      "loss": 0.9357,
+      "step": 65960
+    },
+    {
+      "epoch": 0.4775347998870768,
+      "grad_norm": 0.16803127527236938,
+      "learning_rate": 4.522472438778982e-06,
+      "loss": 0.9284,
+      "step": 65970
+    },
+    {
+      "epoch": 0.477607186547663,
+      "grad_norm": 0.16002456843852997,
+      "learning_rate": 4.522400052118396e-06,
+      "loss": 0.9268,
+      "step": 65980
+    },
+    {
+      "epoch": 0.47767957320824916,
+      "grad_norm": 0.17356227338314056,
+      "learning_rate": 4.52232766545781e-06,
+      "loss": 0.9188,
+      "step": 65990
+    },
+    {
+      "epoch": 0.4777519598688354,
+      "grad_norm": 0.15143811702728271,
+      "learning_rate": 4.522255278797224e-06,
+      "loss": 0.925,
+      "step": 66000
+    },
+    {
+      "epoch": 0.4778243465294216,
+      "grad_norm": 0.16914960741996765,
+      "learning_rate": 4.522182892136637e-06,
+      "loss": 0.9254,
+      "step": 66010
+    },
+    {
+      "epoch": 0.47789673319000775,
+      "grad_norm": 0.18097072839736938,
+      "learning_rate": 4.522110505476051e-06,
+      "loss": 0.9262,
+      "step": 66020
+    },
+    {
+      "epoch": 0.4779691198505939,
+      "grad_norm": 0.15266379714012146,
+      "learning_rate": 4.522038118815465e-06,
+      "loss": 0.9339,
+      "step": 66030
+    },
+    {
+      "epoch": 0.4780415065111801,
+      "grad_norm": 0.16532346606254578,
+      "learning_rate": 4.521965732154879e-06,
+      "loss": 0.9308,
+      "step": 66040
+    },
+    {
+      "epoch": 0.4781138931717663,
+      "grad_norm": 0.15536561608314514,
+      "learning_rate": 4.521893345494293e-06,
+      "loss": 0.9382,
+      "step": 66050
+    },
+    {
+      "epoch": 0.4781862798323525,
+      "grad_norm": 0.17557422816753387,
+      "learning_rate": 4.521820958833706e-06,
+      "loss": 0.9254,
+      "step": 66060
+    },
+    {
+      "epoch": 0.4782586664929387,
+      "grad_norm": 0.1694338172674179,
+      "learning_rate": 4.52174857217312e-06,
+      "loss": 0.9163,
+      "step": 66070
+    },
+    {
+      "epoch": 0.47833105315352487,
+      "grad_norm": 0.1753464937210083,
+      "learning_rate": 4.521676185512534e-06,
+      "loss": 0.9185,
+      "step": 66080
+    },
+    {
+      "epoch": 0.47840343981411104,
+      "grad_norm": 0.17190249264240265,
+      "learning_rate": 4.521603798851948e-06,
+      "loss": 0.9321,
+      "step": 66090
+    },
+    {
+      "epoch": 0.4784758264746972,
+      "grad_norm": 0.19599749147891998,
+      "learning_rate": 4.5215314121913615e-06,
+      "loss": 0.9146,
+      "step": 66100
+    },
+    {
+      "epoch": 0.47854821313528345,
+      "grad_norm": 0.15936939418315887,
+      "learning_rate": 4.521459025530775e-06,
+      "loss": 0.9135,
+      "step": 66110
+    },
+    {
+      "epoch": 0.47862059979586963,
+      "grad_norm": 0.1538030058145523,
+      "learning_rate": 4.52138663887019e-06,
+      "loss": 0.9398,
+      "step": 66120
+    },
+    {
+      "epoch": 0.4786929864564558,
+      "grad_norm": 0.15321306884288788,
+      "learning_rate": 4.521314252209603e-06,
+      "loss": 0.9364,
+      "step": 66130
+    },
+    {
+      "epoch": 0.478765373117042,
+      "grad_norm": 0.17162488400936127,
+      "learning_rate": 4.521241865549017e-06,
+      "loss": 0.9273,
+      "step": 66140
+    },
+    {
+      "epoch": 0.47883775977762816,
+      "grad_norm": 0.14776068925857544,
+      "learning_rate": 4.5211694788884305e-06,
+      "loss": 0.9183,
+      "step": 66150
+    },
+    {
+      "epoch": 0.4789101464382144,
+      "grad_norm": 0.16864345967769623,
+      "learning_rate": 4.521097092227845e-06,
+      "loss": 0.9167,
+      "step": 66160
+    },
+    {
+      "epoch": 0.47898253309880057,
+      "grad_norm": 0.1763739287853241,
+      "learning_rate": 4.5210247055672585e-06,
+      "loss": 0.9372,
+      "step": 66170
+    },
+    {
+      "epoch": 0.47905491975938674,
+      "grad_norm": 0.1718112826347351,
+      "learning_rate": 4.520952318906672e-06,
+      "loss": 0.9237,
+      "step": 66180
+    },
+    {
+      "epoch": 0.4791273064199729,
+      "grad_norm": 0.1696150302886963,
+      "learning_rate": 4.520879932246086e-06,
+      "loss": 0.9025,
+      "step": 66190
+    },
+    {
+      "epoch": 0.4791996930805591,
+      "grad_norm": 0.15991909801959991,
+      "learning_rate": 4.5208075455855e-06,
+      "loss": 0.9252,
+      "step": 66200
+    },
+    {
+      "epoch": 0.47927207974114533,
+      "grad_norm": 0.15851566195487976,
+      "learning_rate": 4.520735158924914e-06,
+      "loss": 0.9346,
+      "step": 66210
+    },
+    {
+      "epoch": 0.4793444664017315,
+      "grad_norm": 0.1567084640264511,
+      "learning_rate": 4.5206627722643275e-06,
+      "loss": 0.9178,
+      "step": 66220
+    },
+    {
+      "epoch": 0.4794168530623177,
+      "grad_norm": 0.20427605509757996,
+      "learning_rate": 4.520590385603741e-06,
+      "loss": 0.9117,
+      "step": 66230
+    },
+    {
+      "epoch": 0.47948923972290386,
+      "grad_norm": 0.17024697363376617,
+      "learning_rate": 4.5205179989431555e-06,
+      "loss": 0.9149,
+      "step": 66240
+    },
+    {
+      "epoch": 0.47956162638349004,
+      "grad_norm": 0.15808874368667603,
+      "learning_rate": 4.520445612282569e-06,
+      "loss": 0.9303,
+      "step": 66250
+    },
+    {
+      "epoch": 0.4796340130440762,
+      "grad_norm": 0.15470923483371735,
+      "learning_rate": 4.520373225621983e-06,
+      "loss": 0.9306,
+      "step": 66260
+    },
+    {
+      "epoch": 0.47970639970466245,
+      "grad_norm": 0.15453216433525085,
+      "learning_rate": 4.520300838961396e-06,
+      "loss": 0.9206,
+      "step": 66270
+    },
+    {
+      "epoch": 0.4797787863652486,
+      "grad_norm": 0.15643690526485443,
+      "learning_rate": 4.520228452300811e-06,
+      "loss": 0.9228,
+      "step": 66280
+    },
+    {
+      "epoch": 0.4798511730258348,
+      "grad_norm": 0.16580794751644135,
+      "learning_rate": 4.5201560656402245e-06,
+      "loss": 0.9398,
+      "step": 66290
+    },
+    {
+      "epoch": 0.479923559686421,
+      "grad_norm": 0.15789580345153809,
+      "learning_rate": 4.520083678979638e-06,
+      "loss": 0.9263,
+      "step": 66300
+    },
+    {
+      "epoch": 0.47999594634700715,
+      "grad_norm": 0.1554155945777893,
+      "learning_rate": 4.520011292319052e-06,
+      "loss": 0.9264,
+      "step": 66310
+    },
+    {
+      "epoch": 0.4800683330075934,
+      "grad_norm": 0.15116317570209503,
+      "learning_rate": 4.519938905658466e-06,
+      "loss": 0.9056,
+      "step": 66320
+    },
+    {
+      "epoch": 0.48014071966817956,
+      "grad_norm": 0.1687956303358078,
+      "learning_rate": 4.51986651899788e-06,
+      "loss": 0.9263,
+      "step": 66330
+    },
+    {
+      "epoch": 0.48021310632876574,
+      "grad_norm": 0.16320586204528809,
+      "learning_rate": 4.519794132337293e-06,
+      "loss": 0.9258,
+      "step": 66340
+    },
+    {
+      "epoch": 0.4802854929893519,
+      "grad_norm": 0.16672076284885406,
+      "learning_rate": 4.519721745676707e-06,
+      "loss": 0.9144,
+      "step": 66350
+    },
+    {
+      "epoch": 0.4803578796499381,
+      "grad_norm": 0.1586775779724121,
+      "learning_rate": 4.5196493590161215e-06,
+      "loss": 0.9242,
+      "step": 66360
+    },
+    {
+      "epoch": 0.4804302663105243,
+      "grad_norm": 0.15324345231056213,
+      "learning_rate": 4.519576972355535e-06,
+      "loss": 0.9235,
+      "step": 66370
+    },
+    {
+      "epoch": 0.4805026529711105,
+      "grad_norm": 0.1635131537914276,
+      "learning_rate": 4.519504585694949e-06,
+      "loss": 0.9354,
+      "step": 66380
+    },
+    {
+      "epoch": 0.4805750396316967,
+      "grad_norm": 0.16541409492492676,
+      "learning_rate": 4.519432199034362e-06,
+      "loss": 0.9113,
+      "step": 66390
+    },
+    {
+      "epoch": 0.48064742629228285,
+      "grad_norm": 0.1589830070734024,
+      "learning_rate": 4.519359812373776e-06,
+      "loss": 0.9271,
+      "step": 66400
+    },
+    {
+      "epoch": 0.48071981295286903,
+      "grad_norm": 0.1620132029056549,
+      "learning_rate": 4.5192874257131896e-06,
+      "loss": 0.9318,
+      "step": 66410
+    },
+    {
+      "epoch": 0.4807921996134552,
+      "grad_norm": 0.1495799869298935,
+      "learning_rate": 4.519215039052603e-06,
+      "loss": 0.908,
+      "step": 66420
+    },
+    {
+      "epoch": 0.48086458627404144,
+      "grad_norm": 0.17354585230350494,
+      "learning_rate": 4.519142652392018e-06,
+      "loss": 0.9334,
+      "step": 66430
+    },
+    {
+      "epoch": 0.4809369729346276,
+      "grad_norm": 0.18685661256313324,
+      "learning_rate": 4.519070265731431e-06,
+      "loss": 0.9207,
+      "step": 66440
+    },
+    {
+      "epoch": 0.4810093595952138,
+      "grad_norm": 0.16093280911445618,
+      "learning_rate": 4.518997879070845e-06,
+      "loss": 0.9236,
+      "step": 66450
+    },
+    {
+      "epoch": 0.48108174625579997,
+      "grad_norm": 0.15578316152095795,
+      "learning_rate": 4.5189254924102585e-06,
+      "loss": 0.918,
+      "step": 66460
+    },
+    {
+      "epoch": 0.48115413291638615,
+      "grad_norm": 0.1621652990579605,
+      "learning_rate": 4.518853105749673e-06,
+      "loss": 0.9276,
+      "step": 66470
+    },
+    {
+      "epoch": 0.4812265195769724,
+      "grad_norm": 0.16943508386611938,
+      "learning_rate": 4.5187807190890866e-06,
+      "loss": 0.9465,
+      "step": 66480
+    },
+    {
+      "epoch": 0.48129890623755855,
+      "grad_norm": 0.17045237123966217,
+      "learning_rate": 4.5187083324285e-06,
+      "loss": 0.9317,
+      "step": 66490
+    },
+    {
+      "epoch": 0.48137129289814473,
+      "grad_norm": 0.16214360296726227,
+      "learning_rate": 4.518635945767914e-06,
+      "loss": 0.9179,
+      "step": 66500
+    },
+    {
+      "epoch": 0.4814436795587309,
+      "grad_norm": 0.15441229939460754,
+      "learning_rate": 4.518563559107328e-06,
+      "loss": 0.9354,
+      "step": 66510
+    },
+    {
+      "epoch": 0.4815160662193171,
+      "grad_norm": 0.14522579312324524,
+      "learning_rate": 4.518491172446742e-06,
+      "loss": 0.9209,
+      "step": 66520
+    },
+    {
+      "epoch": 0.4815884528799033,
+      "grad_norm": 0.15296971797943115,
+      "learning_rate": 4.5184187857861555e-06,
+      "loss": 0.9226,
+      "step": 66530
+    },
+    {
+      "epoch": 0.4816608395404895,
+      "grad_norm": 0.15403321385383606,
+      "learning_rate": 4.518346399125569e-06,
+      "loss": 0.9182,
+      "step": 66540
+    },
+    {
+      "epoch": 0.48173322620107567,
+      "grad_norm": 0.16520103812217712,
+      "learning_rate": 4.5182740124649836e-06,
+      "loss": 0.9209,
+      "step": 66550
+    },
+    {
+      "epoch": 0.48180561286166185,
+      "grad_norm": 0.16668029129505157,
+      "learning_rate": 4.518201625804397e-06,
+      "loss": 0.9384,
+      "step": 66560
+    },
+    {
+      "epoch": 0.481877999522248,
+      "grad_norm": 0.16062092781066895,
+      "learning_rate": 4.518129239143811e-06,
+      "loss": 0.9313,
+      "step": 66570
+    },
+    {
+      "epoch": 0.4819503861828342,
+      "grad_norm": 0.15920652449131012,
+      "learning_rate": 4.518056852483224e-06,
+      "loss": 0.9146,
+      "step": 66580
+    },
+    {
+      "epoch": 0.48202277284342043,
+      "grad_norm": 0.1484328955411911,
+      "learning_rate": 4.517984465822639e-06,
+      "loss": 0.9161,
+      "step": 66590
+    },
+    {
+      "epoch": 0.4820951595040066,
+      "grad_norm": 0.16604140400886536,
+      "learning_rate": 4.5179120791620525e-06,
+      "loss": 0.9102,
+      "step": 66600
+    },
+    {
+      "epoch": 0.4821675461645928,
+      "grad_norm": 0.16721735894680023,
+      "learning_rate": 4.517839692501466e-06,
+      "loss": 0.9208,
+      "step": 66610
+    },
+    {
+      "epoch": 0.48223993282517896,
+      "grad_norm": 0.17339125275611877,
+      "learning_rate": 4.51776730584088e-06,
+      "loss": 0.9244,
+      "step": 66620
+    },
+    {
+      "epoch": 0.48231231948576514,
+      "grad_norm": 0.1743367463350296,
+      "learning_rate": 4.517694919180294e-06,
+      "loss": 0.9294,
+      "step": 66630
+    },
+    {
+      "epoch": 0.48238470614635137,
+      "grad_norm": 0.1753169298171997,
+      "learning_rate": 4.517622532519708e-06,
+      "loss": 0.9206,
+      "step": 66640
+    },
+    {
+      "epoch": 0.48245709280693755,
+      "grad_norm": 0.16888897120952606,
+      "learning_rate": 4.5175501458591214e-06,
+      "loss": 0.9175,
+      "step": 66650
+    },
+    {
+      "epoch": 0.4825294794675237,
+      "grad_norm": 0.18841847777366638,
+      "learning_rate": 4.517477759198535e-06,
+      "loss": 0.9228,
+      "step": 66660
+    },
+    {
+      "epoch": 0.4826018661281099,
+      "grad_norm": 0.1644550859928131,
+      "learning_rate": 4.517405372537949e-06,
+      "loss": 0.9295,
+      "step": 66670
+    },
+    {
+      "epoch": 0.4826742527886961,
+      "grad_norm": 0.15950177609920502,
+      "learning_rate": 4.517332985877363e-06,
+      "loss": 0.9303,
+      "step": 66680
+    },
+    {
+      "epoch": 0.4827466394492823,
+      "grad_norm": 0.1527274250984192,
+      "learning_rate": 4.517260599216777e-06,
+      "loss": 0.924,
+      "step": 66690
+    },
+    {
+      "epoch": 0.4828190261098685,
+      "grad_norm": 0.1690022051334381,
+      "learning_rate": 4.51718821255619e-06,
+      "loss": 0.9212,
+      "step": 66700
+    },
+    {
+      "epoch": 0.48289141277045466,
+      "grad_norm": 0.17802709341049194,
+      "learning_rate": 4.517115825895604e-06,
+      "loss": 0.9246,
+      "step": 66710
+    },
+    {
+      "epoch": 0.48296379943104084,
+      "grad_norm": 0.15989567339420319,
+      "learning_rate": 4.5170434392350184e-06,
+      "loss": 0.9205,
+      "step": 66720
+    },
+    {
+      "epoch": 0.483036186091627,
+      "grad_norm": 0.16513442993164062,
+      "learning_rate": 4.516971052574432e-06,
+      "loss": 0.9294,
+      "step": 66730
+    },
+    {
+      "epoch": 0.4831085727522132,
+      "grad_norm": 0.157196044921875,
+      "learning_rate": 4.516898665913846e-06,
+      "loss": 0.9186,
+      "step": 66740
+    },
+    {
+      "epoch": 0.4831809594127994,
+      "grad_norm": 0.17779704928398132,
+      "learning_rate": 4.516826279253259e-06,
+      "loss": 0.9111,
+      "step": 66750
+    },
+    {
+      "epoch": 0.4832533460733856,
+      "grad_norm": 0.15791267156600952,
+      "learning_rate": 4.516753892592674e-06,
+      "loss": 0.938,
+      "step": 66760
+    },
+    {
+      "epoch": 0.4833257327339718,
+      "grad_norm": 0.16559448838233948,
+      "learning_rate": 4.516681505932087e-06,
+      "loss": 0.9158,
+      "step": 66770
+    },
+    {
+      "epoch": 0.48339811939455796,
+      "grad_norm": 0.1510489284992218,
+      "learning_rate": 4.516609119271501e-06,
+      "loss": 0.9308,
+      "step": 66780
+    },
+    {
+      "epoch": 0.48347050605514413,
+      "grad_norm": 0.16266123950481415,
+      "learning_rate": 4.516536732610915e-06,
+      "loss": 0.9223,
+      "step": 66790
+    },
+    {
+      "epoch": 0.48354289271573037,
+      "grad_norm": 0.16285644471645355,
+      "learning_rate": 4.516464345950329e-06,
+      "loss": 0.9035,
+      "step": 66800
+    },
+    {
+      "epoch": 0.48361527937631654,
+      "grad_norm": 0.16523510217666626,
+      "learning_rate": 4.516391959289743e-06,
+      "loss": 0.923,
+      "step": 66810
+    },
+    {
+      "epoch": 0.4836876660369027,
+      "grad_norm": 0.1816108375787735,
+      "learning_rate": 4.516319572629156e-06,
+      "loss": 0.9195,
+      "step": 66820
+    },
+    {
+      "epoch": 0.4837600526974889,
+      "grad_norm": 0.1596839874982834,
+      "learning_rate": 4.51624718596857e-06,
+      "loss": 0.932,
+      "step": 66830
+    },
+    {
+      "epoch": 0.4838324393580751,
+      "grad_norm": 0.16277308762073517,
+      "learning_rate": 4.516174799307984e-06,
+      "loss": 0.9247,
+      "step": 66840
+    },
+    {
+      "epoch": 0.4839048260186613,
+      "grad_norm": 0.1530946046113968,
+      "learning_rate": 4.516102412647398e-06,
+      "loss": 0.9213,
+      "step": 66850
+    },
+    {
+      "epoch": 0.4839772126792475,
+      "grad_norm": 0.16606640815734863,
+      "learning_rate": 4.516030025986812e-06,
+      "loss": 0.9209,
+      "step": 66860
+    },
+    {
+      "epoch": 0.48404959933983366,
+      "grad_norm": 0.180929034948349,
+      "learning_rate": 4.515957639326225e-06,
+      "loss": 0.9209,
+      "step": 66870
+    },
+    {
+      "epoch": 0.48412198600041983,
+      "grad_norm": 0.15386876463890076,
+      "learning_rate": 4.51588525266564e-06,
+      "loss": 0.9208,
+      "step": 66880
+    },
+    {
+      "epoch": 0.484194372661006,
+      "grad_norm": 0.17483779788017273,
+      "learning_rate": 4.515812866005053e-06,
+      "loss": 0.9147,
+      "step": 66890
+    },
+    {
+      "epoch": 0.48426675932159224,
+      "grad_norm": 0.17675867676734924,
+      "learning_rate": 4.515740479344467e-06,
+      "loss": 0.9216,
+      "step": 66900
+    },
+    {
+      "epoch": 0.4843391459821784,
+      "grad_norm": 0.16373442113399506,
+      "learning_rate": 4.5156680926838805e-06,
+      "loss": 0.9273,
+      "step": 66910
+    },
+    {
+      "epoch": 0.4844115326427646,
+      "grad_norm": 0.16552437841892242,
+      "learning_rate": 4.515595706023295e-06,
+      "loss": 0.9343,
+      "step": 66920
+    },
+    {
+      "epoch": 0.4844839193033508,
+      "grad_norm": 0.16545453667640686,
+      "learning_rate": 4.515523319362708e-06,
+      "loss": 0.9205,
+      "step": 66930
+    },
+    {
+      "epoch": 0.48455630596393695,
+      "grad_norm": 0.17034026980400085,
+      "learning_rate": 4.515450932702121e-06,
+      "loss": 0.9213,
+      "step": 66940
+    },
+    {
+      "epoch": 0.4846286926245231,
+      "grad_norm": 0.2646459639072418,
+      "learning_rate": 4.515378546041536e-06,
+      "loss": 0.9172,
+      "step": 66950
+    },
+    {
+      "epoch": 0.48470107928510936,
+      "grad_norm": 0.16209463775157928,
+      "learning_rate": 4.5153061593809495e-06,
+      "loss": 0.9155,
+      "step": 66960
+    },
+    {
+      "epoch": 0.48477346594569554,
+      "grad_norm": 0.1605486124753952,
+      "learning_rate": 4.515233772720363e-06,
+      "loss": 0.9242,
+      "step": 66970
+    },
+    {
+      "epoch": 0.4848458526062817,
+      "grad_norm": 0.16971814632415771,
+      "learning_rate": 4.515161386059777e-06,
+      "loss": 0.9143,
+      "step": 66980
+    },
+    {
+      "epoch": 0.4849182392668679,
+      "grad_norm": 0.17342694103717804,
+      "learning_rate": 4.515088999399191e-06,
+      "loss": 0.9302,
+      "step": 66990
+    },
+    {
+      "epoch": 0.48499062592745407,
+      "grad_norm": 0.15016865730285645,
+      "learning_rate": 4.515016612738605e-06,
+      "loss": 0.9206,
+      "step": 67000
+    },
+    {
+      "epoch": 0.4850630125880403,
+      "grad_norm": 0.15992531180381775,
+      "learning_rate": 4.514944226078018e-06,
+      "loss": 0.9153,
+      "step": 67010
+    },
+    {
+      "epoch": 0.4851353992486265,
+      "grad_norm": 0.1591133326292038,
+      "learning_rate": 4.514871839417432e-06,
+      "loss": 0.9185,
+      "step": 67020
+    },
+    {
+      "epoch": 0.48520778590921265,
+      "grad_norm": 0.16176247596740723,
+      "learning_rate": 4.5147994527568465e-06,
+      "loss": 0.9246,
+      "step": 67030
+    },
+    {
+      "epoch": 0.48528017256979883,
+      "grad_norm": 0.15282560884952545,
+      "learning_rate": 4.51472706609626e-06,
+      "loss": 0.9323,
+      "step": 67040
+    },
+    {
+      "epoch": 0.485352559230385,
+      "grad_norm": 0.1511177271604538,
+      "learning_rate": 4.514654679435674e-06,
+      "loss": 0.9358,
+      "step": 67050
+    },
+    {
+      "epoch": 0.48542494589097124,
+      "grad_norm": 0.15375123918056488,
+      "learning_rate": 4.514582292775087e-06,
+      "loss": 0.9322,
+      "step": 67060
+    },
+    {
+      "epoch": 0.4854973325515574,
+      "grad_norm": 0.21929968893527985,
+      "learning_rate": 4.514509906114502e-06,
+      "loss": 0.9129,
+      "step": 67070
+    },
+    {
+      "epoch": 0.4855697192121436,
+      "grad_norm": 0.16412512958049774,
+      "learning_rate": 4.514437519453915e-06,
+      "loss": 0.9264,
+      "step": 67080
+    },
+    {
+      "epoch": 0.48564210587272977,
+      "grad_norm": 0.15203996002674103,
+      "learning_rate": 4.514365132793329e-06,
+      "loss": 0.922,
+      "step": 67090
+    },
+    {
+      "epoch": 0.48571449253331594,
+      "grad_norm": 0.15428316593170166,
+      "learning_rate": 4.514292746132743e-06,
+      "loss": 0.9196,
+      "step": 67100
+    },
+    {
+      "epoch": 0.4857868791939021,
+      "grad_norm": 0.171985924243927,
+      "learning_rate": 4.514220359472157e-06,
+      "loss": 0.9349,
+      "step": 67110
+    },
+    {
+      "epoch": 0.48585926585448835,
+      "grad_norm": 0.16062787175178528,
+      "learning_rate": 4.514147972811571e-06,
+      "loss": 0.9108,
+      "step": 67120
+    },
+    {
+      "epoch": 0.48593165251507453,
+      "grad_norm": 0.15661896765232086,
+      "learning_rate": 4.514075586150984e-06,
+      "loss": 0.9236,
+      "step": 67130
+    },
+    {
+      "epoch": 0.4860040391756607,
+      "grad_norm": 0.16929559409618378,
+      "learning_rate": 4.514003199490398e-06,
+      "loss": 0.9322,
+      "step": 67140
+    },
+    {
+      "epoch": 0.4860764258362469,
+      "grad_norm": 0.1664290726184845,
+      "learning_rate": 4.513930812829812e-06,
+      "loss": 0.9166,
+      "step": 67150
+    },
+    {
+      "epoch": 0.48614881249683306,
+      "grad_norm": 0.17513130605220795,
+      "learning_rate": 4.513858426169226e-06,
+      "loss": 0.917,
+      "step": 67160
+    },
+    {
+      "epoch": 0.4862211991574193,
+      "grad_norm": 0.16142228245735168,
+      "learning_rate": 4.51378603950864e-06,
+      "loss": 0.9296,
+      "step": 67170
+    },
+    {
+      "epoch": 0.48629358581800547,
+      "grad_norm": 0.17332206666469574,
+      "learning_rate": 4.513713652848053e-06,
+      "loss": 0.9092,
+      "step": 67180
+    },
+    {
+      "epoch": 0.48636597247859165,
+      "grad_norm": 0.15262171626091003,
+      "learning_rate": 4.513641266187468e-06,
+      "loss": 0.935,
+      "step": 67190
+    },
+    {
+      "epoch": 0.4864383591391778,
+      "grad_norm": 0.1601051539182663,
+      "learning_rate": 4.513568879526881e-06,
+      "loss": 0.9203,
+      "step": 67200
+    },
+    {
+      "epoch": 0.486510745799764,
+      "grad_norm": 0.16139155626296997,
+      "learning_rate": 4.513496492866295e-06,
+      "loss": 0.9336,
+      "step": 67210
+    },
+    {
+      "epoch": 0.48658313246035023,
+      "grad_norm": 0.182013601064682,
+      "learning_rate": 4.5134241062057086e-06,
+      "loss": 0.9269,
+      "step": 67220
+    },
+    {
+      "epoch": 0.4866555191209364,
+      "grad_norm": 0.19671277701854706,
+      "learning_rate": 4.513351719545123e-06,
+      "loss": 0.924,
+      "step": 67230
+    },
+    {
+      "epoch": 0.4867279057815226,
+      "grad_norm": 0.17239131033420563,
+      "learning_rate": 4.513279332884537e-06,
+      "loss": 0.9346,
+      "step": 67240
+    },
+    {
+      "epoch": 0.48680029244210876,
+      "grad_norm": 0.1634131819009781,
+      "learning_rate": 4.51320694622395e-06,
+      "loss": 0.9198,
+      "step": 67250
+    },
+    {
+      "epoch": 0.48687267910269494,
+      "grad_norm": 0.1805330216884613,
+      "learning_rate": 4.513134559563364e-06,
+      "loss": 0.9201,
+      "step": 67260
+    },
+    {
+      "epoch": 0.4869450657632811,
+      "grad_norm": 0.15409274399280548,
+      "learning_rate": 4.513062172902778e-06,
+      "loss": 0.9268,
+      "step": 67270
+    },
+    {
+      "epoch": 0.48701745242386735,
+      "grad_norm": 0.14754238724708557,
+      "learning_rate": 4.512989786242192e-06,
+      "loss": 0.92,
+      "step": 67280
+    },
+    {
+      "epoch": 0.4870898390844535,
+      "grad_norm": 0.14895077049732208,
+      "learning_rate": 4.5129173995816056e-06,
+      "loss": 0.926,
+      "step": 67290
+    },
+    {
+      "epoch": 0.4871622257450397,
+      "grad_norm": 0.16120226681232452,
+      "learning_rate": 4.512845012921019e-06,
+      "loss": 0.9242,
+      "step": 67300
+    },
+    {
+      "epoch": 0.4872346124056259,
+      "grad_norm": 0.16055871546268463,
+      "learning_rate": 4.512772626260433e-06,
+      "loss": 0.9222,
+      "step": 67310
+    },
+    {
+      "epoch": 0.48730699906621205,
+      "grad_norm": 0.16574375331401825,
+      "learning_rate": 4.512700239599847e-06,
+      "loss": 0.9205,
+      "step": 67320
+    },
+    {
+      "epoch": 0.4873793857267983,
+      "grad_norm": 0.15389981865882874,
+      "learning_rate": 4.512627852939261e-06,
+      "loss": 0.9151,
+      "step": 67330
+    },
+    {
+      "epoch": 0.48745177238738446,
+      "grad_norm": 0.15907318890094757,
+      "learning_rate": 4.5125554662786745e-06,
+      "loss": 0.9225,
+      "step": 67340
+    },
+    {
+      "epoch": 0.48752415904797064,
+      "grad_norm": 0.16848038136959076,
+      "learning_rate": 4.512483079618088e-06,
+      "loss": 0.9333,
+      "step": 67350
+    },
+    {
+      "epoch": 0.4875965457085568,
+      "grad_norm": 0.15980379283428192,
+      "learning_rate": 4.512410692957503e-06,
+      "loss": 0.9227,
+      "step": 67360
+    },
+    {
+      "epoch": 0.487668932369143,
+      "grad_norm": 0.15230245888233185,
+      "learning_rate": 4.512338306296916e-06,
+      "loss": 0.9143,
+      "step": 67370
+    },
+    {
+      "epoch": 0.4877413190297292,
+      "grad_norm": 0.16187123954296112,
+      "learning_rate": 4.51226591963633e-06,
+      "loss": 0.9224,
+      "step": 67380
+    },
+    {
+      "epoch": 0.4878137056903154,
+      "grad_norm": 0.16506314277648926,
+      "learning_rate": 4.5121935329757434e-06,
+      "loss": 0.9233,
+      "step": 67390
+    },
+    {
+      "epoch": 0.4878860923509016,
+      "grad_norm": 0.21859927475452423,
+      "learning_rate": 4.512121146315158e-06,
+      "loss": 0.9181,
+      "step": 67400
+    },
+    {
+      "epoch": 0.48795847901148776,
+      "grad_norm": 0.16471314430236816,
+      "learning_rate": 4.5120487596545715e-06,
+      "loss": 0.9317,
+      "step": 67410
+    },
+    {
+      "epoch": 0.48803086567207393,
+      "grad_norm": 0.16795067489147186,
+      "learning_rate": 4.511976372993985e-06,
+      "loss": 0.9162,
+      "step": 67420
+    },
+    {
+      "epoch": 0.48810325233266016,
+      "grad_norm": 0.20342256128787994,
+      "learning_rate": 4.511903986333399e-06,
+      "loss": 0.9284,
+      "step": 67430
+    },
+    {
+      "epoch": 0.48817563899324634,
+      "grad_norm": 0.31586503982543945,
+      "learning_rate": 4.511831599672813e-06,
+      "loss": 0.9252,
+      "step": 67440
+    },
+    {
+      "epoch": 0.4882480256538325,
+      "grad_norm": 0.18031372129917145,
+      "learning_rate": 4.511759213012227e-06,
+      "loss": 0.9218,
+      "step": 67450
+    },
+    {
+      "epoch": 0.4883204123144187,
+      "grad_norm": 0.15259408950805664,
+      "learning_rate": 4.5116868263516404e-06,
+      "loss": 0.9252,
+      "step": 67460
+    },
+    {
+      "epoch": 0.48839279897500487,
+      "grad_norm": 0.15039771795272827,
+      "learning_rate": 4.511614439691054e-06,
+      "loss": 0.9228,
+      "step": 67470
+    },
+    {
+      "epoch": 0.48846518563559105,
+      "grad_norm": 0.15142498910427094,
+      "learning_rate": 4.511542053030468e-06,
+      "loss": 0.8998,
+      "step": 67480
+    },
+    {
+      "epoch": 0.4885375722961773,
+      "grad_norm": 0.15406163036823273,
+      "learning_rate": 4.511469666369881e-06,
+      "loss": 0.9219,
+      "step": 67490
+    },
+    {
+      "epoch": 0.48860995895676346,
+      "grad_norm": 0.16428889334201813,
+      "learning_rate": 4.511397279709295e-06,
+      "loss": 0.9298,
+      "step": 67500
+    },
+    {
+      "epoch": 0.48868234561734963,
+      "grad_norm": 0.15274979174137115,
+      "learning_rate": 4.511324893048709e-06,
+      "loss": 0.9374,
+      "step": 67510
+    },
+    {
+      "epoch": 0.4887547322779358,
+      "grad_norm": 0.15351366996765137,
+      "learning_rate": 4.511252506388123e-06,
+      "loss": 0.9262,
+      "step": 67520
+    },
+    {
+      "epoch": 0.488827118938522,
+      "grad_norm": 0.18971464037895203,
+      "learning_rate": 4.511180119727537e-06,
+      "loss": 0.9329,
+      "step": 67530
+    },
+    {
+      "epoch": 0.4888995055991082,
+      "grad_norm": 0.15166032314300537,
+      "learning_rate": 4.51110773306695e-06,
+      "loss": 0.9232,
+      "step": 67540
+    },
+    {
+      "epoch": 0.4889718922596944,
+      "grad_norm": 0.15017013251781464,
+      "learning_rate": 4.511035346406365e-06,
+      "loss": 0.9279,
+      "step": 67550
+    },
+    {
+      "epoch": 0.4890442789202806,
+      "grad_norm": 0.16144691407680511,
+      "learning_rate": 4.510962959745778e-06,
+      "loss": 0.9267,
+      "step": 67560
+    },
+    {
+      "epoch": 0.48911666558086675,
+      "grad_norm": 0.1623421162366867,
+      "learning_rate": 4.510890573085192e-06,
+      "loss": 0.9219,
+      "step": 67570
+    },
+    {
+      "epoch": 0.4891890522414529,
+      "grad_norm": 0.15922850370407104,
+      "learning_rate": 4.5108181864246055e-06,
+      "loss": 0.919,
+      "step": 67580
+    },
+    {
+      "epoch": 0.48926143890203916,
+      "grad_norm": 0.16896429657936096,
+      "learning_rate": 4.51074579976402e-06,
+      "loss": 0.9263,
+      "step": 67590
+    },
+    {
+      "epoch": 0.48933382556262534,
+      "grad_norm": 0.1584937423467636,
+      "learning_rate": 4.510673413103434e-06,
+      "loss": 0.93,
+      "step": 67600
+    },
+    {
+      "epoch": 0.4894062122232115,
+      "grad_norm": 0.15415899455547333,
+      "learning_rate": 4.510601026442847e-06,
+      "loss": 0.9178,
+      "step": 67610
+    },
+    {
+      "epoch": 0.4894785988837977,
+      "grad_norm": 0.15234239399433136,
+      "learning_rate": 4.510528639782261e-06,
+      "loss": 0.9174,
+      "step": 67620
+    },
+    {
+      "epoch": 0.48955098554438387,
+      "grad_norm": 0.1630704402923584,
+      "learning_rate": 4.510456253121675e-06,
+      "loss": 0.926,
+      "step": 67630
+    },
+    {
+      "epoch": 0.48962337220497004,
+      "grad_norm": 0.15613876283168793,
+      "learning_rate": 4.510383866461089e-06,
+      "loss": 0.9233,
+      "step": 67640
+    },
+    {
+      "epoch": 0.4896957588655563,
+      "grad_norm": 0.16620004177093506,
+      "learning_rate": 4.5103114798005025e-06,
+      "loss": 0.9264,
+      "step": 67650
+    },
+    {
+      "epoch": 0.48976814552614245,
+      "grad_norm": 0.15522471070289612,
+      "learning_rate": 4.510239093139916e-06,
+      "loss": 0.923,
+      "step": 67660
+    },
+    {
+      "epoch": 0.4898405321867286,
+      "grad_norm": 0.15571705996990204,
+      "learning_rate": 4.510166706479331e-06,
+      "loss": 0.927,
+      "step": 67670
+    },
+    {
+      "epoch": 0.4899129188473148,
+      "grad_norm": 0.15183138847351074,
+      "learning_rate": 4.510094319818744e-06,
+      "loss": 0.9162,
+      "step": 67680
+    },
+    {
+      "epoch": 0.489985305507901,
+      "grad_norm": 0.1504392772912979,
+      "learning_rate": 4.510021933158158e-06,
+      "loss": 0.9206,
+      "step": 67690
+    },
+    {
+      "epoch": 0.4900576921684872,
+      "grad_norm": 0.2166612446308136,
+      "learning_rate": 4.5099495464975715e-06,
+      "loss": 0.9206,
+      "step": 67700
+    },
+    {
+      "epoch": 0.4901300788290734,
+      "grad_norm": 0.19607232511043549,
+      "learning_rate": 4.509877159836986e-06,
+      "loss": 0.9225,
+      "step": 67710
+    },
+    {
+      "epoch": 0.49020246548965957,
+      "grad_norm": 0.1525793820619583,
+      "learning_rate": 4.5098047731763995e-06,
+      "loss": 0.9161,
+      "step": 67720
+    },
+    {
+      "epoch": 0.49027485215024574,
+      "grad_norm": 0.16443735361099243,
+      "learning_rate": 4.509732386515813e-06,
+      "loss": 0.9173,
+      "step": 67730
+    },
+    {
+      "epoch": 0.4903472388108319,
+      "grad_norm": 0.17903602123260498,
+      "learning_rate": 4.509659999855227e-06,
+      "loss": 0.9225,
+      "step": 67740
+    },
+    {
+      "epoch": 0.49041962547141815,
+      "grad_norm": 0.1686403751373291,
+      "learning_rate": 4.509587613194641e-06,
+      "loss": 0.9224,
+      "step": 67750
+    },
+    {
+      "epoch": 0.49049201213200433,
+      "grad_norm": 0.36816513538360596,
+      "learning_rate": 4.509515226534055e-06,
+      "loss": 0.9186,
+      "step": 67760
+    },
+    {
+      "epoch": 0.4905643987925905,
+      "grad_norm": 0.15775349736213684,
+      "learning_rate": 4.5094428398734685e-06,
+      "loss": 0.9361,
+      "step": 67770
+    },
+    {
+      "epoch": 0.4906367854531767,
+      "grad_norm": 0.17417992651462555,
+      "learning_rate": 4.509370453212882e-06,
+      "loss": 0.9245,
+      "step": 67780
+    },
+    {
+      "epoch": 0.49070917211376286,
+      "grad_norm": 0.1487663984298706,
+      "learning_rate": 4.5092980665522965e-06,
+      "loss": 0.9302,
+      "step": 67790
+    },
+    {
+      "epoch": 0.49078155877434904,
+      "grad_norm": 0.1552221029996872,
+      "learning_rate": 4.50922567989171e-06,
+      "loss": 0.926,
+      "step": 67800
+    },
+    {
+      "epoch": 0.49085394543493527,
+      "grad_norm": 0.16190670430660248,
+      "learning_rate": 4.509153293231124e-06,
+      "loss": 0.9284,
+      "step": 67810
+    },
+    {
+      "epoch": 0.49092633209552144,
+      "grad_norm": 0.17371685802936554,
+      "learning_rate": 4.509080906570537e-06,
+      "loss": 0.9413,
+      "step": 67820
+    },
+    {
+      "epoch": 0.4909987187561076,
+      "grad_norm": 0.1726863831281662,
+      "learning_rate": 4.509008519909952e-06,
+      "loss": 0.9315,
+      "step": 67830
+    },
+    {
+      "epoch": 0.4910711054166938,
+      "grad_norm": 0.15722721815109253,
+      "learning_rate": 4.5089361332493655e-06,
+      "loss": 0.9261,
+      "step": 67840
+    },
+    {
+      "epoch": 0.49114349207728,
+      "grad_norm": 0.15430855751037598,
+      "learning_rate": 4.508863746588779e-06,
+      "loss": 0.9288,
+      "step": 67850
+    },
+    {
+      "epoch": 0.4912158787378662,
+      "grad_norm": 0.1556122750043869,
+      "learning_rate": 4.508791359928193e-06,
+      "loss": 0.9227,
+      "step": 67860
+    },
+    {
+      "epoch": 0.4912882653984524,
+      "grad_norm": 0.1647377461194992,
+      "learning_rate": 4.508718973267607e-06,
+      "loss": 0.9096,
+      "step": 67870
+    },
+    {
+      "epoch": 0.49136065205903856,
+      "grad_norm": 0.1618121862411499,
+      "learning_rate": 4.508646586607021e-06,
+      "loss": 0.9199,
+      "step": 67880
+    },
+    {
+      "epoch": 0.49143303871962474,
+      "grad_norm": 0.16955524682998657,
+      "learning_rate": 4.508574199946434e-06,
+      "loss": 0.9285,
+      "step": 67890
+    },
+    {
+      "epoch": 0.4915054253802109,
+      "grad_norm": 0.15921622514724731,
+      "learning_rate": 4.508501813285848e-06,
+      "loss": 0.9237,
+      "step": 67900
+    },
+    {
+      "epoch": 0.49157781204079715,
+      "grad_norm": 0.17753024399280548,
+      "learning_rate": 4.5084294266252625e-06,
+      "loss": 0.9264,
+      "step": 67910
+    },
+    {
+      "epoch": 0.4916501987013833,
+      "grad_norm": 0.15799757838249207,
+      "learning_rate": 4.508357039964676e-06,
+      "loss": 0.914,
+      "step": 67920
+    },
+    {
+      "epoch": 0.4917225853619695,
+      "grad_norm": 0.17389525473117828,
+      "learning_rate": 4.50828465330409e-06,
+      "loss": 0.935,
+      "step": 67930
+    },
+    {
+      "epoch": 0.4917949720225557,
+      "grad_norm": 0.15797245502471924,
+      "learning_rate": 4.508212266643503e-06,
+      "loss": 0.9193,
+      "step": 67940
+    },
+    {
+      "epoch": 0.49186735868314185,
+      "grad_norm": 0.1706414818763733,
+      "learning_rate": 4.508139879982917e-06,
+      "loss": 0.9221,
+      "step": 67950
+    },
+    {
+      "epoch": 0.4919397453437281,
+      "grad_norm": 0.17515702545642853,
+      "learning_rate": 4.508067493322331e-06,
+      "loss": 0.924,
+      "step": 67960
+    },
+    {
+      "epoch": 0.49201213200431426,
+      "grad_norm": 0.16736344993114471,
+      "learning_rate": 4.507995106661745e-06,
+      "loss": 0.9147,
+      "step": 67970
+    },
+    {
+      "epoch": 0.49208451866490044,
+      "grad_norm": 0.24362386763095856,
+      "learning_rate": 4.507922720001159e-06,
+      "loss": 0.9217,
+      "step": 67980
+    },
+    {
+      "epoch": 0.4921569053254866,
+      "grad_norm": 0.1629912108182907,
+      "learning_rate": 4.507850333340572e-06,
+      "loss": 0.921,
+      "step": 67990
+    },
+    {
+      "epoch": 0.4922292919860728,
+      "grad_norm": 0.16399963200092316,
+      "learning_rate": 4.507777946679986e-06,
+      "loss": 0.9234,
+      "step": 68000
+    },
+    {
+      "epoch": 0.49230167864665897,
+      "grad_norm": 0.15990719199180603,
+      "learning_rate": 4.5077055600193995e-06,
+      "loss": 0.9207,
+      "step": 68010
+    },
+    {
+      "epoch": 0.4923740653072452,
+      "grad_norm": 0.1633741706609726,
+      "learning_rate": 4.507633173358814e-06,
+      "loss": 0.9153,
+      "step": 68020
+    },
+    {
+      "epoch": 0.4924464519678314,
+      "grad_norm": 0.15352605283260345,
+      "learning_rate": 4.5075607866982276e-06,
+      "loss": 0.9147,
+      "step": 68030
+    },
+    {
+      "epoch": 0.49251883862841755,
+      "grad_norm": 0.15996570885181427,
+      "learning_rate": 4.507488400037641e-06,
+      "loss": 0.9193,
+      "step": 68040
+    },
+    {
+      "epoch": 0.49259122528900373,
+      "grad_norm": 0.15362408757209778,
+      "learning_rate": 4.507416013377055e-06,
+      "loss": 0.9308,
+      "step": 68050
+    },
+    {
+      "epoch": 0.4926636119495899,
+      "grad_norm": 0.1646050661802292,
+      "learning_rate": 4.507343626716469e-06,
+      "loss": 0.9204,
+      "step": 68060
+    },
+    {
+      "epoch": 0.49273599861017614,
+      "grad_norm": 0.1549074798822403,
+      "learning_rate": 4.507271240055883e-06,
+      "loss": 0.9312,
+      "step": 68070
+    },
+    {
+      "epoch": 0.4928083852707623,
+      "grad_norm": 0.16783104836940765,
+      "learning_rate": 4.5071988533952965e-06,
+      "loss": 0.9225,
+      "step": 68080
+    },
+    {
+      "epoch": 0.4928807719313485,
+      "grad_norm": 0.16458235681056976,
+      "learning_rate": 4.50712646673471e-06,
+      "loss": 0.936,
+      "step": 68090
+    },
+    {
+      "epoch": 0.49295315859193467,
+      "grad_norm": 0.16786985099315643,
+      "learning_rate": 4.507054080074124e-06,
+      "loss": 0.9155,
+      "step": 68100
+    },
+    {
+      "epoch": 0.49302554525252085,
+      "grad_norm": 0.16184143722057343,
+      "learning_rate": 4.506981693413538e-06,
+      "loss": 0.9124,
+      "step": 68110
+    },
+    {
+      "epoch": 0.4930979319131071,
+      "grad_norm": 0.1618042141199112,
+      "learning_rate": 4.506909306752952e-06,
+      "loss": 0.9182,
+      "step": 68120
+    },
+    {
+      "epoch": 0.49317031857369326,
+      "grad_norm": 0.47443485260009766,
+      "learning_rate": 4.5068369200923654e-06,
+      "loss": 0.9171,
+      "step": 68130
+    },
+    {
+      "epoch": 0.49324270523427943,
+      "grad_norm": 0.16205495595932007,
+      "learning_rate": 4.506764533431779e-06,
+      "loss": 0.9191,
+      "step": 68140
+    },
+    {
+      "epoch": 0.4933150918948656,
+      "grad_norm": 0.15748542547225952,
+      "learning_rate": 4.5066921467711935e-06,
+      "loss": 0.929,
+      "step": 68150
+    },
+    {
+      "epoch": 0.4933874785554518,
+      "grad_norm": 0.15539519488811493,
+      "learning_rate": 4.506619760110607e-06,
+      "loss": 0.9244,
+      "step": 68160
+    },
+    {
+      "epoch": 0.49345986521603796,
+      "grad_norm": 0.15781527757644653,
+      "learning_rate": 4.506547373450021e-06,
+      "loss": 0.9215,
+      "step": 68170
+    },
+    {
+      "epoch": 0.4935322518766242,
+      "grad_norm": 0.15772010385990143,
+      "learning_rate": 4.506474986789434e-06,
+      "loss": 0.9173,
+      "step": 68180
+    },
+    {
+      "epoch": 0.49360463853721037,
+      "grad_norm": 0.15966345369815826,
+      "learning_rate": 4.506402600128849e-06,
+      "loss": 0.9222,
+      "step": 68190
+    },
+    {
+      "epoch": 0.49367702519779655,
+      "grad_norm": 0.15698853135108948,
+      "learning_rate": 4.5063302134682624e-06,
+      "loss": 0.9244,
+      "step": 68200
+    },
+    {
+      "epoch": 0.4937494118583827,
+      "grad_norm": 0.16982385516166687,
+      "learning_rate": 4.506257826807676e-06,
+      "loss": 0.9258,
+      "step": 68210
+    },
+    {
+      "epoch": 0.4938217985189689,
+      "grad_norm": 0.19245341420173645,
+      "learning_rate": 4.50618544014709e-06,
+      "loss": 0.9141,
+      "step": 68220
+    },
+    {
+      "epoch": 0.49389418517955513,
+      "grad_norm": 0.14945447444915771,
+      "learning_rate": 4.506113053486504e-06,
+      "loss": 0.9339,
+      "step": 68230
+    },
+    {
+      "epoch": 0.4939665718401413,
+      "grad_norm": 0.16145005822181702,
+      "learning_rate": 4.506040666825918e-06,
+      "loss": 0.9186,
+      "step": 68240
+    },
+    {
+      "epoch": 0.4940389585007275,
+      "grad_norm": 0.15670140087604523,
+      "learning_rate": 4.505968280165331e-06,
+      "loss": 0.9277,
+      "step": 68250
+    },
+    {
+      "epoch": 0.49411134516131366,
+      "grad_norm": 0.4954744577407837,
+      "learning_rate": 4.505895893504745e-06,
+      "loss": 0.9307,
+      "step": 68260
+    },
+    {
+      "epoch": 0.49418373182189984,
+      "grad_norm": 0.15574707090854645,
+      "learning_rate": 4.5058235068441594e-06,
+      "loss": 0.9254,
+      "step": 68270
+    },
+    {
+      "epoch": 0.4942561184824861,
+      "grad_norm": 0.1754721999168396,
+      "learning_rate": 4.505751120183573e-06,
+      "loss": 0.9294,
+      "step": 68280
+    },
+    {
+      "epoch": 0.49432850514307225,
+      "grad_norm": 0.18038561940193176,
+      "learning_rate": 4.505678733522987e-06,
+      "loss": 0.9175,
+      "step": 68290
+    },
+    {
+      "epoch": 0.4944008918036584,
+      "grad_norm": 0.16185778379440308,
+      "learning_rate": 4.5056063468624e-06,
+      "loss": 0.9344,
+      "step": 68300
+    },
+    {
+      "epoch": 0.4944732784642446,
+      "grad_norm": 0.1598776876926422,
+      "learning_rate": 4.505533960201815e-06,
+      "loss": 0.9228,
+      "step": 68310
+    },
+    {
+      "epoch": 0.4945456651248308,
+      "grad_norm": 0.4865586459636688,
+      "learning_rate": 4.505461573541228e-06,
+      "loss": 0.9308,
+      "step": 68320
+    },
+    {
+      "epoch": 0.49461805178541696,
+      "grad_norm": 0.1811351776123047,
+      "learning_rate": 4.505389186880642e-06,
+      "loss": 0.9199,
+      "step": 68330
+    },
+    {
+      "epoch": 0.4946904384460032,
+      "grad_norm": 0.15315648913383484,
+      "learning_rate": 4.505316800220056e-06,
+      "loss": 0.9264,
+      "step": 68340
+    },
+    {
+      "epoch": 0.49476282510658937,
+      "grad_norm": 0.1656007021665573,
+      "learning_rate": 4.50524441355947e-06,
+      "loss": 0.9213,
+      "step": 68350
+    },
+    {
+      "epoch": 0.49483521176717554,
+      "grad_norm": 0.17960675060749054,
+      "learning_rate": 4.505172026898884e-06,
+      "loss": 0.9258,
+      "step": 68360
+    },
+    {
+      "epoch": 0.4949075984277617,
+      "grad_norm": 0.18778973817825317,
+      "learning_rate": 4.505099640238297e-06,
+      "loss": 0.9233,
+      "step": 68370
+    },
+    {
+      "epoch": 0.4949799850883479,
+      "grad_norm": 0.1481725573539734,
+      "learning_rate": 4.505027253577711e-06,
+      "loss": 0.9133,
+      "step": 68380
+    },
+    {
+      "epoch": 0.49505237174893413,
+      "grad_norm": 0.171866312623024,
+      "learning_rate": 4.504954866917125e-06,
+      "loss": 0.9296,
+      "step": 68390
+    },
+    {
+      "epoch": 0.4951247584095203,
+      "grad_norm": 0.16532866656780243,
+      "learning_rate": 4.504882480256539e-06,
+      "loss": 0.8981,
+      "step": 68400
+    },
+    {
+      "epoch": 0.4951971450701065,
+      "grad_norm": 0.1649603694677353,
+      "learning_rate": 4.504810093595953e-06,
+      "loss": 0.9199,
+      "step": 68410
+    },
+    {
+      "epoch": 0.49526953173069266,
+      "grad_norm": 0.1862333118915558,
+      "learning_rate": 4.504737706935366e-06,
+      "loss": 0.9284,
+      "step": 68420
+    },
+    {
+      "epoch": 0.49534191839127883,
+      "grad_norm": 0.17143741250038147,
+      "learning_rate": 4.504665320274781e-06,
+      "loss": 0.914,
+      "step": 68430
+    },
+    {
+      "epoch": 0.49541430505186507,
+      "grad_norm": 0.16334125399589539,
+      "learning_rate": 4.504592933614194e-06,
+      "loss": 0.9182,
+      "step": 68440
+    },
+    {
+      "epoch": 0.49548669171245124,
+      "grad_norm": 0.18255673348903656,
+      "learning_rate": 4.504520546953608e-06,
+      "loss": 0.9099,
+      "step": 68450
+    },
+    {
+      "epoch": 0.4955590783730374,
+      "grad_norm": 0.16726234555244446,
+      "learning_rate": 4.5044481602930215e-06,
+      "loss": 0.9263,
+      "step": 68460
+    },
+    {
+      "epoch": 0.4956314650336236,
+      "grad_norm": 0.16593337059020996,
+      "learning_rate": 4.504375773632436e-06,
+      "loss": 0.9165,
+      "step": 68470
+    },
+    {
+      "epoch": 0.4957038516942098,
+      "grad_norm": 0.173760324716568,
+      "learning_rate": 4.50430338697185e-06,
+      "loss": 0.9228,
+      "step": 68480
+    },
+    {
+      "epoch": 0.49577623835479595,
+      "grad_norm": 0.1431112438440323,
+      "learning_rate": 4.504231000311263e-06,
+      "loss": 0.9308,
+      "step": 68490
+    },
+    {
+      "epoch": 0.4958486250153822,
+      "grad_norm": 0.22450746595859528,
+      "learning_rate": 4.504158613650677e-06,
+      "loss": 0.9277,
+      "step": 68500
+    },
+    {
+      "epoch": 0.49592101167596836,
+      "grad_norm": 0.16971862316131592,
+      "learning_rate": 4.504086226990091e-06,
+      "loss": 0.924,
+      "step": 68510
+    },
+    {
+      "epoch": 0.49599339833655454,
+      "grad_norm": 0.15365462005138397,
+      "learning_rate": 4.504013840329504e-06,
+      "loss": 0.9172,
+      "step": 68520
+    },
+    {
+      "epoch": 0.4960657849971407,
+      "grad_norm": 0.17213942110538483,
+      "learning_rate": 4.503941453668918e-06,
+      "loss": 0.912,
+      "step": 68530
+    },
+    {
+      "epoch": 0.4961381716577269,
+      "grad_norm": 0.16275885701179504,
+      "learning_rate": 4.503869067008332e-06,
+      "loss": 0.9113,
+      "step": 68540
+    },
+    {
+      "epoch": 0.4962105583183131,
+      "grad_norm": 0.1540212631225586,
+      "learning_rate": 4.503796680347746e-06,
+      "loss": 0.9075,
+      "step": 68550
+    },
+    {
+      "epoch": 0.4962829449788993,
+      "grad_norm": 0.15798866748809814,
+      "learning_rate": 4.503724293687159e-06,
+      "loss": 0.9271,
+      "step": 68560
+    },
+    {
+      "epoch": 0.4963553316394855,
+      "grad_norm": 0.1525816023349762,
+      "learning_rate": 4.503651907026573e-06,
+      "loss": 0.913,
+      "step": 68570
+    },
+    {
+      "epoch": 0.49642771830007165,
+      "grad_norm": 0.1655115783214569,
+      "learning_rate": 4.5035795203659875e-06,
+      "loss": 0.9094,
+      "step": 68580
+    },
+    {
+      "epoch": 0.49650010496065783,
+      "grad_norm": 0.16454185545444489,
+      "learning_rate": 4.503507133705401e-06,
+      "loss": 0.9232,
+      "step": 68590
+    },
+    {
+      "epoch": 0.49657249162124406,
+      "grad_norm": 0.15450453758239746,
+      "learning_rate": 4.503434747044815e-06,
+      "loss": 0.9256,
+      "step": 68600
+    },
+    {
+      "epoch": 0.49664487828183024,
+      "grad_norm": 0.15737608075141907,
+      "learning_rate": 4.503362360384228e-06,
+      "loss": 0.913,
+      "step": 68610
+    },
+    {
+      "epoch": 0.4967172649424164,
+      "grad_norm": 0.16449777781963348,
+      "learning_rate": 4.503289973723643e-06,
+      "loss": 0.9236,
+      "step": 68620
+    },
+    {
+      "epoch": 0.4967896516030026,
+      "grad_norm": 0.15372395515441895,
+      "learning_rate": 4.503217587063056e-06,
+      "loss": 0.9174,
+      "step": 68630
+    },
+    {
+      "epoch": 0.49686203826358877,
+      "grad_norm": 0.1783866584300995,
+      "learning_rate": 4.50314520040247e-06,
+      "loss": 0.9295,
+      "step": 68640
+    },
+    {
+      "epoch": 0.496934424924175,
+      "grad_norm": 0.1594487428665161,
+      "learning_rate": 4.503072813741884e-06,
+      "loss": 0.9224,
+      "step": 68650
+    },
+    {
+      "epoch": 0.4970068115847612,
+      "grad_norm": 0.16570931673049927,
+      "learning_rate": 4.503000427081298e-06,
+      "loss": 0.9139,
+      "step": 68660
+    },
+    {
+      "epoch": 0.49707919824534735,
+      "grad_norm": 0.16301476955413818,
+      "learning_rate": 4.502928040420712e-06,
+      "loss": 0.9185,
+      "step": 68670
+    },
+    {
+      "epoch": 0.49715158490593353,
+      "grad_norm": 0.17501094937324524,
+      "learning_rate": 4.502855653760125e-06,
+      "loss": 0.9253,
+      "step": 68680
+    },
+    {
+      "epoch": 0.4972239715665197,
+      "grad_norm": 0.1674194633960724,
+      "learning_rate": 4.502783267099539e-06,
+      "loss": 0.9193,
+      "step": 68690
+    },
+    {
+      "epoch": 0.4972963582271059,
+      "grad_norm": 0.16636104881763458,
+      "learning_rate": 4.502710880438953e-06,
+      "loss": 0.9098,
+      "step": 68700
+    },
+    {
+      "epoch": 0.4973687448876921,
+      "grad_norm": 0.16473974287509918,
+      "learning_rate": 4.502638493778367e-06,
+      "loss": 0.9239,
+      "step": 68710
+    },
+    {
+      "epoch": 0.4974411315482783,
+      "grad_norm": 0.17471805214881897,
+      "learning_rate": 4.502566107117781e-06,
+      "loss": 0.9208,
+      "step": 68720
+    },
+    {
+      "epoch": 0.49751351820886447,
+      "grad_norm": 0.16630442440509796,
+      "learning_rate": 4.502493720457194e-06,
+      "loss": 0.9233,
+      "step": 68730
+    },
+    {
+      "epoch": 0.49758590486945065,
+      "grad_norm": 0.14597173035144806,
+      "learning_rate": 4.502421333796608e-06,
+      "loss": 0.916,
+      "step": 68740
+    },
+    {
+      "epoch": 0.4976582915300368,
+      "grad_norm": 0.1661393642425537,
+      "learning_rate": 4.502348947136022e-06,
+      "loss": 0.9205,
+      "step": 68750
+    },
+    {
+      "epoch": 0.49773067819062305,
+      "grad_norm": 0.15321460366249084,
+      "learning_rate": 4.502276560475436e-06,
+      "loss": 0.9276,
+      "step": 68760
+    },
+    {
+      "epoch": 0.49780306485120923,
+      "grad_norm": 0.15991902351379395,
+      "learning_rate": 4.5022041738148496e-06,
+      "loss": 0.9238,
+      "step": 68770
+    },
+    {
+      "epoch": 0.4978754515117954,
+      "grad_norm": 0.17004790902137756,
+      "learning_rate": 4.502131787154263e-06,
+      "loss": 0.9171,
+      "step": 68780
+    },
+    {
+      "epoch": 0.4979478381723816,
+      "grad_norm": 0.17030557990074158,
+      "learning_rate": 4.502059400493678e-06,
+      "loss": 0.9158,
+      "step": 68790
+    },
+    {
+      "epoch": 0.49802022483296776,
+      "grad_norm": 0.16426852345466614,
+      "learning_rate": 4.501987013833091e-06,
+      "loss": 0.927,
+      "step": 68800
+    },
+    {
+      "epoch": 0.498092611493554,
+      "grad_norm": 0.14977115392684937,
+      "learning_rate": 4.501914627172505e-06,
+      "loss": 0.9155,
+      "step": 68810
+    },
+    {
+      "epoch": 0.49816499815414017,
+      "grad_norm": 0.16252385079860687,
+      "learning_rate": 4.5018422405119185e-06,
+      "loss": 0.9082,
+      "step": 68820
+    },
+    {
+      "epoch": 0.49823738481472635,
+      "grad_norm": 0.15293535590171814,
+      "learning_rate": 4.501769853851333e-06,
+      "loss": 0.9093,
+      "step": 68830
+    },
+    {
+      "epoch": 0.4983097714753125,
+      "grad_norm": 0.14948716759681702,
+      "learning_rate": 4.501697467190747e-06,
+      "loss": 0.9099,
+      "step": 68840
+    },
+    {
+      "epoch": 0.4983821581358987,
+      "grad_norm": 0.14924530684947968,
+      "learning_rate": 4.50162508053016e-06,
+      "loss": 0.9288,
+      "step": 68850
+    },
+    {
+      "epoch": 0.4984545447964849,
+      "grad_norm": 0.16742658615112305,
+      "learning_rate": 4.501552693869574e-06,
+      "loss": 0.9359,
+      "step": 68860
+    },
+    {
+      "epoch": 0.4985269314570711,
+      "grad_norm": 0.1566425859928131,
+      "learning_rate": 4.501480307208988e-06,
+      "loss": 0.9298,
+      "step": 68870
+    },
+    {
+      "epoch": 0.4985993181176573,
+      "grad_norm": 0.15507277846336365,
+      "learning_rate": 4.501407920548402e-06,
+      "loss": 0.9209,
+      "step": 68880
+    },
+    {
+      "epoch": 0.49867170477824346,
+      "grad_norm": 0.2236766368150711,
+      "learning_rate": 4.5013355338878155e-06,
+      "loss": 0.9195,
+      "step": 68890
+    },
+    {
+      "epoch": 0.49874409143882964,
+      "grad_norm": 0.14580419659614563,
+      "learning_rate": 4.501263147227229e-06,
+      "loss": 0.9099,
+      "step": 68900
+    },
+    {
+      "epoch": 0.4988164780994158,
+      "grad_norm": 0.1595289558172226,
+      "learning_rate": 4.501190760566644e-06,
+      "loss": 0.9197,
+      "step": 68910
+    },
+    {
+      "epoch": 0.49888886476000205,
+      "grad_norm": 0.15660600364208221,
+      "learning_rate": 4.501118373906057e-06,
+      "loss": 0.9246,
+      "step": 68920
+    },
+    {
+      "epoch": 0.4989612514205882,
+      "grad_norm": 0.1797994077205658,
+      "learning_rate": 4.501045987245471e-06,
+      "loss": 0.9251,
+      "step": 68930
+    },
+    {
+      "epoch": 0.4990336380811744,
+      "grad_norm": 0.16895923018455505,
+      "learning_rate": 4.5009736005848844e-06,
+      "loss": 0.9208,
+      "step": 68940
+    },
+    {
+      "epoch": 0.4991060247417606,
+      "grad_norm": 0.17603491246700287,
+      "learning_rate": 4.500901213924299e-06,
+      "loss": 0.9205,
+      "step": 68950
+    },
+    {
+      "epoch": 0.49917841140234676,
+      "grad_norm": 0.1691259890794754,
+      "learning_rate": 4.5008288272637125e-06,
+      "loss": 0.9282,
+      "step": 68960
+    },
+    {
+      "epoch": 0.499250798062933,
+      "grad_norm": 0.1619505137205124,
+      "learning_rate": 4.500756440603126e-06,
+      "loss": 0.9324,
+      "step": 68970
+    },
+    {
+      "epoch": 0.49932318472351916,
+      "grad_norm": 0.16095148026943207,
+      "learning_rate": 4.50068405394254e-06,
+      "loss": 0.9167,
+      "step": 68980
+    },
+    {
+      "epoch": 0.49939557138410534,
+      "grad_norm": 0.14836084842681885,
+      "learning_rate": 4.500611667281954e-06,
+      "loss": 0.9127,
+      "step": 68990
+    },
+    {
+      "epoch": 0.4994679580446915,
+      "grad_norm": 0.15759669244289398,
+      "learning_rate": 4.500539280621368e-06,
+      "loss": 0.9219,
+      "step": 69000
+    },
+    {
+      "epoch": 0.4995403447052777,
+      "grad_norm": 0.17617014050483704,
+      "learning_rate": 4.5004668939607814e-06,
+      "loss": 0.9272,
+      "step": 69010
+    },
+    {
+      "epoch": 0.49961273136586387,
+      "grad_norm": 0.15363344550132751,
+      "learning_rate": 4.500394507300195e-06,
+      "loss": 0.914,
+      "step": 69020
+    },
+    {
+      "epoch": 0.4996851180264501,
+      "grad_norm": 0.163772851228714,
+      "learning_rate": 4.5003221206396095e-06,
+      "loss": 0.9352,
+      "step": 69030
+    },
+    {
+      "epoch": 0.4997575046870363,
+      "grad_norm": 0.16267885267734528,
+      "learning_rate": 4.500249733979023e-06,
+      "loss": 0.9052,
+      "step": 69040
+    },
+    {
+      "epoch": 0.49982989134762246,
+      "grad_norm": 0.18130633234977722,
+      "learning_rate": 4.500177347318436e-06,
+      "loss": 0.9282,
+      "step": 69050
+    },
+    {
+      "epoch": 0.49990227800820863,
+      "grad_norm": 0.1978328377008438,
+      "learning_rate": 4.50010496065785e-06,
+      "loss": 0.9345,
+      "step": 69060
+    },
+    {
+      "epoch": 0.4999746646687948,
+      "grad_norm": 0.15893995761871338,
+      "learning_rate": 4.500032573997264e-06,
+      "loss": 0.9136,
+      "step": 69070
+    },
+    {
+      "epoch": 0.500047051329381,
+      "grad_norm": 0.17726098001003265,
+      "learning_rate": 4.499960187336678e-06,
+      "loss": 0.9239,
+      "step": 69080
+    },
+    {
+      "epoch": 0.5001194379899672,
+      "grad_norm": 0.20466522872447968,
+      "learning_rate": 4.499887800676091e-06,
+      "loss": 0.916,
+      "step": 69090
+    },
+    {
+      "epoch": 0.5001918246505533,
+      "grad_norm": 0.1550358682870865,
+      "learning_rate": 4.499815414015506e-06,
+      "loss": 0.9195,
+      "step": 69100
+    },
+    {
+      "epoch": 0.5002642113111396,
+      "grad_norm": 0.14892955124378204,
+      "learning_rate": 4.499743027354919e-06,
+      "loss": 0.9103,
+      "step": 69110
+    },
+    {
+      "epoch": 0.5003365979717258,
+      "grad_norm": 0.1611659675836563,
+      "learning_rate": 4.499670640694333e-06,
+      "loss": 0.9123,
+      "step": 69120
+    },
+    {
+      "epoch": 0.500408984632312,
+      "grad_norm": 0.19101987779140472,
+      "learning_rate": 4.4995982540337465e-06,
+      "loss": 0.9135,
+      "step": 69130
+    },
+    {
+      "epoch": 0.5004813712928982,
+      "grad_norm": 0.1674869805574417,
+      "learning_rate": 4.499525867373161e-06,
+      "loss": 0.9236,
+      "step": 69140
+    },
+    {
+      "epoch": 0.5005537579534843,
+      "grad_norm": 0.15739434957504272,
+      "learning_rate": 4.499453480712575e-06,
+      "loss": 0.9267,
+      "step": 69150
+    },
+    {
+      "epoch": 0.5006261446140705,
+      "grad_norm": 0.17267724871635437,
+      "learning_rate": 4.499381094051988e-06,
+      "loss": 0.9148,
+      "step": 69160
+    },
+    {
+      "epoch": 0.5006985312746567,
+      "grad_norm": 0.1497461199760437,
+      "learning_rate": 4.499308707391402e-06,
+      "loss": 0.9141,
+      "step": 69170
+    },
+    {
+      "epoch": 0.5007709179352429,
+      "grad_norm": 0.15880879759788513,
+      "learning_rate": 4.499236320730816e-06,
+      "loss": 0.9333,
+      "step": 69180
+    },
+    {
+      "epoch": 0.500843304595829,
+      "grad_norm": 0.1679236739873886,
+      "learning_rate": 4.49916393407023e-06,
+      "loss": 0.902,
+      "step": 69190
+    },
+    {
+      "epoch": 0.5009156912564152,
+      "grad_norm": 0.1678447723388672,
+      "learning_rate": 4.4990915474096435e-06,
+      "loss": 0.9136,
+      "step": 69200
+    },
+    {
+      "epoch": 0.5009880779170015,
+      "grad_norm": 0.16553092002868652,
+      "learning_rate": 4.499019160749057e-06,
+      "loss": 0.9205,
+      "step": 69210
+    },
+    {
+      "epoch": 0.5010604645775877,
+      "grad_norm": 0.15976086258888245,
+      "learning_rate": 4.498946774088472e-06,
+      "loss": 0.9124,
+      "step": 69220
+    },
+    {
+      "epoch": 0.5011328512381739,
+      "grad_norm": 0.1608666628599167,
+      "learning_rate": 4.498874387427885e-06,
+      "loss": 0.9233,
+      "step": 69230
+    },
+    {
+      "epoch": 0.50120523789876,
+      "grad_norm": 0.19113683700561523,
+      "learning_rate": 4.498802000767299e-06,
+      "loss": 0.9208,
+      "step": 69240
+    },
+    {
+      "epoch": 0.5012776245593462,
+      "grad_norm": 0.1584874838590622,
+      "learning_rate": 4.4987296141067125e-06,
+      "loss": 0.9106,
+      "step": 69250
+    },
+    {
+      "epoch": 0.5013500112199324,
+      "grad_norm": 0.17393231391906738,
+      "learning_rate": 4.498657227446127e-06,
+      "loss": 0.915,
+      "step": 69260
+    },
+    {
+      "epoch": 0.5014223978805186,
+      "grad_norm": 0.1741904765367508,
+      "learning_rate": 4.4985848407855405e-06,
+      "loss": 0.9318,
+      "step": 69270
+    },
+    {
+      "epoch": 0.5014947845411047,
+      "grad_norm": 0.15421254932880402,
+      "learning_rate": 4.498512454124954e-06,
+      "loss": 0.9146,
+      "step": 69280
+    },
+    {
+      "epoch": 0.5015671712016909,
+      "grad_norm": 0.19196033477783203,
+      "learning_rate": 4.498440067464368e-06,
+      "loss": 0.9116,
+      "step": 69290
+    },
+    {
+      "epoch": 0.5016395578622771,
+      "grad_norm": 0.15833470225334167,
+      "learning_rate": 4.498367680803782e-06,
+      "loss": 0.9232,
+      "step": 69300
+    },
+    {
+      "epoch": 0.5017119445228633,
+      "grad_norm": 0.1550120860338211,
+      "learning_rate": 4.498295294143196e-06,
+      "loss": 0.9222,
+      "step": 69310
+    },
+    {
+      "epoch": 0.5017843311834496,
+      "grad_norm": 0.15786628425121307,
+      "learning_rate": 4.4982229074826095e-06,
+      "loss": 0.936,
+      "step": 69320
+    },
+    {
+      "epoch": 0.5018567178440357,
+      "grad_norm": 0.1613067090511322,
+      "learning_rate": 4.498150520822023e-06,
+      "loss": 0.9028,
+      "step": 69330
+    },
+    {
+      "epoch": 0.5019291045046219,
+      "grad_norm": 0.20446714758872986,
+      "learning_rate": 4.498078134161437e-06,
+      "loss": 0.8997,
+      "step": 69340
+    },
+    {
+      "epoch": 0.5020014911652081,
+      "grad_norm": 0.1550551950931549,
+      "learning_rate": 4.498005747500851e-06,
+      "loss": 0.9199,
+      "step": 69350
+    },
+    {
+      "epoch": 0.5020738778257943,
+      "grad_norm": 0.15930670499801636,
+      "learning_rate": 4.497933360840265e-06,
+      "loss": 0.9126,
+      "step": 69360
+    },
+    {
+      "epoch": 0.5021462644863804,
+      "grad_norm": 0.16248895227909088,
+      "learning_rate": 4.497860974179678e-06,
+      "loss": 0.9354,
+      "step": 69370
+    },
+    {
+      "epoch": 0.5022186511469666,
+      "grad_norm": 0.1518125683069229,
+      "learning_rate": 4.497788587519092e-06,
+      "loss": 0.9201,
+      "step": 69380
+    },
+    {
+      "epoch": 0.5022910378075528,
+      "grad_norm": 0.16246938705444336,
+      "learning_rate": 4.4977162008585065e-06,
+      "loss": 0.9231,
+      "step": 69390
+    },
+    {
+      "epoch": 0.502363424468139,
+      "grad_norm": 0.17307987809181213,
+      "learning_rate": 4.49764381419792e-06,
+      "loss": 0.9279,
+      "step": 69400
+    },
+    {
+      "epoch": 0.5024358111287252,
+      "grad_norm": 0.1634259670972824,
+      "learning_rate": 4.497571427537334e-06,
+      "loss": 0.9189,
+      "step": 69410
+    },
+    {
+      "epoch": 0.5025081977893114,
+      "grad_norm": 0.1655704826116562,
+      "learning_rate": 4.497499040876747e-06,
+      "loss": 0.8979,
+      "step": 69420
+    },
+    {
+      "epoch": 0.5025805844498976,
+      "grad_norm": 0.16415700316429138,
+      "learning_rate": 4.497426654216162e-06,
+      "loss": 0.9202,
+      "step": 69430
+    },
+    {
+      "epoch": 0.5026529711104838,
+      "grad_norm": 0.16875231266021729,
+      "learning_rate": 4.497354267555575e-06,
+      "loss": 0.9152,
+      "step": 69440
+    },
+    {
+      "epoch": 0.50272535777107,
+      "grad_norm": 0.19432953000068665,
+      "learning_rate": 4.497281880894989e-06,
+      "loss": 0.9217,
+      "step": 69450
+    },
+    {
+      "epoch": 0.5027977444316561,
+      "grad_norm": 0.15906265377998352,
+      "learning_rate": 4.497209494234403e-06,
+      "loss": 0.9239,
+      "step": 69460
+    },
+    {
+      "epoch": 0.5028701310922423,
+      "grad_norm": 0.16084323823451996,
+      "learning_rate": 4.497137107573817e-06,
+      "loss": 0.9084,
+      "step": 69470
+    },
+    {
+      "epoch": 0.5029425177528285,
+      "grad_norm": 0.16277502477169037,
+      "learning_rate": 4.497064720913231e-06,
+      "loss": 0.9318,
+      "step": 69480
+    },
+    {
+      "epoch": 0.5030149044134147,
+      "grad_norm": 0.16054733097553253,
+      "learning_rate": 4.496992334252644e-06,
+      "loss": 0.9187,
+      "step": 69490
+    },
+    {
+      "epoch": 0.5030872910740009,
+      "grad_norm": 0.14251933991909027,
+      "learning_rate": 4.496919947592058e-06,
+      "loss": 0.9271,
+      "step": 69500
+    },
+    {
+      "epoch": 0.503159677734587,
+      "grad_norm": 0.1685430407524109,
+      "learning_rate": 4.496847560931472e-06,
+      "loss": 0.9193,
+      "step": 69510
+    },
+    {
+      "epoch": 0.5032320643951732,
+      "grad_norm": 0.15440599620342255,
+      "learning_rate": 4.496775174270886e-06,
+      "loss": 0.9206,
+      "step": 69520
+    },
+    {
+      "epoch": 0.5033044510557595,
+      "grad_norm": 0.15421442687511444,
+      "learning_rate": 4.4967027876103e-06,
+      "loss": 0.923,
+      "step": 69530
+    },
+    {
+      "epoch": 0.5033768377163457,
+      "grad_norm": 0.1564491093158722,
+      "learning_rate": 4.496630400949713e-06,
+      "loss": 0.9216,
+      "step": 69540
+    },
+    {
+      "epoch": 0.5034492243769318,
+      "grad_norm": 0.16401636600494385,
+      "learning_rate": 4.496558014289128e-06,
+      "loss": 0.9185,
+      "step": 69550
+    },
+    {
+      "epoch": 0.503521611037518,
+      "grad_norm": 0.1628132462501526,
+      "learning_rate": 4.496485627628541e-06,
+      "loss": 0.9235,
+      "step": 69560
+    },
+    {
+      "epoch": 0.5035939976981042,
+      "grad_norm": 0.16760440170764923,
+      "learning_rate": 4.496413240967955e-06,
+      "loss": 0.9184,
+      "step": 69570
+    },
+    {
+      "epoch": 0.5036663843586904,
+      "grad_norm": 0.15936897695064545,
+      "learning_rate": 4.496340854307369e-06,
+      "loss": 0.9206,
+      "step": 69580
+    },
+    {
+      "epoch": 0.5037387710192766,
+      "grad_norm": 0.15765340626239777,
+      "learning_rate": 4.496268467646782e-06,
+      "loss": 0.9306,
+      "step": 69590
+    },
+    {
+      "epoch": 0.5038111576798627,
+      "grad_norm": 0.17861410975456238,
+      "learning_rate": 4.496196080986196e-06,
+      "loss": 0.9187,
+      "step": 69600
+    },
+    {
+      "epoch": 0.5038835443404489,
+      "grad_norm": 0.24393630027770996,
+      "learning_rate": 4.4961236943256094e-06,
+      "loss": 0.9149,
+      "step": 69610
+    },
+    {
+      "epoch": 0.5039559310010351,
+      "grad_norm": 0.20668023824691772,
+      "learning_rate": 4.496051307665024e-06,
+      "loss": 0.9286,
+      "step": 69620
+    },
+    {
+      "epoch": 0.5040283176616213,
+      "grad_norm": 0.15350449085235596,
+      "learning_rate": 4.4959789210044375e-06,
+      "loss": 0.9045,
+      "step": 69630
+    },
+    {
+      "epoch": 0.5041007043222075,
+      "grad_norm": 0.16766700148582458,
+      "learning_rate": 4.495906534343851e-06,
+      "loss": 0.9155,
+      "step": 69640
+    },
+    {
+      "epoch": 0.5041730909827937,
+      "grad_norm": 0.15862970054149628,
+      "learning_rate": 4.495834147683265e-06,
+      "loss": 0.916,
+      "step": 69650
+    },
+    {
+      "epoch": 0.5042454776433799,
+      "grad_norm": 0.1742592453956604,
+      "learning_rate": 4.495761761022679e-06,
+      "loss": 0.9168,
+      "step": 69660
+    },
+    {
+      "epoch": 0.5043178643039661,
+      "grad_norm": 0.15328940749168396,
+      "learning_rate": 4.495689374362093e-06,
+      "loss": 0.9151,
+      "step": 69670
+    },
+    {
+      "epoch": 0.5043902509645523,
+      "grad_norm": 0.15543635189533234,
+      "learning_rate": 4.4956169877015064e-06,
+      "loss": 0.917,
+      "step": 69680
+    },
+    {
+      "epoch": 0.5044626376251384,
+      "grad_norm": 0.1722266972064972,
+      "learning_rate": 4.49554460104092e-06,
+      "loss": 0.9218,
+      "step": 69690
+    },
+    {
+      "epoch": 0.5045350242857246,
+      "grad_norm": 0.1570882499217987,
+      "learning_rate": 4.4954722143803345e-06,
+      "loss": 0.9202,
+      "step": 69700
+    },
+    {
+      "epoch": 0.5046074109463108,
+      "grad_norm": 0.15502651035785675,
+      "learning_rate": 4.495399827719748e-06,
+      "loss": 0.9209,
+      "step": 69710
+    },
+    {
+      "epoch": 0.504679797606897,
+      "grad_norm": 0.1683962345123291,
+      "learning_rate": 4.495327441059162e-06,
+      "loss": 0.9239,
+      "step": 69720
+    },
+    {
+      "epoch": 0.5047521842674831,
+      "grad_norm": 0.15740440785884857,
+      "learning_rate": 4.495255054398575e-06,
+      "loss": 0.9221,
+      "step": 69730
+    },
+    {
+      "epoch": 0.5048245709280694,
+      "grad_norm": 0.15507307648658752,
+      "learning_rate": 4.49518266773799e-06,
+      "loss": 0.9148,
+      "step": 69740
+    },
+    {
+      "epoch": 0.5048969575886556,
+      "grad_norm": 0.14979131519794464,
+      "learning_rate": 4.4951102810774034e-06,
+      "loss": 0.9056,
+      "step": 69750
+    },
+    {
+      "epoch": 0.5049693442492418,
+      "grad_norm": 0.1572350412607193,
+      "learning_rate": 4.495037894416817e-06,
+      "loss": 0.9233,
+      "step": 69760
+    },
+    {
+      "epoch": 0.505041730909828,
+      "grad_norm": 0.15610437095165253,
+      "learning_rate": 4.494965507756231e-06,
+      "loss": 0.9252,
+      "step": 69770
+    },
+    {
+      "epoch": 0.5051141175704141,
+      "grad_norm": 0.15537583827972412,
+      "learning_rate": 4.494893121095645e-06,
+      "loss": 0.9243,
+      "step": 69780
+    },
+    {
+      "epoch": 0.5051865042310003,
+      "grad_norm": 0.19823171198368073,
+      "learning_rate": 4.494820734435059e-06,
+      "loss": 0.9286,
+      "step": 69790
+    },
+    {
+      "epoch": 0.5052588908915865,
+      "grad_norm": 0.15276482701301575,
+      "learning_rate": 4.494748347774472e-06,
+      "loss": 0.9221,
+      "step": 69800
+    },
+    {
+      "epoch": 0.5053312775521727,
+      "grad_norm": 0.15829552710056305,
+      "learning_rate": 4.494675961113886e-06,
+      "loss": 0.9058,
+      "step": 69810
+    },
+    {
+      "epoch": 0.5054036642127588,
+      "grad_norm": 0.1616862267255783,
+      "learning_rate": 4.4946035744533004e-06,
+      "loss": 0.9224,
+      "step": 69820
+    },
+    {
+      "epoch": 0.505476050873345,
+      "grad_norm": 0.15852747857570648,
+      "learning_rate": 4.494531187792714e-06,
+      "loss": 0.9288,
+      "step": 69830
+    },
+    {
+      "epoch": 0.5055484375339312,
+      "grad_norm": 0.16059504449367523,
+      "learning_rate": 4.494458801132128e-06,
+      "loss": 0.9153,
+      "step": 69840
+    },
+    {
+      "epoch": 0.5056208241945175,
+      "grad_norm": 0.16823314130306244,
+      "learning_rate": 4.494386414471541e-06,
+      "loss": 0.9232,
+      "step": 69850
+    },
+    {
+      "epoch": 0.5056932108551037,
+      "grad_norm": 0.14530618488788605,
+      "learning_rate": 4.494314027810956e-06,
+      "loss": 0.9221,
+      "step": 69860
+    },
+    {
+      "epoch": 0.5057655975156898,
+      "grad_norm": 0.15647926926612854,
+      "learning_rate": 4.494241641150369e-06,
+      "loss": 0.9297,
+      "step": 69870
+    },
+    {
+      "epoch": 0.505837984176276,
+      "grad_norm": 0.18810240924358368,
+      "learning_rate": 4.494169254489783e-06,
+      "loss": 0.9132,
+      "step": 69880
+    },
+    {
+      "epoch": 0.5059103708368622,
+      "grad_norm": 0.16139045357704163,
+      "learning_rate": 4.494096867829197e-06,
+      "loss": 0.9025,
+      "step": 69890
+    },
+    {
+      "epoch": 0.5059827574974484,
+      "grad_norm": 0.14970675110816956,
+      "learning_rate": 4.494024481168611e-06,
+      "loss": 0.9192,
+      "step": 69900
+    },
+    {
+      "epoch": 0.5060551441580345,
+      "grad_norm": 0.17512159049510956,
+      "learning_rate": 4.493952094508025e-06,
+      "loss": 0.9159,
+      "step": 69910
+    },
+    {
+      "epoch": 0.5061275308186207,
+      "grad_norm": 0.1656593680381775,
+      "learning_rate": 4.493879707847438e-06,
+      "loss": 0.9287,
+      "step": 69920
+    },
+    {
+      "epoch": 0.5061999174792069,
+      "grad_norm": 0.14336960017681122,
+      "learning_rate": 4.493807321186852e-06,
+      "loss": 0.913,
+      "step": 69930
+    },
+    {
+      "epoch": 0.5062723041397931,
+      "grad_norm": 0.15477393567562103,
+      "learning_rate": 4.493734934526266e-06,
+      "loss": 0.9236,
+      "step": 69940
+    },
+    {
+      "epoch": 0.5063446908003794,
+      "grad_norm": 0.16996079683303833,
+      "learning_rate": 4.49366254786568e-06,
+      "loss": 0.9288,
+      "step": 69950
+    },
+    {
+      "epoch": 0.5064170774609655,
+      "grad_norm": 0.16285812854766846,
+      "learning_rate": 4.493590161205094e-06,
+      "loss": 0.9243,
+      "step": 69960
+    },
+    {
+      "epoch": 0.5064894641215517,
+      "grad_norm": 0.16963927447795868,
+      "learning_rate": 4.493517774544507e-06,
+      "loss": 0.9154,
+      "step": 69970
+    },
+    {
+      "epoch": 0.5065618507821379,
+      "grad_norm": 0.17112816870212555,
+      "learning_rate": 4.493445387883921e-06,
+      "loss": 0.923,
+      "step": 69980
+    },
+    {
+      "epoch": 0.5066342374427241,
+      "grad_norm": 0.17348773777484894,
+      "learning_rate": 4.493373001223335e-06,
+      "loss": 0.9348,
+      "step": 69990
+    },
+    {
+      "epoch": 0.5067066241033102,
+      "grad_norm": 0.16326965391635895,
+      "learning_rate": 4.493300614562749e-06,
+      "loss": 0.9317,
+      "step": 70000
+    },
+    {
+      "epoch": 0.5067790107638964,
+      "grad_norm": 0.16024067997932434,
+      "learning_rate": 4.4932282279021625e-06,
+      "loss": 0.9252,
+      "step": 70010
+    },
+    {
+      "epoch": 0.5068513974244826,
+      "grad_norm": 0.1615953892469406,
+      "learning_rate": 4.493155841241576e-06,
+      "loss": 0.9135,
+      "step": 70020
+    },
+    {
+      "epoch": 0.5069237840850688,
+      "grad_norm": 0.14714552462100983,
+      "learning_rate": 4.493083454580991e-06,
+      "loss": 0.9307,
+      "step": 70030
+    },
+    {
+      "epoch": 0.506996170745655,
+      "grad_norm": 0.15000593662261963,
+      "learning_rate": 4.493011067920404e-06,
+      "loss": 0.9198,
+      "step": 70040
+    },
+    {
+      "epoch": 0.5070685574062411,
+      "grad_norm": 0.16868354380130768,
+      "learning_rate": 4.492938681259818e-06,
+      "loss": 0.9192,
+      "step": 70050
+    },
+    {
+      "epoch": 0.5071409440668274,
+      "grad_norm": 0.1756649762392044,
+      "learning_rate": 4.4928662945992315e-06,
+      "loss": 0.9145,
+      "step": 70060
+    },
+    {
+      "epoch": 0.5072133307274136,
+      "grad_norm": 0.14338277280330658,
+      "learning_rate": 4.492793907938646e-06,
+      "loss": 0.9162,
+      "step": 70070
+    },
+    {
+      "epoch": 0.5072857173879998,
+      "grad_norm": 0.15879562497138977,
+      "learning_rate": 4.4927215212780596e-06,
+      "loss": 0.9136,
+      "step": 70080
+    },
+    {
+      "epoch": 0.507358104048586,
+      "grad_norm": 0.16323305666446686,
+      "learning_rate": 4.492649134617473e-06,
+      "loss": 0.9207,
+      "step": 70090
+    },
+    {
+      "epoch": 0.5074304907091721,
+      "grad_norm": 0.15792213380336761,
+      "learning_rate": 4.492576747956887e-06,
+      "loss": 0.9112,
+      "step": 70100
+    },
+    {
+      "epoch": 0.5075028773697583,
+      "grad_norm": 0.15665297210216522,
+      "learning_rate": 4.4925043612963e-06,
+      "loss": 0.9182,
+      "step": 70110
+    },
+    {
+      "epoch": 0.5075752640303445,
+      "grad_norm": 0.16642248630523682,
+      "learning_rate": 4.492431974635714e-06,
+      "loss": 0.9249,
+      "step": 70120
+    },
+    {
+      "epoch": 0.5076476506909307,
+      "grad_norm": 0.1554255336523056,
+      "learning_rate": 4.492359587975128e-06,
+      "loss": 0.9203,
+      "step": 70130
+    },
+    {
+      "epoch": 0.5077200373515168,
+      "grad_norm": 0.17137949168682098,
+      "learning_rate": 4.492287201314542e-06,
+      "loss": 0.9248,
+      "step": 70140
+    },
+    {
+      "epoch": 0.507792424012103,
+      "grad_norm": 0.16739679872989655,
+      "learning_rate": 4.492214814653956e-06,
+      "loss": 0.9192,
+      "step": 70150
+    },
+    {
+      "epoch": 0.5078648106726892,
+      "grad_norm": 0.16433091461658478,
+      "learning_rate": 4.492142427993369e-06,
+      "loss": 0.919,
+      "step": 70160
+    },
+    {
+      "epoch": 0.5079371973332755,
+      "grad_norm": 0.16552671790122986,
+      "learning_rate": 4.492070041332783e-06,
+      "loss": 0.9291,
+      "step": 70170
+    },
+    {
+      "epoch": 0.5080095839938616,
+      "grad_norm": 0.15935184061527252,
+      "learning_rate": 4.491997654672197e-06,
+      "loss": 0.9255,
+      "step": 70180
+    },
+    {
+      "epoch": 0.5080819706544478,
+      "grad_norm": 0.16075530648231506,
+      "learning_rate": 4.491925268011611e-06,
+      "loss": 0.9286,
+      "step": 70190
+    },
+    {
+      "epoch": 0.508154357315034,
+      "grad_norm": 0.16693368554115295,
+      "learning_rate": 4.491852881351025e-06,
+      "loss": 0.9312,
+      "step": 70200
+    },
+    {
+      "epoch": 0.5082267439756202,
+      "grad_norm": 0.1627953052520752,
+      "learning_rate": 4.491780494690438e-06,
+      "loss": 0.914,
+      "step": 70210
+    },
+    {
+      "epoch": 0.5082991306362064,
+      "grad_norm": 0.17071197926998138,
+      "learning_rate": 4.491708108029853e-06,
+      "loss": 0.9145,
+      "step": 70220
+    },
+    {
+      "epoch": 0.5083715172967925,
+      "grad_norm": 0.1666022539138794,
+      "learning_rate": 4.491635721369266e-06,
+      "loss": 0.9181,
+      "step": 70230
+    },
+    {
+      "epoch": 0.5084439039573787,
+      "grad_norm": 0.16290847957134247,
+      "learning_rate": 4.49156333470868e-06,
+      "loss": 0.9289,
+      "step": 70240
+    },
+    {
+      "epoch": 0.5085162906179649,
+      "grad_norm": 0.2862056791782379,
+      "learning_rate": 4.4914909480480936e-06,
+      "loss": 0.9133,
+      "step": 70250
+    },
+    {
+      "epoch": 0.5085886772785511,
+      "grad_norm": 0.1533055603504181,
+      "learning_rate": 4.491418561387508e-06,
+      "loss": 0.9179,
+      "step": 70260
+    },
+    {
+      "epoch": 0.5086610639391373,
+      "grad_norm": 0.16260363161563873,
+      "learning_rate": 4.491346174726922e-06,
+      "loss": 0.9203,
+      "step": 70270
+    },
+    {
+      "epoch": 0.5087334505997235,
+      "grad_norm": 0.17325079441070557,
+      "learning_rate": 4.491273788066335e-06,
+      "loss": 0.9229,
+      "step": 70280
+    },
+    {
+      "epoch": 0.5088058372603097,
+      "grad_norm": 0.17838117480278015,
+      "learning_rate": 4.491201401405749e-06,
+      "loss": 0.9184,
+      "step": 70290
+    },
+    {
+      "epoch": 0.5088782239208959,
+      "grad_norm": 0.17130929231643677,
+      "learning_rate": 4.491129014745163e-06,
+      "loss": 0.9226,
+      "step": 70300
+    },
+    {
+      "epoch": 0.508950610581482,
+      "grad_norm": 0.1562015265226364,
+      "learning_rate": 4.491056628084577e-06,
+      "loss": 0.9384,
+      "step": 70310
+    },
+    {
+      "epoch": 0.5090229972420682,
+      "grad_norm": 0.1642456352710724,
+      "learning_rate": 4.490984241423991e-06,
+      "loss": 0.9392,
+      "step": 70320
+    },
+    {
+      "epoch": 0.5090953839026544,
+      "grad_norm": 0.16102585196495056,
+      "learning_rate": 4.490911854763404e-06,
+      "loss": 0.9269,
+      "step": 70330
+    },
+    {
+      "epoch": 0.5091677705632406,
+      "grad_norm": 0.16825130581855774,
+      "learning_rate": 4.490839468102819e-06,
+      "loss": 0.9312,
+      "step": 70340
+    },
+    {
+      "epoch": 0.5092401572238268,
+      "grad_norm": 0.18951553106307983,
+      "learning_rate": 4.490767081442232e-06,
+      "loss": 0.9149,
+      "step": 70350
+    },
+    {
+      "epoch": 0.5093125438844129,
+      "grad_norm": 0.1539865881204605,
+      "learning_rate": 4.490694694781646e-06,
+      "loss": 0.9176,
+      "step": 70360
+    },
+    {
+      "epoch": 0.5093849305449991,
+      "grad_norm": 0.16370318830013275,
+      "learning_rate": 4.4906223081210595e-06,
+      "loss": 0.9208,
+      "step": 70370
+    },
+    {
+      "epoch": 0.5094573172055854,
+      "grad_norm": 0.1988501250743866,
+      "learning_rate": 4.490549921460474e-06,
+      "loss": 0.9306,
+      "step": 70380
+    },
+    {
+      "epoch": 0.5095297038661716,
+      "grad_norm": 0.15401297807693481,
+      "learning_rate": 4.490477534799888e-06,
+      "loss": 0.931,
+      "step": 70390
+    },
+    {
+      "epoch": 0.5096020905267578,
+      "grad_norm": 0.15856841206550598,
+      "learning_rate": 4.490405148139301e-06,
+      "loss": 0.9152,
+      "step": 70400
+    },
+    {
+      "epoch": 0.5096744771873439,
+      "grad_norm": 0.15859901905059814,
+      "learning_rate": 4.490332761478715e-06,
+      "loss": 0.9092,
+      "step": 70410
+    },
+    {
+      "epoch": 0.5097468638479301,
+      "grad_norm": 0.16617213189601898,
+      "learning_rate": 4.490260374818129e-06,
+      "loss": 0.9061,
+      "step": 70420
+    },
+    {
+      "epoch": 0.5098192505085163,
+      "grad_norm": 0.159602552652359,
+      "learning_rate": 4.490187988157543e-06,
+      "loss": 0.9199,
+      "step": 70430
+    },
+    {
+      "epoch": 0.5098916371691025,
+      "grad_norm": 0.17478512227535248,
+      "learning_rate": 4.4901156014969565e-06,
+      "loss": 0.9304,
+      "step": 70440
+    },
+    {
+      "epoch": 0.5099640238296886,
+      "grad_norm": 0.1556614488363266,
+      "learning_rate": 4.49004321483637e-06,
+      "loss": 0.9188,
+      "step": 70450
+    },
+    {
+      "epoch": 0.5100364104902748,
+      "grad_norm": 0.15821588039398193,
+      "learning_rate": 4.489970828175785e-06,
+      "loss": 0.9139,
+      "step": 70460
+    },
+    {
+      "epoch": 0.510108797150861,
+      "grad_norm": 0.1509390026330948,
+      "learning_rate": 4.489898441515198e-06,
+      "loss": 0.9246,
+      "step": 70470
+    },
+    {
+      "epoch": 0.5101811838114473,
+      "grad_norm": 0.15749859809875488,
+      "learning_rate": 4.489826054854612e-06,
+      "loss": 0.9151,
+      "step": 70480
+    },
+    {
+      "epoch": 0.5102535704720335,
+      "grad_norm": 0.1502166986465454,
+      "learning_rate": 4.4897536681940254e-06,
+      "loss": 0.9084,
+      "step": 70490
+    },
+    {
+      "epoch": 0.5103259571326196,
+      "grad_norm": 0.15976519882678986,
+      "learning_rate": 4.48968128153344e-06,
+      "loss": 0.9106,
+      "step": 70500
+    },
+    {
+      "epoch": 0.5103983437932058,
+      "grad_norm": 0.15369580686092377,
+      "learning_rate": 4.4896088948728535e-06,
+      "loss": 0.9203,
+      "step": 70510
+    },
+    {
+      "epoch": 0.510470730453792,
+      "grad_norm": 0.17383261024951935,
+      "learning_rate": 4.489536508212267e-06,
+      "loss": 0.9298,
+      "step": 70520
+    },
+    {
+      "epoch": 0.5105431171143782,
+      "grad_norm": 0.1546318680047989,
+      "learning_rate": 4.489464121551681e-06,
+      "loss": 0.9357,
+      "step": 70530
+    },
+    {
+      "epoch": 0.5106155037749643,
+      "grad_norm": 0.15537676215171814,
+      "learning_rate": 4.489391734891095e-06,
+      "loss": 0.9096,
+      "step": 70540
+    },
+    {
+      "epoch": 0.5106878904355505,
+      "grad_norm": 0.14807634055614471,
+      "learning_rate": 4.489319348230509e-06,
+      "loss": 0.9225,
+      "step": 70550
+    },
+    {
+      "epoch": 0.5107602770961367,
+      "grad_norm": 0.17068465054035187,
+      "learning_rate": 4.4892469615699224e-06,
+      "loss": 0.9229,
+      "step": 70560
+    },
+    {
+      "epoch": 0.5108326637567229,
+      "grad_norm": 0.16985973715782166,
+      "learning_rate": 4.489174574909336e-06,
+      "loss": 0.918,
+      "step": 70570
+    },
+    {
+      "epoch": 0.510905050417309,
+      "grad_norm": 0.1610952913761139,
+      "learning_rate": 4.48910218824875e-06,
+      "loss": 0.9191,
+      "step": 70580
+    },
+    {
+      "epoch": 0.5109774370778953,
+      "grad_norm": 0.19032908976078033,
+      "learning_rate": 4.489029801588164e-06,
+      "loss": 0.9158,
+      "step": 70590
+    },
+    {
+      "epoch": 0.5110498237384815,
+      "grad_norm": 0.1557486355304718,
+      "learning_rate": 4.488957414927578e-06,
+      "loss": 0.9177,
+      "step": 70600
+    },
+    {
+      "epoch": 0.5111222103990677,
+      "grad_norm": 0.15214890241622925,
+      "learning_rate": 4.488885028266991e-06,
+      "loss": 0.9175,
+      "step": 70610
+    },
+    {
+      "epoch": 0.5111945970596539,
+      "grad_norm": 0.16953812539577484,
+      "learning_rate": 4.488812641606405e-06,
+      "loss": 0.9178,
+      "step": 70620
+    },
+    {
+      "epoch": 0.51126698372024,
+      "grad_norm": 0.16255024075508118,
+      "learning_rate": 4.4887402549458195e-06,
+      "loss": 0.9245,
+      "step": 70630
+    },
+    {
+      "epoch": 0.5113393703808262,
+      "grad_norm": 0.15664204955101013,
+      "learning_rate": 4.488667868285232e-06,
+      "loss": 0.9172,
+      "step": 70640
+    },
+    {
+      "epoch": 0.5114117570414124,
+      "grad_norm": 0.1526808738708496,
+      "learning_rate": 4.488595481624647e-06,
+      "loss": 0.9152,
+      "step": 70650
+    },
+    {
+      "epoch": 0.5114841437019986,
+      "grad_norm": 0.18234345316886902,
+      "learning_rate": 4.48852309496406e-06,
+      "loss": 0.9152,
+      "step": 70660
+    },
+    {
+      "epoch": 0.5115565303625847,
+      "grad_norm": 0.15891316533088684,
+      "learning_rate": 4.488450708303474e-06,
+      "loss": 0.9184,
+      "step": 70670
+    },
+    {
+      "epoch": 0.5116289170231709,
+      "grad_norm": 0.15766972303390503,
+      "learning_rate": 4.4883783216428875e-06,
+      "loss": 0.9197,
+      "step": 70680
+    },
+    {
+      "epoch": 0.5117013036837571,
+      "grad_norm": 0.16315676271915436,
+      "learning_rate": 4.488305934982302e-06,
+      "loss": 0.9245,
+      "step": 70690
+    },
+    {
+      "epoch": 0.5117736903443434,
+      "grad_norm": 0.15983930230140686,
+      "learning_rate": 4.488233548321716e-06,
+      "loss": 0.9192,
+      "step": 70700
+    },
+    {
+      "epoch": 0.5118460770049296,
+      "grad_norm": 0.14471487700939178,
+      "learning_rate": 4.488161161661129e-06,
+      "loss": 0.9174,
+      "step": 70710
+    },
+    {
+      "epoch": 0.5119184636655157,
+      "grad_norm": 0.1613370180130005,
+      "learning_rate": 4.488088775000543e-06,
+      "loss": 0.9036,
+      "step": 70720
+    },
+    {
+      "epoch": 0.5119908503261019,
+      "grad_norm": 0.16052919626235962,
+      "learning_rate": 4.488016388339957e-06,
+      "loss": 0.9103,
+      "step": 70730
+    },
+    {
+      "epoch": 0.5120632369866881,
+      "grad_norm": 0.16880005598068237,
+      "learning_rate": 4.487944001679371e-06,
+      "loss": 0.9251,
+      "step": 70740
+    },
+    {
+      "epoch": 0.5121356236472743,
+      "grad_norm": 0.1468336135149002,
+      "learning_rate": 4.4878716150187845e-06,
+      "loss": 0.9009,
+      "step": 70750
+    },
+    {
+      "epoch": 0.5122080103078605,
+      "grad_norm": 0.16562113165855408,
+      "learning_rate": 4.487799228358198e-06,
+      "loss": 0.9216,
+      "step": 70760
+    },
+    {
+      "epoch": 0.5122803969684466,
+      "grad_norm": 0.15913574397563934,
+      "learning_rate": 4.487726841697612e-06,
+      "loss": 0.9216,
+      "step": 70770
+    },
+    {
+      "epoch": 0.5123527836290328,
+      "grad_norm": 0.1655045747756958,
+      "learning_rate": 4.487654455037026e-06,
+      "loss": 0.9086,
+      "step": 70780
+    },
+    {
+      "epoch": 0.512425170289619,
+      "grad_norm": 0.15185129642486572,
+      "learning_rate": 4.48758206837644e-06,
+      "loss": 0.9242,
+      "step": 70790
+    },
+    {
+      "epoch": 0.5124975569502053,
+      "grad_norm": 0.1671728938817978,
+      "learning_rate": 4.4875096817158535e-06,
+      "loss": 0.9166,
+      "step": 70800
+    },
+    {
+      "epoch": 0.5125699436107914,
+      "grad_norm": 0.1865541785955429,
+      "learning_rate": 4.487437295055267e-06,
+      "loss": 0.9236,
+      "step": 70810
+    },
+    {
+      "epoch": 0.5126423302713776,
+      "grad_norm": 0.1669655591249466,
+      "learning_rate": 4.4873649083946816e-06,
+      "loss": 0.9196,
+      "step": 70820
+    },
+    {
+      "epoch": 0.5127147169319638,
+      "grad_norm": 0.17508293688297272,
+      "learning_rate": 4.487292521734095e-06,
+      "loss": 0.9211,
+      "step": 70830
+    },
+    {
+      "epoch": 0.51278710359255,
+      "grad_norm": 0.1667715311050415,
+      "learning_rate": 4.487220135073509e-06,
+      "loss": 0.9144,
+      "step": 70840
+    },
+    {
+      "epoch": 0.5128594902531362,
+      "grad_norm": 0.17849986255168915,
+      "learning_rate": 4.487147748412922e-06,
+      "loss": 0.9205,
+      "step": 70850
+    },
+    {
+      "epoch": 0.5129318769137223,
+      "grad_norm": 0.15619997680187225,
+      "learning_rate": 4.487075361752337e-06,
+      "loss": 0.9379,
+      "step": 70860
+    },
+    {
+      "epoch": 0.5130042635743085,
+      "grad_norm": 0.16368311643600464,
+      "learning_rate": 4.4870029750917505e-06,
+      "loss": 0.9072,
+      "step": 70870
+    },
+    {
+      "epoch": 0.5130766502348947,
+      "grad_norm": 0.16930024325847626,
+      "learning_rate": 4.486930588431164e-06,
+      "loss": 0.91,
+      "step": 70880
+    },
+    {
+      "epoch": 0.5131490368954809,
+      "grad_norm": 0.17948120832443237,
+      "learning_rate": 4.486858201770578e-06,
+      "loss": 0.9104,
+      "step": 70890
+    },
+    {
+      "epoch": 0.513221423556067,
+      "grad_norm": 0.19163572788238525,
+      "learning_rate": 4.486785815109992e-06,
+      "loss": 0.8944,
+      "step": 70900
+    },
+    {
+      "epoch": 0.5132938102166533,
+      "grad_norm": 0.15345646440982819,
+      "learning_rate": 4.486713428449406e-06,
+      "loss": 0.9268,
+      "step": 70910
+    },
+    {
+      "epoch": 0.5133661968772395,
+      "grad_norm": 0.16913288831710815,
+      "learning_rate": 4.486641041788819e-06,
+      "loss": 0.9235,
+      "step": 70920
+    },
+    {
+      "epoch": 0.5134385835378257,
+      "grad_norm": 0.17389893531799316,
+      "learning_rate": 4.486568655128233e-06,
+      "loss": 0.9248,
+      "step": 70930
+    },
+    {
+      "epoch": 0.5135109701984119,
+      "grad_norm": 0.16675566136837006,
+      "learning_rate": 4.4864962684676475e-06,
+      "loss": 0.9113,
+      "step": 70940
+    },
+    {
+      "epoch": 0.513583356858998,
+      "grad_norm": 0.15058213472366333,
+      "learning_rate": 4.486423881807061e-06,
+      "loss": 0.9248,
+      "step": 70950
+    },
+    {
+      "epoch": 0.5136557435195842,
+      "grad_norm": 0.33218497037887573,
+      "learning_rate": 4.486351495146475e-06,
+      "loss": 0.9035,
+      "step": 70960
+    },
+    {
+      "epoch": 0.5137281301801704,
+      "grad_norm": 0.1726812720298767,
+      "learning_rate": 4.486279108485888e-06,
+      "loss": 0.904,
+      "step": 70970
+    },
+    {
+      "epoch": 0.5138005168407566,
+      "grad_norm": 0.1540847271680832,
+      "learning_rate": 4.486206721825303e-06,
+      "loss": 0.9153,
+      "step": 70980
+    },
+    {
+      "epoch": 0.5138729035013427,
+      "grad_norm": 0.1801622062921524,
+      "learning_rate": 4.486134335164716e-06,
+      "loss": 0.9145,
+      "step": 70990
+    },
+    {
+      "epoch": 0.5139452901619289,
+      "grad_norm": 0.14557209610939026,
+      "learning_rate": 4.48606194850413e-06,
+      "loss": 0.9214,
+      "step": 71000
+    },
+    {
+      "epoch": 0.5140176768225152,
+      "grad_norm": 0.1920100599527359,
+      "learning_rate": 4.485989561843544e-06,
+      "loss": 0.9251,
+      "step": 71010
+    },
+    {
+      "epoch": 0.5140900634831014,
+      "grad_norm": 0.1555316001176834,
+      "learning_rate": 4.485917175182958e-06,
+      "loss": 0.9179,
+      "step": 71020
+    },
+    {
+      "epoch": 0.5141624501436876,
+      "grad_norm": 0.16274884343147278,
+      "learning_rate": 4.485844788522372e-06,
+      "loss": 0.9098,
+      "step": 71030
+    },
+    {
+      "epoch": 0.5142348368042737,
+      "grad_norm": 0.17160306870937347,
+      "learning_rate": 4.485772401861785e-06,
+      "loss": 0.918,
+      "step": 71040
+    },
+    {
+      "epoch": 0.5143072234648599,
+      "grad_norm": 0.1791687160730362,
+      "learning_rate": 4.485700015201199e-06,
+      "loss": 0.921,
+      "step": 71050
+    },
+    {
+      "epoch": 0.5143796101254461,
+      "grad_norm": 0.1617984026670456,
+      "learning_rate": 4.485627628540613e-06,
+      "loss": 0.9106,
+      "step": 71060
+    },
+    {
+      "epoch": 0.5144519967860323,
+      "grad_norm": 0.1553729772567749,
+      "learning_rate": 4.485555241880027e-06,
+      "loss": 0.9223,
+      "step": 71070
+    },
+    {
+      "epoch": 0.5145243834466184,
+      "grad_norm": 0.2470700591802597,
+      "learning_rate": 4.485482855219441e-06,
+      "loss": 0.9151,
+      "step": 71080
+    },
+    {
+      "epoch": 0.5145967701072046,
+      "grad_norm": 0.16699855029582977,
+      "learning_rate": 4.485410468558854e-06,
+      "loss": 0.9129,
+      "step": 71090
+    },
+    {
+      "epoch": 0.5146691567677908,
+      "grad_norm": 0.21680283546447754,
+      "learning_rate": 4.485338081898269e-06,
+      "loss": 0.9169,
+      "step": 71100
+    },
+    {
+      "epoch": 0.514741543428377,
+      "grad_norm": 0.16673427820205688,
+      "learning_rate": 4.485265695237682e-06,
+      "loss": 0.9362,
+      "step": 71110
+    },
+    {
+      "epoch": 0.5148139300889633,
+      "grad_norm": 0.1613384336233139,
+      "learning_rate": 4.485193308577096e-06,
+      "loss": 0.9249,
+      "step": 71120
+    },
+    {
+      "epoch": 0.5148863167495494,
+      "grad_norm": 0.1817154735326767,
+      "learning_rate": 4.48512092191651e-06,
+      "loss": 0.9287,
+      "step": 71130
+    },
+    {
+      "epoch": 0.5149587034101356,
+      "grad_norm": 0.1700638085603714,
+      "learning_rate": 4.485048535255924e-06,
+      "loss": 0.9126,
+      "step": 71140
+    },
+    {
+      "epoch": 0.5150310900707218,
+      "grad_norm": 0.17622961103916168,
+      "learning_rate": 4.484976148595338e-06,
+      "loss": 0.9131,
+      "step": 71150
+    },
+    {
+      "epoch": 0.515103476731308,
+      "grad_norm": 0.1659720242023468,
+      "learning_rate": 4.484903761934751e-06,
+      "loss": 0.909,
+      "step": 71160
+    },
+    {
+      "epoch": 0.5151758633918941,
+      "grad_norm": 0.16669756174087524,
+      "learning_rate": 4.484831375274165e-06,
+      "loss": 0.9083,
+      "step": 71170
+    },
+    {
+      "epoch": 0.5152482500524803,
+      "grad_norm": 0.16893939673900604,
+      "learning_rate": 4.4847589886135785e-06,
+      "loss": 0.9122,
+      "step": 71180
+    },
+    {
+      "epoch": 0.5153206367130665,
+      "grad_norm": 0.15991714596748352,
+      "learning_rate": 4.484686601952992e-06,
+      "loss": 0.915,
+      "step": 71190
+    },
+    {
+      "epoch": 0.5153930233736527,
+      "grad_norm": 0.15963132679462433,
+      "learning_rate": 4.484614215292406e-06,
+      "loss": 0.9193,
+      "step": 71200
+    },
+    {
+      "epoch": 0.5154654100342388,
+      "grad_norm": 0.1559552103281021,
+      "learning_rate": 4.48454182863182e-06,
+      "loss": 0.9138,
+      "step": 71210
+    },
+    {
+      "epoch": 0.515537796694825,
+      "grad_norm": 0.17159438133239746,
+      "learning_rate": 4.484469441971234e-06,
+      "loss": 0.9046,
+      "step": 71220
+    },
+    {
+      "epoch": 0.5156101833554113,
+      "grad_norm": 0.15353378653526306,
+      "learning_rate": 4.4843970553106474e-06,
+      "loss": 0.9144,
+      "step": 71230
+    },
+    {
+      "epoch": 0.5156825700159975,
+      "grad_norm": 0.16255958378314972,
+      "learning_rate": 4.484324668650061e-06,
+      "loss": 0.9019,
+      "step": 71240
+    },
+    {
+      "epoch": 0.5157549566765837,
+      "grad_norm": 0.16722968220710754,
+      "learning_rate": 4.4842522819894755e-06,
+      "loss": 0.9364,
+      "step": 71250
+    },
+    {
+      "epoch": 0.5158273433371698,
+      "grad_norm": 0.16965439915657043,
+      "learning_rate": 4.484179895328889e-06,
+      "loss": 0.9201,
+      "step": 71260
+    },
+    {
+      "epoch": 0.515899729997756,
+      "grad_norm": 0.16112767159938812,
+      "learning_rate": 4.484107508668303e-06,
+      "loss": 0.9206,
+      "step": 71270
+    },
+    {
+      "epoch": 0.5159721166583422,
+      "grad_norm": 0.1666184961795807,
+      "learning_rate": 4.484035122007716e-06,
+      "loss": 0.9133,
+      "step": 71280
+    },
+    {
+      "epoch": 0.5160445033189284,
+      "grad_norm": 0.16528722643852234,
+      "learning_rate": 4.483962735347131e-06,
+      "loss": 0.9199,
+      "step": 71290
+    },
+    {
+      "epoch": 0.5161168899795145,
+      "grad_norm": 0.1569664627313614,
+      "learning_rate": 4.4838903486865444e-06,
+      "loss": 0.9239,
+      "step": 71300
+    },
+    {
+      "epoch": 0.5161892766401007,
+      "grad_norm": 0.1667070984840393,
+      "learning_rate": 4.483817962025958e-06,
+      "loss": 0.9269,
+      "step": 71310
+    },
+    {
+      "epoch": 0.5162616633006869,
+      "grad_norm": 0.15985529124736786,
+      "learning_rate": 4.483745575365372e-06,
+      "loss": 0.913,
+      "step": 71320
+    },
+    {
+      "epoch": 0.5163340499612732,
+      "grad_norm": 0.17231838405132294,
+      "learning_rate": 4.483673188704786e-06,
+      "loss": 0.9246,
+      "step": 71330
+    },
+    {
+      "epoch": 0.5164064366218594,
+      "grad_norm": 0.16398270428180695,
+      "learning_rate": 4.4836008020442e-06,
+      "loss": 0.919,
+      "step": 71340
+    },
+    {
+      "epoch": 0.5164788232824455,
+      "grad_norm": 0.25354573130607605,
+      "learning_rate": 4.483528415383613e-06,
+      "loss": 0.9369,
+      "step": 71350
+    },
+    {
+      "epoch": 0.5165512099430317,
+      "grad_norm": 0.15637996792793274,
+      "learning_rate": 4.483456028723027e-06,
+      "loss": 0.9173,
+      "step": 71360
+    },
+    {
+      "epoch": 0.5166235966036179,
+      "grad_norm": 0.1540151685476303,
+      "learning_rate": 4.4833836420624415e-06,
+      "loss": 0.9172,
+      "step": 71370
+    },
+    {
+      "epoch": 0.5166959832642041,
+      "grad_norm": 0.1704026758670807,
+      "learning_rate": 4.483311255401855e-06,
+      "loss": 0.9171,
+      "step": 71380
+    },
+    {
+      "epoch": 0.5167683699247902,
+      "grad_norm": 0.152908593416214,
+      "learning_rate": 4.483238868741269e-06,
+      "loss": 0.9149,
+      "step": 71390
+    },
+    {
+      "epoch": 0.5168407565853764,
+      "grad_norm": 0.15788275003433228,
+      "learning_rate": 4.483166482080682e-06,
+      "loss": 0.9224,
+      "step": 71400
+    },
+    {
+      "epoch": 0.5169131432459626,
+      "grad_norm": 0.1816600263118744,
+      "learning_rate": 4.483094095420096e-06,
+      "loss": 0.9313,
+      "step": 71410
+    },
+    {
+      "epoch": 0.5169855299065488,
+      "grad_norm": 0.3124025762081146,
+      "learning_rate": 4.48302170875951e-06,
+      "loss": 0.9189,
+      "step": 71420
+    },
+    {
+      "epoch": 0.517057916567135,
+      "grad_norm": 0.16039884090423584,
+      "learning_rate": 4.482949322098924e-06,
+      "loss": 0.929,
+      "step": 71430
+    },
+    {
+      "epoch": 0.5171303032277212,
+      "grad_norm": 0.1553005576133728,
+      "learning_rate": 4.482876935438338e-06,
+      "loss": 0.9131,
+      "step": 71440
+    },
+    {
+      "epoch": 0.5172026898883074,
+      "grad_norm": 0.16902105510234833,
+      "learning_rate": 4.482804548777751e-06,
+      "loss": 0.9111,
+      "step": 71450
+    },
+    {
+      "epoch": 0.5172750765488936,
+      "grad_norm": 0.17077742516994476,
+      "learning_rate": 4.482732162117166e-06,
+      "loss": 0.9269,
+      "step": 71460
+    },
+    {
+      "epoch": 0.5173474632094798,
+      "grad_norm": 0.17665700614452362,
+      "learning_rate": 4.482659775456579e-06,
+      "loss": 0.908,
+      "step": 71470
+    },
+    {
+      "epoch": 0.517419849870066,
+      "grad_norm": 0.16787657141685486,
+      "learning_rate": 4.482587388795993e-06,
+      "loss": 0.9067,
+      "step": 71480
+    },
+    {
+      "epoch": 0.5174922365306521,
+      "grad_norm": 0.168588787317276,
+      "learning_rate": 4.4825150021354065e-06,
+      "loss": 0.9111,
+      "step": 71490
+    },
+    {
+      "epoch": 0.5175646231912383,
+      "grad_norm": 0.16920819878578186,
+      "learning_rate": 4.482442615474821e-06,
+      "loss": 0.9262,
+      "step": 71500
+    },
+    {
+      "epoch": 0.5176370098518245,
+      "grad_norm": 0.17827217280864716,
+      "learning_rate": 4.482370228814235e-06,
+      "loss": 0.9352,
+      "step": 71510
+    },
+    {
+      "epoch": 0.5177093965124107,
+      "grad_norm": 0.18853449821472168,
+      "learning_rate": 4.482297842153648e-06,
+      "loss": 0.9177,
+      "step": 71520
+    },
+    {
+      "epoch": 0.5177817831729968,
+      "grad_norm": 0.1763719618320465,
+      "learning_rate": 4.482225455493062e-06,
+      "loss": 0.9231,
+      "step": 71530
+    },
+    {
+      "epoch": 0.5178541698335831,
+      "grad_norm": 0.15667404234409332,
+      "learning_rate": 4.482153068832476e-06,
+      "loss": 0.9272,
+      "step": 71540
+    },
+    {
+      "epoch": 0.5179265564941693,
+      "grad_norm": 0.15414214134216309,
+      "learning_rate": 4.48208068217189e-06,
+      "loss": 0.9289,
+      "step": 71550
+    },
+    {
+      "epoch": 0.5179989431547555,
+      "grad_norm": 0.15999796986579895,
+      "learning_rate": 4.4820082955113036e-06,
+      "loss": 0.9122,
+      "step": 71560
+    },
+    {
+      "epoch": 0.5180713298153417,
+      "grad_norm": 0.1718224138021469,
+      "learning_rate": 4.481935908850717e-06,
+      "loss": 0.9087,
+      "step": 71570
+    },
+    {
+      "epoch": 0.5181437164759278,
+      "grad_norm": 0.14979562163352966,
+      "learning_rate": 4.481863522190132e-06,
+      "loss": 0.9066,
+      "step": 71580
+    },
+    {
+      "epoch": 0.518216103136514,
+      "grad_norm": 0.16340069472789764,
+      "learning_rate": 4.481791135529545e-06,
+      "loss": 0.923,
+      "step": 71590
+    },
+    {
+      "epoch": 0.5182884897971002,
+      "grad_norm": 0.15064361691474915,
+      "learning_rate": 4.481718748868959e-06,
+      "loss": 0.9123,
+      "step": 71600
+    },
+    {
+      "epoch": 0.5183608764576864,
+      "grad_norm": 0.16256332397460938,
+      "learning_rate": 4.4816463622083725e-06,
+      "loss": 0.9243,
+      "step": 71610
+    },
+    {
+      "epoch": 0.5184332631182725,
+      "grad_norm": 0.16027949750423431,
+      "learning_rate": 4.481573975547787e-06,
+      "loss": 0.9185,
+      "step": 71620
+    },
+    {
+      "epoch": 0.5185056497788587,
+      "grad_norm": 0.14856921136379242,
+      "learning_rate": 4.4815015888872006e-06,
+      "loss": 0.914,
+      "step": 71630
+    },
+    {
+      "epoch": 0.5185780364394449,
+      "grad_norm": 0.17712818086147308,
+      "learning_rate": 4.481429202226614e-06,
+      "loss": 0.9299,
+      "step": 71640
+    },
+    {
+      "epoch": 0.5186504231000312,
+      "grad_norm": 0.15580609440803528,
+      "learning_rate": 4.481356815566028e-06,
+      "loss": 0.924,
+      "step": 71650
+    },
+    {
+      "epoch": 0.5187228097606174,
+      "grad_norm": 0.15209463238716125,
+      "learning_rate": 4.481284428905442e-06,
+      "loss": 0.9227,
+      "step": 71660
+    },
+    {
+      "epoch": 0.5187951964212035,
+      "grad_norm": 0.1534406542778015,
+      "learning_rate": 4.481212042244856e-06,
+      "loss": 0.9206,
+      "step": 71670
+    },
+    {
+      "epoch": 0.5188675830817897,
+      "grad_norm": 0.1686408370733261,
+      "learning_rate": 4.4811396555842695e-06,
+      "loss": 0.9148,
+      "step": 71680
+    },
+    {
+      "epoch": 0.5189399697423759,
+      "grad_norm": 0.1665174961090088,
+      "learning_rate": 4.481067268923683e-06,
+      "loss": 0.9151,
+      "step": 71690
+    },
+    {
+      "epoch": 0.5190123564029621,
+      "grad_norm": 0.17360328137874603,
+      "learning_rate": 4.480994882263097e-06,
+      "loss": 0.9171,
+      "step": 71700
+    },
+    {
+      "epoch": 0.5190847430635482,
+      "grad_norm": 0.14976924657821655,
+      "learning_rate": 4.48092249560251e-06,
+      "loss": 0.9203,
+      "step": 71710
+    },
+    {
+      "epoch": 0.5191571297241344,
+      "grad_norm": 0.16571447253227234,
+      "learning_rate": 4.480850108941924e-06,
+      "loss": 0.9062,
+      "step": 71720
+    },
+    {
+      "epoch": 0.5192295163847206,
+      "grad_norm": 0.1702362298965454,
+      "learning_rate": 4.480777722281338e-06,
+      "loss": 0.9272,
+      "step": 71730
+    },
+    {
+      "epoch": 0.5193019030453068,
+      "grad_norm": 0.16296398639678955,
+      "learning_rate": 4.480705335620752e-06,
+      "loss": 0.915,
+      "step": 71740
+    },
+    {
+      "epoch": 0.5193742897058929,
+      "grad_norm": 0.15773184597492218,
+      "learning_rate": 4.480632948960166e-06,
+      "loss": 0.9159,
+      "step": 71750
+    },
+    {
+      "epoch": 0.5194466763664792,
+      "grad_norm": 0.15796466171741486,
+      "learning_rate": 4.480560562299579e-06,
+      "loss": 0.923,
+      "step": 71760
+    },
+    {
+      "epoch": 0.5195190630270654,
+      "grad_norm": 0.16592220962047577,
+      "learning_rate": 4.480488175638994e-06,
+      "loss": 0.9204,
+      "step": 71770
+    },
+    {
+      "epoch": 0.5195914496876516,
+      "grad_norm": 0.161960631608963,
+      "learning_rate": 4.480415788978407e-06,
+      "loss": 0.9235,
+      "step": 71780
+    },
+    {
+      "epoch": 0.5196638363482378,
+      "grad_norm": 0.17310741543769836,
+      "learning_rate": 4.480343402317821e-06,
+      "loss": 0.9062,
+      "step": 71790
+    },
+    {
+      "epoch": 0.5197362230088239,
+      "grad_norm": 0.1718437820672989,
+      "learning_rate": 4.4802710156572346e-06,
+      "loss": 0.9064,
+      "step": 71800
+    },
+    {
+      "epoch": 0.5198086096694101,
+      "grad_norm": 0.16727685928344727,
+      "learning_rate": 4.480198628996649e-06,
+      "loss": 0.9363,
+      "step": 71810
+    },
+    {
+      "epoch": 0.5198809963299963,
+      "grad_norm": 0.17579713463783264,
+      "learning_rate": 4.480126242336063e-06,
+      "loss": 0.9249,
+      "step": 71820
+    },
+    {
+      "epoch": 0.5199533829905825,
+      "grad_norm": 0.15877732634544373,
+      "learning_rate": 4.480053855675476e-06,
+      "loss": 0.9297,
+      "step": 71830
+    },
+    {
+      "epoch": 0.5200257696511686,
+      "grad_norm": 0.1620732545852661,
+      "learning_rate": 4.47998146901489e-06,
+      "loss": 0.9168,
+      "step": 71840
+    },
+    {
+      "epoch": 0.5200981563117548,
+      "grad_norm": 0.17274366319179535,
+      "learning_rate": 4.479909082354304e-06,
+      "loss": 0.9086,
+      "step": 71850
+    },
+    {
+      "epoch": 0.5201705429723411,
+      "grad_norm": 0.1550474315881729,
+      "learning_rate": 4.479836695693718e-06,
+      "loss": 0.9206,
+      "step": 71860
+    },
+    {
+      "epoch": 0.5202429296329273,
+      "grad_norm": 0.17657595872879028,
+      "learning_rate": 4.479764309033132e-06,
+      "loss": 0.9213,
+      "step": 71870
+    },
+    {
+      "epoch": 0.5203153162935135,
+      "grad_norm": 0.16544876992702484,
+      "learning_rate": 4.479691922372545e-06,
+      "loss": 0.929,
+      "step": 71880
+    },
+    {
+      "epoch": 0.5203877029540996,
+      "grad_norm": 0.1544915735721588,
+      "learning_rate": 4.47961953571196e-06,
+      "loss": 0.9057,
+      "step": 71890
+    },
+    {
+      "epoch": 0.5204600896146858,
+      "grad_norm": 0.15595407783985138,
+      "learning_rate": 4.479547149051373e-06,
+      "loss": 0.9144,
+      "step": 71900
+    },
+    {
+      "epoch": 0.520532476275272,
+      "grad_norm": 0.1735614836215973,
+      "learning_rate": 4.479474762390787e-06,
+      "loss": 0.9156,
+      "step": 71910
+    },
+    {
+      "epoch": 0.5206048629358582,
+      "grad_norm": 0.1538870483636856,
+      "learning_rate": 4.4794023757302005e-06,
+      "loss": 0.9117,
+      "step": 71920
+    },
+    {
+      "epoch": 0.5206772495964443,
+      "grad_norm": 0.15817460417747498,
+      "learning_rate": 4.479329989069615e-06,
+      "loss": 0.9233,
+      "step": 71930
+    },
+    {
+      "epoch": 0.5207496362570305,
+      "grad_norm": 0.16701272130012512,
+      "learning_rate": 4.479257602409029e-06,
+      "loss": 0.9218,
+      "step": 71940
+    },
+    {
+      "epoch": 0.5208220229176167,
+      "grad_norm": 0.15565438568592072,
+      "learning_rate": 4.479185215748442e-06,
+      "loss": 0.9113,
+      "step": 71950
+    },
+    {
+      "epoch": 0.5208944095782029,
+      "grad_norm": 0.15946054458618164,
+      "learning_rate": 4.479112829087856e-06,
+      "loss": 0.9165,
+      "step": 71960
+    },
+    {
+      "epoch": 0.5209667962387892,
+      "grad_norm": 0.1618921309709549,
+      "learning_rate": 4.47904044242727e-06,
+      "loss": 0.9241,
+      "step": 71970
+    },
+    {
+      "epoch": 0.5210391828993753,
+      "grad_norm": 0.19003738462924957,
+      "learning_rate": 4.478968055766684e-06,
+      "loss": 0.9101,
+      "step": 71980
+    },
+    {
+      "epoch": 0.5211115695599615,
+      "grad_norm": 0.16518519818782806,
+      "learning_rate": 4.4788956691060975e-06,
+      "loss": 0.9038,
+      "step": 71990
+    },
+    {
+      "epoch": 0.5211839562205477,
+      "grad_norm": 0.15592487156391144,
+      "learning_rate": 4.478823282445511e-06,
+      "loss": 0.9162,
+      "step": 72000
+    },
+    {
+      "epoch": 0.5212563428811339,
+      "grad_norm": 0.17761656641960144,
+      "learning_rate": 4.478750895784925e-06,
+      "loss": 0.9126,
+      "step": 72010
+    },
+    {
+      "epoch": 0.52132872954172,
+      "grad_norm": 0.1610282063484192,
+      "learning_rate": 4.478678509124339e-06,
+      "loss": 0.9199,
+      "step": 72020
+    },
+    {
+      "epoch": 0.5214011162023062,
+      "grad_norm": 0.16208316385746002,
+      "learning_rate": 4.478606122463753e-06,
+      "loss": 0.9129,
+      "step": 72030
+    },
+    {
+      "epoch": 0.5214735028628924,
+      "grad_norm": 0.17312684655189514,
+      "learning_rate": 4.4785337358031664e-06,
+      "loss": 0.9022,
+      "step": 72040
+    },
+    {
+      "epoch": 0.5215458895234786,
+      "grad_norm": 0.16080522537231445,
+      "learning_rate": 4.47846134914258e-06,
+      "loss": 0.908,
+      "step": 72050
+    },
+    {
+      "epoch": 0.5216182761840648,
+      "grad_norm": 0.17630934715270996,
+      "learning_rate": 4.4783889624819945e-06,
+      "loss": 0.9169,
+      "step": 72060
+    },
+    {
+      "epoch": 0.521690662844651,
+      "grad_norm": 0.1585136502981186,
+      "learning_rate": 4.478316575821408e-06,
+      "loss": 0.9079,
+      "step": 72070
+    },
+    {
+      "epoch": 0.5217630495052372,
+      "grad_norm": 0.15829670429229736,
+      "learning_rate": 4.478244189160822e-06,
+      "loss": 0.9209,
+      "step": 72080
+    },
+    {
+      "epoch": 0.5218354361658234,
+      "grad_norm": 0.1633504331111908,
+      "learning_rate": 4.478171802500235e-06,
+      "loss": 0.9202,
+      "step": 72090
+    },
+    {
+      "epoch": 0.5219078228264096,
+      "grad_norm": 0.1850529909133911,
+      "learning_rate": 4.47809941583965e-06,
+      "loss": 0.9244,
+      "step": 72100
+    },
+    {
+      "epoch": 0.5219802094869957,
+      "grad_norm": 0.16101692616939545,
+      "learning_rate": 4.4780270291790635e-06,
+      "loss": 0.9122,
+      "step": 72110
+    },
+    {
+      "epoch": 0.5220525961475819,
+      "grad_norm": 0.16224358975887299,
+      "learning_rate": 4.477954642518477e-06,
+      "loss": 0.9149,
+      "step": 72120
+    },
+    {
+      "epoch": 0.5221249828081681,
+      "grad_norm": 0.14560241997241974,
+      "learning_rate": 4.477882255857891e-06,
+      "loss": 0.9203,
+      "step": 72130
+    },
+    {
+      "epoch": 0.5221973694687543,
+      "grad_norm": 0.15428347885608673,
+      "learning_rate": 4.477809869197305e-06,
+      "loss": 0.9212,
+      "step": 72140
+    },
+    {
+      "epoch": 0.5222697561293405,
+      "grad_norm": 0.1555236577987671,
+      "learning_rate": 4.477737482536719e-06,
+      "loss": 0.9223,
+      "step": 72150
+    },
+    {
+      "epoch": 0.5223421427899266,
+      "grad_norm": 0.16414514183998108,
+      "learning_rate": 4.477665095876132e-06,
+      "loss": 0.9153,
+      "step": 72160
+    },
+    {
+      "epoch": 0.5224145294505128,
+      "grad_norm": 0.1530592143535614,
+      "learning_rate": 4.477592709215546e-06,
+      "loss": 0.9168,
+      "step": 72170
+    },
+    {
+      "epoch": 0.5224869161110991,
+      "grad_norm": 0.17093126475811005,
+      "learning_rate": 4.4775203225549605e-06,
+      "loss": 0.9175,
+      "step": 72180
+    },
+    {
+      "epoch": 0.5225593027716853,
+      "grad_norm": 0.16496798396110535,
+      "learning_rate": 4.477447935894374e-06,
+      "loss": 0.9173,
+      "step": 72190
+    },
+    {
+      "epoch": 0.5226316894322715,
+      "grad_norm": 0.2403537482023239,
+      "learning_rate": 4.477375549233788e-06,
+      "loss": 0.9239,
+      "step": 72200
+    },
+    {
+      "epoch": 0.5227040760928576,
+      "grad_norm": 0.1915457397699356,
+      "learning_rate": 4.477303162573201e-06,
+      "loss": 0.9279,
+      "step": 72210
+    },
+    {
+      "epoch": 0.5227764627534438,
+      "grad_norm": 0.1616457849740982,
+      "learning_rate": 4.477230775912616e-06,
+      "loss": 0.9078,
+      "step": 72220
+    },
+    {
+      "epoch": 0.52284884941403,
+      "grad_norm": 0.19186237454414368,
+      "learning_rate": 4.4771583892520285e-06,
+      "loss": 0.9307,
+      "step": 72230
+    },
+    {
+      "epoch": 0.5229212360746162,
+      "grad_norm": 0.16343003511428833,
+      "learning_rate": 4.477086002591442e-06,
+      "loss": 0.9148,
+      "step": 72240
+    },
+    {
+      "epoch": 0.5229936227352023,
+      "grad_norm": 0.15922591090202332,
+      "learning_rate": 4.477013615930857e-06,
+      "loss": 0.9101,
+      "step": 72250
+    },
+    {
+      "epoch": 0.5230660093957885,
+      "grad_norm": 0.1629839390516281,
+      "learning_rate": 4.47694122927027e-06,
+      "loss": 0.9133,
+      "step": 72260
+    },
+    {
+      "epoch": 0.5231383960563747,
+      "grad_norm": 0.1665726751089096,
+      "learning_rate": 4.476868842609684e-06,
+      "loss": 0.9105,
+      "step": 72270
+    },
+    {
+      "epoch": 0.5232107827169609,
+      "grad_norm": 0.22242015600204468,
+      "learning_rate": 4.4767964559490975e-06,
+      "loss": 0.9184,
+      "step": 72280
+    },
+    {
+      "epoch": 0.5232831693775472,
+      "grad_norm": 0.15692086517810822,
+      "learning_rate": 4.476724069288512e-06,
+      "loss": 0.9071,
+      "step": 72290
+    },
+    {
+      "epoch": 0.5233555560381333,
+      "grad_norm": 0.16244766116142273,
+      "learning_rate": 4.4766516826279255e-06,
+      "loss": 0.924,
+      "step": 72300
+    },
+    {
+      "epoch": 0.5234279426987195,
+      "grad_norm": 0.1637459546327591,
+      "learning_rate": 4.476579295967339e-06,
+      "loss": 0.9181,
+      "step": 72310
+    },
+    {
+      "epoch": 0.5235003293593057,
+      "grad_norm": 0.16626401245594025,
+      "learning_rate": 4.476506909306753e-06,
+      "loss": 0.9203,
+      "step": 72320
+    },
+    {
+      "epoch": 0.5235727160198919,
+      "grad_norm": 0.15717044472694397,
+      "learning_rate": 4.476434522646167e-06,
+      "loss": 0.9104,
+      "step": 72330
+    },
+    {
+      "epoch": 0.523645102680478,
+      "grad_norm": 0.15641281008720398,
+      "learning_rate": 4.476362135985581e-06,
+      "loss": 0.9045,
+      "step": 72340
+    },
+    {
+      "epoch": 0.5237174893410642,
+      "grad_norm": 0.15991607308387756,
+      "learning_rate": 4.4762897493249945e-06,
+      "loss": 0.9179,
+      "step": 72350
+    },
+    {
+      "epoch": 0.5237898760016504,
+      "grad_norm": 0.14962872862815857,
+      "learning_rate": 4.476217362664408e-06,
+      "loss": 0.9184,
+      "step": 72360
+    },
+    {
+      "epoch": 0.5238622626622366,
+      "grad_norm": 0.14683006703853607,
+      "learning_rate": 4.4761449760038226e-06,
+      "loss": 0.9173,
+      "step": 72370
+    },
+    {
+      "epoch": 0.5239346493228227,
+      "grad_norm": 0.17192301154136658,
+      "learning_rate": 4.476072589343236e-06,
+      "loss": 0.9299,
+      "step": 72380
+    },
+    {
+      "epoch": 0.524007035983409,
+      "grad_norm": 0.16342616081237793,
+      "learning_rate": 4.47600020268265e-06,
+      "loss": 0.9071,
+      "step": 72390
+    },
+    {
+      "epoch": 0.5240794226439952,
+      "grad_norm": 0.1558561623096466,
+      "learning_rate": 4.475927816022063e-06,
+      "loss": 0.9077,
+      "step": 72400
+    },
+    {
+      "epoch": 0.5241518093045814,
+      "grad_norm": 0.18777623772621155,
+      "learning_rate": 4.475855429361478e-06,
+      "loss": 0.912,
+      "step": 72410
+    },
+    {
+      "epoch": 0.5242241959651676,
+      "grad_norm": 0.16315610706806183,
+      "learning_rate": 4.4757830427008915e-06,
+      "loss": 0.9415,
+      "step": 72420
+    },
+    {
+      "epoch": 0.5242965826257537,
+      "grad_norm": 0.17795760929584503,
+      "learning_rate": 4.475710656040305e-06,
+      "loss": 0.9197,
+      "step": 72430
+    },
+    {
+      "epoch": 0.5243689692863399,
+      "grad_norm": 0.16951826214790344,
+      "learning_rate": 4.475638269379719e-06,
+      "loss": 0.908,
+      "step": 72440
+    },
+    {
+      "epoch": 0.5244413559469261,
+      "grad_norm": 0.18716560304164886,
+      "learning_rate": 4.475565882719133e-06,
+      "loss": 0.9293,
+      "step": 72450
+    },
+    {
+      "epoch": 0.5245137426075123,
+      "grad_norm": 0.24817276000976562,
+      "learning_rate": 4.475493496058547e-06,
+      "loss": 0.9103,
+      "step": 72460
+    },
+    {
+      "epoch": 0.5245861292680984,
+      "grad_norm": 0.15385325253009796,
+      "learning_rate": 4.47542110939796e-06,
+      "loss": 0.9171,
+      "step": 72470
+    },
+    {
+      "epoch": 0.5246585159286846,
+      "grad_norm": 0.15869303047657013,
+      "learning_rate": 4.475348722737374e-06,
+      "loss": 0.9223,
+      "step": 72480
+    },
+    {
+      "epoch": 0.5247309025892708,
+      "grad_norm": 0.14980655908584595,
+      "learning_rate": 4.4752763360767885e-06,
+      "loss": 0.9182,
+      "step": 72490
+    },
+    {
+      "epoch": 0.5248032892498571,
+      "grad_norm": 0.16237030923366547,
+      "learning_rate": 4.475203949416202e-06,
+      "loss": 0.9184,
+      "step": 72500
+    },
+    {
+      "epoch": 0.5248756759104433,
+      "grad_norm": 0.15368181467056274,
+      "learning_rate": 4.475131562755616e-06,
+      "loss": 0.9164,
+      "step": 72510
+    },
+    {
+      "epoch": 0.5249480625710294,
+      "grad_norm": 0.16227343678474426,
+      "learning_rate": 4.475059176095029e-06,
+      "loss": 0.9022,
+      "step": 72520
+    },
+    {
+      "epoch": 0.5250204492316156,
+      "grad_norm": 0.15485748648643494,
+      "learning_rate": 4.474986789434444e-06,
+      "loss": 0.9167,
+      "step": 72530
+    },
+    {
+      "epoch": 0.5250928358922018,
+      "grad_norm": 0.17310456931591034,
+      "learning_rate": 4.474914402773857e-06,
+      "loss": 0.928,
+      "step": 72540
+    },
+    {
+      "epoch": 0.525165222552788,
+      "grad_norm": 0.1546664834022522,
+      "learning_rate": 4.474842016113271e-06,
+      "loss": 0.9162,
+      "step": 72550
+    },
+    {
+      "epoch": 0.5252376092133741,
+      "grad_norm": 0.17586833238601685,
+      "learning_rate": 4.474769629452685e-06,
+      "loss": 0.9126,
+      "step": 72560
+    },
+    {
+      "epoch": 0.5253099958739603,
+      "grad_norm": 0.1783231645822525,
+      "learning_rate": 4.474697242792099e-06,
+      "loss": 0.9093,
+      "step": 72570
+    },
+    {
+      "epoch": 0.5253823825345465,
+      "grad_norm": 0.17169401049613953,
+      "learning_rate": 4.474624856131513e-06,
+      "loss": 0.9267,
+      "step": 72580
+    },
+    {
+      "epoch": 0.5254547691951327,
+      "grad_norm": 0.478100448846817,
+      "learning_rate": 4.474552469470926e-06,
+      "loss": 0.9089,
+      "step": 72590
+    },
+    {
+      "epoch": 0.5255271558557189,
+      "grad_norm": 0.1514614224433899,
+      "learning_rate": 4.47448008281034e-06,
+      "loss": 0.9243,
+      "step": 72600
+    },
+    {
+      "epoch": 0.5255995425163051,
+      "grad_norm": 0.15338928997516632,
+      "learning_rate": 4.4744076961497544e-06,
+      "loss": 0.9111,
+      "step": 72610
+    },
+    {
+      "epoch": 0.5256719291768913,
+      "grad_norm": 0.16090184450149536,
+      "learning_rate": 4.474335309489168e-06,
+      "loss": 0.9188,
+      "step": 72620
+    },
+    {
+      "epoch": 0.5257443158374775,
+      "grad_norm": 0.1639074832201004,
+      "learning_rate": 4.474262922828582e-06,
+      "loss": 0.9235,
+      "step": 72630
+    },
+    {
+      "epoch": 0.5258167024980637,
+      "grad_norm": 0.16050408780574799,
+      "learning_rate": 4.474190536167995e-06,
+      "loss": 0.9176,
+      "step": 72640
+    },
+    {
+      "epoch": 0.5258890891586498,
+      "grad_norm": 0.1604773849248886,
+      "learning_rate": 4.474118149507409e-06,
+      "loss": 0.9258,
+      "step": 72650
+    },
+    {
+      "epoch": 0.525961475819236,
+      "grad_norm": 0.15453247725963593,
+      "learning_rate": 4.474045762846823e-06,
+      "loss": 0.9273,
+      "step": 72660
+    },
+    {
+      "epoch": 0.5260338624798222,
+      "grad_norm": 0.17497363686561584,
+      "learning_rate": 4.473973376186237e-06,
+      "loss": 0.9219,
+      "step": 72670
+    },
+    {
+      "epoch": 0.5261062491404084,
+      "grad_norm": 0.15889976918697357,
+      "learning_rate": 4.473900989525651e-06,
+      "loss": 0.9325,
+      "step": 72680
+    },
+    {
+      "epoch": 0.5261786358009946,
+      "grad_norm": 0.18632720410823822,
+      "learning_rate": 4.473828602865064e-06,
+      "loss": 0.9188,
+      "step": 72690
+    },
+    {
+      "epoch": 0.5262510224615807,
+      "grad_norm": 0.16045477986335754,
+      "learning_rate": 4.473756216204479e-06,
+      "loss": 0.9261,
+      "step": 72700
+    },
+    {
+      "epoch": 0.526323409122167,
+      "grad_norm": 0.16817103326320648,
+      "learning_rate": 4.473683829543892e-06,
+      "loss": 0.9212,
+      "step": 72710
+    },
+    {
+      "epoch": 0.5263957957827532,
+      "grad_norm": 0.1566298007965088,
+      "learning_rate": 4.473611442883306e-06,
+      "loss": 0.9188,
+      "step": 72720
+    },
+    {
+      "epoch": 0.5264681824433394,
+      "grad_norm": 0.1550060659646988,
+      "learning_rate": 4.4735390562227195e-06,
+      "loss": 0.9263,
+      "step": 72730
+    },
+    {
+      "epoch": 0.5265405691039255,
+      "grad_norm": 0.15189586579799652,
+      "learning_rate": 4.473466669562134e-06,
+      "loss": 0.9202,
+      "step": 72740
+    },
+    {
+      "epoch": 0.5266129557645117,
+      "grad_norm": 0.14698752760887146,
+      "learning_rate": 4.473394282901548e-06,
+      "loss": 0.9105,
+      "step": 72750
+    },
+    {
+      "epoch": 0.5266853424250979,
+      "grad_norm": 0.16903573274612427,
+      "learning_rate": 4.473321896240961e-06,
+      "loss": 0.9135,
+      "step": 72760
+    },
+    {
+      "epoch": 0.5267577290856841,
+      "grad_norm": 0.16911180317401886,
+      "learning_rate": 4.473249509580375e-06,
+      "loss": 0.9249,
+      "step": 72770
+    },
+    {
+      "epoch": 0.5268301157462703,
+      "grad_norm": 0.1543751209974289,
+      "learning_rate": 4.4731771229197884e-06,
+      "loss": 0.9126,
+      "step": 72780
+    },
+    {
+      "epoch": 0.5269025024068564,
+      "grad_norm": 0.15658721327781677,
+      "learning_rate": 4.473104736259202e-06,
+      "loss": 0.9152,
+      "step": 72790
+    },
+    {
+      "epoch": 0.5269748890674426,
+      "grad_norm": 0.16438843309879303,
+      "learning_rate": 4.473032349598616e-06,
+      "loss": 0.9226,
+      "step": 72800
+    },
+    {
+      "epoch": 0.5270472757280288,
+      "grad_norm": 0.17810095846652985,
+      "learning_rate": 4.47295996293803e-06,
+      "loss": 0.9093,
+      "step": 72810
+    },
+    {
+      "epoch": 0.5271196623886151,
+      "grad_norm": 0.1657225638628006,
+      "learning_rate": 4.472887576277444e-06,
+      "loss": 0.9233,
+      "step": 72820
+    },
+    {
+      "epoch": 0.5271920490492013,
+      "grad_norm": 0.16005633771419525,
+      "learning_rate": 4.472815189616857e-06,
+      "loss": 0.9201,
+      "step": 72830
+    },
+    {
+      "epoch": 0.5272644357097874,
+      "grad_norm": 0.16548793017864227,
+      "learning_rate": 4.472742802956271e-06,
+      "loss": 0.9304,
+      "step": 72840
+    },
+    {
+      "epoch": 0.5273368223703736,
+      "grad_norm": 0.14762021601200104,
+      "learning_rate": 4.4726704162956855e-06,
+      "loss": 0.9137,
+      "step": 72850
+    },
+    {
+      "epoch": 0.5274092090309598,
+      "grad_norm": 0.17930978536605835,
+      "learning_rate": 4.472598029635099e-06,
+      "loss": 0.9203,
+      "step": 72860
+    },
+    {
+      "epoch": 0.527481595691546,
+      "grad_norm": 0.1988268494606018,
+      "learning_rate": 4.472525642974513e-06,
+      "loss": 0.9219,
+      "step": 72870
+    },
+    {
+      "epoch": 0.5275539823521321,
+      "grad_norm": 0.1670868992805481,
+      "learning_rate": 4.472453256313926e-06,
+      "loss": 0.9083,
+      "step": 72880
+    },
+    {
+      "epoch": 0.5276263690127183,
+      "grad_norm": 0.14923833310604095,
+      "learning_rate": 4.472380869653341e-06,
+      "loss": 0.9207,
+      "step": 72890
+    },
+    {
+      "epoch": 0.5276987556733045,
+      "grad_norm": 0.15868178009986877,
+      "learning_rate": 4.472308482992754e-06,
+      "loss": 0.9055,
+      "step": 72900
+    },
+    {
+      "epoch": 0.5277711423338907,
+      "grad_norm": 0.1550142616033554,
+      "learning_rate": 4.472236096332168e-06,
+      "loss": 0.906,
+      "step": 72910
+    },
+    {
+      "epoch": 0.527843528994477,
+      "grad_norm": 0.15354187786579132,
+      "learning_rate": 4.472163709671582e-06,
+      "loss": 0.9138,
+      "step": 72920
+    },
+    {
+      "epoch": 0.5279159156550631,
+      "grad_norm": 0.15525110065937042,
+      "learning_rate": 4.472091323010996e-06,
+      "loss": 0.9195,
+      "step": 72930
+    },
+    {
+      "epoch": 0.5279883023156493,
+      "grad_norm": 0.15831002593040466,
+      "learning_rate": 4.47201893635041e-06,
+      "loss": 0.9161,
+      "step": 72940
+    },
+    {
+      "epoch": 0.5280606889762355,
+      "grad_norm": 0.169021874666214,
+      "learning_rate": 4.471946549689823e-06,
+      "loss": 0.9256,
+      "step": 72950
+    },
+    {
+      "epoch": 0.5281330756368217,
+      "grad_norm": 0.14959047734737396,
+      "learning_rate": 4.471874163029237e-06,
+      "loss": 0.8988,
+      "step": 72960
+    },
+    {
+      "epoch": 0.5282054622974078,
+      "grad_norm": 0.1501626968383789,
+      "learning_rate": 4.471801776368651e-06,
+      "loss": 0.9179,
+      "step": 72970
+    },
+    {
+      "epoch": 0.528277848957994,
+      "grad_norm": 0.15244172513484955,
+      "learning_rate": 4.471729389708065e-06,
+      "loss": 0.9218,
+      "step": 72980
+    },
+    {
+      "epoch": 0.5283502356185802,
+      "grad_norm": 0.21361258625984192,
+      "learning_rate": 4.471657003047479e-06,
+      "loss": 0.9205,
+      "step": 72990
+    },
+    {
+      "epoch": 0.5284226222791664,
+      "grad_norm": 0.1592887043952942,
+      "learning_rate": 4.471584616386892e-06,
+      "loss": 0.9192,
+      "step": 73000
+    },
+    {
+      "epoch": 0.5284950089397525,
+      "grad_norm": 0.15822738409042358,
+      "learning_rate": 4.471512229726307e-06,
+      "loss": 0.9095,
+      "step": 73010
+    },
+    {
+      "epoch": 0.5285673956003387,
+      "grad_norm": 0.18526582419872284,
+      "learning_rate": 4.47143984306572e-06,
+      "loss": 0.922,
+      "step": 73020
+    },
+    {
+      "epoch": 0.528639782260925,
+      "grad_norm": 0.16115911304950714,
+      "learning_rate": 4.471367456405134e-06,
+      "loss": 0.9071,
+      "step": 73030
+    },
+    {
+      "epoch": 0.5287121689215112,
+      "grad_norm": 0.16934490203857422,
+      "learning_rate": 4.4712950697445475e-06,
+      "loss": 0.9206,
+      "step": 73040
+    },
+    {
+      "epoch": 0.5287845555820974,
+      "grad_norm": 0.15864363312721252,
+      "learning_rate": 4.471222683083962e-06,
+      "loss": 0.9344,
+      "step": 73050
+    },
+    {
+      "epoch": 0.5288569422426835,
+      "grad_norm": 0.16181667149066925,
+      "learning_rate": 4.471150296423376e-06,
+      "loss": 0.9177,
+      "step": 73060
+    },
+    {
+      "epoch": 0.5289293289032697,
+      "grad_norm": 0.15380777418613434,
+      "learning_rate": 4.471077909762789e-06,
+      "loss": 0.9181,
+      "step": 73070
+    },
+    {
+      "epoch": 0.5290017155638559,
+      "grad_norm": 0.1445995569229126,
+      "learning_rate": 4.471005523102203e-06,
+      "loss": 0.9232,
+      "step": 73080
+    },
+    {
+      "epoch": 0.5290741022244421,
+      "grad_norm": 0.1677917093038559,
+      "learning_rate": 4.470933136441617e-06,
+      "loss": 0.9231,
+      "step": 73090
+    },
+    {
+      "epoch": 0.5291464888850282,
+      "grad_norm": 0.19374777376651764,
+      "learning_rate": 4.470860749781031e-06,
+      "loss": 0.9182,
+      "step": 73100
+    },
+    {
+      "epoch": 0.5292188755456144,
+      "grad_norm": 0.18379537761211395,
+      "learning_rate": 4.4707883631204446e-06,
+      "loss": 0.9132,
+      "step": 73110
+    },
+    {
+      "epoch": 0.5292912622062006,
+      "grad_norm": 0.1719239056110382,
+      "learning_rate": 4.470715976459858e-06,
+      "loss": 0.9188,
+      "step": 73120
+    },
+    {
+      "epoch": 0.5293636488667868,
+      "grad_norm": 0.15979036688804626,
+      "learning_rate": 4.470643589799273e-06,
+      "loss": 0.9093,
+      "step": 73130
+    },
+    {
+      "epoch": 0.5294360355273731,
+      "grad_norm": 0.15309134125709534,
+      "learning_rate": 4.470571203138686e-06,
+      "loss": 0.9195,
+      "step": 73140
+    },
+    {
+      "epoch": 0.5295084221879592,
+      "grad_norm": 0.1547461748123169,
+      "learning_rate": 4.4704988164781e-06,
+      "loss": 0.9073,
+      "step": 73150
+    },
+    {
+      "epoch": 0.5295808088485454,
+      "grad_norm": 0.17491716146469116,
+      "learning_rate": 4.4704264298175135e-06,
+      "loss": 0.9356,
+      "step": 73160
+    },
+    {
+      "epoch": 0.5296531955091316,
+      "grad_norm": 0.15330342948436737,
+      "learning_rate": 4.470354043156928e-06,
+      "loss": 0.9124,
+      "step": 73170
+    },
+    {
+      "epoch": 0.5297255821697178,
+      "grad_norm": 0.1625756025314331,
+      "learning_rate": 4.4702816564963416e-06,
+      "loss": 0.9254,
+      "step": 73180
+    },
+    {
+      "epoch": 0.529797968830304,
+      "grad_norm": 0.1609082669019699,
+      "learning_rate": 4.470209269835755e-06,
+      "loss": 0.9108,
+      "step": 73190
+    },
+    {
+      "epoch": 0.5298703554908901,
+      "grad_norm": 0.15871219336986542,
+      "learning_rate": 4.470136883175169e-06,
+      "loss": 0.9182,
+      "step": 73200
+    },
+    {
+      "epoch": 0.5299427421514763,
+      "grad_norm": 0.20798322558403015,
+      "learning_rate": 4.470064496514583e-06,
+      "loss": 0.9201,
+      "step": 73210
+    },
+    {
+      "epoch": 0.5300151288120625,
+      "grad_norm": 0.15819989144802094,
+      "learning_rate": 4.469992109853997e-06,
+      "loss": 0.9164,
+      "step": 73220
+    },
+    {
+      "epoch": 0.5300875154726487,
+      "grad_norm": 0.16910409927368164,
+      "learning_rate": 4.4699197231934105e-06,
+      "loss": 0.912,
+      "step": 73230
+    },
+    {
+      "epoch": 0.5301599021332349,
+      "grad_norm": 0.16933558881282806,
+      "learning_rate": 4.469847336532824e-06,
+      "loss": 0.9084,
+      "step": 73240
+    },
+    {
+      "epoch": 0.5302322887938211,
+      "grad_norm": 0.1705089658498764,
+      "learning_rate": 4.469774949872238e-06,
+      "loss": 0.9249,
+      "step": 73250
+    },
+    {
+      "epoch": 0.5303046754544073,
+      "grad_norm": 0.16231994330883026,
+      "learning_rate": 4.469702563211652e-06,
+      "loss": 0.9106,
+      "step": 73260
+    },
+    {
+      "epoch": 0.5303770621149935,
+      "grad_norm": 0.1603698879480362,
+      "learning_rate": 4.469630176551066e-06,
+      "loss": 0.9157,
+      "step": 73270
+    },
+    {
+      "epoch": 0.5304494487755796,
+      "grad_norm": 0.1552126109600067,
+      "learning_rate": 4.469557789890479e-06,
+      "loss": 0.9121,
+      "step": 73280
+    },
+    {
+      "epoch": 0.5305218354361658,
+      "grad_norm": 0.16099673509597778,
+      "learning_rate": 4.469485403229893e-06,
+      "loss": 0.9216,
+      "step": 73290
+    },
+    {
+      "epoch": 0.530594222096752,
+      "grad_norm": 0.16103273630142212,
+      "learning_rate": 4.469413016569307e-06,
+      "loss": 0.9206,
+      "step": 73300
+    },
+    {
+      "epoch": 0.5306666087573382,
+      "grad_norm": 0.16438160836696625,
+      "learning_rate": 4.46934062990872e-06,
+      "loss": 0.9218,
+      "step": 73310
+    },
+    {
+      "epoch": 0.5307389954179244,
+      "grad_norm": 0.17345352470874786,
+      "learning_rate": 4.469268243248135e-06,
+      "loss": 0.9085,
+      "step": 73320
+    },
+    {
+      "epoch": 0.5308113820785105,
+      "grad_norm": 0.15736123919487,
+      "learning_rate": 4.469195856587548e-06,
+      "loss": 0.9238,
+      "step": 73330
+    },
+    {
+      "epoch": 0.5308837687390967,
+      "grad_norm": 0.17544583976268768,
+      "learning_rate": 4.469123469926962e-06,
+      "loss": 0.9176,
+      "step": 73340
+    },
+    {
+      "epoch": 0.530956155399683,
+      "grad_norm": 0.17113934457302094,
+      "learning_rate": 4.469051083266376e-06,
+      "loss": 0.9092,
+      "step": 73350
+    },
+    {
+      "epoch": 0.5310285420602692,
+      "grad_norm": 0.1495228260755539,
+      "learning_rate": 4.46897869660579e-06,
+      "loss": 0.9151,
+      "step": 73360
+    },
+    {
+      "epoch": 0.5311009287208553,
+      "grad_norm": 0.16176773607730865,
+      "learning_rate": 4.468906309945204e-06,
+      "loss": 0.8991,
+      "step": 73370
+    },
+    {
+      "epoch": 0.5311733153814415,
+      "grad_norm": 0.15699774026870728,
+      "learning_rate": 4.468833923284617e-06,
+      "loss": 0.9167,
+      "step": 73380
+    },
+    {
+      "epoch": 0.5312457020420277,
+      "grad_norm": 0.15788871049880981,
+      "learning_rate": 4.468761536624031e-06,
+      "loss": 0.8985,
+      "step": 73390
+    },
+    {
+      "epoch": 0.5313180887026139,
+      "grad_norm": 0.155131995677948,
+      "learning_rate": 4.468689149963445e-06,
+      "loss": 0.9163,
+      "step": 73400
+    },
+    {
+      "epoch": 0.5313904753632,
+      "grad_norm": 0.15978825092315674,
+      "learning_rate": 4.468616763302859e-06,
+      "loss": 0.911,
+      "step": 73410
+    },
+    {
+      "epoch": 0.5314628620237862,
+      "grad_norm": 0.18127478659152985,
+      "learning_rate": 4.468544376642273e-06,
+      "loss": 0.9225,
+      "step": 73420
+    },
+    {
+      "epoch": 0.5315352486843724,
+      "grad_norm": 0.15009476244449615,
+      "learning_rate": 4.468471989981686e-06,
+      "loss": 0.9069,
+      "step": 73430
+    },
+    {
+      "epoch": 0.5316076353449586,
+      "grad_norm": 0.15387962758541107,
+      "learning_rate": 4.4683996033211e-06,
+      "loss": 0.9209,
+      "step": 73440
+    },
+    {
+      "epoch": 0.5316800220055449,
+      "grad_norm": 0.15109896659851074,
+      "learning_rate": 4.468327216660514e-06,
+      "loss": 0.9285,
+      "step": 73450
+    },
+    {
+      "epoch": 0.531752408666131,
+      "grad_norm": 0.20869199931621552,
+      "learning_rate": 4.468254829999928e-06,
+      "loss": 0.9189,
+      "step": 73460
+    },
+    {
+      "epoch": 0.5318247953267172,
+      "grad_norm": 0.15885184705257416,
+      "learning_rate": 4.4681824433393415e-06,
+      "loss": 0.9301,
+      "step": 73470
+    },
+    {
+      "epoch": 0.5318971819873034,
+      "grad_norm": 0.2871398627758026,
+      "learning_rate": 4.468110056678755e-06,
+      "loss": 0.9291,
+      "step": 73480
+    },
+    {
+      "epoch": 0.5319695686478896,
+      "grad_norm": 0.14929638803005219,
+      "learning_rate": 4.46803767001817e-06,
+      "loss": 0.9143,
+      "step": 73490
+    },
+    {
+      "epoch": 0.5320419553084758,
+      "grad_norm": 0.15041810274124146,
+      "learning_rate": 4.467965283357583e-06,
+      "loss": 0.9164,
+      "step": 73500
+    },
+    {
+      "epoch": 0.5321143419690619,
+      "grad_norm": 0.1545180082321167,
+      "learning_rate": 4.467892896696997e-06,
+      "loss": 0.9144,
+      "step": 73510
+    },
+    {
+      "epoch": 0.5321867286296481,
+      "grad_norm": 0.15915392339229584,
+      "learning_rate": 4.4678205100364104e-06,
+      "loss": 0.9088,
+      "step": 73520
+    },
+    {
+      "epoch": 0.5322591152902343,
+      "grad_norm": 0.1939249187707901,
+      "learning_rate": 4.467748123375825e-06,
+      "loss": 0.9066,
+      "step": 73530
+    },
+    {
+      "epoch": 0.5323315019508205,
+      "grad_norm": 0.1595548391342163,
+      "learning_rate": 4.4676757367152385e-06,
+      "loss": 0.9067,
+      "step": 73540
+    },
+    {
+      "epoch": 0.5324038886114066,
+      "grad_norm": 0.15864749252796173,
+      "learning_rate": 4.467603350054652e-06,
+      "loss": 0.9134,
+      "step": 73550
+    },
+    {
+      "epoch": 0.5324762752719929,
+      "grad_norm": 0.171446293592453,
+      "learning_rate": 4.467530963394066e-06,
+      "loss": 0.9157,
+      "step": 73560
+    },
+    {
+      "epoch": 0.5325486619325791,
+      "grad_norm": 0.15038402378559113,
+      "learning_rate": 4.46745857673348e-06,
+      "loss": 0.9159,
+      "step": 73570
+    },
+    {
+      "epoch": 0.5326210485931653,
+      "grad_norm": 0.15996231138706207,
+      "learning_rate": 4.467386190072894e-06,
+      "loss": 0.9059,
+      "step": 73580
+    },
+    {
+      "epoch": 0.5326934352537515,
+      "grad_norm": 0.16054002940654755,
+      "learning_rate": 4.4673138034123075e-06,
+      "loss": 0.9136,
+      "step": 73590
+    },
+    {
+      "epoch": 0.5327658219143376,
+      "grad_norm": 0.15865951776504517,
+      "learning_rate": 4.467241416751721e-06,
+      "loss": 0.9048,
+      "step": 73600
+    },
+    {
+      "epoch": 0.5328382085749238,
+      "grad_norm": 0.15602022409439087,
+      "learning_rate": 4.4671690300911355e-06,
+      "loss": 0.901,
+      "step": 73610
+    },
+    {
+      "epoch": 0.53291059523551,
+      "grad_norm": 0.16050325334072113,
+      "learning_rate": 4.467096643430549e-06,
+      "loss": 0.9012,
+      "step": 73620
+    },
+    {
+      "epoch": 0.5329829818960962,
+      "grad_norm": 0.15464062988758087,
+      "learning_rate": 4.467024256769963e-06,
+      "loss": 0.918,
+      "step": 73630
+    },
+    {
+      "epoch": 0.5330553685566823,
+      "grad_norm": 0.16735030710697174,
+      "learning_rate": 4.466951870109376e-06,
+      "loss": 0.9048,
+      "step": 73640
+    },
+    {
+      "epoch": 0.5331277552172685,
+      "grad_norm": 0.17225484549999237,
+      "learning_rate": 4.466879483448791e-06,
+      "loss": 0.9131,
+      "step": 73650
+    },
+    {
+      "epoch": 0.5332001418778547,
+      "grad_norm": 0.17411339282989502,
+      "learning_rate": 4.4668070967882045e-06,
+      "loss": 0.9063,
+      "step": 73660
+    },
+    {
+      "epoch": 0.533272528538441,
+      "grad_norm": 0.19198282063007355,
+      "learning_rate": 4.466734710127618e-06,
+      "loss": 0.9239,
+      "step": 73670
+    },
+    {
+      "epoch": 0.5333449151990272,
+      "grad_norm": 0.16125676035881042,
+      "learning_rate": 4.466662323467032e-06,
+      "loss": 0.9251,
+      "step": 73680
+    },
+    {
+      "epoch": 0.5334173018596133,
+      "grad_norm": 0.1499517261981964,
+      "learning_rate": 4.466589936806446e-06,
+      "loss": 0.913,
+      "step": 73690
+    },
+    {
+      "epoch": 0.5334896885201995,
+      "grad_norm": 0.16353091597557068,
+      "learning_rate": 4.46651755014586e-06,
+      "loss": 0.9166,
+      "step": 73700
+    },
+    {
+      "epoch": 0.5335620751807857,
+      "grad_norm": 0.14786536991596222,
+      "learning_rate": 4.466445163485273e-06,
+      "loss": 0.9174,
+      "step": 73710
+    },
+    {
+      "epoch": 0.5336344618413719,
+      "grad_norm": 0.15844851732254028,
+      "learning_rate": 4.466372776824687e-06,
+      "loss": 0.9162,
+      "step": 73720
+    },
+    {
+      "epoch": 0.533706848501958,
+      "grad_norm": 0.18850095570087433,
+      "learning_rate": 4.4663003901641015e-06,
+      "loss": 0.9141,
+      "step": 73730
+    },
+    {
+      "epoch": 0.5337792351625442,
+      "grad_norm": 0.1814296692609787,
+      "learning_rate": 4.466228003503515e-06,
+      "loss": 0.9087,
+      "step": 73740
+    },
+    {
+      "epoch": 0.5338516218231304,
+      "grad_norm": 0.1553386151790619,
+      "learning_rate": 4.466155616842929e-06,
+      "loss": 0.9115,
+      "step": 73750
+    },
+    {
+      "epoch": 0.5339240084837166,
+      "grad_norm": 0.1544741988182068,
+      "learning_rate": 4.466083230182342e-06,
+      "loss": 0.9066,
+      "step": 73760
+    },
+    {
+      "epoch": 0.5339963951443029,
+      "grad_norm": 0.15202949941158295,
+      "learning_rate": 4.466010843521757e-06,
+      "loss": 0.9164,
+      "step": 73770
+    },
+    {
+      "epoch": 0.534068781804889,
+      "grad_norm": 0.157356858253479,
+      "learning_rate": 4.46593845686117e-06,
+      "loss": 0.9272,
+      "step": 73780
+    },
+    {
+      "epoch": 0.5341411684654752,
+      "grad_norm": 0.1528637409210205,
+      "learning_rate": 4.465866070200584e-06,
+      "loss": 0.9253,
+      "step": 73790
+    },
+    {
+      "epoch": 0.5342135551260614,
+      "grad_norm": 0.1515146940946579,
+      "learning_rate": 4.465793683539998e-06,
+      "loss": 0.9025,
+      "step": 73800
+    },
+    {
+      "epoch": 0.5342859417866476,
+      "grad_norm": 0.1715908944606781,
+      "learning_rate": 4.465721296879412e-06,
+      "loss": 0.9274,
+      "step": 73810
+    },
+    {
+      "epoch": 0.5343583284472337,
+      "grad_norm": 0.17339174449443817,
+      "learning_rate": 4.465648910218826e-06,
+      "loss": 0.9196,
+      "step": 73820
+    },
+    {
+      "epoch": 0.5344307151078199,
+      "grad_norm": 0.18492698669433594,
+      "learning_rate": 4.4655765235582385e-06,
+      "loss": 0.9117,
+      "step": 73830
+    },
+    {
+      "epoch": 0.5345031017684061,
+      "grad_norm": 0.18511684238910675,
+      "learning_rate": 4.465504136897653e-06,
+      "loss": 0.9192,
+      "step": 73840
+    },
+    {
+      "epoch": 0.5345754884289923,
+      "grad_norm": 0.15983940660953522,
+      "learning_rate": 4.4654317502370666e-06,
+      "loss": 0.9144,
+      "step": 73850
+    },
+    {
+      "epoch": 0.5346478750895785,
+      "grad_norm": 0.1514192372560501,
+      "learning_rate": 4.46535936357648e-06,
+      "loss": 0.9289,
+      "step": 73860
+    },
+    {
+      "epoch": 0.5347202617501646,
+      "grad_norm": 0.1741372048854828,
+      "learning_rate": 4.465286976915894e-06,
+      "loss": 0.9224,
+      "step": 73870
+    },
+    {
+      "epoch": 0.5347926484107509,
+      "grad_norm": 0.1461644023656845,
+      "learning_rate": 4.465214590255308e-06,
+      "loss": 0.9044,
+      "step": 73880
+    },
+    {
+      "epoch": 0.5348650350713371,
+      "grad_norm": 0.1568828821182251,
+      "learning_rate": 4.465142203594722e-06,
+      "loss": 0.9098,
+      "step": 73890
+    },
+    {
+      "epoch": 0.5349374217319233,
+      "grad_norm": 0.16432945430278778,
+      "learning_rate": 4.4650698169341355e-06,
+      "loss": 0.9187,
+      "step": 73900
+    },
+    {
+      "epoch": 0.5350098083925094,
+      "grad_norm": 0.16285908222198486,
+      "learning_rate": 4.464997430273549e-06,
+      "loss": 0.9172,
+      "step": 73910
+    },
+    {
+      "epoch": 0.5350821950530956,
+      "grad_norm": 0.15735489130020142,
+      "learning_rate": 4.4649250436129636e-06,
+      "loss": 0.9198,
+      "step": 73920
+    },
+    {
+      "epoch": 0.5351545817136818,
+      "grad_norm": 0.1569201946258545,
+      "learning_rate": 4.464852656952377e-06,
+      "loss": 0.9162,
+      "step": 73930
+    },
+    {
+      "epoch": 0.535226968374268,
+      "grad_norm": 0.1620626151561737,
+      "learning_rate": 4.464780270291791e-06,
+      "loss": 0.9158,
+      "step": 73940
+    },
+    {
+      "epoch": 0.5352993550348542,
+      "grad_norm": 0.16271735727787018,
+      "learning_rate": 4.464707883631204e-06,
+      "loss": 0.9015,
+      "step": 73950
+    },
+    {
+      "epoch": 0.5353717416954403,
+      "grad_norm": 0.23674176633358002,
+      "learning_rate": 4.464635496970619e-06,
+      "loss": 0.9143,
+      "step": 73960
+    },
+    {
+      "epoch": 0.5354441283560265,
+      "grad_norm": 0.18053658306598663,
+      "learning_rate": 4.4645631103100325e-06,
+      "loss": 0.9125,
+      "step": 73970
+    },
+    {
+      "epoch": 0.5355165150166128,
+      "grad_norm": 0.16268426179885864,
+      "learning_rate": 4.464490723649446e-06,
+      "loss": 0.9208,
+      "step": 73980
+    },
+    {
+      "epoch": 0.535588901677199,
+      "grad_norm": 0.15412411093711853,
+      "learning_rate": 4.46441833698886e-06,
+      "loss": 0.9183,
+      "step": 73990
+    },
+    {
+      "epoch": 0.5356612883377851,
+      "grad_norm": 0.18611110746860504,
+      "learning_rate": 4.464345950328274e-06,
+      "loss": 0.9177,
+      "step": 74000
+    },
+    {
+      "epoch": 0.5357336749983713,
+      "grad_norm": 0.14643755555152893,
+      "learning_rate": 4.464273563667688e-06,
+      "loss": 0.9048,
+      "step": 74010
+    },
+    {
+      "epoch": 0.5358060616589575,
+      "grad_norm": 0.17977450788021088,
+      "learning_rate": 4.464201177007101e-06,
+      "loss": 0.9166,
+      "step": 74020
+    },
+    {
+      "epoch": 0.5358784483195437,
+      "grad_norm": 0.159438818693161,
+      "learning_rate": 4.464128790346515e-06,
+      "loss": 0.9098,
+      "step": 74030
+    },
+    {
+      "epoch": 0.5359508349801299,
+      "grad_norm": 0.1566077172756195,
+      "learning_rate": 4.464056403685929e-06,
+      "loss": 0.9198,
+      "step": 74040
+    },
+    {
+      "epoch": 0.536023221640716,
+      "grad_norm": 0.17558586597442627,
+      "learning_rate": 4.463984017025343e-06,
+      "loss": 0.9303,
+      "step": 74050
+    },
+    {
+      "epoch": 0.5360956083013022,
+      "grad_norm": 0.15491542220115662,
+      "learning_rate": 4.463911630364757e-06,
+      "loss": 0.9193,
+      "step": 74060
+    },
+    {
+      "epoch": 0.5361679949618884,
+      "grad_norm": 0.17469727993011475,
+      "learning_rate": 4.46383924370417e-06,
+      "loss": 0.9117,
+      "step": 74070
+    },
+    {
+      "epoch": 0.5362403816224746,
+      "grad_norm": 0.1587485820055008,
+      "learning_rate": 4.463766857043584e-06,
+      "loss": 0.9028,
+      "step": 74080
+    },
+    {
+      "epoch": 0.5363127682830608,
+      "grad_norm": 0.15624359250068665,
+      "learning_rate": 4.4636944703829984e-06,
+      "loss": 0.9028,
+      "step": 74090
+    },
+    {
+      "epoch": 0.536385154943647,
+      "grad_norm": 0.17197231948375702,
+      "learning_rate": 4.463622083722412e-06,
+      "loss": 0.9141,
+      "step": 74100
+    },
+    {
+      "epoch": 0.5364575416042332,
+      "grad_norm": 0.1522126942873001,
+      "learning_rate": 4.463549697061826e-06,
+      "loss": 0.9279,
+      "step": 74110
+    },
+    {
+      "epoch": 0.5365299282648194,
+      "grad_norm": 0.1569167822599411,
+      "learning_rate": 4.463477310401239e-06,
+      "loss": 0.9168,
+      "step": 74120
+    },
+    {
+      "epoch": 0.5366023149254056,
+      "grad_norm": 0.18380215764045715,
+      "learning_rate": 4.463404923740654e-06,
+      "loss": 0.9146,
+      "step": 74130
+    },
+    {
+      "epoch": 0.5366747015859917,
+      "grad_norm": 0.1586330085992813,
+      "learning_rate": 4.463332537080067e-06,
+      "loss": 0.8948,
+      "step": 74140
+    },
+    {
+      "epoch": 0.5367470882465779,
+      "grad_norm": 0.1561429798603058,
+      "learning_rate": 4.463260150419481e-06,
+      "loss": 0.9113,
+      "step": 74150
+    },
+    {
+      "epoch": 0.5368194749071641,
+      "grad_norm": 0.1750173717737198,
+      "learning_rate": 4.463187763758895e-06,
+      "loss": 0.9277,
+      "step": 74160
+    },
+    {
+      "epoch": 0.5368918615677503,
+      "grad_norm": 0.15227210521697998,
+      "learning_rate": 4.463115377098309e-06,
+      "loss": 0.9042,
+      "step": 74170
+    },
+    {
+      "epoch": 0.5369642482283364,
+      "grad_norm": 0.15368933975696564,
+      "learning_rate": 4.463042990437723e-06,
+      "loss": 0.9009,
+      "step": 74180
+    },
+    {
+      "epoch": 0.5370366348889226,
+      "grad_norm": 0.1501675844192505,
+      "learning_rate": 4.462970603777136e-06,
+      "loss": 0.9168,
+      "step": 74190
+    },
+    {
+      "epoch": 0.5371090215495089,
+      "grad_norm": 0.16025373339653015,
+      "learning_rate": 4.46289821711655e-06,
+      "loss": 0.9125,
+      "step": 74200
+    },
+    {
+      "epoch": 0.5371814082100951,
+      "grad_norm": 0.1600399762392044,
+      "learning_rate": 4.462825830455964e-06,
+      "loss": 0.8988,
+      "step": 74210
+    },
+    {
+      "epoch": 0.5372537948706813,
+      "grad_norm": 0.1532607227563858,
+      "learning_rate": 4.462753443795378e-06,
+      "loss": 0.9254,
+      "step": 74220
+    },
+    {
+      "epoch": 0.5373261815312674,
+      "grad_norm": 0.1633194237947464,
+      "learning_rate": 4.462681057134792e-06,
+      "loss": 0.9143,
+      "step": 74230
+    },
+    {
+      "epoch": 0.5373985681918536,
+      "grad_norm": 0.15434294939041138,
+      "learning_rate": 4.462608670474205e-06,
+      "loss": 0.9139,
+      "step": 74240
+    },
+    {
+      "epoch": 0.5374709548524398,
+      "grad_norm": 0.15733137726783752,
+      "learning_rate": 4.46253628381362e-06,
+      "loss": 0.9254,
+      "step": 74250
+    },
+    {
+      "epoch": 0.537543341513026,
+      "grad_norm": 0.16783663630485535,
+      "learning_rate": 4.462463897153033e-06,
+      "loss": 0.9132,
+      "step": 74260
+    },
+    {
+      "epoch": 0.5376157281736121,
+      "grad_norm": 0.16016171872615814,
+      "learning_rate": 4.462391510492447e-06,
+      "loss": 0.9161,
+      "step": 74270
+    },
+    {
+      "epoch": 0.5376881148341983,
+      "grad_norm": 0.1455632746219635,
+      "learning_rate": 4.4623191238318605e-06,
+      "loss": 0.9178,
+      "step": 74280
+    },
+    {
+      "epoch": 0.5377605014947845,
+      "grad_norm": 0.16944627463817596,
+      "learning_rate": 4.462246737171275e-06,
+      "loss": 0.9171,
+      "step": 74290
+    },
+    {
+      "epoch": 0.5378328881553708,
+      "grad_norm": 0.16209137439727783,
+      "learning_rate": 4.462174350510689e-06,
+      "loss": 0.9148,
+      "step": 74300
+    },
+    {
+      "epoch": 0.537905274815957,
+      "grad_norm": 0.16766659915447235,
+      "learning_rate": 4.462101963850102e-06,
+      "loss": 0.9054,
+      "step": 74310
+    },
+    {
+      "epoch": 0.5379776614765431,
+      "grad_norm": 0.15169428288936615,
+      "learning_rate": 4.462029577189516e-06,
+      "loss": 0.916,
+      "step": 74320
+    },
+    {
+      "epoch": 0.5380500481371293,
+      "grad_norm": 0.17565733194351196,
+      "learning_rate": 4.46195719052893e-06,
+      "loss": 0.9056,
+      "step": 74330
+    },
+    {
+      "epoch": 0.5381224347977155,
+      "grad_norm": 0.1468237191438675,
+      "learning_rate": 4.461884803868344e-06,
+      "loss": 0.9199,
+      "step": 74340
+    },
+    {
+      "epoch": 0.5381948214583017,
+      "grad_norm": 0.15995702147483826,
+      "learning_rate": 4.4618124172077575e-06,
+      "loss": 0.9205,
+      "step": 74350
+    },
+    {
+      "epoch": 0.5382672081188878,
+      "grad_norm": 0.28701356053352356,
+      "learning_rate": 4.461740030547171e-06,
+      "loss": 0.9032,
+      "step": 74360
+    },
+    {
+      "epoch": 0.538339594779474,
+      "grad_norm": 0.15283870697021484,
+      "learning_rate": 4.461667643886585e-06,
+      "loss": 0.919,
+      "step": 74370
+    },
+    {
+      "epoch": 0.5384119814400602,
+      "grad_norm": 0.2051803320646286,
+      "learning_rate": 4.461595257225998e-06,
+      "loss": 0.9115,
+      "step": 74380
+    },
+    {
+      "epoch": 0.5384843681006464,
+      "grad_norm": 0.16268160939216614,
+      "learning_rate": 4.461522870565412e-06,
+      "loss": 0.9043,
+      "step": 74390
+    },
+    {
+      "epoch": 0.5385567547612325,
+      "grad_norm": 0.16522175073623657,
+      "learning_rate": 4.4614504839048265e-06,
+      "loss": 0.9193,
+      "step": 74400
+    },
+    {
+      "epoch": 0.5386291414218188,
+      "grad_norm": 0.1583668738603592,
+      "learning_rate": 4.46137809724424e-06,
+      "loss": 0.9083,
+      "step": 74410
+    },
+    {
+      "epoch": 0.538701528082405,
+      "grad_norm": 0.15010447800159454,
+      "learning_rate": 4.461305710583654e-06,
+      "loss": 0.9025,
+      "step": 74420
+    },
+    {
+      "epoch": 0.5387739147429912,
+      "grad_norm": 0.16112802922725677,
+      "learning_rate": 4.461233323923067e-06,
+      "loss": 0.9152,
+      "step": 74430
+    },
+    {
+      "epoch": 0.5388463014035774,
+      "grad_norm": 0.16676628589630127,
+      "learning_rate": 4.461160937262482e-06,
+      "loss": 0.9022,
+      "step": 74440
+    },
+    {
+      "epoch": 0.5389186880641635,
+      "grad_norm": 0.1548004299402237,
+      "learning_rate": 4.461088550601895e-06,
+      "loss": 0.9131,
+      "step": 74450
+    },
+    {
+      "epoch": 0.5389910747247497,
+      "grad_norm": 0.17125369608402252,
+      "learning_rate": 4.461016163941309e-06,
+      "loss": 0.9151,
+      "step": 74460
+    },
+    {
+      "epoch": 0.5390634613853359,
+      "grad_norm": 0.1690664291381836,
+      "learning_rate": 4.460943777280723e-06,
+      "loss": 0.9138,
+      "step": 74470
+    },
+    {
+      "epoch": 0.5391358480459221,
+      "grad_norm": 0.17033086717128754,
+      "learning_rate": 4.460871390620137e-06,
+      "loss": 0.8982,
+      "step": 74480
+    },
+    {
+      "epoch": 0.5392082347065082,
+      "grad_norm": 0.15503831207752228,
+      "learning_rate": 4.460799003959551e-06,
+      "loss": 0.9268,
+      "step": 74490
+    },
+    {
+      "epoch": 0.5392806213670944,
+      "grad_norm": 0.15129579603672028,
+      "learning_rate": 4.460726617298964e-06,
+      "loss": 0.9067,
+      "step": 74500
+    },
+    {
+      "epoch": 0.5393530080276807,
+      "grad_norm": 0.16346019506454468,
+      "learning_rate": 4.460654230638378e-06,
+      "loss": 0.9138,
+      "step": 74510
+    },
+    {
+      "epoch": 0.5394253946882669,
+      "grad_norm": 0.15942130982875824,
+      "learning_rate": 4.460581843977792e-06,
+      "loss": 0.9109,
+      "step": 74520
+    },
+    {
+      "epoch": 0.5394977813488531,
+      "grad_norm": 0.14833667874336243,
+      "learning_rate": 4.460509457317206e-06,
+      "loss": 0.9065,
+      "step": 74530
+    },
+    {
+      "epoch": 0.5395701680094392,
+      "grad_norm": 0.15694496035575867,
+      "learning_rate": 4.46043707065662e-06,
+      "loss": 0.9225,
+      "step": 74540
+    },
+    {
+      "epoch": 0.5396425546700254,
+      "grad_norm": 0.15417462587356567,
+      "learning_rate": 4.460364683996033e-06,
+      "loss": 0.8993,
+      "step": 74550
+    },
+    {
+      "epoch": 0.5397149413306116,
+      "grad_norm": 0.16096115112304688,
+      "learning_rate": 4.460292297335448e-06,
+      "loss": 0.9104,
+      "step": 74560
+    },
+    {
+      "epoch": 0.5397873279911978,
+      "grad_norm": 0.1507575362920761,
+      "learning_rate": 4.460219910674861e-06,
+      "loss": 0.9165,
+      "step": 74570
+    },
+    {
+      "epoch": 0.539859714651784,
+      "grad_norm": 0.16418473422527313,
+      "learning_rate": 4.460147524014275e-06,
+      "loss": 0.9192,
+      "step": 74580
+    },
+    {
+      "epoch": 0.5399321013123701,
+      "grad_norm": 0.1715206801891327,
+      "learning_rate": 4.4600751373536886e-06,
+      "loss": 0.9093,
+      "step": 74590
+    },
+    {
+      "epoch": 0.5400044879729563,
+      "grad_norm": 0.1661735624074936,
+      "learning_rate": 4.460002750693103e-06,
+      "loss": 0.9224,
+      "step": 74600
+    },
+    {
+      "epoch": 0.5400768746335425,
+      "grad_norm": 0.15840347111225128,
+      "learning_rate": 4.459930364032517e-06,
+      "loss": 0.9192,
+      "step": 74610
+    },
+    {
+      "epoch": 0.5401492612941288,
+      "grad_norm": 0.15837319195270538,
+      "learning_rate": 4.45985797737193e-06,
+      "loss": 0.9191,
+      "step": 74620
+    },
+    {
+      "epoch": 0.540221647954715,
+      "grad_norm": 0.1872469186782837,
+      "learning_rate": 4.459785590711344e-06,
+      "loss": 0.9116,
+      "step": 74630
+    },
+    {
+      "epoch": 0.5402940346153011,
+      "grad_norm": 0.16932313144207,
+      "learning_rate": 4.459713204050758e-06,
+      "loss": 0.8963,
+      "step": 74640
+    },
+    {
+      "epoch": 0.5403664212758873,
+      "grad_norm": 0.1552034318447113,
+      "learning_rate": 4.459640817390172e-06,
+      "loss": 0.9058,
+      "step": 74650
+    },
+    {
+      "epoch": 0.5404388079364735,
+      "grad_norm": 0.1520053595304489,
+      "learning_rate": 4.4595684307295856e-06,
+      "loss": 0.9006,
+      "step": 74660
+    },
+    {
+      "epoch": 0.5405111945970597,
+      "grad_norm": 0.15014509856700897,
+      "learning_rate": 4.459496044068999e-06,
+      "loss": 0.9045,
+      "step": 74670
+    },
+    {
+      "epoch": 0.5405835812576458,
+      "grad_norm": 0.18093499541282654,
+      "learning_rate": 4.459423657408413e-06,
+      "loss": 0.9013,
+      "step": 74680
+    },
+    {
+      "epoch": 0.540655967918232,
+      "grad_norm": 0.1501227766275406,
+      "learning_rate": 4.459351270747827e-06,
+      "loss": 0.9043,
+      "step": 74690
+    },
+    {
+      "epoch": 0.5407283545788182,
+      "grad_norm": 0.1506822109222412,
+      "learning_rate": 4.459278884087241e-06,
+      "loss": 0.9163,
+      "step": 74700
+    },
+    {
+      "epoch": 0.5408007412394044,
+      "grad_norm": 0.1608346849679947,
+      "learning_rate": 4.4592064974266545e-06,
+      "loss": 0.911,
+      "step": 74710
+    },
+    {
+      "epoch": 0.5408731278999905,
+      "grad_norm": 0.1549476683139801,
+      "learning_rate": 4.459134110766068e-06,
+      "loss": 0.9132,
+      "step": 74720
+    },
+    {
+      "epoch": 0.5409455145605768,
+      "grad_norm": 0.15968433022499084,
+      "learning_rate": 4.4590617241054826e-06,
+      "loss": 0.9119,
+      "step": 74730
+    },
+    {
+      "epoch": 0.541017901221163,
+      "grad_norm": 0.17671333253383636,
+      "learning_rate": 4.458989337444896e-06,
+      "loss": 0.9038,
+      "step": 74740
+    },
+    {
+      "epoch": 0.5410902878817492,
+      "grad_norm": 0.15897606313228607,
+      "learning_rate": 4.45891695078431e-06,
+      "loss": 0.9038,
+      "step": 74750
+    },
+    {
+      "epoch": 0.5411626745423354,
+      "grad_norm": 0.15166926383972168,
+      "learning_rate": 4.458844564123723e-06,
+      "loss": 0.9155,
+      "step": 74760
+    },
+    {
+      "epoch": 0.5412350612029215,
+      "grad_norm": 0.14908485114574432,
+      "learning_rate": 4.458772177463138e-06,
+      "loss": 0.9107,
+      "step": 74770
+    },
+    {
+      "epoch": 0.5413074478635077,
+      "grad_norm": 0.15309011936187744,
+      "learning_rate": 4.4586997908025515e-06,
+      "loss": 0.9124,
+      "step": 74780
+    },
+    {
+      "epoch": 0.5413798345240939,
+      "grad_norm": 0.15603891015052795,
+      "learning_rate": 4.458627404141965e-06,
+      "loss": 0.9081,
+      "step": 74790
+    },
+    {
+      "epoch": 0.5414522211846801,
+      "grad_norm": 0.17669987678527832,
+      "learning_rate": 4.458555017481379e-06,
+      "loss": 0.9093,
+      "step": 74800
+    },
+    {
+      "epoch": 0.5415246078452662,
+      "grad_norm": 0.1566879153251648,
+      "learning_rate": 4.458482630820793e-06,
+      "loss": 0.9079,
+      "step": 74810
+    },
+    {
+      "epoch": 0.5415969945058524,
+      "grad_norm": 0.17799584567546844,
+      "learning_rate": 4.458410244160207e-06,
+      "loss": 0.9145,
+      "step": 74820
+    },
+    {
+      "epoch": 0.5416693811664387,
+      "grad_norm": 0.1597883254289627,
+      "learning_rate": 4.4583378574996204e-06,
+      "loss": 0.9082,
+      "step": 74830
+    },
+    {
+      "epoch": 0.5417417678270249,
+      "grad_norm": 0.16738228499889374,
+      "learning_rate": 4.458265470839034e-06,
+      "loss": 0.9198,
+      "step": 74840
+    },
+    {
+      "epoch": 0.541814154487611,
+      "grad_norm": 0.16331464052200317,
+      "learning_rate": 4.4581930841784485e-06,
+      "loss": 0.9269,
+      "step": 74850
+    },
+    {
+      "epoch": 0.5418865411481972,
+      "grad_norm": 0.1695605218410492,
+      "learning_rate": 4.458120697517862e-06,
+      "loss": 0.9043,
+      "step": 74860
+    },
+    {
+      "epoch": 0.5419589278087834,
+      "grad_norm": 0.15982206165790558,
+      "learning_rate": 4.458048310857276e-06,
+      "loss": 0.9224,
+      "step": 74870
+    },
+    {
+      "epoch": 0.5420313144693696,
+      "grad_norm": 0.16445820033550262,
+      "learning_rate": 4.457975924196689e-06,
+      "loss": 0.9069,
+      "step": 74880
+    },
+    {
+      "epoch": 0.5421037011299558,
+      "grad_norm": 0.18443961441516876,
+      "learning_rate": 4.457903537536103e-06,
+      "loss": 0.9108,
+      "step": 74890
+    },
+    {
+      "epoch": 0.5421760877905419,
+      "grad_norm": 0.1700538992881775,
+      "learning_rate": 4.457831150875517e-06,
+      "loss": 0.924,
+      "step": 74900
+    },
+    {
+      "epoch": 0.5422484744511281,
+      "grad_norm": 0.174394428730011,
+      "learning_rate": 4.45775876421493e-06,
+      "loss": 0.9144,
+      "step": 74910
+    },
+    {
+      "epoch": 0.5423208611117143,
+      "grad_norm": 0.1507367342710495,
+      "learning_rate": 4.457686377554345e-06,
+      "loss": 0.9172,
+      "step": 74920
+    },
+    {
+      "epoch": 0.5423932477723005,
+      "grad_norm": 0.1828458607196808,
+      "learning_rate": 4.457613990893758e-06,
+      "loss": 0.9018,
+      "step": 74930
+    },
+    {
+      "epoch": 0.5424656344328868,
+      "grad_norm": 0.15712609887123108,
+      "learning_rate": 4.457541604233172e-06,
+      "loss": 0.9145,
+      "step": 74940
+    },
+    {
+      "epoch": 0.5425380210934729,
+      "grad_norm": 0.1707613319158554,
+      "learning_rate": 4.4574692175725855e-06,
+      "loss": 0.9128,
+      "step": 74950
+    },
+    {
+      "epoch": 0.5426104077540591,
+      "grad_norm": 0.17930477857589722,
+      "learning_rate": 4.457396830912e-06,
+      "loss": 0.9068,
+      "step": 74960
+    },
+    {
+      "epoch": 0.5426827944146453,
+      "grad_norm": 0.1521557718515396,
+      "learning_rate": 4.457324444251414e-06,
+      "loss": 0.9035,
+      "step": 74970
+    },
+    {
+      "epoch": 0.5427551810752315,
+      "grad_norm": 0.17281807959079742,
+      "learning_rate": 4.457252057590827e-06,
+      "loss": 0.9113,
+      "step": 74980
+    },
+    {
+      "epoch": 0.5428275677358176,
+      "grad_norm": 0.147269606590271,
+      "learning_rate": 4.457179670930241e-06,
+      "loss": 0.9171,
+      "step": 74990
+    },
+    {
+      "epoch": 0.5428999543964038,
+      "grad_norm": 0.1692981719970703,
+      "learning_rate": 4.457107284269655e-06,
+      "loss": 0.9118,
+      "step": 75000
+    },
+    {
+      "epoch": 0.54297234105699,
+      "grad_norm": 0.16155540943145752,
+      "learning_rate": 4.457034897609069e-06,
+      "loss": 0.9116,
+      "step": 75010
+    },
+    {
+      "epoch": 0.5430447277175762,
+      "grad_norm": 0.15603849291801453,
+      "learning_rate": 4.4569625109484825e-06,
+      "loss": 0.9051,
+      "step": 75020
+    },
+    {
+      "epoch": 0.5431171143781623,
+      "grad_norm": 0.16101206839084625,
+      "learning_rate": 4.456890124287896e-06,
+      "loss": 0.9154,
+      "step": 75030
+    },
+    {
+      "epoch": 0.5431895010387486,
+      "grad_norm": 0.15974144637584686,
+      "learning_rate": 4.456817737627311e-06,
+      "loss": 0.9182,
+      "step": 75040
+    },
+    {
+      "epoch": 0.5432618876993348,
+      "grad_norm": 0.169901043176651,
+      "learning_rate": 4.456745350966724e-06,
+      "loss": 0.9084,
+      "step": 75050
+    },
+    {
+      "epoch": 0.543334274359921,
+      "grad_norm": 0.16322177648544312,
+      "learning_rate": 4.456672964306138e-06,
+      "loss": 0.9178,
+      "step": 75060
+    },
+    {
+      "epoch": 0.5434066610205072,
+      "grad_norm": 0.16943545639514923,
+      "learning_rate": 4.4566005776455514e-06,
+      "loss": 0.9102,
+      "step": 75070
+    },
+    {
+      "epoch": 0.5434790476810933,
+      "grad_norm": 0.16489852964878082,
+      "learning_rate": 4.456528190984966e-06,
+      "loss": 0.9265,
+      "step": 75080
+    },
+    {
+      "epoch": 0.5435514343416795,
+      "grad_norm": 0.198415607213974,
+      "learning_rate": 4.4564558043243795e-06,
+      "loss": 0.9159,
+      "step": 75090
+    },
+    {
+      "epoch": 0.5436238210022657,
+      "grad_norm": 0.15088145434856415,
+      "learning_rate": 4.456383417663793e-06,
+      "loss": 0.9112,
+      "step": 75100
+    },
+    {
+      "epoch": 0.5436962076628519,
+      "grad_norm": 0.1523219645023346,
+      "learning_rate": 4.456311031003207e-06,
+      "loss": 0.9206,
+      "step": 75110
+    },
+    {
+      "epoch": 0.543768594323438,
+      "grad_norm": 0.1473480463027954,
+      "learning_rate": 4.456238644342621e-06,
+      "loss": 0.918,
+      "step": 75120
+    },
+    {
+      "epoch": 0.5438409809840242,
+      "grad_norm": 0.1520926058292389,
+      "learning_rate": 4.456166257682035e-06,
+      "loss": 0.9063,
+      "step": 75130
+    },
+    {
+      "epoch": 0.5439133676446104,
+      "grad_norm": 0.15161623060703278,
+      "learning_rate": 4.4560938710214485e-06,
+      "loss": 0.9107,
+      "step": 75140
+    },
+    {
+      "epoch": 0.5439857543051967,
+      "grad_norm": 0.15313850343227386,
+      "learning_rate": 4.456021484360862e-06,
+      "loss": 0.9073,
+      "step": 75150
+    },
+    {
+      "epoch": 0.5440581409657829,
+      "grad_norm": 0.14688560366630554,
+      "learning_rate": 4.4559490977002765e-06,
+      "loss": 0.9086,
+      "step": 75160
+    },
+    {
+      "epoch": 0.544130527626369,
+      "grad_norm": 0.15404599905014038,
+      "learning_rate": 4.45587671103969e-06,
+      "loss": 0.9174,
+      "step": 75170
+    },
+    {
+      "epoch": 0.5442029142869552,
+      "grad_norm": 0.16538633406162262,
+      "learning_rate": 4.455804324379104e-06,
+      "loss": 0.9209,
+      "step": 75180
+    },
+    {
+      "epoch": 0.5442753009475414,
+      "grad_norm": 0.16180939972400665,
+      "learning_rate": 4.455731937718517e-06,
+      "loss": 0.9071,
+      "step": 75190
+    },
+    {
+      "epoch": 0.5443476876081276,
+      "grad_norm": 0.1648901402950287,
+      "learning_rate": 4.455659551057932e-06,
+      "loss": 0.9119,
+      "step": 75200
+    },
+    {
+      "epoch": 0.5444200742687137,
+      "grad_norm": 0.1480080932378769,
+      "learning_rate": 4.4555871643973455e-06,
+      "loss": 0.9057,
+      "step": 75210
+    },
+    {
+      "epoch": 0.5444924609292999,
+      "grad_norm": 0.16776438057422638,
+      "learning_rate": 4.455514777736759e-06,
+      "loss": 0.9274,
+      "step": 75220
+    },
+    {
+      "epoch": 0.5445648475898861,
+      "grad_norm": 0.1602822095155716,
+      "learning_rate": 4.455442391076173e-06,
+      "loss": 0.9214,
+      "step": 75230
+    },
+    {
+      "epoch": 0.5446372342504723,
+      "grad_norm": 0.14920540153980255,
+      "learning_rate": 4.455370004415587e-06,
+      "loss": 0.922,
+      "step": 75240
+    },
+    {
+      "epoch": 0.5447096209110585,
+      "grad_norm": 0.14813263714313507,
+      "learning_rate": 4.455297617755001e-06,
+      "loss": 0.9304,
+      "step": 75250
+    },
+    {
+      "epoch": 0.5447820075716447,
+      "grad_norm": 0.16396941244602203,
+      "learning_rate": 4.455225231094414e-06,
+      "loss": 0.9075,
+      "step": 75260
+    },
+    {
+      "epoch": 0.5448543942322309,
+      "grad_norm": 0.15782274305820465,
+      "learning_rate": 4.455152844433828e-06,
+      "loss": 0.9049,
+      "step": 75270
+    },
+    {
+      "epoch": 0.5449267808928171,
+      "grad_norm": 0.15898077189922333,
+      "learning_rate": 4.4550804577732425e-06,
+      "loss": 0.9095,
+      "step": 75280
+    },
+    {
+      "epoch": 0.5449991675534033,
+      "grad_norm": 0.15519340336322784,
+      "learning_rate": 4.455008071112656e-06,
+      "loss": 0.9105,
+      "step": 75290
+    },
+    {
+      "epoch": 0.5450715542139895,
+      "grad_norm": 0.1506635993719101,
+      "learning_rate": 4.45493568445207e-06,
+      "loss": 0.9096,
+      "step": 75300
+    },
+    {
+      "epoch": 0.5451439408745756,
+      "grad_norm": 0.15947678685188293,
+      "learning_rate": 4.454863297791483e-06,
+      "loss": 0.9099,
+      "step": 75310
+    },
+    {
+      "epoch": 0.5452163275351618,
+      "grad_norm": 0.1622234433889389,
+      "learning_rate": 4.454790911130897e-06,
+      "loss": 0.9208,
+      "step": 75320
+    },
+    {
+      "epoch": 0.545288714195748,
+      "grad_norm": 0.16322208940982819,
+      "learning_rate": 4.454718524470311e-06,
+      "loss": 0.8956,
+      "step": 75330
+    },
+    {
+      "epoch": 0.5453611008563342,
+      "grad_norm": 0.16314153373241425,
+      "learning_rate": 4.454646137809725e-06,
+      "loss": 0.9099,
+      "step": 75340
+    },
+    {
+      "epoch": 0.5454334875169203,
+      "grad_norm": 0.15650366246700287,
+      "learning_rate": 4.454573751149139e-06,
+      "loss": 0.9143,
+      "step": 75350
+    },
+    {
+      "epoch": 0.5455058741775066,
+      "grad_norm": 0.16160249710083008,
+      "learning_rate": 4.454501364488552e-06,
+      "loss": 0.9086,
+      "step": 75360
+    },
+    {
+      "epoch": 0.5455782608380928,
+      "grad_norm": 0.18455857038497925,
+      "learning_rate": 4.454428977827967e-06,
+      "loss": 0.916,
+      "step": 75370
+    },
+    {
+      "epoch": 0.545650647498679,
+      "grad_norm": 0.16547948122024536,
+      "learning_rate": 4.45435659116738e-06,
+      "loss": 0.9188,
+      "step": 75380
+    },
+    {
+      "epoch": 0.5457230341592652,
+      "grad_norm": 0.15037128329277039,
+      "learning_rate": 4.454284204506794e-06,
+      "loss": 0.9107,
+      "step": 75390
+    },
+    {
+      "epoch": 0.5457954208198513,
+      "grad_norm": 0.15458130836486816,
+      "learning_rate": 4.4542118178462076e-06,
+      "loss": 0.9162,
+      "step": 75400
+    },
+    {
+      "epoch": 0.5458678074804375,
+      "grad_norm": 0.17746815085411072,
+      "learning_rate": 4.454139431185622e-06,
+      "loss": 0.9108,
+      "step": 75410
+    },
+    {
+      "epoch": 0.5459401941410237,
+      "grad_norm": 0.15650658309459686,
+      "learning_rate": 4.454067044525035e-06,
+      "loss": 0.9192,
+      "step": 75420
+    },
+    {
+      "epoch": 0.5460125808016099,
+      "grad_norm": 0.16250117123126984,
+      "learning_rate": 4.453994657864449e-06,
+      "loss": 0.9132,
+      "step": 75430
+    },
+    {
+      "epoch": 0.546084967462196,
+      "grad_norm": 0.1670369803905487,
+      "learning_rate": 4.453922271203863e-06,
+      "loss": 0.9135,
+      "step": 75440
+    },
+    {
+      "epoch": 0.5461573541227822,
+      "grad_norm": 1.3092010021209717,
+      "learning_rate": 4.4538498845432765e-06,
+      "loss": 0.9078,
+      "step": 75450
+    },
+    {
+      "epoch": 0.5462297407833684,
+      "grad_norm": 0.17468515038490295,
+      "learning_rate": 4.45377749788269e-06,
+      "loss": 0.9231,
+      "step": 75460
+    },
+    {
+      "epoch": 0.5463021274439547,
+      "grad_norm": 0.17053230106830597,
+      "learning_rate": 4.453705111222104e-06,
+      "loss": 0.9126,
+      "step": 75470
+    },
+    {
+      "epoch": 0.5463745141045409,
+      "grad_norm": 0.17557454109191895,
+      "learning_rate": 4.453632724561518e-06,
+      "loss": 0.9228,
+      "step": 75480
+    },
+    {
+      "epoch": 0.546446900765127,
+      "grad_norm": 0.17532938718795776,
+      "learning_rate": 4.453560337900932e-06,
+      "loss": 0.899,
+      "step": 75490
+    },
+    {
+      "epoch": 0.5465192874257132,
+      "grad_norm": 0.15332704782485962,
+      "learning_rate": 4.453487951240345e-06,
+      "loss": 0.9157,
+      "step": 75500
+    },
+    {
+      "epoch": 0.5465916740862994,
+      "grad_norm": 0.18341203033924103,
+      "learning_rate": 4.453415564579759e-06,
+      "loss": 0.9216,
+      "step": 75510
+    },
+    {
+      "epoch": 0.5466640607468856,
+      "grad_norm": 0.15997199714183807,
+      "learning_rate": 4.4533431779191735e-06,
+      "loss": 0.9244,
+      "step": 75520
+    },
+    {
+      "epoch": 0.5467364474074717,
+      "grad_norm": 0.1543131321668625,
+      "learning_rate": 4.453270791258587e-06,
+      "loss": 0.9075,
+      "step": 75530
+    },
+    {
+      "epoch": 0.5468088340680579,
+      "grad_norm": 0.16920112073421478,
+      "learning_rate": 4.453198404598001e-06,
+      "loss": 0.912,
+      "step": 75540
+    },
+    {
+      "epoch": 0.5468812207286441,
+      "grad_norm": 0.15737563371658325,
+      "learning_rate": 4.453126017937414e-06,
+      "loss": 0.9132,
+      "step": 75550
+    },
+    {
+      "epoch": 0.5469536073892303,
+      "grad_norm": 0.175185427069664,
+      "learning_rate": 4.453053631276829e-06,
+      "loss": 0.9211,
+      "step": 75560
+    },
+    {
+      "epoch": 0.5470259940498166,
+      "grad_norm": 0.163910910487175,
+      "learning_rate": 4.4529812446162424e-06,
+      "loss": 0.9075,
+      "step": 75570
+    },
+    {
+      "epoch": 0.5470983807104027,
+      "grad_norm": 0.185495063662529,
+      "learning_rate": 4.452908857955656e-06,
+      "loss": 0.929,
+      "step": 75580
+    },
+    {
+      "epoch": 0.5471707673709889,
+      "grad_norm": 0.159888356924057,
+      "learning_rate": 4.45283647129507e-06,
+      "loss": 0.9159,
+      "step": 75590
+    },
+    {
+      "epoch": 0.5472431540315751,
+      "grad_norm": 0.14591063559055328,
+      "learning_rate": 4.452764084634484e-06,
+      "loss": 0.9206,
+      "step": 75600
+    },
+    {
+      "epoch": 0.5473155406921613,
+      "grad_norm": 0.16549073159694672,
+      "learning_rate": 4.452691697973898e-06,
+      "loss": 0.8996,
+      "step": 75610
+    },
+    {
+      "epoch": 0.5473879273527474,
+      "grad_norm": 0.16032549738883972,
+      "learning_rate": 4.452619311313311e-06,
+      "loss": 0.9198,
+      "step": 75620
+    },
+    {
+      "epoch": 0.5474603140133336,
+      "grad_norm": 0.15921436250209808,
+      "learning_rate": 4.452546924652725e-06,
+      "loss": 0.8986,
+      "step": 75630
+    },
+    {
+      "epoch": 0.5475327006739198,
+      "grad_norm": 0.15605443716049194,
+      "learning_rate": 4.4524745379921394e-06,
+      "loss": 0.911,
+      "step": 75640
+    },
+    {
+      "epoch": 0.547605087334506,
+      "grad_norm": 0.1790502518415451,
+      "learning_rate": 4.452402151331553e-06,
+      "loss": 0.9153,
+      "step": 75650
+    },
+    {
+      "epoch": 0.5476774739950921,
+      "grad_norm": 0.16580620408058167,
+      "learning_rate": 4.452329764670967e-06,
+      "loss": 0.9162,
+      "step": 75660
+    },
+    {
+      "epoch": 0.5477498606556783,
+      "grad_norm": 0.19182774424552917,
+      "learning_rate": 4.45225737801038e-06,
+      "loss": 0.9188,
+      "step": 75670
+    },
+    {
+      "epoch": 0.5478222473162646,
+      "grad_norm": 0.16452908515930176,
+      "learning_rate": 4.452184991349795e-06,
+      "loss": 0.9187,
+      "step": 75680
+    },
+    {
+      "epoch": 0.5478946339768508,
+      "grad_norm": 0.15444496273994446,
+      "learning_rate": 4.452112604689208e-06,
+      "loss": 0.9169,
+      "step": 75690
+    },
+    {
+      "epoch": 0.547967020637437,
+      "grad_norm": 0.149451345205307,
+      "learning_rate": 4.452040218028622e-06,
+      "loss": 0.9236,
+      "step": 75700
+    },
+    {
+      "epoch": 0.5480394072980231,
+      "grad_norm": 0.16976690292358398,
+      "learning_rate": 4.451967831368036e-06,
+      "loss": 0.9086,
+      "step": 75710
+    },
+    {
+      "epoch": 0.5481117939586093,
+      "grad_norm": 0.14885394275188446,
+      "learning_rate": 4.45189544470745e-06,
+      "loss": 0.9012,
+      "step": 75720
+    },
+    {
+      "epoch": 0.5481841806191955,
+      "grad_norm": 0.1846446394920349,
+      "learning_rate": 4.451823058046864e-06,
+      "loss": 0.9183,
+      "step": 75730
+    },
+    {
+      "epoch": 0.5482565672797817,
+      "grad_norm": 0.15907852351665497,
+      "learning_rate": 4.451750671386277e-06,
+      "loss": 0.9219,
+      "step": 75740
+    },
+    {
+      "epoch": 0.5483289539403678,
+      "grad_norm": 0.1452861726284027,
+      "learning_rate": 4.451678284725691e-06,
+      "loss": 0.9102,
+      "step": 75750
+    },
+    {
+      "epoch": 0.548401340600954,
+      "grad_norm": 0.16507689654827118,
+      "learning_rate": 4.451605898065105e-06,
+      "loss": 0.9243,
+      "step": 75760
+    },
+    {
+      "epoch": 0.5484737272615402,
+      "grad_norm": 0.15828417241573334,
+      "learning_rate": 4.451533511404519e-06,
+      "loss": 0.9028,
+      "step": 75770
+    },
+    {
+      "epoch": 0.5485461139221264,
+      "grad_norm": 0.18392127752304077,
+      "learning_rate": 4.451461124743933e-06,
+      "loss": 0.908,
+      "step": 75780
+    },
+    {
+      "epoch": 0.5486185005827127,
+      "grad_norm": 0.15240317583084106,
+      "learning_rate": 4.451388738083346e-06,
+      "loss": 0.9013,
+      "step": 75790
+    },
+    {
+      "epoch": 0.5486908872432988,
+      "grad_norm": 0.14687864482402802,
+      "learning_rate": 4.451316351422761e-06,
+      "loss": 0.9184,
+      "step": 75800
+    },
+    {
+      "epoch": 0.548763273903885,
+      "grad_norm": 0.15710905194282532,
+      "learning_rate": 4.451243964762174e-06,
+      "loss": 0.9165,
+      "step": 75810
+    },
+    {
+      "epoch": 0.5488356605644712,
+      "grad_norm": 0.1660161018371582,
+      "learning_rate": 4.451171578101588e-06,
+      "loss": 0.9045,
+      "step": 75820
+    },
+    {
+      "epoch": 0.5489080472250574,
+      "grad_norm": 0.1465510129928589,
+      "learning_rate": 4.4510991914410015e-06,
+      "loss": 0.8998,
+      "step": 75830
+    },
+    {
+      "epoch": 0.5489804338856435,
+      "grad_norm": 0.17025746405124664,
+      "learning_rate": 4.451026804780416e-06,
+      "loss": 0.9093,
+      "step": 75840
+    },
+    {
+      "epoch": 0.5490528205462297,
+      "grad_norm": 0.18562918901443481,
+      "learning_rate": 4.45095441811983e-06,
+      "loss": 0.9195,
+      "step": 75850
+    },
+    {
+      "epoch": 0.5491252072068159,
+      "grad_norm": 0.1641816943883896,
+      "learning_rate": 4.450882031459243e-06,
+      "loss": 0.9039,
+      "step": 75860
+    },
+    {
+      "epoch": 0.5491975938674021,
+      "grad_norm": 0.19206029176712036,
+      "learning_rate": 4.450809644798657e-06,
+      "loss": 0.9224,
+      "step": 75870
+    },
+    {
+      "epoch": 0.5492699805279883,
+      "grad_norm": 0.15584112703800201,
+      "learning_rate": 4.450737258138071e-06,
+      "loss": 0.9171,
+      "step": 75880
+    },
+    {
+      "epoch": 0.5493423671885745,
+      "grad_norm": 0.1531420797109604,
+      "learning_rate": 4.450664871477485e-06,
+      "loss": 0.8936,
+      "step": 75890
+    },
+    {
+      "epoch": 0.5494147538491607,
+      "grad_norm": 0.17042067646980286,
+      "learning_rate": 4.4505924848168985e-06,
+      "loss": 0.9002,
+      "step": 75900
+    },
+    {
+      "epoch": 0.5494871405097469,
+      "grad_norm": 0.15222638845443726,
+      "learning_rate": 4.450520098156312e-06,
+      "loss": 0.9058,
+      "step": 75910
+    },
+    {
+      "epoch": 0.5495595271703331,
+      "grad_norm": 0.1593593806028366,
+      "learning_rate": 4.450447711495726e-06,
+      "loss": 0.8956,
+      "step": 75920
+    },
+    {
+      "epoch": 0.5496319138309192,
+      "grad_norm": 0.16259454190731049,
+      "learning_rate": 4.45037532483514e-06,
+      "loss": 0.9124,
+      "step": 75930
+    },
+    {
+      "epoch": 0.5497043004915054,
+      "grad_norm": 0.23078764975070953,
+      "learning_rate": 4.450302938174554e-06,
+      "loss": 0.9127,
+      "step": 75940
+    },
+    {
+      "epoch": 0.5497766871520916,
+      "grad_norm": 0.1593347191810608,
+      "learning_rate": 4.4502305515139675e-06,
+      "loss": 0.8963,
+      "step": 75950
+    },
+    {
+      "epoch": 0.5498490738126778,
+      "grad_norm": 0.1672598272562027,
+      "learning_rate": 4.450158164853381e-06,
+      "loss": 0.9212,
+      "step": 75960
+    },
+    {
+      "epoch": 0.549921460473264,
+      "grad_norm": 0.1680530607700348,
+      "learning_rate": 4.450085778192795e-06,
+      "loss": 0.9194,
+      "step": 75970
+    },
+    {
+      "epoch": 0.5499938471338501,
+      "grad_norm": 0.15082144737243652,
+      "learning_rate": 4.450013391532208e-06,
+      "loss": 0.9028,
+      "step": 75980
+    },
+    {
+      "epoch": 0.5500662337944363,
+      "grad_norm": 0.1934494823217392,
+      "learning_rate": 4.449941004871623e-06,
+      "loss": 0.907,
+      "step": 75990
+    },
+    {
+      "epoch": 0.5501386204550226,
+      "grad_norm": 0.15488490462303162,
+      "learning_rate": 4.449868618211036e-06,
+      "loss": 0.9231,
+      "step": 76000
+    },
+    {
+      "epoch": 0.5502110071156088,
+      "grad_norm": 0.1557077169418335,
+      "learning_rate": 4.44979623155045e-06,
+      "loss": 0.9252,
+      "step": 76010
+    },
+    {
+      "epoch": 0.550283393776195,
+      "grad_norm": 0.16591419279575348,
+      "learning_rate": 4.449723844889864e-06,
+      "loss": 0.9057,
+      "step": 76020
+    },
+    {
+      "epoch": 0.5503557804367811,
+      "grad_norm": 0.16087841987609863,
+      "learning_rate": 4.449651458229278e-06,
+      "loss": 0.9173,
+      "step": 76030
+    },
+    {
+      "epoch": 0.5504281670973673,
+      "grad_norm": 0.16236597299575806,
+      "learning_rate": 4.449579071568692e-06,
+      "loss": 0.9211,
+      "step": 76040
+    },
+    {
+      "epoch": 0.5505005537579535,
+      "grad_norm": 0.17655915021896362,
+      "learning_rate": 4.449506684908105e-06,
+      "loss": 0.9211,
+      "step": 76050
+    },
+    {
+      "epoch": 0.5505729404185397,
+      "grad_norm": 0.1689653992652893,
+      "learning_rate": 4.449434298247519e-06,
+      "loss": 0.9139,
+      "step": 76060
+    },
+    {
+      "epoch": 0.5506453270791258,
+      "grad_norm": 0.16405482590198517,
+      "learning_rate": 4.449361911586933e-06,
+      "loss": 0.9165,
+      "step": 76070
+    },
+    {
+      "epoch": 0.550717713739712,
+      "grad_norm": 0.16635659337043762,
+      "learning_rate": 4.449289524926347e-06,
+      "loss": 0.8991,
+      "step": 76080
+    },
+    {
+      "epoch": 0.5507901004002982,
+      "grad_norm": 0.1841815710067749,
+      "learning_rate": 4.449217138265761e-06,
+      "loss": 0.9175,
+      "step": 76090
+    },
+    {
+      "epoch": 0.5508624870608844,
+      "grad_norm": 0.15842583775520325,
+      "learning_rate": 4.449144751605174e-06,
+      "loss": 0.9144,
+      "step": 76100
+    },
+    {
+      "epoch": 0.5509348737214707,
+      "grad_norm": 0.16369058191776276,
+      "learning_rate": 4.449072364944588e-06,
+      "loss": 0.9071,
+      "step": 76110
+    },
+    {
+      "epoch": 0.5510072603820568,
+      "grad_norm": 0.16025404632091522,
+      "learning_rate": 4.448999978284002e-06,
+      "loss": 0.9108,
+      "step": 76120
+    },
+    {
+      "epoch": 0.551079647042643,
+      "grad_norm": 0.15726248919963837,
+      "learning_rate": 4.448927591623416e-06,
+      "loss": 0.9048,
+      "step": 76130
+    },
+    {
+      "epoch": 0.5511520337032292,
+      "grad_norm": 0.149103045463562,
+      "learning_rate": 4.4488552049628296e-06,
+      "loss": 0.9041,
+      "step": 76140
+    },
+    {
+      "epoch": 0.5512244203638154,
+      "grad_norm": 0.1646459996700287,
+      "learning_rate": 4.448782818302243e-06,
+      "loss": 0.9207,
+      "step": 76150
+    },
+    {
+      "epoch": 0.5512968070244015,
+      "grad_norm": 0.16346558928489685,
+      "learning_rate": 4.448710431641658e-06,
+      "loss": 0.9152,
+      "step": 76160
+    },
+    {
+      "epoch": 0.5513691936849877,
+      "grad_norm": 0.15449324250221252,
+      "learning_rate": 4.448638044981071e-06,
+      "loss": 0.9028,
+      "step": 76170
+    },
+    {
+      "epoch": 0.5514415803455739,
+      "grad_norm": 0.1610867977142334,
+      "learning_rate": 4.448565658320485e-06,
+      "loss": 0.9255,
+      "step": 76180
+    },
+    {
+      "epoch": 0.5515139670061601,
+      "grad_norm": 0.15043221414089203,
+      "learning_rate": 4.4484932716598985e-06,
+      "loss": 0.9077,
+      "step": 76190
+    },
+    {
+      "epoch": 0.5515863536667462,
+      "grad_norm": 0.18590010702610016,
+      "learning_rate": 4.448420884999313e-06,
+      "loss": 0.9125,
+      "step": 76200
+    },
+    {
+      "epoch": 0.5516587403273325,
+      "grad_norm": 0.16317640244960785,
+      "learning_rate": 4.4483484983387266e-06,
+      "loss": 0.9219,
+      "step": 76210
+    },
+    {
+      "epoch": 0.5517311269879187,
+      "grad_norm": 0.15565603971481323,
+      "learning_rate": 4.44827611167814e-06,
+      "loss": 0.9128,
+      "step": 76220
+    },
+    {
+      "epoch": 0.5518035136485049,
+      "grad_norm": 0.1605430394411087,
+      "learning_rate": 4.448203725017554e-06,
+      "loss": 0.9104,
+      "step": 76230
+    },
+    {
+      "epoch": 0.5518759003090911,
+      "grad_norm": 0.17234712839126587,
+      "learning_rate": 4.448131338356968e-06,
+      "loss": 0.9054,
+      "step": 76240
+    },
+    {
+      "epoch": 0.5519482869696772,
+      "grad_norm": 0.15464408695697784,
+      "learning_rate": 4.448058951696382e-06,
+      "loss": 0.9115,
+      "step": 76250
+    },
+    {
+      "epoch": 0.5520206736302634,
+      "grad_norm": 0.15122352540493011,
+      "learning_rate": 4.4479865650357955e-06,
+      "loss": 0.9039,
+      "step": 76260
+    },
+    {
+      "epoch": 0.5520930602908496,
+      "grad_norm": 0.1647047996520996,
+      "learning_rate": 4.447914178375209e-06,
+      "loss": 0.9161,
+      "step": 76270
+    },
+    {
+      "epoch": 0.5521654469514358,
+      "grad_norm": 0.16340136528015137,
+      "learning_rate": 4.4478417917146236e-06,
+      "loss": 0.9097,
+      "step": 76280
+    },
+    {
+      "epoch": 0.5522378336120219,
+      "grad_norm": 0.1576181948184967,
+      "learning_rate": 4.447769405054037e-06,
+      "loss": 0.9111,
+      "step": 76290
+    },
+    {
+      "epoch": 0.5523102202726081,
+      "grad_norm": 0.17520330846309662,
+      "learning_rate": 4.447697018393451e-06,
+      "loss": 0.9019,
+      "step": 76300
+    },
+    {
+      "epoch": 0.5523826069331943,
+      "grad_norm": 0.16088417172431946,
+      "learning_rate": 4.447624631732864e-06,
+      "loss": 0.9159,
+      "step": 76310
+    },
+    {
+      "epoch": 0.5524549935937806,
+      "grad_norm": 0.15434597432613373,
+      "learning_rate": 4.447552245072279e-06,
+      "loss": 0.9033,
+      "step": 76320
+    },
+    {
+      "epoch": 0.5525273802543668,
+      "grad_norm": 0.16620974242687225,
+      "learning_rate": 4.4474798584116925e-06,
+      "loss": 0.9203,
+      "step": 76330
+    },
+    {
+      "epoch": 0.5525997669149529,
+      "grad_norm": 0.159925177693367,
+      "learning_rate": 4.447407471751106e-06,
+      "loss": 0.9086,
+      "step": 76340
+    },
+    {
+      "epoch": 0.5526721535755391,
+      "grad_norm": 0.19963112473487854,
+      "learning_rate": 4.44733508509052e-06,
+      "loss": 0.9212,
+      "step": 76350
+    },
+    {
+      "epoch": 0.5527445402361253,
+      "grad_norm": 0.1715710163116455,
+      "learning_rate": 4.447262698429934e-06,
+      "loss": 0.9205,
+      "step": 76360
+    },
+    {
+      "epoch": 0.5528169268967115,
+      "grad_norm": 0.1506585329771042,
+      "learning_rate": 4.447190311769348e-06,
+      "loss": 0.9063,
+      "step": 76370
+    },
+    {
+      "epoch": 0.5528893135572976,
+      "grad_norm": 0.16564472019672394,
+      "learning_rate": 4.4471179251087614e-06,
+      "loss": 0.9322,
+      "step": 76380
+    },
+    {
+      "epoch": 0.5529617002178838,
+      "grad_norm": 0.1629706174135208,
+      "learning_rate": 4.447045538448175e-06,
+      "loss": 0.8981,
+      "step": 76390
+    },
+    {
+      "epoch": 0.55303408687847,
+      "grad_norm": 0.15394346415996552,
+      "learning_rate": 4.4469731517875895e-06,
+      "loss": 0.918,
+      "step": 76400
+    },
+    {
+      "epoch": 0.5531064735390562,
+      "grad_norm": 0.1582847237586975,
+      "learning_rate": 4.446900765127003e-06,
+      "loss": 0.9083,
+      "step": 76410
+    },
+    {
+      "epoch": 0.5531788601996425,
+      "grad_norm": 0.16456353664398193,
+      "learning_rate": 4.446828378466417e-06,
+      "loss": 0.9225,
+      "step": 76420
+    },
+    {
+      "epoch": 0.5532512468602286,
+      "grad_norm": 0.1710321456193924,
+      "learning_rate": 4.44675599180583e-06,
+      "loss": 0.916,
+      "step": 76430
+    },
+    {
+      "epoch": 0.5533236335208148,
+      "grad_norm": 0.15686561167240143,
+      "learning_rate": 4.446683605145245e-06,
+      "loss": 0.9092,
+      "step": 76440
+    },
+    {
+      "epoch": 0.553396020181401,
+      "grad_norm": 0.15640173852443695,
+      "learning_rate": 4.4466112184846584e-06,
+      "loss": 0.9061,
+      "step": 76450
+    },
+    {
+      "epoch": 0.5534684068419872,
+      "grad_norm": 0.1626739203929901,
+      "learning_rate": 4.446538831824072e-06,
+      "loss": 0.9156,
+      "step": 76460
+    },
+    {
+      "epoch": 0.5535407935025733,
+      "grad_norm": 0.15522697567939758,
+      "learning_rate": 4.446466445163486e-06,
+      "loss": 0.8936,
+      "step": 76470
+    },
+    {
+      "epoch": 0.5536131801631595,
+      "grad_norm": 0.15257471799850464,
+      "learning_rate": 4.446394058502899e-06,
+      "loss": 0.9053,
+      "step": 76480
+    },
+    {
+      "epoch": 0.5536855668237457,
+      "grad_norm": 0.16405083239078522,
+      "learning_rate": 4.446321671842313e-06,
+      "loss": 0.9124,
+      "step": 76490
+    },
+    {
+      "epoch": 0.5537579534843319,
+      "grad_norm": 0.1680813431739807,
+      "learning_rate": 4.4462492851817265e-06,
+      "loss": 0.9134,
+      "step": 76500
+    },
+    {
+      "epoch": 0.553830340144918,
+      "grad_norm": 0.16853106021881104,
+      "learning_rate": 4.446176898521141e-06,
+      "loss": 0.926,
+      "step": 76510
+    },
+    {
+      "epoch": 0.5539027268055042,
+      "grad_norm": 0.164813831448555,
+      "learning_rate": 4.446104511860555e-06,
+      "loss": 0.9091,
+      "step": 76520
+    },
+    {
+      "epoch": 0.5539751134660905,
+      "grad_norm": 0.1515646129846573,
+      "learning_rate": 4.446032125199968e-06,
+      "loss": 0.9083,
+      "step": 76530
+    },
+    {
+      "epoch": 0.5540475001266767,
+      "grad_norm": 0.16372545063495636,
+      "learning_rate": 4.445959738539382e-06,
+      "loss": 0.8965,
+      "step": 76540
+    },
+    {
+      "epoch": 0.5541198867872629,
+      "grad_norm": 0.1647547334432602,
+      "learning_rate": 4.445887351878796e-06,
+      "loss": 0.8986,
+      "step": 76550
+    },
+    {
+      "epoch": 0.554192273447849,
+      "grad_norm": 0.1702655404806137,
+      "learning_rate": 4.44581496521821e-06,
+      "loss": 0.9111,
+      "step": 76560
+    },
+    {
+      "epoch": 0.5542646601084352,
+      "grad_norm": 0.1671968400478363,
+      "learning_rate": 4.4457425785576235e-06,
+      "loss": 0.9038,
+      "step": 76570
+    },
+    {
+      "epoch": 0.5543370467690214,
+      "grad_norm": 0.23342488706111908,
+      "learning_rate": 4.445670191897037e-06,
+      "loss": 0.8895,
+      "step": 76580
+    },
+    {
+      "epoch": 0.5544094334296076,
+      "grad_norm": 0.17012055218219757,
+      "learning_rate": 4.445597805236452e-06,
+      "loss": 0.9223,
+      "step": 76590
+    },
+    {
+      "epoch": 0.5544818200901938,
+      "grad_norm": 0.153640478849411,
+      "learning_rate": 4.445525418575865e-06,
+      "loss": 0.9189,
+      "step": 76600
+    },
+    {
+      "epoch": 0.5545542067507799,
+      "grad_norm": 0.1719810962677002,
+      "learning_rate": 4.445453031915279e-06,
+      "loss": 0.8992,
+      "step": 76610
+    },
+    {
+      "epoch": 0.5546265934113661,
+      "grad_norm": 0.16828425228595734,
+      "learning_rate": 4.4453806452546925e-06,
+      "loss": 0.9154,
+      "step": 76620
+    },
+    {
+      "epoch": 0.5546989800719523,
+      "grad_norm": 0.1485888808965683,
+      "learning_rate": 4.445308258594107e-06,
+      "loss": 0.9044,
+      "step": 76630
+    },
+    {
+      "epoch": 0.5547713667325386,
+      "grad_norm": 0.15807074308395386,
+      "learning_rate": 4.4452358719335205e-06,
+      "loss": 0.9193,
+      "step": 76640
+    },
+    {
+      "epoch": 0.5548437533931248,
+      "grad_norm": 0.2003334015607834,
+      "learning_rate": 4.445163485272934e-06,
+      "loss": 0.9226,
+      "step": 76650
+    },
+    {
+      "epoch": 0.5549161400537109,
+      "grad_norm": 0.16058491170406342,
+      "learning_rate": 4.445091098612348e-06,
+      "loss": 0.9208,
+      "step": 76660
+    },
+    {
+      "epoch": 0.5549885267142971,
+      "grad_norm": 0.22712364792823792,
+      "learning_rate": 4.445018711951762e-06,
+      "loss": 0.9116,
+      "step": 76670
+    },
+    {
+      "epoch": 0.5550609133748833,
+      "grad_norm": 0.16164818406105042,
+      "learning_rate": 4.444946325291176e-06,
+      "loss": 0.9143,
+      "step": 76680
+    },
+    {
+      "epoch": 0.5551333000354695,
+      "grad_norm": 0.1746838092803955,
+      "learning_rate": 4.4448739386305895e-06,
+      "loss": 0.9179,
+      "step": 76690
+    },
+    {
+      "epoch": 0.5552056866960556,
+      "grad_norm": 0.1547100841999054,
+      "learning_rate": 4.444801551970003e-06,
+      "loss": 0.9122,
+      "step": 76700
+    },
+    {
+      "epoch": 0.5552780733566418,
+      "grad_norm": 0.1596509963274002,
+      "learning_rate": 4.444729165309417e-06,
+      "loss": 0.908,
+      "step": 76710
+    },
+    {
+      "epoch": 0.555350460017228,
+      "grad_norm": 0.16364139318466187,
+      "learning_rate": 4.444656778648831e-06,
+      "loss": 0.9158,
+      "step": 76720
+    },
+    {
+      "epoch": 0.5554228466778142,
+      "grad_norm": 0.17858783900737762,
+      "learning_rate": 4.444584391988245e-06,
+      "loss": 0.9134,
+      "step": 76730
+    },
+    {
+      "epoch": 0.5554952333384005,
+      "grad_norm": 0.14799803495407104,
+      "learning_rate": 4.444512005327658e-06,
+      "loss": 0.9138,
+      "step": 76740
+    },
+    {
+      "epoch": 0.5555676199989866,
+      "grad_norm": 0.156327486038208,
+      "learning_rate": 4.444439618667072e-06,
+      "loss": 0.9222,
+      "step": 76750
+    },
+    {
+      "epoch": 0.5556400066595728,
+      "grad_norm": 0.16460415720939636,
+      "learning_rate": 4.4443672320064865e-06,
+      "loss": 0.9107,
+      "step": 76760
+    },
+    {
+      "epoch": 0.555712393320159,
+      "grad_norm": 0.1606915295124054,
+      "learning_rate": 4.4442948453459e-06,
+      "loss": 0.8984,
+      "step": 76770
+    },
+    {
+      "epoch": 0.5557847799807452,
+      "grad_norm": 0.15452681481838226,
+      "learning_rate": 4.444222458685314e-06,
+      "loss": 0.8986,
+      "step": 76780
+    },
+    {
+      "epoch": 0.5558571666413313,
+      "grad_norm": 0.16996990144252777,
+      "learning_rate": 4.444150072024727e-06,
+      "loss": 0.919,
+      "step": 76790
+    },
+    {
+      "epoch": 0.5559295533019175,
+      "grad_norm": 0.16095289587974548,
+      "learning_rate": 4.444077685364142e-06,
+      "loss": 0.9196,
+      "step": 76800
+    },
+    {
+      "epoch": 0.5560019399625037,
+      "grad_norm": 0.16664938628673553,
+      "learning_rate": 4.444005298703555e-06,
+      "loss": 0.915,
+      "step": 76810
+    },
+    {
+      "epoch": 0.5560743266230899,
+      "grad_norm": 0.15813925862312317,
+      "learning_rate": 4.443932912042969e-06,
+      "loss": 0.9096,
+      "step": 76820
+    },
+    {
+      "epoch": 0.556146713283676,
+      "grad_norm": 0.1628894805908203,
+      "learning_rate": 4.443860525382383e-06,
+      "loss": 0.9181,
+      "step": 76830
+    },
+    {
+      "epoch": 0.5562190999442622,
+      "grad_norm": 0.16485193371772766,
+      "learning_rate": 4.443788138721797e-06,
+      "loss": 0.8998,
+      "step": 76840
+    },
+    {
+      "epoch": 0.5562914866048485,
+      "grad_norm": 0.15740840137004852,
+      "learning_rate": 4.443715752061211e-06,
+      "loss": 0.9215,
+      "step": 76850
+    },
+    {
+      "epoch": 0.5563638732654347,
+      "grad_norm": 0.16431400179862976,
+      "learning_rate": 4.443643365400624e-06,
+      "loss": 0.9179,
+      "step": 76860
+    },
+    {
+      "epoch": 0.5564362599260209,
+      "grad_norm": 0.17064224183559418,
+      "learning_rate": 4.443570978740038e-06,
+      "loss": 0.9115,
+      "step": 76870
+    },
+    {
+      "epoch": 0.556508646586607,
+      "grad_norm": 0.15299154818058014,
+      "learning_rate": 4.443498592079452e-06,
+      "loss": 0.9046,
+      "step": 76880
+    },
+    {
+      "epoch": 0.5565810332471932,
+      "grad_norm": 0.19055993854999542,
+      "learning_rate": 4.443426205418866e-06,
+      "loss": 0.9223,
+      "step": 76890
+    },
+    {
+      "epoch": 0.5566534199077794,
+      "grad_norm": 0.16058650612831116,
+      "learning_rate": 4.44335381875828e-06,
+      "loss": 0.9083,
+      "step": 76900
+    },
+    {
+      "epoch": 0.5567258065683656,
+      "grad_norm": 0.15559203922748566,
+      "learning_rate": 4.443281432097693e-06,
+      "loss": 0.9158,
+      "step": 76910
+    },
+    {
+      "epoch": 0.5567981932289517,
+      "grad_norm": 0.4795348346233368,
+      "learning_rate": 4.443209045437108e-06,
+      "loss": 0.9116,
+      "step": 76920
+    },
+    {
+      "epoch": 0.5568705798895379,
+      "grad_norm": 0.1700964719057083,
+      "learning_rate": 4.443136658776521e-06,
+      "loss": 0.907,
+      "step": 76930
+    },
+    {
+      "epoch": 0.5569429665501241,
+      "grad_norm": 0.15994440019130707,
+      "learning_rate": 4.443064272115935e-06,
+      "loss": 0.9222,
+      "step": 76940
+    },
+    {
+      "epoch": 0.5570153532107104,
+      "grad_norm": 0.17749464511871338,
+      "learning_rate": 4.4429918854553486e-06,
+      "loss": 0.9154,
+      "step": 76950
+    },
+    {
+      "epoch": 0.5570877398712966,
+      "grad_norm": 0.16541603207588196,
+      "learning_rate": 4.442919498794763e-06,
+      "loss": 0.9177,
+      "step": 76960
+    },
+    {
+      "epoch": 0.5571601265318827,
+      "grad_norm": 0.16451773047447205,
+      "learning_rate": 4.442847112134177e-06,
+      "loss": 0.9144,
+      "step": 76970
+    },
+    {
+      "epoch": 0.5572325131924689,
+      "grad_norm": 0.16148695349693298,
+      "learning_rate": 4.44277472547359e-06,
+      "loss": 0.9084,
+      "step": 76980
+    },
+    {
+      "epoch": 0.5573048998530551,
+      "grad_norm": 0.18457120656967163,
+      "learning_rate": 4.442702338813004e-06,
+      "loss": 0.9008,
+      "step": 76990
+    },
+    {
+      "epoch": 0.5573772865136413,
+      "grad_norm": 0.1575312465429306,
+      "learning_rate": 4.442629952152418e-06,
+      "loss": 0.9129,
+      "step": 77000
+    },
+    {
+      "epoch": 0.5574496731742274,
+      "grad_norm": 0.16345492005348206,
+      "learning_rate": 4.442557565491831e-06,
+      "loss": 0.9194,
+      "step": 77010
+    },
+    {
+      "epoch": 0.5575220598348136,
+      "grad_norm": 0.17905868589878082,
+      "learning_rate": 4.442485178831245e-06,
+      "loss": 0.9271,
+      "step": 77020
+    },
+    {
+      "epoch": 0.5575944464953998,
+      "grad_norm": 0.15946638584136963,
+      "learning_rate": 4.442412792170659e-06,
+      "loss": 0.9158,
+      "step": 77030
+    },
+    {
+      "epoch": 0.557666833155986,
+      "grad_norm": 0.1656288057565689,
+      "learning_rate": 4.442340405510073e-06,
+      "loss": 0.897,
+      "step": 77040
+    },
+    {
+      "epoch": 0.5577392198165722,
+      "grad_norm": 0.15466462075710297,
+      "learning_rate": 4.442268018849486e-06,
+      "loss": 0.9194,
+      "step": 77050
+    },
+    {
+      "epoch": 0.5578116064771584,
+      "grad_norm": 0.17180196940898895,
+      "learning_rate": 4.4421956321889e-06,
+      "loss": 0.9056,
+      "step": 77060
+    },
+    {
+      "epoch": 0.5578839931377446,
+      "grad_norm": 0.16797685623168945,
+      "learning_rate": 4.4421232455283145e-06,
+      "loss": 0.9069,
+      "step": 77070
+    },
+    {
+      "epoch": 0.5579563797983308,
+      "grad_norm": 0.15991143882274628,
+      "learning_rate": 4.442050858867728e-06,
+      "loss": 0.9102,
+      "step": 77080
+    },
+    {
+      "epoch": 0.558028766458917,
+      "grad_norm": 0.1536579132080078,
+      "learning_rate": 4.441978472207142e-06,
+      "loss": 0.9134,
+      "step": 77090
+    },
+    {
+      "epoch": 0.5581011531195031,
+      "grad_norm": 0.15524393320083618,
+      "learning_rate": 4.441906085546555e-06,
+      "loss": 0.9237,
+      "step": 77100
+    },
+    {
+      "epoch": 0.5581735397800893,
+      "grad_norm": 0.15506607294082642,
+      "learning_rate": 4.44183369888597e-06,
+      "loss": 0.925,
+      "step": 77110
+    },
+    {
+      "epoch": 0.5582459264406755,
+      "grad_norm": 0.1472669243812561,
+      "learning_rate": 4.4417613122253834e-06,
+      "loss": 0.8972,
+      "step": 77120
+    },
+    {
+      "epoch": 0.5583183131012617,
+      "grad_norm": 0.17821435630321503,
+      "learning_rate": 4.441688925564797e-06,
+      "loss": 0.9146,
+      "step": 77130
+    },
+    {
+      "epoch": 0.5583906997618479,
+      "grad_norm": 0.152136892080307,
+      "learning_rate": 4.441616538904211e-06,
+      "loss": 0.9101,
+      "step": 77140
+    },
+    {
+      "epoch": 0.558463086422434,
+      "grad_norm": 0.16555529832839966,
+      "learning_rate": 4.441544152243625e-06,
+      "loss": 0.915,
+      "step": 77150
+    },
+    {
+      "epoch": 0.5585354730830202,
+      "grad_norm": 0.18034303188323975,
+      "learning_rate": 4.441471765583039e-06,
+      "loss": 0.9174,
+      "step": 77160
+    },
+    {
+      "epoch": 0.5586078597436065,
+      "grad_norm": 0.1521337479352951,
+      "learning_rate": 4.441399378922452e-06,
+      "loss": 0.9194,
+      "step": 77170
+    },
+    {
+      "epoch": 0.5586802464041927,
+      "grad_norm": 0.1529371738433838,
+      "learning_rate": 4.441326992261866e-06,
+      "loss": 0.893,
+      "step": 77180
+    },
+    {
+      "epoch": 0.5587526330647788,
+      "grad_norm": 0.18080340325832367,
+      "learning_rate": 4.4412546056012804e-06,
+      "loss": 0.921,
+      "step": 77190
+    },
+    {
+      "epoch": 0.558825019725365,
+      "grad_norm": 0.15729157626628876,
+      "learning_rate": 4.441182218940694e-06,
+      "loss": 0.905,
+      "step": 77200
+    },
+    {
+      "epoch": 0.5588974063859512,
+      "grad_norm": 0.16402077674865723,
+      "learning_rate": 4.441109832280108e-06,
+      "loss": 0.9141,
+      "step": 77210
+    },
+    {
+      "epoch": 0.5589697930465374,
+      "grad_norm": 0.14309856295585632,
+      "learning_rate": 4.441037445619521e-06,
+      "loss": 0.909,
+      "step": 77220
+    },
+    {
+      "epoch": 0.5590421797071236,
+      "grad_norm": 0.168483704328537,
+      "learning_rate": 4.440965058958936e-06,
+      "loss": 0.9038,
+      "step": 77230
+    },
+    {
+      "epoch": 0.5591145663677097,
+      "grad_norm": 0.15855084359645844,
+      "learning_rate": 4.440892672298349e-06,
+      "loss": 0.9107,
+      "step": 77240
+    },
+    {
+      "epoch": 0.5591869530282959,
+      "grad_norm": 0.2141026258468628,
+      "learning_rate": 4.440820285637763e-06,
+      "loss": 0.917,
+      "step": 77250
+    },
+    {
+      "epoch": 0.5592593396888821,
+      "grad_norm": 0.14907251298427582,
+      "learning_rate": 4.440747898977177e-06,
+      "loss": 0.9142,
+      "step": 77260
+    },
+    {
+      "epoch": 0.5593317263494684,
+      "grad_norm": 0.16853876411914825,
+      "learning_rate": 4.440675512316591e-06,
+      "loss": 0.9273,
+      "step": 77270
+    },
+    {
+      "epoch": 0.5594041130100545,
+      "grad_norm": 0.15877121686935425,
+      "learning_rate": 4.440603125656005e-06,
+      "loss": 0.9126,
+      "step": 77280
+    },
+    {
+      "epoch": 0.5594764996706407,
+      "grad_norm": 0.1590542197227478,
+      "learning_rate": 4.440530738995418e-06,
+      "loss": 0.9254,
+      "step": 77290
+    },
+    {
+      "epoch": 0.5595488863312269,
+      "grad_norm": 0.15544173121452332,
+      "learning_rate": 4.440458352334832e-06,
+      "loss": 0.9253,
+      "step": 77300
+    },
+    {
+      "epoch": 0.5596212729918131,
+      "grad_norm": 0.15453557670116425,
+      "learning_rate": 4.440385965674246e-06,
+      "loss": 0.9185,
+      "step": 77310
+    },
+    {
+      "epoch": 0.5596936596523993,
+      "grad_norm": 0.16094282269477844,
+      "learning_rate": 4.44031357901366e-06,
+      "loss": 0.9194,
+      "step": 77320
+    },
+    {
+      "epoch": 0.5597660463129854,
+      "grad_norm": 0.15466651320457458,
+      "learning_rate": 4.440241192353074e-06,
+      "loss": 0.9163,
+      "step": 77330
+    },
+    {
+      "epoch": 0.5598384329735716,
+      "grad_norm": 0.1567269265651703,
+      "learning_rate": 4.440168805692487e-06,
+      "loss": 0.9179,
+      "step": 77340
+    },
+    {
+      "epoch": 0.5599108196341578,
+      "grad_norm": 0.2014339566230774,
+      "learning_rate": 4.440096419031901e-06,
+      "loss": 0.9073,
+      "step": 77350
+    },
+    {
+      "epoch": 0.559983206294744,
+      "grad_norm": 0.16425397992134094,
+      "learning_rate": 4.440024032371315e-06,
+      "loss": 0.8897,
+      "step": 77360
+    },
+    {
+      "epoch": 0.5600555929553301,
+      "grad_norm": 0.16576462984085083,
+      "learning_rate": 4.439951645710729e-06,
+      "loss": 0.9062,
+      "step": 77370
+    },
+    {
+      "epoch": 0.5601279796159164,
+      "grad_norm": 0.15141081809997559,
+      "learning_rate": 4.4398792590501425e-06,
+      "loss": 0.9214,
+      "step": 77380
+    },
+    {
+      "epoch": 0.5602003662765026,
+      "grad_norm": 0.16057878732681274,
+      "learning_rate": 4.439806872389556e-06,
+      "loss": 0.9085,
+      "step": 77390
+    },
+    {
+      "epoch": 0.5602727529370888,
+      "grad_norm": 0.1669279783964157,
+      "learning_rate": 4.439734485728971e-06,
+      "loss": 0.9171,
+      "step": 77400
+    },
+    {
+      "epoch": 0.560345139597675,
+      "grad_norm": 0.17052066326141357,
+      "learning_rate": 4.439662099068384e-06,
+      "loss": 0.9034,
+      "step": 77410
+    },
+    {
+      "epoch": 0.5604175262582611,
+      "grad_norm": 0.16968189179897308,
+      "learning_rate": 4.439589712407798e-06,
+      "loss": 0.9182,
+      "step": 77420
+    },
+    {
+      "epoch": 0.5604899129188473,
+      "grad_norm": 0.1690019816160202,
+      "learning_rate": 4.4395173257472115e-06,
+      "loss": 0.9208,
+      "step": 77430
+    },
+    {
+      "epoch": 0.5605622995794335,
+      "grad_norm": 0.22552861273288727,
+      "learning_rate": 4.439444939086626e-06,
+      "loss": 0.9157,
+      "step": 77440
+    },
+    {
+      "epoch": 0.5606346862400197,
+      "grad_norm": 0.16138851642608643,
+      "learning_rate": 4.4393725524260395e-06,
+      "loss": 0.9054,
+      "step": 77450
+    },
+    {
+      "epoch": 0.5607070729006058,
+      "grad_norm": 0.15333735942840576,
+      "learning_rate": 4.439300165765453e-06,
+      "loss": 0.9084,
+      "step": 77460
+    },
+    {
+      "epoch": 0.560779459561192,
+      "grad_norm": 0.26510128378868103,
+      "learning_rate": 4.439227779104867e-06,
+      "loss": 0.9149,
+      "step": 77470
+    },
+    {
+      "epoch": 0.5608518462217783,
+      "grad_norm": 0.14910390973091125,
+      "learning_rate": 4.439155392444281e-06,
+      "loss": 0.9052,
+      "step": 77480
+    },
+    {
+      "epoch": 0.5609242328823645,
+      "grad_norm": 0.15675272047519684,
+      "learning_rate": 4.439083005783695e-06,
+      "loss": 0.9103,
+      "step": 77490
+    },
+    {
+      "epoch": 0.5609966195429507,
+      "grad_norm": 0.16790616512298584,
+      "learning_rate": 4.4390106191231085e-06,
+      "loss": 0.8932,
+      "step": 77500
+    },
+    {
+      "epoch": 0.5610690062035368,
+      "grad_norm": 0.15522590279579163,
+      "learning_rate": 4.438938232462522e-06,
+      "loss": 0.9203,
+      "step": 77510
+    },
+    {
+      "epoch": 0.561141392864123,
+      "grad_norm": 0.1486007571220398,
+      "learning_rate": 4.4388658458019365e-06,
+      "loss": 0.893,
+      "step": 77520
+    },
+    {
+      "epoch": 0.5612137795247092,
+      "grad_norm": 0.16432353854179382,
+      "learning_rate": 4.43879345914135e-06,
+      "loss": 0.9105,
+      "step": 77530
+    },
+    {
+      "epoch": 0.5612861661852954,
+      "grad_norm": 0.16237597167491913,
+      "learning_rate": 4.438721072480763e-06,
+      "loss": 0.9097,
+      "step": 77540
+    },
+    {
+      "epoch": 0.5613585528458815,
+      "grad_norm": 0.15193864703178406,
+      "learning_rate": 4.438648685820177e-06,
+      "loss": 0.9242,
+      "step": 77550
+    },
+    {
+      "epoch": 0.5614309395064677,
+      "grad_norm": 0.18787966668605804,
+      "learning_rate": 4.438576299159591e-06,
+      "loss": 0.9021,
+      "step": 77560
+    },
+    {
+      "epoch": 0.5615033261670539,
+      "grad_norm": 0.17147335410118103,
+      "learning_rate": 4.438503912499005e-06,
+      "loss": 0.9167,
+      "step": 77570
+    },
+    {
+      "epoch": 0.5615757128276401,
+      "grad_norm": 0.16556903719902039,
+      "learning_rate": 4.438431525838418e-06,
+      "loss": 0.9053,
+      "step": 77580
+    },
+    {
+      "epoch": 0.5616480994882264,
+      "grad_norm": 0.17590190470218658,
+      "learning_rate": 4.438359139177833e-06,
+      "loss": 0.9204,
+      "step": 77590
+    },
+    {
+      "epoch": 0.5617204861488125,
+      "grad_norm": 0.173880934715271,
+      "learning_rate": 4.438286752517246e-06,
+      "loss": 0.9062,
+      "step": 77600
+    },
+    {
+      "epoch": 0.5617928728093987,
+      "grad_norm": 0.16164129972457886,
+      "learning_rate": 4.43821436585666e-06,
+      "loss": 0.9107,
+      "step": 77610
+    },
+    {
+      "epoch": 0.5618652594699849,
+      "grad_norm": 0.1576722264289856,
+      "learning_rate": 4.4381419791960736e-06,
+      "loss": 0.9131,
+      "step": 77620
+    },
+    {
+      "epoch": 0.5619376461305711,
+      "grad_norm": 0.17891205847263336,
+      "learning_rate": 4.438069592535488e-06,
+      "loss": 0.9132,
+      "step": 77630
+    },
+    {
+      "epoch": 0.5620100327911572,
+      "grad_norm": 0.149394229054451,
+      "learning_rate": 4.437997205874902e-06,
+      "loss": 0.9029,
+      "step": 77640
+    },
+    {
+      "epoch": 0.5620824194517434,
+      "grad_norm": 0.1493150144815445,
+      "learning_rate": 4.437924819214315e-06,
+      "loss": 0.9246,
+      "step": 77650
+    },
+    {
+      "epoch": 0.5621548061123296,
+      "grad_norm": 0.1570667326450348,
+      "learning_rate": 4.437852432553729e-06,
+      "loss": 0.9175,
+      "step": 77660
+    },
+    {
+      "epoch": 0.5622271927729158,
+      "grad_norm": 0.1554926335811615,
+      "learning_rate": 4.437780045893143e-06,
+      "loss": 0.907,
+      "step": 77670
+    },
+    {
+      "epoch": 0.562299579433502,
+      "grad_norm": 0.18290413916110992,
+      "learning_rate": 4.437707659232557e-06,
+      "loss": 0.9032,
+      "step": 77680
+    },
+    {
+      "epoch": 0.5623719660940881,
+      "grad_norm": 0.1693306416273117,
+      "learning_rate": 4.4376352725719706e-06,
+      "loss": 0.9073,
+      "step": 77690
+    },
+    {
+      "epoch": 0.5624443527546744,
+      "grad_norm": 0.16810953617095947,
+      "learning_rate": 4.437562885911384e-06,
+      "loss": 0.9203,
+      "step": 77700
+    },
+    {
+      "epoch": 0.5625167394152606,
+      "grad_norm": 0.1598738580942154,
+      "learning_rate": 4.437490499250799e-06,
+      "loss": 0.919,
+      "step": 77710
+    },
+    {
+      "epoch": 0.5625891260758468,
+      "grad_norm": 0.17527292668819427,
+      "learning_rate": 4.437418112590212e-06,
+      "loss": 0.9263,
+      "step": 77720
+    },
+    {
+      "epoch": 0.562661512736433,
+      "grad_norm": 0.14632010459899902,
+      "learning_rate": 4.437345725929626e-06,
+      "loss": 0.9096,
+      "step": 77730
+    },
+    {
+      "epoch": 0.5627338993970191,
+      "grad_norm": 0.16045227646827698,
+      "learning_rate": 4.4372733392690395e-06,
+      "loss": 0.9228,
+      "step": 77740
+    },
+    {
+      "epoch": 0.5628062860576053,
+      "grad_norm": 0.1670956164598465,
+      "learning_rate": 4.437200952608454e-06,
+      "loss": 0.9122,
+      "step": 77750
+    },
+    {
+      "epoch": 0.5628786727181915,
+      "grad_norm": 0.15484373271465302,
+      "learning_rate": 4.4371285659478676e-06,
+      "loss": 0.9101,
+      "step": 77760
+    },
+    {
+      "epoch": 0.5629510593787777,
+      "grad_norm": 0.1494317650794983,
+      "learning_rate": 4.437056179287281e-06,
+      "loss": 0.9119,
+      "step": 77770
+    },
+    {
+      "epoch": 0.5630234460393638,
+      "grad_norm": 0.15059392154216766,
+      "learning_rate": 4.436983792626695e-06,
+      "loss": 0.9092,
+      "step": 77780
+    },
+    {
+      "epoch": 0.56309583269995,
+      "grad_norm": 0.19551579654216766,
+      "learning_rate": 4.436911405966109e-06,
+      "loss": 0.9181,
+      "step": 77790
+    },
+    {
+      "epoch": 0.5631682193605363,
+      "grad_norm": 0.14669376611709595,
+      "learning_rate": 4.436839019305523e-06,
+      "loss": 0.9141,
+      "step": 77800
+    },
+    {
+      "epoch": 0.5632406060211225,
+      "grad_norm": 0.16056394577026367,
+      "learning_rate": 4.4367666326449365e-06,
+      "loss": 0.9127,
+      "step": 77810
+    },
+    {
+      "epoch": 0.5633129926817086,
+      "grad_norm": 0.15070192515850067,
+      "learning_rate": 4.43669424598435e-06,
+      "loss": 0.8972,
+      "step": 77820
+    },
+    {
+      "epoch": 0.5633853793422948,
+      "grad_norm": 0.16001105308532715,
+      "learning_rate": 4.436621859323765e-06,
+      "loss": 0.9019,
+      "step": 77830
+    },
+    {
+      "epoch": 0.563457766002881,
+      "grad_norm": 0.14969342947006226,
+      "learning_rate": 4.436549472663178e-06,
+      "loss": 0.9138,
+      "step": 77840
+    },
+    {
+      "epoch": 0.5635301526634672,
+      "grad_norm": 0.15888501703739166,
+      "learning_rate": 4.436477086002592e-06,
+      "loss": 0.9168,
+      "step": 77850
+    },
+    {
+      "epoch": 0.5636025393240534,
+      "grad_norm": 0.16971467435359955,
+      "learning_rate": 4.4364046993420054e-06,
+      "loss": 0.9011,
+      "step": 77860
+    },
+    {
+      "epoch": 0.5636749259846395,
+      "grad_norm": 0.18575704097747803,
+      "learning_rate": 4.43633231268142e-06,
+      "loss": 0.8922,
+      "step": 77870
+    },
+    {
+      "epoch": 0.5637473126452257,
+      "grad_norm": 0.161865696310997,
+      "learning_rate": 4.4362599260208335e-06,
+      "loss": 0.9045,
+      "step": 77880
+    },
+    {
+      "epoch": 0.5638196993058119,
+      "grad_norm": 0.22121845185756683,
+      "learning_rate": 4.436187539360247e-06,
+      "loss": 0.9184,
+      "step": 77890
+    },
+    {
+      "epoch": 0.5638920859663981,
+      "grad_norm": 0.15823763608932495,
+      "learning_rate": 4.436115152699661e-06,
+      "loss": 0.898,
+      "step": 77900
+    },
+    {
+      "epoch": 0.5639644726269843,
+      "grad_norm": 0.1676093190908432,
+      "learning_rate": 4.436042766039075e-06,
+      "loss": 0.9173,
+      "step": 77910
+    },
+    {
+      "epoch": 0.5640368592875705,
+      "grad_norm": 0.15236294269561768,
+      "learning_rate": 4.435970379378489e-06,
+      "loss": 0.9052,
+      "step": 77920
+    },
+    {
+      "epoch": 0.5641092459481567,
+      "grad_norm": 0.16066612303256989,
+      "learning_rate": 4.4358979927179024e-06,
+      "loss": 0.9102,
+      "step": 77930
+    },
+    {
+      "epoch": 0.5641816326087429,
+      "grad_norm": 0.162504181265831,
+      "learning_rate": 4.435825606057316e-06,
+      "loss": 0.8935,
+      "step": 77940
+    },
+    {
+      "epoch": 0.564254019269329,
+      "grad_norm": 0.1564251035451889,
+      "learning_rate": 4.4357532193967305e-06,
+      "loss": 0.9219,
+      "step": 77950
+    },
+    {
+      "epoch": 0.5643264059299152,
+      "grad_norm": 0.14474442601203918,
+      "learning_rate": 4.435680832736144e-06,
+      "loss": 0.9118,
+      "step": 77960
+    },
+    {
+      "epoch": 0.5643987925905014,
+      "grad_norm": 0.1615808606147766,
+      "learning_rate": 4.435608446075558e-06,
+      "loss": 0.9025,
+      "step": 77970
+    },
+    {
+      "epoch": 0.5644711792510876,
+      "grad_norm": 0.16067107021808624,
+      "learning_rate": 4.435536059414971e-06,
+      "loss": 0.9064,
+      "step": 77980
+    },
+    {
+      "epoch": 0.5645435659116738,
+      "grad_norm": 0.1663515865802765,
+      "learning_rate": 4.435463672754385e-06,
+      "loss": 0.8929,
+      "step": 77990
+    },
+    {
+      "epoch": 0.5646159525722599,
+      "grad_norm": 0.15532717108726501,
+      "learning_rate": 4.4353912860937994e-06,
+      "loss": 0.9231,
+      "step": 78000
+    },
+    {
+      "epoch": 0.5646883392328462,
+      "grad_norm": 0.16401223838329315,
+      "learning_rate": 4.435318899433213e-06,
+      "loss": 0.9048,
+      "step": 78010
+    },
+    {
+      "epoch": 0.5647607258934324,
+      "grad_norm": 0.15295349061489105,
+      "learning_rate": 4.435246512772627e-06,
+      "loss": 0.9135,
+      "step": 78020
+    },
+    {
+      "epoch": 0.5648331125540186,
+      "grad_norm": 0.16399818658828735,
+      "learning_rate": 4.43517412611204e-06,
+      "loss": 0.9169,
+      "step": 78030
+    },
+    {
+      "epoch": 0.5649054992146048,
+      "grad_norm": 0.15173108875751495,
+      "learning_rate": 4.435101739451455e-06,
+      "loss": 0.9184,
+      "step": 78040
+    },
+    {
+      "epoch": 0.5649778858751909,
+      "grad_norm": 0.15851818025112152,
+      "learning_rate": 4.435029352790868e-06,
+      "loss": 0.907,
+      "step": 78050
+    },
+    {
+      "epoch": 0.5650502725357771,
+      "grad_norm": 0.14833968877792358,
+      "learning_rate": 4.434956966130282e-06,
+      "loss": 0.9057,
+      "step": 78060
+    },
+    {
+      "epoch": 0.5651226591963633,
+      "grad_norm": 0.15308476984500885,
+      "learning_rate": 4.434884579469696e-06,
+      "loss": 0.9124,
+      "step": 78070
+    },
+    {
+      "epoch": 0.5651950458569495,
+      "grad_norm": 0.1614348590373993,
+      "learning_rate": 4.434812192809109e-06,
+      "loss": 0.9196,
+      "step": 78080
+    },
+    {
+      "epoch": 0.5652674325175356,
+      "grad_norm": 0.15268519520759583,
+      "learning_rate": 4.434739806148523e-06,
+      "loss": 0.9167,
+      "step": 78090
+    },
+    {
+      "epoch": 0.5653398191781218,
+      "grad_norm": 0.20000995695590973,
+      "learning_rate": 4.434667419487937e-06,
+      "loss": 0.8982,
+      "step": 78100
+    },
+    {
+      "epoch": 0.565412205838708,
+      "grad_norm": 0.19799144566059113,
+      "learning_rate": 4.434595032827351e-06,
+      "loss": 0.9041,
+      "step": 78110
+    },
+    {
+      "epoch": 0.5654845924992943,
+      "grad_norm": 0.16405373811721802,
+      "learning_rate": 4.4345226461667645e-06,
+      "loss": 0.9125,
+      "step": 78120
+    },
+    {
+      "epoch": 0.5655569791598805,
+      "grad_norm": 0.14727067947387695,
+      "learning_rate": 4.434450259506178e-06,
+      "loss": 0.9143,
+      "step": 78130
+    },
+    {
+      "epoch": 0.5656293658204666,
+      "grad_norm": 0.15156599879264832,
+      "learning_rate": 4.434377872845592e-06,
+      "loss": 0.9086,
+      "step": 78140
+    },
+    {
+      "epoch": 0.5657017524810528,
+      "grad_norm": 0.16756990551948547,
+      "learning_rate": 4.434305486185006e-06,
+      "loss": 0.919,
+      "step": 78150
+    },
+    {
+      "epoch": 0.565774139141639,
+      "grad_norm": 0.159501850605011,
+      "learning_rate": 4.43423309952442e-06,
+      "loss": 0.8931,
+      "step": 78160
+    },
+    {
+      "epoch": 0.5658465258022252,
+      "grad_norm": 0.1752566546201706,
+      "learning_rate": 4.4341607128638335e-06,
+      "loss": 0.9072,
+      "step": 78170
+    },
+    {
+      "epoch": 0.5659189124628113,
+      "grad_norm": 0.15048715472221375,
+      "learning_rate": 4.434088326203247e-06,
+      "loss": 0.8956,
+      "step": 78180
+    },
+    {
+      "epoch": 0.5659912991233975,
+      "grad_norm": 0.18843425810337067,
+      "learning_rate": 4.4340159395426615e-06,
+      "loss": 0.9209,
+      "step": 78190
+    },
+    {
+      "epoch": 0.5660636857839837,
+      "grad_norm": 0.23466771841049194,
+      "learning_rate": 4.433943552882075e-06,
+      "loss": 0.9072,
+      "step": 78200
+    },
+    {
+      "epoch": 0.5661360724445699,
+      "grad_norm": 0.1651633083820343,
+      "learning_rate": 4.433871166221489e-06,
+      "loss": 0.9061,
+      "step": 78210
+    },
+    {
+      "epoch": 0.566208459105156,
+      "grad_norm": 0.18153083324432373,
+      "learning_rate": 4.433798779560902e-06,
+      "loss": 0.9043,
+      "step": 78220
+    },
+    {
+      "epoch": 0.5662808457657423,
+      "grad_norm": 0.14733073115348816,
+      "learning_rate": 4.433726392900317e-06,
+      "loss": 0.9011,
+      "step": 78230
+    },
+    {
+      "epoch": 0.5663532324263285,
+      "grad_norm": 0.16242343187332153,
+      "learning_rate": 4.4336540062397305e-06,
+      "loss": 0.9121,
+      "step": 78240
+    },
+    {
+      "epoch": 0.5664256190869147,
+      "grad_norm": 0.2215704470872879,
+      "learning_rate": 4.433581619579144e-06,
+      "loss": 0.9137,
+      "step": 78250
+    },
+    {
+      "epoch": 0.5664980057475009,
+      "grad_norm": 0.15027424693107605,
+      "learning_rate": 4.433509232918558e-06,
+      "loss": 0.911,
+      "step": 78260
+    },
+    {
+      "epoch": 0.566570392408087,
+      "grad_norm": 0.15450255572795868,
+      "learning_rate": 4.433436846257972e-06,
+      "loss": 0.9016,
+      "step": 78270
+    },
+    {
+      "epoch": 0.5666427790686732,
+      "grad_norm": 0.15546143054962158,
+      "learning_rate": 4.433364459597386e-06,
+      "loss": 0.9196,
+      "step": 78280
+    },
+    {
+      "epoch": 0.5667151657292594,
+      "grad_norm": 0.2212473452091217,
+      "learning_rate": 4.433292072936799e-06,
+      "loss": 0.9095,
+      "step": 78290
+    },
+    {
+      "epoch": 0.5667875523898456,
+      "grad_norm": 0.21143324673175812,
+      "learning_rate": 4.433219686276213e-06,
+      "loss": 0.9151,
+      "step": 78300
+    },
+    {
+      "epoch": 0.5668599390504317,
+      "grad_norm": 0.15912073850631714,
+      "learning_rate": 4.4331472996156275e-06,
+      "loss": 0.9131,
+      "step": 78310
+    },
+    {
+      "epoch": 0.5669323257110179,
+      "grad_norm": 0.16560253500938416,
+      "learning_rate": 4.433074912955041e-06,
+      "loss": 0.9257,
+      "step": 78320
+    },
+    {
+      "epoch": 0.5670047123716042,
+      "grad_norm": 0.15693403780460358,
+      "learning_rate": 4.433002526294455e-06,
+      "loss": 0.9085,
+      "step": 78330
+    },
+    {
+      "epoch": 0.5670770990321904,
+      "grad_norm": 0.15633919835090637,
+      "learning_rate": 4.432930139633868e-06,
+      "loss": 0.9304,
+      "step": 78340
+    },
+    {
+      "epoch": 0.5671494856927766,
+      "grad_norm": 0.2166070193052292,
+      "learning_rate": 4.432857752973283e-06,
+      "loss": 0.9092,
+      "step": 78350
+    },
+    {
+      "epoch": 0.5672218723533627,
+      "grad_norm": 0.154709130525589,
+      "learning_rate": 4.432785366312696e-06,
+      "loss": 0.9122,
+      "step": 78360
+    },
+    {
+      "epoch": 0.5672942590139489,
+      "grad_norm": 0.1582053154706955,
+      "learning_rate": 4.43271297965211e-06,
+      "loss": 0.9011,
+      "step": 78370
+    },
+    {
+      "epoch": 0.5673666456745351,
+      "grad_norm": 0.16059152781963348,
+      "learning_rate": 4.432640592991524e-06,
+      "loss": 0.9082,
+      "step": 78380
+    },
+    {
+      "epoch": 0.5674390323351213,
+      "grad_norm": 0.1610398292541504,
+      "learning_rate": 4.432568206330938e-06,
+      "loss": 0.9072,
+      "step": 78390
+    },
+    {
+      "epoch": 0.5675114189957075,
+      "grad_norm": 0.1508665531873703,
+      "learning_rate": 4.432495819670352e-06,
+      "loss": 0.9011,
+      "step": 78400
+    },
+    {
+      "epoch": 0.5675838056562936,
+      "grad_norm": 0.16580963134765625,
+      "learning_rate": 4.432423433009765e-06,
+      "loss": 0.9156,
+      "step": 78410
+    },
+    {
+      "epoch": 0.5676561923168798,
+      "grad_norm": 0.17013192176818848,
+      "learning_rate": 4.432351046349179e-06,
+      "loss": 0.9119,
+      "step": 78420
+    },
+    {
+      "epoch": 0.567728578977466,
+      "grad_norm": 0.15810072422027588,
+      "learning_rate": 4.432278659688593e-06,
+      "loss": 0.9018,
+      "step": 78430
+    },
+    {
+      "epoch": 0.5678009656380523,
+      "grad_norm": 0.1526464819908142,
+      "learning_rate": 4.432206273028007e-06,
+      "loss": 0.9058,
+      "step": 78440
+    },
+    {
+      "epoch": 0.5678733522986384,
+      "grad_norm": 0.15462712943553925,
+      "learning_rate": 4.432133886367421e-06,
+      "loss": 0.9117,
+      "step": 78450
+    },
+    {
+      "epoch": 0.5679457389592246,
+      "grad_norm": 0.1546926200389862,
+      "learning_rate": 4.432061499706834e-06,
+      "loss": 0.9001,
+      "step": 78460
+    },
+    {
+      "epoch": 0.5680181256198108,
+      "grad_norm": 0.152653768658638,
+      "learning_rate": 4.431989113046249e-06,
+      "loss": 0.8992,
+      "step": 78470
+    },
+    {
+      "epoch": 0.568090512280397,
+      "grad_norm": 0.20859988033771515,
+      "learning_rate": 4.431916726385662e-06,
+      "loss": 0.9182,
+      "step": 78480
+    },
+    {
+      "epoch": 0.5681628989409832,
+      "grad_norm": 0.16723498702049255,
+      "learning_rate": 4.431844339725076e-06,
+      "loss": 0.898,
+      "step": 78490
+    },
+    {
+      "epoch": 0.5682352856015693,
+      "grad_norm": 0.16271011531352997,
+      "learning_rate": 4.4317719530644896e-06,
+      "loss": 0.9084,
+      "step": 78500
+    },
+    {
+      "epoch": 0.5683076722621555,
+      "grad_norm": 0.15765948593616486,
+      "learning_rate": 4.431699566403904e-06,
+      "loss": 0.9245,
+      "step": 78510
+    },
+    {
+      "epoch": 0.5683800589227417,
+      "grad_norm": 0.15575428307056427,
+      "learning_rate": 4.431627179743318e-06,
+      "loss": 0.9282,
+      "step": 78520
+    },
+    {
+      "epoch": 0.5684524455833279,
+      "grad_norm": 0.17210975289344788,
+      "learning_rate": 4.431554793082731e-06,
+      "loss": 0.9234,
+      "step": 78530
+    },
+    {
+      "epoch": 0.5685248322439141,
+      "grad_norm": 0.15548361837863922,
+      "learning_rate": 4.431482406422145e-06,
+      "loss": 0.9141,
+      "step": 78540
+    },
+    {
+      "epoch": 0.5685972189045003,
+      "grad_norm": 0.15747930109500885,
+      "learning_rate": 4.431410019761559e-06,
+      "loss": 0.9105,
+      "step": 78550
+    },
+    {
+      "epoch": 0.5686696055650865,
+      "grad_norm": 0.19277207553386688,
+      "learning_rate": 4.431337633100973e-06,
+      "loss": 0.8969,
+      "step": 78560
+    },
+    {
+      "epoch": 0.5687419922256727,
+      "grad_norm": 0.15888191759586334,
+      "learning_rate": 4.431265246440387e-06,
+      "loss": 0.9133,
+      "step": 78570
+    },
+    {
+      "epoch": 0.5688143788862589,
+      "grad_norm": 0.18703103065490723,
+      "learning_rate": 4.4311928597798e-06,
+      "loss": 0.9204,
+      "step": 78580
+    },
+    {
+      "epoch": 0.568886765546845,
+      "grad_norm": 0.15291064977645874,
+      "learning_rate": 4.431120473119214e-06,
+      "loss": 0.9055,
+      "step": 78590
+    },
+    {
+      "epoch": 0.5689591522074312,
+      "grad_norm": 0.15586498379707336,
+      "learning_rate": 4.4310480864586274e-06,
+      "loss": 0.9007,
+      "step": 78600
+    },
+    {
+      "epoch": 0.5690315388680174,
+      "grad_norm": 0.17706800997257233,
+      "learning_rate": 4.430975699798041e-06,
+      "loss": 0.9223,
+      "step": 78610
+    },
+    {
+      "epoch": 0.5691039255286036,
+      "grad_norm": 0.1869611293077469,
+      "learning_rate": 4.4309033131374555e-06,
+      "loss": 0.9168,
+      "step": 78620
+    },
+    {
+      "epoch": 0.5691763121891897,
+      "grad_norm": 0.20068153738975525,
+      "learning_rate": 4.430830926476869e-06,
+      "loss": 0.9092,
+      "step": 78630
+    },
+    {
+      "epoch": 0.5692486988497759,
+      "grad_norm": 0.15373751521110535,
+      "learning_rate": 4.430758539816283e-06,
+      "loss": 0.9068,
+      "step": 78640
+    },
+    {
+      "epoch": 0.5693210855103622,
+      "grad_norm": 0.161079540848732,
+      "learning_rate": 4.430686153155696e-06,
+      "loss": 0.9072,
+      "step": 78650
+    },
+    {
+      "epoch": 0.5693934721709484,
+      "grad_norm": 0.17405648529529572,
+      "learning_rate": 4.430613766495111e-06,
+      "loss": 0.9152,
+      "step": 78660
+    },
+    {
+      "epoch": 0.5694658588315346,
+      "grad_norm": 0.20295019447803497,
+      "learning_rate": 4.4305413798345244e-06,
+      "loss": 0.9121,
+      "step": 78670
+    },
+    {
+      "epoch": 0.5695382454921207,
+      "grad_norm": 0.16853387653827667,
+      "learning_rate": 4.430468993173938e-06,
+      "loss": 0.9134,
+      "step": 78680
+    },
+    {
+      "epoch": 0.5696106321527069,
+      "grad_norm": 0.21727709472179413,
+      "learning_rate": 4.430396606513352e-06,
+      "loss": 0.9049,
+      "step": 78690
+    },
+    {
+      "epoch": 0.5696830188132931,
+      "grad_norm": 0.15580429136753082,
+      "learning_rate": 4.430324219852766e-06,
+      "loss": 0.9035,
+      "step": 78700
+    },
+    {
+      "epoch": 0.5697554054738793,
+      "grad_norm": 0.1704913228750229,
+      "learning_rate": 4.43025183319218e-06,
+      "loss": 0.9269,
+      "step": 78710
+    },
+    {
+      "epoch": 0.5698277921344654,
+      "grad_norm": 0.1460624784231186,
+      "learning_rate": 4.430179446531593e-06,
+      "loss": 0.9113,
+      "step": 78720
+    },
+    {
+      "epoch": 0.5699001787950516,
+      "grad_norm": 0.16129523515701294,
+      "learning_rate": 4.430107059871007e-06,
+      "loss": 0.9033,
+      "step": 78730
+    },
+    {
+      "epoch": 0.5699725654556378,
+      "grad_norm": 0.15936952829360962,
+      "learning_rate": 4.4300346732104214e-06,
+      "loss": 0.9164,
+      "step": 78740
+    },
+    {
+      "epoch": 0.570044952116224,
+      "grad_norm": 0.15949364006519318,
+      "learning_rate": 4.429962286549835e-06,
+      "loss": 0.9013,
+      "step": 78750
+    },
+    {
+      "epoch": 0.5701173387768103,
+      "grad_norm": 0.16319353878498077,
+      "learning_rate": 4.429889899889249e-06,
+      "loss": 0.9086,
+      "step": 78760
+    },
+    {
+      "epoch": 0.5701897254373964,
+      "grad_norm": 0.16823211312294006,
+      "learning_rate": 4.429817513228662e-06,
+      "loss": 0.9163,
+      "step": 78770
+    },
+    {
+      "epoch": 0.5702621120979826,
+      "grad_norm": 0.16069024801254272,
+      "learning_rate": 4.429745126568076e-06,
+      "loss": 0.9148,
+      "step": 78780
+    },
+    {
+      "epoch": 0.5703344987585688,
+      "grad_norm": 0.15689004957675934,
+      "learning_rate": 4.42967273990749e-06,
+      "loss": 0.9087,
+      "step": 78790
+    },
+    {
+      "epoch": 0.570406885419155,
+      "grad_norm": 0.19310441613197327,
+      "learning_rate": 4.429600353246904e-06,
+      "loss": 0.9031,
+      "step": 78800
+    },
+    {
+      "epoch": 0.5704792720797411,
+      "grad_norm": 0.15190792083740234,
+      "learning_rate": 4.429527966586318e-06,
+      "loss": 0.908,
+      "step": 78810
+    },
+    {
+      "epoch": 0.5705516587403273,
+      "grad_norm": 0.1643807291984558,
+      "learning_rate": 4.429455579925731e-06,
+      "loss": 0.9052,
+      "step": 78820
+    },
+    {
+      "epoch": 0.5706240454009135,
+      "grad_norm": 0.16099394857883453,
+      "learning_rate": 4.429383193265146e-06,
+      "loss": 0.8997,
+      "step": 78830
+    },
+    {
+      "epoch": 0.5706964320614997,
+      "grad_norm": 0.1543867588043213,
+      "learning_rate": 4.429310806604559e-06,
+      "loss": 0.9139,
+      "step": 78840
+    },
+    {
+      "epoch": 0.5707688187220858,
+      "grad_norm": 0.1512497216463089,
+      "learning_rate": 4.429238419943973e-06,
+      "loss": 0.8972,
+      "step": 78850
+    },
+    {
+      "epoch": 0.5708412053826721,
+      "grad_norm": 0.1591501533985138,
+      "learning_rate": 4.4291660332833865e-06,
+      "loss": 0.9097,
+      "step": 78860
+    },
+    {
+      "epoch": 0.5709135920432583,
+      "grad_norm": 0.1647246778011322,
+      "learning_rate": 4.429093646622801e-06,
+      "loss": 0.903,
+      "step": 78870
+    },
+    {
+      "epoch": 0.5709859787038445,
+      "grad_norm": 0.1652226597070694,
+      "learning_rate": 4.429021259962215e-06,
+      "loss": 0.9128,
+      "step": 78880
+    },
+    {
+      "epoch": 0.5710583653644307,
+      "grad_norm": 0.16023515164852142,
+      "learning_rate": 4.428948873301628e-06,
+      "loss": 0.9068,
+      "step": 78890
+    },
+    {
+      "epoch": 0.5711307520250168,
+      "grad_norm": 0.16677772998809814,
+      "learning_rate": 4.428876486641042e-06,
+      "loss": 0.9067,
+      "step": 78900
+    },
+    {
+      "epoch": 0.571203138685603,
+      "grad_norm": 0.15754565596580505,
+      "learning_rate": 4.428804099980456e-06,
+      "loss": 0.8981,
+      "step": 78910
+    },
+    {
+      "epoch": 0.5712755253461892,
+      "grad_norm": 0.1568695604801178,
+      "learning_rate": 4.42873171331987e-06,
+      "loss": 0.9127,
+      "step": 78920
+    },
+    {
+      "epoch": 0.5713479120067754,
+      "grad_norm": 0.14470389485359192,
+      "learning_rate": 4.4286593266592835e-06,
+      "loss": 0.908,
+      "step": 78930
+    },
+    {
+      "epoch": 0.5714202986673615,
+      "grad_norm": 0.17773394286632538,
+      "learning_rate": 4.428586939998697e-06,
+      "loss": 0.9145,
+      "step": 78940
+    },
+    {
+      "epoch": 0.5714926853279477,
+      "grad_norm": 0.18091265857219696,
+      "learning_rate": 4.428514553338112e-06,
+      "loss": 0.9129,
+      "step": 78950
+    },
+    {
+      "epoch": 0.5715650719885339,
+      "grad_norm": 0.17283041775226593,
+      "learning_rate": 4.428442166677525e-06,
+      "loss": 0.9103,
+      "step": 78960
+    },
+    {
+      "epoch": 0.5716374586491202,
+      "grad_norm": 0.17135843634605408,
+      "learning_rate": 4.428369780016939e-06,
+      "loss": 0.9144,
+      "step": 78970
+    },
+    {
+      "epoch": 0.5717098453097064,
+      "grad_norm": 0.15438847243785858,
+      "learning_rate": 4.4282973933563525e-06,
+      "loss": 0.9039,
+      "step": 78980
+    },
+    {
+      "epoch": 0.5717822319702925,
+      "grad_norm": 0.16643427312374115,
+      "learning_rate": 4.428225006695767e-06,
+      "loss": 0.9042,
+      "step": 78990
+    },
+    {
+      "epoch": 0.5718546186308787,
+      "grad_norm": 0.17289410531520844,
+      "learning_rate": 4.4281526200351805e-06,
+      "loss": 0.909,
+      "step": 79000
+    },
+    {
+      "epoch": 0.5719270052914649,
+      "grad_norm": 0.15444664657115936,
+      "learning_rate": 4.428080233374594e-06,
+      "loss": 0.9129,
+      "step": 79010
+    },
+    {
+      "epoch": 0.5719993919520511,
+      "grad_norm": 0.1560271829366684,
+      "learning_rate": 4.428007846714008e-06,
+      "loss": 0.9051,
+      "step": 79020
+    },
+    {
+      "epoch": 0.5720717786126372,
+      "grad_norm": 0.1660393625497818,
+      "learning_rate": 4.427935460053422e-06,
+      "loss": 0.9067,
+      "step": 79030
+    },
+    {
+      "epoch": 0.5721441652732234,
+      "grad_norm": 0.16919827461242676,
+      "learning_rate": 4.427863073392836e-06,
+      "loss": 0.9075,
+      "step": 79040
+    },
+    {
+      "epoch": 0.5722165519338096,
+      "grad_norm": 0.16163021326065063,
+      "learning_rate": 4.4277906867322495e-06,
+      "loss": 0.9167,
+      "step": 79050
+    },
+    {
+      "epoch": 0.5722889385943958,
+      "grad_norm": 0.15363195538520813,
+      "learning_rate": 4.427718300071663e-06,
+      "loss": 0.9091,
+      "step": 79060
+    },
+    {
+      "epoch": 0.5723613252549821,
+      "grad_norm": 0.22884926199913025,
+      "learning_rate": 4.4276459134110776e-06,
+      "loss": 0.9061,
+      "step": 79070
+    },
+    {
+      "epoch": 0.5724337119155682,
+      "grad_norm": 0.17008480429649353,
+      "learning_rate": 4.427573526750491e-06,
+      "loss": 0.8936,
+      "step": 79080
+    },
+    {
+      "epoch": 0.5725060985761544,
+      "grad_norm": 0.1554049849510193,
+      "learning_rate": 4.427501140089905e-06,
+      "loss": 0.9041,
+      "step": 79090
+    },
+    {
+      "epoch": 0.5725784852367406,
+      "grad_norm": 0.15605086088180542,
+      "learning_rate": 4.427428753429318e-06,
+      "loss": 0.9061,
+      "step": 79100
+    },
+    {
+      "epoch": 0.5726508718973268,
+      "grad_norm": 0.18999440968036652,
+      "learning_rate": 4.427356366768733e-06,
+      "loss": 0.918,
+      "step": 79110
+    },
+    {
+      "epoch": 0.572723258557913,
+      "grad_norm": 0.14956019818782806,
+      "learning_rate": 4.4272839801081465e-06,
+      "loss": 0.9107,
+      "step": 79120
+    },
+    {
+      "epoch": 0.5727956452184991,
+      "grad_norm": 0.16016794741153717,
+      "learning_rate": 4.427211593447559e-06,
+      "loss": 0.9077,
+      "step": 79130
+    },
+    {
+      "epoch": 0.5728680318790853,
+      "grad_norm": 0.16640208661556244,
+      "learning_rate": 4.427139206786974e-06,
+      "loss": 0.9131,
+      "step": 79140
+    },
+    {
+      "epoch": 0.5729404185396715,
+      "grad_norm": 0.14928004145622253,
+      "learning_rate": 4.427066820126387e-06,
+      "loss": 0.9216,
+      "step": 79150
+    },
+    {
+      "epoch": 0.5730128052002577,
+      "grad_norm": 0.17138387262821198,
+      "learning_rate": 4.426994433465801e-06,
+      "loss": 0.912,
+      "step": 79160
+    },
+    {
+      "epoch": 0.5730851918608438,
+      "grad_norm": 0.16592368483543396,
+      "learning_rate": 4.4269220468052146e-06,
+      "loss": 0.9082,
+      "step": 79170
+    },
+    {
+      "epoch": 0.5731575785214301,
+      "grad_norm": 0.15400424599647522,
+      "learning_rate": 4.426849660144629e-06,
+      "loss": 0.8956,
+      "step": 79180
+    },
+    {
+      "epoch": 0.5732299651820163,
+      "grad_norm": 0.16034607589244843,
+      "learning_rate": 4.426777273484043e-06,
+      "loss": 0.9104,
+      "step": 79190
+    },
+    {
+      "epoch": 0.5733023518426025,
+      "grad_norm": 0.16131481528282166,
+      "learning_rate": 4.426704886823456e-06,
+      "loss": 0.9076,
+      "step": 79200
+    },
+    {
+      "epoch": 0.5733747385031887,
+      "grad_norm": 0.15663005411624908,
+      "learning_rate": 4.42663250016287e-06,
+      "loss": 0.9027,
+      "step": 79210
+    },
+    {
+      "epoch": 0.5734471251637748,
+      "grad_norm": 0.15593023598194122,
+      "learning_rate": 4.426560113502284e-06,
+      "loss": 0.8946,
+      "step": 79220
+    },
+    {
+      "epoch": 0.573519511824361,
+      "grad_norm": 0.15859454870224,
+      "learning_rate": 4.426487726841698e-06,
+      "loss": 0.908,
+      "step": 79230
+    },
+    {
+      "epoch": 0.5735918984849472,
+      "grad_norm": 0.2031136453151703,
+      "learning_rate": 4.4264153401811116e-06,
+      "loss": 0.903,
+      "step": 79240
+    },
+    {
+      "epoch": 0.5736642851455334,
+      "grad_norm": 0.1812455952167511,
+      "learning_rate": 4.426342953520525e-06,
+      "loss": 0.9156,
+      "step": 79250
+    },
+    {
+      "epoch": 0.5737366718061195,
+      "grad_norm": 0.16677896678447723,
+      "learning_rate": 4.42627056685994e-06,
+      "loss": 0.9107,
+      "step": 79260
+    },
+    {
+      "epoch": 0.5738090584667057,
+      "grad_norm": 0.19645458459854126,
+      "learning_rate": 4.426198180199353e-06,
+      "loss": 0.9055,
+      "step": 79270
+    },
+    {
+      "epoch": 0.5738814451272919,
+      "grad_norm": 0.15948496758937836,
+      "learning_rate": 4.426125793538767e-06,
+      "loss": 0.9226,
+      "step": 79280
+    },
+    {
+      "epoch": 0.5739538317878782,
+      "grad_norm": 0.152080699801445,
+      "learning_rate": 4.4260534068781805e-06,
+      "loss": 0.9106,
+      "step": 79290
+    },
+    {
+      "epoch": 0.5740262184484644,
+      "grad_norm": 0.16088615357875824,
+      "learning_rate": 4.425981020217595e-06,
+      "loss": 0.9034,
+      "step": 79300
+    },
+    {
+      "epoch": 0.5740986051090505,
+      "grad_norm": 0.1661704033613205,
+      "learning_rate": 4.425908633557009e-06,
+      "loss": 0.9064,
+      "step": 79310
+    },
+    {
+      "epoch": 0.5741709917696367,
+      "grad_norm": 0.175064355134964,
+      "learning_rate": 4.425836246896422e-06,
+      "loss": 0.9014,
+      "step": 79320
+    },
+    {
+      "epoch": 0.5742433784302229,
+      "grad_norm": 0.1588856726884842,
+      "learning_rate": 4.425763860235836e-06,
+      "loss": 0.904,
+      "step": 79330
+    },
+    {
+      "epoch": 0.5743157650908091,
+      "grad_norm": 0.16363567113876343,
+      "learning_rate": 4.42569147357525e-06,
+      "loss": 0.9076,
+      "step": 79340
+    },
+    {
+      "epoch": 0.5743881517513952,
+      "grad_norm": 0.1526729166507721,
+      "learning_rate": 4.425619086914664e-06,
+      "loss": 0.9174,
+      "step": 79350
+    },
+    {
+      "epoch": 0.5744605384119814,
+      "grad_norm": 0.15896102786064148,
+      "learning_rate": 4.4255467002540775e-06,
+      "loss": 0.9168,
+      "step": 79360
+    },
+    {
+      "epoch": 0.5745329250725676,
+      "grad_norm": 0.15912756323814392,
+      "learning_rate": 4.425474313593491e-06,
+      "loss": 0.9176,
+      "step": 79370
+    },
+    {
+      "epoch": 0.5746053117331538,
+      "grad_norm": 0.15687337517738342,
+      "learning_rate": 4.425401926932905e-06,
+      "loss": 0.9177,
+      "step": 79380
+    },
+    {
+      "epoch": 0.57467769839374,
+      "grad_norm": 0.1485118865966797,
+      "learning_rate": 4.425329540272319e-06,
+      "loss": 0.9097,
+      "step": 79390
+    },
+    {
+      "epoch": 0.5747500850543262,
+      "grad_norm": 0.1573365032672882,
+      "learning_rate": 4.425257153611733e-06,
+      "loss": 0.9106,
+      "step": 79400
+    },
+    {
+      "epoch": 0.5748224717149124,
+      "grad_norm": 0.16917355358600616,
+      "learning_rate": 4.4251847669511464e-06,
+      "loss": 0.9,
+      "step": 79410
+    },
+    {
+      "epoch": 0.5748948583754986,
+      "grad_norm": 0.15793545544147491,
+      "learning_rate": 4.42511238029056e-06,
+      "loss": 0.9028,
+      "step": 79420
+    },
+    {
+      "epoch": 0.5749672450360848,
+      "grad_norm": 0.16056816279888153,
+      "learning_rate": 4.4250399936299745e-06,
+      "loss": 0.9103,
+      "step": 79430
+    },
+    {
+      "epoch": 0.5750396316966709,
+      "grad_norm": 0.21023601293563843,
+      "learning_rate": 4.424967606969388e-06,
+      "loss": 0.9131,
+      "step": 79440
+    },
+    {
+      "epoch": 0.5751120183572571,
+      "grad_norm": 0.16455906629562378,
+      "learning_rate": 4.424895220308802e-06,
+      "loss": 0.9136,
+      "step": 79450
+    },
+    {
+      "epoch": 0.5751844050178433,
+      "grad_norm": 0.15624572336673737,
+      "learning_rate": 4.424822833648215e-06,
+      "loss": 0.8955,
+      "step": 79460
+    },
+    {
+      "epoch": 0.5752567916784295,
+      "grad_norm": 0.15977784991264343,
+      "learning_rate": 4.42475044698763e-06,
+      "loss": 0.9127,
+      "step": 79470
+    },
+    {
+      "epoch": 0.5753291783390156,
+      "grad_norm": 0.15706151723861694,
+      "learning_rate": 4.4246780603270434e-06,
+      "loss": 0.9218,
+      "step": 79480
+    },
+    {
+      "epoch": 0.5754015649996018,
+      "grad_norm": 0.1580117791891098,
+      "learning_rate": 4.424605673666457e-06,
+      "loss": 0.9063,
+      "step": 79490
+    },
+    {
+      "epoch": 0.5754739516601881,
+      "grad_norm": 0.1586533933877945,
+      "learning_rate": 4.424533287005871e-06,
+      "loss": 0.9001,
+      "step": 79500
+    },
+    {
+      "epoch": 0.5755463383207743,
+      "grad_norm": 0.15484578907489777,
+      "learning_rate": 4.424460900345285e-06,
+      "loss": 0.9188,
+      "step": 79510
+    },
+    {
+      "epoch": 0.5756187249813605,
+      "grad_norm": 0.16581545770168304,
+      "learning_rate": 4.424388513684699e-06,
+      "loss": 0.9029,
+      "step": 79520
+    },
+    {
+      "epoch": 0.5756911116419466,
+      "grad_norm": 0.2024676501750946,
+      "learning_rate": 4.424316127024112e-06,
+      "loss": 0.9066,
+      "step": 79530
+    },
+    {
+      "epoch": 0.5757634983025328,
+      "grad_norm": 0.17313936352729797,
+      "learning_rate": 4.424243740363526e-06,
+      "loss": 0.9075,
+      "step": 79540
+    },
+    {
+      "epoch": 0.575835884963119,
+      "grad_norm": 0.15734641253948212,
+      "learning_rate": 4.4241713537029404e-06,
+      "loss": 0.891,
+      "step": 79550
+    },
+    {
+      "epoch": 0.5759082716237052,
+      "grad_norm": 0.16273874044418335,
+      "learning_rate": 4.424098967042354e-06,
+      "loss": 0.8987,
+      "step": 79560
+    },
+    {
+      "epoch": 0.5759806582842913,
+      "grad_norm": 0.1653600037097931,
+      "learning_rate": 4.424026580381768e-06,
+      "loss": 0.9086,
+      "step": 79570
+    },
+    {
+      "epoch": 0.5760530449448775,
+      "grad_norm": 0.14946460723876953,
+      "learning_rate": 4.423954193721181e-06,
+      "loss": 0.9125,
+      "step": 79580
+    },
+    {
+      "epoch": 0.5761254316054637,
+      "grad_norm": 0.15209373831748962,
+      "learning_rate": 4.423881807060596e-06,
+      "loss": 0.9055,
+      "step": 79590
+    },
+    {
+      "epoch": 0.5761978182660499,
+      "grad_norm": 0.15624649822711945,
+      "learning_rate": 4.423809420400009e-06,
+      "loss": 0.9077,
+      "step": 79600
+    },
+    {
+      "epoch": 0.5762702049266362,
+      "grad_norm": 0.15583081543445587,
+      "learning_rate": 4.423737033739423e-06,
+      "loss": 0.8949,
+      "step": 79610
+    },
+    {
+      "epoch": 0.5763425915872223,
+      "grad_norm": 0.17833974957466125,
+      "learning_rate": 4.423664647078837e-06,
+      "loss": 0.9099,
+      "step": 79620
+    },
+    {
+      "epoch": 0.5764149782478085,
+      "grad_norm": 0.15480463206768036,
+      "learning_rate": 4.423592260418251e-06,
+      "loss": 0.9146,
+      "step": 79630
+    },
+    {
+      "epoch": 0.5764873649083947,
+      "grad_norm": 0.17085537314414978,
+      "learning_rate": 4.423519873757665e-06,
+      "loss": 0.9139,
+      "step": 79640
+    },
+    {
+      "epoch": 0.5765597515689809,
+      "grad_norm": 0.14300264418125153,
+      "learning_rate": 4.423447487097078e-06,
+      "loss": 0.9212,
+      "step": 79650
+    },
+    {
+      "epoch": 0.576632138229567,
+      "grad_norm": 0.1670181304216385,
+      "learning_rate": 4.423375100436492e-06,
+      "loss": 0.8988,
+      "step": 79660
+    },
+    {
+      "epoch": 0.5767045248901532,
+      "grad_norm": 0.16258497536182404,
+      "learning_rate": 4.4233027137759055e-06,
+      "loss": 0.9033,
+      "step": 79670
+    },
+    {
+      "epoch": 0.5767769115507394,
+      "grad_norm": 0.16200661659240723,
+      "learning_rate": 4.423230327115319e-06,
+      "loss": 0.9015,
+      "step": 79680
+    },
+    {
+      "epoch": 0.5768492982113256,
+      "grad_norm": 0.1577218770980835,
+      "learning_rate": 4.423157940454733e-06,
+      "loss": 0.9133,
+      "step": 79690
+    },
+    {
+      "epoch": 0.5769216848719118,
+      "grad_norm": 0.1620890349149704,
+      "learning_rate": 4.423085553794147e-06,
+      "loss": 0.9146,
+      "step": 79700
+    },
+    {
+      "epoch": 0.576994071532498,
+      "grad_norm": 0.18708209693431854,
+      "learning_rate": 4.423013167133561e-06,
+      "loss": 0.8999,
+      "step": 79710
+    },
+    {
+      "epoch": 0.5770664581930842,
+      "grad_norm": 0.1733100414276123,
+      "learning_rate": 4.4229407804729745e-06,
+      "loss": 0.9122,
+      "step": 79720
+    },
+    {
+      "epoch": 0.5771388448536704,
+      "grad_norm": 0.15812858939170837,
+      "learning_rate": 4.422868393812388e-06,
+      "loss": 0.907,
+      "step": 79730
+    },
+    {
+      "epoch": 0.5772112315142566,
+      "grad_norm": 0.1608811616897583,
+      "learning_rate": 4.4227960071518025e-06,
+      "loss": 0.9055,
+      "step": 79740
+    },
+    {
+      "epoch": 0.5772836181748427,
+      "grad_norm": 0.17216205596923828,
+      "learning_rate": 4.422723620491216e-06,
+      "loss": 0.919,
+      "step": 79750
+    },
+    {
+      "epoch": 0.5773560048354289,
+      "grad_norm": 0.36230647563934326,
+      "learning_rate": 4.42265123383063e-06,
+      "loss": 0.9139,
+      "step": 79760
+    },
+    {
+      "epoch": 0.5774283914960151,
+      "grad_norm": 0.15506798028945923,
+      "learning_rate": 4.422578847170043e-06,
+      "loss": 0.9009,
+      "step": 79770
+    },
+    {
+      "epoch": 0.5775007781566013,
+      "grad_norm": 0.1626974493265152,
+      "learning_rate": 4.422506460509458e-06,
+      "loss": 0.8987,
+      "step": 79780
+    },
+    {
+      "epoch": 0.5775731648171875,
+      "grad_norm": 0.1675325185060501,
+      "learning_rate": 4.4224340738488715e-06,
+      "loss": 0.9121,
+      "step": 79790
+    },
+    {
+      "epoch": 0.5776455514777736,
+      "grad_norm": 0.15305018424987793,
+      "learning_rate": 4.422361687188285e-06,
+      "loss": 0.9116,
+      "step": 79800
+    },
+    {
+      "epoch": 0.5777179381383598,
+      "grad_norm": 0.16215454041957855,
+      "learning_rate": 4.422289300527699e-06,
+      "loss": 0.9109,
+      "step": 79810
+    },
+    {
+      "epoch": 0.5777903247989461,
+      "grad_norm": 0.1520712971687317,
+      "learning_rate": 4.422216913867113e-06,
+      "loss": 0.9134,
+      "step": 79820
+    },
+    {
+      "epoch": 0.5778627114595323,
+      "grad_norm": 0.16776016354560852,
+      "learning_rate": 4.422144527206527e-06,
+      "loss": 0.9024,
+      "step": 79830
+    },
+    {
+      "epoch": 0.5779350981201185,
+      "grad_norm": 0.16810259222984314,
+      "learning_rate": 4.42207214054594e-06,
+      "loss": 0.9109,
+      "step": 79840
+    },
+    {
+      "epoch": 0.5780074847807046,
+      "grad_norm": 0.16688929498195648,
+      "learning_rate": 4.421999753885354e-06,
+      "loss": 0.9119,
+      "step": 79850
+    },
+    {
+      "epoch": 0.5780798714412908,
+      "grad_norm": 0.15468403697013855,
+      "learning_rate": 4.4219273672247685e-06,
+      "loss": 0.9151,
+      "step": 79860
+    },
+    {
+      "epoch": 0.578152258101877,
+      "grad_norm": 0.14916957914829254,
+      "learning_rate": 4.421854980564182e-06,
+      "loss": 0.9155,
+      "step": 79870
+    },
+    {
+      "epoch": 0.5782246447624632,
+      "grad_norm": 0.15850642323493958,
+      "learning_rate": 4.421782593903596e-06,
+      "loss": 0.9105,
+      "step": 79880
+    },
+    {
+      "epoch": 0.5782970314230493,
+      "grad_norm": 0.15652024745941162,
+      "learning_rate": 4.421710207243009e-06,
+      "loss": 0.9018,
+      "step": 79890
+    },
+    {
+      "epoch": 0.5783694180836355,
+      "grad_norm": 0.15815021097660065,
+      "learning_rate": 4.421637820582424e-06,
+      "loss": 0.9002,
+      "step": 79900
+    },
+    {
+      "epoch": 0.5784418047442217,
+      "grad_norm": 0.15399345755577087,
+      "learning_rate": 4.421565433921837e-06,
+      "loss": 0.9184,
+      "step": 79910
+    },
+    {
+      "epoch": 0.578514191404808,
+      "grad_norm": 0.16379056870937347,
+      "learning_rate": 4.421493047261251e-06,
+      "loss": 0.8932,
+      "step": 79920
+    },
+    {
+      "epoch": 0.5785865780653942,
+      "grad_norm": 0.1618824154138565,
+      "learning_rate": 4.421420660600665e-06,
+      "loss": 0.9028,
+      "step": 79930
+    },
+    {
+      "epoch": 0.5786589647259803,
+      "grad_norm": 0.1554708182811737,
+      "learning_rate": 4.421348273940079e-06,
+      "loss": 0.8998,
+      "step": 79940
+    },
+    {
+      "epoch": 0.5787313513865665,
+      "grad_norm": 0.16406132280826569,
+      "learning_rate": 4.421275887279493e-06,
+      "loss": 0.9058,
+      "step": 79950
+    },
+    {
+      "epoch": 0.5788037380471527,
+      "grad_norm": 0.15660354495048523,
+      "learning_rate": 4.421203500618906e-06,
+      "loss": 0.9134,
+      "step": 79960
+    },
+    {
+      "epoch": 0.5788761247077389,
+      "grad_norm": 0.15395161509513855,
+      "learning_rate": 4.42113111395832e-06,
+      "loss": 0.8982,
+      "step": 79970
+    },
+    {
+      "epoch": 0.578948511368325,
+      "grad_norm": 0.15442846715450287,
+      "learning_rate": 4.421058727297734e-06,
+      "loss": 0.9148,
+      "step": 79980
+    },
+    {
+      "epoch": 0.5790208980289112,
+      "grad_norm": 0.156655415892601,
+      "learning_rate": 4.420986340637148e-06,
+      "loss": 0.9125,
+      "step": 79990
+    },
+    {
+      "epoch": 0.5790932846894974,
+      "grad_norm": 0.16474436223506927,
+      "learning_rate": 4.420913953976562e-06,
+      "loss": 0.9103,
+      "step": 80000
+    },
+    {
+      "epoch": 0.5791656713500836,
+      "grad_norm": 0.17078536748886108,
+      "learning_rate": 4.420841567315975e-06,
+      "loss": 0.9094,
+      "step": 80010
+    },
+    {
+      "epoch": 0.5792380580106697,
+      "grad_norm": 0.1581786423921585,
+      "learning_rate": 4.420769180655389e-06,
+      "loss": 0.8923,
+      "step": 80020
+    },
+    {
+      "epoch": 0.579310444671256,
+      "grad_norm": 0.16456280648708344,
+      "learning_rate": 4.420696793994803e-06,
+      "loss": 0.9152,
+      "step": 80030
+    },
+    {
+      "epoch": 0.5793828313318422,
+      "grad_norm": 0.1432550996541977,
+      "learning_rate": 4.420624407334217e-06,
+      "loss": 0.8975,
+      "step": 80040
+    },
+    {
+      "epoch": 0.5794552179924284,
+      "grad_norm": 0.1518087089061737,
+      "learning_rate": 4.4205520206736306e-06,
+      "loss": 0.9096,
+      "step": 80050
+    },
+    {
+      "epoch": 0.5795276046530146,
+      "grad_norm": 0.15951353311538696,
+      "learning_rate": 4.420479634013044e-06,
+      "loss": 0.8996,
+      "step": 80060
+    },
+    {
+      "epoch": 0.5795999913136007,
+      "grad_norm": 0.16211937367916107,
+      "learning_rate": 4.420407247352459e-06,
+      "loss": 0.912,
+      "step": 80070
+    },
+    {
+      "epoch": 0.5796723779741869,
+      "grad_norm": 0.14895065128803253,
+      "learning_rate": 4.420334860691872e-06,
+      "loss": 0.9152,
+      "step": 80080
+    },
+    {
+      "epoch": 0.5797447646347731,
+      "grad_norm": 0.14833849668502808,
+      "learning_rate": 4.420262474031286e-06,
+      "loss": 0.9243,
+      "step": 80090
+    },
+    {
+      "epoch": 0.5798171512953593,
+      "grad_norm": 0.1672501266002655,
+      "learning_rate": 4.4201900873706995e-06,
+      "loss": 0.9072,
+      "step": 80100
+    },
+    {
+      "epoch": 0.5798895379559454,
+      "grad_norm": 0.16042830049991608,
+      "learning_rate": 4.420117700710114e-06,
+      "loss": 0.9099,
+      "step": 80110
+    },
+    {
+      "epoch": 0.5799619246165316,
+      "grad_norm": 0.16218118369579315,
+      "learning_rate": 4.420045314049528e-06,
+      "loss": 0.9036,
+      "step": 80120
+    },
+    {
+      "epoch": 0.5800343112771178,
+      "grad_norm": 0.1452185958623886,
+      "learning_rate": 4.419972927388941e-06,
+      "loss": 0.9013,
+      "step": 80130
+    },
+    {
+      "epoch": 0.5801066979377041,
+      "grad_norm": 0.1564382016658783,
+      "learning_rate": 4.419900540728355e-06,
+      "loss": 0.8933,
+      "step": 80140
+    },
+    {
+      "epoch": 0.5801790845982903,
+      "grad_norm": 0.1605406403541565,
+      "learning_rate": 4.419828154067769e-06,
+      "loss": 0.8964,
+      "step": 80150
+    },
+    {
+      "epoch": 0.5802514712588764,
+      "grad_norm": 0.16458991169929504,
+      "learning_rate": 4.419755767407183e-06,
+      "loss": 0.9147,
+      "step": 80160
+    },
+    {
+      "epoch": 0.5803238579194626,
+      "grad_norm": 0.16406084597110748,
+      "learning_rate": 4.4196833807465965e-06,
+      "loss": 0.9032,
+      "step": 80170
+    },
+    {
+      "epoch": 0.5803962445800488,
+      "grad_norm": 0.16171269118785858,
+      "learning_rate": 4.41961099408601e-06,
+      "loss": 0.8972,
+      "step": 80180
+    },
+    {
+      "epoch": 0.580468631240635,
+      "grad_norm": 0.1607089638710022,
+      "learning_rate": 4.419538607425424e-06,
+      "loss": 0.8913,
+      "step": 80190
+    },
+    {
+      "epoch": 0.5805410179012211,
+      "grad_norm": 0.15454769134521484,
+      "learning_rate": 4.419466220764837e-06,
+      "loss": 0.9014,
+      "step": 80200
+    },
+    {
+      "epoch": 0.5806134045618073,
+      "grad_norm": 0.15976151823997498,
+      "learning_rate": 4.419393834104251e-06,
+      "loss": 0.8891,
+      "step": 80210
+    },
+    {
+      "epoch": 0.5806857912223935,
+      "grad_norm": 0.15922151505947113,
+      "learning_rate": 4.4193214474436654e-06,
+      "loss": 0.9069,
+      "step": 80220
+    },
+    {
+      "epoch": 0.5807581778829797,
+      "grad_norm": 0.1480971723794937,
+      "learning_rate": 4.419249060783079e-06,
+      "loss": 0.9023,
+      "step": 80230
+    },
+    {
+      "epoch": 0.580830564543566,
+      "grad_norm": 0.3126855492591858,
+      "learning_rate": 4.419176674122493e-06,
+      "loss": 0.9113,
+      "step": 80240
+    },
+    {
+      "epoch": 0.5809029512041521,
+      "grad_norm": 0.15098969638347626,
+      "learning_rate": 4.419104287461906e-06,
+      "loss": 0.9212,
+      "step": 80250
+    },
+    {
+      "epoch": 0.5809753378647383,
+      "grad_norm": 0.16232796013355255,
+      "learning_rate": 4.419031900801321e-06,
+      "loss": 0.8948,
+      "step": 80260
+    },
+    {
+      "epoch": 0.5810477245253245,
+      "grad_norm": 0.16059771180152893,
+      "learning_rate": 4.418959514140734e-06,
+      "loss": 0.9039,
+      "step": 80270
+    },
+    {
+      "epoch": 0.5811201111859107,
+      "grad_norm": 0.15262576937675476,
+      "learning_rate": 4.418887127480148e-06,
+      "loss": 0.9075,
+      "step": 80280
+    },
+    {
+      "epoch": 0.5811924978464968,
+      "grad_norm": 0.15464602410793304,
+      "learning_rate": 4.418814740819562e-06,
+      "loss": 0.9119,
+      "step": 80290
+    },
+    {
+      "epoch": 0.581264884507083,
+      "grad_norm": 0.1607392579317093,
+      "learning_rate": 4.418742354158976e-06,
+      "loss": 0.903,
+      "step": 80300
+    },
+    {
+      "epoch": 0.5813372711676692,
+      "grad_norm": 0.16211529076099396,
+      "learning_rate": 4.41866996749839e-06,
+      "loss": 0.9033,
+      "step": 80310
+    },
+    {
+      "epoch": 0.5814096578282554,
+      "grad_norm": 0.1596810221672058,
+      "learning_rate": 4.418597580837803e-06,
+      "loss": 0.9072,
+      "step": 80320
+    },
+    {
+      "epoch": 0.5814820444888416,
+      "grad_norm": 0.15583199262619019,
+      "learning_rate": 4.418525194177217e-06,
+      "loss": 0.9146,
+      "step": 80330
+    },
+    {
+      "epoch": 0.5815544311494277,
+      "grad_norm": 0.1569822132587433,
+      "learning_rate": 4.418452807516631e-06,
+      "loss": 0.9046,
+      "step": 80340
+    },
+    {
+      "epoch": 0.581626817810014,
+      "grad_norm": 0.15323781967163086,
+      "learning_rate": 4.418380420856045e-06,
+      "loss": 0.9146,
+      "step": 80350
+    },
+    {
+      "epoch": 0.5816992044706002,
+      "grad_norm": 0.161820650100708,
+      "learning_rate": 4.418308034195459e-06,
+      "loss": 0.9147,
+      "step": 80360
+    },
+    {
+      "epoch": 0.5817715911311864,
+      "grad_norm": 0.1547018438577652,
+      "learning_rate": 4.418235647534872e-06,
+      "loss": 0.9052,
+      "step": 80370
+    },
+    {
+      "epoch": 0.5818439777917725,
+      "grad_norm": 0.16685166954994202,
+      "learning_rate": 4.418163260874287e-06,
+      "loss": 0.9076,
+      "step": 80380
+    },
+    {
+      "epoch": 0.5819163644523587,
+      "grad_norm": 0.1596868634223938,
+      "learning_rate": 4.4180908742137e-06,
+      "loss": 0.8918,
+      "step": 80390
+    },
+    {
+      "epoch": 0.5819887511129449,
+      "grad_norm": 0.16917213797569275,
+      "learning_rate": 4.418018487553114e-06,
+      "loss": 0.8946,
+      "step": 80400
+    },
+    {
+      "epoch": 0.5820611377735311,
+      "grad_norm": 0.15056735277175903,
+      "learning_rate": 4.4179461008925275e-06,
+      "loss": 0.9045,
+      "step": 80410
+    },
+    {
+      "epoch": 0.5821335244341173,
+      "grad_norm": 0.1574123352766037,
+      "learning_rate": 4.417873714231942e-06,
+      "loss": 0.9067,
+      "step": 80420
+    },
+    {
+      "epoch": 0.5822059110947034,
+      "grad_norm": 0.15226131677627563,
+      "learning_rate": 4.417801327571356e-06,
+      "loss": 0.9183,
+      "step": 80430
+    },
+    {
+      "epoch": 0.5822782977552896,
+      "grad_norm": 0.15113304555416107,
+      "learning_rate": 4.417728940910769e-06,
+      "loss": 0.9102,
+      "step": 80440
+    },
+    {
+      "epoch": 0.5823506844158759,
+      "grad_norm": 0.14975102245807648,
+      "learning_rate": 4.417656554250183e-06,
+      "loss": 0.902,
+      "step": 80450
+    },
+    {
+      "epoch": 0.5824230710764621,
+      "grad_norm": 0.1500435620546341,
+      "learning_rate": 4.417584167589597e-06,
+      "loss": 0.9046,
+      "step": 80460
+    },
+    {
+      "epoch": 0.5824954577370483,
+      "grad_norm": 0.15473416447639465,
+      "learning_rate": 4.417511780929011e-06,
+      "loss": 0.9146,
+      "step": 80470
+    },
+    {
+      "epoch": 0.5825678443976344,
+      "grad_norm": 0.1687445044517517,
+      "learning_rate": 4.4174393942684245e-06,
+      "loss": 0.9172,
+      "step": 80480
+    },
+    {
+      "epoch": 0.5826402310582206,
+      "grad_norm": 0.18456241488456726,
+      "learning_rate": 4.417367007607838e-06,
+      "loss": 0.9059,
+      "step": 80490
+    },
+    {
+      "epoch": 0.5827126177188068,
+      "grad_norm": 0.16103129088878632,
+      "learning_rate": 4.417294620947253e-06,
+      "loss": 0.9129,
+      "step": 80500
+    },
+    {
+      "epoch": 0.582785004379393,
+      "grad_norm": 0.15495523810386658,
+      "learning_rate": 4.417222234286666e-06,
+      "loss": 0.8998,
+      "step": 80510
+    },
+    {
+      "epoch": 0.5828573910399791,
+      "grad_norm": 0.1617932915687561,
+      "learning_rate": 4.41714984762608e-06,
+      "loss": 0.9129,
+      "step": 80520
+    },
+    {
+      "epoch": 0.5829297777005653,
+      "grad_norm": 0.1493421494960785,
+      "learning_rate": 4.4170774609654935e-06,
+      "loss": 0.8968,
+      "step": 80530
+    },
+    {
+      "epoch": 0.5830021643611515,
+      "grad_norm": 0.15675191581249237,
+      "learning_rate": 4.417005074304908e-06,
+      "loss": 0.9182,
+      "step": 80540
+    },
+    {
+      "epoch": 0.5830745510217377,
+      "grad_norm": 0.16273775696754456,
+      "learning_rate": 4.4169326876443216e-06,
+      "loss": 0.9103,
+      "step": 80550
+    },
+    {
+      "epoch": 0.583146937682324,
+      "grad_norm": 0.171338751912117,
+      "learning_rate": 4.416860300983735e-06,
+      "loss": 0.9151,
+      "step": 80560
+    },
+    {
+      "epoch": 0.5832193243429101,
+      "grad_norm": 0.16506828367710114,
+      "learning_rate": 4.416787914323149e-06,
+      "loss": 0.9137,
+      "step": 80570
+    },
+    {
+      "epoch": 0.5832917110034963,
+      "grad_norm": 0.1581031233072281,
+      "learning_rate": 4.416715527662563e-06,
+      "loss": 0.8992,
+      "step": 80580
+    },
+    {
+      "epoch": 0.5833640976640825,
+      "grad_norm": 0.1575007438659668,
+      "learning_rate": 4.416643141001977e-06,
+      "loss": 0.9053,
+      "step": 80590
+    },
+    {
+      "epoch": 0.5834364843246687,
+      "grad_norm": 0.17189882695674896,
+      "learning_rate": 4.4165707543413905e-06,
+      "loss": 0.9148,
+      "step": 80600
+    },
+    {
+      "epoch": 0.5835088709852548,
+      "grad_norm": 0.16765065491199493,
+      "learning_rate": 4.416498367680804e-06,
+      "loss": 0.9063,
+      "step": 80610
+    },
+    {
+      "epoch": 0.583581257645841,
+      "grad_norm": 0.15300755202770233,
+      "learning_rate": 4.416425981020218e-06,
+      "loss": 0.9059,
+      "step": 80620
+    },
+    {
+      "epoch": 0.5836536443064272,
+      "grad_norm": 0.16409176588058472,
+      "learning_rate": 4.416353594359632e-06,
+      "loss": 0.9074,
+      "step": 80630
+    },
+    {
+      "epoch": 0.5837260309670134,
+      "grad_norm": 0.1527976095676422,
+      "learning_rate": 4.416281207699046e-06,
+      "loss": 0.9057,
+      "step": 80640
+    },
+    {
+      "epoch": 0.5837984176275995,
+      "grad_norm": 0.17555196583271027,
+      "learning_rate": 4.416208821038459e-06,
+      "loss": 0.918,
+      "step": 80650
+    },
+    {
+      "epoch": 0.5838708042881857,
+      "grad_norm": 0.15772706270217896,
+      "learning_rate": 4.416136434377873e-06,
+      "loss": 0.9064,
+      "step": 80660
+    },
+    {
+      "epoch": 0.583943190948772,
+      "grad_norm": 0.16107985377311707,
+      "learning_rate": 4.4160640477172875e-06,
+      "loss": 0.9153,
+      "step": 80670
+    },
+    {
+      "epoch": 0.5840155776093582,
+      "grad_norm": 0.27143919467926025,
+      "learning_rate": 4.415991661056701e-06,
+      "loss": 0.9147,
+      "step": 80680
+    },
+    {
+      "epoch": 0.5840879642699444,
+      "grad_norm": 0.1891665756702423,
+      "learning_rate": 4.415919274396115e-06,
+      "loss": 0.9206,
+      "step": 80690
+    },
+    {
+      "epoch": 0.5841603509305305,
+      "grad_norm": 0.15576356649398804,
+      "learning_rate": 4.415846887735528e-06,
+      "loss": 0.9045,
+      "step": 80700
+    },
+    {
+      "epoch": 0.5842327375911167,
+      "grad_norm": 0.16257961094379425,
+      "learning_rate": 4.415774501074943e-06,
+      "loss": 0.8978,
+      "step": 80710
+    },
+    {
+      "epoch": 0.5843051242517029,
+      "grad_norm": 0.17086383700370789,
+      "learning_rate": 4.4157021144143556e-06,
+      "loss": 0.9039,
+      "step": 80720
+    },
+    {
+      "epoch": 0.5843775109122891,
+      "grad_norm": 0.205778568983078,
+      "learning_rate": 4.41562972775377e-06,
+      "loss": 0.9195,
+      "step": 80730
+    },
+    {
+      "epoch": 0.5844498975728752,
+      "grad_norm": 0.15765634179115295,
+      "learning_rate": 4.415557341093184e-06,
+      "loss": 0.9038,
+      "step": 80740
+    },
+    {
+      "epoch": 0.5845222842334614,
+      "grad_norm": 0.14532588422298431,
+      "learning_rate": 4.415484954432597e-06,
+      "loss": 0.9122,
+      "step": 80750
+    },
+    {
+      "epoch": 0.5845946708940476,
+      "grad_norm": 0.172866091132164,
+      "learning_rate": 4.415412567772011e-06,
+      "loss": 0.9122,
+      "step": 80760
+    },
+    {
+      "epoch": 0.5846670575546339,
+      "grad_norm": 0.16551901400089264,
+      "learning_rate": 4.415340181111425e-06,
+      "loss": 0.9049,
+      "step": 80770
+    },
+    {
+      "epoch": 0.5847394442152201,
+      "grad_norm": 0.15034866333007812,
+      "learning_rate": 4.415267794450839e-06,
+      "loss": 0.9006,
+      "step": 80780
+    },
+    {
+      "epoch": 0.5848118308758062,
+      "grad_norm": 0.1626167744398117,
+      "learning_rate": 4.4151954077902526e-06,
+      "loss": 0.9122,
+      "step": 80790
+    },
+    {
+      "epoch": 0.5848842175363924,
+      "grad_norm": 0.2411961555480957,
+      "learning_rate": 4.415123021129666e-06,
+      "loss": 0.9039,
+      "step": 80800
+    },
+    {
+      "epoch": 0.5849566041969786,
+      "grad_norm": 0.16269677877426147,
+      "learning_rate": 4.41505063446908e-06,
+      "loss": 0.923,
+      "step": 80810
+    },
+    {
+      "epoch": 0.5850289908575648,
+      "grad_norm": 0.15651679039001465,
+      "learning_rate": 4.414978247808494e-06,
+      "loss": 0.9018,
+      "step": 80820
+    },
+    {
+      "epoch": 0.585101377518151,
+      "grad_norm": 0.1721392273902893,
+      "learning_rate": 4.414905861147908e-06,
+      "loss": 0.9058,
+      "step": 80830
+    },
+    {
+      "epoch": 0.5851737641787371,
+      "grad_norm": 0.1486913561820984,
+      "learning_rate": 4.4148334744873215e-06,
+      "loss": 0.9067,
+      "step": 80840
+    },
+    {
+      "epoch": 0.5852461508393233,
+      "grad_norm": 0.16526785492897034,
+      "learning_rate": 4.414761087826735e-06,
+      "loss": 0.9126,
+      "step": 80850
+    },
+    {
+      "epoch": 0.5853185374999095,
+      "grad_norm": 0.18631219863891602,
+      "learning_rate": 4.41468870116615e-06,
+      "loss": 0.9067,
+      "step": 80860
+    },
+    {
+      "epoch": 0.5853909241604957,
+      "grad_norm": 0.15394018590450287,
+      "learning_rate": 4.414616314505563e-06,
+      "loss": 0.9065,
+      "step": 80870
+    },
+    {
+      "epoch": 0.5854633108210819,
+      "grad_norm": 0.15232212841510773,
+      "learning_rate": 4.414543927844977e-06,
+      "loss": 0.907,
+      "step": 80880
+    },
+    {
+      "epoch": 0.5855356974816681,
+      "grad_norm": 0.15810738503932953,
+      "learning_rate": 4.4144715411843904e-06,
+      "loss": 0.9133,
+      "step": 80890
+    },
+    {
+      "epoch": 0.5856080841422543,
+      "grad_norm": 0.15687547624111176,
+      "learning_rate": 4.414399154523805e-06,
+      "loss": 0.9222,
+      "step": 80900
+    },
+    {
+      "epoch": 0.5856804708028405,
+      "grad_norm": 0.1554637998342514,
+      "learning_rate": 4.4143267678632185e-06,
+      "loss": 0.9004,
+      "step": 80910
+    },
+    {
+      "epoch": 0.5857528574634266,
+      "grad_norm": 0.14952871203422546,
+      "learning_rate": 4.414254381202632e-06,
+      "loss": 0.9071,
+      "step": 80920
+    },
+    {
+      "epoch": 0.5858252441240128,
+      "grad_norm": 0.15082089602947235,
+      "learning_rate": 4.414181994542046e-06,
+      "loss": 0.9173,
+      "step": 80930
+    },
+    {
+      "epoch": 0.585897630784599,
+      "grad_norm": 0.1659436672925949,
+      "learning_rate": 4.41410960788146e-06,
+      "loss": 0.9038,
+      "step": 80940
+    },
+    {
+      "epoch": 0.5859700174451852,
+      "grad_norm": 0.16675877571105957,
+      "learning_rate": 4.414037221220874e-06,
+      "loss": 0.9022,
+      "step": 80950
+    },
+    {
+      "epoch": 0.5860424041057714,
+      "grad_norm": 0.17415930330753326,
+      "learning_rate": 4.4139648345602874e-06,
+      "loss": 0.912,
+      "step": 80960
+    },
+    {
+      "epoch": 0.5861147907663575,
+      "grad_norm": 0.16868185997009277,
+      "learning_rate": 4.413892447899701e-06,
+      "loss": 0.9139,
+      "step": 80970
+    },
+    {
+      "epoch": 0.5861871774269438,
+      "grad_norm": 0.1576145589351654,
+      "learning_rate": 4.4138200612391155e-06,
+      "loss": 0.9084,
+      "step": 80980
+    },
+    {
+      "epoch": 0.58625956408753,
+      "grad_norm": 0.16312648355960846,
+      "learning_rate": 4.413747674578529e-06,
+      "loss": 0.9107,
+      "step": 80990
+    },
+    {
+      "epoch": 0.5863319507481162,
+      "grad_norm": 0.1908407360315323,
+      "learning_rate": 4.413675287917943e-06,
+      "loss": 0.9057,
+      "step": 81000
+    },
+    {
+      "epoch": 0.5864043374087023,
+      "grad_norm": 0.19931037724018097,
+      "learning_rate": 4.413602901257356e-06,
+      "loss": 0.9036,
+      "step": 81010
+    },
+    {
+      "epoch": 0.5864767240692885,
+      "grad_norm": 0.15866592526435852,
+      "learning_rate": 4.413530514596771e-06,
+      "loss": 0.9027,
+      "step": 81020
+    },
+    {
+      "epoch": 0.5865491107298747,
+      "grad_norm": 0.1934008151292801,
+      "learning_rate": 4.4134581279361844e-06,
+      "loss": 0.9087,
+      "step": 81030
+    },
+    {
+      "epoch": 0.5866214973904609,
+      "grad_norm": 0.1468328833580017,
+      "learning_rate": 4.413385741275598e-06,
+      "loss": 0.9099,
+      "step": 81040
+    },
+    {
+      "epoch": 0.586693884051047,
+      "grad_norm": 0.16777294874191284,
+      "learning_rate": 4.413313354615012e-06,
+      "loss": 0.8982,
+      "step": 81050
+    },
+    {
+      "epoch": 0.5867662707116332,
+      "grad_norm": 0.15594753623008728,
+      "learning_rate": 4.413240967954426e-06,
+      "loss": 0.9146,
+      "step": 81060
+    },
+    {
+      "epoch": 0.5868386573722194,
+      "grad_norm": 0.15250714123249054,
+      "learning_rate": 4.41316858129384e-06,
+      "loss": 0.9137,
+      "step": 81070
+    },
+    {
+      "epoch": 0.5869110440328056,
+      "grad_norm": 0.15024009346961975,
+      "learning_rate": 4.413096194633253e-06,
+      "loss": 0.9134,
+      "step": 81080
+    },
+    {
+      "epoch": 0.5869834306933919,
+      "grad_norm": 0.1556905061006546,
+      "learning_rate": 4.413023807972667e-06,
+      "loss": 0.9178,
+      "step": 81090
+    },
+    {
+      "epoch": 0.587055817353978,
+      "grad_norm": 0.1638568490743637,
+      "learning_rate": 4.4129514213120815e-06,
+      "loss": 0.9126,
+      "step": 81100
+    },
+    {
+      "epoch": 0.5871282040145642,
+      "grad_norm": 0.15844887495040894,
+      "learning_rate": 4.412879034651495e-06,
+      "loss": 0.9135,
+      "step": 81110
+    },
+    {
+      "epoch": 0.5872005906751504,
+      "grad_norm": 0.1853482723236084,
+      "learning_rate": 4.412806647990909e-06,
+      "loss": 0.897,
+      "step": 81120
+    },
+    {
+      "epoch": 0.5872729773357366,
+      "grad_norm": 0.15754495561122894,
+      "learning_rate": 4.412734261330322e-06,
+      "loss": 0.9092,
+      "step": 81130
+    },
+    {
+      "epoch": 0.5873453639963228,
+      "grad_norm": 0.16885913908481598,
+      "learning_rate": 4.412661874669737e-06,
+      "loss": 0.9082,
+      "step": 81140
+    },
+    {
+      "epoch": 0.5874177506569089,
+      "grad_norm": 0.16286881268024445,
+      "learning_rate": 4.41258948800915e-06,
+      "loss": 0.893,
+      "step": 81150
+    },
+    {
+      "epoch": 0.5874901373174951,
+      "grad_norm": 0.14878816902637482,
+      "learning_rate": 4.412517101348564e-06,
+      "loss": 0.9028,
+      "step": 81160
+    },
+    {
+      "epoch": 0.5875625239780813,
+      "grad_norm": 0.17409685254096985,
+      "learning_rate": 4.412444714687978e-06,
+      "loss": 0.9155,
+      "step": 81170
+    },
+    {
+      "epoch": 0.5876349106386675,
+      "grad_norm": 0.1579102873802185,
+      "learning_rate": 4.412372328027392e-06,
+      "loss": 0.9146,
+      "step": 81180
+    },
+    {
+      "epoch": 0.5877072972992536,
+      "grad_norm": 0.15453718602657318,
+      "learning_rate": 4.412299941366806e-06,
+      "loss": 0.8998,
+      "step": 81190
+    },
+    {
+      "epoch": 0.5877796839598399,
+      "grad_norm": 0.16412879526615143,
+      "learning_rate": 4.412227554706219e-06,
+      "loss": 0.907,
+      "step": 81200
+    },
+    {
+      "epoch": 0.5878520706204261,
+      "grad_norm": 0.1556544452905655,
+      "learning_rate": 4.412155168045633e-06,
+      "loss": 0.9067,
+      "step": 81210
+    },
+    {
+      "epoch": 0.5879244572810123,
+      "grad_norm": 0.1514187902212143,
+      "learning_rate": 4.412082781385047e-06,
+      "loss": 0.909,
+      "step": 81220
+    },
+    {
+      "epoch": 0.5879968439415985,
+      "grad_norm": 0.154129296541214,
+      "learning_rate": 4.412010394724461e-06,
+      "loss": 0.9017,
+      "step": 81230
+    },
+    {
+      "epoch": 0.5880692306021846,
+      "grad_norm": 0.16333553194999695,
+      "learning_rate": 4.411938008063875e-06,
+      "loss": 0.9039,
+      "step": 81240
+    },
+    {
+      "epoch": 0.5881416172627708,
+      "grad_norm": 0.1498739868402481,
+      "learning_rate": 4.411865621403288e-06,
+      "loss": 0.8942,
+      "step": 81250
+    },
+    {
+      "epoch": 0.588214003923357,
+      "grad_norm": 0.1580556333065033,
+      "learning_rate": 4.411793234742702e-06,
+      "loss": 0.8984,
+      "step": 81260
+    },
+    {
+      "epoch": 0.5882863905839432,
+      "grad_norm": 0.15687990188598633,
+      "learning_rate": 4.4117208480821155e-06,
+      "loss": 0.9096,
+      "step": 81270
+    },
+    {
+      "epoch": 0.5883587772445293,
+      "grad_norm": 0.17736192047595978,
+      "learning_rate": 4.411648461421529e-06,
+      "loss": 0.9031,
+      "step": 81280
+    },
+    {
+      "epoch": 0.5884311639051155,
+      "grad_norm": 0.1527496576309204,
+      "learning_rate": 4.4115760747609435e-06,
+      "loss": 0.8972,
+      "step": 81290
+    },
+    {
+      "epoch": 0.5885035505657018,
+      "grad_norm": 0.17745208740234375,
+      "learning_rate": 4.411503688100357e-06,
+      "loss": 0.917,
+      "step": 81300
+    },
+    {
+      "epoch": 0.588575937226288,
+      "grad_norm": 0.20360904932022095,
+      "learning_rate": 4.411431301439771e-06,
+      "loss": 0.9189,
+      "step": 81310
+    },
+    {
+      "epoch": 0.5886483238868742,
+      "grad_norm": 0.1483248472213745,
+      "learning_rate": 4.411358914779184e-06,
+      "loss": 0.9098,
+      "step": 81320
+    },
+    {
+      "epoch": 0.5887207105474603,
+      "grad_norm": 0.1655409038066864,
+      "learning_rate": 4.411286528118599e-06,
+      "loss": 0.9164,
+      "step": 81330
+    },
+    {
+      "epoch": 0.5887930972080465,
+      "grad_norm": 0.1597280353307724,
+      "learning_rate": 4.4112141414580125e-06,
+      "loss": 0.9058,
+      "step": 81340
+    },
+    {
+      "epoch": 0.5888654838686327,
+      "grad_norm": 0.16111089289188385,
+      "learning_rate": 4.411141754797426e-06,
+      "loss": 0.9048,
+      "step": 81350
+    },
+    {
+      "epoch": 0.5889378705292189,
+      "grad_norm": 0.16747364401817322,
+      "learning_rate": 4.41106936813684e-06,
+      "loss": 0.9183,
+      "step": 81360
+    },
+    {
+      "epoch": 0.589010257189805,
+      "grad_norm": 0.14955104887485504,
+      "learning_rate": 4.410996981476254e-06,
+      "loss": 0.9064,
+      "step": 81370
+    },
+    {
+      "epoch": 0.5890826438503912,
+      "grad_norm": 0.15736231207847595,
+      "learning_rate": 4.410924594815668e-06,
+      "loss": 0.9048,
+      "step": 81380
+    },
+    {
+      "epoch": 0.5891550305109774,
+      "grad_norm": 0.20818248391151428,
+      "learning_rate": 4.410852208155081e-06,
+      "loss": 0.8868,
+      "step": 81390
+    },
+    {
+      "epoch": 0.5892274171715636,
+      "grad_norm": 0.15236660838127136,
+      "learning_rate": 4.410779821494495e-06,
+      "loss": 0.9082,
+      "step": 81400
+    },
+    {
+      "epoch": 0.5892998038321499,
+      "grad_norm": 0.1863219141960144,
+      "learning_rate": 4.4107074348339095e-06,
+      "loss": 0.9106,
+      "step": 81410
+    },
+    {
+      "epoch": 0.589372190492736,
+      "grad_norm": 0.14689262211322784,
+      "learning_rate": 4.410635048173323e-06,
+      "loss": 0.9158,
+      "step": 81420
+    },
+    {
+      "epoch": 0.5894445771533222,
+      "grad_norm": 0.15385663509368896,
+      "learning_rate": 4.410562661512737e-06,
+      "loss": 0.9033,
+      "step": 81430
+    },
+    {
+      "epoch": 0.5895169638139084,
+      "grad_norm": 0.18120580911636353,
+      "learning_rate": 4.41049027485215e-06,
+      "loss": 0.9147,
+      "step": 81440
+    },
+    {
+      "epoch": 0.5895893504744946,
+      "grad_norm": 0.16570447385311127,
+      "learning_rate": 4.410417888191564e-06,
+      "loss": 0.9152,
+      "step": 81450
+    },
+    {
+      "epoch": 0.5896617371350807,
+      "grad_norm": 0.15214945375919342,
+      "learning_rate": 4.410345501530978e-06,
+      "loss": 0.9046,
+      "step": 81460
+    },
+    {
+      "epoch": 0.5897341237956669,
+      "grad_norm": 0.1923670470714569,
+      "learning_rate": 4.410273114870392e-06,
+      "loss": 0.918,
+      "step": 81470
+    },
+    {
+      "epoch": 0.5898065104562531,
+      "grad_norm": 0.1635698527097702,
+      "learning_rate": 4.410200728209806e-06,
+      "loss": 0.9058,
+      "step": 81480
+    },
+    {
+      "epoch": 0.5898788971168393,
+      "grad_norm": 0.1771516501903534,
+      "learning_rate": 4.410128341549219e-06,
+      "loss": 0.9068,
+      "step": 81490
+    },
+    {
+      "epoch": 0.5899512837774255,
+      "grad_norm": 0.16354140639305115,
+      "learning_rate": 4.410055954888634e-06,
+      "loss": 0.9066,
+      "step": 81500
+    },
+    {
+      "epoch": 0.5900236704380117,
+      "grad_norm": 0.16545943915843964,
+      "learning_rate": 4.409983568228047e-06,
+      "loss": 0.9138,
+      "step": 81510
+    },
+    {
+      "epoch": 0.5900960570985979,
+      "grad_norm": 0.15025997161865234,
+      "learning_rate": 4.409911181567461e-06,
+      "loss": 0.8911,
+      "step": 81520
+    },
+    {
+      "epoch": 0.5901684437591841,
+      "grad_norm": 0.17573940753936768,
+      "learning_rate": 4.4098387949068746e-06,
+      "loss": 0.9041,
+      "step": 81530
+    },
+    {
+      "epoch": 0.5902408304197703,
+      "grad_norm": 0.17626507580280304,
+      "learning_rate": 4.409766408246289e-06,
+      "loss": 0.9036,
+      "step": 81540
+    },
+    {
+      "epoch": 0.5903132170803564,
+      "grad_norm": 0.16366511583328247,
+      "learning_rate": 4.409694021585703e-06,
+      "loss": 0.9016,
+      "step": 81550
+    },
+    {
+      "epoch": 0.5903856037409426,
+      "grad_norm": 0.17350682616233826,
+      "learning_rate": 4.409621634925116e-06,
+      "loss": 0.9112,
+      "step": 81560
+    },
+    {
+      "epoch": 0.5904579904015288,
+      "grad_norm": 0.16151121258735657,
+      "learning_rate": 4.40954924826453e-06,
+      "loss": 0.9051,
+      "step": 81570
+    },
+    {
+      "epoch": 0.590530377062115,
+      "grad_norm": 0.16566744446754456,
+      "learning_rate": 4.409476861603944e-06,
+      "loss": 0.9037,
+      "step": 81580
+    },
+    {
+      "epoch": 0.5906027637227012,
+      "grad_norm": 0.16139444708824158,
+      "learning_rate": 4.409404474943358e-06,
+      "loss": 0.907,
+      "step": 81590
+    },
+    {
+      "epoch": 0.5906751503832873,
+      "grad_norm": 0.16810381412506104,
+      "learning_rate": 4.409332088282772e-06,
+      "loss": 0.8919,
+      "step": 81600
+    },
+    {
+      "epoch": 0.5907475370438735,
+      "grad_norm": 0.1652746945619583,
+      "learning_rate": 4.409259701622185e-06,
+      "loss": 0.9017,
+      "step": 81610
+    },
+    {
+      "epoch": 0.5908199237044598,
+      "grad_norm": 0.1639404594898224,
+      "learning_rate": 4.4091873149616e-06,
+      "loss": 0.8995,
+      "step": 81620
+    },
+    {
+      "epoch": 0.590892310365046,
+      "grad_norm": 0.1466279774904251,
+      "learning_rate": 4.409114928301013e-06,
+      "loss": 0.9177,
+      "step": 81630
+    },
+    {
+      "epoch": 0.5909646970256321,
+      "grad_norm": 0.16004787385463715,
+      "learning_rate": 4.409042541640427e-06,
+      "loss": 0.9039,
+      "step": 81640
+    },
+    {
+      "epoch": 0.5910370836862183,
+      "grad_norm": 0.15307894349098206,
+      "learning_rate": 4.4089701549798405e-06,
+      "loss": 0.9036,
+      "step": 81650
+    },
+    {
+      "epoch": 0.5911094703468045,
+      "grad_norm": 0.19654370844364166,
+      "learning_rate": 4.408897768319255e-06,
+      "loss": 0.9107,
+      "step": 81660
+    },
+    {
+      "epoch": 0.5911818570073907,
+      "grad_norm": 0.17902538180351257,
+      "learning_rate": 4.408825381658669e-06,
+      "loss": 0.9045,
+      "step": 81670
+    },
+    {
+      "epoch": 0.5912542436679769,
+      "grad_norm": 0.1651085615158081,
+      "learning_rate": 4.408752994998082e-06,
+      "loss": 0.9106,
+      "step": 81680
+    },
+    {
+      "epoch": 0.591326630328563,
+      "grad_norm": 0.1591925323009491,
+      "learning_rate": 4.408680608337496e-06,
+      "loss": 0.904,
+      "step": 81690
+    },
+    {
+      "epoch": 0.5913990169891492,
+      "grad_norm": 0.1673898696899414,
+      "learning_rate": 4.40860822167691e-06,
+      "loss": 0.9005,
+      "step": 81700
+    },
+    {
+      "epoch": 0.5914714036497354,
+      "grad_norm": 0.16726025938987732,
+      "learning_rate": 4.408535835016324e-06,
+      "loss": 0.9051,
+      "step": 81710
+    },
+    {
+      "epoch": 0.5915437903103216,
+      "grad_norm": 0.16342008113861084,
+      "learning_rate": 4.4084634483557375e-06,
+      "loss": 0.9207,
+      "step": 81720
+    },
+    {
+      "epoch": 0.5916161769709078,
+      "grad_norm": 0.15118926763534546,
+      "learning_rate": 4.408391061695151e-06,
+      "loss": 0.9022,
+      "step": 81730
+    },
+    {
+      "epoch": 0.591688563631494,
+      "grad_norm": 0.15618331730365753,
+      "learning_rate": 4.408318675034566e-06,
+      "loss": 0.9089,
+      "step": 81740
+    },
+    {
+      "epoch": 0.5917609502920802,
+      "grad_norm": 0.16020120680332184,
+      "learning_rate": 4.408246288373979e-06,
+      "loss": 0.9077,
+      "step": 81750
+    },
+    {
+      "epoch": 0.5918333369526664,
+      "grad_norm": 0.15385285019874573,
+      "learning_rate": 4.408173901713393e-06,
+      "loss": 0.909,
+      "step": 81760
+    },
+    {
+      "epoch": 0.5919057236132526,
+      "grad_norm": 0.15946872532367706,
+      "learning_rate": 4.4081015150528064e-06,
+      "loss": 0.9251,
+      "step": 81770
+    },
+    {
+      "epoch": 0.5919781102738387,
+      "grad_norm": 0.16759130358695984,
+      "learning_rate": 4.40802912839222e-06,
+      "loss": 0.9045,
+      "step": 81780
+    },
+    {
+      "epoch": 0.5920504969344249,
+      "grad_norm": 0.1478552371263504,
+      "learning_rate": 4.407956741731634e-06,
+      "loss": 0.9143,
+      "step": 81790
+    },
+    {
+      "epoch": 0.5921228835950111,
+      "grad_norm": 0.18161530792713165,
+      "learning_rate": 4.407884355071047e-06,
+      "loss": 0.9115,
+      "step": 81800
+    },
+    {
+      "epoch": 0.5921952702555973,
+      "grad_norm": 0.16163845360279083,
+      "learning_rate": 4.407811968410462e-06,
+      "loss": 0.8982,
+      "step": 81810
+    },
+    {
+      "epoch": 0.5922676569161834,
+      "grad_norm": 0.15115785598754883,
+      "learning_rate": 4.407739581749875e-06,
+      "loss": 0.9166,
+      "step": 81820
+    },
+    {
+      "epoch": 0.5923400435767697,
+      "grad_norm": 0.1659201681613922,
+      "learning_rate": 4.407667195089289e-06,
+      "loss": 0.9008,
+      "step": 81830
+    },
+    {
+      "epoch": 0.5924124302373559,
+      "grad_norm": 0.1570078432559967,
+      "learning_rate": 4.407594808428703e-06,
+      "loss": 0.8988,
+      "step": 81840
+    },
+    {
+      "epoch": 0.5924848168979421,
+      "grad_norm": 0.1584419459104538,
+      "learning_rate": 4.407522421768117e-06,
+      "loss": 0.9097,
+      "step": 81850
+    },
+    {
+      "epoch": 0.5925572035585283,
+      "grad_norm": 0.18121954798698425,
+      "learning_rate": 4.407450035107531e-06,
+      "loss": 0.8932,
+      "step": 81860
+    },
+    {
+      "epoch": 0.5926295902191144,
+      "grad_norm": 0.15848958492279053,
+      "learning_rate": 4.407377648446944e-06,
+      "loss": 0.9048,
+      "step": 81870
+    },
+    {
+      "epoch": 0.5927019768797006,
+      "grad_norm": 0.17608241736888885,
+      "learning_rate": 4.407305261786358e-06,
+      "loss": 0.9077,
+      "step": 81880
+    },
+    {
+      "epoch": 0.5927743635402868,
+      "grad_norm": 0.16470035910606384,
+      "learning_rate": 4.407232875125772e-06,
+      "loss": 0.9114,
+      "step": 81890
+    },
+    {
+      "epoch": 0.592846750200873,
+      "grad_norm": 0.14142341911792755,
+      "learning_rate": 4.407160488465186e-06,
+      "loss": 0.8911,
+      "step": 81900
+    },
+    {
+      "epoch": 0.5929191368614591,
+      "grad_norm": 0.151789128780365,
+      "learning_rate": 4.4070881018046e-06,
+      "loss": 0.9079,
+      "step": 81910
+    },
+    {
+      "epoch": 0.5929915235220453,
+      "grad_norm": 0.15422183275222778,
+      "learning_rate": 4.407015715144013e-06,
+      "loss": 0.9028,
+      "step": 81920
+    },
+    {
+      "epoch": 0.5930639101826315,
+      "grad_norm": 0.1466582864522934,
+      "learning_rate": 4.406943328483428e-06,
+      "loss": 0.9025,
+      "step": 81930
+    },
+    {
+      "epoch": 0.5931362968432178,
+      "grad_norm": 0.16983705759048462,
+      "learning_rate": 4.406870941822841e-06,
+      "loss": 0.9048,
+      "step": 81940
+    },
+    {
+      "epoch": 0.593208683503804,
+      "grad_norm": 0.1635872721672058,
+      "learning_rate": 4.406798555162255e-06,
+      "loss": 0.901,
+      "step": 81950
+    },
+    {
+      "epoch": 0.5932810701643901,
+      "grad_norm": 0.17607009410858154,
+      "learning_rate": 4.4067261685016685e-06,
+      "loss": 0.9248,
+      "step": 81960
+    },
+    {
+      "epoch": 0.5933534568249763,
+      "grad_norm": 0.1671140491962433,
+      "learning_rate": 4.406653781841083e-06,
+      "loss": 0.9005,
+      "step": 81970
+    },
+    {
+      "epoch": 0.5934258434855625,
+      "grad_norm": 0.17591434717178345,
+      "learning_rate": 4.406581395180497e-06,
+      "loss": 0.909,
+      "step": 81980
+    },
+    {
+      "epoch": 0.5934982301461487,
+      "grad_norm": 0.17302151024341583,
+      "learning_rate": 4.40650900851991e-06,
+      "loss": 0.9135,
+      "step": 81990
+    },
+    {
+      "epoch": 0.5935706168067348,
+      "grad_norm": 0.169133722782135,
+      "learning_rate": 4.406436621859324e-06,
+      "loss": 0.8986,
+      "step": 82000
+    },
+    {
+      "epoch": 0.593643003467321,
+      "grad_norm": 0.1573852002620697,
+      "learning_rate": 4.406364235198738e-06,
+      "loss": 0.9064,
+      "step": 82010
+    },
+    {
+      "epoch": 0.5937153901279072,
+      "grad_norm": 0.1580667495727539,
+      "learning_rate": 4.406291848538152e-06,
+      "loss": 0.9076,
+      "step": 82020
+    },
+    {
+      "epoch": 0.5937877767884934,
+      "grad_norm": 0.15899869799613953,
+      "learning_rate": 4.4062194618775655e-06,
+      "loss": 0.9142,
+      "step": 82030
+    },
+    {
+      "epoch": 0.5938601634490797,
+      "grad_norm": 0.17980434000492096,
+      "learning_rate": 4.406147075216979e-06,
+      "loss": 0.9041,
+      "step": 82040
+    },
+    {
+      "epoch": 0.5939325501096658,
+      "grad_norm": 0.1649356186389923,
+      "learning_rate": 4.406074688556393e-06,
+      "loss": 0.9299,
+      "step": 82050
+    },
+    {
+      "epoch": 0.594004936770252,
+      "grad_norm": 0.15136809647083282,
+      "learning_rate": 4.406002301895807e-06,
+      "loss": 0.9082,
+      "step": 82060
+    },
+    {
+      "epoch": 0.5940773234308382,
+      "grad_norm": 0.15223738551139832,
+      "learning_rate": 4.405929915235221e-06,
+      "loss": 0.9107,
+      "step": 82070
+    },
+    {
+      "epoch": 0.5941497100914244,
+      "grad_norm": 0.1510484218597412,
+      "learning_rate": 4.4058575285746345e-06,
+      "loss": 0.9013,
+      "step": 82080
+    },
+    {
+      "epoch": 0.5942220967520105,
+      "grad_norm": 0.17146191000938416,
+      "learning_rate": 4.405785141914048e-06,
+      "loss": 0.9101,
+      "step": 82090
+    },
+    {
+      "epoch": 0.5942944834125967,
+      "grad_norm": 0.19103531539440155,
+      "learning_rate": 4.4057127552534626e-06,
+      "loss": 0.9044,
+      "step": 82100
+    },
+    {
+      "epoch": 0.5943668700731829,
+      "grad_norm": 0.1852683573961258,
+      "learning_rate": 4.405640368592876e-06,
+      "loss": 0.905,
+      "step": 82110
+    },
+    {
+      "epoch": 0.5944392567337691,
+      "grad_norm": 0.17115405201911926,
+      "learning_rate": 4.40556798193229e-06,
+      "loss": 0.903,
+      "step": 82120
+    },
+    {
+      "epoch": 0.5945116433943552,
+      "grad_norm": 0.1565488874912262,
+      "learning_rate": 4.405495595271703e-06,
+      "loss": 0.8996,
+      "step": 82130
+    },
+    {
+      "epoch": 0.5945840300549414,
+      "grad_norm": 0.16356854140758514,
+      "learning_rate": 4.405423208611118e-06,
+      "loss": 0.9087,
+      "step": 82140
+    },
+    {
+      "epoch": 0.5946564167155277,
+      "grad_norm": 0.16093464195728302,
+      "learning_rate": 4.4053508219505315e-06,
+      "loss": 0.9107,
+      "step": 82150
+    },
+    {
+      "epoch": 0.5947288033761139,
+      "grad_norm": 0.16113322973251343,
+      "learning_rate": 4.405278435289945e-06,
+      "loss": 0.9077,
+      "step": 82160
+    },
+    {
+      "epoch": 0.5948011900367001,
+      "grad_norm": 0.19098584353923798,
+      "learning_rate": 4.405206048629359e-06,
+      "loss": 0.8965,
+      "step": 82170
+    },
+    {
+      "epoch": 0.5948735766972862,
+      "grad_norm": 0.46058592200279236,
+      "learning_rate": 4.405133661968773e-06,
+      "loss": 0.9037,
+      "step": 82180
+    },
+    {
+      "epoch": 0.5949459633578724,
+      "grad_norm": 0.15246206521987915,
+      "learning_rate": 4.405061275308187e-06,
+      "loss": 0.9013,
+      "step": 82190
+    },
+    {
+      "epoch": 0.5950183500184586,
+      "grad_norm": 0.16424813866615295,
+      "learning_rate": 4.4049888886476e-06,
+      "loss": 0.8992,
+      "step": 82200
+    },
+    {
+      "epoch": 0.5950907366790448,
+      "grad_norm": 0.16102401912212372,
+      "learning_rate": 4.404916501987014e-06,
+      "loss": 0.9074,
+      "step": 82210
+    },
+    {
+      "epoch": 0.595163123339631,
+      "grad_norm": 0.16575241088867188,
+      "learning_rate": 4.4048441153264285e-06,
+      "loss": 0.9125,
+      "step": 82220
+    },
+    {
+      "epoch": 0.5952355100002171,
+      "grad_norm": 0.15276582539081573,
+      "learning_rate": 4.404771728665842e-06,
+      "loss": 0.8948,
+      "step": 82230
+    },
+    {
+      "epoch": 0.5953078966608033,
+      "grad_norm": 0.16686709225177765,
+      "learning_rate": 4.404699342005256e-06,
+      "loss": 0.9082,
+      "step": 82240
+    },
+    {
+      "epoch": 0.5953802833213895,
+      "grad_norm": 0.19125671684741974,
+      "learning_rate": 4.404626955344669e-06,
+      "loss": 0.9021,
+      "step": 82250
+    },
+    {
+      "epoch": 0.5954526699819758,
+      "grad_norm": 0.15619054436683655,
+      "learning_rate": 4.404554568684084e-06,
+      "loss": 0.9076,
+      "step": 82260
+    },
+    {
+      "epoch": 0.595525056642562,
+      "grad_norm": 0.1474914848804474,
+      "learning_rate": 4.404482182023497e-06,
+      "loss": 0.9121,
+      "step": 82270
+    },
+    {
+      "epoch": 0.5955974433031481,
+      "grad_norm": 0.1519433706998825,
+      "learning_rate": 4.404409795362911e-06,
+      "loss": 0.9154,
+      "step": 82280
+    },
+    {
+      "epoch": 0.5956698299637343,
+      "grad_norm": 0.15717659890651703,
+      "learning_rate": 4.404337408702325e-06,
+      "loss": 0.9113,
+      "step": 82290
+    },
+    {
+      "epoch": 0.5957422166243205,
+      "grad_norm": 0.1966000646352768,
+      "learning_rate": 4.404265022041739e-06,
+      "loss": 0.9156,
+      "step": 82300
+    },
+    {
+      "epoch": 0.5958146032849067,
+      "grad_norm": 0.16030791401863098,
+      "learning_rate": 4.404192635381152e-06,
+      "loss": 0.9195,
+      "step": 82310
+    },
+    {
+      "epoch": 0.5958869899454928,
+      "grad_norm": 0.16544552147388458,
+      "learning_rate": 4.4041202487205655e-06,
+      "loss": 0.9117,
+      "step": 82320
+    },
+    {
+      "epoch": 0.595959376606079,
+      "grad_norm": 0.15709573030471802,
+      "learning_rate": 4.40404786205998e-06,
+      "loss": 0.9049,
+      "step": 82330
+    },
+    {
+      "epoch": 0.5960317632666652,
+      "grad_norm": 0.15841509401798248,
+      "learning_rate": 4.403975475399394e-06,
+      "loss": 0.9054,
+      "step": 82340
+    },
+    {
+      "epoch": 0.5961041499272514,
+      "grad_norm": 0.15718714892864227,
+      "learning_rate": 4.403903088738807e-06,
+      "loss": 0.8977,
+      "step": 82350
+    },
+    {
+      "epoch": 0.5961765365878376,
+      "grad_norm": 0.1640692800283432,
+      "learning_rate": 4.403830702078221e-06,
+      "loss": 0.9121,
+      "step": 82360
+    },
+    {
+      "epoch": 0.5962489232484238,
+      "grad_norm": 0.1538037806749344,
+      "learning_rate": 4.403758315417635e-06,
+      "loss": 0.9113,
+      "step": 82370
+    },
+    {
+      "epoch": 0.59632130990901,
+      "grad_norm": 0.18401135504245758,
+      "learning_rate": 4.403685928757049e-06,
+      "loss": 0.8984,
+      "step": 82380
+    },
+    {
+      "epoch": 0.5963936965695962,
+      "grad_norm": 0.1522941142320633,
+      "learning_rate": 4.4036135420964625e-06,
+      "loss": 0.9004,
+      "step": 82390
+    },
+    {
+      "epoch": 0.5964660832301824,
+      "grad_norm": 0.16432693600654602,
+      "learning_rate": 4.403541155435876e-06,
+      "loss": 0.9088,
+      "step": 82400
+    },
+    {
+      "epoch": 0.5965384698907685,
+      "grad_norm": 0.16186141967773438,
+      "learning_rate": 4.403468768775291e-06,
+      "loss": 0.8907,
+      "step": 82410
+    },
+    {
+      "epoch": 0.5966108565513547,
+      "grad_norm": 0.16021370887756348,
+      "learning_rate": 4.403396382114704e-06,
+      "loss": 0.906,
+      "step": 82420
+    },
+    {
+      "epoch": 0.5966832432119409,
+      "grad_norm": 0.1591646671295166,
+      "learning_rate": 4.403323995454118e-06,
+      "loss": 0.9104,
+      "step": 82430
+    },
+    {
+      "epoch": 0.5967556298725271,
+      "grad_norm": 0.16218051314353943,
+      "learning_rate": 4.4032516087935314e-06,
+      "loss": 0.8933,
+      "step": 82440
+    },
+    {
+      "epoch": 0.5968280165331132,
+      "grad_norm": 0.1592613160610199,
+      "learning_rate": 4.403179222132946e-06,
+      "loss": 0.9027,
+      "step": 82450
+    },
+    {
+      "epoch": 0.5969004031936994,
+      "grad_norm": 0.15859758853912354,
+      "learning_rate": 4.4031068354723595e-06,
+      "loss": 0.9131,
+      "step": 82460
+    },
+    {
+      "epoch": 0.5969727898542857,
+      "grad_norm": 0.1566203236579895,
+      "learning_rate": 4.403034448811773e-06,
+      "loss": 0.9032,
+      "step": 82470
+    },
+    {
+      "epoch": 0.5970451765148719,
+      "grad_norm": 0.14999212324619293,
+      "learning_rate": 4.402962062151187e-06,
+      "loss": 0.9117,
+      "step": 82480
+    },
+    {
+      "epoch": 0.597117563175458,
+      "grad_norm": 0.18934468924999237,
+      "learning_rate": 4.402889675490601e-06,
+      "loss": 0.909,
+      "step": 82490
+    },
+    {
+      "epoch": 0.5971899498360442,
+      "grad_norm": 0.15167193114757538,
+      "learning_rate": 4.402817288830015e-06,
+      "loss": 0.8999,
+      "step": 82500
+    },
+    {
+      "epoch": 0.5972623364966304,
+      "grad_norm": 0.16032280027866364,
+      "learning_rate": 4.4027449021694284e-06,
+      "loss": 0.8981,
+      "step": 82510
+    },
+    {
+      "epoch": 0.5973347231572166,
+      "grad_norm": 0.16216394305229187,
+      "learning_rate": 4.402672515508842e-06,
+      "loss": 0.9033,
+      "step": 82520
+    },
+    {
+      "epoch": 0.5974071098178028,
+      "grad_norm": 0.15446747839450836,
+      "learning_rate": 4.4026001288482565e-06,
+      "loss": 0.9062,
+      "step": 82530
+    },
+    {
+      "epoch": 0.5974794964783889,
+      "grad_norm": 0.1542857438325882,
+      "learning_rate": 4.40252774218767e-06,
+      "loss": 0.9097,
+      "step": 82540
+    },
+    {
+      "epoch": 0.5975518831389751,
+      "grad_norm": 0.1710725724697113,
+      "learning_rate": 4.402455355527084e-06,
+      "loss": 0.9131,
+      "step": 82550
+    },
+    {
+      "epoch": 0.5976242697995613,
+      "grad_norm": 0.16666147112846375,
+      "learning_rate": 4.402382968866497e-06,
+      "loss": 0.905,
+      "step": 82560
+    },
+    {
+      "epoch": 0.5976966564601475,
+      "grad_norm": 0.17649833858013153,
+      "learning_rate": 4.402310582205912e-06,
+      "loss": 0.8965,
+      "step": 82570
+    },
+    {
+      "epoch": 0.5977690431207338,
+      "grad_norm": 0.15569385886192322,
+      "learning_rate": 4.4022381955453255e-06,
+      "loss": 0.8984,
+      "step": 82580
+    },
+    {
+      "epoch": 0.5978414297813199,
+      "grad_norm": 0.1647312492132187,
+      "learning_rate": 4.402165808884739e-06,
+      "loss": 0.9114,
+      "step": 82590
+    },
+    {
+      "epoch": 0.5979138164419061,
+      "grad_norm": 0.15604494512081146,
+      "learning_rate": 4.402093422224153e-06,
+      "loss": 0.9011,
+      "step": 82600
+    },
+    {
+      "epoch": 0.5979862031024923,
+      "grad_norm": 0.16003774106502533,
+      "learning_rate": 4.402021035563567e-06,
+      "loss": 0.9036,
+      "step": 82610
+    },
+    {
+      "epoch": 0.5980585897630785,
+      "grad_norm": 0.15925242006778717,
+      "learning_rate": 4.401948648902981e-06,
+      "loss": 0.8946,
+      "step": 82620
+    },
+    {
+      "epoch": 0.5981309764236646,
+      "grad_norm": 0.1530761569738388,
+      "learning_rate": 4.401876262242394e-06,
+      "loss": 0.9071,
+      "step": 82630
+    },
+    {
+      "epoch": 0.5982033630842508,
+      "grad_norm": 0.15942499041557312,
+      "learning_rate": 4.401803875581808e-06,
+      "loss": 0.9162,
+      "step": 82640
+    },
+    {
+      "epoch": 0.598275749744837,
+      "grad_norm": 0.1493762582540512,
+      "learning_rate": 4.4017314889212225e-06,
+      "loss": 0.905,
+      "step": 82650
+    },
+    {
+      "epoch": 0.5983481364054232,
+      "grad_norm": 0.1528092622756958,
+      "learning_rate": 4.401659102260636e-06,
+      "loss": 0.9097,
+      "step": 82660
+    },
+    {
+      "epoch": 0.5984205230660093,
+      "grad_norm": 0.18498297035694122,
+      "learning_rate": 4.40158671560005e-06,
+      "loss": 0.8937,
+      "step": 82670
+    },
+    {
+      "epoch": 0.5984929097265956,
+      "grad_norm": 0.19264182448387146,
+      "learning_rate": 4.401514328939463e-06,
+      "loss": 0.9167,
+      "step": 82680
+    },
+    {
+      "epoch": 0.5985652963871818,
+      "grad_norm": 0.14750641584396362,
+      "learning_rate": 4.401441942278877e-06,
+      "loss": 0.913,
+      "step": 82690
+    },
+    {
+      "epoch": 0.598637683047768,
+      "grad_norm": 0.2046324908733368,
+      "learning_rate": 4.401369555618291e-06,
+      "loss": 0.9094,
+      "step": 82700
+    },
+    {
+      "epoch": 0.5987100697083542,
+      "grad_norm": 0.16712640225887299,
+      "learning_rate": 4.401297168957705e-06,
+      "loss": 0.9057,
+      "step": 82710
+    },
+    {
+      "epoch": 0.5987824563689403,
+      "grad_norm": 0.1629089117050171,
+      "learning_rate": 4.401224782297119e-06,
+      "loss": 0.9194,
+      "step": 82720
+    },
+    {
+      "epoch": 0.5988548430295265,
+      "grad_norm": 0.15542463958263397,
+      "learning_rate": 4.401152395636532e-06,
+      "loss": 0.8874,
+      "step": 82730
+    },
+    {
+      "epoch": 0.5989272296901127,
+      "grad_norm": 0.16220228374004364,
+      "learning_rate": 4.401080008975947e-06,
+      "loss": 0.9143,
+      "step": 82740
+    },
+    {
+      "epoch": 0.5989996163506989,
+      "grad_norm": 0.15400896966457367,
+      "learning_rate": 4.40100762231536e-06,
+      "loss": 0.8985,
+      "step": 82750
+    },
+    {
+      "epoch": 0.599072003011285,
+      "grad_norm": 0.16913191974163055,
+      "learning_rate": 4.400935235654774e-06,
+      "loss": 0.9012,
+      "step": 82760
+    },
+    {
+      "epoch": 0.5991443896718712,
+      "grad_norm": 0.1506996899843216,
+      "learning_rate": 4.4008628489941875e-06,
+      "loss": 0.9109,
+      "step": 82770
+    },
+    {
+      "epoch": 0.5992167763324574,
+      "grad_norm": 0.15028440952301025,
+      "learning_rate": 4.400790462333602e-06,
+      "loss": 0.9002,
+      "step": 82780
+    },
+    {
+      "epoch": 0.5992891629930437,
+      "grad_norm": 0.15182295441627502,
+      "learning_rate": 4.400718075673016e-06,
+      "loss": 0.8906,
+      "step": 82790
+    },
+    {
+      "epoch": 0.5993615496536299,
+      "grad_norm": 0.1721028983592987,
+      "learning_rate": 4.400645689012429e-06,
+      "loss": 0.9059,
+      "step": 82800
+    },
+    {
+      "epoch": 0.599433936314216,
+      "grad_norm": 0.1762334257364273,
+      "learning_rate": 4.400573302351843e-06,
+      "loss": 0.8982,
+      "step": 82810
+    },
+    {
+      "epoch": 0.5995063229748022,
+      "grad_norm": 0.15542656183242798,
+      "learning_rate": 4.400500915691257e-06,
+      "loss": 0.9063,
+      "step": 82820
+    },
+    {
+      "epoch": 0.5995787096353884,
+      "grad_norm": 0.1622530221939087,
+      "learning_rate": 4.400428529030671e-06,
+      "loss": 0.906,
+      "step": 82830
+    },
+    {
+      "epoch": 0.5996510962959746,
+      "grad_norm": 0.19038499891757965,
+      "learning_rate": 4.400356142370084e-06,
+      "loss": 0.9196,
+      "step": 82840
+    },
+    {
+      "epoch": 0.5997234829565607,
+      "grad_norm": 0.15594981610774994,
+      "learning_rate": 4.400283755709498e-06,
+      "loss": 0.8982,
+      "step": 82850
+    },
+    {
+      "epoch": 0.5997958696171469,
+      "grad_norm": 0.1520778387784958,
+      "learning_rate": 4.400211369048912e-06,
+      "loss": 0.912,
+      "step": 82860
+    },
+    {
+      "epoch": 0.5998682562777331,
+      "grad_norm": 0.15902015566825867,
+      "learning_rate": 4.400138982388325e-06,
+      "loss": 0.9152,
+      "step": 82870
+    },
+    {
+      "epoch": 0.5999406429383193,
+      "grad_norm": 0.1633022576570511,
+      "learning_rate": 4.400066595727739e-06,
+      "loss": 0.8891,
+      "step": 82880
+    },
+    {
+      "epoch": 0.6000130295989056,
+      "grad_norm": 0.1755109429359436,
+      "learning_rate": 4.3999942090671535e-06,
+      "loss": 0.906,
+      "step": 82890
+    },
+    {
+      "epoch": 0.6000854162594917,
+      "grad_norm": 0.15575723350048065,
+      "learning_rate": 4.399921822406567e-06,
+      "loss": 0.9105,
+      "step": 82900
+    },
+    {
+      "epoch": 0.6001578029200779,
+      "grad_norm": 0.1682191789150238,
+      "learning_rate": 4.399849435745981e-06,
+      "loss": 0.8904,
+      "step": 82910
+    },
+    {
+      "epoch": 0.6002301895806641,
+      "grad_norm": 0.1746690571308136,
+      "learning_rate": 4.399777049085394e-06,
+      "loss": 0.9284,
+      "step": 82920
+    },
+    {
+      "epoch": 0.6003025762412503,
+      "grad_norm": 0.16141672432422638,
+      "learning_rate": 4.399704662424809e-06,
+      "loss": 0.8947,
+      "step": 82930
+    },
+    {
+      "epoch": 0.6003749629018365,
+      "grad_norm": 0.17700308561325073,
+      "learning_rate": 4.399632275764222e-06,
+      "loss": 0.9184,
+      "step": 82940
+    },
+    {
+      "epoch": 0.6004473495624226,
+      "grad_norm": 0.16985797882080078,
+      "learning_rate": 4.399559889103636e-06,
+      "loss": 0.905,
+      "step": 82950
+    },
+    {
+      "epoch": 0.6005197362230088,
+      "grad_norm": 0.17443716526031494,
+      "learning_rate": 4.39948750244305e-06,
+      "loss": 0.9197,
+      "step": 82960
+    },
+    {
+      "epoch": 0.600592122883595,
+      "grad_norm": 0.1507861465215683,
+      "learning_rate": 4.399415115782464e-06,
+      "loss": 0.9189,
+      "step": 82970
+    },
+    {
+      "epoch": 0.6006645095441812,
+      "grad_norm": 0.16039422154426575,
+      "learning_rate": 4.399342729121878e-06,
+      "loss": 0.9017,
+      "step": 82980
+    },
+    {
+      "epoch": 0.6007368962047673,
+      "grad_norm": 0.15274271368980408,
+      "learning_rate": 4.399270342461291e-06,
+      "loss": 0.9117,
+      "step": 82990
+    },
+    {
+      "epoch": 0.6008092828653536,
+      "grad_norm": 0.18588446080684662,
+      "learning_rate": 4.399197955800705e-06,
+      "loss": 0.9046,
+      "step": 83000
+    },
+    {
+      "epoch": 0.6008816695259398,
+      "grad_norm": 0.15456266701221466,
+      "learning_rate": 4.399125569140119e-06,
+      "loss": 0.8961,
+      "step": 83010
+    },
+    {
+      "epoch": 0.600954056186526,
+      "grad_norm": 0.17130137979984283,
+      "learning_rate": 4.399053182479533e-06,
+      "loss": 0.9042,
+      "step": 83020
+    },
+    {
+      "epoch": 0.6010264428471122,
+      "grad_norm": 0.15653465688228607,
+      "learning_rate": 4.398980795818947e-06,
+      "loss": 0.9129,
+      "step": 83030
+    },
+    {
+      "epoch": 0.6010988295076983,
+      "grad_norm": 0.1832706183195114,
+      "learning_rate": 4.39890840915836e-06,
+      "loss": 0.9146,
+      "step": 83040
+    },
+    {
+      "epoch": 0.6011712161682845,
+      "grad_norm": 0.1502702534198761,
+      "learning_rate": 4.398836022497775e-06,
+      "loss": 0.8918,
+      "step": 83050
+    },
+    {
+      "epoch": 0.6012436028288707,
+      "grad_norm": 0.19595682621002197,
+      "learning_rate": 4.398763635837188e-06,
+      "loss": 0.8992,
+      "step": 83060
+    },
+    {
+      "epoch": 0.6013159894894569,
+      "grad_norm": 0.16274172067642212,
+      "learning_rate": 4.398691249176602e-06,
+      "loss": 0.8992,
+      "step": 83070
+    },
+    {
+      "epoch": 0.601388376150043,
+      "grad_norm": 0.15060241520404816,
+      "learning_rate": 4.398618862516016e-06,
+      "loss": 0.9031,
+      "step": 83080
+    },
+    {
+      "epoch": 0.6014607628106292,
+      "grad_norm": 0.19672121107578278,
+      "learning_rate": 4.39854647585543e-06,
+      "loss": 0.9027,
+      "step": 83090
+    },
+    {
+      "epoch": 0.6015331494712154,
+      "grad_norm": 0.1655958890914917,
+      "learning_rate": 4.398474089194844e-06,
+      "loss": 0.9133,
+      "step": 83100
+    },
+    {
+      "epoch": 0.6016055361318017,
+      "grad_norm": 0.16771474480628967,
+      "learning_rate": 4.398401702534257e-06,
+      "loss": 0.9074,
+      "step": 83110
+    },
+    {
+      "epoch": 0.6016779227923879,
+      "grad_norm": 0.16556835174560547,
+      "learning_rate": 4.398329315873671e-06,
+      "loss": 0.8982,
+      "step": 83120
+    },
+    {
+      "epoch": 0.601750309452974,
+      "grad_norm": 0.15332920849323273,
+      "learning_rate": 4.398256929213085e-06,
+      "loss": 0.8985,
+      "step": 83130
+    },
+    {
+      "epoch": 0.6018226961135602,
+      "grad_norm": 0.1574867218732834,
+      "learning_rate": 4.398184542552499e-06,
+      "loss": 0.9085,
+      "step": 83140
+    },
+    {
+      "epoch": 0.6018950827741464,
+      "grad_norm": 0.17546038329601288,
+      "learning_rate": 4.398112155891913e-06,
+      "loss": 0.9062,
+      "step": 83150
+    },
+    {
+      "epoch": 0.6019674694347326,
+      "grad_norm": 0.1600867062807083,
+      "learning_rate": 4.398039769231326e-06,
+      "loss": 0.914,
+      "step": 83160
+    },
+    {
+      "epoch": 0.6020398560953187,
+      "grad_norm": 0.1494045853614807,
+      "learning_rate": 4.397967382570741e-06,
+      "loss": 0.8953,
+      "step": 83170
+    },
+    {
+      "epoch": 0.6021122427559049,
+      "grad_norm": 0.15468309819698334,
+      "learning_rate": 4.397894995910154e-06,
+      "loss": 0.9284,
+      "step": 83180
+    },
+    {
+      "epoch": 0.6021846294164911,
+      "grad_norm": 0.2398003190755844,
+      "learning_rate": 4.397822609249568e-06,
+      "loss": 0.9161,
+      "step": 83190
+    },
+    {
+      "epoch": 0.6022570160770773,
+      "grad_norm": 0.1596948802471161,
+      "learning_rate": 4.3977502225889815e-06,
+      "loss": 0.8963,
+      "step": 83200
+    },
+    {
+      "epoch": 0.6023294027376636,
+      "grad_norm": 0.16264113783836365,
+      "learning_rate": 4.397677835928396e-06,
+      "loss": 0.9142,
+      "step": 83210
+    },
+    {
+      "epoch": 0.6024017893982497,
+      "grad_norm": 0.1623883843421936,
+      "learning_rate": 4.39760544926781e-06,
+      "loss": 0.8912,
+      "step": 83220
+    },
+    {
+      "epoch": 0.6024741760588359,
+      "grad_norm": 0.16414840519428253,
+      "learning_rate": 4.397533062607223e-06,
+      "loss": 0.9061,
+      "step": 83230
+    },
+    {
+      "epoch": 0.6025465627194221,
+      "grad_norm": 0.16708609461784363,
+      "learning_rate": 4.397460675946637e-06,
+      "loss": 0.9081,
+      "step": 83240
+    },
+    {
+      "epoch": 0.6026189493800083,
+      "grad_norm": 0.15145167708396912,
+      "learning_rate": 4.397388289286051e-06,
+      "loss": 0.8966,
+      "step": 83250
+    },
+    {
+      "epoch": 0.6026913360405944,
+      "grad_norm": 0.31119629740715027,
+      "learning_rate": 4.397315902625465e-06,
+      "loss": 0.9119,
+      "step": 83260
+    },
+    {
+      "epoch": 0.6027637227011806,
+      "grad_norm": 0.16512273252010345,
+      "learning_rate": 4.3972435159648785e-06,
+      "loss": 0.9058,
+      "step": 83270
+    },
+    {
+      "epoch": 0.6028361093617668,
+      "grad_norm": 0.2106596678495407,
+      "learning_rate": 4.397171129304292e-06,
+      "loss": 0.896,
+      "step": 83280
+    },
+    {
+      "epoch": 0.602908496022353,
+      "grad_norm": 0.16792406141757965,
+      "learning_rate": 4.397098742643706e-06,
+      "loss": 0.9028,
+      "step": 83290
+    },
+    {
+      "epoch": 0.6029808826829391,
+      "grad_norm": 0.1586570292711258,
+      "learning_rate": 4.39702635598312e-06,
+      "loss": 0.9083,
+      "step": 83300
+    },
+    {
+      "epoch": 0.6030532693435253,
+      "grad_norm": 0.18892773985862732,
+      "learning_rate": 4.396953969322534e-06,
+      "loss": 0.9072,
+      "step": 83310
+    },
+    {
+      "epoch": 0.6031256560041116,
+      "grad_norm": 0.17914731800556183,
+      "learning_rate": 4.3968815826619475e-06,
+      "loss": 0.8988,
+      "step": 83320
+    },
+    {
+      "epoch": 0.6031980426646978,
+      "grad_norm": 0.15073609352111816,
+      "learning_rate": 4.396809196001361e-06,
+      "loss": 0.9003,
+      "step": 83330
+    },
+    {
+      "epoch": 0.603270429325284,
+      "grad_norm": 0.16799314320087433,
+      "learning_rate": 4.3967368093407755e-06,
+      "loss": 0.9045,
+      "step": 83340
+    },
+    {
+      "epoch": 0.6033428159858701,
+      "grad_norm": 0.16455517709255219,
+      "learning_rate": 4.396664422680189e-06,
+      "loss": 0.9059,
+      "step": 83350
+    },
+    {
+      "epoch": 0.6034152026464563,
+      "grad_norm": 0.22611026465892792,
+      "learning_rate": 4.396592036019603e-06,
+      "loss": 0.9087,
+      "step": 83360
+    },
+    {
+      "epoch": 0.6034875893070425,
+      "grad_norm": 0.15722450613975525,
+      "learning_rate": 4.396519649359016e-06,
+      "loss": 0.8941,
+      "step": 83370
+    },
+    {
+      "epoch": 0.6035599759676287,
+      "grad_norm": 0.1701124906539917,
+      "learning_rate": 4.39644726269843e-06,
+      "loss": 0.9138,
+      "step": 83380
+    },
+    {
+      "epoch": 0.6036323626282148,
+      "grad_norm": 0.1639043539762497,
+      "learning_rate": 4.396374876037844e-06,
+      "loss": 0.91,
+      "step": 83390
+    },
+    {
+      "epoch": 0.603704749288801,
+      "grad_norm": 0.15046794712543488,
+      "learning_rate": 4.396302489377258e-06,
+      "loss": 0.898,
+      "step": 83400
+    },
+    {
+      "epoch": 0.6037771359493872,
+      "grad_norm": 0.1862155944108963,
+      "learning_rate": 4.396230102716672e-06,
+      "loss": 0.9124,
+      "step": 83410
+    },
+    {
+      "epoch": 0.6038495226099735,
+      "grad_norm": 0.16142509877681732,
+      "learning_rate": 4.396157716056085e-06,
+      "loss": 0.8921,
+      "step": 83420
+    },
+    {
+      "epoch": 0.6039219092705597,
+      "grad_norm": 0.14882884919643402,
+      "learning_rate": 4.396085329395499e-06,
+      "loss": 0.9176,
+      "step": 83430
+    },
+    {
+      "epoch": 0.6039942959311458,
+      "grad_norm": 0.1499185711145401,
+      "learning_rate": 4.396012942734913e-06,
+      "loss": 0.9056,
+      "step": 83440
+    },
+    {
+      "epoch": 0.604066682591732,
+      "grad_norm": 0.15857931971549988,
+      "learning_rate": 4.395940556074327e-06,
+      "loss": 0.8983,
+      "step": 83450
+    },
+    {
+      "epoch": 0.6041390692523182,
+      "grad_norm": 0.16343648731708527,
+      "learning_rate": 4.395868169413741e-06,
+      "loss": 0.9089,
+      "step": 83460
+    },
+    {
+      "epoch": 0.6042114559129044,
+      "grad_norm": 0.1648421734571457,
+      "learning_rate": 4.395795782753154e-06,
+      "loss": 0.9001,
+      "step": 83470
+    },
+    {
+      "epoch": 0.6042838425734905,
+      "grad_norm": 0.15347935259342194,
+      "learning_rate": 4.395723396092568e-06,
+      "loss": 0.9173,
+      "step": 83480
+    },
+    {
+      "epoch": 0.6043562292340767,
+      "grad_norm": 0.1825142502784729,
+      "learning_rate": 4.395651009431982e-06,
+      "loss": 0.8914,
+      "step": 83490
+    },
+    {
+      "epoch": 0.6044286158946629,
+      "grad_norm": 0.17557092010974884,
+      "learning_rate": 4.395578622771396e-06,
+      "loss": 0.8938,
+      "step": 83500
+    },
+    {
+      "epoch": 0.6045010025552491,
+      "grad_norm": 0.15079079568386078,
+      "learning_rate": 4.3955062361108095e-06,
+      "loss": 0.8986,
+      "step": 83510
+    },
+    {
+      "epoch": 0.6045733892158353,
+      "grad_norm": 0.16225622594356537,
+      "learning_rate": 4.395433849450223e-06,
+      "loss": 0.8948,
+      "step": 83520
+    },
+    {
+      "epoch": 0.6046457758764215,
+      "grad_norm": 0.14633281528949738,
+      "learning_rate": 4.395361462789638e-06,
+      "loss": 0.9021,
+      "step": 83530
+    },
+    {
+      "epoch": 0.6047181625370077,
+      "grad_norm": 0.15504805743694305,
+      "learning_rate": 4.395289076129051e-06,
+      "loss": 0.9164,
+      "step": 83540
+    },
+    {
+      "epoch": 0.6047905491975939,
+      "grad_norm": 0.1608634889125824,
+      "learning_rate": 4.395216689468465e-06,
+      "loss": 0.9065,
+      "step": 83550
+    },
+    {
+      "epoch": 0.6048629358581801,
+      "grad_norm": 0.15265637636184692,
+      "learning_rate": 4.3951443028078785e-06,
+      "loss": 0.9109,
+      "step": 83560
+    },
+    {
+      "epoch": 0.6049353225187662,
+      "grad_norm": 0.16638517379760742,
+      "learning_rate": 4.395071916147293e-06,
+      "loss": 0.8959,
+      "step": 83570
+    },
+    {
+      "epoch": 0.6050077091793524,
+      "grad_norm": 0.15558472275733948,
+      "learning_rate": 4.3949995294867066e-06,
+      "loss": 0.8929,
+      "step": 83580
+    },
+    {
+      "epoch": 0.6050800958399386,
+      "grad_norm": 0.1535881906747818,
+      "learning_rate": 4.39492714282612e-06,
+      "loss": 0.9172,
+      "step": 83590
+    },
+    {
+      "epoch": 0.6051524825005248,
+      "grad_norm": 0.14769315719604492,
+      "learning_rate": 4.394854756165534e-06,
+      "loss": 0.9152,
+      "step": 83600
+    },
+    {
+      "epoch": 0.605224869161111,
+      "grad_norm": 0.1421438306570053,
+      "learning_rate": 4.394782369504948e-06,
+      "loss": 0.91,
+      "step": 83610
+    },
+    {
+      "epoch": 0.6052972558216971,
+      "grad_norm": 0.18984383344650269,
+      "learning_rate": 4.394709982844362e-06,
+      "loss": 0.9114,
+      "step": 83620
+    },
+    {
+      "epoch": 0.6053696424822833,
+      "grad_norm": 0.15073958039283752,
+      "learning_rate": 4.3946375961837755e-06,
+      "loss": 0.9094,
+      "step": 83630
+    },
+    {
+      "epoch": 0.6054420291428696,
+      "grad_norm": 0.15686622262001038,
+      "learning_rate": 4.394565209523189e-06,
+      "loss": 0.9015,
+      "step": 83640
+    },
+    {
+      "epoch": 0.6055144158034558,
+      "grad_norm": 0.15890394151210785,
+      "learning_rate": 4.3944928228626036e-06,
+      "loss": 0.9264,
+      "step": 83650
+    },
+    {
+      "epoch": 0.605586802464042,
+      "grad_norm": 0.15338025987148285,
+      "learning_rate": 4.394420436202017e-06,
+      "loss": 0.8971,
+      "step": 83660
+    },
+    {
+      "epoch": 0.6056591891246281,
+      "grad_norm": 0.1526980698108673,
+      "learning_rate": 4.394348049541431e-06,
+      "loss": 0.9015,
+      "step": 83670
+    },
+    {
+      "epoch": 0.6057315757852143,
+      "grad_norm": 0.15152287483215332,
+      "learning_rate": 4.394275662880844e-06,
+      "loss": 0.8988,
+      "step": 83680
+    },
+    {
+      "epoch": 0.6058039624458005,
+      "grad_norm": 0.15338623523712158,
+      "learning_rate": 4.394203276220259e-06,
+      "loss": 0.9021,
+      "step": 83690
+    },
+    {
+      "epoch": 0.6058763491063867,
+      "grad_norm": 0.1615828573703766,
+      "learning_rate": 4.3941308895596725e-06,
+      "loss": 0.9109,
+      "step": 83700
+    },
+    {
+      "epoch": 0.6059487357669728,
+      "grad_norm": 0.16938403248786926,
+      "learning_rate": 4.394058502899086e-06,
+      "loss": 0.8999,
+      "step": 83710
+    },
+    {
+      "epoch": 0.606021122427559,
+      "grad_norm": 0.154579296708107,
+      "learning_rate": 4.3939861162385e-06,
+      "loss": 0.8897,
+      "step": 83720
+    },
+    {
+      "epoch": 0.6060935090881452,
+      "grad_norm": 0.1727149486541748,
+      "learning_rate": 4.393913729577914e-06,
+      "loss": 0.9162,
+      "step": 83730
+    },
+    {
+      "epoch": 0.6061658957487315,
+      "grad_norm": 0.16071633994579315,
+      "learning_rate": 4.393841342917328e-06,
+      "loss": 0.9169,
+      "step": 83740
+    },
+    {
+      "epoch": 0.6062382824093177,
+      "grad_norm": 0.17092163860797882,
+      "learning_rate": 4.393768956256741e-06,
+      "loss": 0.9072,
+      "step": 83750
+    },
+    {
+      "epoch": 0.6063106690699038,
+      "grad_norm": 0.15256325900554657,
+      "learning_rate": 4.393696569596155e-06,
+      "loss": 0.8937,
+      "step": 83760
+    },
+    {
+      "epoch": 0.60638305573049,
+      "grad_norm": 0.16219750046730042,
+      "learning_rate": 4.3936241829355695e-06,
+      "loss": 0.9149,
+      "step": 83770
+    },
+    {
+      "epoch": 0.6064554423910762,
+      "grad_norm": 0.15577305853366852,
+      "learning_rate": 4.393551796274983e-06,
+      "loss": 0.9136,
+      "step": 83780
+    },
+    {
+      "epoch": 0.6065278290516624,
+      "grad_norm": 0.20615074038505554,
+      "learning_rate": 4.393479409614397e-06,
+      "loss": 0.9098,
+      "step": 83790
+    },
+    {
+      "epoch": 0.6066002157122485,
+      "grad_norm": 0.1500859558582306,
+      "learning_rate": 4.39340702295381e-06,
+      "loss": 0.908,
+      "step": 83800
+    },
+    {
+      "epoch": 0.6066726023728347,
+      "grad_norm": 0.1555171012878418,
+      "learning_rate": 4.393334636293225e-06,
+      "loss": 0.8914,
+      "step": 83810
+    },
+    {
+      "epoch": 0.6067449890334209,
+      "grad_norm": 0.16960106790065765,
+      "learning_rate": 4.3932622496326384e-06,
+      "loss": 0.9178,
+      "step": 83820
+    },
+    {
+      "epoch": 0.6068173756940071,
+      "grad_norm": 0.1492023915052414,
+      "learning_rate": 4.393189862972052e-06,
+      "loss": 0.9121,
+      "step": 83830
+    },
+    {
+      "epoch": 0.6068897623545932,
+      "grad_norm": 0.16274462640285492,
+      "learning_rate": 4.393117476311466e-06,
+      "loss": 0.9152,
+      "step": 83840
+    },
+    {
+      "epoch": 0.6069621490151795,
+      "grad_norm": 0.1625206470489502,
+      "learning_rate": 4.39304508965088e-06,
+      "loss": 0.9116,
+      "step": 83850
+    },
+    {
+      "epoch": 0.6070345356757657,
+      "grad_norm": 0.15123465657234192,
+      "learning_rate": 4.392972702990294e-06,
+      "loss": 0.9061,
+      "step": 83860
+    },
+    {
+      "epoch": 0.6071069223363519,
+      "grad_norm": 0.17103265225887299,
+      "learning_rate": 4.392900316329707e-06,
+      "loss": 0.9021,
+      "step": 83870
+    },
+    {
+      "epoch": 0.6071793089969381,
+      "grad_norm": 0.19635428488254547,
+      "learning_rate": 4.392827929669121e-06,
+      "loss": 0.8978,
+      "step": 83880
+    },
+    {
+      "epoch": 0.6072516956575242,
+      "grad_norm": 0.14300402998924255,
+      "learning_rate": 4.3927555430085354e-06,
+      "loss": 0.9072,
+      "step": 83890
+    },
+    {
+      "epoch": 0.6073240823181104,
+      "grad_norm": 0.1507839858531952,
+      "learning_rate": 4.392683156347948e-06,
+      "loss": 0.9102,
+      "step": 83900
+    },
+    {
+      "epoch": 0.6073964689786966,
+      "grad_norm": 0.16211600601673126,
+      "learning_rate": 4.392610769687362e-06,
+      "loss": 0.9038,
+      "step": 83910
+    },
+    {
+      "epoch": 0.6074688556392828,
+      "grad_norm": 0.1655724048614502,
+      "learning_rate": 4.392538383026776e-06,
+      "loss": 0.9158,
+      "step": 83920
+    },
+    {
+      "epoch": 0.6075412422998689,
+      "grad_norm": 0.16653649508953094,
+      "learning_rate": 4.39246599636619e-06,
+      "loss": 0.8987,
+      "step": 83930
+    },
+    {
+      "epoch": 0.6076136289604551,
+      "grad_norm": 0.1796729415655136,
+      "learning_rate": 4.3923936097056035e-06,
+      "loss": 0.9065,
+      "step": 83940
+    },
+    {
+      "epoch": 0.6076860156210414,
+      "grad_norm": 0.16883957386016846,
+      "learning_rate": 4.392321223045017e-06,
+      "loss": 0.9116,
+      "step": 83950
+    },
+    {
+      "epoch": 0.6077584022816276,
+      "grad_norm": 0.18032489717006683,
+      "learning_rate": 4.392248836384432e-06,
+      "loss": 0.8919,
+      "step": 83960
+    },
+    {
+      "epoch": 0.6078307889422138,
+      "grad_norm": 0.18664510548114777,
+      "learning_rate": 4.392176449723845e-06,
+      "loss": 0.9099,
+      "step": 83970
+    },
+    {
+      "epoch": 0.6079031756027999,
+      "grad_norm": 0.16412018239498138,
+      "learning_rate": 4.392104063063259e-06,
+      "loss": 0.902,
+      "step": 83980
+    },
+    {
+      "epoch": 0.6079755622633861,
+      "grad_norm": 0.17010684311389923,
+      "learning_rate": 4.3920316764026724e-06,
+      "loss": 0.9389,
+      "step": 83990
+    },
+    {
+      "epoch": 0.6080479489239723,
+      "grad_norm": 0.150782972574234,
+      "learning_rate": 4.391959289742087e-06,
+      "loss": 0.8912,
+      "step": 84000
+    },
+    {
+      "epoch": 0.6081203355845585,
+      "grad_norm": 0.17593304812908173,
+      "learning_rate": 4.3918869030815005e-06,
+      "loss": 0.8924,
+      "step": 84010
+    },
+    {
+      "epoch": 0.6081927222451446,
+      "grad_norm": 0.1518871784210205,
+      "learning_rate": 4.391814516420914e-06,
+      "loss": 0.9028,
+      "step": 84020
+    },
+    {
+      "epoch": 0.6082651089057308,
+      "grad_norm": 0.1600775122642517,
+      "learning_rate": 4.391742129760328e-06,
+      "loss": 0.9032,
+      "step": 84030
+    },
+    {
+      "epoch": 0.608337495566317,
+      "grad_norm": 0.1590421050786972,
+      "learning_rate": 4.391669743099742e-06,
+      "loss": 0.8984,
+      "step": 84040
+    },
+    {
+      "epoch": 0.6084098822269032,
+      "grad_norm": 0.17040304839611053,
+      "learning_rate": 4.391597356439156e-06,
+      "loss": 0.8969,
+      "step": 84050
+    },
+    {
+      "epoch": 0.6084822688874895,
+      "grad_norm": 0.1723751425743103,
+      "learning_rate": 4.3915249697785694e-06,
+      "loss": 0.8904,
+      "step": 84060
+    },
+    {
+      "epoch": 0.6085546555480756,
+      "grad_norm": 0.14594772458076477,
+      "learning_rate": 4.391452583117983e-06,
+      "loss": 0.9142,
+      "step": 84070
+    },
+    {
+      "epoch": 0.6086270422086618,
+      "grad_norm": 0.16909512877464294,
+      "learning_rate": 4.391380196457397e-06,
+      "loss": 0.8943,
+      "step": 84080
+    },
+    {
+      "epoch": 0.608699428869248,
+      "grad_norm": 0.1714186817407608,
+      "learning_rate": 4.391307809796811e-06,
+      "loss": 0.9039,
+      "step": 84090
+    },
+    {
+      "epoch": 0.6087718155298342,
+      "grad_norm": 0.16077734529972076,
+      "learning_rate": 4.391235423136225e-06,
+      "loss": 0.912,
+      "step": 84100
+    },
+    {
+      "epoch": 0.6088442021904203,
+      "grad_norm": 0.1786806732416153,
+      "learning_rate": 4.391163036475638e-06,
+      "loss": 0.915,
+      "step": 84110
+    },
+    {
+      "epoch": 0.6089165888510065,
+      "grad_norm": 0.15787489712238312,
+      "learning_rate": 4.391090649815052e-06,
+      "loss": 0.9037,
+      "step": 84120
+    },
+    {
+      "epoch": 0.6089889755115927,
+      "grad_norm": 0.14963692426681519,
+      "learning_rate": 4.3910182631544665e-06,
+      "loss": 0.8954,
+      "step": 84130
+    },
+    {
+      "epoch": 0.6090613621721789,
+      "grad_norm": 0.15370801091194153,
+      "learning_rate": 4.39094587649388e-06,
+      "loss": 0.9128,
+      "step": 84140
+    },
+    {
+      "epoch": 0.609133748832765,
+      "grad_norm": 0.16809116303920746,
+      "learning_rate": 4.390873489833294e-06,
+      "loss": 0.9006,
+      "step": 84150
+    },
+    {
+      "epoch": 0.6092061354933512,
+      "grad_norm": 0.16092905402183533,
+      "learning_rate": 4.390801103172707e-06,
+      "loss": 0.8976,
+      "step": 84160
+    },
+    {
+      "epoch": 0.6092785221539375,
+      "grad_norm": 0.1607845276594162,
+      "learning_rate": 4.390728716512122e-06,
+      "loss": 0.9273,
+      "step": 84170
+    },
+    {
+      "epoch": 0.6093509088145237,
+      "grad_norm": 0.15682317316532135,
+      "learning_rate": 4.390656329851535e-06,
+      "loss": 0.9062,
+      "step": 84180
+    },
+    {
+      "epoch": 0.6094232954751099,
+      "grad_norm": 0.154841810464859,
+      "learning_rate": 4.390583943190949e-06,
+      "loss": 0.9231,
+      "step": 84190
+    },
+    {
+      "epoch": 0.609495682135696,
+      "grad_norm": 0.16527412831783295,
+      "learning_rate": 4.390511556530363e-06,
+      "loss": 0.9122,
+      "step": 84200
+    },
+    {
+      "epoch": 0.6095680687962822,
+      "grad_norm": 0.15766416490077972,
+      "learning_rate": 4.390439169869777e-06,
+      "loss": 0.8958,
+      "step": 84210
+    },
+    {
+      "epoch": 0.6096404554568684,
+      "grad_norm": 0.17075148224830627,
+      "learning_rate": 4.390366783209191e-06,
+      "loss": 0.9055,
+      "step": 84220
+    },
+    {
+      "epoch": 0.6097128421174546,
+      "grad_norm": 0.15440765023231506,
+      "learning_rate": 4.390294396548604e-06,
+      "loss": 0.9209,
+      "step": 84230
+    },
+    {
+      "epoch": 0.6097852287780408,
+      "grad_norm": 0.1687108278274536,
+      "learning_rate": 4.390222009888018e-06,
+      "loss": 0.9068,
+      "step": 84240
+    },
+    {
+      "epoch": 0.6098576154386269,
+      "grad_norm": 0.15964575111865997,
+      "learning_rate": 4.390149623227432e-06,
+      "loss": 0.9099,
+      "step": 84250
+    },
+    {
+      "epoch": 0.6099300020992131,
+      "grad_norm": 0.15133626759052277,
+      "learning_rate": 4.390077236566846e-06,
+      "loss": 0.912,
+      "step": 84260
+    },
+    {
+      "epoch": 0.6100023887597994,
+      "grad_norm": 0.16388767957687378,
+      "learning_rate": 4.39000484990626e-06,
+      "loss": 0.9091,
+      "step": 84270
+    },
+    {
+      "epoch": 0.6100747754203856,
+      "grad_norm": 0.16977226734161377,
+      "learning_rate": 4.389932463245673e-06,
+      "loss": 0.9041,
+      "step": 84280
+    },
+    {
+      "epoch": 0.6101471620809717,
+      "grad_norm": 0.2612716555595398,
+      "learning_rate": 4.389860076585088e-06,
+      "loss": 0.9024,
+      "step": 84290
+    },
+    {
+      "epoch": 0.6102195487415579,
+      "grad_norm": 0.17897847294807434,
+      "learning_rate": 4.389787689924501e-06,
+      "loss": 0.898,
+      "step": 84300
+    },
+    {
+      "epoch": 0.6102919354021441,
+      "grad_norm": 0.16913948953151703,
+      "learning_rate": 4.389715303263915e-06,
+      "loss": 0.9011,
+      "step": 84310
+    },
+    {
+      "epoch": 0.6103643220627303,
+      "grad_norm": 0.1693422794342041,
+      "learning_rate": 4.3896429166033286e-06,
+      "loss": 0.91,
+      "step": 84320
+    },
+    {
+      "epoch": 0.6104367087233165,
+      "grad_norm": 0.14860454201698303,
+      "learning_rate": 4.389570529942743e-06,
+      "loss": 0.9087,
+      "step": 84330
+    },
+    {
+      "epoch": 0.6105090953839026,
+      "grad_norm": 0.17605887353420258,
+      "learning_rate": 4.389498143282157e-06,
+      "loss": 0.8988,
+      "step": 84340
+    },
+    {
+      "epoch": 0.6105814820444888,
+      "grad_norm": 0.2861873507499695,
+      "learning_rate": 4.38942575662157e-06,
+      "loss": 0.8955,
+      "step": 84350
+    },
+    {
+      "epoch": 0.610653868705075,
+      "grad_norm": 0.1683700680732727,
+      "learning_rate": 4.389353369960984e-06,
+      "loss": 0.8973,
+      "step": 84360
+    },
+    {
+      "epoch": 0.6107262553656612,
+      "grad_norm": 0.15339066088199615,
+      "learning_rate": 4.389280983300398e-06,
+      "loss": 0.9137,
+      "step": 84370
+    },
+    {
+      "epoch": 0.6107986420262475,
+      "grad_norm": 0.15601833164691925,
+      "learning_rate": 4.389208596639812e-06,
+      "loss": 0.9166,
+      "step": 84380
+    },
+    {
+      "epoch": 0.6108710286868336,
+      "grad_norm": 0.16695159673690796,
+      "learning_rate": 4.3891362099792256e-06,
+      "loss": 0.902,
+      "step": 84390
+    },
+    {
+      "epoch": 0.6109434153474198,
+      "grad_norm": 0.1545722484588623,
+      "learning_rate": 4.389063823318639e-06,
+      "loss": 0.9112,
+      "step": 84400
+    },
+    {
+      "epoch": 0.611015802008006,
+      "grad_norm": 0.17581118643283844,
+      "learning_rate": 4.388991436658054e-06,
+      "loss": 0.9104,
+      "step": 84410
+    },
+    {
+      "epoch": 0.6110881886685922,
+      "grad_norm": 0.16836950182914734,
+      "learning_rate": 4.388919049997467e-06,
+      "loss": 0.9007,
+      "step": 84420
+    },
+    {
+      "epoch": 0.6111605753291783,
+      "grad_norm": 0.1791825294494629,
+      "learning_rate": 4.38884666333688e-06,
+      "loss": 0.8978,
+      "step": 84430
+    },
+    {
+      "epoch": 0.6112329619897645,
+      "grad_norm": 0.17053841054439545,
+      "learning_rate": 4.3887742766762945e-06,
+      "loss": 0.9075,
+      "step": 84440
+    },
+    {
+      "epoch": 0.6113053486503507,
+      "grad_norm": 0.15153911709785461,
+      "learning_rate": 4.388701890015708e-06,
+      "loss": 0.9048,
+      "step": 84450
+    },
+    {
+      "epoch": 0.6113777353109369,
+      "grad_norm": 0.15731872618198395,
+      "learning_rate": 4.388629503355122e-06,
+      "loss": 0.9043,
+      "step": 84460
+    },
+    {
+      "epoch": 0.611450121971523,
+      "grad_norm": 0.15405894815921783,
+      "learning_rate": 4.388557116694535e-06,
+      "loss": 0.9059,
+      "step": 84470
+    },
+    {
+      "epoch": 0.6115225086321093,
+      "grad_norm": 0.15265247225761414,
+      "learning_rate": 4.38848473003395e-06,
+      "loss": 0.903,
+      "step": 84480
+    },
+    {
+      "epoch": 0.6115948952926955,
+      "grad_norm": 0.16418184340000153,
+      "learning_rate": 4.388412343373363e-06,
+      "loss": 0.898,
+      "step": 84490
+    },
+    {
+      "epoch": 0.6116672819532817,
+      "grad_norm": 0.16227209568023682,
+      "learning_rate": 4.388339956712777e-06,
+      "loss": 0.9145,
+      "step": 84500
+    },
+    {
+      "epoch": 0.6117396686138679,
+      "grad_norm": 0.20757746696472168,
+      "learning_rate": 4.388267570052191e-06,
+      "loss": 0.9047,
+      "step": 84510
+    },
+    {
+      "epoch": 0.611812055274454,
+      "grad_norm": 0.1717897206544876,
+      "learning_rate": 4.388195183391605e-06,
+      "loss": 0.8976,
+      "step": 84520
+    },
+    {
+      "epoch": 0.6118844419350402,
+      "grad_norm": 0.1659894585609436,
+      "learning_rate": 4.388122796731019e-06,
+      "loss": 0.9076,
+      "step": 84530
+    },
+    {
+      "epoch": 0.6119568285956264,
+      "grad_norm": 0.15432026982307434,
+      "learning_rate": 4.388050410070432e-06,
+      "loss": 0.9111,
+      "step": 84540
+    },
+    {
+      "epoch": 0.6120292152562126,
+      "grad_norm": 0.15473110973834991,
+      "learning_rate": 4.387978023409846e-06,
+      "loss": 0.9041,
+      "step": 84550
+    },
+    {
+      "epoch": 0.6121016019167987,
+      "grad_norm": 0.18219618499279022,
+      "learning_rate": 4.3879056367492604e-06,
+      "loss": 0.8956,
+      "step": 84560
+    },
+    {
+      "epoch": 0.6121739885773849,
+      "grad_norm": 0.15730948746204376,
+      "learning_rate": 4.387833250088674e-06,
+      "loss": 0.9085,
+      "step": 84570
+    },
+    {
+      "epoch": 0.6122463752379711,
+      "grad_norm": 0.14876548945903778,
+      "learning_rate": 4.387760863428088e-06,
+      "loss": 0.9082,
+      "step": 84580
+    },
+    {
+      "epoch": 0.6123187618985574,
+      "grad_norm": 0.1581803560256958,
+      "learning_rate": 4.387688476767501e-06,
+      "loss": 0.9161,
+      "step": 84590
+    },
+    {
+      "epoch": 0.6123911485591436,
+      "grad_norm": 0.1603010892868042,
+      "learning_rate": 4.387616090106916e-06,
+      "loss": 0.9022,
+      "step": 84600
+    },
+    {
+      "epoch": 0.6124635352197297,
+      "grad_norm": 0.17957136034965515,
+      "learning_rate": 4.387543703446329e-06,
+      "loss": 0.9077,
+      "step": 84610
+    },
+    {
+      "epoch": 0.6125359218803159,
+      "grad_norm": 0.1617906540632248,
+      "learning_rate": 4.387471316785743e-06,
+      "loss": 0.9247,
+      "step": 84620
+    },
+    {
+      "epoch": 0.6126083085409021,
+      "grad_norm": 0.1665765643119812,
+      "learning_rate": 4.387398930125157e-06,
+      "loss": 0.9073,
+      "step": 84630
+    },
+    {
+      "epoch": 0.6126806952014883,
+      "grad_norm": 0.1667291522026062,
+      "learning_rate": 4.387326543464571e-06,
+      "loss": 0.8995,
+      "step": 84640
+    },
+    {
+      "epoch": 0.6127530818620744,
+      "grad_norm": 0.1612149178981781,
+      "learning_rate": 4.387254156803985e-06,
+      "loss": 0.9066,
+      "step": 84650
+    },
+    {
+      "epoch": 0.6128254685226606,
+      "grad_norm": 0.15186448395252228,
+      "learning_rate": 4.387181770143398e-06,
+      "loss": 0.9118,
+      "step": 84660
+    },
+    {
+      "epoch": 0.6128978551832468,
+      "grad_norm": 0.15720537304878235,
+      "learning_rate": 4.387109383482812e-06,
+      "loss": 0.8964,
+      "step": 84670
+    },
+    {
+      "epoch": 0.612970241843833,
+      "grad_norm": 0.1654808670282364,
+      "learning_rate": 4.387036996822226e-06,
+      "loss": 0.9022,
+      "step": 84680
+    },
+    {
+      "epoch": 0.6130426285044192,
+      "grad_norm": 0.15336300432682037,
+      "learning_rate": 4.38696461016164e-06,
+      "loss": 0.9128,
+      "step": 84690
+    },
+    {
+      "epoch": 0.6131150151650054,
+      "grad_norm": 0.15341922640800476,
+      "learning_rate": 4.386892223501054e-06,
+      "loss": 0.9072,
+      "step": 84700
+    },
+    {
+      "epoch": 0.6131874018255916,
+      "grad_norm": 0.15757760405540466,
+      "learning_rate": 4.386819836840467e-06,
+      "loss": 0.895,
+      "step": 84710
+    },
+    {
+      "epoch": 0.6132597884861778,
+      "grad_norm": 0.1582041084766388,
+      "learning_rate": 4.386747450179881e-06,
+      "loss": 0.9047,
+      "step": 84720
+    },
+    {
+      "epoch": 0.613332175146764,
+      "grad_norm": 0.19254916906356812,
+      "learning_rate": 4.386675063519295e-06,
+      "loss": 0.8918,
+      "step": 84730
+    },
+    {
+      "epoch": 0.6134045618073501,
+      "grad_norm": 0.14358896017074585,
+      "learning_rate": 4.386602676858709e-06,
+      "loss": 0.8909,
+      "step": 84740
+    },
+    {
+      "epoch": 0.6134769484679363,
+      "grad_norm": 0.15055955946445465,
+      "learning_rate": 4.3865302901981225e-06,
+      "loss": 0.9093,
+      "step": 84750
+    },
+    {
+      "epoch": 0.6135493351285225,
+      "grad_norm": 0.1698206514120102,
+      "learning_rate": 4.386457903537536e-06,
+      "loss": 0.8958,
+      "step": 84760
+    },
+    {
+      "epoch": 0.6136217217891087,
+      "grad_norm": 0.1727665215730667,
+      "learning_rate": 4.386385516876951e-06,
+      "loss": 0.9116,
+      "step": 84770
+    },
+    {
+      "epoch": 0.6136941084496949,
+      "grad_norm": 0.1568523496389389,
+      "learning_rate": 4.386313130216364e-06,
+      "loss": 0.9059,
+      "step": 84780
+    },
+    {
+      "epoch": 0.613766495110281,
+      "grad_norm": 0.17023316025733948,
+      "learning_rate": 4.386240743555778e-06,
+      "loss": 0.9019,
+      "step": 84790
+    },
+    {
+      "epoch": 0.6138388817708673,
+      "grad_norm": 0.15724539756774902,
+      "learning_rate": 4.3861683568951914e-06,
+      "loss": 0.9167,
+      "step": 84800
+    },
+    {
+      "epoch": 0.6139112684314535,
+      "grad_norm": 0.17113758623600006,
+      "learning_rate": 4.386095970234606e-06,
+      "loss": 0.8977,
+      "step": 84810
+    },
+    {
+      "epoch": 0.6139836550920397,
+      "grad_norm": 0.166314959526062,
+      "learning_rate": 4.3860235835740195e-06,
+      "loss": 0.9104,
+      "step": 84820
+    },
+    {
+      "epoch": 0.6140560417526258,
+      "grad_norm": 0.1817166805267334,
+      "learning_rate": 4.385951196913433e-06,
+      "loss": 0.902,
+      "step": 84830
+    },
+    {
+      "epoch": 0.614128428413212,
+      "grad_norm": 0.1507468968629837,
+      "learning_rate": 4.385878810252847e-06,
+      "loss": 0.8974,
+      "step": 84840
+    },
+    {
+      "epoch": 0.6142008150737982,
+      "grad_norm": 0.15710368752479553,
+      "learning_rate": 4.385806423592261e-06,
+      "loss": 0.9036,
+      "step": 84850
+    },
+    {
+      "epoch": 0.6142732017343844,
+      "grad_norm": 0.14728538691997528,
+      "learning_rate": 4.385734036931675e-06,
+      "loss": 0.9031,
+      "step": 84860
+    },
+    {
+      "epoch": 0.6143455883949706,
+      "grad_norm": 0.15462151169776917,
+      "learning_rate": 4.3856616502710885e-06,
+      "loss": 0.9021,
+      "step": 84870
+    },
+    {
+      "epoch": 0.6144179750555567,
+      "grad_norm": 0.1638130396604538,
+      "learning_rate": 4.385589263610502e-06,
+      "loss": 0.92,
+      "step": 84880
+    },
+    {
+      "epoch": 0.6144903617161429,
+      "grad_norm": 0.1549723744392395,
+      "learning_rate": 4.3855168769499165e-06,
+      "loss": 0.9028,
+      "step": 84890
+    },
+    {
+      "epoch": 0.6145627483767291,
+      "grad_norm": 0.15805752575397491,
+      "learning_rate": 4.38544449028933e-06,
+      "loss": 0.899,
+      "step": 84900
+    },
+    {
+      "epoch": 0.6146351350373154,
+      "grad_norm": 0.17812702059745789,
+      "learning_rate": 4.385372103628744e-06,
+      "loss": 0.9045,
+      "step": 84910
+    },
+    {
+      "epoch": 0.6147075216979015,
+      "grad_norm": 0.15838688611984253,
+      "learning_rate": 4.385299716968157e-06,
+      "loss": 0.9154,
+      "step": 84920
+    },
+    {
+      "epoch": 0.6147799083584877,
+      "grad_norm": 0.16311374306678772,
+      "learning_rate": 4.385227330307572e-06,
+      "loss": 0.8976,
+      "step": 84930
+    },
+    {
+      "epoch": 0.6148522950190739,
+      "grad_norm": 0.15058547258377075,
+      "learning_rate": 4.3851549436469855e-06,
+      "loss": 0.8918,
+      "step": 84940
+    },
+    {
+      "epoch": 0.6149246816796601,
+      "grad_norm": 0.15357309579849243,
+      "learning_rate": 4.385082556986399e-06,
+      "loss": 0.9086,
+      "step": 84950
+    },
+    {
+      "epoch": 0.6149970683402463,
+      "grad_norm": 0.15655238926410675,
+      "learning_rate": 4.385010170325813e-06,
+      "loss": 0.9006,
+      "step": 84960
+    },
+    {
+      "epoch": 0.6150694550008324,
+      "grad_norm": 0.17502596974372864,
+      "learning_rate": 4.384937783665226e-06,
+      "loss": 0.8993,
+      "step": 84970
+    },
+    {
+      "epoch": 0.6151418416614186,
+      "grad_norm": 0.1721956878900528,
+      "learning_rate": 4.38486539700464e-06,
+      "loss": 0.8934,
+      "step": 84980
+    },
+    {
+      "epoch": 0.6152142283220048,
+      "grad_norm": 0.1560133844614029,
+      "learning_rate": 4.3847930103440535e-06,
+      "loss": 0.9097,
+      "step": 84990
+    },
+    {
+      "epoch": 0.615286614982591,
+      "grad_norm": 0.15889939665794373,
+      "learning_rate": 4.384720623683468e-06,
+      "loss": 0.8942,
+      "step": 85000
+    },
+    {
+      "epoch": 0.6153590016431773,
+      "grad_norm": 0.16215567290782928,
+      "learning_rate": 4.384648237022882e-06,
+      "loss": 0.9017,
+      "step": 85010
+    },
+    {
+      "epoch": 0.6154313883037634,
+      "grad_norm": 0.16176781058311462,
+      "learning_rate": 4.384575850362295e-06,
+      "loss": 0.8989,
+      "step": 85020
+    },
+    {
+      "epoch": 0.6155037749643496,
+      "grad_norm": 0.1544492542743683,
+      "learning_rate": 4.384503463701709e-06,
+      "loss": 0.9044,
+      "step": 85030
+    },
+    {
+      "epoch": 0.6155761616249358,
+      "grad_norm": 0.14962157607078552,
+      "learning_rate": 4.384431077041123e-06,
+      "loss": 0.8937,
+      "step": 85040
+    },
+    {
+      "epoch": 0.615648548285522,
+      "grad_norm": 0.16228331625461578,
+      "learning_rate": 4.384358690380537e-06,
+      "loss": 0.9054,
+      "step": 85050
+    },
+    {
+      "epoch": 0.6157209349461081,
+      "grad_norm": 0.1617509126663208,
+      "learning_rate": 4.3842863037199506e-06,
+      "loss": 0.8999,
+      "step": 85060
+    },
+    {
+      "epoch": 0.6157933216066943,
+      "grad_norm": 0.15040123462677002,
+      "learning_rate": 4.384213917059364e-06,
+      "loss": 0.9004,
+      "step": 85070
+    },
+    {
+      "epoch": 0.6158657082672805,
+      "grad_norm": 0.15811018645763397,
+      "learning_rate": 4.384141530398779e-06,
+      "loss": 0.8944,
+      "step": 85080
+    },
+    {
+      "epoch": 0.6159380949278667,
+      "grad_norm": 0.17127995193004608,
+      "learning_rate": 4.384069143738192e-06,
+      "loss": 0.8957,
+      "step": 85090
+    },
+    {
+      "epoch": 0.6160104815884528,
+      "grad_norm": 0.16223451495170593,
+      "learning_rate": 4.383996757077606e-06,
+      "loss": 0.8946,
+      "step": 85100
+    },
+    {
+      "epoch": 0.616082868249039,
+      "grad_norm": 0.1593468189239502,
+      "learning_rate": 4.3839243704170195e-06,
+      "loss": 0.908,
+      "step": 85110
+    },
+    {
+      "epoch": 0.6161552549096253,
+      "grad_norm": 0.16201543807983398,
+      "learning_rate": 4.383851983756434e-06,
+      "loss": 0.9083,
+      "step": 85120
+    },
+    {
+      "epoch": 0.6162276415702115,
+      "grad_norm": 0.15319061279296875,
+      "learning_rate": 4.3837795970958476e-06,
+      "loss": 0.9078,
+      "step": 85130
+    },
+    {
+      "epoch": 0.6163000282307977,
+      "grad_norm": 0.15424944460391998,
+      "learning_rate": 4.383707210435261e-06,
+      "loss": 0.9044,
+      "step": 85140
+    },
+    {
+      "epoch": 0.6163724148913838,
+      "grad_norm": 0.17157965898513794,
+      "learning_rate": 4.383634823774675e-06,
+      "loss": 0.9082,
+      "step": 85150
+    },
+    {
+      "epoch": 0.61644480155197,
+      "grad_norm": 0.15128067135810852,
+      "learning_rate": 4.383562437114089e-06,
+      "loss": 0.9109,
+      "step": 85160
+    },
+    {
+      "epoch": 0.6165171882125562,
+      "grad_norm": 0.170233815908432,
+      "learning_rate": 4.383490050453503e-06,
+      "loss": 0.901,
+      "step": 85170
+    },
+    {
+      "epoch": 0.6165895748731424,
+      "grad_norm": 0.16256168484687805,
+      "learning_rate": 4.3834176637929165e-06,
+      "loss": 0.9104,
+      "step": 85180
+    },
+    {
+      "epoch": 0.6166619615337285,
+      "grad_norm": 0.16661547124385834,
+      "learning_rate": 4.38334527713233e-06,
+      "loss": 0.9079,
+      "step": 85190
+    },
+    {
+      "epoch": 0.6167343481943147,
+      "grad_norm": 0.17816273868083954,
+      "learning_rate": 4.3832728904717446e-06,
+      "loss": 0.9076,
+      "step": 85200
+    },
+    {
+      "epoch": 0.6168067348549009,
+      "grad_norm": 0.16999828815460205,
+      "learning_rate": 4.383200503811158e-06,
+      "loss": 0.9152,
+      "step": 85210
+    },
+    {
+      "epoch": 0.6168791215154871,
+      "grad_norm": 0.14718665182590485,
+      "learning_rate": 4.383128117150572e-06,
+      "loss": 0.8982,
+      "step": 85220
+    },
+    {
+      "epoch": 0.6169515081760734,
+      "grad_norm": 0.2667122781276703,
+      "learning_rate": 4.383055730489985e-06,
+      "loss": 0.9059,
+      "step": 85230
+    },
+    {
+      "epoch": 0.6170238948366595,
+      "grad_norm": 0.1408415138721466,
+      "learning_rate": 4.3829833438294e-06,
+      "loss": 0.9018,
+      "step": 85240
+    },
+    {
+      "epoch": 0.6170962814972457,
+      "grad_norm": 0.16204825043678284,
+      "learning_rate": 4.3829109571688135e-06,
+      "loss": 0.8952,
+      "step": 85250
+    },
+    {
+      "epoch": 0.6171686681578319,
+      "grad_norm": 0.1726893186569214,
+      "learning_rate": 4.382838570508227e-06,
+      "loss": 0.9127,
+      "step": 85260
+    },
+    {
+      "epoch": 0.6172410548184181,
+      "grad_norm": 0.1730419099330902,
+      "learning_rate": 4.382766183847641e-06,
+      "loss": 0.9003,
+      "step": 85270
+    },
+    {
+      "epoch": 0.6173134414790042,
+      "grad_norm": 0.17287707328796387,
+      "learning_rate": 4.382693797187055e-06,
+      "loss": 0.8961,
+      "step": 85280
+    },
+    {
+      "epoch": 0.6173858281395904,
+      "grad_norm": 0.1501106470823288,
+      "learning_rate": 4.382621410526469e-06,
+      "loss": 0.9104,
+      "step": 85290
+    },
+    {
+      "epoch": 0.6174582148001766,
+      "grad_norm": 0.16863562166690826,
+      "learning_rate": 4.382549023865882e-06,
+      "loss": 0.8988,
+      "step": 85300
+    },
+    {
+      "epoch": 0.6175306014607628,
+      "grad_norm": 0.1819497048854828,
+      "learning_rate": 4.382476637205296e-06,
+      "loss": 0.9,
+      "step": 85310
+    },
+    {
+      "epoch": 0.617602988121349,
+      "grad_norm": 0.15243135392665863,
+      "learning_rate": 4.3824042505447105e-06,
+      "loss": 0.9003,
+      "step": 85320
+    },
+    {
+      "epoch": 0.6176753747819352,
+      "grad_norm": 0.1578928381204605,
+      "learning_rate": 4.382331863884124e-06,
+      "loss": 0.9003,
+      "step": 85330
+    },
+    {
+      "epoch": 0.6177477614425214,
+      "grad_norm": 0.15596744418144226,
+      "learning_rate": 4.382259477223538e-06,
+      "loss": 0.9036,
+      "step": 85340
+    },
+    {
+      "epoch": 0.6178201481031076,
+      "grad_norm": 0.16625262796878815,
+      "learning_rate": 4.382187090562951e-06,
+      "loss": 0.9022,
+      "step": 85350
+    },
+    {
+      "epoch": 0.6178925347636938,
+      "grad_norm": 0.14543043076992035,
+      "learning_rate": 4.382114703902365e-06,
+      "loss": 0.8922,
+      "step": 85360
+    },
+    {
+      "epoch": 0.61796492142428,
+      "grad_norm": 0.14668937027454376,
+      "learning_rate": 4.3820423172417794e-06,
+      "loss": 0.8963,
+      "step": 85370
+    },
+    {
+      "epoch": 0.6180373080848661,
+      "grad_norm": 0.15720996260643005,
+      "learning_rate": 4.381969930581193e-06,
+      "loss": 0.9104,
+      "step": 85380
+    },
+    {
+      "epoch": 0.6181096947454523,
+      "grad_norm": 0.20788639783859253,
+      "learning_rate": 4.381897543920607e-06,
+      "loss": 0.908,
+      "step": 85390
+    },
+    {
+      "epoch": 0.6181820814060385,
+      "grad_norm": 0.1491943746805191,
+      "learning_rate": 4.38182515726002e-06,
+      "loss": 0.9125,
+      "step": 85400
+    },
+    {
+      "epoch": 0.6182544680666247,
+      "grad_norm": 0.16570164263248444,
+      "learning_rate": 4.381752770599435e-06,
+      "loss": 0.8988,
+      "step": 85410
+    },
+    {
+      "epoch": 0.6183268547272108,
+      "grad_norm": 0.15333351492881775,
+      "learning_rate": 4.381680383938848e-06,
+      "loss": 0.9015,
+      "step": 85420
+    },
+    {
+      "epoch": 0.618399241387797,
+      "grad_norm": 0.15450377762317657,
+      "learning_rate": 4.381607997278262e-06,
+      "loss": 0.9206,
+      "step": 85430
+    },
+    {
+      "epoch": 0.6184716280483833,
+      "grad_norm": 0.15417149662971497,
+      "learning_rate": 4.381535610617676e-06,
+      "loss": 0.9062,
+      "step": 85440
+    },
+    {
+      "epoch": 0.6185440147089695,
+      "grad_norm": 0.1625894010066986,
+      "learning_rate": 4.38146322395709e-06,
+      "loss": 0.9074,
+      "step": 85450
+    },
+    {
+      "epoch": 0.6186164013695556,
+      "grad_norm": 0.15775802731513977,
+      "learning_rate": 4.381390837296504e-06,
+      "loss": 0.9068,
+      "step": 85460
+    },
+    {
+      "epoch": 0.6186887880301418,
+      "grad_norm": 0.16794852912425995,
+      "learning_rate": 4.381318450635917e-06,
+      "loss": 0.896,
+      "step": 85470
+    },
+    {
+      "epoch": 0.618761174690728,
+      "grad_norm": 0.158334419131279,
+      "learning_rate": 4.381246063975331e-06,
+      "loss": 0.8872,
+      "step": 85480
+    },
+    {
+      "epoch": 0.6188335613513142,
+      "grad_norm": 0.15617749094963074,
+      "learning_rate": 4.3811736773147445e-06,
+      "loss": 0.9026,
+      "step": 85490
+    },
+    {
+      "epoch": 0.6189059480119004,
+      "grad_norm": 0.16775725781917572,
+      "learning_rate": 4.381101290654158e-06,
+      "loss": 0.9062,
+      "step": 85500
+    },
+    {
+      "epoch": 0.6189783346724865,
+      "grad_norm": 0.16923066973686218,
+      "learning_rate": 4.381028903993572e-06,
+      "loss": 0.9077,
+      "step": 85510
+    },
+    {
+      "epoch": 0.6190507213330727,
+      "grad_norm": 0.16332891583442688,
+      "learning_rate": 4.380956517332986e-06,
+      "loss": 0.8984,
+      "step": 85520
+    },
+    {
+      "epoch": 0.6191231079936589,
+      "grad_norm": 0.1725182682275772,
+      "learning_rate": 4.3808841306724e-06,
+      "loss": 0.8937,
+      "step": 85530
+    },
+    {
+      "epoch": 0.6191954946542452,
+      "grad_norm": 0.17352226376533508,
+      "learning_rate": 4.3808117440118134e-06,
+      "loss": 0.9188,
+      "step": 85540
+    },
+    {
+      "epoch": 0.6192678813148313,
+      "grad_norm": 0.1709306836128235,
+      "learning_rate": 4.380739357351227e-06,
+      "loss": 0.8912,
+      "step": 85550
+    },
+    {
+      "epoch": 0.6193402679754175,
+      "grad_norm": 0.2021743208169937,
+      "learning_rate": 4.3806669706906415e-06,
+      "loss": 0.9081,
+      "step": 85560
+    },
+    {
+      "epoch": 0.6194126546360037,
+      "grad_norm": 0.14920145273208618,
+      "learning_rate": 4.380594584030055e-06,
+      "loss": 0.8997,
+      "step": 85570
+    },
+    {
+      "epoch": 0.6194850412965899,
+      "grad_norm": 0.1603516936302185,
+      "learning_rate": 4.380522197369469e-06,
+      "loss": 0.9109,
+      "step": 85580
+    },
+    {
+      "epoch": 0.619557427957176,
+      "grad_norm": 0.14569969475269318,
+      "learning_rate": 4.380449810708882e-06,
+      "loss": 0.896,
+      "step": 85590
+    },
+    {
+      "epoch": 0.6196298146177622,
+      "grad_norm": 0.16068822145462036,
+      "learning_rate": 4.380377424048297e-06,
+      "loss": 0.8974,
+      "step": 85600
+    },
+    {
+      "epoch": 0.6197022012783484,
+      "grad_norm": 0.16494551301002502,
+      "learning_rate": 4.3803050373877105e-06,
+      "loss": 0.8991,
+      "step": 85610
+    },
+    {
+      "epoch": 0.6197745879389346,
+      "grad_norm": 0.15262630581855774,
+      "learning_rate": 4.380232650727124e-06,
+      "loss": 0.8966,
+      "step": 85620
+    },
+    {
+      "epoch": 0.6198469745995208,
+      "grad_norm": 0.15079142153263092,
+      "learning_rate": 4.380160264066538e-06,
+      "loss": 0.8946,
+      "step": 85630
+    },
+    {
+      "epoch": 0.6199193612601069,
+      "grad_norm": 0.15024054050445557,
+      "learning_rate": 4.380087877405952e-06,
+      "loss": 0.9155,
+      "step": 85640
+    },
+    {
+      "epoch": 0.6199917479206932,
+      "grad_norm": 0.17157381772994995,
+      "learning_rate": 4.380015490745366e-06,
+      "loss": 0.9006,
+      "step": 85650
+    },
+    {
+      "epoch": 0.6200641345812794,
+      "grad_norm": 0.1453818827867508,
+      "learning_rate": 4.379943104084779e-06,
+      "loss": 0.9009,
+      "step": 85660
+    },
+    {
+      "epoch": 0.6201365212418656,
+      "grad_norm": 0.15377745032310486,
+      "learning_rate": 4.379870717424193e-06,
+      "loss": 0.8998,
+      "step": 85670
+    },
+    {
+      "epoch": 0.6202089079024518,
+      "grad_norm": 0.16755272448062897,
+      "learning_rate": 4.3797983307636075e-06,
+      "loss": 0.8894,
+      "step": 85680
+    },
+    {
+      "epoch": 0.6202812945630379,
+      "grad_norm": 0.1771673858165741,
+      "learning_rate": 4.379725944103021e-06,
+      "loss": 0.9014,
+      "step": 85690
+    },
+    {
+      "epoch": 0.6203536812236241,
+      "grad_norm": 0.1448373943567276,
+      "learning_rate": 4.379653557442435e-06,
+      "loss": 0.8946,
+      "step": 85700
+    },
+    {
+      "epoch": 0.6204260678842103,
+      "grad_norm": 0.15097440779209137,
+      "learning_rate": 4.379581170781848e-06,
+      "loss": 0.8906,
+      "step": 85710
+    },
+    {
+      "epoch": 0.6204984545447965,
+      "grad_norm": 0.1607360988855362,
+      "learning_rate": 4.379508784121263e-06,
+      "loss": 0.9011,
+      "step": 85720
+    },
+    {
+      "epoch": 0.6205708412053826,
+      "grad_norm": 0.16542325913906097,
+      "learning_rate": 4.379436397460676e-06,
+      "loss": 0.9062,
+      "step": 85730
+    },
+    {
+      "epoch": 0.6206432278659688,
+      "grad_norm": 0.18820612132549286,
+      "learning_rate": 4.37936401080009e-06,
+      "loss": 0.9047,
+      "step": 85740
+    },
+    {
+      "epoch": 0.620715614526555,
+      "grad_norm": 0.15945078432559967,
+      "learning_rate": 4.379291624139504e-06,
+      "loss": 0.9064,
+      "step": 85750
+    },
+    {
+      "epoch": 0.6207880011871413,
+      "grad_norm": 0.2132113128900528,
+      "learning_rate": 4.379219237478918e-06,
+      "loss": 0.9078,
+      "step": 85760
+    },
+    {
+      "epoch": 0.6208603878477275,
+      "grad_norm": 0.15202796459197998,
+      "learning_rate": 4.379146850818332e-06,
+      "loss": 0.901,
+      "step": 85770
+    },
+    {
+      "epoch": 0.6209327745083136,
+      "grad_norm": 0.15480341017246246,
+      "learning_rate": 4.379074464157745e-06,
+      "loss": 0.9037,
+      "step": 85780
+    },
+    {
+      "epoch": 0.6210051611688998,
+      "grad_norm": 0.22787299752235413,
+      "learning_rate": 4.379002077497159e-06,
+      "loss": 0.9112,
+      "step": 85790
+    },
+    {
+      "epoch": 0.621077547829486,
+      "grad_norm": 0.16583016514778137,
+      "learning_rate": 4.378929690836573e-06,
+      "loss": 0.9095,
+      "step": 85800
+    },
+    {
+      "epoch": 0.6211499344900722,
+      "grad_norm": 0.1868494153022766,
+      "learning_rate": 4.378857304175987e-06,
+      "loss": 0.8924,
+      "step": 85810
+    },
+    {
+      "epoch": 0.6212223211506583,
+      "grad_norm": 0.18490175902843475,
+      "learning_rate": 4.378784917515401e-06,
+      "loss": 0.8871,
+      "step": 85820
+    },
+    {
+      "epoch": 0.6212947078112445,
+      "grad_norm": 0.15634137392044067,
+      "learning_rate": 4.378712530854814e-06,
+      "loss": 0.9004,
+      "step": 85830
+    },
+    {
+      "epoch": 0.6213670944718307,
+      "grad_norm": 0.24686381220817566,
+      "learning_rate": 4.378640144194229e-06,
+      "loss": 0.8762,
+      "step": 85840
+    },
+    {
+      "epoch": 0.6214394811324169,
+      "grad_norm": 0.1837376058101654,
+      "learning_rate": 4.378567757533642e-06,
+      "loss": 0.9133,
+      "step": 85850
+    },
+    {
+      "epoch": 0.6215118677930032,
+      "grad_norm": 0.16701044142246246,
+      "learning_rate": 4.378495370873056e-06,
+      "loss": 0.9067,
+      "step": 85860
+    },
+    {
+      "epoch": 0.6215842544535893,
+      "grad_norm": 0.15616366267204285,
+      "learning_rate": 4.3784229842124696e-06,
+      "loss": 0.9053,
+      "step": 85870
+    },
+    {
+      "epoch": 0.6216566411141755,
+      "grad_norm": 0.1612260639667511,
+      "learning_rate": 4.378350597551884e-06,
+      "loss": 0.8892,
+      "step": 85880
+    },
+    {
+      "epoch": 0.6217290277747617,
+      "grad_norm": 0.17558954656124115,
+      "learning_rate": 4.378278210891298e-06,
+      "loss": 0.9048,
+      "step": 85890
+    },
+    {
+      "epoch": 0.6218014144353479,
+      "grad_norm": 0.175954669713974,
+      "learning_rate": 4.378205824230711e-06,
+      "loss": 0.8994,
+      "step": 85900
+    },
+    {
+      "epoch": 0.621873801095934,
+      "grad_norm": 0.1546930968761444,
+      "learning_rate": 4.378133437570125e-06,
+      "loss": 0.8875,
+      "step": 85910
+    },
+    {
+      "epoch": 0.6219461877565202,
+      "grad_norm": 0.1583179086446762,
+      "learning_rate": 4.378061050909539e-06,
+      "loss": 0.8928,
+      "step": 85920
+    },
+    {
+      "epoch": 0.6220185744171064,
+      "grad_norm": 0.15575550496578217,
+      "learning_rate": 4.377988664248953e-06,
+      "loss": 0.9023,
+      "step": 85930
+    },
+    {
+      "epoch": 0.6220909610776926,
+      "grad_norm": 0.1556166708469391,
+      "learning_rate": 4.3779162775883666e-06,
+      "loss": 0.9032,
+      "step": 85940
+    },
+    {
+      "epoch": 0.6221633477382787,
+      "grad_norm": 0.16194866597652435,
+      "learning_rate": 4.37784389092778e-06,
+      "loss": 0.9187,
+      "step": 85950
+    },
+    {
+      "epoch": 0.6222357343988649,
+      "grad_norm": 0.15536653995513916,
+      "learning_rate": 4.377771504267194e-06,
+      "loss": 0.907,
+      "step": 85960
+    },
+    {
+      "epoch": 0.6223081210594512,
+      "grad_norm": 0.15262702107429504,
+      "learning_rate": 4.377699117606608e-06,
+      "loss": 0.8914,
+      "step": 85970
+    },
+    {
+      "epoch": 0.6223805077200374,
+      "grad_norm": 0.26444554328918457,
+      "learning_rate": 4.377626730946022e-06,
+      "loss": 0.9096,
+      "step": 85980
+    },
+    {
+      "epoch": 0.6224528943806236,
+      "grad_norm": 0.14140740036964417,
+      "learning_rate": 4.3775543442854355e-06,
+      "loss": 0.898,
+      "step": 85990
+    },
+    {
+      "epoch": 0.6225252810412097,
+      "grad_norm": 0.17096258699893951,
+      "learning_rate": 4.377481957624849e-06,
+      "loss": 0.9021,
+      "step": 86000
+    },
+    {
+      "epoch": 0.6225976677017959,
+      "grad_norm": 0.16008111834526062,
+      "learning_rate": 4.3774095709642636e-06,
+      "loss": 0.8982,
+      "step": 86010
+    },
+    {
+      "epoch": 0.6226700543623821,
+      "grad_norm": 0.1551225781440735,
+      "learning_rate": 4.377337184303676e-06,
+      "loss": 0.9093,
+      "step": 86020
+    },
+    {
+      "epoch": 0.6227424410229683,
+      "grad_norm": 0.14841316640377045,
+      "learning_rate": 4.377264797643091e-06,
+      "loss": 0.902,
+      "step": 86030
+    },
+    {
+      "epoch": 0.6228148276835545,
+      "grad_norm": 0.18173016607761383,
+      "learning_rate": 4.377192410982504e-06,
+      "loss": 0.9144,
+      "step": 86040
+    },
+    {
+      "epoch": 0.6228872143441406,
+      "grad_norm": 0.14934363961219788,
+      "learning_rate": 4.377120024321918e-06,
+      "loss": 0.8974,
+      "step": 86050
+    },
+    {
+      "epoch": 0.6229596010047268,
+      "grad_norm": 0.17030176520347595,
+      "learning_rate": 4.377047637661332e-06,
+      "loss": 0.8943,
+      "step": 86060
+    },
+    {
+      "epoch": 0.623031987665313,
+      "grad_norm": 0.16661794483661652,
+      "learning_rate": 4.376975251000746e-06,
+      "loss": 0.9049,
+      "step": 86070
+    },
+    {
+      "epoch": 0.6231043743258993,
+      "grad_norm": 0.16051805019378662,
+      "learning_rate": 4.37690286434016e-06,
+      "loss": 0.899,
+      "step": 86080
+    },
+    {
+      "epoch": 0.6231767609864854,
+      "grad_norm": 0.16653110086917877,
+      "learning_rate": 4.376830477679573e-06,
+      "loss": 0.9057,
+      "step": 86090
+    },
+    {
+      "epoch": 0.6232491476470716,
+      "grad_norm": 0.16507036983966827,
+      "learning_rate": 4.376758091018987e-06,
+      "loss": 0.907,
+      "step": 86100
+    },
+    {
+      "epoch": 0.6233215343076578,
+      "grad_norm": 0.16348743438720703,
+      "learning_rate": 4.3766857043584014e-06,
+      "loss": 0.9077,
+      "step": 86110
+    },
+    {
+      "epoch": 0.623393920968244,
+      "grad_norm": 0.15697596967220306,
+      "learning_rate": 4.376613317697815e-06,
+      "loss": 0.9028,
+      "step": 86120
+    },
+    {
+      "epoch": 0.6234663076288302,
+      "grad_norm": 0.15257367491722107,
+      "learning_rate": 4.376540931037229e-06,
+      "loss": 0.896,
+      "step": 86130
+    },
+    {
+      "epoch": 0.6235386942894163,
+      "grad_norm": 0.16256658732891083,
+      "learning_rate": 4.376468544376642e-06,
+      "loss": 0.9138,
+      "step": 86140
+    },
+    {
+      "epoch": 0.6236110809500025,
+      "grad_norm": 0.19833579659461975,
+      "learning_rate": 4.376396157716056e-06,
+      "loss": 0.9078,
+      "step": 86150
+    },
+    {
+      "epoch": 0.6236834676105887,
+      "grad_norm": 0.15901575982570648,
+      "learning_rate": 4.37632377105547e-06,
+      "loss": 0.8936,
+      "step": 86160
+    },
+    {
+      "epoch": 0.6237558542711749,
+      "grad_norm": 0.16411714255809784,
+      "learning_rate": 4.376251384394884e-06,
+      "loss": 0.9072,
+      "step": 86170
+    },
+    {
+      "epoch": 0.6238282409317611,
+      "grad_norm": 0.1543898582458496,
+      "learning_rate": 4.376178997734298e-06,
+      "loss": 0.91,
+      "step": 86180
+    },
+    {
+      "epoch": 0.6239006275923473,
+      "grad_norm": 0.15518857538700104,
+      "learning_rate": 4.376106611073711e-06,
+      "loss": 0.9141,
+      "step": 86190
+    },
+    {
+      "epoch": 0.6239730142529335,
+      "grad_norm": 0.15521982312202454,
+      "learning_rate": 4.376034224413126e-06,
+      "loss": 0.9045,
+      "step": 86200
+    },
+    {
+      "epoch": 0.6240454009135197,
+      "grad_norm": 0.1619000881910324,
+      "learning_rate": 4.375961837752539e-06,
+      "loss": 0.8963,
+      "step": 86210
+    },
+    {
+      "epoch": 0.6241177875741059,
+      "grad_norm": 0.15840552747249603,
+      "learning_rate": 4.375889451091953e-06,
+      "loss": 0.8982,
+      "step": 86220
+    },
+    {
+      "epoch": 0.624190174234692,
+      "grad_norm": 0.1774725317955017,
+      "learning_rate": 4.3758170644313665e-06,
+      "loss": 0.9161,
+      "step": 86230
+    },
+    {
+      "epoch": 0.6242625608952782,
+      "grad_norm": 0.160127192735672,
+      "learning_rate": 4.375744677770781e-06,
+      "loss": 0.8993,
+      "step": 86240
+    },
+    {
+      "epoch": 0.6243349475558644,
+      "grad_norm": 0.1587544083595276,
+      "learning_rate": 4.375672291110195e-06,
+      "loss": 0.892,
+      "step": 86250
+    },
+    {
+      "epoch": 0.6244073342164506,
+      "grad_norm": 0.1745489090681076,
+      "learning_rate": 4.375599904449608e-06,
+      "loss": 0.9089,
+      "step": 86260
+    },
+    {
+      "epoch": 0.6244797208770367,
+      "grad_norm": 0.1463412344455719,
+      "learning_rate": 4.375527517789022e-06,
+      "loss": 0.8934,
+      "step": 86270
+    },
+    {
+      "epoch": 0.6245521075376229,
+      "grad_norm": 0.16872276365756989,
+      "learning_rate": 4.375455131128436e-06,
+      "loss": 0.9004,
+      "step": 86280
+    },
+    {
+      "epoch": 0.6246244941982092,
+      "grad_norm": 0.1642751842737198,
+      "learning_rate": 4.37538274446785e-06,
+      "loss": 0.8958,
+      "step": 86290
+    },
+    {
+      "epoch": 0.6246968808587954,
+      "grad_norm": 0.17172108590602875,
+      "learning_rate": 4.3753103578072635e-06,
+      "loss": 0.9021,
+      "step": 86300
+    },
+    {
+      "epoch": 0.6247692675193816,
+      "grad_norm": 0.1646968573331833,
+      "learning_rate": 4.375237971146677e-06,
+      "loss": 0.8952,
+      "step": 86310
+    },
+    {
+      "epoch": 0.6248416541799677,
+      "grad_norm": 0.15084651112556458,
+      "learning_rate": 4.375165584486092e-06,
+      "loss": 0.894,
+      "step": 86320
+    },
+    {
+      "epoch": 0.6249140408405539,
+      "grad_norm": 0.15649060904979706,
+      "learning_rate": 4.375093197825505e-06,
+      "loss": 0.9019,
+      "step": 86330
+    },
+    {
+      "epoch": 0.6249864275011401,
+      "grad_norm": 0.16130280494689941,
+      "learning_rate": 4.375020811164919e-06,
+      "loss": 0.9091,
+      "step": 86340
+    },
+    {
+      "epoch": 0.6250588141617263,
+      "grad_norm": 0.15296019613742828,
+      "learning_rate": 4.3749484245043325e-06,
+      "loss": 0.9127,
+      "step": 86350
+    },
+    {
+      "epoch": 0.6251312008223124,
+      "grad_norm": 0.15848460793495178,
+      "learning_rate": 4.374876037843747e-06,
+      "loss": 0.8889,
+      "step": 86360
+    },
+    {
+      "epoch": 0.6252035874828986,
+      "grad_norm": 0.15533143281936646,
+      "learning_rate": 4.3748036511831605e-06,
+      "loss": 0.9016,
+      "step": 86370
+    },
+    {
+      "epoch": 0.6252759741434848,
+      "grad_norm": 0.15972629189491272,
+      "learning_rate": 4.374731264522574e-06,
+      "loss": 0.9124,
+      "step": 86380
+    },
+    {
+      "epoch": 0.6253483608040711,
+      "grad_norm": 0.18219304084777832,
+      "learning_rate": 4.374658877861988e-06,
+      "loss": 0.8992,
+      "step": 86390
+    },
+    {
+      "epoch": 0.6254207474646573,
+      "grad_norm": 0.14863041043281555,
+      "learning_rate": 4.374586491201402e-06,
+      "loss": 0.8854,
+      "step": 86400
+    },
+    {
+      "epoch": 0.6254931341252434,
+      "grad_norm": 0.1658528745174408,
+      "learning_rate": 4.374514104540816e-06,
+      "loss": 0.8979,
+      "step": 86410
+    },
+    {
+      "epoch": 0.6255655207858296,
+      "grad_norm": 0.15939390659332275,
+      "learning_rate": 4.3744417178802295e-06,
+      "loss": 0.9003,
+      "step": 86420
+    },
+    {
+      "epoch": 0.6256379074464158,
+      "grad_norm": 0.17238759994506836,
+      "learning_rate": 4.374369331219643e-06,
+      "loss": 0.9056,
+      "step": 86430
+    },
+    {
+      "epoch": 0.625710294107002,
+      "grad_norm": 0.1758257895708084,
+      "learning_rate": 4.3742969445590575e-06,
+      "loss": 0.9052,
+      "step": 86440
+    },
+    {
+      "epoch": 0.6257826807675881,
+      "grad_norm": 0.1517690122127533,
+      "learning_rate": 4.374224557898471e-06,
+      "loss": 0.8987,
+      "step": 86450
+    },
+    {
+      "epoch": 0.6258550674281743,
+      "grad_norm": 0.16682100296020508,
+      "learning_rate": 4.374152171237885e-06,
+      "loss": 0.9072,
+      "step": 86460
+    },
+    {
+      "epoch": 0.6259274540887605,
+      "grad_norm": 0.14764508605003357,
+      "learning_rate": 4.374079784577298e-06,
+      "loss": 0.8948,
+      "step": 86470
+    },
+    {
+      "epoch": 0.6259998407493467,
+      "grad_norm": 0.1788049340248108,
+      "learning_rate": 4.374007397916713e-06,
+      "loss": 0.8991,
+      "step": 86480
+    },
+    {
+      "epoch": 0.6260722274099328,
+      "grad_norm": 0.15142115950584412,
+      "learning_rate": 4.3739350112561265e-06,
+      "loss": 0.9062,
+      "step": 86490
+    },
+    {
+      "epoch": 0.6261446140705191,
+      "grad_norm": 0.1992928683757782,
+      "learning_rate": 4.37386262459554e-06,
+      "loss": 0.9139,
+      "step": 86500
+    },
+    {
+      "epoch": 0.6262170007311053,
+      "grad_norm": 0.1716473549604416,
+      "learning_rate": 4.373790237934954e-06,
+      "loss": 0.8933,
+      "step": 86510
+    },
+    {
+      "epoch": 0.6262893873916915,
+      "grad_norm": 0.19439038634300232,
+      "learning_rate": 4.373717851274368e-06,
+      "loss": 0.8969,
+      "step": 86520
+    },
+    {
+      "epoch": 0.6263617740522777,
+      "grad_norm": 0.13993479311466217,
+      "learning_rate": 4.373645464613782e-06,
+      "loss": 0.9131,
+      "step": 86530
+    },
+    {
+      "epoch": 0.6264341607128638,
+      "grad_norm": 0.1706664115190506,
+      "learning_rate": 4.373573077953195e-06,
+      "loss": 0.9075,
+      "step": 86540
+    },
+    {
+      "epoch": 0.62650654737345,
+      "grad_norm": 0.17721545696258545,
+      "learning_rate": 4.373500691292609e-06,
+      "loss": 0.9067,
+      "step": 86550
+    },
+    {
+      "epoch": 0.6265789340340362,
+      "grad_norm": 0.1569727510213852,
+      "learning_rate": 4.373428304632023e-06,
+      "loss": 0.9073,
+      "step": 86560
+    },
+    {
+      "epoch": 0.6266513206946224,
+      "grad_norm": 0.15974320471286774,
+      "learning_rate": 4.373355917971436e-06,
+      "loss": 0.9023,
+      "step": 86570
+    },
+    {
+      "epoch": 0.6267237073552085,
+      "grad_norm": 0.14838136732578278,
+      "learning_rate": 4.37328353131085e-06,
+      "loss": 0.8885,
+      "step": 86580
+    },
+    {
+      "epoch": 0.6267960940157947,
+      "grad_norm": 0.1551826149225235,
+      "learning_rate": 4.373211144650264e-06,
+      "loss": 0.905,
+      "step": 86590
+    },
+    {
+      "epoch": 0.6268684806763809,
+      "grad_norm": 0.158196359872818,
+      "learning_rate": 4.373138757989678e-06,
+      "loss": 0.8934,
+      "step": 86600
+    },
+    {
+      "epoch": 0.6269408673369672,
+      "grad_norm": 0.17901091277599335,
+      "learning_rate": 4.3730663713290916e-06,
+      "loss": 0.919,
+      "step": 86610
+    },
+    {
+      "epoch": 0.6270132539975534,
+      "grad_norm": 0.1554396152496338,
+      "learning_rate": 4.372993984668505e-06,
+      "loss": 0.896,
+      "step": 86620
+    },
+    {
+      "epoch": 0.6270856406581395,
+      "grad_norm": 0.1537899672985077,
+      "learning_rate": 4.37292159800792e-06,
+      "loss": 0.8975,
+      "step": 86630
+    },
+    {
+      "epoch": 0.6271580273187257,
+      "grad_norm": 0.16763430833816528,
+      "learning_rate": 4.372849211347333e-06,
+      "loss": 0.8944,
+      "step": 86640
+    },
+    {
+      "epoch": 0.6272304139793119,
+      "grad_norm": 0.15677796304225922,
+      "learning_rate": 4.372776824686747e-06,
+      "loss": 0.9166,
+      "step": 86650
+    },
+    {
+      "epoch": 0.6273028006398981,
+      "grad_norm": 0.15729351341724396,
+      "learning_rate": 4.3727044380261605e-06,
+      "loss": 0.9052,
+      "step": 86660
+    },
+    {
+      "epoch": 0.6273751873004842,
+      "grad_norm": 0.16490468382835388,
+      "learning_rate": 4.372632051365575e-06,
+      "loss": 0.8986,
+      "step": 86670
+    },
+    {
+      "epoch": 0.6274475739610704,
+      "grad_norm": 0.16459213197231293,
+      "learning_rate": 4.3725596647049886e-06,
+      "loss": 0.8967,
+      "step": 86680
+    },
+    {
+      "epoch": 0.6275199606216566,
+      "grad_norm": 0.16037428379058838,
+      "learning_rate": 4.372487278044402e-06,
+      "loss": 0.9144,
+      "step": 86690
+    },
+    {
+      "epoch": 0.6275923472822428,
+      "grad_norm": 0.16006019711494446,
+      "learning_rate": 4.372414891383816e-06,
+      "loss": 0.9089,
+      "step": 86700
+    },
+    {
+      "epoch": 0.6276647339428291,
+      "grad_norm": 0.1548503190279007,
+      "learning_rate": 4.37234250472323e-06,
+      "loss": 0.9,
+      "step": 86710
+    },
+    {
+      "epoch": 0.6277371206034152,
+      "grad_norm": 0.21827849745750427,
+      "learning_rate": 4.372270118062644e-06,
+      "loss": 0.8975,
+      "step": 86720
+    },
+    {
+      "epoch": 0.6278095072640014,
+      "grad_norm": 0.16216625273227692,
+      "learning_rate": 4.3721977314020575e-06,
+      "loss": 0.9022,
+      "step": 86730
+    },
+    {
+      "epoch": 0.6278818939245876,
+      "grad_norm": 0.17740298807621002,
+      "learning_rate": 4.372125344741471e-06,
+      "loss": 0.9114,
+      "step": 86740
+    },
+    {
+      "epoch": 0.6279542805851738,
+      "grad_norm": 0.15223225951194763,
+      "learning_rate": 4.372052958080885e-06,
+      "loss": 0.886,
+      "step": 86750
+    },
+    {
+      "epoch": 0.62802666724576,
+      "grad_norm": 0.1714017689228058,
+      "learning_rate": 4.371980571420299e-06,
+      "loss": 0.8929,
+      "step": 86760
+    },
+    {
+      "epoch": 0.6280990539063461,
+      "grad_norm": 0.17561236023902893,
+      "learning_rate": 4.371908184759713e-06,
+      "loss": 0.9036,
+      "step": 86770
+    },
+    {
+      "epoch": 0.6281714405669323,
+      "grad_norm": 0.16735588014125824,
+      "learning_rate": 4.371835798099126e-06,
+      "loss": 0.9127,
+      "step": 86780
+    },
+    {
+      "epoch": 0.6282438272275185,
+      "grad_norm": 0.15860114991664886,
+      "learning_rate": 4.37176341143854e-06,
+      "loss": 0.9031,
+      "step": 86790
+    },
+    {
+      "epoch": 0.6283162138881047,
+      "grad_norm": 0.23566380143165588,
+      "learning_rate": 4.3716910247779545e-06,
+      "loss": 0.9014,
+      "step": 86800
+    },
+    {
+      "epoch": 0.6283886005486908,
+      "grad_norm": 0.18236784636974335,
+      "learning_rate": 4.371618638117368e-06,
+      "loss": 0.8992,
+      "step": 86810
+    },
+    {
+      "epoch": 0.6284609872092771,
+      "grad_norm": 0.1653483808040619,
+      "learning_rate": 4.371546251456782e-06,
+      "loss": 0.9042,
+      "step": 86820
+    },
+    {
+      "epoch": 0.6285333738698633,
+      "grad_norm": 0.15193116664886475,
+      "learning_rate": 4.371473864796195e-06,
+      "loss": 0.8964,
+      "step": 86830
+    },
+    {
+      "epoch": 0.6286057605304495,
+      "grad_norm": 0.15298867225646973,
+      "learning_rate": 4.37140147813561e-06,
+      "loss": 0.8856,
+      "step": 86840
+    },
+    {
+      "epoch": 0.6286781471910357,
+      "grad_norm": 0.15323877334594727,
+      "learning_rate": 4.3713290914750234e-06,
+      "loss": 0.9021,
+      "step": 86850
+    },
+    {
+      "epoch": 0.6287505338516218,
+      "grad_norm": 0.159663587808609,
+      "learning_rate": 4.371256704814437e-06,
+      "loss": 0.8852,
+      "step": 86860
+    },
+    {
+      "epoch": 0.628822920512208,
+      "grad_norm": 0.1943221390247345,
+      "learning_rate": 4.371184318153851e-06,
+      "loss": 0.9021,
+      "step": 86870
+    },
+    {
+      "epoch": 0.6288953071727942,
+      "grad_norm": 0.16797471046447754,
+      "learning_rate": 4.371111931493265e-06,
+      "loss": 0.9143,
+      "step": 86880
+    },
+    {
+      "epoch": 0.6289676938333804,
+      "grad_norm": 0.17109879851341248,
+      "learning_rate": 4.371039544832679e-06,
+      "loss": 0.905,
+      "step": 86890
+    },
+    {
+      "epoch": 0.6290400804939665,
+      "grad_norm": 0.17516149580478668,
+      "learning_rate": 4.370967158172092e-06,
+      "loss": 0.9148,
+      "step": 86900
+    },
+    {
+      "epoch": 0.6291124671545527,
+      "grad_norm": 0.15488648414611816,
+      "learning_rate": 4.370894771511506e-06,
+      "loss": 0.9063,
+      "step": 86910
+    },
+    {
+      "epoch": 0.629184853815139,
+      "grad_norm": 0.16343455016613007,
+      "learning_rate": 4.3708223848509204e-06,
+      "loss": 0.9098,
+      "step": 86920
+    },
+    {
+      "epoch": 0.6292572404757252,
+      "grad_norm": 0.1545439064502716,
+      "learning_rate": 4.370749998190334e-06,
+      "loss": 0.9044,
+      "step": 86930
+    },
+    {
+      "epoch": 0.6293296271363114,
+      "grad_norm": 0.14563705027103424,
+      "learning_rate": 4.370677611529748e-06,
+      "loss": 0.8909,
+      "step": 86940
+    },
+    {
+      "epoch": 0.6294020137968975,
+      "grad_norm": 0.16632063686847687,
+      "learning_rate": 4.370605224869161e-06,
+      "loss": 0.9051,
+      "step": 86950
+    },
+    {
+      "epoch": 0.6294744004574837,
+      "grad_norm": 0.1433282047510147,
+      "learning_rate": 4.370532838208576e-06,
+      "loss": 0.9143,
+      "step": 86960
+    },
+    {
+      "epoch": 0.6295467871180699,
+      "grad_norm": 0.16106697916984558,
+      "learning_rate": 4.370460451547989e-06,
+      "loss": 0.8937,
+      "step": 86970
+    },
+    {
+      "epoch": 0.6296191737786561,
+      "grad_norm": 0.15034160017967224,
+      "learning_rate": 4.370388064887403e-06,
+      "loss": 0.9059,
+      "step": 86980
+    },
+    {
+      "epoch": 0.6296915604392422,
+      "grad_norm": 0.15533864498138428,
+      "learning_rate": 4.370315678226817e-06,
+      "loss": 0.9062,
+      "step": 86990
+    },
+    {
+      "epoch": 0.6297639470998284,
+      "grad_norm": 0.17288139462471008,
+      "learning_rate": 4.370243291566231e-06,
+      "loss": 0.896,
+      "step": 87000
+    },
+    {
+      "epoch": 0.6298363337604146,
+      "grad_norm": 0.1603853702545166,
+      "learning_rate": 4.370170904905645e-06,
+      "loss": 0.9013,
+      "step": 87010
+    },
+    {
+      "epoch": 0.6299087204210008,
+      "grad_norm": 0.1669483482837677,
+      "learning_rate": 4.370098518245058e-06,
+      "loss": 0.8986,
+      "step": 87020
+    },
+    {
+      "epoch": 0.629981107081587,
+      "grad_norm": 0.15310995280742645,
+      "learning_rate": 4.370026131584472e-06,
+      "loss": 0.9085,
+      "step": 87030
+    },
+    {
+      "epoch": 0.6300534937421732,
+      "grad_norm": 0.1482187807559967,
+      "learning_rate": 4.369953744923886e-06,
+      "loss": 0.9044,
+      "step": 87040
+    },
+    {
+      "epoch": 0.6301258804027594,
+      "grad_norm": 0.16005839407444,
+      "learning_rate": 4.3698813582633e-06,
+      "loss": 0.9033,
+      "step": 87050
+    },
+    {
+      "epoch": 0.6301982670633456,
+      "grad_norm": 0.1608879268169403,
+      "learning_rate": 4.369808971602714e-06,
+      "loss": 0.8938,
+      "step": 87060
+    },
+    {
+      "epoch": 0.6302706537239318,
+      "grad_norm": 0.15144924819469452,
+      "learning_rate": 4.369736584942127e-06,
+      "loss": 0.905,
+      "step": 87070
+    },
+    {
+      "epoch": 0.6303430403845179,
+      "grad_norm": 0.1928846538066864,
+      "learning_rate": 4.369664198281541e-06,
+      "loss": 0.897,
+      "step": 87080
+    },
+    {
+      "epoch": 0.6304154270451041,
+      "grad_norm": 0.15390628576278687,
+      "learning_rate": 4.3695918116209545e-06,
+      "loss": 0.908,
+      "step": 87090
+    },
+    {
+      "epoch": 0.6304878137056903,
+      "grad_norm": 0.15502314269542694,
+      "learning_rate": 4.369519424960368e-06,
+      "loss": 0.8921,
+      "step": 87100
+    },
+    {
+      "epoch": 0.6305602003662765,
+      "grad_norm": 0.1564696878194809,
+      "learning_rate": 4.3694470382997825e-06,
+      "loss": 0.8971,
+      "step": 87110
+    },
+    {
+      "epoch": 0.6306325870268626,
+      "grad_norm": 0.1649816632270813,
+      "learning_rate": 4.369374651639196e-06,
+      "loss": 0.898,
+      "step": 87120
+    },
+    {
+      "epoch": 0.6307049736874488,
+      "grad_norm": 0.16276071965694427,
+      "learning_rate": 4.36930226497861e-06,
+      "loss": 0.8981,
+      "step": 87130
+    },
+    {
+      "epoch": 0.6307773603480351,
+      "grad_norm": 0.14670120179653168,
+      "learning_rate": 4.369229878318023e-06,
+      "loss": 0.8952,
+      "step": 87140
+    },
+    {
+      "epoch": 0.6308497470086213,
+      "grad_norm": 0.15809543430805206,
+      "learning_rate": 4.369157491657438e-06,
+      "loss": 0.9064,
+      "step": 87150
+    },
+    {
+      "epoch": 0.6309221336692075,
+      "grad_norm": 0.16645848751068115,
+      "learning_rate": 4.3690851049968515e-06,
+      "loss": 0.8973,
+      "step": 87160
+    },
+    {
+      "epoch": 0.6309945203297936,
+      "grad_norm": 0.1631806641817093,
+      "learning_rate": 4.369012718336265e-06,
+      "loss": 0.8984,
+      "step": 87170
+    },
+    {
+      "epoch": 0.6310669069903798,
+      "grad_norm": 0.20448337495326996,
+      "learning_rate": 4.368940331675679e-06,
+      "loss": 0.9061,
+      "step": 87180
+    },
+    {
+      "epoch": 0.631139293650966,
+      "grad_norm": 0.3554750084877014,
+      "learning_rate": 4.368867945015093e-06,
+      "loss": 0.9172,
+      "step": 87190
+    },
+    {
+      "epoch": 0.6312116803115522,
+      "grad_norm": 0.15781621634960175,
+      "learning_rate": 4.368795558354507e-06,
+      "loss": 0.9017,
+      "step": 87200
+    },
+    {
+      "epoch": 0.6312840669721383,
+      "grad_norm": 0.16704939305782318,
+      "learning_rate": 4.36872317169392e-06,
+      "loss": 0.8875,
+      "step": 87210
+    },
+    {
+      "epoch": 0.6313564536327245,
+      "grad_norm": 0.16684958338737488,
+      "learning_rate": 4.368650785033334e-06,
+      "loss": 0.9216,
+      "step": 87220
+    },
+    {
+      "epoch": 0.6314288402933107,
+      "grad_norm": 0.16323356330394745,
+      "learning_rate": 4.3685783983727485e-06,
+      "loss": 0.9072,
+      "step": 87230
+    },
+    {
+      "epoch": 0.631501226953897,
+      "grad_norm": 0.16825905442237854,
+      "learning_rate": 4.368506011712162e-06,
+      "loss": 0.8985,
+      "step": 87240
+    },
+    {
+      "epoch": 0.6315736136144832,
+      "grad_norm": 0.1925610899925232,
+      "learning_rate": 4.368433625051576e-06,
+      "loss": 0.9055,
+      "step": 87250
+    },
+    {
+      "epoch": 0.6316460002750693,
+      "grad_norm": 0.16499613225460052,
+      "learning_rate": 4.368361238390989e-06,
+      "loss": 0.9002,
+      "step": 87260
+    },
+    {
+      "epoch": 0.6317183869356555,
+      "grad_norm": 0.16235573589801788,
+      "learning_rate": 4.368288851730404e-06,
+      "loss": 0.9047,
+      "step": 87270
+    },
+    {
+      "epoch": 0.6317907735962417,
+      "grad_norm": 0.15934467315673828,
+      "learning_rate": 4.368216465069817e-06,
+      "loss": 0.8983,
+      "step": 87280
+    },
+    {
+      "epoch": 0.6318631602568279,
+      "grad_norm": 0.15878109633922577,
+      "learning_rate": 4.368144078409231e-06,
+      "loss": 0.9026,
+      "step": 87290
+    },
+    {
+      "epoch": 0.631935546917414,
+      "grad_norm": 0.14932307600975037,
+      "learning_rate": 4.368071691748645e-06,
+      "loss": 0.8983,
+      "step": 87300
+    },
+    {
+      "epoch": 0.6320079335780002,
+      "grad_norm": 0.1681598424911499,
+      "learning_rate": 4.367999305088059e-06,
+      "loss": 0.9061,
+      "step": 87310
+    },
+    {
+      "epoch": 0.6320803202385864,
+      "grad_norm": 0.14897821843624115,
+      "learning_rate": 4.367926918427473e-06,
+      "loss": 0.8846,
+      "step": 87320
+    },
+    {
+      "epoch": 0.6321527068991726,
+      "grad_norm": 0.4469514787197113,
+      "learning_rate": 4.367854531766886e-06,
+      "loss": 0.892,
+      "step": 87330
+    },
+    {
+      "epoch": 0.6322250935597588,
+      "grad_norm": 0.15947720408439636,
+      "learning_rate": 4.3677821451063e-06,
+      "loss": 0.8925,
+      "step": 87340
+    },
+    {
+      "epoch": 0.632297480220345,
+      "grad_norm": 0.15741553902626038,
+      "learning_rate": 4.367709758445714e-06,
+      "loss": 0.8912,
+      "step": 87350
+    },
+    {
+      "epoch": 0.6323698668809312,
+      "grad_norm": 0.1720518171787262,
+      "learning_rate": 4.367637371785128e-06,
+      "loss": 0.9073,
+      "step": 87360
+    },
+    {
+      "epoch": 0.6324422535415174,
+      "grad_norm": 0.14960655570030212,
+      "learning_rate": 4.367564985124542e-06,
+      "loss": 0.8995,
+      "step": 87370
+    },
+    {
+      "epoch": 0.6325146402021036,
+      "grad_norm": 0.16579873859882355,
+      "learning_rate": 4.367492598463955e-06,
+      "loss": 0.8921,
+      "step": 87380
+    },
+    {
+      "epoch": 0.6325870268626897,
+      "grad_norm": 0.15160775184631348,
+      "learning_rate": 4.367420211803369e-06,
+      "loss": 0.8939,
+      "step": 87390
+    },
+    {
+      "epoch": 0.6326594135232759,
+      "grad_norm": 0.15509670972824097,
+      "learning_rate": 4.367347825142783e-06,
+      "loss": 0.8962,
+      "step": 87400
+    },
+    {
+      "epoch": 0.6327318001838621,
+      "grad_norm": 0.16785472631454468,
+      "learning_rate": 4.367275438482197e-06,
+      "loss": 0.9102,
+      "step": 87410
+    },
+    {
+      "epoch": 0.6328041868444483,
+      "grad_norm": 0.18528971076011658,
+      "learning_rate": 4.3672030518216106e-06,
+      "loss": 0.901,
+      "step": 87420
+    },
+    {
+      "epoch": 0.6328765735050345,
+      "grad_norm": 0.1553627848625183,
+      "learning_rate": 4.367130665161024e-06,
+      "loss": 0.9006,
+      "step": 87430
+    },
+    {
+      "epoch": 0.6329489601656206,
+      "grad_norm": 0.15704572200775146,
+      "learning_rate": 4.367058278500439e-06,
+      "loss": 0.8956,
+      "step": 87440
+    },
+    {
+      "epoch": 0.6330213468262069,
+      "grad_norm": 0.16988041996955872,
+      "learning_rate": 4.366985891839852e-06,
+      "loss": 0.8945,
+      "step": 87450
+    },
+    {
+      "epoch": 0.6330937334867931,
+      "grad_norm": 0.15681475400924683,
+      "learning_rate": 4.366913505179266e-06,
+      "loss": 0.8865,
+      "step": 87460
+    },
+    {
+      "epoch": 0.6331661201473793,
+      "grad_norm": 0.15510614216327667,
+      "learning_rate": 4.3668411185186795e-06,
+      "loss": 0.8965,
+      "step": 87470
+    },
+    {
+      "epoch": 0.6332385068079655,
+      "grad_norm": 0.16827818751335144,
+      "learning_rate": 4.366768731858094e-06,
+      "loss": 0.892,
+      "step": 87480
+    },
+    {
+      "epoch": 0.6333108934685516,
+      "grad_norm": 0.1560061275959015,
+      "learning_rate": 4.3666963451975076e-06,
+      "loss": 0.9041,
+      "step": 87490
+    },
+    {
+      "epoch": 0.6333832801291378,
+      "grad_norm": 0.15482322871685028,
+      "learning_rate": 4.366623958536921e-06,
+      "loss": 0.9012,
+      "step": 87500
+    },
+    {
+      "epoch": 0.633455666789724,
+      "grad_norm": 0.1474655419588089,
+      "learning_rate": 4.366551571876335e-06,
+      "loss": 0.896,
+      "step": 87510
+    },
+    {
+      "epoch": 0.6335280534503102,
+      "grad_norm": 0.15873880684375763,
+      "learning_rate": 4.366479185215749e-06,
+      "loss": 0.8961,
+      "step": 87520
+    },
+    {
+      "epoch": 0.6336004401108963,
+      "grad_norm": 0.2057919055223465,
+      "learning_rate": 4.366406798555163e-06,
+      "loss": 0.9079,
+      "step": 87530
+    },
+    {
+      "epoch": 0.6336728267714825,
+      "grad_norm": 0.1464945524930954,
+      "learning_rate": 4.3663344118945765e-06,
+      "loss": 0.9039,
+      "step": 87540
+    },
+    {
+      "epoch": 0.6337452134320687,
+      "grad_norm": 0.15978756546974182,
+      "learning_rate": 4.36626202523399e-06,
+      "loss": 0.9076,
+      "step": 87550
+    },
+    {
+      "epoch": 0.633817600092655,
+      "grad_norm": 0.15238073468208313,
+      "learning_rate": 4.366189638573405e-06,
+      "loss": 0.8772,
+      "step": 87560
+    },
+    {
+      "epoch": 0.6338899867532412,
+      "grad_norm": 0.1541658192873001,
+      "learning_rate": 4.366117251912818e-06,
+      "loss": 0.9014,
+      "step": 87570
+    },
+    {
+      "epoch": 0.6339623734138273,
+      "grad_norm": 0.15558382868766785,
+      "learning_rate": 4.366044865252232e-06,
+      "loss": 0.9151,
+      "step": 87580
+    },
+    {
+      "epoch": 0.6340347600744135,
+      "grad_norm": 0.22495369613170624,
+      "learning_rate": 4.3659724785916454e-06,
+      "loss": 0.9017,
+      "step": 87590
+    },
+    {
+      "epoch": 0.6341071467349997,
+      "grad_norm": 0.14666642248630524,
+      "learning_rate": 4.36590009193106e-06,
+      "loss": 0.906,
+      "step": 87600
+    },
+    {
+      "epoch": 0.6341795333955859,
+      "grad_norm": 0.1527988612651825,
+      "learning_rate": 4.365827705270473e-06,
+      "loss": 0.8978,
+      "step": 87610
+    },
+    {
+      "epoch": 0.634251920056172,
+      "grad_norm": 0.2135559469461441,
+      "learning_rate": 4.365755318609886e-06,
+      "loss": 0.9118,
+      "step": 87620
+    },
+    {
+      "epoch": 0.6343243067167582,
+      "grad_norm": 0.15013383328914642,
+      "learning_rate": 4.365682931949301e-06,
+      "loss": 0.8956,
+      "step": 87630
+    },
+    {
+      "epoch": 0.6343966933773444,
+      "grad_norm": 0.15739452838897705,
+      "learning_rate": 4.365610545288714e-06,
+      "loss": 0.8912,
+      "step": 87640
+    },
+    {
+      "epoch": 0.6344690800379306,
+      "grad_norm": 0.16141614317893982,
+      "learning_rate": 4.365538158628128e-06,
+      "loss": 0.9124,
+      "step": 87650
+    },
+    {
+      "epoch": 0.6345414666985167,
+      "grad_norm": 0.1528901308774948,
+      "learning_rate": 4.365465771967542e-06,
+      "loss": 0.903,
+      "step": 87660
+    },
+    {
+      "epoch": 0.634613853359103,
+      "grad_norm": 0.1768021136522293,
+      "learning_rate": 4.365393385306956e-06,
+      "loss": 0.8949,
+      "step": 87670
+    },
+    {
+      "epoch": 0.6346862400196892,
+      "grad_norm": 0.15449446439743042,
+      "learning_rate": 4.36532099864637e-06,
+      "loss": 0.9019,
+      "step": 87680
+    },
+    {
+      "epoch": 0.6347586266802754,
+      "grad_norm": 0.19363653659820557,
+      "learning_rate": 4.365248611985783e-06,
+      "loss": 0.9071,
+      "step": 87690
+    },
+    {
+      "epoch": 0.6348310133408616,
+      "grad_norm": 0.1683684140443802,
+      "learning_rate": 4.365176225325197e-06,
+      "loss": 0.895,
+      "step": 87700
+    },
+    {
+      "epoch": 0.6349034000014477,
+      "grad_norm": 0.15621653199195862,
+      "learning_rate": 4.365103838664611e-06,
+      "loss": 0.8939,
+      "step": 87710
+    },
+    {
+      "epoch": 0.6349757866620339,
+      "grad_norm": 0.16487151384353638,
+      "learning_rate": 4.365031452004025e-06,
+      "loss": 0.8915,
+      "step": 87720
+    },
+    {
+      "epoch": 0.6350481733226201,
+      "grad_norm": 0.1636609584093094,
+      "learning_rate": 4.364959065343439e-06,
+      "loss": 0.895,
+      "step": 87730
+    },
+    {
+      "epoch": 0.6351205599832063,
+      "grad_norm": 0.1616763025522232,
+      "learning_rate": 4.364886678682852e-06,
+      "loss": 0.9072,
+      "step": 87740
+    },
+    {
+      "epoch": 0.6351929466437924,
+      "grad_norm": 0.1616402268409729,
+      "learning_rate": 4.364814292022267e-06,
+      "loss": 0.8905,
+      "step": 87750
+    },
+    {
+      "epoch": 0.6352653333043786,
+      "grad_norm": 0.1978462189435959,
+      "learning_rate": 4.36474190536168e-06,
+      "loss": 0.9042,
+      "step": 87760
+    },
+    {
+      "epoch": 0.6353377199649649,
+      "grad_norm": 0.16519363224506378,
+      "learning_rate": 4.364669518701094e-06,
+      "loss": 0.8818,
+      "step": 87770
+    },
+    {
+      "epoch": 0.6354101066255511,
+      "grad_norm": 0.16044704616069794,
+      "learning_rate": 4.3645971320405075e-06,
+      "loss": 0.8968,
+      "step": 87780
+    },
+    {
+      "epoch": 0.6354824932861373,
+      "grad_norm": 0.16723746061325073,
+      "learning_rate": 4.364524745379922e-06,
+      "loss": 0.9008,
+      "step": 87790
+    },
+    {
+      "epoch": 0.6355548799467234,
+      "grad_norm": 0.15763086080551147,
+      "learning_rate": 4.364452358719336e-06,
+      "loss": 0.9046,
+      "step": 87800
+    },
+    {
+      "epoch": 0.6356272666073096,
+      "grad_norm": 0.16559143364429474,
+      "learning_rate": 4.364379972058749e-06,
+      "loss": 0.9074,
+      "step": 87810
+    },
+    {
+      "epoch": 0.6356996532678958,
+      "grad_norm": 0.17605692148208618,
+      "learning_rate": 4.364307585398163e-06,
+      "loss": 0.8845,
+      "step": 87820
+    },
+    {
+      "epoch": 0.635772039928482,
+      "grad_norm": 0.1688181757926941,
+      "learning_rate": 4.364235198737577e-06,
+      "loss": 0.9096,
+      "step": 87830
+    },
+    {
+      "epoch": 0.6358444265890681,
+      "grad_norm": 0.15427358448505402,
+      "learning_rate": 4.364162812076991e-06,
+      "loss": 0.9048,
+      "step": 87840
+    },
+    {
+      "epoch": 0.6359168132496543,
+      "grad_norm": 0.17729778587818146,
+      "learning_rate": 4.3640904254164045e-06,
+      "loss": 0.8993,
+      "step": 87850
+    },
+    {
+      "epoch": 0.6359891999102405,
+      "grad_norm": 0.17065225541591644,
+      "learning_rate": 4.364018038755818e-06,
+      "loss": 0.8954,
+      "step": 87860
+    },
+    {
+      "epoch": 0.6360615865708267,
+      "grad_norm": 0.15446516871452332,
+      "learning_rate": 4.363945652095233e-06,
+      "loss": 0.9051,
+      "step": 87870
+    },
+    {
+      "epoch": 0.636133973231413,
+      "grad_norm": 0.19369186460971832,
+      "learning_rate": 4.363873265434646e-06,
+      "loss": 0.8988,
+      "step": 87880
+    },
+    {
+      "epoch": 0.6362063598919991,
+      "grad_norm": 0.15911339223384857,
+      "learning_rate": 4.36380087877406e-06,
+      "loss": 0.9014,
+      "step": 87890
+    },
+    {
+      "epoch": 0.6362787465525853,
+      "grad_norm": 0.1556956171989441,
+      "learning_rate": 4.3637284921134735e-06,
+      "loss": 0.8952,
+      "step": 87900
+    },
+    {
+      "epoch": 0.6363511332131715,
+      "grad_norm": 0.176447331905365,
+      "learning_rate": 4.363656105452888e-06,
+      "loss": 0.8821,
+      "step": 87910
+    },
+    {
+      "epoch": 0.6364235198737577,
+      "grad_norm": 0.15189889073371887,
+      "learning_rate": 4.3635837187923015e-06,
+      "loss": 0.896,
+      "step": 87920
+    },
+    {
+      "epoch": 0.6364959065343438,
+      "grad_norm": 0.20345427095890045,
+      "learning_rate": 4.363511332131715e-06,
+      "loss": 0.9084,
+      "step": 87930
+    },
+    {
+      "epoch": 0.63656829319493,
+      "grad_norm": 0.1482352763414383,
+      "learning_rate": 4.363438945471129e-06,
+      "loss": 0.8937,
+      "step": 87940
+    },
+    {
+      "epoch": 0.6366406798555162,
+      "grad_norm": 0.16106894612312317,
+      "learning_rate": 4.363366558810543e-06,
+      "loss": 0.8992,
+      "step": 87950
+    },
+    {
+      "epoch": 0.6367130665161024,
+      "grad_norm": 0.16146771609783173,
+      "learning_rate": 4.363294172149957e-06,
+      "loss": 0.8968,
+      "step": 87960
+    },
+    {
+      "epoch": 0.6367854531766886,
+      "grad_norm": 0.15578705072402954,
+      "learning_rate": 4.3632217854893705e-06,
+      "loss": 0.9023,
+      "step": 87970
+    },
+    {
+      "epoch": 0.6368578398372748,
+      "grad_norm": 0.19550129771232605,
+      "learning_rate": 4.363149398828784e-06,
+      "loss": 0.9035,
+      "step": 87980
+    },
+    {
+      "epoch": 0.636930226497861,
+      "grad_norm": 0.18265287578105927,
+      "learning_rate": 4.3630770121681985e-06,
+      "loss": 0.9067,
+      "step": 87990
+    },
+    {
+      "epoch": 0.6370026131584472,
+      "grad_norm": 0.14970242977142334,
+      "learning_rate": 4.363004625507612e-06,
+      "loss": 0.9117,
+      "step": 88000
+    },
+    {
+      "epoch": 0.6370749998190334,
+      "grad_norm": 0.1472383439540863,
+      "learning_rate": 4.362932238847026e-06,
+      "loss": 0.8982,
+      "step": 88010
+    },
+    {
+      "epoch": 0.6371473864796195,
+      "grad_norm": 0.14995744824409485,
+      "learning_rate": 4.362859852186439e-06,
+      "loss": 0.8951,
+      "step": 88020
+    },
+    {
+      "epoch": 0.6372197731402057,
+      "grad_norm": 0.15421155095100403,
+      "learning_rate": 4.362787465525853e-06,
+      "loss": 0.8772,
+      "step": 88030
+    },
+    {
+      "epoch": 0.6372921598007919,
+      "grad_norm": 0.1862330138683319,
+      "learning_rate": 4.3627150788652675e-06,
+      "loss": 0.9068,
+      "step": 88040
+    },
+    {
+      "epoch": 0.6373645464613781,
+      "grad_norm": 0.15791213512420654,
+      "learning_rate": 4.362642692204681e-06,
+      "loss": 0.8989,
+      "step": 88050
+    },
+    {
+      "epoch": 0.6374369331219643,
+      "grad_norm": 0.1469462811946869,
+      "learning_rate": 4.362570305544095e-06,
+      "loss": 0.8983,
+      "step": 88060
+    },
+    {
+      "epoch": 0.6375093197825504,
+      "grad_norm": 0.15179064869880676,
+      "learning_rate": 4.362497918883508e-06,
+      "loss": 0.893,
+      "step": 88070
+    },
+    {
+      "epoch": 0.6375817064431366,
+      "grad_norm": 0.15898427367210388,
+      "learning_rate": 4.362425532222923e-06,
+      "loss": 0.8959,
+      "step": 88080
+    },
+    {
+      "epoch": 0.6376540931037229,
+      "grad_norm": 0.1530866026878357,
+      "learning_rate": 4.362353145562336e-06,
+      "loss": 0.8991,
+      "step": 88090
+    },
+    {
+      "epoch": 0.6377264797643091,
+      "grad_norm": 0.15585550665855408,
+      "learning_rate": 4.36228075890175e-06,
+      "loss": 0.8949,
+      "step": 88100
+    },
+    {
+      "epoch": 0.6377988664248952,
+      "grad_norm": 0.2193233221769333,
+      "learning_rate": 4.362208372241164e-06,
+      "loss": 0.9041,
+      "step": 88110
+    },
+    {
+      "epoch": 0.6378712530854814,
+      "grad_norm": 0.144108384847641,
+      "learning_rate": 4.362135985580578e-06,
+      "loss": 0.9028,
+      "step": 88120
+    },
+    {
+      "epoch": 0.6379436397460676,
+      "grad_norm": 0.18072602152824402,
+      "learning_rate": 4.362063598919992e-06,
+      "loss": 0.9006,
+      "step": 88130
+    },
+    {
+      "epoch": 0.6380160264066538,
+      "grad_norm": 0.15635623037815094,
+      "learning_rate": 4.361991212259405e-06,
+      "loss": 0.9059,
+      "step": 88140
+    },
+    {
+      "epoch": 0.63808841306724,
+      "grad_norm": 0.16746172308921814,
+      "learning_rate": 4.361918825598819e-06,
+      "loss": 0.8977,
+      "step": 88150
+    },
+    {
+      "epoch": 0.6381607997278261,
+      "grad_norm": 0.1613510549068451,
+      "learning_rate": 4.3618464389382326e-06,
+      "loss": 0.9092,
+      "step": 88160
+    },
+    {
+      "epoch": 0.6382331863884123,
+      "grad_norm": 0.1660555601119995,
+      "learning_rate": 4.361774052277646e-06,
+      "loss": 0.908,
+      "step": 88170
+    },
+    {
+      "epoch": 0.6383055730489985,
+      "grad_norm": 0.19197924435138702,
+      "learning_rate": 4.36170166561706e-06,
+      "loss": 0.8946,
+      "step": 88180
+    },
+    {
+      "epoch": 0.6383779597095847,
+      "grad_norm": 0.1495658904314041,
+      "learning_rate": 4.361629278956474e-06,
+      "loss": 0.9006,
+      "step": 88190
+    },
+    {
+      "epoch": 0.638450346370171,
+      "grad_norm": 0.15957580506801605,
+      "learning_rate": 4.361556892295888e-06,
+      "loss": 0.9024,
+      "step": 88200
+    },
+    {
+      "epoch": 0.6385227330307571,
+      "grad_norm": 0.1745164394378662,
+      "learning_rate": 4.3614845056353015e-06,
+      "loss": 0.8863,
+      "step": 88210
+    },
+    {
+      "epoch": 0.6385951196913433,
+      "grad_norm": 0.15032242238521576,
+      "learning_rate": 4.361412118974715e-06,
+      "loss": 0.8906,
+      "step": 88220
+    },
+    {
+      "epoch": 0.6386675063519295,
+      "grad_norm": 0.15983961522579193,
+      "learning_rate": 4.3613397323141296e-06,
+      "loss": 0.9114,
+      "step": 88230
+    },
+    {
+      "epoch": 0.6387398930125157,
+      "grad_norm": 0.21465584635734558,
+      "learning_rate": 4.361267345653543e-06,
+      "loss": 0.8914,
+      "step": 88240
+    },
+    {
+      "epoch": 0.6388122796731018,
+      "grad_norm": 0.15455862879753113,
+      "learning_rate": 4.361194958992957e-06,
+      "loss": 0.9126,
+      "step": 88250
+    },
+    {
+      "epoch": 0.638884666333688,
+      "grad_norm": 0.14769776165485382,
+      "learning_rate": 4.36112257233237e-06,
+      "loss": 0.8858,
+      "step": 88260
+    },
+    {
+      "epoch": 0.6389570529942742,
+      "grad_norm": 1.265124797821045,
+      "learning_rate": 4.361050185671785e-06,
+      "loss": 0.8948,
+      "step": 88270
+    },
+    {
+      "epoch": 0.6390294396548604,
+      "grad_norm": 0.1922004669904709,
+      "learning_rate": 4.3609777990111985e-06,
+      "loss": 0.9017,
+      "step": 88280
+    },
+    {
+      "epoch": 0.6391018263154465,
+      "grad_norm": 0.16533154249191284,
+      "learning_rate": 4.360905412350612e-06,
+      "loss": 0.8987,
+      "step": 88290
+    },
+    {
+      "epoch": 0.6391742129760328,
+      "grad_norm": 0.15893571078777313,
+      "learning_rate": 4.360833025690026e-06,
+      "loss": 0.8976,
+      "step": 88300
+    },
+    {
+      "epoch": 0.639246599636619,
+      "grad_norm": 0.150678813457489,
+      "learning_rate": 4.36076063902944e-06,
+      "loss": 0.894,
+      "step": 88310
+    },
+    {
+      "epoch": 0.6393189862972052,
+      "grad_norm": 0.14992178976535797,
+      "learning_rate": 4.360688252368854e-06,
+      "loss": 0.9037,
+      "step": 88320
+    },
+    {
+      "epoch": 0.6393913729577914,
+      "grad_norm": 0.1573602557182312,
+      "learning_rate": 4.3606158657082674e-06,
+      "loss": 0.9089,
+      "step": 88330
+    },
+    {
+      "epoch": 0.6394637596183775,
+      "grad_norm": 0.1497740000486374,
+      "learning_rate": 4.360543479047681e-06,
+      "loss": 0.9014,
+      "step": 88340
+    },
+    {
+      "epoch": 0.6395361462789637,
+      "grad_norm": 0.1509128361940384,
+      "learning_rate": 4.3604710923870955e-06,
+      "loss": 0.9033,
+      "step": 88350
+    },
+    {
+      "epoch": 0.6396085329395499,
+      "grad_norm": 0.15733207762241364,
+      "learning_rate": 4.360398705726509e-06,
+      "loss": 0.9074,
+      "step": 88360
+    },
+    {
+      "epoch": 0.6396809196001361,
+      "grad_norm": 0.16280068457126617,
+      "learning_rate": 4.360326319065923e-06,
+      "loss": 0.9005,
+      "step": 88370
+    },
+    {
+      "epoch": 0.6397533062607222,
+      "grad_norm": 0.15421195328235626,
+      "learning_rate": 4.360253932405336e-06,
+      "loss": 0.9046,
+      "step": 88380
+    },
+    {
+      "epoch": 0.6398256929213084,
+      "grad_norm": 0.22842425107955933,
+      "learning_rate": 4.360181545744751e-06,
+      "loss": 0.9101,
+      "step": 88390
+    },
+    {
+      "epoch": 0.6398980795818946,
+      "grad_norm": 0.1527772694826126,
+      "learning_rate": 4.3601091590841644e-06,
+      "loss": 0.9021,
+      "step": 88400
+    },
+    {
+      "epoch": 0.6399704662424809,
+      "grad_norm": 0.15494686365127563,
+      "learning_rate": 4.360036772423578e-06,
+      "loss": 0.8911,
+      "step": 88410
+    },
+    {
+      "epoch": 0.6400428529030671,
+      "grad_norm": 0.16015708446502686,
+      "learning_rate": 4.359964385762992e-06,
+      "loss": 0.916,
+      "step": 88420
+    },
+    {
+      "epoch": 0.6401152395636532,
+      "grad_norm": 0.15892557799816132,
+      "learning_rate": 4.359891999102406e-06,
+      "loss": 0.8768,
+      "step": 88430
+    },
+    {
+      "epoch": 0.6401876262242394,
+      "grad_norm": 0.1498948186635971,
+      "learning_rate": 4.35981961244182e-06,
+      "loss": 0.9047,
+      "step": 88440
+    },
+    {
+      "epoch": 0.6402600128848256,
+      "grad_norm": 0.1665375530719757,
+      "learning_rate": 4.359747225781233e-06,
+      "loss": 0.8941,
+      "step": 88450
+    },
+    {
+      "epoch": 0.6403323995454118,
+      "grad_norm": 0.15249140560626984,
+      "learning_rate": 4.359674839120647e-06,
+      "loss": 0.8964,
+      "step": 88460
+    },
+    {
+      "epoch": 0.640404786205998,
+      "grad_norm": 0.1530652791261673,
+      "learning_rate": 4.3596024524600614e-06,
+      "loss": 0.9092,
+      "step": 88470
+    },
+    {
+      "epoch": 0.6404771728665841,
+      "grad_norm": 0.1995309293270111,
+      "learning_rate": 4.359530065799475e-06,
+      "loss": 0.8985,
+      "step": 88480
+    },
+    {
+      "epoch": 0.6405495595271703,
+      "grad_norm": 0.17197419703006744,
+      "learning_rate": 4.359457679138889e-06,
+      "loss": 0.9123,
+      "step": 88490
+    },
+    {
+      "epoch": 0.6406219461877565,
+      "grad_norm": 0.17037421464920044,
+      "learning_rate": 4.359385292478302e-06,
+      "loss": 0.8954,
+      "step": 88500
+    },
+    {
+      "epoch": 0.6406943328483428,
+      "grad_norm": 0.14726117253303528,
+      "learning_rate": 4.359312905817717e-06,
+      "loss": 0.9045,
+      "step": 88510
+    },
+    {
+      "epoch": 0.6407667195089289,
+      "grad_norm": 0.14530207216739655,
+      "learning_rate": 4.35924051915713e-06,
+      "loss": 0.8884,
+      "step": 88520
+    },
+    {
+      "epoch": 0.6408391061695151,
+      "grad_norm": 0.15528494119644165,
+      "learning_rate": 4.359168132496544e-06,
+      "loss": 0.8854,
+      "step": 88530
+    },
+    {
+      "epoch": 0.6409114928301013,
+      "grad_norm": 0.16812381148338318,
+      "learning_rate": 4.359095745835958e-06,
+      "loss": 0.9137,
+      "step": 88540
+    },
+    {
+      "epoch": 0.6409838794906875,
+      "grad_norm": 0.15406684577465057,
+      "learning_rate": 4.359023359175372e-06,
+      "loss": 0.9013,
+      "step": 88550
+    },
+    {
+      "epoch": 0.6410562661512736,
+      "grad_norm": 0.15804636478424072,
+      "learning_rate": 4.358950972514786e-06,
+      "loss": 0.903,
+      "step": 88560
+    },
+    {
+      "epoch": 0.6411286528118598,
+      "grad_norm": 0.15636689960956573,
+      "learning_rate": 4.358878585854199e-06,
+      "loss": 0.9006,
+      "step": 88570
+    },
+    {
+      "epoch": 0.641201039472446,
+      "grad_norm": 0.18669186532497406,
+      "learning_rate": 4.358806199193613e-06,
+      "loss": 0.9171,
+      "step": 88580
+    },
+    {
+      "epoch": 0.6412734261330322,
+      "grad_norm": 0.17967639863491058,
+      "learning_rate": 4.358733812533027e-06,
+      "loss": 0.8925,
+      "step": 88590
+    },
+    {
+      "epoch": 0.6413458127936184,
+      "grad_norm": 0.15877005457878113,
+      "learning_rate": 4.358661425872441e-06,
+      "loss": 0.8923,
+      "step": 88600
+    },
+    {
+      "epoch": 0.6414181994542045,
+      "grad_norm": 0.1480860710144043,
+      "learning_rate": 4.358589039211855e-06,
+      "loss": 0.9054,
+      "step": 88610
+    },
+    {
+      "epoch": 0.6414905861147908,
+      "grad_norm": 0.15211063623428345,
+      "learning_rate": 4.358516652551268e-06,
+      "loss": 0.9001,
+      "step": 88620
+    },
+    {
+      "epoch": 0.641562972775377,
+      "grad_norm": 0.1518232822418213,
+      "learning_rate": 4.358444265890682e-06,
+      "loss": 0.9031,
+      "step": 88630
+    },
+    {
+      "epoch": 0.6416353594359632,
+      "grad_norm": 0.1446765810251236,
+      "learning_rate": 4.358371879230096e-06,
+      "loss": 0.9166,
+      "step": 88640
+    },
+    {
+      "epoch": 0.6417077460965493,
+      "grad_norm": 0.17465344071388245,
+      "learning_rate": 4.35829949256951e-06,
+      "loss": 0.8882,
+      "step": 88650
+    },
+    {
+      "epoch": 0.6417801327571355,
+      "grad_norm": 0.1594177931547165,
+      "learning_rate": 4.3582271059089235e-06,
+      "loss": 0.9004,
+      "step": 88660
+    },
+    {
+      "epoch": 0.6418525194177217,
+      "grad_norm": 0.1680103987455368,
+      "learning_rate": 4.358154719248337e-06,
+      "loss": 0.891,
+      "step": 88670
+    },
+    {
+      "epoch": 0.6419249060783079,
+      "grad_norm": 0.1576722413301468,
+      "learning_rate": 4.358082332587751e-06,
+      "loss": 0.9137,
+      "step": 88680
+    },
+    {
+      "epoch": 0.641997292738894,
+      "grad_norm": 0.14920315146446228,
+      "learning_rate": 4.358009945927164e-06,
+      "loss": 0.8867,
+      "step": 88690
+    },
+    {
+      "epoch": 0.6420696793994802,
+      "grad_norm": 0.15702413022518158,
+      "learning_rate": 4.357937559266579e-06,
+      "loss": 0.8899,
+      "step": 88700
+    },
+    {
+      "epoch": 0.6421420660600664,
+      "grad_norm": 0.14902466535568237,
+      "learning_rate": 4.3578651726059925e-06,
+      "loss": 0.9046,
+      "step": 88710
+    },
+    {
+      "epoch": 0.6422144527206526,
+      "grad_norm": 0.1598339080810547,
+      "learning_rate": 4.357792785945406e-06,
+      "loss": 0.896,
+      "step": 88720
+    },
+    {
+      "epoch": 0.6422868393812389,
+      "grad_norm": 0.15564772486686707,
+      "learning_rate": 4.35772039928482e-06,
+      "loss": 0.9102,
+      "step": 88730
+    },
+    {
+      "epoch": 0.642359226041825,
+      "grad_norm": 0.14430661499500275,
+      "learning_rate": 4.357648012624234e-06,
+      "loss": 0.8899,
+      "step": 88740
+    },
+    {
+      "epoch": 0.6424316127024112,
+      "grad_norm": 0.15602971613407135,
+      "learning_rate": 4.357575625963648e-06,
+      "loss": 0.8982,
+      "step": 88750
+    },
+    {
+      "epoch": 0.6425039993629974,
+      "grad_norm": 0.16379621624946594,
+      "learning_rate": 4.357503239303061e-06,
+      "loss": 0.9031,
+      "step": 88760
+    },
+    {
+      "epoch": 0.6425763860235836,
+      "grad_norm": 0.15604230761528015,
+      "learning_rate": 4.357430852642475e-06,
+      "loss": 0.8916,
+      "step": 88770
+    },
+    {
+      "epoch": 0.6426487726841698,
+      "grad_norm": 0.15678240358829498,
+      "learning_rate": 4.3573584659818895e-06,
+      "loss": 0.8947,
+      "step": 88780
+    },
+    {
+      "epoch": 0.6427211593447559,
+      "grad_norm": 0.1667027622461319,
+      "learning_rate": 4.357286079321303e-06,
+      "loss": 0.8967,
+      "step": 88790
+    },
+    {
+      "epoch": 0.6427935460053421,
+      "grad_norm": 0.1623423546552658,
+      "learning_rate": 4.357213692660717e-06,
+      "loss": 0.8882,
+      "step": 88800
+    },
+    {
+      "epoch": 0.6428659326659283,
+      "grad_norm": 0.15376125276088715,
+      "learning_rate": 4.35714130600013e-06,
+      "loss": 0.8956,
+      "step": 88810
+    },
+    {
+      "epoch": 0.6429383193265145,
+      "grad_norm": 0.1492316871881485,
+      "learning_rate": 4.357068919339544e-06,
+      "loss": 0.8974,
+      "step": 88820
+    },
+    {
+      "epoch": 0.6430107059871008,
+      "grad_norm": 0.1571977287530899,
+      "learning_rate": 4.356996532678958e-06,
+      "loss": 0.9096,
+      "step": 88830
+    },
+    {
+      "epoch": 0.6430830926476869,
+      "grad_norm": 0.1929909884929657,
+      "learning_rate": 4.356924146018372e-06,
+      "loss": 0.8849,
+      "step": 88840
+    },
+    {
+      "epoch": 0.6431554793082731,
+      "grad_norm": 0.15631908178329468,
+      "learning_rate": 4.356851759357786e-06,
+      "loss": 0.9151,
+      "step": 88850
+    },
+    {
+      "epoch": 0.6432278659688593,
+      "grad_norm": 0.14949972927570343,
+      "learning_rate": 4.356779372697199e-06,
+      "loss": 0.8962,
+      "step": 88860
+    },
+    {
+      "epoch": 0.6433002526294455,
+      "grad_norm": 0.16994498670101166,
+      "learning_rate": 4.356706986036614e-06,
+      "loss": 0.8988,
+      "step": 88870
+    },
+    {
+      "epoch": 0.6433726392900316,
+      "grad_norm": 0.16940979659557343,
+      "learning_rate": 4.356634599376027e-06,
+      "loss": 0.9085,
+      "step": 88880
+    },
+    {
+      "epoch": 0.6434450259506178,
+      "grad_norm": 0.15808720886707306,
+      "learning_rate": 4.356562212715441e-06,
+      "loss": 0.9043,
+      "step": 88890
+    },
+    {
+      "epoch": 0.643517412611204,
+      "grad_norm": 0.15764841437339783,
+      "learning_rate": 4.3564898260548546e-06,
+      "loss": 0.8932,
+      "step": 88900
+    },
+    {
+      "epoch": 0.6435897992717902,
+      "grad_norm": 0.15186621248722076,
+      "learning_rate": 4.356417439394269e-06,
+      "loss": 0.9061,
+      "step": 88910
+    },
+    {
+      "epoch": 0.6436621859323763,
+      "grad_norm": 0.14821207523345947,
+      "learning_rate": 4.356345052733683e-06,
+      "loss": 0.8926,
+      "step": 88920
+    },
+    {
+      "epoch": 0.6437345725929625,
+      "grad_norm": 0.15951776504516602,
+      "learning_rate": 4.356272666073096e-06,
+      "loss": 0.887,
+      "step": 88930
+    },
+    {
+      "epoch": 0.6438069592535488,
+      "grad_norm": 0.1618291735649109,
+      "learning_rate": 4.35620027941251e-06,
+      "loss": 0.8978,
+      "step": 88940
+    },
+    {
+      "epoch": 0.643879345914135,
+      "grad_norm": 0.1718137413263321,
+      "learning_rate": 4.356127892751924e-06,
+      "loss": 0.902,
+      "step": 88950
+    },
+    {
+      "epoch": 0.6439517325747212,
+      "grad_norm": 0.1487615555524826,
+      "learning_rate": 4.356055506091338e-06,
+      "loss": 0.897,
+      "step": 88960
+    },
+    {
+      "epoch": 0.6440241192353073,
+      "grad_norm": 0.1924024522304535,
+      "learning_rate": 4.3559831194307516e-06,
+      "loss": 0.9099,
+      "step": 88970
+    },
+    {
+      "epoch": 0.6440965058958935,
+      "grad_norm": 0.15762527287006378,
+      "learning_rate": 4.355910732770165e-06,
+      "loss": 0.891,
+      "step": 88980
+    },
+    {
+      "epoch": 0.6441688925564797,
+      "grad_norm": 0.16442440450191498,
+      "learning_rate": 4.35583834610958e-06,
+      "loss": 0.8914,
+      "step": 88990
+    },
+    {
+      "epoch": 0.6442412792170659,
+      "grad_norm": 0.15198470652103424,
+      "learning_rate": 4.355765959448993e-06,
+      "loss": 0.9036,
+      "step": 89000
+    },
+    {
+      "epoch": 0.644313665877652,
+      "grad_norm": 0.16357183456420898,
+      "learning_rate": 4.355693572788407e-06,
+      "loss": 0.9018,
+      "step": 89010
+    },
+    {
+      "epoch": 0.6443860525382382,
+      "grad_norm": 0.15125641226768494,
+      "learning_rate": 4.3556211861278205e-06,
+      "loss": 0.9005,
+      "step": 89020
+    },
+    {
+      "epoch": 0.6444584391988244,
+      "grad_norm": 0.1542976200580597,
+      "learning_rate": 4.355548799467235e-06,
+      "loss": 0.9062,
+      "step": 89030
+    },
+    {
+      "epoch": 0.6445308258594107,
+      "grad_norm": 0.16405089199543,
+      "learning_rate": 4.3554764128066486e-06,
+      "loss": 0.8959,
+      "step": 89040
+    },
+    {
+      "epoch": 0.6446032125199969,
+      "grad_norm": 0.15201352536678314,
+      "learning_rate": 4.355404026146062e-06,
+      "loss": 0.8916,
+      "step": 89050
+    },
+    {
+      "epoch": 0.644675599180583,
+      "grad_norm": 0.1623867303133011,
+      "learning_rate": 4.355331639485476e-06,
+      "loss": 0.889,
+      "step": 89060
+    },
+    {
+      "epoch": 0.6447479858411692,
+      "grad_norm": 0.1591363102197647,
+      "learning_rate": 4.35525925282489e-06,
+      "loss": 0.9057,
+      "step": 89070
+    },
+    {
+      "epoch": 0.6448203725017554,
+      "grad_norm": 0.16368667781352997,
+      "learning_rate": 4.355186866164304e-06,
+      "loss": 0.8983,
+      "step": 89080
+    },
+    {
+      "epoch": 0.6448927591623416,
+      "grad_norm": 0.15838485956192017,
+      "learning_rate": 4.3551144795037175e-06,
+      "loss": 0.8975,
+      "step": 89090
+    },
+    {
+      "epoch": 0.6449651458229277,
+      "grad_norm": 0.14503420889377594,
+      "learning_rate": 4.355042092843131e-06,
+      "loss": 0.8942,
+      "step": 89100
+    },
+    {
+      "epoch": 0.6450375324835139,
+      "grad_norm": 0.1557312160730362,
+      "learning_rate": 4.354969706182546e-06,
+      "loss": 0.8893,
+      "step": 89110
+    },
+    {
+      "epoch": 0.6451099191441001,
+      "grad_norm": 0.16098271310329437,
+      "learning_rate": 4.354897319521959e-06,
+      "loss": 0.9019,
+      "step": 89120
+    },
+    {
+      "epoch": 0.6451823058046863,
+      "grad_norm": 0.2473098188638687,
+      "learning_rate": 4.354824932861373e-06,
+      "loss": 0.8977,
+      "step": 89130
+    },
+    {
+      "epoch": 0.6452546924652725,
+      "grad_norm": 0.15116575360298157,
+      "learning_rate": 4.3547525462007864e-06,
+      "loss": 0.896,
+      "step": 89140
+    },
+    {
+      "epoch": 0.6453270791258587,
+      "grad_norm": 0.16768385469913483,
+      "learning_rate": 4.354680159540201e-06,
+      "loss": 0.8977,
+      "step": 89150
+    },
+    {
+      "epoch": 0.6453994657864449,
+      "grad_norm": 0.15651331841945648,
+      "learning_rate": 4.3546077728796145e-06,
+      "loss": 0.8983,
+      "step": 89160
+    },
+    {
+      "epoch": 0.6454718524470311,
+      "grad_norm": 0.18052604794502258,
+      "learning_rate": 4.354535386219028e-06,
+      "loss": 0.9187,
+      "step": 89170
+    },
+    {
+      "epoch": 0.6455442391076173,
+      "grad_norm": 0.15085764229297638,
+      "learning_rate": 4.354462999558442e-06,
+      "loss": 0.8947,
+      "step": 89180
+    },
+    {
+      "epoch": 0.6456166257682034,
+      "grad_norm": 0.15222260355949402,
+      "learning_rate": 4.354390612897856e-06,
+      "loss": 0.8996,
+      "step": 89190
+    },
+    {
+      "epoch": 0.6456890124287896,
+      "grad_norm": 0.16957657039165497,
+      "learning_rate": 4.354318226237269e-06,
+      "loss": 0.9059,
+      "step": 89200
+    },
+    {
+      "epoch": 0.6457613990893758,
+      "grad_norm": 0.17988649010658264,
+      "learning_rate": 4.354245839576683e-06,
+      "loss": 0.9038,
+      "step": 89210
+    },
+    {
+      "epoch": 0.645833785749962,
+      "grad_norm": 0.15836915373802185,
+      "learning_rate": 4.354173452916097e-06,
+      "loss": 0.9061,
+      "step": 89220
+    },
+    {
+      "epoch": 0.6459061724105482,
+      "grad_norm": 0.16917668282985687,
+      "learning_rate": 4.354101066255511e-06,
+      "loss": 0.9056,
+      "step": 89230
+    },
+    {
+      "epoch": 0.6459785590711343,
+      "grad_norm": 0.17027825117111206,
+      "learning_rate": 4.354028679594924e-06,
+      "loss": 0.8929,
+      "step": 89240
+    },
+    {
+      "epoch": 0.6460509457317205,
+      "grad_norm": 0.15539249777793884,
+      "learning_rate": 4.353956292934338e-06,
+      "loss": 0.8926,
+      "step": 89250
+    },
+    {
+      "epoch": 0.6461233323923068,
+      "grad_norm": 0.17285488545894623,
+      "learning_rate": 4.353883906273752e-06,
+      "loss": 0.9006,
+      "step": 89260
+    },
+    {
+      "epoch": 0.646195719052893,
+      "grad_norm": 0.15279121696949005,
+      "learning_rate": 4.353811519613166e-06,
+      "loss": 0.9059,
+      "step": 89270
+    },
+    {
+      "epoch": 0.6462681057134791,
+      "grad_norm": 0.16332434117794037,
+      "learning_rate": 4.35373913295258e-06,
+      "loss": 0.8942,
+      "step": 89280
+    },
+    {
+      "epoch": 0.6463404923740653,
+      "grad_norm": 0.15591338276863098,
+      "learning_rate": 4.353666746291993e-06,
+      "loss": 0.8902,
+      "step": 89290
+    },
+    {
+      "epoch": 0.6464128790346515,
+      "grad_norm": 0.15603341162204742,
+      "learning_rate": 4.353594359631408e-06,
+      "loss": 0.8933,
+      "step": 89300
+    },
+    {
+      "epoch": 0.6464852656952377,
+      "grad_norm": 0.17558693885803223,
+      "learning_rate": 4.353521972970821e-06,
+      "loss": 0.8918,
+      "step": 89310
+    },
+    {
+      "epoch": 0.6465576523558239,
+      "grad_norm": 0.1703271120786667,
+      "learning_rate": 4.353449586310235e-06,
+      "loss": 0.8969,
+      "step": 89320
+    },
+    {
+      "epoch": 0.64663003901641,
+      "grad_norm": 0.15776929259300232,
+      "learning_rate": 4.3533771996496485e-06,
+      "loss": 0.8974,
+      "step": 89330
+    },
+    {
+      "epoch": 0.6467024256769962,
+      "grad_norm": 0.1533646285533905,
+      "learning_rate": 4.353304812989063e-06,
+      "loss": 0.8994,
+      "step": 89340
+    },
+    {
+      "epoch": 0.6467748123375824,
+      "grad_norm": 0.3359949290752411,
+      "learning_rate": 4.353232426328477e-06,
+      "loss": 0.9038,
+      "step": 89350
+    },
+    {
+      "epoch": 0.6468471989981687,
+      "grad_norm": 0.1611877977848053,
+      "learning_rate": 4.35316003966789e-06,
+      "loss": 0.8968,
+      "step": 89360
+    },
+    {
+      "epoch": 0.6469195856587548,
+      "grad_norm": 0.1490904539823532,
+      "learning_rate": 4.353087653007304e-06,
+      "loss": 0.8771,
+      "step": 89370
+    },
+    {
+      "epoch": 0.646991972319341,
+      "grad_norm": 0.15412895381450653,
+      "learning_rate": 4.353015266346718e-06,
+      "loss": 0.9061,
+      "step": 89380
+    },
+    {
+      "epoch": 0.6470643589799272,
+      "grad_norm": 0.16382527351379395,
+      "learning_rate": 4.352942879686132e-06,
+      "loss": 0.8989,
+      "step": 89390
+    },
+    {
+      "epoch": 0.6471367456405134,
+      "grad_norm": 0.15625497698783875,
+      "learning_rate": 4.3528704930255455e-06,
+      "loss": 0.9106,
+      "step": 89400
+    },
+    {
+      "epoch": 0.6472091323010996,
+      "grad_norm": 0.149552121758461,
+      "learning_rate": 4.352798106364959e-06,
+      "loss": 0.9041,
+      "step": 89410
+    },
+    {
+      "epoch": 0.6472815189616857,
+      "grad_norm": 0.16269992291927338,
+      "learning_rate": 4.352725719704373e-06,
+      "loss": 0.9069,
+      "step": 89420
+    },
+    {
+      "epoch": 0.6473539056222719,
+      "grad_norm": 0.15287677943706512,
+      "learning_rate": 4.352653333043787e-06,
+      "loss": 0.8955,
+      "step": 89430
+    },
+    {
+      "epoch": 0.6474262922828581,
+      "grad_norm": 0.1846160590648651,
+      "learning_rate": 4.352580946383201e-06,
+      "loss": 0.8977,
+      "step": 89440
+    },
+    {
+      "epoch": 0.6474986789434443,
+      "grad_norm": 0.15473540127277374,
+      "learning_rate": 4.3525085597226145e-06,
+      "loss": 0.8942,
+      "step": 89450
+    },
+    {
+      "epoch": 0.6475710656040304,
+      "grad_norm": 0.16567891836166382,
+      "learning_rate": 4.352436173062028e-06,
+      "loss": 0.9138,
+      "step": 89460
+    },
+    {
+      "epoch": 0.6476434522646167,
+      "grad_norm": 0.15014810860157013,
+      "learning_rate": 4.3523637864014425e-06,
+      "loss": 0.9014,
+      "step": 89470
+    },
+    {
+      "epoch": 0.6477158389252029,
+      "grad_norm": 0.15278102457523346,
+      "learning_rate": 4.352291399740856e-06,
+      "loss": 0.8891,
+      "step": 89480
+    },
+    {
+      "epoch": 0.6477882255857891,
+      "grad_norm": 0.15812081098556519,
+      "learning_rate": 4.35221901308027e-06,
+      "loss": 0.9111,
+      "step": 89490
+    },
+    {
+      "epoch": 0.6478606122463753,
+      "grad_norm": 0.15760880708694458,
+      "learning_rate": 4.352146626419683e-06,
+      "loss": 0.8998,
+      "step": 89500
+    },
+    {
+      "epoch": 0.6479329989069614,
+      "grad_norm": 0.15096792578697205,
+      "learning_rate": 4.352074239759098e-06,
+      "loss": 0.8996,
+      "step": 89510
+    },
+    {
+      "epoch": 0.6480053855675476,
+      "grad_norm": 0.16857214272022247,
+      "learning_rate": 4.3520018530985115e-06,
+      "loss": 0.8953,
+      "step": 89520
+    },
+    {
+      "epoch": 0.6480777722281338,
+      "grad_norm": 0.16111542284488678,
+      "learning_rate": 4.351929466437925e-06,
+      "loss": 0.8887,
+      "step": 89530
+    },
+    {
+      "epoch": 0.64815015888872,
+      "grad_norm": 0.15649932622909546,
+      "learning_rate": 4.351857079777339e-06,
+      "loss": 0.8869,
+      "step": 89540
+    },
+    {
+      "epoch": 0.6482225455493061,
+      "grad_norm": 0.1554279923439026,
+      "learning_rate": 4.351784693116753e-06,
+      "loss": 0.8913,
+      "step": 89550
+    },
+    {
+      "epoch": 0.6482949322098923,
+      "grad_norm": 0.15444056689739227,
+      "learning_rate": 4.351712306456167e-06,
+      "loss": 0.8978,
+      "step": 89560
+    },
+    {
+      "epoch": 0.6483673188704785,
+      "grad_norm": 0.16410863399505615,
+      "learning_rate": 4.35163991979558e-06,
+      "loss": 0.9021,
+      "step": 89570
+    },
+    {
+      "epoch": 0.6484397055310648,
+      "grad_norm": 0.181893453001976,
+      "learning_rate": 4.351567533134994e-06,
+      "loss": 0.8952,
+      "step": 89580
+    },
+    {
+      "epoch": 0.648512092191651,
+      "grad_norm": 0.15560483932495117,
+      "learning_rate": 4.3514951464744085e-06,
+      "loss": 0.8824,
+      "step": 89590
+    },
+    {
+      "epoch": 0.6485844788522371,
+      "grad_norm": 0.16299884021282196,
+      "learning_rate": 4.351422759813822e-06,
+      "loss": 0.8884,
+      "step": 89600
+    },
+    {
+      "epoch": 0.6486568655128233,
+      "grad_norm": 0.16149957478046417,
+      "learning_rate": 4.351350373153236e-06,
+      "loss": 0.8942,
+      "step": 89610
+    },
+    {
+      "epoch": 0.6487292521734095,
+      "grad_norm": 0.15844495594501495,
+      "learning_rate": 4.351277986492649e-06,
+      "loss": 0.8916,
+      "step": 89620
+    },
+    {
+      "epoch": 0.6488016388339957,
+      "grad_norm": 0.15929342806339264,
+      "learning_rate": 4.351205599832064e-06,
+      "loss": 0.9046,
+      "step": 89630
+    },
+    {
+      "epoch": 0.6488740254945818,
+      "grad_norm": 0.14573869109153748,
+      "learning_rate": 4.351133213171477e-06,
+      "loss": 0.8981,
+      "step": 89640
+    },
+    {
+      "epoch": 0.648946412155168,
+      "grad_norm": 0.14629115164279938,
+      "learning_rate": 4.351060826510891e-06,
+      "loss": 0.9027,
+      "step": 89650
+    },
+    {
+      "epoch": 0.6490187988157542,
+      "grad_norm": 0.16613242030143738,
+      "learning_rate": 4.350988439850305e-06,
+      "loss": 0.8883,
+      "step": 89660
+    },
+    {
+      "epoch": 0.6490911854763404,
+      "grad_norm": 0.18370899558067322,
+      "learning_rate": 4.350916053189719e-06,
+      "loss": 0.8997,
+      "step": 89670
+    },
+    {
+      "epoch": 0.6491635721369267,
+      "grad_norm": 0.15053485333919525,
+      "learning_rate": 4.350843666529133e-06,
+      "loss": 0.8909,
+      "step": 89680
+    },
+    {
+      "epoch": 0.6492359587975128,
+      "grad_norm": 0.1765306442975998,
+      "learning_rate": 4.350771279868546e-06,
+      "loss": 0.8805,
+      "step": 89690
+    },
+    {
+      "epoch": 0.649308345458099,
+      "grad_norm": 0.1697237193584442,
+      "learning_rate": 4.35069889320796e-06,
+      "loss": 0.9051,
+      "step": 89700
+    },
+    {
+      "epoch": 0.6493807321186852,
+      "grad_norm": 0.15003421902656555,
+      "learning_rate": 4.350626506547374e-06,
+      "loss": 0.9008,
+      "step": 89710
+    },
+    {
+      "epoch": 0.6494531187792714,
+      "grad_norm": 0.15888899564743042,
+      "learning_rate": 4.350554119886788e-06,
+      "loss": 0.9043,
+      "step": 89720
+    },
+    {
+      "epoch": 0.6495255054398575,
+      "grad_norm": 0.15244218707084656,
+      "learning_rate": 4.350481733226202e-06,
+      "loss": 0.898,
+      "step": 89730
+    },
+    {
+      "epoch": 0.6495978921004437,
+      "grad_norm": 0.1715979427099228,
+      "learning_rate": 4.350409346565615e-06,
+      "loss": 0.8955,
+      "step": 89740
+    },
+    {
+      "epoch": 0.6496702787610299,
+      "grad_norm": 0.15656600892543793,
+      "learning_rate": 4.350336959905029e-06,
+      "loss": 0.8979,
+      "step": 89750
+    },
+    {
+      "epoch": 0.6497426654216161,
+      "grad_norm": 0.1492892950773239,
+      "learning_rate": 4.3502645732444425e-06,
+      "loss": 0.8935,
+      "step": 89760
+    },
+    {
+      "epoch": 0.6498150520822022,
+      "grad_norm": 0.15817688405513763,
+      "learning_rate": 4.350192186583856e-06,
+      "loss": 0.8975,
+      "step": 89770
+    },
+    {
+      "epoch": 0.6498874387427884,
+      "grad_norm": 0.17220188677310944,
+      "learning_rate": 4.3501197999232706e-06,
+      "loss": 0.904,
+      "step": 89780
+    },
+    {
+      "epoch": 0.6499598254033747,
+      "grad_norm": 0.15756267309188843,
+      "learning_rate": 4.350047413262684e-06,
+      "loss": 0.8945,
+      "step": 89790
+    },
+    {
+      "epoch": 0.6500322120639609,
+      "grad_norm": 0.14775143563747406,
+      "learning_rate": 4.349975026602098e-06,
+      "loss": 0.892,
+      "step": 89800
+    },
+    {
+      "epoch": 0.6501045987245471,
+      "grad_norm": 0.14889222383499146,
+      "learning_rate": 4.3499026399415114e-06,
+      "loss": 0.883,
+      "step": 89810
+    },
+    {
+      "epoch": 0.6501769853851332,
+      "grad_norm": 0.15578392148017883,
+      "learning_rate": 4.349830253280926e-06,
+      "loss": 0.8863,
+      "step": 89820
+    },
+    {
+      "epoch": 0.6502493720457194,
+      "grad_norm": 0.15579867362976074,
+      "learning_rate": 4.3497578666203395e-06,
+      "loss": 0.9008,
+      "step": 89830
+    },
+    {
+      "epoch": 0.6503217587063056,
+      "grad_norm": 0.1711921989917755,
+      "learning_rate": 4.349685479959753e-06,
+      "loss": 0.9018,
+      "step": 89840
+    },
+    {
+      "epoch": 0.6503941453668918,
+      "grad_norm": 0.17373885214328766,
+      "learning_rate": 4.349613093299167e-06,
+      "loss": 0.8876,
+      "step": 89850
+    },
+    {
+      "epoch": 0.650466532027478,
+      "grad_norm": 0.15983055531978607,
+      "learning_rate": 4.349540706638581e-06,
+      "loss": 0.9056,
+      "step": 89860
+    },
+    {
+      "epoch": 0.6505389186880641,
+      "grad_norm": 0.17014628648757935,
+      "learning_rate": 4.349468319977995e-06,
+      "loss": 0.8816,
+      "step": 89870
+    },
+    {
+      "epoch": 0.6506113053486503,
+      "grad_norm": 0.15592381358146667,
+      "learning_rate": 4.3493959333174084e-06,
+      "loss": 0.8784,
+      "step": 89880
+    },
+    {
+      "epoch": 0.6506836920092366,
+      "grad_norm": 0.16757424175739288,
+      "learning_rate": 4.349323546656822e-06,
+      "loss": 0.8985,
+      "step": 89890
+    },
+    {
+      "epoch": 0.6507560786698228,
+      "grad_norm": 0.15414102375507355,
+      "learning_rate": 4.3492511599962365e-06,
+      "loss": 0.8938,
+      "step": 89900
+    },
+    {
+      "epoch": 0.650828465330409,
+      "grad_norm": 0.16440652310848236,
+      "learning_rate": 4.34917877333565e-06,
+      "loss": 0.9051,
+      "step": 89910
+    },
+    {
+      "epoch": 0.6509008519909951,
+      "grad_norm": 0.17210443317890167,
+      "learning_rate": 4.349106386675064e-06,
+      "loss": 0.8878,
+      "step": 89920
+    },
+    {
+      "epoch": 0.6509732386515813,
+      "grad_norm": 0.17794910073280334,
+      "learning_rate": 4.349034000014477e-06,
+      "loss": 0.9026,
+      "step": 89930
+    },
+    {
+      "epoch": 0.6510456253121675,
+      "grad_norm": 0.15437255799770355,
+      "learning_rate": 4.348961613353892e-06,
+      "loss": 0.9072,
+      "step": 89940
+    },
+    {
+      "epoch": 0.6511180119727537,
+      "grad_norm": 0.15506230294704437,
+      "learning_rate": 4.3488892266933054e-06,
+      "loss": 0.8998,
+      "step": 89950
+    },
+    {
+      "epoch": 0.6511903986333398,
+      "grad_norm": 0.1564485728740692,
+      "learning_rate": 4.348816840032719e-06,
+      "loss": 0.8746,
+      "step": 89960
+    },
+    {
+      "epoch": 0.651262785293926,
+      "grad_norm": 0.15721721947193146,
+      "learning_rate": 4.348744453372133e-06,
+      "loss": 0.8953,
+      "step": 89970
+    },
+    {
+      "epoch": 0.6513351719545122,
+      "grad_norm": 0.16001653671264648,
+      "learning_rate": 4.348672066711547e-06,
+      "loss": 0.9131,
+      "step": 89980
+    },
+    {
+      "epoch": 0.6514075586150984,
+      "grad_norm": 0.15154358744621277,
+      "learning_rate": 4.348599680050961e-06,
+      "loss": 0.8969,
+      "step": 89990
+    },
+    {
+      "epoch": 0.6514799452756846,
+      "grad_norm": 0.17069634795188904,
+      "learning_rate": 4.348527293390374e-06,
+      "loss": 0.9152,
+      "step": 90000
+    },
+    {
+      "epoch": 0.6515523319362708,
+      "grad_norm": 0.16475622355937958,
+      "learning_rate": 4.348454906729788e-06,
+      "loss": 0.9022,
+      "step": 90010
+    },
+    {
+      "epoch": 0.651624718596857,
+      "grad_norm": 0.16313569247722626,
+      "learning_rate": 4.3483825200692024e-06,
+      "loss": 0.8895,
+      "step": 90020
+    },
+    {
+      "epoch": 0.6516971052574432,
+      "grad_norm": 0.15203681588172913,
+      "learning_rate": 4.348310133408616e-06,
+      "loss": 0.8818,
+      "step": 90030
+    },
+    {
+      "epoch": 0.6517694919180294,
+      "grad_norm": 0.1589788943529129,
+      "learning_rate": 4.34823774674803e-06,
+      "loss": 0.9019,
+      "step": 90040
+    },
+    {
+      "epoch": 0.6518418785786155,
+      "grad_norm": 0.1522030234336853,
+      "learning_rate": 4.348165360087443e-06,
+      "loss": 0.8986,
+      "step": 90050
+    },
+    {
+      "epoch": 0.6519142652392017,
+      "grad_norm": 0.15555304288864136,
+      "learning_rate": 4.348092973426857e-06,
+      "loss": 0.9048,
+      "step": 90060
+    },
+    {
+      "epoch": 0.6519866518997879,
+      "grad_norm": 0.15009433031082153,
+      "learning_rate": 4.348020586766271e-06,
+      "loss": 0.9054,
+      "step": 90070
+    },
+    {
+      "epoch": 0.6520590385603741,
+      "grad_norm": 0.1470109075307846,
+      "learning_rate": 4.347948200105685e-06,
+      "loss": 0.9064,
+      "step": 90080
+    },
+    {
+      "epoch": 0.6521314252209602,
+      "grad_norm": 0.17592494189739227,
+      "learning_rate": 4.347875813445099e-06,
+      "loss": 0.8949,
+      "step": 90090
+    },
+    {
+      "epoch": 0.6522038118815464,
+      "grad_norm": 0.14667287468910217,
+      "learning_rate": 4.347803426784512e-06,
+      "loss": 0.9068,
+      "step": 90100
+    },
+    {
+      "epoch": 0.6522761985421327,
+      "grad_norm": 0.15610606968402863,
+      "learning_rate": 4.347731040123927e-06,
+      "loss": 0.8909,
+      "step": 90110
+    },
+    {
+      "epoch": 0.6523485852027189,
+      "grad_norm": 0.17158207297325134,
+      "learning_rate": 4.34765865346334e-06,
+      "loss": 0.8969,
+      "step": 90120
+    },
+    {
+      "epoch": 0.652420971863305,
+      "grad_norm": 0.15405572950839996,
+      "learning_rate": 4.347586266802754e-06,
+      "loss": 0.9042,
+      "step": 90130
+    },
+    {
+      "epoch": 0.6524933585238912,
+      "grad_norm": 0.1571338176727295,
+      "learning_rate": 4.3475138801421675e-06,
+      "loss": 0.8955,
+      "step": 90140
+    },
+    {
+      "epoch": 0.6525657451844774,
+      "grad_norm": 0.14696629345417023,
+      "learning_rate": 4.347441493481582e-06,
+      "loss": 0.8909,
+      "step": 90150
+    },
+    {
+      "epoch": 0.6526381318450636,
+      "grad_norm": 0.16139934957027435,
+      "learning_rate": 4.347369106820996e-06,
+      "loss": 0.8968,
+      "step": 90160
+    },
+    {
+      "epoch": 0.6527105185056498,
+      "grad_norm": 0.15784136950969696,
+      "learning_rate": 4.347296720160409e-06,
+      "loss": 0.9019,
+      "step": 90170
+    },
+    {
+      "epoch": 0.6527829051662359,
+      "grad_norm": 0.1769580841064453,
+      "learning_rate": 4.347224333499823e-06,
+      "loss": 0.9009,
+      "step": 90180
+    },
+    {
+      "epoch": 0.6528552918268221,
+      "grad_norm": 0.15210415422916412,
+      "learning_rate": 4.347151946839237e-06,
+      "loss": 0.8969,
+      "step": 90190
+    },
+    {
+      "epoch": 0.6529276784874083,
+      "grad_norm": 0.15724553167819977,
+      "learning_rate": 4.347079560178651e-06,
+      "loss": 0.8949,
+      "step": 90200
+    },
+    {
+      "epoch": 0.6530000651479946,
+      "grad_norm": 0.15788982808589935,
+      "learning_rate": 4.3470071735180645e-06,
+      "loss": 0.8914,
+      "step": 90210
+    },
+    {
+      "epoch": 0.6530724518085808,
+      "grad_norm": 0.16926227509975433,
+      "learning_rate": 4.346934786857478e-06,
+      "loss": 0.8868,
+      "step": 90220
+    },
+    {
+      "epoch": 0.6531448384691669,
+      "grad_norm": 0.16521447896957397,
+      "learning_rate": 4.346862400196893e-06,
+      "loss": 0.8966,
+      "step": 90230
+    },
+    {
+      "epoch": 0.6532172251297531,
+      "grad_norm": 0.17298093438148499,
+      "learning_rate": 4.346790013536306e-06,
+      "loss": 0.8916,
+      "step": 90240
+    },
+    {
+      "epoch": 0.6532896117903393,
+      "grad_norm": 0.1661769151687622,
+      "learning_rate": 4.34671762687572e-06,
+      "loss": 0.8891,
+      "step": 90250
+    },
+    {
+      "epoch": 0.6533619984509255,
+      "grad_norm": 0.16557759046554565,
+      "learning_rate": 4.3466452402151335e-06,
+      "loss": 0.9243,
+      "step": 90260
+    },
+    {
+      "epoch": 0.6534343851115116,
+      "grad_norm": 0.15999002754688263,
+      "learning_rate": 4.346572853554547e-06,
+      "loss": 0.8908,
+      "step": 90270
+    },
+    {
+      "epoch": 0.6535067717720978,
+      "grad_norm": 0.15427353978157043,
+      "learning_rate": 4.346500466893961e-06,
+      "loss": 0.9055,
+      "step": 90280
+    },
+    {
+      "epoch": 0.653579158432684,
+      "grad_norm": 0.16352301836013794,
+      "learning_rate": 4.346428080233374e-06,
+      "loss": 0.8789,
+      "step": 90290
+    },
+    {
+      "epoch": 0.6536515450932702,
+      "grad_norm": 0.15975026786327362,
+      "learning_rate": 4.346355693572789e-06,
+      "loss": 0.9023,
+      "step": 90300
+    },
+    {
+      "epoch": 0.6537239317538563,
+      "grad_norm": 0.17519934475421906,
+      "learning_rate": 4.346283306912202e-06,
+      "loss": 0.903,
+      "step": 90310
+    },
+    {
+      "epoch": 0.6537963184144426,
+      "grad_norm": 0.1745595932006836,
+      "learning_rate": 4.346210920251616e-06,
+      "loss": 0.9025,
+      "step": 90320
+    },
+    {
+      "epoch": 0.6538687050750288,
+      "grad_norm": 0.16801311075687408,
+      "learning_rate": 4.34613853359103e-06,
+      "loss": 0.8968,
+      "step": 90330
+    },
+    {
+      "epoch": 0.653941091735615,
+      "grad_norm": 0.1629055142402649,
+      "learning_rate": 4.346066146930444e-06,
+      "loss": 0.8973,
+      "step": 90340
+    },
+    {
+      "epoch": 0.6540134783962012,
+      "grad_norm": 0.17443734407424927,
+      "learning_rate": 4.345993760269858e-06,
+      "loss": 0.8937,
+      "step": 90350
+    },
+    {
+      "epoch": 0.6540858650567873,
+      "grad_norm": 0.15698504447937012,
+      "learning_rate": 4.345921373609271e-06,
+      "loss": 0.902,
+      "step": 90360
+    },
+    {
+      "epoch": 0.6541582517173735,
+      "grad_norm": 0.1658298522233963,
+      "learning_rate": 4.345848986948685e-06,
+      "loss": 0.9065,
+      "step": 90370
+    },
+    {
+      "epoch": 0.6542306383779597,
+      "grad_norm": 0.17639166116714478,
+      "learning_rate": 4.345776600288099e-06,
+      "loss": 0.8894,
+      "step": 90380
+    },
+    {
+      "epoch": 0.6543030250385459,
+      "grad_norm": 0.17266248166561127,
+      "learning_rate": 4.345704213627513e-06,
+      "loss": 0.8957,
+      "step": 90390
+    },
+    {
+      "epoch": 0.654375411699132,
+      "grad_norm": 0.1632227897644043,
+      "learning_rate": 4.345631826966927e-06,
+      "loss": 0.8963,
+      "step": 90400
+    },
+    {
+      "epoch": 0.6544477983597182,
+      "grad_norm": 0.16069155931472778,
+      "learning_rate": 4.34555944030634e-06,
+      "loss": 0.9013,
+      "step": 90410
+    },
+    {
+      "epoch": 0.6545201850203045,
+      "grad_norm": 0.1587107628583908,
+      "learning_rate": 4.345487053645755e-06,
+      "loss": 0.8857,
+      "step": 90420
+    },
+    {
+      "epoch": 0.6545925716808907,
+      "grad_norm": 0.16049619019031525,
+      "learning_rate": 4.345414666985168e-06,
+      "loss": 0.9042,
+      "step": 90430
+    },
+    {
+      "epoch": 0.6546649583414769,
+      "grad_norm": 0.16169418394565582,
+      "learning_rate": 4.345342280324582e-06,
+      "loss": 0.8816,
+      "step": 90440
+    },
+    {
+      "epoch": 0.654737345002063,
+      "grad_norm": 0.17584997415542603,
+      "learning_rate": 4.3452698936639956e-06,
+      "loss": 0.8948,
+      "step": 90450
+    },
+    {
+      "epoch": 0.6548097316626492,
+      "grad_norm": 0.15903721749782562,
+      "learning_rate": 4.34519750700341e-06,
+      "loss": 0.8819,
+      "step": 90460
+    },
+    {
+      "epoch": 0.6548821183232354,
+      "grad_norm": 0.1641675978899002,
+      "learning_rate": 4.345125120342824e-06,
+      "loss": 0.8808,
+      "step": 90470
+    },
+    {
+      "epoch": 0.6549545049838216,
+      "grad_norm": 0.15496835112571716,
+      "learning_rate": 4.345052733682237e-06,
+      "loss": 0.8954,
+      "step": 90480
+    },
+    {
+      "epoch": 0.6550268916444077,
+      "grad_norm": 0.15261691808700562,
+      "learning_rate": 4.344980347021651e-06,
+      "loss": 0.9,
+      "step": 90490
+    },
+    {
+      "epoch": 0.6550992783049939,
+      "grad_norm": 0.15243463218212128,
+      "learning_rate": 4.344907960361065e-06,
+      "loss": 0.9019,
+      "step": 90500
+    },
+    {
+      "epoch": 0.6551716649655801,
+      "grad_norm": 0.16035859286785126,
+      "learning_rate": 4.344835573700479e-06,
+      "loss": 0.9019,
+      "step": 90510
+    },
+    {
+      "epoch": 0.6552440516261663,
+      "grad_norm": 0.17638221383094788,
+      "learning_rate": 4.3447631870398926e-06,
+      "loss": 0.9142,
+      "step": 90520
+    },
+    {
+      "epoch": 0.6553164382867526,
+      "grad_norm": 0.15220975875854492,
+      "learning_rate": 4.344690800379306e-06,
+      "loss": 0.9064,
+      "step": 90530
+    },
+    {
+      "epoch": 0.6553888249473387,
+      "grad_norm": 0.14705011248588562,
+      "learning_rate": 4.344618413718721e-06,
+      "loss": 0.8898,
+      "step": 90540
+    },
+    {
+      "epoch": 0.6554612116079249,
+      "grad_norm": 0.164079487323761,
+      "learning_rate": 4.344546027058134e-06,
+      "loss": 0.905,
+      "step": 90550
+    },
+    {
+      "epoch": 0.6555335982685111,
+      "grad_norm": 0.15466266870498657,
+      "learning_rate": 4.344473640397548e-06,
+      "loss": 0.9071,
+      "step": 90560
+    },
+    {
+      "epoch": 0.6556059849290973,
+      "grad_norm": 0.1539478898048401,
+      "learning_rate": 4.3444012537369615e-06,
+      "loss": 0.8988,
+      "step": 90570
+    },
+    {
+      "epoch": 0.6556783715896835,
+      "grad_norm": 0.15907570719718933,
+      "learning_rate": 4.344328867076376e-06,
+      "loss": 0.8937,
+      "step": 90580
+    },
+    {
+      "epoch": 0.6557507582502696,
+      "grad_norm": 0.16037173569202423,
+      "learning_rate": 4.34425648041579e-06,
+      "loss": 0.8899,
+      "step": 90590
+    },
+    {
+      "epoch": 0.6558231449108558,
+      "grad_norm": 0.1608818769454956,
+      "learning_rate": 4.344184093755203e-06,
+      "loss": 0.8924,
+      "step": 90600
+    },
+    {
+      "epoch": 0.655895531571442,
+      "grad_norm": 0.17891094088554382,
+      "learning_rate": 4.344111707094617e-06,
+      "loss": 0.904,
+      "step": 90610
+    },
+    {
+      "epoch": 0.6559679182320282,
+      "grad_norm": 0.15963943302631378,
+      "learning_rate": 4.344039320434031e-06,
+      "loss": 0.8965,
+      "step": 90620
+    },
+    {
+      "epoch": 0.6560403048926143,
+      "grad_norm": 0.1626274734735489,
+      "learning_rate": 4.343966933773445e-06,
+      "loss": 0.8977,
+      "step": 90630
+    },
+    {
+      "epoch": 0.6561126915532006,
+      "grad_norm": 0.14938683807849884,
+      "learning_rate": 4.3438945471128585e-06,
+      "loss": 0.8972,
+      "step": 90640
+    },
+    {
+      "epoch": 0.6561850782137868,
+      "grad_norm": 0.15236683189868927,
+      "learning_rate": 4.343822160452272e-06,
+      "loss": 0.8866,
+      "step": 90650
+    },
+    {
+      "epoch": 0.656257464874373,
+      "grad_norm": 0.1745002120733261,
+      "learning_rate": 4.343749773791686e-06,
+      "loss": 0.8951,
+      "step": 90660
+    },
+    {
+      "epoch": 0.6563298515349592,
+      "grad_norm": 0.15428422391414642,
+      "learning_rate": 4.3436773871311e-06,
+      "loss": 0.9125,
+      "step": 90670
+    },
+    {
+      "epoch": 0.6564022381955453,
+      "grad_norm": 0.18317624926567078,
+      "learning_rate": 4.343605000470514e-06,
+      "loss": 0.8919,
+      "step": 90680
+    },
+    {
+      "epoch": 0.6564746248561315,
+      "grad_norm": 0.15780498087406158,
+      "learning_rate": 4.3435326138099274e-06,
+      "loss": 0.9001,
+      "step": 90690
+    },
+    {
+      "epoch": 0.6565470115167177,
+      "grad_norm": 0.1752166748046875,
+      "learning_rate": 4.343460227149341e-06,
+      "loss": 0.9079,
+      "step": 90700
+    },
+    {
+      "epoch": 0.6566193981773039,
+      "grad_norm": 0.17048229277133942,
+      "learning_rate": 4.3433878404887555e-06,
+      "loss": 0.8877,
+      "step": 90710
+    },
+    {
+      "epoch": 0.65669178483789,
+      "grad_norm": 0.15478329360485077,
+      "learning_rate": 4.343315453828169e-06,
+      "loss": 0.8885,
+      "step": 90720
+    },
+    {
+      "epoch": 0.6567641714984762,
+      "grad_norm": 0.15634417533874512,
+      "learning_rate": 4.343243067167583e-06,
+      "loss": 0.8944,
+      "step": 90730
+    },
+    {
+      "epoch": 0.6568365581590625,
+      "grad_norm": 0.15285581350326538,
+      "learning_rate": 4.343170680506996e-06,
+      "loss": 0.9014,
+      "step": 90740
+    },
+    {
+      "epoch": 0.6569089448196487,
+      "grad_norm": 0.1574464589357376,
+      "learning_rate": 4.343098293846411e-06,
+      "loss": 0.8962,
+      "step": 90750
+    },
+    {
+      "epoch": 0.6569813314802349,
+      "grad_norm": 0.19475358724594116,
+      "learning_rate": 4.3430259071858244e-06,
+      "loss": 0.8974,
+      "step": 90760
+    },
+    {
+      "epoch": 0.657053718140821,
+      "grad_norm": 0.21129895746707916,
+      "learning_rate": 4.342953520525238e-06,
+      "loss": 0.8898,
+      "step": 90770
+    },
+    {
+      "epoch": 0.6571261048014072,
+      "grad_norm": 0.1641937643289566,
+      "learning_rate": 4.342881133864652e-06,
+      "loss": 0.8914,
+      "step": 90780
+    },
+    {
+      "epoch": 0.6571984914619934,
+      "grad_norm": 0.158699169754982,
+      "learning_rate": 4.342808747204066e-06,
+      "loss": 0.8855,
+      "step": 90790
+    },
+    {
+      "epoch": 0.6572708781225796,
+      "grad_norm": 0.1632438600063324,
+      "learning_rate": 4.342736360543479e-06,
+      "loss": 0.8882,
+      "step": 90800
+    },
+    {
+      "epoch": 0.6573432647831657,
+      "grad_norm": 0.16449430584907532,
+      "learning_rate": 4.342663973882893e-06,
+      "loss": 0.8886,
+      "step": 90810
+    },
+    {
+      "epoch": 0.6574156514437519,
+      "grad_norm": 0.16445982456207275,
+      "learning_rate": 4.342591587222307e-06,
+      "loss": 0.8873,
+      "step": 90820
+    },
+    {
+      "epoch": 0.6574880381043381,
+      "grad_norm": 0.18143968284130096,
+      "learning_rate": 4.342519200561721e-06,
+      "loss": 0.9047,
+      "step": 90830
+    },
+    {
+      "epoch": 0.6575604247649243,
+      "grad_norm": 0.15910190343856812,
+      "learning_rate": 4.342446813901134e-06,
+      "loss": 0.8998,
+      "step": 90840
+    },
+    {
+      "epoch": 0.6576328114255106,
+      "grad_norm": 0.1646713763475418,
+      "learning_rate": 4.342374427240548e-06,
+      "loss": 0.9012,
+      "step": 90850
+    },
+    {
+      "epoch": 0.6577051980860967,
+      "grad_norm": 0.1441211849451065,
+      "learning_rate": 4.342302040579962e-06,
+      "loss": 0.897,
+      "step": 90860
+    },
+    {
+      "epoch": 0.6577775847466829,
+      "grad_norm": 0.15770696103572845,
+      "learning_rate": 4.342229653919376e-06,
+      "loss": 0.9049,
+      "step": 90870
+    },
+    {
+      "epoch": 0.6578499714072691,
+      "grad_norm": 0.1501597911119461,
+      "learning_rate": 4.3421572672587895e-06,
+      "loss": 0.901,
+      "step": 90880
+    },
+    {
+      "epoch": 0.6579223580678553,
+      "grad_norm": 0.16416409611701965,
+      "learning_rate": 4.342084880598203e-06,
+      "loss": 0.9032,
+      "step": 90890
+    },
+    {
+      "epoch": 0.6579947447284414,
+      "grad_norm": 0.16274628043174744,
+      "learning_rate": 4.342012493937618e-06,
+      "loss": 0.887,
+      "step": 90900
+    },
+    {
+      "epoch": 0.6580671313890276,
+      "grad_norm": 0.14996199309825897,
+      "learning_rate": 4.341940107277031e-06,
+      "loss": 0.8815,
+      "step": 90910
+    },
+    {
+      "epoch": 0.6581395180496138,
+      "grad_norm": 0.2303011268377304,
+      "learning_rate": 4.341867720616445e-06,
+      "loss": 0.8799,
+      "step": 90920
+    },
+    {
+      "epoch": 0.6582119047102,
+      "grad_norm": 0.14957544207572937,
+      "learning_rate": 4.3417953339558585e-06,
+      "loss": 0.903,
+      "step": 90930
+    },
+    {
+      "epoch": 0.6582842913707861,
+      "grad_norm": 0.15761502087116241,
+      "learning_rate": 4.341722947295273e-06,
+      "loss": 0.9141,
+      "step": 90940
+    },
+    {
+      "epoch": 0.6583566780313724,
+      "grad_norm": 0.17332914471626282,
+      "learning_rate": 4.3416505606346865e-06,
+      "loss": 0.8998,
+      "step": 90950
+    },
+    {
+      "epoch": 0.6584290646919586,
+      "grad_norm": 0.1537407785654068,
+      "learning_rate": 4.3415781739741e-06,
+      "loss": 0.8852,
+      "step": 90960
+    },
+    {
+      "epoch": 0.6585014513525448,
+      "grad_norm": 0.14254407584667206,
+      "learning_rate": 4.341505787313514e-06,
+      "loss": 0.8973,
+      "step": 90970
+    },
+    {
+      "epoch": 0.658573838013131,
+      "grad_norm": 0.1535002887248993,
+      "learning_rate": 4.341433400652928e-06,
+      "loss": 0.8981,
+      "step": 90980
+    },
+    {
+      "epoch": 0.6586462246737171,
+      "grad_norm": 0.15489903092384338,
+      "learning_rate": 4.341361013992342e-06,
+      "loss": 0.8854,
+      "step": 90990
+    },
+    {
+      "epoch": 0.6587186113343033,
+      "grad_norm": 0.16978639364242554,
+      "learning_rate": 4.3412886273317555e-06,
+      "loss": 0.9023,
+      "step": 91000
+    },
+    {
+      "epoch": 0.6587909979948895,
+      "grad_norm": 0.1680452525615692,
+      "learning_rate": 4.341216240671169e-06,
+      "loss": 0.9008,
+      "step": 91010
+    },
+    {
+      "epoch": 0.6588633846554757,
+      "grad_norm": 0.15573959052562714,
+      "learning_rate": 4.3411438540105835e-06,
+      "loss": 0.889,
+      "step": 91020
+    },
+    {
+      "epoch": 0.6589357713160618,
+      "grad_norm": 0.1434423178434372,
+      "learning_rate": 4.341071467349997e-06,
+      "loss": 0.8804,
+      "step": 91030
+    },
+    {
+      "epoch": 0.659008157976648,
+      "grad_norm": 0.16832110285758972,
+      "learning_rate": 4.340999080689411e-06,
+      "loss": 0.9065,
+      "step": 91040
+    },
+    {
+      "epoch": 0.6590805446372342,
+      "grad_norm": 0.15029281377792358,
+      "learning_rate": 4.340926694028824e-06,
+      "loss": 0.8851,
+      "step": 91050
+    },
+    {
+      "epoch": 0.6591529312978205,
+      "grad_norm": 0.1768171489238739,
+      "learning_rate": 4.340854307368239e-06,
+      "loss": 0.9025,
+      "step": 91060
+    },
+    {
+      "epoch": 0.6592253179584067,
+      "grad_norm": 0.14737635850906372,
+      "learning_rate": 4.3407819207076525e-06,
+      "loss": 0.9048,
+      "step": 91070
+    },
+    {
+      "epoch": 0.6592977046189928,
+      "grad_norm": 0.1616610735654831,
+      "learning_rate": 4.340709534047066e-06,
+      "loss": 0.9036,
+      "step": 91080
+    },
+    {
+      "epoch": 0.659370091279579,
+      "grad_norm": 0.15665307641029358,
+      "learning_rate": 4.34063714738648e-06,
+      "loss": 0.8956,
+      "step": 91090
+    },
+    {
+      "epoch": 0.6594424779401652,
+      "grad_norm": 0.15417052805423737,
+      "learning_rate": 4.340564760725894e-06,
+      "loss": 0.9022,
+      "step": 91100
+    },
+    {
+      "epoch": 0.6595148646007514,
+      "grad_norm": 0.14735175669193268,
+      "learning_rate": 4.340492374065308e-06,
+      "loss": 0.8875,
+      "step": 91110
+    },
+    {
+      "epoch": 0.6595872512613375,
+      "grad_norm": 0.18449713289737701,
+      "learning_rate": 4.340419987404721e-06,
+      "loss": 0.8886,
+      "step": 91120
+    },
+    {
+      "epoch": 0.6596596379219237,
+      "grad_norm": 0.15880465507507324,
+      "learning_rate": 4.340347600744135e-06,
+      "loss": 0.8961,
+      "step": 91130
+    },
+    {
+      "epoch": 0.6597320245825099,
+      "grad_norm": 0.1495341658592224,
+      "learning_rate": 4.3402752140835495e-06,
+      "loss": 0.8861,
+      "step": 91140
+    },
+    {
+      "epoch": 0.6598044112430961,
+      "grad_norm": 0.1608497053384781,
+      "learning_rate": 4.340202827422963e-06,
+      "loss": 0.8905,
+      "step": 91150
+    },
+    {
+      "epoch": 0.6598767979036823,
+      "grad_norm": 0.18891900777816772,
+      "learning_rate": 4.340130440762377e-06,
+      "loss": 0.894,
+      "step": 91160
+    },
+    {
+      "epoch": 0.6599491845642685,
+      "grad_norm": 0.1456620842218399,
+      "learning_rate": 4.34005805410179e-06,
+      "loss": 0.8951,
+      "step": 91170
+    },
+    {
+      "epoch": 0.6600215712248547,
+      "grad_norm": 0.16527323424816132,
+      "learning_rate": 4.339985667441205e-06,
+      "loss": 0.8914,
+      "step": 91180
+    },
+    {
+      "epoch": 0.6600939578854409,
+      "grad_norm": 0.15947329998016357,
+      "learning_rate": 4.339913280780618e-06,
+      "loss": 0.8964,
+      "step": 91190
+    },
+    {
+      "epoch": 0.6601663445460271,
+      "grad_norm": 0.1482989341020584,
+      "learning_rate": 4.339840894120032e-06,
+      "loss": 0.8923,
+      "step": 91200
+    },
+    {
+      "epoch": 0.6602387312066132,
+      "grad_norm": 0.15714868903160095,
+      "learning_rate": 4.339768507459446e-06,
+      "loss": 0.892,
+      "step": 91210
+    },
+    {
+      "epoch": 0.6603111178671994,
+      "grad_norm": 0.18459376692771912,
+      "learning_rate": 4.33969612079886e-06,
+      "loss": 0.8944,
+      "step": 91220
+    },
+    {
+      "epoch": 0.6603835045277856,
+      "grad_norm": 0.1577734649181366,
+      "learning_rate": 4.339623734138274e-06,
+      "loss": 0.8919,
+      "step": 91230
+    },
+    {
+      "epoch": 0.6604558911883718,
+      "grad_norm": 0.19559817016124725,
+      "learning_rate": 4.339551347477687e-06,
+      "loss": 0.9021,
+      "step": 91240
+    },
+    {
+      "epoch": 0.660528277848958,
+      "grad_norm": 0.16929206252098083,
+      "learning_rate": 4.339478960817101e-06,
+      "loss": 0.9089,
+      "step": 91250
+    },
+    {
+      "epoch": 0.6606006645095441,
+      "grad_norm": 0.1482090950012207,
+      "learning_rate": 4.339406574156515e-06,
+      "loss": 0.8995,
+      "step": 91260
+    },
+    {
+      "epoch": 0.6606730511701304,
+      "grad_norm": 0.17333853244781494,
+      "learning_rate": 4.339334187495929e-06,
+      "loss": 0.8998,
+      "step": 91270
+    },
+    {
+      "epoch": 0.6607454378307166,
+      "grad_norm": 0.16415739059448242,
+      "learning_rate": 4.339261800835343e-06,
+      "loss": 0.8956,
+      "step": 91280
+    },
+    {
+      "epoch": 0.6608178244913028,
+      "grad_norm": 0.14835873246192932,
+      "learning_rate": 4.339189414174756e-06,
+      "loss": 0.8809,
+      "step": 91290
+    },
+    {
+      "epoch": 0.660890211151889,
+      "grad_norm": 0.16368795931339264,
+      "learning_rate": 4.33911702751417e-06,
+      "loss": 0.9006,
+      "step": 91300
+    },
+    {
+      "epoch": 0.6609625978124751,
+      "grad_norm": 0.1738886833190918,
+      "learning_rate": 4.339044640853584e-06,
+      "loss": 0.8993,
+      "step": 91310
+    },
+    {
+      "epoch": 0.6610349844730613,
+      "grad_norm": 0.1818006932735443,
+      "learning_rate": 4.338972254192998e-06,
+      "loss": 0.8997,
+      "step": 91320
+    },
+    {
+      "epoch": 0.6611073711336475,
+      "grad_norm": 0.16817954182624817,
+      "learning_rate": 4.338899867532412e-06,
+      "loss": 0.8994,
+      "step": 91330
+    },
+    {
+      "epoch": 0.6611797577942337,
+      "grad_norm": 0.15932884812355042,
+      "learning_rate": 4.338827480871825e-06,
+      "loss": 0.8976,
+      "step": 91340
+    },
+    {
+      "epoch": 0.6612521444548198,
+      "grad_norm": 0.15406905114650726,
+      "learning_rate": 4.338755094211239e-06,
+      "loss": 0.8906,
+      "step": 91350
+    },
+    {
+      "epoch": 0.661324531115406,
+      "grad_norm": 0.16480666399002075,
+      "learning_rate": 4.3386827075506524e-06,
+      "loss": 0.9069,
+      "step": 91360
+    },
+    {
+      "epoch": 0.6613969177759922,
+      "grad_norm": 0.17040395736694336,
+      "learning_rate": 4.338610320890067e-06,
+      "loss": 0.8925,
+      "step": 91370
+    },
+    {
+      "epoch": 0.6614693044365785,
+      "grad_norm": 0.1497850865125656,
+      "learning_rate": 4.3385379342294805e-06,
+      "loss": 0.8956,
+      "step": 91380
+    },
+    {
+      "epoch": 0.6615416910971647,
+      "grad_norm": 0.326239675283432,
+      "learning_rate": 4.338465547568894e-06,
+      "loss": 0.8959,
+      "step": 91390
+    },
+    {
+      "epoch": 0.6616140777577508,
+      "grad_norm": 0.17361514270305634,
+      "learning_rate": 4.338393160908308e-06,
+      "loss": 0.9122,
+      "step": 91400
+    },
+    {
+      "epoch": 0.661686464418337,
+      "grad_norm": 0.16059482097625732,
+      "learning_rate": 4.338320774247722e-06,
+      "loss": 0.8971,
+      "step": 91410
+    },
+    {
+      "epoch": 0.6617588510789232,
+      "grad_norm": 0.16254779696464539,
+      "learning_rate": 4.338248387587136e-06,
+      "loss": 0.8937,
+      "step": 91420
+    },
+    {
+      "epoch": 0.6618312377395094,
+      "grad_norm": 0.16201868653297424,
+      "learning_rate": 4.3381760009265494e-06,
+      "loss": 0.8988,
+      "step": 91430
+    },
+    {
+      "epoch": 0.6619036244000955,
+      "grad_norm": 0.15529035031795502,
+      "learning_rate": 4.338103614265963e-06,
+      "loss": 0.8974,
+      "step": 91440
+    },
+    {
+      "epoch": 0.6619760110606817,
+      "grad_norm": 0.18519388139247894,
+      "learning_rate": 4.3380312276053775e-06,
+      "loss": 0.8969,
+      "step": 91450
+    },
+    {
+      "epoch": 0.6620483977212679,
+      "grad_norm": 0.14886817336082458,
+      "learning_rate": 4.337958840944791e-06,
+      "loss": 0.8927,
+      "step": 91460
+    },
+    {
+      "epoch": 0.6621207843818541,
+      "grad_norm": 0.18947485089302063,
+      "learning_rate": 4.337886454284205e-06,
+      "loss": 0.8954,
+      "step": 91470
+    },
+    {
+      "epoch": 0.6621931710424404,
+      "grad_norm": 0.14915327727794647,
+      "learning_rate": 4.337814067623618e-06,
+      "loss": 0.904,
+      "step": 91480
+    },
+    {
+      "epoch": 0.6622655577030265,
+      "grad_norm": 0.3526378273963928,
+      "learning_rate": 4.337741680963032e-06,
+      "loss": 0.907,
+      "step": 91490
+    },
+    {
+      "epoch": 0.6623379443636127,
+      "grad_norm": 0.17231549322605133,
+      "learning_rate": 4.3376692943024464e-06,
+      "loss": 0.8878,
+      "step": 91500
+    },
+    {
+      "epoch": 0.6624103310241989,
+      "grad_norm": 0.15187296271324158,
+      "learning_rate": 4.33759690764186e-06,
+      "loss": 0.8893,
+      "step": 91510
+    },
+    {
+      "epoch": 0.6624827176847851,
+      "grad_norm": 0.16822843253612518,
+      "learning_rate": 4.337524520981274e-06,
+      "loss": 0.893,
+      "step": 91520
+    },
+    {
+      "epoch": 0.6625551043453712,
+      "grad_norm": 0.16215179860591888,
+      "learning_rate": 4.337452134320687e-06,
+      "loss": 0.893,
+      "step": 91530
+    },
+    {
+      "epoch": 0.6626274910059574,
+      "grad_norm": 0.15672698616981506,
+      "learning_rate": 4.337379747660102e-06,
+      "loss": 0.8875,
+      "step": 91540
+    },
+    {
+      "epoch": 0.6626998776665436,
+      "grad_norm": 0.14577756822109222,
+      "learning_rate": 4.337307360999515e-06,
+      "loss": 0.9028,
+      "step": 91550
+    },
+    {
+      "epoch": 0.6627722643271298,
+      "grad_norm": 0.1537584662437439,
+      "learning_rate": 4.337234974338929e-06,
+      "loss": 0.8978,
+      "step": 91560
+    },
+    {
+      "epoch": 0.6628446509877159,
+      "grad_norm": 0.15644492208957672,
+      "learning_rate": 4.337162587678343e-06,
+      "loss": 0.886,
+      "step": 91570
+    },
+    {
+      "epoch": 0.6629170376483021,
+      "grad_norm": 0.16348648071289062,
+      "learning_rate": 4.337090201017757e-06,
+      "loss": 0.9081,
+      "step": 91580
+    },
+    {
+      "epoch": 0.6629894243088884,
+      "grad_norm": 0.15281915664672852,
+      "learning_rate": 4.337017814357171e-06,
+      "loss": 0.8931,
+      "step": 91590
+    },
+    {
+      "epoch": 0.6630618109694746,
+      "grad_norm": 0.1640012115240097,
+      "learning_rate": 4.336945427696584e-06,
+      "loss": 0.9045,
+      "step": 91600
+    },
+    {
+      "epoch": 0.6631341976300608,
+      "grad_norm": 0.17783017456531525,
+      "learning_rate": 4.336873041035998e-06,
+      "loss": 0.8932,
+      "step": 91610
+    },
+    {
+      "epoch": 0.6632065842906469,
+      "grad_norm": 0.23584623634815216,
+      "learning_rate": 4.336800654375412e-06,
+      "loss": 0.9074,
+      "step": 91620
+    },
+    {
+      "epoch": 0.6632789709512331,
+      "grad_norm": 0.15153855085372925,
+      "learning_rate": 4.336728267714826e-06,
+      "loss": 0.8817,
+      "step": 91630
+    },
+    {
+      "epoch": 0.6633513576118193,
+      "grad_norm": 0.1572607457637787,
+      "learning_rate": 4.33665588105424e-06,
+      "loss": 0.8969,
+      "step": 91640
+    },
+    {
+      "epoch": 0.6634237442724055,
+      "grad_norm": 0.15882018208503723,
+      "learning_rate": 4.336583494393653e-06,
+      "loss": 0.8956,
+      "step": 91650
+    },
+    {
+      "epoch": 0.6634961309329916,
+      "grad_norm": 0.19259294867515564,
+      "learning_rate": 4.336511107733068e-06,
+      "loss": 0.8947,
+      "step": 91660
+    },
+    {
+      "epoch": 0.6635685175935778,
+      "grad_norm": 0.15729625523090363,
+      "learning_rate": 4.336438721072481e-06,
+      "loss": 0.8938,
+      "step": 91670
+    },
+    {
+      "epoch": 0.663640904254164,
+      "grad_norm": 0.18913884460926056,
+      "learning_rate": 4.336366334411895e-06,
+      "loss": 0.898,
+      "step": 91680
+    },
+    {
+      "epoch": 0.6637132909147502,
+      "grad_norm": 0.23208874464035034,
+      "learning_rate": 4.3362939477513085e-06,
+      "loss": 0.8831,
+      "step": 91690
+    },
+    {
+      "epoch": 0.6637856775753365,
+      "grad_norm": 0.15006226301193237,
+      "learning_rate": 4.336221561090723e-06,
+      "loss": 0.8933,
+      "step": 91700
+    },
+    {
+      "epoch": 0.6638580642359226,
+      "grad_norm": 0.17437438666820526,
+      "learning_rate": 4.336149174430137e-06,
+      "loss": 0.8922,
+      "step": 91710
+    },
+    {
+      "epoch": 0.6639304508965088,
+      "grad_norm": 0.1585850864648819,
+      "learning_rate": 4.33607678776955e-06,
+      "loss": 0.8929,
+      "step": 91720
+    },
+    {
+      "epoch": 0.664002837557095,
+      "grad_norm": 0.15892000496387482,
+      "learning_rate": 4.336004401108964e-06,
+      "loss": 0.8981,
+      "step": 91730
+    },
+    {
+      "epoch": 0.6640752242176812,
+      "grad_norm": 0.15951044857501984,
+      "learning_rate": 4.335932014448378e-06,
+      "loss": 0.8968,
+      "step": 91740
+    },
+    {
+      "epoch": 0.6641476108782673,
+      "grad_norm": 0.16389575600624084,
+      "learning_rate": 4.335859627787792e-06,
+      "loss": 0.9076,
+      "step": 91750
+    },
+    {
+      "epoch": 0.6642199975388535,
+      "grad_norm": 0.23026235401630402,
+      "learning_rate": 4.3357872411272055e-06,
+      "loss": 0.8934,
+      "step": 91760
+    },
+    {
+      "epoch": 0.6642923841994397,
+      "grad_norm": 0.15285776555538177,
+      "learning_rate": 4.335714854466619e-06,
+      "loss": 0.8961,
+      "step": 91770
+    },
+    {
+      "epoch": 0.6643647708600259,
+      "grad_norm": 0.15406441688537598,
+      "learning_rate": 4.335642467806034e-06,
+      "loss": 0.8853,
+      "step": 91780
+    },
+    {
+      "epoch": 0.664437157520612,
+      "grad_norm": 0.1550244837999344,
+      "learning_rate": 4.335570081145447e-06,
+      "loss": 0.9001,
+      "step": 91790
+    },
+    {
+      "epoch": 0.6645095441811983,
+      "grad_norm": 0.16414350271224976,
+      "learning_rate": 4.335497694484861e-06,
+      "loss": 0.9078,
+      "step": 91800
+    },
+    {
+      "epoch": 0.6645819308417845,
+      "grad_norm": 0.15932433307170868,
+      "learning_rate": 4.3354253078242745e-06,
+      "loss": 0.8989,
+      "step": 91810
+    },
+    {
+      "epoch": 0.6646543175023707,
+      "grad_norm": 0.1505679041147232,
+      "learning_rate": 4.335352921163689e-06,
+      "loss": 0.8885,
+      "step": 91820
+    },
+    {
+      "epoch": 0.6647267041629569,
+      "grad_norm": 0.16631951928138733,
+      "learning_rate": 4.3352805345031026e-06,
+      "loss": 0.8972,
+      "step": 91830
+    },
+    {
+      "epoch": 0.664799090823543,
+      "grad_norm": 0.17262007296085358,
+      "learning_rate": 4.335208147842516e-06,
+      "loss": 0.8933,
+      "step": 91840
+    },
+    {
+      "epoch": 0.6648714774841292,
+      "grad_norm": 0.16659270226955414,
+      "learning_rate": 4.33513576118193e-06,
+      "loss": 0.9059,
+      "step": 91850
+    },
+    {
+      "epoch": 0.6649438641447154,
+      "grad_norm": 0.15563061833381653,
+      "learning_rate": 4.335063374521343e-06,
+      "loss": 0.9071,
+      "step": 91860
+    },
+    {
+      "epoch": 0.6650162508053016,
+      "grad_norm": 0.16777054965496063,
+      "learning_rate": 4.334990987860757e-06,
+      "loss": 0.8994,
+      "step": 91870
+    },
+    {
+      "epoch": 0.6650886374658878,
+      "grad_norm": 0.15433162450790405,
+      "learning_rate": 4.334918601200171e-06,
+      "loss": 0.9064,
+      "step": 91880
+    },
+    {
+      "epoch": 0.6651610241264739,
+      "grad_norm": 0.16042456030845642,
+      "learning_rate": 4.334846214539585e-06,
+      "loss": 0.8952,
+      "step": 91890
+    },
+    {
+      "epoch": 0.6652334107870601,
+      "grad_norm": 0.189911887049675,
+      "learning_rate": 4.334773827878999e-06,
+      "loss": 0.8903,
+      "step": 91900
+    },
+    {
+      "epoch": 0.6653057974476464,
+      "grad_norm": 0.1497250199317932,
+      "learning_rate": 4.334701441218412e-06,
+      "loss": 0.8836,
+      "step": 91910
+    },
+    {
+      "epoch": 0.6653781841082326,
+      "grad_norm": 0.16076631844043732,
+      "learning_rate": 4.334629054557826e-06,
+      "loss": 0.8971,
+      "step": 91920
+    },
+    {
+      "epoch": 0.6654505707688187,
+      "grad_norm": 0.15558098256587982,
+      "learning_rate": 4.33455666789724e-06,
+      "loss": 0.8838,
+      "step": 91930
+    },
+    {
+      "epoch": 0.6655229574294049,
+      "grad_norm": 0.1555255949497223,
+      "learning_rate": 4.334484281236654e-06,
+      "loss": 0.9026,
+      "step": 91940
+    },
+    {
+      "epoch": 0.6655953440899911,
+      "grad_norm": 0.15714456140995026,
+      "learning_rate": 4.334411894576068e-06,
+      "loss": 0.8889,
+      "step": 91950
+    },
+    {
+      "epoch": 0.6656677307505773,
+      "grad_norm": 0.15416781604290009,
+      "learning_rate": 4.334339507915481e-06,
+      "loss": 0.8943,
+      "step": 91960
+    },
+    {
+      "epoch": 0.6657401174111635,
+      "grad_norm": 0.16116608679294586,
+      "learning_rate": 4.334267121254896e-06,
+      "loss": 0.8844,
+      "step": 91970
+    },
+    {
+      "epoch": 0.6658125040717496,
+      "grad_norm": 0.141585111618042,
+      "learning_rate": 4.334194734594309e-06,
+      "loss": 0.8845,
+      "step": 91980
+    },
+    {
+      "epoch": 0.6658848907323358,
+      "grad_norm": 0.1526327133178711,
+      "learning_rate": 4.334122347933723e-06,
+      "loss": 0.8922,
+      "step": 91990
+    },
+    {
+      "epoch": 0.665957277392922,
+      "grad_norm": 0.16217315196990967,
+      "learning_rate": 4.3340499612731366e-06,
+      "loss": 0.889,
+      "step": 92000
+    },
+    {
+      "epoch": 0.6660296640535083,
+      "grad_norm": 0.16622312366962433,
+      "learning_rate": 4.333977574612551e-06,
+      "loss": 0.8993,
+      "step": 92010
+    },
+    {
+      "epoch": 0.6661020507140945,
+      "grad_norm": 0.19960935413837433,
+      "learning_rate": 4.333905187951965e-06,
+      "loss": 0.8982,
+      "step": 92020
+    },
+    {
+      "epoch": 0.6661744373746806,
+      "grad_norm": 0.16707240045070648,
+      "learning_rate": 4.333832801291378e-06,
+      "loss": 0.9088,
+      "step": 92030
+    },
+    {
+      "epoch": 0.6662468240352668,
+      "grad_norm": 0.15086571872234344,
+      "learning_rate": 4.333760414630792e-06,
+      "loss": 0.8991,
+      "step": 92040
+    },
+    {
+      "epoch": 0.666319210695853,
+      "grad_norm": 0.17651046812534332,
+      "learning_rate": 4.333688027970206e-06,
+      "loss": 0.8987,
+      "step": 92050
+    },
+    {
+      "epoch": 0.6663915973564392,
+      "grad_norm": 0.1702297031879425,
+      "learning_rate": 4.33361564130962e-06,
+      "loss": 0.898,
+      "step": 92060
+    },
+    {
+      "epoch": 0.6664639840170253,
+      "grad_norm": 0.17619454860687256,
+      "learning_rate": 4.333543254649034e-06,
+      "loss": 0.894,
+      "step": 92070
+    },
+    {
+      "epoch": 0.6665363706776115,
+      "grad_norm": 0.16067981719970703,
+      "learning_rate": 4.333470867988447e-06,
+      "loss": 0.9031,
+      "step": 92080
+    },
+    {
+      "epoch": 0.6666087573381977,
+      "grad_norm": 0.16957126557826996,
+      "learning_rate": 4.333398481327861e-06,
+      "loss": 0.8966,
+      "step": 92090
+    },
+    {
+      "epoch": 0.6666811439987839,
+      "grad_norm": 0.1614599972963333,
+      "learning_rate": 4.333326094667275e-06,
+      "loss": 0.8976,
+      "step": 92100
+    },
+    {
+      "epoch": 0.66675353065937,
+      "grad_norm": 0.17003393173217773,
+      "learning_rate": 4.333253708006689e-06,
+      "loss": 0.9072,
+      "step": 92110
+    },
+    {
+      "epoch": 0.6668259173199563,
+      "grad_norm": 0.17658734321594238,
+      "learning_rate": 4.3331813213461025e-06,
+      "loss": 0.8964,
+      "step": 92120
+    },
+    {
+      "epoch": 0.6668983039805425,
+      "grad_norm": 0.15573443472385406,
+      "learning_rate": 4.333108934685516e-06,
+      "loss": 0.8984,
+      "step": 92130
+    },
+    {
+      "epoch": 0.6669706906411287,
+      "grad_norm": 0.17280741035938263,
+      "learning_rate": 4.333036548024931e-06,
+      "loss": 0.8982,
+      "step": 92140
+    },
+    {
+      "epoch": 0.6670430773017149,
+      "grad_norm": 0.1615643948316574,
+      "learning_rate": 4.332964161364344e-06,
+      "loss": 0.8958,
+      "step": 92150
+    },
+    {
+      "epoch": 0.667115463962301,
+      "grad_norm": 0.16449616849422455,
+      "learning_rate": 4.332891774703758e-06,
+      "loss": 0.8938,
+      "step": 92160
+    },
+    {
+      "epoch": 0.6671878506228872,
+      "grad_norm": 0.1627790331840515,
+      "learning_rate": 4.3328193880431714e-06,
+      "loss": 0.8861,
+      "step": 92170
+    },
+    {
+      "epoch": 0.6672602372834734,
+      "grad_norm": 0.15564194321632385,
+      "learning_rate": 4.332747001382586e-06,
+      "loss": 0.9061,
+      "step": 92180
+    },
+    {
+      "epoch": 0.6673326239440596,
+      "grad_norm": 0.18332041800022125,
+      "learning_rate": 4.3326746147219995e-06,
+      "loss": 0.9071,
+      "step": 92190
+    },
+    {
+      "epoch": 0.6674050106046457,
+      "grad_norm": 0.165358766913414,
+      "learning_rate": 4.332602228061413e-06,
+      "loss": 0.8964,
+      "step": 92200
+    },
+    {
+      "epoch": 0.6674773972652319,
+      "grad_norm": 0.1612190455198288,
+      "learning_rate": 4.332529841400827e-06,
+      "loss": 0.8976,
+      "step": 92210
+    },
+    {
+      "epoch": 0.6675497839258181,
+      "grad_norm": 0.16194748878479004,
+      "learning_rate": 4.332457454740241e-06,
+      "loss": 0.8953,
+      "step": 92220
+    },
+    {
+      "epoch": 0.6676221705864044,
+      "grad_norm": 0.16928349435329437,
+      "learning_rate": 4.332385068079655e-06,
+      "loss": 0.9029,
+      "step": 92230
+    },
+    {
+      "epoch": 0.6676945572469906,
+      "grad_norm": 0.15044300258159637,
+      "learning_rate": 4.3323126814190684e-06,
+      "loss": 0.8986,
+      "step": 92240
+    },
+    {
+      "epoch": 0.6677669439075767,
+      "grad_norm": 0.1588992476463318,
+      "learning_rate": 4.332240294758482e-06,
+      "loss": 0.9034,
+      "step": 92250
+    },
+    {
+      "epoch": 0.6678393305681629,
+      "grad_norm": 0.16093967854976654,
+      "learning_rate": 4.3321679080978965e-06,
+      "loss": 0.8845,
+      "step": 92260
+    },
+    {
+      "epoch": 0.6679117172287491,
+      "grad_norm": 0.17110575735569,
+      "learning_rate": 4.33209552143731e-06,
+      "loss": 0.8848,
+      "step": 92270
+    },
+    {
+      "epoch": 0.6679841038893353,
+      "grad_norm": 0.15584023296833038,
+      "learning_rate": 4.332023134776724e-06,
+      "loss": 0.8858,
+      "step": 92280
+    },
+    {
+      "epoch": 0.6680564905499214,
+      "grad_norm": 0.16196756064891815,
+      "learning_rate": 4.331950748116137e-06,
+      "loss": 0.8951,
+      "step": 92290
+    },
+    {
+      "epoch": 0.6681288772105076,
+      "grad_norm": 0.1788647472858429,
+      "learning_rate": 4.331878361455552e-06,
+      "loss": 0.8988,
+      "step": 92300
+    },
+    {
+      "epoch": 0.6682012638710938,
+      "grad_norm": 0.1447635442018509,
+      "learning_rate": 4.3318059747949655e-06,
+      "loss": 0.8857,
+      "step": 92310
+    },
+    {
+      "epoch": 0.66827365053168,
+      "grad_norm": 0.16333886981010437,
+      "learning_rate": 4.331733588134379e-06,
+      "loss": 0.8831,
+      "step": 92320
+    },
+    {
+      "epoch": 0.6683460371922663,
+      "grad_norm": 0.16351155936717987,
+      "learning_rate": 4.331661201473793e-06,
+      "loss": 0.9038,
+      "step": 92330
+    },
+    {
+      "epoch": 0.6684184238528524,
+      "grad_norm": 0.16826938092708588,
+      "learning_rate": 4.331588814813207e-06,
+      "loss": 0.8982,
+      "step": 92340
+    },
+    {
+      "epoch": 0.6684908105134386,
+      "grad_norm": 0.16592121124267578,
+      "learning_rate": 4.331516428152621e-06,
+      "loss": 0.8828,
+      "step": 92350
+    },
+    {
+      "epoch": 0.6685631971740248,
+      "grad_norm": 0.1630096733570099,
+      "learning_rate": 4.331444041492034e-06,
+      "loss": 0.8867,
+      "step": 92360
+    },
+    {
+      "epoch": 0.668635583834611,
+      "grad_norm": 0.18398292362689972,
+      "learning_rate": 4.331371654831448e-06,
+      "loss": 0.9008,
+      "step": 92370
+    },
+    {
+      "epoch": 0.6687079704951971,
+      "grad_norm": 0.16176696121692657,
+      "learning_rate": 4.3312992681708625e-06,
+      "loss": 0.9025,
+      "step": 92380
+    },
+    {
+      "epoch": 0.6687803571557833,
+      "grad_norm": 0.15559855103492737,
+      "learning_rate": 4.331226881510275e-06,
+      "loss": 0.9031,
+      "step": 92390
+    },
+    {
+      "epoch": 0.6688527438163695,
+      "grad_norm": 0.16191458702087402,
+      "learning_rate": 4.331154494849689e-06,
+      "loss": 0.8954,
+      "step": 92400
+    },
+    {
+      "epoch": 0.6689251304769557,
+      "grad_norm": 0.15984484553337097,
+      "learning_rate": 4.331082108189103e-06,
+      "loss": 0.8888,
+      "step": 92410
+    },
+    {
+      "epoch": 0.6689975171375419,
+      "grad_norm": 0.15444859862327576,
+      "learning_rate": 4.331009721528517e-06,
+      "loss": 0.8966,
+      "step": 92420
+    },
+    {
+      "epoch": 0.669069903798128,
+      "grad_norm": 0.1950903832912445,
+      "learning_rate": 4.3309373348679305e-06,
+      "loss": 0.901,
+      "step": 92430
+    },
+    {
+      "epoch": 0.6691422904587143,
+      "grad_norm": 0.1582936942577362,
+      "learning_rate": 4.330864948207344e-06,
+      "loss": 0.891,
+      "step": 92440
+    },
+    {
+      "epoch": 0.6692146771193005,
+      "grad_norm": 0.15177102386951447,
+      "learning_rate": 4.330792561546759e-06,
+      "loss": 0.8893,
+      "step": 92450
+    },
+    {
+      "epoch": 0.6692870637798867,
+      "grad_norm": 0.14646144211292267,
+      "learning_rate": 4.330720174886172e-06,
+      "loss": 0.8964,
+      "step": 92460
+    },
+    {
+      "epoch": 0.6693594504404728,
+      "grad_norm": 0.17326298356056213,
+      "learning_rate": 4.330647788225586e-06,
+      "loss": 0.8839,
+      "step": 92470
+    },
+    {
+      "epoch": 0.669431837101059,
+      "grad_norm": 0.16231794655323029,
+      "learning_rate": 4.3305754015649995e-06,
+      "loss": 0.8983,
+      "step": 92480
+    },
+    {
+      "epoch": 0.6695042237616452,
+      "grad_norm": 0.1864873170852661,
+      "learning_rate": 4.330503014904414e-06,
+      "loss": 0.897,
+      "step": 92490
+    },
+    {
+      "epoch": 0.6695766104222314,
+      "grad_norm": 0.16087661683559418,
+      "learning_rate": 4.3304306282438275e-06,
+      "loss": 0.8815,
+      "step": 92500
+    },
+    {
+      "epoch": 0.6696489970828176,
+      "grad_norm": 0.1563531458377838,
+      "learning_rate": 4.330358241583241e-06,
+      "loss": 0.8971,
+      "step": 92510
+    },
+    {
+      "epoch": 0.6697213837434037,
+      "grad_norm": 0.15433651208877563,
+      "learning_rate": 4.330285854922655e-06,
+      "loss": 0.8948,
+      "step": 92520
+    },
+    {
+      "epoch": 0.6697937704039899,
+      "grad_norm": 0.18349480628967285,
+      "learning_rate": 4.330213468262069e-06,
+      "loss": 0.9053,
+      "step": 92530
+    },
+    {
+      "epoch": 0.6698661570645762,
+      "grad_norm": 0.186500683426857,
+      "learning_rate": 4.330141081601483e-06,
+      "loss": 0.8941,
+      "step": 92540
+    },
+    {
+      "epoch": 0.6699385437251624,
+      "grad_norm": 0.17492738366127014,
+      "learning_rate": 4.3300686949408965e-06,
+      "loss": 0.8914,
+      "step": 92550
+    },
+    {
+      "epoch": 0.6700109303857485,
+      "grad_norm": 0.1956550031900406,
+      "learning_rate": 4.32999630828031e-06,
+      "loss": 0.8992,
+      "step": 92560
+    },
+    {
+      "epoch": 0.6700833170463347,
+      "grad_norm": 0.1706896424293518,
+      "learning_rate": 4.3299239216197246e-06,
+      "loss": 0.8879,
+      "step": 92570
+    },
+    {
+      "epoch": 0.6701557037069209,
+      "grad_norm": 0.16048724949359894,
+      "learning_rate": 4.329851534959138e-06,
+      "loss": 0.8983,
+      "step": 92580
+    },
+    {
+      "epoch": 0.6702280903675071,
+      "grad_norm": 0.14965306222438812,
+      "learning_rate": 4.329779148298552e-06,
+      "loss": 0.8997,
+      "step": 92590
+    },
+    {
+      "epoch": 0.6703004770280933,
+      "grad_norm": 0.15428891777992249,
+      "learning_rate": 4.329706761637965e-06,
+      "loss": 0.8933,
+      "step": 92600
+    },
+    {
+      "epoch": 0.6703728636886794,
+      "grad_norm": 0.15892021358013153,
+      "learning_rate": 4.32963437497738e-06,
+      "loss": 0.8951,
+      "step": 92610
+    },
+    {
+      "epoch": 0.6704452503492656,
+      "grad_norm": 0.1577247977256775,
+      "learning_rate": 4.3295619883167935e-06,
+      "loss": 0.8999,
+      "step": 92620
+    },
+    {
+      "epoch": 0.6705176370098518,
+      "grad_norm": 0.15700013935565948,
+      "learning_rate": 4.329489601656207e-06,
+      "loss": 0.8752,
+      "step": 92630
+    },
+    {
+      "epoch": 0.670590023670438,
+      "grad_norm": 0.18034988641738892,
+      "learning_rate": 4.329417214995621e-06,
+      "loss": 0.9094,
+      "step": 92640
+    },
+    {
+      "epoch": 0.6706624103310243,
+      "grad_norm": 0.15093854069709778,
+      "learning_rate": 4.329344828335035e-06,
+      "loss": 0.909,
+      "step": 92650
+    },
+    {
+      "epoch": 0.6707347969916104,
+      "grad_norm": 0.15224012732505798,
+      "learning_rate": 4.329272441674449e-06,
+      "loss": 0.8964,
+      "step": 92660
+    },
+    {
+      "epoch": 0.6708071836521966,
+      "grad_norm": 0.16921748220920563,
+      "learning_rate": 4.329200055013862e-06,
+      "loss": 0.893,
+      "step": 92670
+    },
+    {
+      "epoch": 0.6708795703127828,
+      "grad_norm": 0.15135690569877625,
+      "learning_rate": 4.329127668353276e-06,
+      "loss": 0.8859,
+      "step": 92680
+    },
+    {
+      "epoch": 0.670951956973369,
+      "grad_norm": 0.15335379540920258,
+      "learning_rate": 4.3290552816926905e-06,
+      "loss": 0.8956,
+      "step": 92690
+    },
+    {
+      "epoch": 0.6710243436339551,
+      "grad_norm": 0.1561633050441742,
+      "learning_rate": 4.328982895032104e-06,
+      "loss": 0.8801,
+      "step": 92700
+    },
+    {
+      "epoch": 0.6710967302945413,
+      "grad_norm": 0.1546168178319931,
+      "learning_rate": 4.328910508371518e-06,
+      "loss": 0.9001,
+      "step": 92710
+    },
+    {
+      "epoch": 0.6711691169551275,
+      "grad_norm": 0.16860635578632355,
+      "learning_rate": 4.328838121710931e-06,
+      "loss": 0.9041,
+      "step": 92720
+    },
+    {
+      "epoch": 0.6712415036157137,
+      "grad_norm": 0.1650303453207016,
+      "learning_rate": 4.328765735050345e-06,
+      "loss": 0.8919,
+      "step": 92730
+    },
+    {
+      "epoch": 0.6713138902762998,
+      "grad_norm": 0.15512806177139282,
+      "learning_rate": 4.328693348389759e-06,
+      "loss": 0.884,
+      "step": 92740
+    },
+    {
+      "epoch": 0.671386276936886,
+      "grad_norm": 0.17656339704990387,
+      "learning_rate": 4.328620961729173e-06,
+      "loss": 0.8928,
+      "step": 92750
+    },
+    {
+      "epoch": 0.6714586635974723,
+      "grad_norm": 0.1842651218175888,
+      "learning_rate": 4.328548575068587e-06,
+      "loss": 0.9057,
+      "step": 92760
+    },
+    {
+      "epoch": 0.6715310502580585,
+      "grad_norm": 0.15689325332641602,
+      "learning_rate": 4.328476188408e-06,
+      "loss": 0.8871,
+      "step": 92770
+    },
+    {
+      "epoch": 0.6716034369186447,
+      "grad_norm": 0.16614311933517456,
+      "learning_rate": 4.328403801747415e-06,
+      "loss": 0.8998,
+      "step": 92780
+    },
+    {
+      "epoch": 0.6716758235792308,
+      "grad_norm": 0.16957563161849976,
+      "learning_rate": 4.328331415086828e-06,
+      "loss": 0.8937,
+      "step": 92790
+    },
+    {
+      "epoch": 0.671748210239817,
+      "grad_norm": 0.15638667345046997,
+      "learning_rate": 4.328259028426242e-06,
+      "loss": 0.894,
+      "step": 92800
+    },
+    {
+      "epoch": 0.6718205969004032,
+      "grad_norm": 0.16604597866535187,
+      "learning_rate": 4.328186641765656e-06,
+      "loss": 0.8891,
+      "step": 92810
+    },
+    {
+      "epoch": 0.6718929835609894,
+      "grad_norm": 0.1521102339029312,
+      "learning_rate": 4.32811425510507e-06,
+      "loss": 0.8946,
+      "step": 92820
+    },
+    {
+      "epoch": 0.6719653702215755,
+      "grad_norm": 0.1847507506608963,
+      "learning_rate": 4.328041868444484e-06,
+      "loss": 0.906,
+      "step": 92830
+    },
+    {
+      "epoch": 0.6720377568821617,
+      "grad_norm": 0.15854564309120178,
+      "learning_rate": 4.327969481783897e-06,
+      "loss": 0.8913,
+      "step": 92840
+    },
+    {
+      "epoch": 0.6721101435427479,
+      "grad_norm": 0.16507163643836975,
+      "learning_rate": 4.327897095123311e-06,
+      "loss": 0.8888,
+      "step": 92850
+    },
+    {
+      "epoch": 0.6721825302033342,
+      "grad_norm": 0.1640719473361969,
+      "learning_rate": 4.327824708462725e-06,
+      "loss": 0.8969,
+      "step": 92860
+    },
+    {
+      "epoch": 0.6722549168639204,
+      "grad_norm": 0.15480723977088928,
+      "learning_rate": 4.327752321802139e-06,
+      "loss": 0.9023,
+      "step": 92870
+    },
+    {
+      "epoch": 0.6723273035245065,
+      "grad_norm": 0.1514626145362854,
+      "learning_rate": 4.327679935141553e-06,
+      "loss": 0.8934,
+      "step": 92880
+    },
+    {
+      "epoch": 0.6723996901850927,
+      "grad_norm": 0.16035489737987518,
+      "learning_rate": 4.327607548480966e-06,
+      "loss": 0.8847,
+      "step": 92890
+    },
+    {
+      "epoch": 0.6724720768456789,
+      "grad_norm": 0.15631358325481415,
+      "learning_rate": 4.327535161820381e-06,
+      "loss": 0.9011,
+      "step": 92900
+    },
+    {
+      "epoch": 0.6725444635062651,
+      "grad_norm": 0.16039063036441803,
+      "learning_rate": 4.327462775159794e-06,
+      "loss": 0.8872,
+      "step": 92910
+    },
+    {
+      "epoch": 0.6726168501668512,
+      "grad_norm": 0.15340082347393036,
+      "learning_rate": 4.327390388499207e-06,
+      "loss": 0.8918,
+      "step": 92920
+    },
+    {
+      "epoch": 0.6726892368274374,
+      "grad_norm": 0.16243009269237518,
+      "learning_rate": 4.3273180018386215e-06,
+      "loss": 0.8937,
+      "step": 92930
+    },
+    {
+      "epoch": 0.6727616234880236,
+      "grad_norm": 0.15922003984451294,
+      "learning_rate": 4.327245615178035e-06,
+      "loss": 0.8906,
+      "step": 92940
+    },
+    {
+      "epoch": 0.6728340101486098,
+      "grad_norm": 0.1569734811782837,
+      "learning_rate": 4.327173228517449e-06,
+      "loss": 0.8916,
+      "step": 92950
+    },
+    {
+      "epoch": 0.672906396809196,
+      "grad_norm": 0.1412716507911682,
+      "learning_rate": 4.327100841856862e-06,
+      "loss": 0.881,
+      "step": 92960
+    },
+    {
+      "epoch": 0.6729787834697822,
+      "grad_norm": 0.16564616560935974,
+      "learning_rate": 4.327028455196277e-06,
+      "loss": 0.8876,
+      "step": 92970
+    },
+    {
+      "epoch": 0.6730511701303684,
+      "grad_norm": 0.18585346639156342,
+      "learning_rate": 4.3269560685356904e-06,
+      "loss": 0.8932,
+      "step": 92980
+    },
+    {
+      "epoch": 0.6731235567909546,
+      "grad_norm": 0.15012194216251373,
+      "learning_rate": 4.326883681875104e-06,
+      "loss": 0.8916,
+      "step": 92990
+    },
+    {
+      "epoch": 0.6731959434515408,
+      "grad_norm": 0.14845649898052216,
+      "learning_rate": 4.326811295214518e-06,
+      "loss": 0.8963,
+      "step": 93000
+    },
+    {
+      "epoch": 0.673268330112127,
+      "grad_norm": 0.15878477692604065,
+      "learning_rate": 4.326738908553932e-06,
+      "loss": 0.8983,
+      "step": 93010
+    },
+    {
+      "epoch": 0.6733407167727131,
+      "grad_norm": 0.15004906058311462,
+      "learning_rate": 4.326666521893346e-06,
+      "loss": 0.8929,
+      "step": 93020
+    },
+    {
+      "epoch": 0.6734131034332993,
+      "grad_norm": 0.16124626994132996,
+      "learning_rate": 4.326594135232759e-06,
+      "loss": 0.8837,
+      "step": 93030
+    },
+    {
+      "epoch": 0.6734854900938855,
+      "grad_norm": 0.15695308148860931,
+      "learning_rate": 4.326521748572173e-06,
+      "loss": 0.8991,
+      "step": 93040
+    },
+    {
+      "epoch": 0.6735578767544717,
+      "grad_norm": 0.1830526888370514,
+      "learning_rate": 4.3264493619115874e-06,
+      "loss": 0.8826,
+      "step": 93050
+    },
+    {
+      "epoch": 0.6736302634150578,
+      "grad_norm": 0.1436559408903122,
+      "learning_rate": 4.326376975251001e-06,
+      "loss": 0.8939,
+      "step": 93060
+    },
+    {
+      "epoch": 0.673702650075644,
+      "grad_norm": 0.15827639400959015,
+      "learning_rate": 4.326304588590415e-06,
+      "loss": 0.907,
+      "step": 93070
+    },
+    {
+      "epoch": 0.6737750367362303,
+      "grad_norm": 0.15548427402973175,
+      "learning_rate": 4.326232201929828e-06,
+      "loss": 0.9136,
+      "step": 93080
+    },
+    {
+      "epoch": 0.6738474233968165,
+      "grad_norm": 0.15338309109210968,
+      "learning_rate": 4.326159815269243e-06,
+      "loss": 0.8979,
+      "step": 93090
+    },
+    {
+      "epoch": 0.6739198100574026,
+      "grad_norm": 0.16544108092784882,
+      "learning_rate": 4.326087428608656e-06,
+      "loss": 0.8817,
+      "step": 93100
+    },
+    {
+      "epoch": 0.6739921967179888,
+      "grad_norm": 0.16785688698291779,
+      "learning_rate": 4.32601504194807e-06,
+      "loss": 0.8912,
+      "step": 93110
+    },
+    {
+      "epoch": 0.674064583378575,
+      "grad_norm": 0.17321330308914185,
+      "learning_rate": 4.325942655287484e-06,
+      "loss": 0.8863,
+      "step": 93120
+    },
+    {
+      "epoch": 0.6741369700391612,
+      "grad_norm": 0.16258351504802704,
+      "learning_rate": 4.325870268626898e-06,
+      "loss": 0.9073,
+      "step": 93130
+    },
+    {
+      "epoch": 0.6742093566997474,
+      "grad_norm": 0.16058233380317688,
+      "learning_rate": 4.325797881966312e-06,
+      "loss": 0.9084,
+      "step": 93140
+    },
+    {
+      "epoch": 0.6742817433603335,
+      "grad_norm": 0.16303308308124542,
+      "learning_rate": 4.325725495305725e-06,
+      "loss": 0.8965,
+      "step": 93150
+    },
+    {
+      "epoch": 0.6743541300209197,
+      "grad_norm": 0.15831464529037476,
+      "learning_rate": 4.325653108645139e-06,
+      "loss": 0.8865,
+      "step": 93160
+    },
+    {
+      "epoch": 0.6744265166815059,
+      "grad_norm": 0.1554798036813736,
+      "learning_rate": 4.325580721984553e-06,
+      "loss": 0.8803,
+      "step": 93170
+    },
+    {
+      "epoch": 0.6744989033420922,
+      "grad_norm": 0.16166193783283234,
+      "learning_rate": 4.325508335323967e-06,
+      "loss": 0.9018,
+      "step": 93180
+    },
+    {
+      "epoch": 0.6745712900026783,
+      "grad_norm": 0.15309026837348938,
+      "learning_rate": 4.325435948663381e-06,
+      "loss": 0.8972,
+      "step": 93190
+    },
+    {
+      "epoch": 0.6746436766632645,
+      "grad_norm": 0.15099987387657166,
+      "learning_rate": 4.325363562002794e-06,
+      "loss": 0.9075,
+      "step": 93200
+    },
+    {
+      "epoch": 0.6747160633238507,
+      "grad_norm": 0.17467179894447327,
+      "learning_rate": 4.325291175342209e-06,
+      "loss": 0.9024,
+      "step": 93210
+    },
+    {
+      "epoch": 0.6747884499844369,
+      "grad_norm": 0.15384790301322937,
+      "learning_rate": 4.325218788681622e-06,
+      "loss": 0.9,
+      "step": 93220
+    },
+    {
+      "epoch": 0.674860836645023,
+      "grad_norm": 0.1675538271665573,
+      "learning_rate": 4.325146402021036e-06,
+      "loss": 0.893,
+      "step": 93230
+    },
+    {
+      "epoch": 0.6749332233056092,
+      "grad_norm": 0.15132692456245422,
+      "learning_rate": 4.3250740153604495e-06,
+      "loss": 0.9021,
+      "step": 93240
+    },
+    {
+      "epoch": 0.6750056099661954,
+      "grad_norm": 0.15508437156677246,
+      "learning_rate": 4.325001628699864e-06,
+      "loss": 0.9039,
+      "step": 93250
+    },
+    {
+      "epoch": 0.6750779966267816,
+      "grad_norm": 0.15347804129123688,
+      "learning_rate": 4.324929242039278e-06,
+      "loss": 0.8787,
+      "step": 93260
+    },
+    {
+      "epoch": 0.6751503832873678,
+      "grad_norm": 0.15671610832214355,
+      "learning_rate": 4.324856855378691e-06,
+      "loss": 0.8973,
+      "step": 93270
+    },
+    {
+      "epoch": 0.6752227699479539,
+      "grad_norm": 0.15267059206962585,
+      "learning_rate": 4.324784468718105e-06,
+      "loss": 0.8862,
+      "step": 93280
+    },
+    {
+      "epoch": 0.6752951566085402,
+      "grad_norm": 0.1843479573726654,
+      "learning_rate": 4.324712082057519e-06,
+      "loss": 0.9027,
+      "step": 93290
+    },
+    {
+      "epoch": 0.6753675432691264,
+      "grad_norm": 0.18705353140830994,
+      "learning_rate": 4.324639695396933e-06,
+      "loss": 0.8873,
+      "step": 93300
+    },
+    {
+      "epoch": 0.6754399299297126,
+      "grad_norm": 0.16083325445652008,
+      "learning_rate": 4.3245673087363466e-06,
+      "loss": 0.8973,
+      "step": 93310
+    },
+    {
+      "epoch": 0.6755123165902988,
+      "grad_norm": 0.16239725053310394,
+      "learning_rate": 4.32449492207576e-06,
+      "loss": 0.8963,
+      "step": 93320
+    },
+    {
+      "epoch": 0.6755847032508849,
+      "grad_norm": 0.15586166083812714,
+      "learning_rate": 4.324422535415174e-06,
+      "loss": 0.8988,
+      "step": 93330
+    },
+    {
+      "epoch": 0.6756570899114711,
+      "grad_norm": 0.17255674302577972,
+      "learning_rate": 4.324350148754588e-06,
+      "loss": 0.8979,
+      "step": 93340
+    },
+    {
+      "epoch": 0.6757294765720573,
+      "grad_norm": 0.15488573908805847,
+      "learning_rate": 4.324277762094002e-06,
+      "loss": 0.8938,
+      "step": 93350
+    },
+    {
+      "epoch": 0.6758018632326435,
+      "grad_norm": 0.1733180284500122,
+      "learning_rate": 4.3242053754334155e-06,
+      "loss": 0.8936,
+      "step": 93360
+    },
+    {
+      "epoch": 0.6758742498932296,
+      "grad_norm": 0.16540446877479553,
+      "learning_rate": 4.324132988772829e-06,
+      "loss": 0.8812,
+      "step": 93370
+    },
+    {
+      "epoch": 0.6759466365538158,
+      "grad_norm": 0.16037392616271973,
+      "learning_rate": 4.3240606021122436e-06,
+      "loss": 0.8979,
+      "step": 93380
+    },
+    {
+      "epoch": 0.6760190232144021,
+      "grad_norm": 0.17287902534008026,
+      "learning_rate": 4.323988215451657e-06,
+      "loss": 0.8928,
+      "step": 93390
+    },
+    {
+      "epoch": 0.6760914098749883,
+      "grad_norm": 0.19038693606853485,
+      "learning_rate": 4.323915828791071e-06,
+      "loss": 0.896,
+      "step": 93400
+    },
+    {
+      "epoch": 0.6761637965355745,
+      "grad_norm": 0.14845329523086548,
+      "learning_rate": 4.323843442130484e-06,
+      "loss": 0.8969,
+      "step": 93410
+    },
+    {
+      "epoch": 0.6762361831961606,
+      "grad_norm": 0.15594539046287537,
+      "learning_rate": 4.323771055469899e-06,
+      "loss": 0.8967,
+      "step": 93420
+    },
+    {
+      "epoch": 0.6763085698567468,
+      "grad_norm": 0.1664988100528717,
+      "learning_rate": 4.3236986688093125e-06,
+      "loss": 0.8907,
+      "step": 93430
+    },
+    {
+      "epoch": 0.676380956517333,
+      "grad_norm": 0.16152945160865784,
+      "learning_rate": 4.323626282148726e-06,
+      "loss": 0.8887,
+      "step": 93440
+    },
+    {
+      "epoch": 0.6764533431779192,
+      "grad_norm": 0.16336587071418762,
+      "learning_rate": 4.32355389548814e-06,
+      "loss": 0.8937,
+      "step": 93450
+    },
+    {
+      "epoch": 0.6765257298385053,
+      "grad_norm": 0.17990297079086304,
+      "learning_rate": 4.323481508827553e-06,
+      "loss": 0.8885,
+      "step": 93460
+    },
+    {
+      "epoch": 0.6765981164990915,
+      "grad_norm": 0.1632242649793625,
+      "learning_rate": 4.323409122166967e-06,
+      "loss": 0.9011,
+      "step": 93470
+    },
+    {
+      "epoch": 0.6766705031596777,
+      "grad_norm": 0.1489850878715515,
+      "learning_rate": 4.323336735506381e-06,
+      "loss": 0.8834,
+      "step": 93480
+    },
+    {
+      "epoch": 0.6767428898202639,
+      "grad_norm": 0.16526971757411957,
+      "learning_rate": 4.323264348845795e-06,
+      "loss": 0.8924,
+      "step": 93490
+    },
+    {
+      "epoch": 0.6768152764808502,
+      "grad_norm": 0.14482301473617554,
+      "learning_rate": 4.323191962185209e-06,
+      "loss": 0.8818,
+      "step": 93500
+    },
+    {
+      "epoch": 0.6768876631414363,
+      "grad_norm": 0.19359660148620605,
+      "learning_rate": 4.323119575524622e-06,
+      "loss": 0.9027,
+      "step": 93510
+    },
+    {
+      "epoch": 0.6769600498020225,
+      "grad_norm": 0.15199698507785797,
+      "learning_rate": 4.323047188864036e-06,
+      "loss": 0.9103,
+      "step": 93520
+    },
+    {
+      "epoch": 0.6770324364626087,
+      "grad_norm": 0.1570311337709427,
+      "learning_rate": 4.32297480220345e-06,
+      "loss": 0.8866,
+      "step": 93530
+    },
+    {
+      "epoch": 0.6771048231231949,
+      "grad_norm": 0.16493958234786987,
+      "learning_rate": 4.322902415542864e-06,
+      "loss": 0.8965,
+      "step": 93540
+    },
+    {
+      "epoch": 0.677177209783781,
+      "grad_norm": 0.15966171026229858,
+      "learning_rate": 4.322830028882278e-06,
+      "loss": 0.9059,
+      "step": 93550
+    },
+    {
+      "epoch": 0.6772495964443672,
+      "grad_norm": 0.15570668876171112,
+      "learning_rate": 4.322757642221691e-06,
+      "loss": 0.9031,
+      "step": 93560
+    },
+    {
+      "epoch": 0.6773219831049534,
+      "grad_norm": 0.15496212244033813,
+      "learning_rate": 4.322685255561106e-06,
+      "loss": 0.9177,
+      "step": 93570
+    },
+    {
+      "epoch": 0.6773943697655396,
+      "grad_norm": 0.15817132592201233,
+      "learning_rate": 4.322612868900519e-06,
+      "loss": 0.8852,
+      "step": 93580
+    },
+    {
+      "epoch": 0.6774667564261257,
+      "grad_norm": 0.15422877669334412,
+      "learning_rate": 4.322540482239933e-06,
+      "loss": 0.8944,
+      "step": 93590
+    },
+    {
+      "epoch": 0.6775391430867119,
+      "grad_norm": 0.15881921350955963,
+      "learning_rate": 4.3224680955793465e-06,
+      "loss": 0.9014,
+      "step": 93600
+    },
+    {
+      "epoch": 0.6776115297472982,
+      "grad_norm": 0.24979665875434875,
+      "learning_rate": 4.322395708918761e-06,
+      "loss": 0.8926,
+      "step": 93610
+    },
+    {
+      "epoch": 0.6776839164078844,
+      "grad_norm": 0.19371944665908813,
+      "learning_rate": 4.322323322258175e-06,
+      "loss": 0.879,
+      "step": 93620
+    },
+    {
+      "epoch": 0.6777563030684706,
+      "grad_norm": 0.15385429561138153,
+      "learning_rate": 4.322250935597588e-06,
+      "loss": 0.9084,
+      "step": 93630
+    },
+    {
+      "epoch": 0.6778286897290567,
+      "grad_norm": 0.16214215755462646,
+      "learning_rate": 4.322178548937002e-06,
+      "loss": 0.8968,
+      "step": 93640
+    },
+    {
+      "epoch": 0.6779010763896429,
+      "grad_norm": 0.21187572181224823,
+      "learning_rate": 4.322106162276416e-06,
+      "loss": 0.8969,
+      "step": 93650
+    },
+    {
+      "epoch": 0.6779734630502291,
+      "grad_norm": 0.15346659719944,
+      "learning_rate": 4.32203377561583e-06,
+      "loss": 0.8985,
+      "step": 93660
+    },
+    {
+      "epoch": 0.6780458497108153,
+      "grad_norm": 0.14978167414665222,
+      "learning_rate": 4.3219613889552435e-06,
+      "loss": 0.8998,
+      "step": 93670
+    },
+    {
+      "epoch": 0.6781182363714015,
+      "grad_norm": 0.16475534439086914,
+      "learning_rate": 4.321889002294657e-06,
+      "loss": 0.8935,
+      "step": 93680
+    },
+    {
+      "epoch": 0.6781906230319876,
+      "grad_norm": 0.1509210169315338,
+      "learning_rate": 4.321816615634072e-06,
+      "loss": 0.9166,
+      "step": 93690
+    },
+    {
+      "epoch": 0.6782630096925738,
+      "grad_norm": 0.15706129372119904,
+      "learning_rate": 4.321744228973485e-06,
+      "loss": 0.884,
+      "step": 93700
+    },
+    {
+      "epoch": 0.6783353963531601,
+      "grad_norm": 0.19146853685379028,
+      "learning_rate": 4.321671842312899e-06,
+      "loss": 0.8988,
+      "step": 93710
+    },
+    {
+      "epoch": 0.6784077830137463,
+      "grad_norm": 0.14886470139026642,
+      "learning_rate": 4.3215994556523124e-06,
+      "loss": 0.8954,
+      "step": 93720
+    },
+    {
+      "epoch": 0.6784801696743324,
+      "grad_norm": 0.15223486721515656,
+      "learning_rate": 4.321527068991727e-06,
+      "loss": 0.8994,
+      "step": 93730
+    },
+    {
+      "epoch": 0.6785525563349186,
+      "grad_norm": 0.1696760505437851,
+      "learning_rate": 4.3214546823311405e-06,
+      "loss": 0.9005,
+      "step": 93740
+    },
+    {
+      "epoch": 0.6786249429955048,
+      "grad_norm": 0.17853619158267975,
+      "learning_rate": 4.321382295670554e-06,
+      "loss": 0.883,
+      "step": 93750
+    },
+    {
+      "epoch": 0.678697329656091,
+      "grad_norm": 0.1491808295249939,
+      "learning_rate": 4.321309909009968e-06,
+      "loss": 0.8852,
+      "step": 93760
+    },
+    {
+      "epoch": 0.6787697163166772,
+      "grad_norm": 0.19821669161319733,
+      "learning_rate": 4.321237522349382e-06,
+      "loss": 0.8976,
+      "step": 93770
+    },
+    {
+      "epoch": 0.6788421029772633,
+      "grad_norm": 0.1570088416337967,
+      "learning_rate": 4.321165135688796e-06,
+      "loss": 0.8913,
+      "step": 93780
+    },
+    {
+      "epoch": 0.6789144896378495,
+      "grad_norm": 0.15688063204288483,
+      "learning_rate": 4.3210927490282094e-06,
+      "loss": 0.9008,
+      "step": 93790
+    },
+    {
+      "epoch": 0.6789868762984357,
+      "grad_norm": 0.19363504648208618,
+      "learning_rate": 4.321020362367623e-06,
+      "loss": 0.8946,
+      "step": 93800
+    },
+    {
+      "epoch": 0.6790592629590219,
+      "grad_norm": 0.16203875839710236,
+      "learning_rate": 4.3209479757070375e-06,
+      "loss": 0.8867,
+      "step": 93810
+    },
+    {
+      "epoch": 0.6791316496196081,
+      "grad_norm": 0.15617011487483978,
+      "learning_rate": 4.320875589046451e-06,
+      "loss": 0.9006,
+      "step": 93820
+    },
+    {
+      "epoch": 0.6792040362801943,
+      "grad_norm": 0.1507921814918518,
+      "learning_rate": 4.320803202385865e-06,
+      "loss": 0.9161,
+      "step": 93830
+    },
+    {
+      "epoch": 0.6792764229407805,
+      "grad_norm": 0.16400021314620972,
+      "learning_rate": 4.320730815725278e-06,
+      "loss": 0.8899,
+      "step": 93840
+    },
+    {
+      "epoch": 0.6793488096013667,
+      "grad_norm": 0.15807297825813293,
+      "learning_rate": 4.320658429064693e-06,
+      "loss": 0.9041,
+      "step": 93850
+    },
+    {
+      "epoch": 0.6794211962619529,
+      "grad_norm": 0.15214060246944427,
+      "learning_rate": 4.3205860424041065e-06,
+      "loss": 0.8978,
+      "step": 93860
+    },
+    {
+      "epoch": 0.679493582922539,
+      "grad_norm": 0.16088810563087463,
+      "learning_rate": 4.32051365574352e-06,
+      "loss": 0.8974,
+      "step": 93870
+    },
+    {
+      "epoch": 0.6795659695831252,
+      "grad_norm": 0.1684071570634842,
+      "learning_rate": 4.320441269082934e-06,
+      "loss": 0.8949,
+      "step": 93880
+    },
+    {
+      "epoch": 0.6796383562437114,
+      "grad_norm": 0.15487946569919586,
+      "learning_rate": 4.320368882422348e-06,
+      "loss": 0.8905,
+      "step": 93890
+    },
+    {
+      "epoch": 0.6797107429042976,
+      "grad_norm": 0.1507687270641327,
+      "learning_rate": 4.320296495761762e-06,
+      "loss": 0.8959,
+      "step": 93900
+    },
+    {
+      "epoch": 0.6797831295648837,
+      "grad_norm": 0.16048942506313324,
+      "learning_rate": 4.320224109101175e-06,
+      "loss": 0.8903,
+      "step": 93910
+    },
+    {
+      "epoch": 0.67985551622547,
+      "grad_norm": 0.1652323603630066,
+      "learning_rate": 4.320151722440589e-06,
+      "loss": 0.8897,
+      "step": 93920
+    },
+    {
+      "epoch": 0.6799279028860562,
+      "grad_norm": 0.16126669943332672,
+      "learning_rate": 4.3200793357800035e-06,
+      "loss": 0.8921,
+      "step": 93930
+    },
+    {
+      "epoch": 0.6800002895466424,
+      "grad_norm": 0.15348832309246063,
+      "learning_rate": 4.320006949119417e-06,
+      "loss": 0.8927,
+      "step": 93940
+    },
+    {
+      "epoch": 0.6800726762072286,
+      "grad_norm": 0.2017420381307602,
+      "learning_rate": 4.319934562458831e-06,
+      "loss": 0.8945,
+      "step": 93950
+    },
+    {
+      "epoch": 0.6801450628678147,
+      "grad_norm": 0.17592447996139526,
+      "learning_rate": 4.319862175798244e-06,
+      "loss": 0.8905,
+      "step": 93960
+    },
+    {
+      "epoch": 0.6802174495284009,
+      "grad_norm": 0.15097962319850922,
+      "learning_rate": 4.319789789137658e-06,
+      "loss": 0.8964,
+      "step": 93970
+    },
+    {
+      "epoch": 0.6802898361889871,
+      "grad_norm": 0.16152705252170563,
+      "learning_rate": 4.3197174024770715e-06,
+      "loss": 0.8875,
+      "step": 93980
+    },
+    {
+      "epoch": 0.6803622228495733,
+      "grad_norm": 0.17407093942165375,
+      "learning_rate": 4.319645015816485e-06,
+      "loss": 0.9065,
+      "step": 93990
+    },
+    {
+      "epoch": 0.6804346095101594,
+      "grad_norm": 0.16338643431663513,
+      "learning_rate": 4.3195726291559e-06,
+      "loss": 0.907,
+      "step": 94000
+    },
+    {
+      "epoch": 0.6805069961707456,
+      "grad_norm": 0.15193772315979004,
+      "learning_rate": 4.319500242495313e-06,
+      "loss": 0.8908,
+      "step": 94010
+    },
+    {
+      "epoch": 0.6805793828313318,
+      "grad_norm": 0.15826401114463806,
+      "learning_rate": 4.319427855834727e-06,
+      "loss": 0.889,
+      "step": 94020
+    },
+    {
+      "epoch": 0.6806517694919181,
+      "grad_norm": 0.15474501252174377,
+      "learning_rate": 4.3193554691741405e-06,
+      "loss": 0.8907,
+      "step": 94030
+    },
+    {
+      "epoch": 0.6807241561525043,
+      "grad_norm": 0.1633521169424057,
+      "learning_rate": 4.319283082513555e-06,
+      "loss": 0.8997,
+      "step": 94040
+    },
+    {
+      "epoch": 0.6807965428130904,
+      "grad_norm": 0.19621436297893524,
+      "learning_rate": 4.3192106958529686e-06,
+      "loss": 0.9016,
+      "step": 94050
+    },
+    {
+      "epoch": 0.6808689294736766,
+      "grad_norm": 0.15351617336273193,
+      "learning_rate": 4.319138309192382e-06,
+      "loss": 0.9004,
+      "step": 94060
+    },
+    {
+      "epoch": 0.6809413161342628,
+      "grad_norm": 0.1630709171295166,
+      "learning_rate": 4.319065922531796e-06,
+      "loss": 0.8924,
+      "step": 94070
+    },
+    {
+      "epoch": 0.681013702794849,
+      "grad_norm": 0.17487677931785583,
+      "learning_rate": 4.31899353587121e-06,
+      "loss": 0.8983,
+      "step": 94080
+    },
+    {
+      "epoch": 0.6810860894554351,
+      "grad_norm": 0.15708361566066742,
+      "learning_rate": 4.318921149210624e-06,
+      "loss": 0.8904,
+      "step": 94090
+    },
+    {
+      "epoch": 0.6811584761160213,
+      "grad_norm": 0.15445072948932648,
+      "learning_rate": 4.3188487625500375e-06,
+      "loss": 0.9044,
+      "step": 94100
+    },
+    {
+      "epoch": 0.6812308627766075,
+      "grad_norm": 0.16823984682559967,
+      "learning_rate": 4.318776375889451e-06,
+      "loss": 0.8893,
+      "step": 94110
+    },
+    {
+      "epoch": 0.6813032494371937,
+      "grad_norm": 0.1907905787229538,
+      "learning_rate": 4.318703989228865e-06,
+      "loss": 0.906,
+      "step": 94120
+    },
+    {
+      "epoch": 0.6813756360977798,
+      "grad_norm": 0.1714591085910797,
+      "learning_rate": 4.318631602568279e-06,
+      "loss": 0.8967,
+      "step": 94130
+    },
+    {
+      "epoch": 0.6814480227583661,
+      "grad_norm": 0.15924982726573944,
+      "learning_rate": 4.318559215907693e-06,
+      "loss": 0.8899,
+      "step": 94140
+    },
+    {
+      "epoch": 0.6815204094189523,
+      "grad_norm": 0.18446829915046692,
+      "learning_rate": 4.318486829247106e-06,
+      "loss": 0.9054,
+      "step": 94150
+    },
+    {
+      "epoch": 0.6815927960795385,
+      "grad_norm": 0.1584208905696869,
+      "learning_rate": 4.31841444258652e-06,
+      "loss": 0.9035,
+      "step": 94160
+    },
+    {
+      "epoch": 0.6816651827401247,
+      "grad_norm": 0.15730392932891846,
+      "learning_rate": 4.3183420559259345e-06,
+      "loss": 0.9004,
+      "step": 94170
+    },
+    {
+      "epoch": 0.6817375694007108,
+      "grad_norm": 0.15665248036384583,
+      "learning_rate": 4.318269669265348e-06,
+      "loss": 0.8768,
+      "step": 94180
+    },
+    {
+      "epoch": 0.681809956061297,
+      "grad_norm": 0.15125800669193268,
+      "learning_rate": 4.318197282604762e-06,
+      "loss": 0.9052,
+      "step": 94190
+    },
+    {
+      "epoch": 0.6818823427218832,
+      "grad_norm": 0.15670722723007202,
+      "learning_rate": 4.318124895944175e-06,
+      "loss": 0.8914,
+      "step": 94200
+    },
+    {
+      "epoch": 0.6819547293824694,
+      "grad_norm": 0.16642498970031738,
+      "learning_rate": 4.31805250928359e-06,
+      "loss": 0.8978,
+      "step": 94210
+    },
+    {
+      "epoch": 0.6820271160430555,
+      "grad_norm": 0.17175577580928802,
+      "learning_rate": 4.317980122623003e-06,
+      "loss": 0.9012,
+      "step": 94220
+    },
+    {
+      "epoch": 0.6820995027036417,
+      "grad_norm": 0.16106824576854706,
+      "learning_rate": 4.317907735962417e-06,
+      "loss": 0.8985,
+      "step": 94230
+    },
+    {
+      "epoch": 0.682171889364228,
+      "grad_norm": 0.15547014772891998,
+      "learning_rate": 4.317835349301831e-06,
+      "loss": 0.9019,
+      "step": 94240
+    },
+    {
+      "epoch": 0.6822442760248142,
+      "grad_norm": 0.16611367464065552,
+      "learning_rate": 4.317762962641245e-06,
+      "loss": 0.8933,
+      "step": 94250
+    },
+    {
+      "epoch": 0.6823166626854004,
+      "grad_norm": 0.33474913239479065,
+      "learning_rate": 4.317690575980659e-06,
+      "loss": 0.8996,
+      "step": 94260
+    },
+    {
+      "epoch": 0.6823890493459865,
+      "grad_norm": 0.17985767126083374,
+      "learning_rate": 4.317618189320072e-06,
+      "loss": 0.9111,
+      "step": 94270
+    },
+    {
+      "epoch": 0.6824614360065727,
+      "grad_norm": 0.14204993844032288,
+      "learning_rate": 4.317545802659486e-06,
+      "loss": 0.9077,
+      "step": 94280
+    },
+    {
+      "epoch": 0.6825338226671589,
+      "grad_norm": 0.14709369838237762,
+      "learning_rate": 4.3174734159989e-06,
+      "loss": 0.8779,
+      "step": 94290
+    },
+    {
+      "epoch": 0.6826062093277451,
+      "grad_norm": 0.16410696506500244,
+      "learning_rate": 4.317401029338314e-06,
+      "loss": 0.8993,
+      "step": 94300
+    },
+    {
+      "epoch": 0.6826785959883312,
+      "grad_norm": 0.15575379133224487,
+      "learning_rate": 4.317328642677728e-06,
+      "loss": 0.8887,
+      "step": 94310
+    },
+    {
+      "epoch": 0.6827509826489174,
+      "grad_norm": 0.16469940543174744,
+      "learning_rate": 4.317256256017141e-06,
+      "loss": 0.8954,
+      "step": 94320
+    },
+    {
+      "epoch": 0.6828233693095036,
+      "grad_norm": 0.15340863168239594,
+      "learning_rate": 4.317183869356556e-06,
+      "loss": 0.8975,
+      "step": 94330
+    },
+    {
+      "epoch": 0.6828957559700898,
+      "grad_norm": 0.16109062731266022,
+      "learning_rate": 4.317111482695969e-06,
+      "loss": 0.8972,
+      "step": 94340
+    },
+    {
+      "epoch": 0.6829681426306761,
+      "grad_norm": 0.16391950845718384,
+      "learning_rate": 4.317039096035383e-06,
+      "loss": 0.9028,
+      "step": 94350
+    },
+    {
+      "epoch": 0.6830405292912622,
+      "grad_norm": 0.15392933785915375,
+      "learning_rate": 4.316966709374797e-06,
+      "loss": 0.8991,
+      "step": 94360
+    },
+    {
+      "epoch": 0.6831129159518484,
+      "grad_norm": 0.16828075051307678,
+      "learning_rate": 4.316894322714211e-06,
+      "loss": 0.9025,
+      "step": 94370
+    },
+    {
+      "epoch": 0.6831853026124346,
+      "grad_norm": 0.147588312625885,
+      "learning_rate": 4.316821936053625e-06,
+      "loss": 0.9008,
+      "step": 94380
+    },
+    {
+      "epoch": 0.6832576892730208,
+      "grad_norm": 0.14783324301242828,
+      "learning_rate": 4.316749549393038e-06,
+      "loss": 0.9054,
+      "step": 94390
+    },
+    {
+      "epoch": 0.683330075933607,
+      "grad_norm": 0.1539517641067505,
+      "learning_rate": 4.316677162732452e-06,
+      "loss": 0.8891,
+      "step": 94400
+    },
+    {
+      "epoch": 0.6834024625941931,
+      "grad_norm": 0.15323641896247864,
+      "learning_rate": 4.316604776071866e-06,
+      "loss": 0.9112,
+      "step": 94410
+    },
+    {
+      "epoch": 0.6834748492547793,
+      "grad_norm": 0.1703089028596878,
+      "learning_rate": 4.31653238941128e-06,
+      "loss": 0.8885,
+      "step": 94420
+    },
+    {
+      "epoch": 0.6835472359153655,
+      "grad_norm": 0.15150368213653564,
+      "learning_rate": 4.316460002750694e-06,
+      "loss": 0.8762,
+      "step": 94430
+    },
+    {
+      "epoch": 0.6836196225759517,
+      "grad_norm": 0.15605489909648895,
+      "learning_rate": 4.316387616090107e-06,
+      "loss": 0.9036,
+      "step": 94440
+    },
+    {
+      "epoch": 0.683692009236538,
+      "grad_norm": 0.14873306453227997,
+      "learning_rate": 4.316315229429522e-06,
+      "loss": 0.8936,
+      "step": 94450
+    },
+    {
+      "epoch": 0.6837643958971241,
+      "grad_norm": 0.15489499270915985,
+      "learning_rate": 4.316242842768935e-06,
+      "loss": 0.8927,
+      "step": 94460
+    },
+    {
+      "epoch": 0.6838367825577103,
+      "grad_norm": 0.14702056348323822,
+      "learning_rate": 4.316170456108349e-06,
+      "loss": 0.8919,
+      "step": 94470
+    },
+    {
+      "epoch": 0.6839091692182965,
+      "grad_norm": 0.17012947797775269,
+      "learning_rate": 4.3160980694477625e-06,
+      "loss": 0.9027,
+      "step": 94480
+    },
+    {
+      "epoch": 0.6839815558788827,
+      "grad_norm": 0.23859137296676636,
+      "learning_rate": 4.316025682787177e-06,
+      "loss": 0.8989,
+      "step": 94490
+    },
+    {
+      "epoch": 0.6840539425394688,
+      "grad_norm": 0.14270347356796265,
+      "learning_rate": 4.315953296126591e-06,
+      "loss": 0.8966,
+      "step": 94500
+    },
+    {
+      "epoch": 0.684126329200055,
+      "grad_norm": 0.15070217847824097,
+      "learning_rate": 4.315880909466003e-06,
+      "loss": 0.8967,
+      "step": 94510
+    },
+    {
+      "epoch": 0.6841987158606412,
+      "grad_norm": 0.14996741712093353,
+      "learning_rate": 4.315808522805418e-06,
+      "loss": 0.8973,
+      "step": 94520
+    },
+    {
+      "epoch": 0.6842711025212274,
+      "grad_norm": 0.1512274593114853,
+      "learning_rate": 4.3157361361448314e-06,
+      "loss": 0.8937,
+      "step": 94530
+    },
+    {
+      "epoch": 0.6843434891818135,
+      "grad_norm": 0.1495598703622818,
+      "learning_rate": 4.315663749484245e-06,
+      "loss": 0.894,
+      "step": 94540
+    },
+    {
+      "epoch": 0.6844158758423997,
+      "grad_norm": 0.1526460349559784,
+      "learning_rate": 4.315591362823659e-06,
+      "loss": 0.9058,
+      "step": 94550
+    },
+    {
+      "epoch": 0.684488262502986,
+      "grad_norm": 0.22177492082118988,
+      "learning_rate": 4.315518976163073e-06,
+      "loss": 0.8909,
+      "step": 94560
+    },
+    {
+      "epoch": 0.6845606491635722,
+      "grad_norm": 0.1475263386964798,
+      "learning_rate": 4.315446589502487e-06,
+      "loss": 0.8894,
+      "step": 94570
+    },
+    {
+      "epoch": 0.6846330358241584,
+      "grad_norm": 0.15898674726486206,
+      "learning_rate": 4.3153742028419e-06,
+      "loss": 0.9013,
+      "step": 94580
+    },
+    {
+      "epoch": 0.6847054224847445,
+      "grad_norm": 0.160345658659935,
+      "learning_rate": 4.315301816181314e-06,
+      "loss": 0.8894,
+      "step": 94590
+    },
+    {
+      "epoch": 0.6847778091453307,
+      "grad_norm": 0.16944102942943573,
+      "learning_rate": 4.3152294295207285e-06,
+      "loss": 0.9001,
+      "step": 94600
+    },
+    {
+      "epoch": 0.6848501958059169,
+      "grad_norm": 0.17007231712341309,
+      "learning_rate": 4.315157042860142e-06,
+      "loss": 0.9066,
+      "step": 94610
+    },
+    {
+      "epoch": 0.6849225824665031,
+      "grad_norm": 0.14328494668006897,
+      "learning_rate": 4.315084656199556e-06,
+      "loss": 0.8942,
+      "step": 94620
+    },
+    {
+      "epoch": 0.6849949691270892,
+      "grad_norm": 0.17920292913913727,
+      "learning_rate": 4.315012269538969e-06,
+      "loss": 0.9016,
+      "step": 94630
+    },
+    {
+      "epoch": 0.6850673557876754,
+      "grad_norm": 0.15425662696361542,
+      "learning_rate": 4.314939882878384e-06,
+      "loss": 0.885,
+      "step": 94640
+    },
+    {
+      "epoch": 0.6851397424482616,
+      "grad_norm": 0.15390869975090027,
+      "learning_rate": 4.314867496217797e-06,
+      "loss": 0.8941,
+      "step": 94650
+    },
+    {
+      "epoch": 0.6852121291088478,
+      "grad_norm": 0.15865157544612885,
+      "learning_rate": 4.314795109557211e-06,
+      "loss": 0.9016,
+      "step": 94660
+    },
+    {
+      "epoch": 0.685284515769434,
+      "grad_norm": 0.14358533918857574,
+      "learning_rate": 4.314722722896625e-06,
+      "loss": 0.8833,
+      "step": 94670
+    },
+    {
+      "epoch": 0.6853569024300202,
+      "grad_norm": 0.15622635185718536,
+      "learning_rate": 4.314650336236039e-06,
+      "loss": 0.8932,
+      "step": 94680
+    },
+    {
+      "epoch": 0.6854292890906064,
+      "grad_norm": 0.1485530436038971,
+      "learning_rate": 4.314577949575453e-06,
+      "loss": 0.9085,
+      "step": 94690
+    },
+    {
+      "epoch": 0.6855016757511926,
+      "grad_norm": 0.1571887731552124,
+      "learning_rate": 4.314505562914866e-06,
+      "loss": 0.9022,
+      "step": 94700
+    },
+    {
+      "epoch": 0.6855740624117788,
+      "grad_norm": 0.15338723361492157,
+      "learning_rate": 4.31443317625428e-06,
+      "loss": 0.8992,
+      "step": 94710
+    },
+    {
+      "epoch": 0.6856464490723649,
+      "grad_norm": 0.14642809331417084,
+      "learning_rate": 4.314360789593694e-06,
+      "loss": 0.8907,
+      "step": 94720
+    },
+    {
+      "epoch": 0.6857188357329511,
+      "grad_norm": 0.1597226858139038,
+      "learning_rate": 4.314288402933108e-06,
+      "loss": 0.9018,
+      "step": 94730
+    },
+    {
+      "epoch": 0.6857912223935373,
+      "grad_norm": 0.16934312880039215,
+      "learning_rate": 4.314216016272522e-06,
+      "loss": 0.8962,
+      "step": 94740
+    },
+    {
+      "epoch": 0.6858636090541235,
+      "grad_norm": 0.16490796208381653,
+      "learning_rate": 4.314143629611935e-06,
+      "loss": 0.8977,
+      "step": 94750
+    },
+    {
+      "epoch": 0.6859359957147096,
+      "grad_norm": 0.1624918133020401,
+      "learning_rate": 4.314071242951349e-06,
+      "loss": 0.8968,
+      "step": 94760
+    },
+    {
+      "epoch": 0.6860083823752959,
+      "grad_norm": 0.16102252900600433,
+      "learning_rate": 4.313998856290763e-06,
+      "loss": 0.8991,
+      "step": 94770
+    },
+    {
+      "epoch": 0.6860807690358821,
+      "grad_norm": 0.15310189127922058,
+      "learning_rate": 4.313926469630177e-06,
+      "loss": 0.894,
+      "step": 94780
+    },
+    {
+      "epoch": 0.6861531556964683,
+      "grad_norm": 0.15352487564086914,
+      "learning_rate": 4.3138540829695906e-06,
+      "loss": 0.8779,
+      "step": 94790
+    },
+    {
+      "epoch": 0.6862255423570545,
+      "grad_norm": 0.15823419392108917,
+      "learning_rate": 4.313781696309004e-06,
+      "loss": 0.906,
+      "step": 94800
+    },
+    {
+      "epoch": 0.6862979290176406,
+      "grad_norm": 0.14976033568382263,
+      "learning_rate": 4.313709309648419e-06,
+      "loss": 0.8872,
+      "step": 94810
+    },
+    {
+      "epoch": 0.6863703156782268,
+      "grad_norm": 0.1584375500679016,
+      "learning_rate": 4.313636922987832e-06,
+      "loss": 0.8847,
+      "step": 94820
+    },
+    {
+      "epoch": 0.686442702338813,
+      "grad_norm": 0.16619914770126343,
+      "learning_rate": 4.313564536327246e-06,
+      "loss": 0.8942,
+      "step": 94830
+    },
+    {
+      "epoch": 0.6865150889993992,
+      "grad_norm": 0.16585594415664673,
+      "learning_rate": 4.3134921496666595e-06,
+      "loss": 0.8958,
+      "step": 94840
+    },
+    {
+      "epoch": 0.6865874756599853,
+      "grad_norm": 0.1556151807308197,
+      "learning_rate": 4.313419763006074e-06,
+      "loss": 0.9076,
+      "step": 94850
+    },
+    {
+      "epoch": 0.6866598623205715,
+      "grad_norm": 0.16103480756282806,
+      "learning_rate": 4.3133473763454876e-06,
+      "loss": 0.8927,
+      "step": 94860
+    },
+    {
+      "epoch": 0.6867322489811577,
+      "grad_norm": 0.15770234167575836,
+      "learning_rate": 4.313274989684901e-06,
+      "loss": 0.8901,
+      "step": 94870
+    },
+    {
+      "epoch": 0.686804635641744,
+      "grad_norm": 0.15498584508895874,
+      "learning_rate": 4.313202603024315e-06,
+      "loss": 0.9026,
+      "step": 94880
+    },
+    {
+      "epoch": 0.6868770223023302,
+      "grad_norm": 0.15212376415729523,
+      "learning_rate": 4.313130216363729e-06,
+      "loss": 0.8837,
+      "step": 94890
+    },
+    {
+      "epoch": 0.6869494089629163,
+      "grad_norm": 0.16598445177078247,
+      "learning_rate": 4.313057829703143e-06,
+      "loss": 0.8998,
+      "step": 94900
+    },
+    {
+      "epoch": 0.6870217956235025,
+      "grad_norm": 0.15342400968074799,
+      "learning_rate": 4.3129854430425565e-06,
+      "loss": 0.8858,
+      "step": 94910
+    },
+    {
+      "epoch": 0.6870941822840887,
+      "grad_norm": 0.1750396341085434,
+      "learning_rate": 4.31291305638197e-06,
+      "loss": 0.8823,
+      "step": 94920
+    },
+    {
+      "epoch": 0.6871665689446749,
+      "grad_norm": 0.14746147394180298,
+      "learning_rate": 4.3128406697213846e-06,
+      "loss": 0.8892,
+      "step": 94930
+    },
+    {
+      "epoch": 0.687238955605261,
+      "grad_norm": 0.16607217490673065,
+      "learning_rate": 4.312768283060798e-06,
+      "loss": 0.9003,
+      "step": 94940
+    },
+    {
+      "epoch": 0.6873113422658472,
+      "grad_norm": 0.15274251997470856,
+      "learning_rate": 4.312695896400212e-06,
+      "loss": 0.8967,
+      "step": 94950
+    },
+    {
+      "epoch": 0.6873837289264334,
+      "grad_norm": 0.15561649203300476,
+      "learning_rate": 4.312623509739625e-06,
+      "loss": 0.8926,
+      "step": 94960
+    },
+    {
+      "epoch": 0.6874561155870196,
+      "grad_norm": 0.1617891639471054,
+      "learning_rate": 4.31255112307904e-06,
+      "loss": 0.8919,
+      "step": 94970
+    },
+    {
+      "epoch": 0.6875285022476059,
+      "grad_norm": 0.15291917324066162,
+      "learning_rate": 4.3124787364184535e-06,
+      "loss": 0.8961,
+      "step": 94980
+    },
+    {
+      "epoch": 0.687600888908192,
+      "grad_norm": 0.15395525097846985,
+      "learning_rate": 4.312406349757867e-06,
+      "loss": 0.9067,
+      "step": 94990
+    },
+    {
+      "epoch": 0.6876732755687782,
+      "grad_norm": 0.15367929637432098,
+      "learning_rate": 4.312333963097281e-06,
+      "loss": 0.8899,
+      "step": 95000
+    },
+    {
+      "epoch": 0.6877456622293644,
+      "grad_norm": 0.16362635791301727,
+      "learning_rate": 4.312261576436695e-06,
+      "loss": 0.8971,
+      "step": 95010
+    },
+    {
+      "epoch": 0.6878180488899506,
+      "grad_norm": 0.1670594960451126,
+      "learning_rate": 4.312189189776109e-06,
+      "loss": 0.8869,
+      "step": 95020
+    },
+    {
+      "epoch": 0.6878904355505367,
+      "grad_norm": 0.161848783493042,
+      "learning_rate": 4.312116803115522e-06,
+      "loss": 0.8845,
+      "step": 95030
+    },
+    {
+      "epoch": 0.6879628222111229,
+      "grad_norm": 0.16515681147575378,
+      "learning_rate": 4.312044416454936e-06,
+      "loss": 0.8909,
+      "step": 95040
+    },
+    {
+      "epoch": 0.6880352088717091,
+      "grad_norm": 0.14597424864768982,
+      "learning_rate": 4.31197202979435e-06,
+      "loss": 0.896,
+      "step": 95050
+    },
+    {
+      "epoch": 0.6881075955322953,
+      "grad_norm": 0.1595359891653061,
+      "learning_rate": 4.311899643133763e-06,
+      "loss": 0.8876,
+      "step": 95060
+    },
+    {
+      "epoch": 0.6881799821928815,
+      "grad_norm": 0.1476903259754181,
+      "learning_rate": 4.311827256473177e-06,
+      "loss": 0.8913,
+      "step": 95070
+    },
+    {
+      "epoch": 0.6882523688534676,
+      "grad_norm": 0.18288491666316986,
+      "learning_rate": 4.311754869812591e-06,
+      "loss": 0.9013,
+      "step": 95080
+    },
+    {
+      "epoch": 0.6883247555140539,
+      "grad_norm": 0.16732849180698395,
+      "learning_rate": 4.311682483152005e-06,
+      "loss": 0.8962,
+      "step": 95090
+    },
+    {
+      "epoch": 0.6883971421746401,
+      "grad_norm": 0.15418116748332977,
+      "learning_rate": 4.311610096491419e-06,
+      "loss": 0.8913,
+      "step": 95100
+    },
+    {
+      "epoch": 0.6884695288352263,
+      "grad_norm": 0.15506073832511902,
+      "learning_rate": 4.311537709830832e-06,
+      "loss": 0.9081,
+      "step": 95110
+    },
+    {
+      "epoch": 0.6885419154958125,
+      "grad_norm": 0.15247192978858948,
+      "learning_rate": 4.311465323170247e-06,
+      "loss": 0.9026,
+      "step": 95120
+    },
+    {
+      "epoch": 0.6886143021563986,
+      "grad_norm": 0.15282821655273438,
+      "learning_rate": 4.31139293650966e-06,
+      "loss": 0.8913,
+      "step": 95130
+    },
+    {
+      "epoch": 0.6886866888169848,
+      "grad_norm": 0.14951321482658386,
+      "learning_rate": 4.311320549849074e-06,
+      "loss": 0.8937,
+      "step": 95140
+    },
+    {
+      "epoch": 0.688759075477571,
+      "grad_norm": 0.15877053141593933,
+      "learning_rate": 4.3112481631884875e-06,
+      "loss": 0.889,
+      "step": 95150
+    },
+    {
+      "epoch": 0.6888314621381572,
+      "grad_norm": 0.15322931110858917,
+      "learning_rate": 4.311175776527902e-06,
+      "loss": 0.8997,
+      "step": 95160
+    },
+    {
+      "epoch": 0.6889038487987433,
+      "grad_norm": 0.16098260879516602,
+      "learning_rate": 4.311103389867316e-06,
+      "loss": 0.8823,
+      "step": 95170
+    },
+    {
+      "epoch": 0.6889762354593295,
+      "grad_norm": 0.16160458326339722,
+      "learning_rate": 4.311031003206729e-06,
+      "loss": 0.8972,
+      "step": 95180
+    },
+    {
+      "epoch": 0.6890486221199157,
+      "grad_norm": 0.16651996970176697,
+      "learning_rate": 4.310958616546143e-06,
+      "loss": 0.8879,
+      "step": 95190
+    },
+    {
+      "epoch": 0.689121008780502,
+      "grad_norm": 0.1751406192779541,
+      "learning_rate": 4.310886229885557e-06,
+      "loss": 0.9041,
+      "step": 95200
+    },
+    {
+      "epoch": 0.6891933954410882,
+      "grad_norm": 0.1623631864786148,
+      "learning_rate": 4.310813843224971e-06,
+      "loss": 0.8999,
+      "step": 95210
+    },
+    {
+      "epoch": 0.6892657821016743,
+      "grad_norm": 0.1833324134349823,
+      "learning_rate": 4.3107414565643845e-06,
+      "loss": 0.8994,
+      "step": 95220
+    },
+    {
+      "epoch": 0.6893381687622605,
+      "grad_norm": 0.1521463841199875,
+      "learning_rate": 4.310669069903798e-06,
+      "loss": 0.9037,
+      "step": 95230
+    },
+    {
+      "epoch": 0.6894105554228467,
+      "grad_norm": 0.16113896667957306,
+      "learning_rate": 4.310596683243213e-06,
+      "loss": 0.8856,
+      "step": 95240
+    },
+    {
+      "epoch": 0.6894829420834329,
+      "grad_norm": 0.1714179962873459,
+      "learning_rate": 4.310524296582626e-06,
+      "loss": 0.8825,
+      "step": 95250
+    },
+    {
+      "epoch": 0.689555328744019,
+      "grad_norm": 0.16432242095470428,
+      "learning_rate": 4.31045190992204e-06,
+      "loss": 0.8871,
+      "step": 95260
+    },
+    {
+      "epoch": 0.6896277154046052,
+      "grad_norm": 0.16391827166080475,
+      "learning_rate": 4.3103795232614534e-06,
+      "loss": 0.8978,
+      "step": 95270
+    },
+    {
+      "epoch": 0.6897001020651914,
+      "grad_norm": 0.1498943716287613,
+      "learning_rate": 4.310307136600868e-06,
+      "loss": 0.8863,
+      "step": 95280
+    },
+    {
+      "epoch": 0.6897724887257776,
+      "grad_norm": 0.1618964523077011,
+      "learning_rate": 4.3102347499402815e-06,
+      "loss": 0.8955,
+      "step": 95290
+    },
+    {
+      "epoch": 0.6898448753863639,
+      "grad_norm": 0.1474188268184662,
+      "learning_rate": 4.310162363279695e-06,
+      "loss": 0.8935,
+      "step": 95300
+    },
+    {
+      "epoch": 0.68991726204695,
+      "grad_norm": 0.1580621600151062,
+      "learning_rate": 4.310089976619109e-06,
+      "loss": 0.8884,
+      "step": 95310
+    },
+    {
+      "epoch": 0.6899896487075362,
+      "grad_norm": 0.16119031608104706,
+      "learning_rate": 4.310017589958523e-06,
+      "loss": 0.8994,
+      "step": 95320
+    },
+    {
+      "epoch": 0.6900620353681224,
+      "grad_norm": 0.19750839471817017,
+      "learning_rate": 4.309945203297937e-06,
+      "loss": 0.8847,
+      "step": 95330
+    },
+    {
+      "epoch": 0.6901344220287086,
+      "grad_norm": 0.15879863500595093,
+      "learning_rate": 4.3098728166373505e-06,
+      "loss": 0.8885,
+      "step": 95340
+    },
+    {
+      "epoch": 0.6902068086892947,
+      "grad_norm": 0.14561788737773895,
+      "learning_rate": 4.309800429976764e-06,
+      "loss": 0.8881,
+      "step": 95350
+    },
+    {
+      "epoch": 0.6902791953498809,
+      "grad_norm": 0.16508562862873077,
+      "learning_rate": 4.3097280433161785e-06,
+      "loss": 0.8895,
+      "step": 95360
+    },
+    {
+      "epoch": 0.6903515820104671,
+      "grad_norm": 0.14818058907985687,
+      "learning_rate": 4.309655656655592e-06,
+      "loss": 0.8812,
+      "step": 95370
+    },
+    {
+      "epoch": 0.6904239686710533,
+      "grad_norm": 0.15891733765602112,
+      "learning_rate": 4.309583269995006e-06,
+      "loss": 0.8962,
+      "step": 95380
+    },
+    {
+      "epoch": 0.6904963553316394,
+      "grad_norm": 0.15448154509067535,
+      "learning_rate": 4.309510883334419e-06,
+      "loss": 0.8874,
+      "step": 95390
+    },
+    {
+      "epoch": 0.6905687419922256,
+      "grad_norm": 0.15475550293922424,
+      "learning_rate": 4.309438496673833e-06,
+      "loss": 0.8954,
+      "step": 95400
+    },
+    {
+      "epoch": 0.6906411286528119,
+      "grad_norm": 0.1788392812013626,
+      "learning_rate": 4.3093661100132475e-06,
+      "loss": 0.8905,
+      "step": 95410
+    },
+    {
+      "epoch": 0.6907135153133981,
+      "grad_norm": 0.17869041860103607,
+      "learning_rate": 4.309293723352661e-06,
+      "loss": 0.8919,
+      "step": 95420
+    },
+    {
+      "epoch": 0.6907859019739843,
+      "grad_norm": 0.1565256416797638,
+      "learning_rate": 4.309221336692075e-06,
+      "loss": 0.8814,
+      "step": 95430
+    },
+    {
+      "epoch": 0.6908582886345704,
+      "grad_norm": 0.30670905113220215,
+      "learning_rate": 4.309148950031488e-06,
+      "loss": 0.8798,
+      "step": 95440
+    },
+    {
+      "epoch": 0.6909306752951566,
+      "grad_norm": 0.1644957959651947,
+      "learning_rate": 4.309076563370903e-06,
+      "loss": 0.8952,
+      "step": 95450
+    },
+    {
+      "epoch": 0.6910030619557428,
+      "grad_norm": 0.17856526374816895,
+      "learning_rate": 4.309004176710316e-06,
+      "loss": 0.9032,
+      "step": 95460
+    },
+    {
+      "epoch": 0.691075448616329,
+      "grad_norm": 0.1568627655506134,
+      "learning_rate": 4.30893179004973e-06,
+      "loss": 0.893,
+      "step": 95470
+    },
+    {
+      "epoch": 0.6911478352769151,
+      "grad_norm": 0.2152179330587387,
+      "learning_rate": 4.308859403389144e-06,
+      "loss": 0.8926,
+      "step": 95480
+    },
+    {
+      "epoch": 0.6912202219375013,
+      "grad_norm": 0.16184626519680023,
+      "learning_rate": 4.308787016728558e-06,
+      "loss": 0.8783,
+      "step": 95490
+    },
+    {
+      "epoch": 0.6912926085980875,
+      "grad_norm": 0.1587613821029663,
+      "learning_rate": 4.308714630067972e-06,
+      "loss": 0.8855,
+      "step": 95500
+    },
+    {
+      "epoch": 0.6913649952586738,
+      "grad_norm": 0.18116620182991028,
+      "learning_rate": 4.308642243407385e-06,
+      "loss": 0.8977,
+      "step": 95510
+    },
+    {
+      "epoch": 0.69143738191926,
+      "grad_norm": 0.14871811866760254,
+      "learning_rate": 4.308569856746799e-06,
+      "loss": 0.8962,
+      "step": 95520
+    },
+    {
+      "epoch": 0.6915097685798461,
+      "grad_norm": 0.14631856977939606,
+      "learning_rate": 4.308497470086213e-06,
+      "loss": 0.8987,
+      "step": 95530
+    },
+    {
+      "epoch": 0.6915821552404323,
+      "grad_norm": 0.16045540571212769,
+      "learning_rate": 4.308425083425627e-06,
+      "loss": 0.8954,
+      "step": 95540
+    },
+    {
+      "epoch": 0.6916545419010185,
+      "grad_norm": 0.22909776866436005,
+      "learning_rate": 4.308352696765041e-06,
+      "loss": 0.9038,
+      "step": 95550
+    },
+    {
+      "epoch": 0.6917269285616047,
+      "grad_norm": 0.1496800035238266,
+      "learning_rate": 4.308280310104454e-06,
+      "loss": 0.8881,
+      "step": 95560
+    },
+    {
+      "epoch": 0.6917993152221908,
+      "grad_norm": 0.16365250945091248,
+      "learning_rate": 4.308207923443868e-06,
+      "loss": 0.8915,
+      "step": 95570
+    },
+    {
+      "epoch": 0.691871701882777,
+      "grad_norm": 0.16174180805683136,
+      "learning_rate": 4.3081355367832815e-06,
+      "loss": 0.9061,
+      "step": 95580
+    },
+    {
+      "epoch": 0.6919440885433632,
+      "grad_norm": 0.1649826467037201,
+      "learning_rate": 4.308063150122695e-06,
+      "loss": 0.8954,
+      "step": 95590
+    },
+    {
+      "epoch": 0.6920164752039494,
+      "grad_norm": 0.17469272017478943,
+      "learning_rate": 4.3079907634621096e-06,
+      "loss": 0.887,
+      "step": 95600
+    },
+    {
+      "epoch": 0.6920888618645356,
+      "grad_norm": 0.1564358025789261,
+      "learning_rate": 4.307918376801523e-06,
+      "loss": 0.8889,
+      "step": 95610
+    },
+    {
+      "epoch": 0.6921612485251218,
+      "grad_norm": 0.1583586186170578,
+      "learning_rate": 4.307845990140937e-06,
+      "loss": 0.8859,
+      "step": 95620
+    },
+    {
+      "epoch": 0.692233635185708,
+      "grad_norm": 0.1620096117258072,
+      "learning_rate": 4.30777360348035e-06,
+      "loss": 0.8862,
+      "step": 95630
+    },
+    {
+      "epoch": 0.6923060218462942,
+      "grad_norm": 0.16032646596431732,
+      "learning_rate": 4.307701216819765e-06,
+      "loss": 0.8894,
+      "step": 95640
+    },
+    {
+      "epoch": 0.6923784085068804,
+      "grad_norm": 0.21608605980873108,
+      "learning_rate": 4.3076288301591785e-06,
+      "loss": 0.8925,
+      "step": 95650
+    },
+    {
+      "epoch": 0.6924507951674665,
+      "grad_norm": 0.14863435924053192,
+      "learning_rate": 4.307556443498592e-06,
+      "loss": 0.8989,
+      "step": 95660
+    },
+    {
+      "epoch": 0.6925231818280527,
+      "grad_norm": 0.16338156163692474,
+      "learning_rate": 4.307484056838006e-06,
+      "loss": 0.8888,
+      "step": 95670
+    },
+    {
+      "epoch": 0.6925955684886389,
+      "grad_norm": 0.15900051593780518,
+      "learning_rate": 4.30741167017742e-06,
+      "loss": 0.8827,
+      "step": 95680
+    },
+    {
+      "epoch": 0.6926679551492251,
+      "grad_norm": 0.16256722807884216,
+      "learning_rate": 4.307339283516834e-06,
+      "loss": 0.8878,
+      "step": 95690
+    },
+    {
+      "epoch": 0.6927403418098113,
+      "grad_norm": 0.15759159624576569,
+      "learning_rate": 4.307266896856247e-06,
+      "loss": 0.8929,
+      "step": 95700
+    },
+    {
+      "epoch": 0.6928127284703974,
+      "grad_norm": 0.15648800134658813,
+      "learning_rate": 4.307194510195661e-06,
+      "loss": 0.8818,
+      "step": 95710
+    },
+    {
+      "epoch": 0.6928851151309836,
+      "grad_norm": 0.1834617257118225,
+      "learning_rate": 4.3071221235350755e-06,
+      "loss": 0.8918,
+      "step": 95720
+    },
+    {
+      "epoch": 0.6929575017915699,
+      "grad_norm": 0.162217915058136,
+      "learning_rate": 4.307049736874489e-06,
+      "loss": 0.9029,
+      "step": 95730
+    },
+    {
+      "epoch": 0.6930298884521561,
+      "grad_norm": 0.15800444781780243,
+      "learning_rate": 4.306977350213903e-06,
+      "loss": 0.8949,
+      "step": 95740
+    },
+    {
+      "epoch": 0.6931022751127422,
+      "grad_norm": 0.16938433051109314,
+      "learning_rate": 4.306904963553316e-06,
+      "loss": 0.8983,
+      "step": 95750
+    },
+    {
+      "epoch": 0.6931746617733284,
+      "grad_norm": 0.16872501373291016,
+      "learning_rate": 4.306832576892731e-06,
+      "loss": 0.8944,
+      "step": 95760
+    },
+    {
+      "epoch": 0.6932470484339146,
+      "grad_norm": 0.1726616472005844,
+      "learning_rate": 4.306760190232144e-06,
+      "loss": 0.8868,
+      "step": 95770
+    },
+    {
+      "epoch": 0.6933194350945008,
+      "grad_norm": 0.1653449535369873,
+      "learning_rate": 4.306687803571558e-06,
+      "loss": 0.8915,
+      "step": 95780
+    },
+    {
+      "epoch": 0.693391821755087,
+      "grad_norm": 0.15586352348327637,
+      "learning_rate": 4.306615416910972e-06,
+      "loss": 0.8864,
+      "step": 95790
+    },
+    {
+      "epoch": 0.6934642084156731,
+      "grad_norm": 0.16180157661437988,
+      "learning_rate": 4.306543030250386e-06,
+      "loss": 0.8912,
+      "step": 95800
+    },
+    {
+      "epoch": 0.6935365950762593,
+      "grad_norm": 0.16136401891708374,
+      "learning_rate": 4.3064706435898e-06,
+      "loss": 0.8899,
+      "step": 95810
+    },
+    {
+      "epoch": 0.6936089817368455,
+      "grad_norm": 0.1651676446199417,
+      "learning_rate": 4.306398256929213e-06,
+      "loss": 0.8796,
+      "step": 95820
+    },
+    {
+      "epoch": 0.6936813683974318,
+      "grad_norm": 0.16563965380191803,
+      "learning_rate": 4.306325870268627e-06,
+      "loss": 0.8822,
+      "step": 95830
+    },
+    {
+      "epoch": 0.693753755058018,
+      "grad_norm": 0.15649251639842987,
+      "learning_rate": 4.3062534836080414e-06,
+      "loss": 0.8911,
+      "step": 95840
+    },
+    {
+      "epoch": 0.6938261417186041,
+      "grad_norm": 0.1557883769273758,
+      "learning_rate": 4.306181096947455e-06,
+      "loss": 0.9081,
+      "step": 95850
+    },
+    {
+      "epoch": 0.6938985283791903,
+      "grad_norm": 0.18621514737606049,
+      "learning_rate": 4.306108710286869e-06,
+      "loss": 0.8983,
+      "step": 95860
+    },
+    {
+      "epoch": 0.6939709150397765,
+      "grad_norm": 0.16845759749412537,
+      "learning_rate": 4.306036323626282e-06,
+      "loss": 0.8874,
+      "step": 95870
+    },
+    {
+      "epoch": 0.6940433017003627,
+      "grad_norm": 0.16313624382019043,
+      "learning_rate": 4.305963936965697e-06,
+      "loss": 0.8855,
+      "step": 95880
+    },
+    {
+      "epoch": 0.6941156883609488,
+      "grad_norm": 0.160096675157547,
+      "learning_rate": 4.30589155030511e-06,
+      "loss": 0.8966,
+      "step": 95890
+    },
+    {
+      "epoch": 0.694188075021535,
+      "grad_norm": 0.15571869909763336,
+      "learning_rate": 4.305819163644524e-06,
+      "loss": 0.8865,
+      "step": 95900
+    },
+    {
+      "epoch": 0.6942604616821212,
+      "grad_norm": 0.14915607869625092,
+      "learning_rate": 4.305746776983938e-06,
+      "loss": 0.8954,
+      "step": 95910
+    },
+    {
+      "epoch": 0.6943328483427074,
+      "grad_norm": 0.15486833453178406,
+      "learning_rate": 4.305674390323352e-06,
+      "loss": 0.893,
+      "step": 95920
+    },
+    {
+      "epoch": 0.6944052350032935,
+      "grad_norm": 0.16171999275684357,
+      "learning_rate": 4.305602003662766e-06,
+      "loss": 0.8914,
+      "step": 95930
+    },
+    {
+      "epoch": 0.6944776216638798,
+      "grad_norm": 0.16727794706821442,
+      "learning_rate": 4.305529617002179e-06,
+      "loss": 0.8979,
+      "step": 95940
+    },
+    {
+      "epoch": 0.694550008324466,
+      "grad_norm": 0.19044362008571625,
+      "learning_rate": 4.305457230341593e-06,
+      "loss": 0.8751,
+      "step": 95950
+    },
+    {
+      "epoch": 0.6946223949850522,
+      "grad_norm": 0.16111010313034058,
+      "learning_rate": 4.305384843681007e-06,
+      "loss": 0.9069,
+      "step": 95960
+    },
+    {
+      "epoch": 0.6946947816456384,
+      "grad_norm": 0.20112472772598267,
+      "learning_rate": 4.305312457020421e-06,
+      "loss": 0.8972,
+      "step": 95970
+    },
+    {
+      "epoch": 0.6947671683062245,
+      "grad_norm": 0.19165970385074615,
+      "learning_rate": 4.305240070359835e-06,
+      "loss": 0.9029,
+      "step": 95980
+    },
+    {
+      "epoch": 0.6948395549668107,
+      "grad_norm": 0.1580514758825302,
+      "learning_rate": 4.305167683699248e-06,
+      "loss": 0.8961,
+      "step": 95990
+    },
+    {
+      "epoch": 0.6949119416273969,
+      "grad_norm": 0.1498570591211319,
+      "learning_rate": 4.305095297038662e-06,
+      "loss": 0.8919,
+      "step": 96000
+    },
+    {
+      "epoch": 0.6949843282879831,
+      "grad_norm": 0.15419946610927582,
+      "learning_rate": 4.305022910378076e-06,
+      "loss": 0.888,
+      "step": 96010
+    },
+    {
+      "epoch": 0.6950567149485692,
+      "grad_norm": 0.17555716633796692,
+      "learning_rate": 4.30495052371749e-06,
+      "loss": 0.8925,
+      "step": 96020
+    },
+    {
+      "epoch": 0.6951291016091554,
+      "grad_norm": 0.15583516657352448,
+      "learning_rate": 4.3048781370569035e-06,
+      "loss": 0.8886,
+      "step": 96030
+    },
+    {
+      "epoch": 0.6952014882697416,
+      "grad_norm": 0.15992936491966248,
+      "learning_rate": 4.304805750396317e-06,
+      "loss": 0.9023,
+      "step": 96040
+    },
+    {
+      "epoch": 0.6952738749303279,
+      "grad_norm": 0.16176897287368774,
+      "learning_rate": 4.304733363735732e-06,
+      "loss": 0.8834,
+      "step": 96050
+    },
+    {
+      "epoch": 0.6953462615909141,
+      "grad_norm": 0.14862500131130219,
+      "learning_rate": 4.304660977075145e-06,
+      "loss": 0.9043,
+      "step": 96060
+    },
+    {
+      "epoch": 0.6954186482515002,
+      "grad_norm": 0.16168230772018433,
+      "learning_rate": 4.304588590414559e-06,
+      "loss": 0.9008,
+      "step": 96070
+    },
+    {
+      "epoch": 0.6954910349120864,
+      "grad_norm": 0.15463055670261383,
+      "learning_rate": 4.3045162037539725e-06,
+      "loss": 0.8975,
+      "step": 96080
+    },
+    {
+      "epoch": 0.6955634215726726,
+      "grad_norm": 0.15740707516670227,
+      "learning_rate": 4.304443817093387e-06,
+      "loss": 0.8999,
+      "step": 96090
+    },
+    {
+      "epoch": 0.6956358082332588,
+      "grad_norm": 0.1548217087984085,
+      "learning_rate": 4.3043714304328e-06,
+      "loss": 0.8997,
+      "step": 96100
+    },
+    {
+      "epoch": 0.695708194893845,
+      "grad_norm": 0.17392714321613312,
+      "learning_rate": 4.304299043772214e-06,
+      "loss": 0.8889,
+      "step": 96110
+    },
+    {
+      "epoch": 0.6957805815544311,
+      "grad_norm": 0.16982245445251465,
+      "learning_rate": 4.304226657111628e-06,
+      "loss": 0.8908,
+      "step": 96120
+    },
+    {
+      "epoch": 0.6958529682150173,
+      "grad_norm": 0.33574557304382324,
+      "learning_rate": 4.304154270451041e-06,
+      "loss": 0.8943,
+      "step": 96130
+    },
+    {
+      "epoch": 0.6959253548756035,
+      "grad_norm": 0.1604243814945221,
+      "learning_rate": 4.304081883790455e-06,
+      "loss": 0.891,
+      "step": 96140
+    },
+    {
+      "epoch": 0.6959977415361898,
+      "grad_norm": 0.1695748269557953,
+      "learning_rate": 4.3040094971298695e-06,
+      "loss": 0.9052,
+      "step": 96150
+    },
+    {
+      "epoch": 0.6960701281967759,
+      "grad_norm": 0.15508553385734558,
+      "learning_rate": 4.303937110469283e-06,
+      "loss": 0.8917,
+      "step": 96160
+    },
+    {
+      "epoch": 0.6961425148573621,
+      "grad_norm": 0.16269119083881378,
+      "learning_rate": 4.303864723808697e-06,
+      "loss": 0.891,
+      "step": 96170
+    },
+    {
+      "epoch": 0.6962149015179483,
+      "grad_norm": 0.15743182599544525,
+      "learning_rate": 4.30379233714811e-06,
+      "loss": 0.8944,
+      "step": 96180
+    },
+    {
+      "epoch": 0.6962872881785345,
+      "grad_norm": 0.15545742213726044,
+      "learning_rate": 4.303719950487524e-06,
+      "loss": 0.8862,
+      "step": 96190
+    },
+    {
+      "epoch": 0.6963596748391206,
+      "grad_norm": 0.2056836634874344,
+      "learning_rate": 4.303647563826938e-06,
+      "loss": 0.8911,
+      "step": 96200
+    },
+    {
+      "epoch": 0.6964320614997068,
+      "grad_norm": 0.15505336225032806,
+      "learning_rate": 4.303575177166352e-06,
+      "loss": 0.8959,
+      "step": 96210
+    },
+    {
+      "epoch": 0.696504448160293,
+      "grad_norm": 0.155269056558609,
+      "learning_rate": 4.303502790505766e-06,
+      "loss": 0.8989,
+      "step": 96220
+    },
+    {
+      "epoch": 0.6965768348208792,
+      "grad_norm": 0.16013725101947784,
+      "learning_rate": 4.303430403845179e-06,
+      "loss": 0.8941,
+      "step": 96230
+    },
+    {
+      "epoch": 0.6966492214814654,
+      "grad_norm": 0.15824046730995178,
+      "learning_rate": 4.303358017184594e-06,
+      "loss": 0.8998,
+      "step": 96240
+    },
+    {
+      "epoch": 0.6967216081420515,
+      "grad_norm": 0.17412789165973663,
+      "learning_rate": 4.303285630524007e-06,
+      "loss": 0.9062,
+      "step": 96250
+    },
+    {
+      "epoch": 0.6967939948026378,
+      "grad_norm": 0.22454731166362762,
+      "learning_rate": 4.303213243863421e-06,
+      "loss": 0.8884,
+      "step": 96260
+    },
+    {
+      "epoch": 0.696866381463224,
+      "grad_norm": 0.16182060539722443,
+      "learning_rate": 4.3031408572028345e-06,
+      "loss": 0.8888,
+      "step": 96270
+    },
+    {
+      "epoch": 0.6969387681238102,
+      "grad_norm": 0.17886275053024292,
+      "learning_rate": 4.303068470542249e-06,
+      "loss": 0.8823,
+      "step": 96280
+    },
+    {
+      "epoch": 0.6970111547843963,
+      "grad_norm": 0.1985713690519333,
+      "learning_rate": 4.302996083881663e-06,
+      "loss": 0.8882,
+      "step": 96290
+    },
+    {
+      "epoch": 0.6970835414449825,
+      "grad_norm": 0.1513681411743164,
+      "learning_rate": 4.302923697221076e-06,
+      "loss": 0.9012,
+      "step": 96300
+    },
+    {
+      "epoch": 0.6971559281055687,
+      "grad_norm": 0.15492072701454163,
+      "learning_rate": 4.30285131056049e-06,
+      "loss": 0.899,
+      "step": 96310
+    },
+    {
+      "epoch": 0.6972283147661549,
+      "grad_norm": 0.15350612998008728,
+      "learning_rate": 4.302778923899904e-06,
+      "loss": 0.9022,
+      "step": 96320
+    },
+    {
+      "epoch": 0.697300701426741,
+      "grad_norm": 0.15936022996902466,
+      "learning_rate": 4.302706537239318e-06,
+      "loss": 0.8985,
+      "step": 96330
+    },
+    {
+      "epoch": 0.6973730880873272,
+      "grad_norm": 0.15187810361385345,
+      "learning_rate": 4.3026341505787316e-06,
+      "loss": 0.8867,
+      "step": 96340
+    },
+    {
+      "epoch": 0.6974454747479134,
+      "grad_norm": 0.16198959946632385,
+      "learning_rate": 4.302561763918145e-06,
+      "loss": 0.903,
+      "step": 96350
+    },
+    {
+      "epoch": 0.6975178614084997,
+      "grad_norm": 0.1528894305229187,
+      "learning_rate": 4.30248937725756e-06,
+      "loss": 0.8793,
+      "step": 96360
+    },
+    {
+      "epoch": 0.6975902480690859,
+      "grad_norm": 0.17313164472579956,
+      "learning_rate": 4.302416990596973e-06,
+      "loss": 0.8992,
+      "step": 96370
+    },
+    {
+      "epoch": 0.697662634729672,
+      "grad_norm": 0.15843333303928375,
+      "learning_rate": 4.302344603936387e-06,
+      "loss": 0.8963,
+      "step": 96380
+    },
+    {
+      "epoch": 0.6977350213902582,
+      "grad_norm": 0.1584440916776657,
+      "learning_rate": 4.3022722172758005e-06,
+      "loss": 0.9057,
+      "step": 96390
+    },
+    {
+      "epoch": 0.6978074080508444,
+      "grad_norm": 0.14555677771568298,
+      "learning_rate": 4.302199830615215e-06,
+      "loss": 0.8841,
+      "step": 96400
+    },
+    {
+      "epoch": 0.6978797947114306,
+      "grad_norm": 0.22453060746192932,
+      "learning_rate": 4.3021274439546286e-06,
+      "loss": 0.8912,
+      "step": 96410
+    },
+    {
+      "epoch": 0.6979521813720168,
+      "grad_norm": 0.16137515008449554,
+      "learning_rate": 4.302055057294042e-06,
+      "loss": 0.8757,
+      "step": 96420
+    },
+    {
+      "epoch": 0.6980245680326029,
+      "grad_norm": 0.2278577834367752,
+      "learning_rate": 4.301982670633456e-06,
+      "loss": 0.8918,
+      "step": 96430
+    },
+    {
+      "epoch": 0.6980969546931891,
+      "grad_norm": 0.1604139506816864,
+      "learning_rate": 4.30191028397287e-06,
+      "loss": 0.8988,
+      "step": 96440
+    },
+    {
+      "epoch": 0.6981693413537753,
+      "grad_norm": 0.1709796041250229,
+      "learning_rate": 4.301837897312284e-06,
+      "loss": 0.8999,
+      "step": 96450
+    },
+    {
+      "epoch": 0.6982417280143615,
+      "grad_norm": 0.16104479134082794,
+      "learning_rate": 4.3017655106516975e-06,
+      "loss": 0.8961,
+      "step": 96460
+    },
+    {
+      "epoch": 0.6983141146749478,
+      "grad_norm": 0.1653842329978943,
+      "learning_rate": 4.301693123991111e-06,
+      "loss": 0.8805,
+      "step": 96470
+    },
+    {
+      "epoch": 0.6983865013355339,
+      "grad_norm": 0.16090551018714905,
+      "learning_rate": 4.3016207373305256e-06,
+      "loss": 0.9097,
+      "step": 96480
+    },
+    {
+      "epoch": 0.6984588879961201,
+      "grad_norm": 0.3694705367088318,
+      "learning_rate": 4.301548350669939e-06,
+      "loss": 0.8856,
+      "step": 96490
+    },
+    {
+      "epoch": 0.6985312746567063,
+      "grad_norm": 0.1503128856420517,
+      "learning_rate": 4.301475964009353e-06,
+      "loss": 0.8856,
+      "step": 96500
+    },
+    {
+      "epoch": 0.6986036613172925,
+      "grad_norm": 0.16498462855815887,
+      "learning_rate": 4.301403577348766e-06,
+      "loss": 0.8975,
+      "step": 96510
+    },
+    {
+      "epoch": 0.6986760479778786,
+      "grad_norm": 0.1551860123872757,
+      "learning_rate": 4.301331190688181e-06,
+      "loss": 0.895,
+      "step": 96520
+    },
+    {
+      "epoch": 0.6987484346384648,
+      "grad_norm": 0.16694217920303345,
+      "learning_rate": 4.3012588040275945e-06,
+      "loss": 0.8932,
+      "step": 96530
+    },
+    {
+      "epoch": 0.698820821299051,
+      "grad_norm": 0.9841502904891968,
+      "learning_rate": 4.301186417367008e-06,
+      "loss": 0.9151,
+      "step": 96540
+    },
+    {
+      "epoch": 0.6988932079596372,
+      "grad_norm": 0.3107115626335144,
+      "learning_rate": 4.301114030706422e-06,
+      "loss": 0.8826,
+      "step": 96550
+    },
+    {
+      "epoch": 0.6989655946202233,
+      "grad_norm": 0.3002294898033142,
+      "learning_rate": 4.301041644045836e-06,
+      "loss": 0.8991,
+      "step": 96560
+    },
+    {
+      "epoch": 0.6990379812808095,
+      "grad_norm": 0.1559462547302246,
+      "learning_rate": 4.30096925738525e-06,
+      "loss": 0.8887,
+      "step": 96570
+    },
+    {
+      "epoch": 0.6991103679413958,
+      "grad_norm": 0.1544226109981537,
+      "learning_rate": 4.3008968707246634e-06,
+      "loss": 0.8921,
+      "step": 96580
+    },
+    {
+      "epoch": 0.699182754601982,
+      "grad_norm": 0.18448671698570251,
+      "learning_rate": 4.300824484064077e-06,
+      "loss": 0.8846,
+      "step": 96590
+    },
+    {
+      "epoch": 0.6992551412625682,
+      "grad_norm": 0.1595505028963089,
+      "learning_rate": 4.3007520974034915e-06,
+      "loss": 0.8977,
+      "step": 96600
+    },
+    {
+      "epoch": 0.6993275279231543,
+      "grad_norm": 0.15914291143417358,
+      "learning_rate": 4.300679710742905e-06,
+      "loss": 0.9054,
+      "step": 96610
+    },
+    {
+      "epoch": 0.6993999145837405,
+      "grad_norm": 0.14967529475688934,
+      "learning_rate": 4.300607324082319e-06,
+      "loss": 0.8893,
+      "step": 96620
+    },
+    {
+      "epoch": 0.6994723012443267,
+      "grad_norm": 0.1701008379459381,
+      "learning_rate": 4.300534937421732e-06,
+      "loss": 0.8854,
+      "step": 96630
+    },
+    {
+      "epoch": 0.6995446879049129,
+      "grad_norm": 0.15182742476463318,
+      "learning_rate": 4.300462550761146e-06,
+      "loss": 0.8878,
+      "step": 96640
+    },
+    {
+      "epoch": 0.699617074565499,
+      "grad_norm": 0.16047948598861694,
+      "learning_rate": 4.30039016410056e-06,
+      "loss": 0.8819,
+      "step": 96650
+    },
+    {
+      "epoch": 0.6996894612260852,
+      "grad_norm": 0.15404923260211945,
+      "learning_rate": 4.300317777439973e-06,
+      "loss": 0.8876,
+      "step": 96660
+    },
+    {
+      "epoch": 0.6997618478866714,
+      "grad_norm": 0.14895722270011902,
+      "learning_rate": 4.300245390779388e-06,
+      "loss": 0.8889,
+      "step": 96670
+    },
+    {
+      "epoch": 0.6998342345472577,
+      "grad_norm": 0.1480094939470291,
+      "learning_rate": 4.300173004118801e-06,
+      "loss": 0.8798,
+      "step": 96680
+    },
+    {
+      "epoch": 0.6999066212078439,
+      "grad_norm": 0.15281951427459717,
+      "learning_rate": 4.300100617458215e-06,
+      "loss": 0.8828,
+      "step": 96690
+    },
+    {
+      "epoch": 0.69997900786843,
+      "grad_norm": 0.15964744985103607,
+      "learning_rate": 4.3000282307976285e-06,
+      "loss": 0.8885,
+      "step": 96700
+    },
+    {
+      "epoch": 0.7000513945290162,
+      "grad_norm": 0.15603557229042053,
+      "learning_rate": 4.299955844137043e-06,
+      "loss": 0.8863,
+      "step": 96710
+    },
+    {
+      "epoch": 0.7001237811896024,
+      "grad_norm": 0.1674613654613495,
+      "learning_rate": 4.299883457476457e-06,
+      "loss": 0.8982,
+      "step": 96720
+    },
+    {
+      "epoch": 0.7001961678501886,
+      "grad_norm": 0.15703196823596954,
+      "learning_rate": 4.29981107081587e-06,
+      "loss": 0.8825,
+      "step": 96730
+    },
+    {
+      "epoch": 0.7002685545107747,
+      "grad_norm": 0.17176085710525513,
+      "learning_rate": 4.299738684155284e-06,
+      "loss": 0.8873,
+      "step": 96740
+    },
+    {
+      "epoch": 0.7003409411713609,
+      "grad_norm": 0.16086554527282715,
+      "learning_rate": 4.299666297494698e-06,
+      "loss": 0.8854,
+      "step": 96750
+    },
+    {
+      "epoch": 0.7004133278319471,
+      "grad_norm": 0.16075126826763153,
+      "learning_rate": 4.299593910834112e-06,
+      "loss": 0.9035,
+      "step": 96760
+    },
+    {
+      "epoch": 0.7004857144925333,
+      "grad_norm": 0.14747561514377594,
+      "learning_rate": 4.2995215241735255e-06,
+      "loss": 0.8975,
+      "step": 96770
+    },
+    {
+      "epoch": 0.7005581011531195,
+      "grad_norm": 0.16974276304244995,
+      "learning_rate": 4.299449137512939e-06,
+      "loss": 0.8975,
+      "step": 96780
+    },
+    {
+      "epoch": 0.7006304878137057,
+      "grad_norm": 0.16168342530727386,
+      "learning_rate": 4.299376750852353e-06,
+      "loss": 0.8864,
+      "step": 96790
+    },
+    {
+      "epoch": 0.7007028744742919,
+      "grad_norm": 0.1794668585062027,
+      "learning_rate": 4.299304364191767e-06,
+      "loss": 0.8973,
+      "step": 96800
+    },
+    {
+      "epoch": 0.7007752611348781,
+      "grad_norm": 0.16029122471809387,
+      "learning_rate": 4.299231977531181e-06,
+      "loss": 0.8895,
+      "step": 96810
+    },
+    {
+      "epoch": 0.7008476477954643,
+      "grad_norm": 0.18477699160575867,
+      "learning_rate": 4.2991595908705945e-06,
+      "loss": 0.8888,
+      "step": 96820
+    },
+    {
+      "epoch": 0.7009200344560504,
+      "grad_norm": 0.14988726377487183,
+      "learning_rate": 4.299087204210008e-06,
+      "loss": 0.8774,
+      "step": 96830
+    },
+    {
+      "epoch": 0.7009924211166366,
+      "grad_norm": 0.1730933040380478,
+      "learning_rate": 4.2990148175494225e-06,
+      "loss": 0.9006,
+      "step": 96840
+    },
+    {
+      "epoch": 0.7010648077772228,
+      "grad_norm": 0.1729445457458496,
+      "learning_rate": 4.298942430888836e-06,
+      "loss": 0.892,
+      "step": 96850
+    },
+    {
+      "epoch": 0.701137194437809,
+      "grad_norm": 0.15627533197402954,
+      "learning_rate": 4.29887004422825e-06,
+      "loss": 0.8931,
+      "step": 96860
+    },
+    {
+      "epoch": 0.7012095810983952,
+      "grad_norm": 0.16592872142791748,
+      "learning_rate": 4.298797657567663e-06,
+      "loss": 0.9052,
+      "step": 96870
+    },
+    {
+      "epoch": 0.7012819677589813,
+      "grad_norm": 0.15809662640094757,
+      "learning_rate": 4.298725270907078e-06,
+      "loss": 0.8885,
+      "step": 96880
+    },
+    {
+      "epoch": 0.7013543544195676,
+      "grad_norm": 0.1509890854358673,
+      "learning_rate": 4.2986528842464915e-06,
+      "loss": 0.8884,
+      "step": 96890
+    },
+    {
+      "epoch": 0.7014267410801538,
+      "grad_norm": 0.14822304248809814,
+      "learning_rate": 4.298580497585905e-06,
+      "loss": 0.8768,
+      "step": 96900
+    },
+    {
+      "epoch": 0.70149912774074,
+      "grad_norm": 0.15169335901737213,
+      "learning_rate": 4.298508110925319e-06,
+      "loss": 0.8876,
+      "step": 96910
+    },
+    {
+      "epoch": 0.7015715144013261,
+      "grad_norm": 0.16953204572200775,
+      "learning_rate": 4.298435724264733e-06,
+      "loss": 0.9045,
+      "step": 96920
+    },
+    {
+      "epoch": 0.7016439010619123,
+      "grad_norm": 0.16186213493347168,
+      "learning_rate": 4.298363337604147e-06,
+      "loss": 0.8963,
+      "step": 96930
+    },
+    {
+      "epoch": 0.7017162877224985,
+      "grad_norm": 0.16287651658058167,
+      "learning_rate": 4.29829095094356e-06,
+      "loss": 0.9003,
+      "step": 96940
+    },
+    {
+      "epoch": 0.7017886743830847,
+      "grad_norm": 0.17633283138275146,
+      "learning_rate": 4.298218564282974e-06,
+      "loss": 0.8685,
+      "step": 96950
+    },
+    {
+      "epoch": 0.7018610610436709,
+      "grad_norm": 0.22455504536628723,
+      "learning_rate": 4.2981461776223885e-06,
+      "loss": 0.9052,
+      "step": 96960
+    },
+    {
+      "epoch": 0.701933447704257,
+      "grad_norm": 0.14891460537910461,
+      "learning_rate": 4.298073790961802e-06,
+      "loss": 0.8836,
+      "step": 96970
+    },
+    {
+      "epoch": 0.7020058343648432,
+      "grad_norm": 0.1637570559978485,
+      "learning_rate": 4.298001404301216e-06,
+      "loss": 0.8994,
+      "step": 96980
+    },
+    {
+      "epoch": 0.7020782210254294,
+      "grad_norm": 0.1491258293390274,
+      "learning_rate": 4.297929017640629e-06,
+      "loss": 0.8972,
+      "step": 96990
+    },
+    {
+      "epoch": 0.7021506076860157,
+      "grad_norm": 0.15172043442726135,
+      "learning_rate": 4.297856630980044e-06,
+      "loss": 0.8946,
+      "step": 97000
+    },
+    {
+      "epoch": 0.7022229943466018,
+      "grad_norm": 0.1903785616159439,
+      "learning_rate": 4.297784244319457e-06,
+      "loss": 0.885,
+      "step": 97010
+    },
+    {
+      "epoch": 0.702295381007188,
+      "grad_norm": 0.1513489931821823,
+      "learning_rate": 4.297711857658871e-06,
+      "loss": 0.8886,
+      "step": 97020
+    },
+    {
+      "epoch": 0.7023677676677742,
+      "grad_norm": 0.14946502447128296,
+      "learning_rate": 4.297639470998285e-06,
+      "loss": 0.8916,
+      "step": 97030
+    },
+    {
+      "epoch": 0.7024401543283604,
+      "grad_norm": 0.16224405169487,
+      "learning_rate": 4.297567084337699e-06,
+      "loss": 0.8992,
+      "step": 97040
+    },
+    {
+      "epoch": 0.7025125409889466,
+      "grad_norm": 0.17303119599819183,
+      "learning_rate": 4.297494697677113e-06,
+      "loss": 0.9058,
+      "step": 97050
+    },
+    {
+      "epoch": 0.7025849276495327,
+      "grad_norm": 0.16080045700073242,
+      "learning_rate": 4.297422311016526e-06,
+      "loss": 0.8884,
+      "step": 97060
+    },
+    {
+      "epoch": 0.7026573143101189,
+      "grad_norm": 0.1557897925376892,
+      "learning_rate": 4.29734992435594e-06,
+      "loss": 0.8916,
+      "step": 97070
+    },
+    {
+      "epoch": 0.7027297009707051,
+      "grad_norm": 0.14745230972766876,
+      "learning_rate": 4.297277537695354e-06,
+      "loss": 0.888,
+      "step": 97080
+    },
+    {
+      "epoch": 0.7028020876312913,
+      "grad_norm": 0.16206741333007812,
+      "learning_rate": 4.297205151034768e-06,
+      "loss": 0.889,
+      "step": 97090
+    },
+    {
+      "epoch": 0.7028744742918774,
+      "grad_norm": 0.23554551601409912,
+      "learning_rate": 4.297132764374182e-06,
+      "loss": 0.8866,
+      "step": 97100
+    },
+    {
+      "epoch": 0.7029468609524637,
+      "grad_norm": 0.15564242005348206,
+      "learning_rate": 4.297060377713595e-06,
+      "loss": 0.8908,
+      "step": 97110
+    },
+    {
+      "epoch": 0.7030192476130499,
+      "grad_norm": 0.15459449589252472,
+      "learning_rate": 4.29698799105301e-06,
+      "loss": 0.8898,
+      "step": 97120
+    },
+    {
+      "epoch": 0.7030916342736361,
+      "grad_norm": 0.1636834740638733,
+      "learning_rate": 4.296915604392423e-06,
+      "loss": 0.8917,
+      "step": 97130
+    },
+    {
+      "epoch": 0.7031640209342223,
+      "grad_norm": 0.1554258018732071,
+      "learning_rate": 4.296843217731837e-06,
+      "loss": 0.887,
+      "step": 97140
+    },
+    {
+      "epoch": 0.7032364075948084,
+      "grad_norm": 0.1681572049856186,
+      "learning_rate": 4.2967708310712506e-06,
+      "loss": 0.8967,
+      "step": 97150
+    },
+    {
+      "epoch": 0.7033087942553946,
+      "grad_norm": 0.15658222138881683,
+      "learning_rate": 4.296698444410664e-06,
+      "loss": 0.8905,
+      "step": 97160
+    },
+    {
+      "epoch": 0.7033811809159808,
+      "grad_norm": 0.159566730260849,
+      "learning_rate": 4.296626057750078e-06,
+      "loss": 0.8944,
+      "step": 97170
+    },
+    {
+      "epoch": 0.703453567576567,
+      "grad_norm": 0.16223838925361633,
+      "learning_rate": 4.296553671089491e-06,
+      "loss": 0.8846,
+      "step": 97180
+    },
+    {
+      "epoch": 0.7035259542371531,
+      "grad_norm": 0.16831691563129425,
+      "learning_rate": 4.296481284428906e-06,
+      "loss": 0.8883,
+      "step": 97190
+    },
+    {
+      "epoch": 0.7035983408977393,
+      "grad_norm": 0.17067785561084747,
+      "learning_rate": 4.2964088977683195e-06,
+      "loss": 0.8852,
+      "step": 97200
+    },
+    {
+      "epoch": 0.7036707275583256,
+      "grad_norm": 0.1619025319814682,
+      "learning_rate": 4.296336511107733e-06,
+      "loss": 0.8959,
+      "step": 97210
+    },
+    {
+      "epoch": 0.7037431142189118,
+      "grad_norm": 0.17006492614746094,
+      "learning_rate": 4.296264124447147e-06,
+      "loss": 0.8926,
+      "step": 97220
+    },
+    {
+      "epoch": 0.703815500879498,
+      "grad_norm": 0.16066719591617584,
+      "learning_rate": 4.296191737786561e-06,
+      "loss": 0.8944,
+      "step": 97230
+    },
+    {
+      "epoch": 0.7038878875400841,
+      "grad_norm": 0.16666296124458313,
+      "learning_rate": 4.296119351125975e-06,
+      "loss": 0.8865,
+      "step": 97240
+    },
+    {
+      "epoch": 0.7039602742006703,
+      "grad_norm": 0.152059406042099,
+      "learning_rate": 4.296046964465388e-06,
+      "loss": 0.8949,
+      "step": 97250
+    },
+    {
+      "epoch": 0.7040326608612565,
+      "grad_norm": 0.15612152218818665,
+      "learning_rate": 4.295974577804802e-06,
+      "loss": 0.9075,
+      "step": 97260
+    },
+    {
+      "epoch": 0.7041050475218427,
+      "grad_norm": 0.15522941946983337,
+      "learning_rate": 4.2959021911442165e-06,
+      "loss": 0.8996,
+      "step": 97270
+    },
+    {
+      "epoch": 0.7041774341824288,
+      "grad_norm": 0.1641325205564499,
+      "learning_rate": 4.29582980448363e-06,
+      "loss": 0.8991,
+      "step": 97280
+    },
+    {
+      "epoch": 0.704249820843015,
+      "grad_norm": 0.1451716423034668,
+      "learning_rate": 4.295757417823044e-06,
+      "loss": 0.9109,
+      "step": 97290
+    },
+    {
+      "epoch": 0.7043222075036012,
+      "grad_norm": 0.211027592420578,
+      "learning_rate": 4.295685031162457e-06,
+      "loss": 0.8876,
+      "step": 97300
+    },
+    {
+      "epoch": 0.7043945941641874,
+      "grad_norm": 0.15049049258232117,
+      "learning_rate": 4.295612644501872e-06,
+      "loss": 0.8982,
+      "step": 97310
+    },
+    {
+      "epoch": 0.7044669808247737,
+      "grad_norm": 0.1562298983335495,
+      "learning_rate": 4.2955402578412854e-06,
+      "loss": 0.9015,
+      "step": 97320
+    },
+    {
+      "epoch": 0.7045393674853598,
+      "grad_norm": 0.1571105718612671,
+      "learning_rate": 4.295467871180699e-06,
+      "loss": 0.8958,
+      "step": 97330
+    },
+    {
+      "epoch": 0.704611754145946,
+      "grad_norm": 0.1848883330821991,
+      "learning_rate": 4.295395484520113e-06,
+      "loss": 0.894,
+      "step": 97340
+    },
+    {
+      "epoch": 0.7046841408065322,
+      "grad_norm": 0.15157540142536163,
+      "learning_rate": 4.295323097859527e-06,
+      "loss": 0.8881,
+      "step": 97350
+    },
+    {
+      "epoch": 0.7047565274671184,
+      "grad_norm": 0.16974295675754547,
+      "learning_rate": 4.295250711198941e-06,
+      "loss": 0.8998,
+      "step": 97360
+    },
+    {
+      "epoch": 0.7048289141277045,
+      "grad_norm": 0.15676848590373993,
+      "learning_rate": 4.295178324538354e-06,
+      "loss": 0.8847,
+      "step": 97370
+    },
+    {
+      "epoch": 0.7049013007882907,
+      "grad_norm": 0.15643902122974396,
+      "learning_rate": 4.295105937877768e-06,
+      "loss": 0.9044,
+      "step": 97380
+    },
+    {
+      "epoch": 0.7049736874488769,
+      "grad_norm": 0.15624825656414032,
+      "learning_rate": 4.2950335512171824e-06,
+      "loss": 0.8963,
+      "step": 97390
+    },
+    {
+      "epoch": 0.7050460741094631,
+      "grad_norm": 0.16203194856643677,
+      "learning_rate": 4.294961164556596e-06,
+      "loss": 0.8655,
+      "step": 97400
+    },
+    {
+      "epoch": 0.7051184607700492,
+      "grad_norm": 0.16650012135505676,
+      "learning_rate": 4.29488877789601e-06,
+      "loss": 0.9065,
+      "step": 97410
+    },
+    {
+      "epoch": 0.7051908474306355,
+      "grad_norm": 0.16026648879051208,
+      "learning_rate": 4.294816391235423e-06,
+      "loss": 0.9028,
+      "step": 97420
+    },
+    {
+      "epoch": 0.7052632340912217,
+      "grad_norm": 0.15564897656440735,
+      "learning_rate": 4.294744004574837e-06,
+      "loss": 0.8955,
+      "step": 97430
+    },
+    {
+      "epoch": 0.7053356207518079,
+      "grad_norm": 0.15881158411502838,
+      "learning_rate": 4.294671617914251e-06,
+      "loss": 0.9105,
+      "step": 97440
+    },
+    {
+      "epoch": 0.7054080074123941,
+      "grad_norm": 0.14864668250083923,
+      "learning_rate": 4.294599231253665e-06,
+      "loss": 0.8795,
+      "step": 97450
+    },
+    {
+      "epoch": 0.7054803940729802,
+      "grad_norm": 0.16540266573429108,
+      "learning_rate": 4.294526844593079e-06,
+      "loss": 0.8881,
+      "step": 97460
+    },
+    {
+      "epoch": 0.7055527807335664,
+      "grad_norm": 0.1469249576330185,
+      "learning_rate": 4.294454457932492e-06,
+      "loss": 0.8988,
+      "step": 97470
+    },
+    {
+      "epoch": 0.7056251673941526,
+      "grad_norm": 0.1481475979089737,
+      "learning_rate": 4.294382071271907e-06,
+      "loss": 0.9061,
+      "step": 97480
+    },
+    {
+      "epoch": 0.7056975540547388,
+      "grad_norm": 0.14965546131134033,
+      "learning_rate": 4.29430968461132e-06,
+      "loss": 0.9011,
+      "step": 97490
+    },
+    {
+      "epoch": 0.705769940715325,
+      "grad_norm": 0.1463448852300644,
+      "learning_rate": 4.294237297950734e-06,
+      "loss": 0.8892,
+      "step": 97500
+    },
+    {
+      "epoch": 0.7058423273759111,
+      "grad_norm": 0.157759428024292,
+      "learning_rate": 4.2941649112901475e-06,
+      "loss": 0.8854,
+      "step": 97510
+    },
+    {
+      "epoch": 0.7059147140364973,
+      "grad_norm": 0.15894946455955505,
+      "learning_rate": 4.294092524629562e-06,
+      "loss": 0.8911,
+      "step": 97520
+    },
+    {
+      "epoch": 0.7059871006970836,
+      "grad_norm": 0.16521485149860382,
+      "learning_rate": 4.294020137968976e-06,
+      "loss": 0.8939,
+      "step": 97530
+    },
+    {
+      "epoch": 0.7060594873576698,
+      "grad_norm": 0.16126827895641327,
+      "learning_rate": 4.293947751308389e-06,
+      "loss": 0.8838,
+      "step": 97540
+    },
+    {
+      "epoch": 0.706131874018256,
+      "grad_norm": 0.18010060489177704,
+      "learning_rate": 4.293875364647803e-06,
+      "loss": 0.9076,
+      "step": 97550
+    },
+    {
+      "epoch": 0.7062042606788421,
+      "grad_norm": 0.15842019021511078,
+      "learning_rate": 4.293802977987217e-06,
+      "loss": 0.8983,
+      "step": 97560
+    },
+    {
+      "epoch": 0.7062766473394283,
+      "grad_norm": 0.14757438004016876,
+      "learning_rate": 4.293730591326631e-06,
+      "loss": 0.893,
+      "step": 97570
+    },
+    {
+      "epoch": 0.7063490340000145,
+      "grad_norm": 0.15686437487602234,
+      "learning_rate": 4.2936582046660445e-06,
+      "loss": 0.8923,
+      "step": 97580
+    },
+    {
+      "epoch": 0.7064214206606007,
+      "grad_norm": 0.15694831311702728,
+      "learning_rate": 4.293585818005458e-06,
+      "loss": 0.9027,
+      "step": 97590
+    },
+    {
+      "epoch": 0.7064938073211868,
+      "grad_norm": 0.15480537712574005,
+      "learning_rate": 4.293513431344873e-06,
+      "loss": 0.8853,
+      "step": 97600
+    },
+    {
+      "epoch": 0.706566193981773,
+      "grad_norm": 0.17100360989570618,
+      "learning_rate": 4.293441044684286e-06,
+      "loss": 0.8935,
+      "step": 97610
+    },
+    {
+      "epoch": 0.7066385806423592,
+      "grad_norm": 0.15572969615459442,
+      "learning_rate": 4.2933686580237e-06,
+      "loss": 0.8777,
+      "step": 97620
+    },
+    {
+      "epoch": 0.7067109673029454,
+      "grad_norm": 0.16251128911972046,
+      "learning_rate": 4.2932962713631135e-06,
+      "loss": 0.8977,
+      "step": 97630
+    },
+    {
+      "epoch": 0.7067833539635316,
+      "grad_norm": 0.15367302298545837,
+      "learning_rate": 4.293223884702528e-06,
+      "loss": 0.8939,
+      "step": 97640
+    },
+    {
+      "epoch": 0.7068557406241178,
+      "grad_norm": 0.1467921882867813,
+      "learning_rate": 4.2931514980419415e-06,
+      "loss": 0.8904,
+      "step": 97650
+    },
+    {
+      "epoch": 0.706928127284704,
+      "grad_norm": 0.14249904453754425,
+      "learning_rate": 4.293079111381355e-06,
+      "loss": 0.8977,
+      "step": 97660
+    },
+    {
+      "epoch": 0.7070005139452902,
+      "grad_norm": 0.15507516264915466,
+      "learning_rate": 4.293006724720769e-06,
+      "loss": 0.9006,
+      "step": 97670
+    },
+    {
+      "epoch": 0.7070729006058764,
+      "grad_norm": 0.147030770778656,
+      "learning_rate": 4.292934338060183e-06,
+      "loss": 0.8973,
+      "step": 97680
+    },
+    {
+      "epoch": 0.7071452872664625,
+      "grad_norm": 0.14783620834350586,
+      "learning_rate": 4.292861951399596e-06,
+      "loss": 0.8842,
+      "step": 97690
+    },
+    {
+      "epoch": 0.7072176739270487,
+      "grad_norm": 0.16624851524829865,
+      "learning_rate": 4.29278956473901e-06,
+      "loss": 0.8908,
+      "step": 97700
+    },
+    {
+      "epoch": 0.7072900605876349,
+      "grad_norm": 0.13825160264968872,
+      "learning_rate": 4.292717178078424e-06,
+      "loss": 0.8955,
+      "step": 97710
+    },
+    {
+      "epoch": 0.7073624472482211,
+      "grad_norm": 0.1511533409357071,
+      "learning_rate": 4.292644791417838e-06,
+      "loss": 0.8846,
+      "step": 97720
+    },
+    {
+      "epoch": 0.7074348339088072,
+      "grad_norm": 0.1551799476146698,
+      "learning_rate": 4.292572404757251e-06,
+      "loss": 0.8946,
+      "step": 97730
+    },
+    {
+      "epoch": 0.7075072205693935,
+      "grad_norm": 0.1462963968515396,
+      "learning_rate": 4.292500018096665e-06,
+      "loss": 0.8802,
+      "step": 97740
+    },
+    {
+      "epoch": 0.7075796072299797,
+      "grad_norm": 0.15262961387634277,
+      "learning_rate": 4.292427631436079e-06,
+      "loss": 0.9035,
+      "step": 97750
+    },
+    {
+      "epoch": 0.7076519938905659,
+      "grad_norm": 0.1675291806459427,
+      "learning_rate": 4.292355244775493e-06,
+      "loss": 0.8842,
+      "step": 97760
+    },
+    {
+      "epoch": 0.707724380551152,
+      "grad_norm": 0.15412959456443787,
+      "learning_rate": 4.292282858114907e-06,
+      "loss": 0.8899,
+      "step": 97770
+    },
+    {
+      "epoch": 0.7077967672117382,
+      "grad_norm": 0.16402597725391388,
+      "learning_rate": 4.29221047145432e-06,
+      "loss": 0.8956,
+      "step": 97780
+    },
+    {
+      "epoch": 0.7078691538723244,
+      "grad_norm": 0.15542785823345184,
+      "learning_rate": 4.292138084793735e-06,
+      "loss": 0.8926,
+      "step": 97790
+    },
+    {
+      "epoch": 0.7079415405329106,
+      "grad_norm": 0.16312408447265625,
+      "learning_rate": 4.292065698133148e-06,
+      "loss": 0.9012,
+      "step": 97800
+    },
+    {
+      "epoch": 0.7080139271934968,
+      "grad_norm": 0.16359078884124756,
+      "learning_rate": 4.291993311472562e-06,
+      "loss": 0.889,
+      "step": 97810
+    },
+    {
+      "epoch": 0.7080863138540829,
+      "grad_norm": 0.1581135094165802,
+      "learning_rate": 4.2919209248119756e-06,
+      "loss": 0.8854,
+      "step": 97820
+    },
+    {
+      "epoch": 0.7081587005146691,
+      "grad_norm": 0.16701552271842957,
+      "learning_rate": 4.29184853815139e-06,
+      "loss": 0.8877,
+      "step": 97830
+    },
+    {
+      "epoch": 0.7082310871752553,
+      "grad_norm": 0.18646441400051117,
+      "learning_rate": 4.291776151490804e-06,
+      "loss": 0.8825,
+      "step": 97840
+    },
+    {
+      "epoch": 0.7083034738358416,
+      "grad_norm": 0.15613111853599548,
+      "learning_rate": 4.291703764830217e-06,
+      "loss": 0.8889,
+      "step": 97850
+    },
+    {
+      "epoch": 0.7083758604964278,
+      "grad_norm": 0.15725113451480865,
+      "learning_rate": 4.291631378169631e-06,
+      "loss": 0.892,
+      "step": 97860
+    },
+    {
+      "epoch": 0.7084482471570139,
+      "grad_norm": 0.1500454545021057,
+      "learning_rate": 4.291558991509045e-06,
+      "loss": 0.8845,
+      "step": 97870
+    },
+    {
+      "epoch": 0.7085206338176001,
+      "grad_norm": 0.14962556958198547,
+      "learning_rate": 4.291486604848459e-06,
+      "loss": 0.9005,
+      "step": 97880
+    },
+    {
+      "epoch": 0.7085930204781863,
+      "grad_norm": 0.15193293988704681,
+      "learning_rate": 4.2914142181878726e-06,
+      "loss": 0.898,
+      "step": 97890
+    },
+    {
+      "epoch": 0.7086654071387725,
+      "grad_norm": 0.19393183290958405,
+      "learning_rate": 4.291341831527286e-06,
+      "loss": 0.8808,
+      "step": 97900
+    },
+    {
+      "epoch": 0.7087377937993586,
+      "grad_norm": 0.16170886158943176,
+      "learning_rate": 4.291269444866701e-06,
+      "loss": 0.8892,
+      "step": 97910
+    },
+    {
+      "epoch": 0.7088101804599448,
+      "grad_norm": 0.16696853935718536,
+      "learning_rate": 4.291197058206114e-06,
+      "loss": 0.8952,
+      "step": 97920
+    },
+    {
+      "epoch": 0.708882567120531,
+      "grad_norm": 0.14689412713050842,
+      "learning_rate": 4.291124671545528e-06,
+      "loss": 0.8976,
+      "step": 97930
+    },
+    {
+      "epoch": 0.7089549537811172,
+      "grad_norm": 0.18432478606700897,
+      "learning_rate": 4.2910522848849415e-06,
+      "loss": 0.888,
+      "step": 97940
+    },
+    {
+      "epoch": 0.7090273404417035,
+      "grad_norm": 0.14978989958763123,
+      "learning_rate": 4.290979898224356e-06,
+      "loss": 0.8878,
+      "step": 97950
+    },
+    {
+      "epoch": 0.7090997271022896,
+      "grad_norm": 0.14927172660827637,
+      "learning_rate": 4.2909075115637696e-06,
+      "loss": 0.9018,
+      "step": 97960
+    },
+    {
+      "epoch": 0.7091721137628758,
+      "grad_norm": 0.1598445326089859,
+      "learning_rate": 4.290835124903183e-06,
+      "loss": 0.8776,
+      "step": 97970
+    },
+    {
+      "epoch": 0.709244500423462,
+      "grad_norm": 0.15578649938106537,
+      "learning_rate": 4.290762738242597e-06,
+      "loss": 0.8925,
+      "step": 97980
+    },
+    {
+      "epoch": 0.7093168870840482,
+      "grad_norm": 0.15951481461524963,
+      "learning_rate": 4.290690351582011e-06,
+      "loss": 0.8979,
+      "step": 97990
+    },
+    {
+      "epoch": 0.7093892737446343,
+      "grad_norm": 0.15647555887699127,
+      "learning_rate": 4.290617964921425e-06,
+      "loss": 0.9094,
+      "step": 98000
+    },
+    {
+      "epoch": 0.7094616604052205,
+      "grad_norm": 0.15051069855690002,
+      "learning_rate": 4.2905455782608385e-06,
+      "loss": 0.8905,
+      "step": 98010
+    },
+    {
+      "epoch": 0.7095340470658067,
+      "grad_norm": 0.17179308831691742,
+      "learning_rate": 4.290473191600252e-06,
+      "loss": 0.883,
+      "step": 98020
+    },
+    {
+      "epoch": 0.7096064337263929,
+      "grad_norm": 0.16709086298942566,
+      "learning_rate": 4.290400804939666e-06,
+      "loss": 0.9033,
+      "step": 98030
+    },
+    {
+      "epoch": 0.709678820386979,
+      "grad_norm": 0.17320266366004944,
+      "learning_rate": 4.29032841827908e-06,
+      "loss": 0.9088,
+      "step": 98040
+    },
+    {
+      "epoch": 0.7097512070475652,
+      "grad_norm": 0.16562733054161072,
+      "learning_rate": 4.290256031618494e-06,
+      "loss": 0.8882,
+      "step": 98050
+    },
+    {
+      "epoch": 0.7098235937081515,
+      "grad_norm": 0.15800151228904724,
+      "learning_rate": 4.2901836449579074e-06,
+      "loss": 0.8935,
+      "step": 98060
+    },
+    {
+      "epoch": 0.7098959803687377,
+      "grad_norm": 0.15598443150520325,
+      "learning_rate": 4.290111258297321e-06,
+      "loss": 0.8878,
+      "step": 98070
+    },
+    {
+      "epoch": 0.7099683670293239,
+      "grad_norm": 0.14312607049942017,
+      "learning_rate": 4.2900388716367355e-06,
+      "loss": 0.8901,
+      "step": 98080
+    },
+    {
+      "epoch": 0.71004075368991,
+      "grad_norm": 0.16290390491485596,
+      "learning_rate": 4.289966484976149e-06,
+      "loss": 0.8821,
+      "step": 98090
+    },
+    {
+      "epoch": 0.7101131403504962,
+      "grad_norm": 0.1655735820531845,
+      "learning_rate": 4.289894098315563e-06,
+      "loss": 0.8994,
+      "step": 98100
+    },
+    {
+      "epoch": 0.7101855270110824,
+      "grad_norm": 0.1505131721496582,
+      "learning_rate": 4.289821711654976e-06,
+      "loss": 0.8779,
+      "step": 98110
+    },
+    {
+      "epoch": 0.7102579136716686,
+      "grad_norm": 0.1438320130109787,
+      "learning_rate": 4.289749324994391e-06,
+      "loss": 0.8743,
+      "step": 98120
+    },
+    {
+      "epoch": 0.7103303003322547,
+      "grad_norm": 0.18201330304145813,
+      "learning_rate": 4.2896769383338044e-06,
+      "loss": 0.8858,
+      "step": 98130
+    },
+    {
+      "epoch": 0.7104026869928409,
+      "grad_norm": 0.18367905914783478,
+      "learning_rate": 4.289604551673218e-06,
+      "loss": 0.9101,
+      "step": 98140
+    },
+    {
+      "epoch": 0.7104750736534271,
+      "grad_norm": 0.15578269958496094,
+      "learning_rate": 4.289532165012632e-06,
+      "loss": 0.893,
+      "step": 98150
+    },
+    {
+      "epoch": 0.7105474603140133,
+      "grad_norm": 0.15970683097839355,
+      "learning_rate": 4.289459778352046e-06,
+      "loss": 0.8996,
+      "step": 98160
+    },
+    {
+      "epoch": 0.7106198469745996,
+      "grad_norm": 0.16508597135543823,
+      "learning_rate": 4.28938739169146e-06,
+      "loss": 0.8862,
+      "step": 98170
+    },
+    {
+      "epoch": 0.7106922336351857,
+      "grad_norm": 0.174842968583107,
+      "learning_rate": 4.289315005030873e-06,
+      "loss": 0.9011,
+      "step": 98180
+    },
+    {
+      "epoch": 0.7107646202957719,
+      "grad_norm": 0.17075632512569427,
+      "learning_rate": 4.289242618370287e-06,
+      "loss": 0.8765,
+      "step": 98190
+    },
+    {
+      "epoch": 0.7108370069563581,
+      "grad_norm": 0.15258832275867462,
+      "learning_rate": 4.2891702317097014e-06,
+      "loss": 0.895,
+      "step": 98200
+    },
+    {
+      "epoch": 0.7109093936169443,
+      "grad_norm": 0.14689311385154724,
+      "learning_rate": 4.289097845049115e-06,
+      "loss": 0.8874,
+      "step": 98210
+    },
+    {
+      "epoch": 0.7109817802775305,
+      "grad_norm": 0.16319765150547028,
+      "learning_rate": 4.289025458388528e-06,
+      "loss": 0.8929,
+      "step": 98220
+    },
+    {
+      "epoch": 0.7110541669381166,
+      "grad_norm": 0.16500282287597656,
+      "learning_rate": 4.288953071727942e-06,
+      "loss": 0.887,
+      "step": 98230
+    },
+    {
+      "epoch": 0.7111265535987028,
+      "grad_norm": 0.15623874962329865,
+      "learning_rate": 4.288880685067356e-06,
+      "loss": 0.8959,
+      "step": 98240
+    },
+    {
+      "epoch": 0.711198940259289,
+      "grad_norm": 0.15647245943546295,
+      "learning_rate": 4.2888082984067695e-06,
+      "loss": 0.8864,
+      "step": 98250
+    },
+    {
+      "epoch": 0.7112713269198752,
+      "grad_norm": 0.1665528565645218,
+      "learning_rate": 4.288735911746183e-06,
+      "loss": 0.8903,
+      "step": 98260
+    },
+    {
+      "epoch": 0.7113437135804614,
+      "grad_norm": 0.14643557369709015,
+      "learning_rate": 4.288663525085598e-06,
+      "loss": 0.8978,
+      "step": 98270
+    },
+    {
+      "epoch": 0.7114161002410476,
+      "grad_norm": 0.154141366481781,
+      "learning_rate": 4.288591138425011e-06,
+      "loss": 0.8978,
+      "step": 98280
+    },
+    {
+      "epoch": 0.7114884869016338,
+      "grad_norm": 0.14892645180225372,
+      "learning_rate": 4.288518751764425e-06,
+      "loss": 0.8925,
+      "step": 98290
+    },
+    {
+      "epoch": 0.71156087356222,
+      "grad_norm": 0.17983193695545197,
+      "learning_rate": 4.2884463651038384e-06,
+      "loss": 0.8798,
+      "step": 98300
+    },
+    {
+      "epoch": 0.7116332602228062,
+      "grad_norm": 0.15208591520786285,
+      "learning_rate": 4.288373978443253e-06,
+      "loss": 0.8917,
+      "step": 98310
+    },
+    {
+      "epoch": 0.7117056468833923,
+      "grad_norm": 0.15199977159500122,
+      "learning_rate": 4.2883015917826665e-06,
+      "loss": 0.8878,
+      "step": 98320
+    },
+    {
+      "epoch": 0.7117780335439785,
+      "grad_norm": 0.16912555694580078,
+      "learning_rate": 4.28822920512208e-06,
+      "loss": 0.8876,
+      "step": 98330
+    },
+    {
+      "epoch": 0.7118504202045647,
+      "grad_norm": 0.18677960336208344,
+      "learning_rate": 4.288156818461494e-06,
+      "loss": 0.8971,
+      "step": 98340
+    },
+    {
+      "epoch": 0.7119228068651509,
+      "grad_norm": 0.16484694182872772,
+      "learning_rate": 4.288084431800908e-06,
+      "loss": 0.8867,
+      "step": 98350
+    },
+    {
+      "epoch": 0.711995193525737,
+      "grad_norm": 0.18507961928844452,
+      "learning_rate": 4.288012045140322e-06,
+      "loss": 0.9005,
+      "step": 98360
+    },
+    {
+      "epoch": 0.7120675801863232,
+      "grad_norm": 0.16130656003952026,
+      "learning_rate": 4.2879396584797355e-06,
+      "loss": 0.9067,
+      "step": 98370
+    },
+    {
+      "epoch": 0.7121399668469095,
+      "grad_norm": 0.16613750159740448,
+      "learning_rate": 4.287867271819149e-06,
+      "loss": 0.8972,
+      "step": 98380
+    },
+    {
+      "epoch": 0.7122123535074957,
+      "grad_norm": 0.1898043304681778,
+      "learning_rate": 4.2877948851585635e-06,
+      "loss": 0.8951,
+      "step": 98390
+    },
+    {
+      "epoch": 0.7122847401680819,
+      "grad_norm": 0.15496717393398285,
+      "learning_rate": 4.287722498497977e-06,
+      "loss": 0.8843,
+      "step": 98400
+    },
+    {
+      "epoch": 0.712357126828668,
+      "grad_norm": 0.15950420498847961,
+      "learning_rate": 4.287650111837391e-06,
+      "loss": 0.8907,
+      "step": 98410
+    },
+    {
+      "epoch": 0.7124295134892542,
+      "grad_norm": 0.16205261647701263,
+      "learning_rate": 4.287577725176804e-06,
+      "loss": 0.8871,
+      "step": 98420
+    },
+    {
+      "epoch": 0.7125019001498404,
+      "grad_norm": 0.16304804384708405,
+      "learning_rate": 4.287505338516219e-06,
+      "loss": 0.8884,
+      "step": 98430
+    },
+    {
+      "epoch": 0.7125742868104266,
+      "grad_norm": 0.15807406604290009,
+      "learning_rate": 4.2874329518556325e-06,
+      "loss": 0.8893,
+      "step": 98440
+    },
+    {
+      "epoch": 0.7126466734710127,
+      "grad_norm": 0.16383838653564453,
+      "learning_rate": 4.287360565195046e-06,
+      "loss": 0.9048,
+      "step": 98450
+    },
+    {
+      "epoch": 0.7127190601315989,
+      "grad_norm": 0.17542411386966705,
+      "learning_rate": 4.28728817853446e-06,
+      "loss": 0.8776,
+      "step": 98460
+    },
+    {
+      "epoch": 0.7127914467921851,
+      "grad_norm": 0.18369115889072418,
+      "learning_rate": 4.287215791873874e-06,
+      "loss": 0.888,
+      "step": 98470
+    },
+    {
+      "epoch": 0.7128638334527714,
+      "grad_norm": 0.21220123767852783,
+      "learning_rate": 4.287143405213288e-06,
+      "loss": 0.8996,
+      "step": 98480
+    },
+    {
+      "epoch": 0.7129362201133576,
+      "grad_norm": 0.17068584263324738,
+      "learning_rate": 4.287071018552701e-06,
+      "loss": 0.8882,
+      "step": 98490
+    },
+    {
+      "epoch": 0.7130086067739437,
+      "grad_norm": 0.15225493907928467,
+      "learning_rate": 4.286998631892115e-06,
+      "loss": 0.8924,
+      "step": 98500
+    },
+    {
+      "epoch": 0.7130809934345299,
+      "grad_norm": 0.14265893399715424,
+      "learning_rate": 4.2869262452315295e-06,
+      "loss": 0.8832,
+      "step": 98510
+    },
+    {
+      "epoch": 0.7131533800951161,
+      "grad_norm": 0.18743206560611725,
+      "learning_rate": 4.286853858570943e-06,
+      "loss": 0.9077,
+      "step": 98520
+    },
+    {
+      "epoch": 0.7132257667557023,
+      "grad_norm": 0.15794509649276733,
+      "learning_rate": 4.286781471910357e-06,
+      "loss": 0.8924,
+      "step": 98530
+    },
+    {
+      "epoch": 0.7132981534162884,
+      "grad_norm": 0.16649536788463593,
+      "learning_rate": 4.28670908524977e-06,
+      "loss": 0.8789,
+      "step": 98540
+    },
+    {
+      "epoch": 0.7133705400768746,
+      "grad_norm": 0.1717004030942917,
+      "learning_rate": 4.286636698589185e-06,
+      "loss": 0.8966,
+      "step": 98550
+    },
+    {
+      "epoch": 0.7134429267374608,
+      "grad_norm": 0.1472935825586319,
+      "learning_rate": 4.286564311928598e-06,
+      "loss": 0.8928,
+      "step": 98560
+    },
+    {
+      "epoch": 0.713515313398047,
+      "grad_norm": 0.1435527801513672,
+      "learning_rate": 4.286491925268012e-06,
+      "loss": 0.872,
+      "step": 98570
+    },
+    {
+      "epoch": 0.7135877000586331,
+      "grad_norm": 0.17352406680583954,
+      "learning_rate": 4.286419538607426e-06,
+      "loss": 0.883,
+      "step": 98580
+    },
+    {
+      "epoch": 0.7136600867192194,
+      "grad_norm": 0.14839379489421844,
+      "learning_rate": 4.28634715194684e-06,
+      "loss": 0.8919,
+      "step": 98590
+    },
+    {
+      "epoch": 0.7137324733798056,
+      "grad_norm": 0.1555548906326294,
+      "learning_rate": 4.286274765286254e-06,
+      "loss": 0.8863,
+      "step": 98600
+    },
+    {
+      "epoch": 0.7138048600403918,
+      "grad_norm": 0.1623990833759308,
+      "learning_rate": 4.286202378625667e-06,
+      "loss": 0.8832,
+      "step": 98610
+    },
+    {
+      "epoch": 0.713877246700978,
+      "grad_norm": 0.19231371581554413,
+      "learning_rate": 4.286129991965081e-06,
+      "loss": 0.8975,
+      "step": 98620
+    },
+    {
+      "epoch": 0.7139496333615641,
+      "grad_norm": 0.15244658291339874,
+      "learning_rate": 4.286057605304495e-06,
+      "loss": 0.9031,
+      "step": 98630
+    },
+    {
+      "epoch": 0.7140220200221503,
+      "grad_norm": 0.15277956426143646,
+      "learning_rate": 4.285985218643909e-06,
+      "loss": 0.888,
+      "step": 98640
+    },
+    {
+      "epoch": 0.7140944066827365,
+      "grad_norm": 0.15656965970993042,
+      "learning_rate": 4.285912831983323e-06,
+      "loss": 0.8905,
+      "step": 98650
+    },
+    {
+      "epoch": 0.7141667933433227,
+      "grad_norm": 0.18114785850048065,
+      "learning_rate": 4.285840445322736e-06,
+      "loss": 0.8754,
+      "step": 98660
+    },
+    {
+      "epoch": 0.7142391800039088,
+      "grad_norm": 0.23680664598941803,
+      "learning_rate": 4.28576805866215e-06,
+      "loss": 0.894,
+      "step": 98670
+    },
+    {
+      "epoch": 0.714311566664495,
+      "grad_norm": 0.14668525755405426,
+      "learning_rate": 4.285695672001564e-06,
+      "loss": 0.8903,
+      "step": 98680
+    },
+    {
+      "epoch": 0.7143839533250812,
+      "grad_norm": 0.1626420021057129,
+      "learning_rate": 4.285623285340978e-06,
+      "loss": 0.8976,
+      "step": 98690
+    },
+    {
+      "epoch": 0.7144563399856675,
+      "grad_norm": 0.154447540640831,
+      "learning_rate": 4.2855508986803916e-06,
+      "loss": 0.8961,
+      "step": 98700
+    },
+    {
+      "epoch": 0.7145287266462537,
+      "grad_norm": 0.15234994888305664,
+      "learning_rate": 4.285478512019805e-06,
+      "loss": 0.8995,
+      "step": 98710
+    },
+    {
+      "epoch": 0.7146011133068398,
+      "grad_norm": 0.15986403822898865,
+      "learning_rate": 4.28540612535922e-06,
+      "loss": 0.8968,
+      "step": 98720
+    },
+    {
+      "epoch": 0.714673499967426,
+      "grad_norm": 0.16405628621578217,
+      "learning_rate": 4.285333738698633e-06,
+      "loss": 0.8877,
+      "step": 98730
+    },
+    {
+      "epoch": 0.7147458866280122,
+      "grad_norm": 0.14801183342933655,
+      "learning_rate": 4.285261352038047e-06,
+      "loss": 0.8865,
+      "step": 98740
+    },
+    {
+      "epoch": 0.7148182732885984,
+      "grad_norm": 0.1582917869091034,
+      "learning_rate": 4.2851889653774605e-06,
+      "loss": 0.8968,
+      "step": 98750
+    },
+    {
+      "epoch": 0.7148906599491845,
+      "grad_norm": 0.14904256165027618,
+      "learning_rate": 4.285116578716874e-06,
+      "loss": 0.8825,
+      "step": 98760
+    },
+    {
+      "epoch": 0.7149630466097707,
+      "grad_norm": 0.15857091546058655,
+      "learning_rate": 4.285044192056288e-06,
+      "loss": 0.8883,
+      "step": 98770
+    },
+    {
+      "epoch": 0.7150354332703569,
+      "grad_norm": 0.1512639969587326,
+      "learning_rate": 4.284971805395702e-06,
+      "loss": 0.9077,
+      "step": 98780
+    },
+    {
+      "epoch": 0.7151078199309431,
+      "grad_norm": 0.1620493084192276,
+      "learning_rate": 4.284899418735116e-06,
+      "loss": 0.8921,
+      "step": 98790
+    },
+    {
+      "epoch": 0.7151802065915294,
+      "grad_norm": 0.15523122251033783,
+      "learning_rate": 4.2848270320745294e-06,
+      "loss": 0.8975,
+      "step": 98800
+    },
+    {
+      "epoch": 0.7152525932521155,
+      "grad_norm": 0.15379586815834045,
+      "learning_rate": 4.284754645413943e-06,
+      "loss": 0.9056,
+      "step": 98810
+    },
+    {
+      "epoch": 0.7153249799127017,
+      "grad_norm": 0.15808962285518646,
+      "learning_rate": 4.2846822587533575e-06,
+      "loss": 0.8923,
+      "step": 98820
+    },
+    {
+      "epoch": 0.7153973665732879,
+      "grad_norm": 0.21590928733348846,
+      "learning_rate": 4.284609872092771e-06,
+      "loss": 0.8851,
+      "step": 98830
+    },
+    {
+      "epoch": 0.7154697532338741,
+      "grad_norm": 0.16815008223056793,
+      "learning_rate": 4.284537485432185e-06,
+      "loss": 0.8893,
+      "step": 98840
+    },
+    {
+      "epoch": 0.7155421398944602,
+      "grad_norm": 0.15785428881645203,
+      "learning_rate": 4.284465098771598e-06,
+      "loss": 0.8874,
+      "step": 98850
+    },
+    {
+      "epoch": 0.7156145265550464,
+      "grad_norm": 0.1511503905057907,
+      "learning_rate": 4.284392712111012e-06,
+      "loss": 0.8789,
+      "step": 98860
+    },
+    {
+      "epoch": 0.7156869132156326,
+      "grad_norm": 0.15221667289733887,
+      "learning_rate": 4.2843203254504264e-06,
+      "loss": 0.8909,
+      "step": 98870
+    },
+    {
+      "epoch": 0.7157592998762188,
+      "grad_norm": 0.15201707184314728,
+      "learning_rate": 4.28424793878984e-06,
+      "loss": 0.8908,
+      "step": 98880
+    },
+    {
+      "epoch": 0.715831686536805,
+      "grad_norm": 0.1675233393907547,
+      "learning_rate": 4.284175552129254e-06,
+      "loss": 0.9029,
+      "step": 98890
+    },
+    {
+      "epoch": 0.7159040731973911,
+      "grad_norm": 0.1582505851984024,
+      "learning_rate": 4.284103165468667e-06,
+      "loss": 0.894,
+      "step": 98900
+    },
+    {
+      "epoch": 0.7159764598579774,
+      "grad_norm": 0.14627264440059662,
+      "learning_rate": 4.284030778808082e-06,
+      "loss": 0.8948,
+      "step": 98910
+    },
+    {
+      "epoch": 0.7160488465185636,
+      "grad_norm": 0.15614593029022217,
+      "learning_rate": 4.283958392147495e-06,
+      "loss": 0.8895,
+      "step": 98920
+    },
+    {
+      "epoch": 0.7161212331791498,
+      "grad_norm": 0.17109185457229614,
+      "learning_rate": 4.283886005486909e-06,
+      "loss": 0.9064,
+      "step": 98930
+    },
+    {
+      "epoch": 0.716193619839736,
+      "grad_norm": 0.17292575538158417,
+      "learning_rate": 4.283813618826323e-06,
+      "loss": 0.8936,
+      "step": 98940
+    },
+    {
+      "epoch": 0.7162660065003221,
+      "grad_norm": 0.15228956937789917,
+      "learning_rate": 4.283741232165737e-06,
+      "loss": 0.8948,
+      "step": 98950
+    },
+    {
+      "epoch": 0.7163383931609083,
+      "grad_norm": 0.15983666479587555,
+      "learning_rate": 4.283668845505151e-06,
+      "loss": 0.892,
+      "step": 98960
+    },
+    {
+      "epoch": 0.7164107798214945,
+      "grad_norm": 0.15323320031166077,
+      "learning_rate": 4.283596458844564e-06,
+      "loss": 0.885,
+      "step": 98970
+    },
+    {
+      "epoch": 0.7164831664820807,
+      "grad_norm": 0.1651761531829834,
+      "learning_rate": 4.283524072183978e-06,
+      "loss": 0.8907,
+      "step": 98980
+    },
+    {
+      "epoch": 0.7165555531426668,
+      "grad_norm": 0.1607891172170639,
+      "learning_rate": 4.283451685523392e-06,
+      "loss": 0.8986,
+      "step": 98990
+    },
+    {
+      "epoch": 0.716627939803253,
+      "grad_norm": 0.15800441801548004,
+      "learning_rate": 4.283379298862806e-06,
+      "loss": 0.8871,
+      "step": 99000
+    },
+    {
+      "epoch": 0.7167003264638393,
+      "grad_norm": 0.15280944108963013,
+      "learning_rate": 4.28330691220222e-06,
+      "loss": 0.892,
+      "step": 99010
+    },
+    {
+      "epoch": 0.7167727131244255,
+      "grad_norm": 0.15718095004558563,
+      "learning_rate": 4.283234525541633e-06,
+      "loss": 0.8885,
+      "step": 99020
+    },
+    {
+      "epoch": 0.7168450997850117,
+      "grad_norm": 0.15296663343906403,
+      "learning_rate": 4.283162138881048e-06,
+      "loss": 0.8945,
+      "step": 99030
+    },
+    {
+      "epoch": 0.7169174864455978,
+      "grad_norm": 0.1575295478105545,
+      "learning_rate": 4.283089752220461e-06,
+      "loss": 0.895,
+      "step": 99040
+    },
+    {
+      "epoch": 0.716989873106184,
+      "grad_norm": 0.16054439544677734,
+      "learning_rate": 4.283017365559875e-06,
+      "loss": 0.8788,
+      "step": 99050
+    },
+    {
+      "epoch": 0.7170622597667702,
+      "grad_norm": 0.15393316745758057,
+      "learning_rate": 4.2829449788992885e-06,
+      "loss": 0.881,
+      "step": 99060
+    },
+    {
+      "epoch": 0.7171346464273564,
+      "grad_norm": 0.16581709682941437,
+      "learning_rate": 4.282872592238703e-06,
+      "loss": 0.8949,
+      "step": 99070
+    },
+    {
+      "epoch": 0.7172070330879425,
+      "grad_norm": 0.16176964342594147,
+      "learning_rate": 4.282800205578117e-06,
+      "loss": 0.8834,
+      "step": 99080
+    },
+    {
+      "epoch": 0.7172794197485287,
+      "grad_norm": 0.15313786268234253,
+      "learning_rate": 4.28272781891753e-06,
+      "loss": 0.8849,
+      "step": 99090
+    },
+    {
+      "epoch": 0.7173518064091149,
+      "grad_norm": 0.1565057635307312,
+      "learning_rate": 4.282655432256944e-06,
+      "loss": 0.8898,
+      "step": 99100
+    },
+    {
+      "epoch": 0.7174241930697011,
+      "grad_norm": 0.17131567001342773,
+      "learning_rate": 4.282583045596358e-06,
+      "loss": 0.8981,
+      "step": 99110
+    },
+    {
+      "epoch": 0.7174965797302874,
+      "grad_norm": 0.16490551829338074,
+      "learning_rate": 4.282510658935772e-06,
+      "loss": 0.8875,
+      "step": 99120
+    },
+    {
+      "epoch": 0.7175689663908735,
+      "grad_norm": 0.14752908051013947,
+      "learning_rate": 4.2824382722751855e-06,
+      "loss": 0.8903,
+      "step": 99130
+    },
+    {
+      "epoch": 0.7176413530514597,
+      "grad_norm": 0.1532769799232483,
+      "learning_rate": 4.282365885614599e-06,
+      "loss": 0.8946,
+      "step": 99140
+    },
+    {
+      "epoch": 0.7177137397120459,
+      "grad_norm": 0.16182607412338257,
+      "learning_rate": 4.282293498954014e-06,
+      "loss": 0.8948,
+      "step": 99150
+    },
+    {
+      "epoch": 0.7177861263726321,
+      "grad_norm": 0.1951036900281906,
+      "learning_rate": 4.282221112293427e-06,
+      "loss": 0.8873,
+      "step": 99160
+    },
+    {
+      "epoch": 0.7178585130332182,
+      "grad_norm": 0.16654542088508606,
+      "learning_rate": 4.282148725632841e-06,
+      "loss": 0.8903,
+      "step": 99170
+    },
+    {
+      "epoch": 0.7179308996938044,
+      "grad_norm": 0.17811834812164307,
+      "learning_rate": 4.2820763389722545e-06,
+      "loss": 0.8951,
+      "step": 99180
+    },
+    {
+      "epoch": 0.7180032863543906,
+      "grad_norm": 0.14857879281044006,
+      "learning_rate": 4.282003952311669e-06,
+      "loss": 0.8927,
+      "step": 99190
+    },
+    {
+      "epoch": 0.7180756730149768,
+      "grad_norm": 0.16096261143684387,
+      "learning_rate": 4.2819315656510825e-06,
+      "loss": 0.8919,
+      "step": 99200
+    },
+    {
+      "epoch": 0.7181480596755629,
+      "grad_norm": 0.15119074285030365,
+      "learning_rate": 4.281859178990496e-06,
+      "loss": 0.8969,
+      "step": 99210
+    },
+    {
+      "epoch": 0.7182204463361491,
+      "grad_norm": 0.15937365591526031,
+      "learning_rate": 4.28178679232991e-06,
+      "loss": 0.8907,
+      "step": 99220
+    },
+    {
+      "epoch": 0.7182928329967354,
+      "grad_norm": 0.15309041738510132,
+      "learning_rate": 4.281714405669324e-06,
+      "loss": 0.8832,
+      "step": 99230
+    },
+    {
+      "epoch": 0.7183652196573216,
+      "grad_norm": 0.20644572377204895,
+      "learning_rate": 4.281642019008738e-06,
+      "loss": 0.8935,
+      "step": 99240
+    },
+    {
+      "epoch": 0.7184376063179078,
+      "grad_norm": 0.17840343713760376,
+      "learning_rate": 4.2815696323481515e-06,
+      "loss": 0.8981,
+      "step": 99250
+    },
+    {
+      "epoch": 0.7185099929784939,
+      "grad_norm": 0.16128534078598022,
+      "learning_rate": 4.281497245687565e-06,
+      "loss": 0.8925,
+      "step": 99260
+    },
+    {
+      "epoch": 0.7185823796390801,
+      "grad_norm": 0.16106021404266357,
+      "learning_rate": 4.2814248590269796e-06,
+      "loss": 0.8902,
+      "step": 99270
+    },
+    {
+      "epoch": 0.7186547662996663,
+      "grad_norm": 0.15783260762691498,
+      "learning_rate": 4.281352472366392e-06,
+      "loss": 0.878,
+      "step": 99280
+    },
+    {
+      "epoch": 0.7187271529602525,
+      "grad_norm": 0.17158517241477966,
+      "learning_rate": 4.281280085705806e-06,
+      "loss": 0.895,
+      "step": 99290
+    },
+    {
+      "epoch": 0.7187995396208386,
+      "grad_norm": 0.16213983297348022,
+      "learning_rate": 4.28120769904522e-06,
+      "loss": 0.8746,
+      "step": 99300
+    },
+    {
+      "epoch": 0.7188719262814248,
+      "grad_norm": 0.17305901646614075,
+      "learning_rate": 4.281135312384634e-06,
+      "loss": 0.8907,
+      "step": 99310
+    },
+    {
+      "epoch": 0.718944312942011,
+      "grad_norm": 0.18499404191970825,
+      "learning_rate": 4.281062925724048e-06,
+      "loss": 0.884,
+      "step": 99320
+    },
+    {
+      "epoch": 0.7190166996025973,
+      "grad_norm": 0.15206129848957062,
+      "learning_rate": 4.280990539063461e-06,
+      "loss": 0.8882,
+      "step": 99330
+    },
+    {
+      "epoch": 0.7190890862631835,
+      "grad_norm": 0.15170209109783173,
+      "learning_rate": 4.280918152402876e-06,
+      "loss": 0.9006,
+      "step": 99340
+    },
+    {
+      "epoch": 0.7191614729237696,
+      "grad_norm": 0.14841248095035553,
+      "learning_rate": 4.280845765742289e-06,
+      "loss": 0.8907,
+      "step": 99350
+    },
+    {
+      "epoch": 0.7192338595843558,
+      "grad_norm": 0.16877888143062592,
+      "learning_rate": 4.280773379081703e-06,
+      "loss": 0.8853,
+      "step": 99360
+    },
+    {
+      "epoch": 0.719306246244942,
+      "grad_norm": 0.14914186298847198,
+      "learning_rate": 4.2807009924211166e-06,
+      "loss": 0.9001,
+      "step": 99370
+    },
+    {
+      "epoch": 0.7193786329055282,
+      "grad_norm": 0.15811428427696228,
+      "learning_rate": 4.280628605760531e-06,
+      "loss": 0.8884,
+      "step": 99380
+    },
+    {
+      "epoch": 0.7194510195661143,
+      "grad_norm": 0.15445317327976227,
+      "learning_rate": 4.280556219099945e-06,
+      "loss": 0.8969,
+      "step": 99390
+    },
+    {
+      "epoch": 0.7195234062267005,
+      "grad_norm": 0.1613607257604599,
+      "learning_rate": 4.280483832439358e-06,
+      "loss": 0.8747,
+      "step": 99400
+    },
+    {
+      "epoch": 0.7195957928872867,
+      "grad_norm": 0.1554788202047348,
+      "learning_rate": 4.280411445778772e-06,
+      "loss": 0.8949,
+      "step": 99410
+    },
+    {
+      "epoch": 0.7196681795478729,
+      "grad_norm": 0.15722106397151947,
+      "learning_rate": 4.280339059118186e-06,
+      "loss": 0.888,
+      "step": 99420
+    },
+    {
+      "epoch": 0.719740566208459,
+      "grad_norm": 0.1679772436618805,
+      "learning_rate": 4.2802666724576e-06,
+      "loss": 0.8853,
+      "step": 99430
+    },
+    {
+      "epoch": 0.7198129528690453,
+      "grad_norm": 0.676620602607727,
+      "learning_rate": 4.2801942857970136e-06,
+      "loss": 0.884,
+      "step": 99440
+    },
+    {
+      "epoch": 0.7198853395296315,
+      "grad_norm": 0.15480059385299683,
+      "learning_rate": 4.280121899136427e-06,
+      "loss": 0.8963,
+      "step": 99450
+    },
+    {
+      "epoch": 0.7199577261902177,
+      "grad_norm": 0.15194228291511536,
+      "learning_rate": 4.280049512475841e-06,
+      "loss": 0.8896,
+      "step": 99460
+    },
+    {
+      "epoch": 0.7200301128508039,
+      "grad_norm": 0.17319074273109436,
+      "learning_rate": 4.279977125815255e-06,
+      "loss": 0.8986,
+      "step": 99470
+    },
+    {
+      "epoch": 0.72010249951139,
+      "grad_norm": 0.15072764456272125,
+      "learning_rate": 4.279904739154669e-06,
+      "loss": 0.8776,
+      "step": 99480
+    },
+    {
+      "epoch": 0.7201748861719762,
+      "grad_norm": 0.16070221364498138,
+      "learning_rate": 4.2798323524940825e-06,
+      "loss": 0.8977,
+      "step": 99490
+    },
+    {
+      "epoch": 0.7202472728325624,
+      "grad_norm": 0.156993567943573,
+      "learning_rate": 4.279759965833496e-06,
+      "loss": 0.8952,
+      "step": 99500
+    },
+    {
+      "epoch": 0.7203196594931486,
+      "grad_norm": 0.1609060913324356,
+      "learning_rate": 4.2796875791729106e-06,
+      "loss": 0.896,
+      "step": 99510
+    },
+    {
+      "epoch": 0.7203920461537348,
+      "grad_norm": 0.15467387437820435,
+      "learning_rate": 4.279615192512324e-06,
+      "loss": 0.8862,
+      "step": 99520
+    },
+    {
+      "epoch": 0.7204644328143209,
+      "grad_norm": 0.2120065689086914,
+      "learning_rate": 4.279542805851738e-06,
+      "loss": 0.8828,
+      "step": 99530
+    },
+    {
+      "epoch": 0.7205368194749071,
+      "grad_norm": 0.16634128987789154,
+      "learning_rate": 4.279470419191151e-06,
+      "loss": 0.8791,
+      "step": 99540
+    },
+    {
+      "epoch": 0.7206092061354934,
+      "grad_norm": 0.18684551119804382,
+      "learning_rate": 4.279398032530566e-06,
+      "loss": 0.8796,
+      "step": 99550
+    },
+    {
+      "epoch": 0.7206815927960796,
+      "grad_norm": 0.17714783549308777,
+      "learning_rate": 4.2793256458699795e-06,
+      "loss": 0.8972,
+      "step": 99560
+    },
+    {
+      "epoch": 0.7207539794566657,
+      "grad_norm": 0.14424735307693481,
+      "learning_rate": 4.279253259209393e-06,
+      "loss": 0.8898,
+      "step": 99570
+    },
+    {
+      "epoch": 0.7208263661172519,
+      "grad_norm": 0.1793050915002823,
+      "learning_rate": 4.279180872548807e-06,
+      "loss": 0.8837,
+      "step": 99580
+    },
+    {
+      "epoch": 0.7208987527778381,
+      "grad_norm": 0.15566468238830566,
+      "learning_rate": 4.279108485888221e-06,
+      "loss": 0.8823,
+      "step": 99590
+    },
+    {
+      "epoch": 0.7209711394384243,
+      "grad_norm": 0.14767152070999146,
+      "learning_rate": 4.279036099227635e-06,
+      "loss": 0.8748,
+      "step": 99600
+    },
+    {
+      "epoch": 0.7210435260990105,
+      "grad_norm": 0.1632547825574875,
+      "learning_rate": 4.2789637125670484e-06,
+      "loss": 0.8904,
+      "step": 99610
+    },
+    {
+      "epoch": 0.7211159127595966,
+      "grad_norm": 0.15623623132705688,
+      "learning_rate": 4.278891325906462e-06,
+      "loss": 0.8852,
+      "step": 99620
+    },
+    {
+      "epoch": 0.7211882994201828,
+      "grad_norm": 0.20870515704154968,
+      "learning_rate": 4.2788189392458765e-06,
+      "loss": 0.8955,
+      "step": 99630
+    },
+    {
+      "epoch": 0.721260686080769,
+      "grad_norm": 0.15594063699245453,
+      "learning_rate": 4.27874655258529e-06,
+      "loss": 0.9089,
+      "step": 99640
+    },
+    {
+      "epoch": 0.7213330727413553,
+      "grad_norm": 0.16253912448883057,
+      "learning_rate": 4.278674165924704e-06,
+      "loss": 0.8859,
+      "step": 99650
+    },
+    {
+      "epoch": 0.7214054594019415,
+      "grad_norm": 0.3714236319065094,
+      "learning_rate": 4.278601779264117e-06,
+      "loss": 0.8875,
+      "step": 99660
+    },
+    {
+      "epoch": 0.7214778460625276,
+      "grad_norm": 0.14919225871562958,
+      "learning_rate": 4.278529392603532e-06,
+      "loss": 0.8868,
+      "step": 99670
+    },
+    {
+      "epoch": 0.7215502327231138,
+      "grad_norm": 0.16863149404525757,
+      "learning_rate": 4.2784570059429454e-06,
+      "loss": 0.8998,
+      "step": 99680
+    },
+    {
+      "epoch": 0.7216226193837,
+      "grad_norm": 0.15315097570419312,
+      "learning_rate": 4.278384619282359e-06,
+      "loss": 0.8859,
+      "step": 99690
+    },
+    {
+      "epoch": 0.7216950060442862,
+      "grad_norm": 0.14236433804035187,
+      "learning_rate": 4.278312232621773e-06,
+      "loss": 0.8921,
+      "step": 99700
+    },
+    {
+      "epoch": 0.7217673927048723,
+      "grad_norm": 0.16768266260623932,
+      "learning_rate": 4.278239845961187e-06,
+      "loss": 0.8741,
+      "step": 99710
+    },
+    {
+      "epoch": 0.7218397793654585,
+      "grad_norm": 0.1807376593351364,
+      "learning_rate": 4.278167459300601e-06,
+      "loss": 0.888,
+      "step": 99720
+    },
+    {
+      "epoch": 0.7219121660260447,
+      "grad_norm": 0.16242581605911255,
+      "learning_rate": 4.278095072640014e-06,
+      "loss": 0.8871,
+      "step": 99730
+    },
+    {
+      "epoch": 0.7219845526866309,
+      "grad_norm": 0.1657247543334961,
+      "learning_rate": 4.278022685979428e-06,
+      "loss": 0.8905,
+      "step": 99740
+    },
+    {
+      "epoch": 0.722056939347217,
+      "grad_norm": 0.1562010794878006,
+      "learning_rate": 4.2779502993188424e-06,
+      "loss": 0.865,
+      "step": 99750
+    },
+    {
+      "epoch": 0.7221293260078033,
+      "grad_norm": 0.1501951813697815,
+      "learning_rate": 4.277877912658256e-06,
+      "loss": 0.8991,
+      "step": 99760
+    },
+    {
+      "epoch": 0.7222017126683895,
+      "grad_norm": 0.15544991195201874,
+      "learning_rate": 4.27780552599767e-06,
+      "loss": 0.8812,
+      "step": 99770
+    },
+    {
+      "epoch": 0.7222740993289757,
+      "grad_norm": 0.17391003668308258,
+      "learning_rate": 4.277733139337083e-06,
+      "loss": 0.8923,
+      "step": 99780
+    },
+    {
+      "epoch": 0.7223464859895619,
+      "grad_norm": 0.16744978725910187,
+      "learning_rate": 4.277660752676498e-06,
+      "loss": 0.8973,
+      "step": 99790
+    },
+    {
+      "epoch": 0.722418872650148,
+      "grad_norm": 0.1574922800064087,
+      "learning_rate": 4.277588366015911e-06,
+      "loss": 0.8932,
+      "step": 99800
+    },
+    {
+      "epoch": 0.7224912593107342,
+      "grad_norm": 0.14947380125522614,
+      "learning_rate": 4.277515979355324e-06,
+      "loss": 0.8859,
+      "step": 99810
+    },
+    {
+      "epoch": 0.7225636459713204,
+      "grad_norm": 0.16751673817634583,
+      "learning_rate": 4.277443592694739e-06,
+      "loss": 0.8879,
+      "step": 99820
+    },
+    {
+      "epoch": 0.7226360326319066,
+      "grad_norm": 0.1539609283208847,
+      "learning_rate": 4.277371206034152e-06,
+      "loss": 0.8991,
+      "step": 99830
+    },
+    {
+      "epoch": 0.7227084192924927,
+      "grad_norm": 0.19024856388568878,
+      "learning_rate": 4.277298819373566e-06,
+      "loss": 0.8886,
+      "step": 99840
+    },
+    {
+      "epoch": 0.7227808059530789,
+      "grad_norm": 0.1468716859817505,
+      "learning_rate": 4.2772264327129795e-06,
+      "loss": 0.8967,
+      "step": 99850
+    },
+    {
+      "epoch": 0.7228531926136652,
+      "grad_norm": 0.15970027446746826,
+      "learning_rate": 4.277154046052394e-06,
+      "loss": 0.8892,
+      "step": 99860
+    },
+    {
+      "epoch": 0.7229255792742514,
+      "grad_norm": 0.1565774530172348,
+      "learning_rate": 4.2770816593918075e-06,
+      "loss": 0.8853,
+      "step": 99870
+    },
+    {
+      "epoch": 0.7229979659348376,
+      "grad_norm": 0.19593319296836853,
+      "learning_rate": 4.277009272731221e-06,
+      "loss": 0.8866,
+      "step": 99880
+    },
+    {
+      "epoch": 0.7230703525954237,
+      "grad_norm": 0.14920704066753387,
+      "learning_rate": 4.276936886070635e-06,
+      "loss": 0.8842,
+      "step": 99890
+    },
+    {
+      "epoch": 0.7231427392560099,
+      "grad_norm": 0.18510152399539948,
+      "learning_rate": 4.276864499410049e-06,
+      "loss": 0.8763,
+      "step": 99900
+    },
+    {
+      "epoch": 0.7232151259165961,
+      "grad_norm": 0.16031725704669952,
+      "learning_rate": 4.276792112749463e-06,
+      "loss": 0.8836,
+      "step": 99910
+    },
+    {
+      "epoch": 0.7232875125771823,
+      "grad_norm": 0.18583592772483826,
+      "learning_rate": 4.2767197260888765e-06,
+      "loss": 0.8894,
+      "step": 99920
+    },
+    {
+      "epoch": 0.7233598992377684,
+      "grad_norm": 0.15419869124889374,
+      "learning_rate": 4.27664733942829e-06,
+      "loss": 0.8858,
+      "step": 99930
+    },
+    {
+      "epoch": 0.7234322858983546,
+      "grad_norm": 0.15521828830242157,
+      "learning_rate": 4.2765749527677045e-06,
+      "loss": 0.893,
+      "step": 99940
+    },
+    {
+      "epoch": 0.7235046725589408,
+      "grad_norm": 0.15606524050235748,
+      "learning_rate": 4.276502566107118e-06,
+      "loss": 0.9011,
+      "step": 99950
+    },
+    {
+      "epoch": 0.723577059219527,
+      "grad_norm": 0.15376631915569305,
+      "learning_rate": 4.276430179446532e-06,
+      "loss": 0.8898,
+      "step": 99960
+    },
+    {
+      "epoch": 0.7236494458801133,
+      "grad_norm": 0.14439062774181366,
+      "learning_rate": 4.276357792785945e-06,
+      "loss": 0.8911,
+      "step": 99970
+    },
+    {
+      "epoch": 0.7237218325406994,
+      "grad_norm": 0.1527758240699768,
+      "learning_rate": 4.27628540612536e-06,
+      "loss": 0.8911,
+      "step": 99980
+    },
+    {
+      "epoch": 0.7237942192012856,
+      "grad_norm": 0.15964572131633759,
+      "learning_rate": 4.2762130194647735e-06,
+      "loss": 0.8951,
+      "step": 99990
+    },
+    {
+      "epoch": 0.7238666058618718,
+      "grad_norm": 0.15122631192207336,
+      "learning_rate": 4.276140632804187e-06,
+      "loss": 0.8813,
+      "step": 100000
+    },
+    {
+      "epoch": 0.723938992522458,
+      "grad_norm": 0.16392894089221954,
+      "learning_rate": 4.276068246143601e-06,
+      "loss": 0.8888,
+      "step": 100010
+    },
+    {
+      "epoch": 0.7240113791830441,
+      "grad_norm": 0.1608874499797821,
+      "learning_rate": 4.275995859483015e-06,
+      "loss": 0.8825,
+      "step": 100020
+    },
+    {
+      "epoch": 0.7240837658436303,
+      "grad_norm": 0.16042576730251312,
+      "learning_rate": 4.275923472822429e-06,
+      "loss": 0.8939,
+      "step": 100030
+    },
+    {
+      "epoch": 0.7241561525042165,
+      "grad_norm": 0.15833930671215057,
+      "learning_rate": 4.275851086161842e-06,
+      "loss": 0.8763,
+      "step": 100040
+    },
+    {
+      "epoch": 0.7242285391648027,
+      "grad_norm": 0.1503579467535019,
+      "learning_rate": 4.275778699501256e-06,
+      "loss": 0.8852,
+      "step": 100050
+    },
+    {
+      "epoch": 0.7243009258253889,
+      "grad_norm": 0.1671973466873169,
+      "learning_rate": 4.2757063128406705e-06,
+      "loss": 0.8856,
+      "step": 100060
+    },
+    {
+      "epoch": 0.724373312485975,
+      "grad_norm": 0.14335432648658752,
+      "learning_rate": 4.275633926180084e-06,
+      "loss": 0.8775,
+      "step": 100070
+    },
+    {
+      "epoch": 0.7244456991465613,
+      "grad_norm": 0.1621866524219513,
+      "learning_rate": 4.275561539519498e-06,
+      "loss": 0.8868,
+      "step": 100080
+    },
+    {
+      "epoch": 0.7245180858071475,
+      "grad_norm": 0.1560746431350708,
+      "learning_rate": 4.275489152858911e-06,
+      "loss": 0.8857,
+      "step": 100090
+    },
+    {
+      "epoch": 0.7245904724677337,
+      "grad_norm": 0.1637454777956009,
+      "learning_rate": 4.275416766198325e-06,
+      "loss": 0.8863,
+      "step": 100100
+    },
+    {
+      "epoch": 0.7246628591283198,
+      "grad_norm": 0.16514307260513306,
+      "learning_rate": 4.275344379537739e-06,
+      "loss": 0.893,
+      "step": 100110
+    },
+    {
+      "epoch": 0.724735245788906,
+      "grad_norm": 0.15878024697303772,
+      "learning_rate": 4.275271992877153e-06,
+      "loss": 0.8852,
+      "step": 100120
+    },
+    {
+      "epoch": 0.7248076324494922,
+      "grad_norm": 0.15798494219779968,
+      "learning_rate": 4.275199606216567e-06,
+      "loss": 0.8922,
+      "step": 100130
+    },
+    {
+      "epoch": 0.7248800191100784,
+      "grad_norm": 0.1547410637140274,
+      "learning_rate": 4.27512721955598e-06,
+      "loss": 0.8779,
+      "step": 100140
+    },
+    {
+      "epoch": 0.7249524057706646,
+      "grad_norm": 0.14897017180919647,
+      "learning_rate": 4.275054832895395e-06,
+      "loss": 0.8924,
+      "step": 100150
+    },
+    {
+      "epoch": 0.7250247924312507,
+      "grad_norm": 0.1668456345796585,
+      "learning_rate": 4.274982446234808e-06,
+      "loss": 0.8803,
+      "step": 100160
+    },
+    {
+      "epoch": 0.7250971790918369,
+      "grad_norm": 0.1529316008090973,
+      "learning_rate": 4.274910059574222e-06,
+      "loss": 0.8984,
+      "step": 100170
+    },
+    {
+      "epoch": 0.7251695657524232,
+      "grad_norm": 0.14754995703697205,
+      "learning_rate": 4.2748376729136356e-06,
+      "loss": 0.8735,
+      "step": 100180
+    },
+    {
+      "epoch": 0.7252419524130094,
+      "grad_norm": 0.16086558997631073,
+      "learning_rate": 4.27476528625305e-06,
+      "loss": 0.8852,
+      "step": 100190
+    },
+    {
+      "epoch": 0.7253143390735955,
+      "grad_norm": 0.16596639156341553,
+      "learning_rate": 4.274692899592464e-06,
+      "loss": 0.9067,
+      "step": 100200
+    },
+    {
+      "epoch": 0.7253867257341817,
+      "grad_norm": 0.18739081919193268,
+      "learning_rate": 4.274620512931877e-06,
+      "loss": 0.8872,
+      "step": 100210
+    },
+    {
+      "epoch": 0.7254591123947679,
+      "grad_norm": 0.15418687462806702,
+      "learning_rate": 4.274548126271291e-06,
+      "loss": 0.8934,
+      "step": 100220
+    },
+    {
+      "epoch": 0.7255314990553541,
+      "grad_norm": 0.16099806129932404,
+      "learning_rate": 4.274475739610705e-06,
+      "loss": 0.8943,
+      "step": 100230
+    },
+    {
+      "epoch": 0.7256038857159403,
+      "grad_norm": 0.15143683552742004,
+      "learning_rate": 4.274403352950119e-06,
+      "loss": 0.8825,
+      "step": 100240
+    },
+    {
+      "epoch": 0.7256762723765264,
+      "grad_norm": 0.15264058113098145,
+      "learning_rate": 4.2743309662895326e-06,
+      "loss": 0.8817,
+      "step": 100250
+    },
+    {
+      "epoch": 0.7257486590371126,
+      "grad_norm": 0.1594517081975937,
+      "learning_rate": 4.274258579628946e-06,
+      "loss": 0.8968,
+      "step": 100260
+    },
+    {
+      "epoch": 0.7258210456976988,
+      "grad_norm": 0.1539693921804428,
+      "learning_rate": 4.274186192968361e-06,
+      "loss": 0.8994,
+      "step": 100270
+    },
+    {
+      "epoch": 0.725893432358285,
+      "grad_norm": 0.1556989848613739,
+      "learning_rate": 4.274113806307774e-06,
+      "loss": 0.8952,
+      "step": 100280
+    },
+    {
+      "epoch": 0.7259658190188713,
+      "grad_norm": 0.15270110964775085,
+      "learning_rate": 4.274041419647188e-06,
+      "loss": 0.8867,
+      "step": 100290
+    },
+    {
+      "epoch": 0.7260382056794574,
+      "grad_norm": 0.16615337133407593,
+      "learning_rate": 4.2739690329866015e-06,
+      "loss": 0.881,
+      "step": 100300
+    },
+    {
+      "epoch": 0.7261105923400436,
+      "grad_norm": 0.15638002753257751,
+      "learning_rate": 4.273896646326016e-06,
+      "loss": 0.8889,
+      "step": 100310
+    },
+    {
+      "epoch": 0.7261829790006298,
+      "grad_norm": 0.1828213930130005,
+      "learning_rate": 4.27382425966543e-06,
+      "loss": 0.8921,
+      "step": 100320
+    },
+    {
+      "epoch": 0.726255365661216,
+      "grad_norm": 0.14612999558448792,
+      "learning_rate": 4.273751873004843e-06,
+      "loss": 0.889,
+      "step": 100330
+    },
+    {
+      "epoch": 0.7263277523218021,
+      "grad_norm": 0.15252217650413513,
+      "learning_rate": 4.273679486344257e-06,
+      "loss": 0.8879,
+      "step": 100340
+    },
+    {
+      "epoch": 0.7264001389823883,
+      "grad_norm": 0.1537008434534073,
+      "learning_rate": 4.2736070996836704e-06,
+      "loss": 0.899,
+      "step": 100350
+    },
+    {
+      "epoch": 0.7264725256429745,
+      "grad_norm": 0.1628774255514145,
+      "learning_rate": 4.273534713023084e-06,
+      "loss": 0.8934,
+      "step": 100360
+    },
+    {
+      "epoch": 0.7265449123035607,
+      "grad_norm": 0.1571381837129593,
+      "learning_rate": 4.273462326362498e-06,
+      "loss": 0.8825,
+      "step": 100370
+    },
+    {
+      "epoch": 0.7266172989641468,
+      "grad_norm": 0.16100090742111206,
+      "learning_rate": 4.273389939701912e-06,
+      "loss": 0.8901,
+      "step": 100380
+    },
+    {
+      "epoch": 0.7266896856247331,
+      "grad_norm": 0.14620473980903625,
+      "learning_rate": 4.273317553041326e-06,
+      "loss": 0.8886,
+      "step": 100390
+    },
+    {
+      "epoch": 0.7267620722853193,
+      "grad_norm": 0.15393929183483124,
+      "learning_rate": 4.273245166380739e-06,
+      "loss": 0.8974,
+      "step": 100400
+    },
+    {
+      "epoch": 0.7268344589459055,
+      "grad_norm": 0.16305720806121826,
+      "learning_rate": 4.273172779720153e-06,
+      "loss": 0.884,
+      "step": 100410
+    },
+    {
+      "epoch": 0.7269068456064917,
+      "grad_norm": 0.1579166203737259,
+      "learning_rate": 4.2731003930595674e-06,
+      "loss": 0.8762,
+      "step": 100420
+    },
+    {
+      "epoch": 0.7269792322670778,
+      "grad_norm": 0.14451733231544495,
+      "learning_rate": 4.273028006398981e-06,
+      "loss": 0.8893,
+      "step": 100430
+    },
+    {
+      "epoch": 0.727051618927664,
+      "grad_norm": 0.14748463034629822,
+      "learning_rate": 4.272955619738395e-06,
+      "loss": 0.8861,
+      "step": 100440
+    },
+    {
+      "epoch": 0.7271240055882502,
+      "grad_norm": 0.16086864471435547,
+      "learning_rate": 4.272883233077808e-06,
+      "loss": 0.8885,
+      "step": 100450
+    },
+    {
+      "epoch": 0.7271963922488364,
+      "grad_norm": 0.14903460443019867,
+      "learning_rate": 4.272810846417223e-06,
+      "loss": 0.8832,
+      "step": 100460
+    },
+    {
+      "epoch": 0.7272687789094225,
+      "grad_norm": 0.16146931052207947,
+      "learning_rate": 4.272738459756636e-06,
+      "loss": 0.8856,
+      "step": 100470
+    },
+    {
+      "epoch": 0.7273411655700087,
+      "grad_norm": 0.1918579488992691,
+      "learning_rate": 4.27266607309605e-06,
+      "loss": 0.8953,
+      "step": 100480
+    },
+    {
+      "epoch": 0.7274135522305949,
+      "grad_norm": 0.1559261530637741,
+      "learning_rate": 4.272593686435464e-06,
+      "loss": 0.8855,
+      "step": 100490
+    },
+    {
+      "epoch": 0.7274859388911812,
+      "grad_norm": 0.15730531513690948,
+      "learning_rate": 4.272521299774878e-06,
+      "loss": 0.8946,
+      "step": 100500
+    },
+    {
+      "epoch": 0.7275583255517674,
+      "grad_norm": 0.1542884111404419,
+      "learning_rate": 4.272448913114292e-06,
+      "loss": 0.8932,
+      "step": 100510
+    },
+    {
+      "epoch": 0.7276307122123535,
+      "grad_norm": 0.15835349261760712,
+      "learning_rate": 4.272376526453705e-06,
+      "loss": 0.903,
+      "step": 100520
+    },
+    {
+      "epoch": 0.7277030988729397,
+      "grad_norm": 0.1627766340970993,
+      "learning_rate": 4.272304139793119e-06,
+      "loss": 0.8901,
+      "step": 100530
+    },
+    {
+      "epoch": 0.7277754855335259,
+      "grad_norm": 0.15794923901557922,
+      "learning_rate": 4.272231753132533e-06,
+      "loss": 0.8939,
+      "step": 100540
+    },
+    {
+      "epoch": 0.7278478721941121,
+      "grad_norm": 0.15838219225406647,
+      "learning_rate": 4.272159366471947e-06,
+      "loss": 0.8933,
+      "step": 100550
+    },
+    {
+      "epoch": 0.7279202588546982,
+      "grad_norm": 0.15067455172538757,
+      "learning_rate": 4.272086979811361e-06,
+      "loss": 0.8892,
+      "step": 100560
+    },
+    {
+      "epoch": 0.7279926455152844,
+      "grad_norm": 0.158193901181221,
+      "learning_rate": 4.272014593150774e-06,
+      "loss": 0.8793,
+      "step": 100570
+    },
+    {
+      "epoch": 0.7280650321758706,
+      "grad_norm": 0.15686438977718353,
+      "learning_rate": 4.271942206490189e-06,
+      "loss": 0.8775,
+      "step": 100580
+    },
+    {
+      "epoch": 0.7281374188364568,
+      "grad_norm": 0.1513579934835434,
+      "learning_rate": 4.271869819829602e-06,
+      "loss": 0.8954,
+      "step": 100590
+    },
+    {
+      "epoch": 0.728209805497043,
+      "grad_norm": 0.15484221279621124,
+      "learning_rate": 4.271797433169016e-06,
+      "loss": 0.8944,
+      "step": 100600
+    },
+    {
+      "epoch": 0.7282821921576292,
+      "grad_norm": 0.14668524265289307,
+      "learning_rate": 4.2717250465084295e-06,
+      "loss": 0.8753,
+      "step": 100610
+    },
+    {
+      "epoch": 0.7283545788182154,
+      "grad_norm": 0.1556718647480011,
+      "learning_rate": 4.271652659847844e-06,
+      "loss": 0.8875,
+      "step": 100620
+    },
+    {
+      "epoch": 0.7284269654788016,
+      "grad_norm": 0.16115809977054596,
+      "learning_rate": 4.271580273187258e-06,
+      "loss": 0.8909,
+      "step": 100630
+    },
+    {
+      "epoch": 0.7284993521393878,
+      "grad_norm": 0.152574360370636,
+      "learning_rate": 4.271507886526671e-06,
+      "loss": 0.8916,
+      "step": 100640
+    },
+    {
+      "epoch": 0.728571738799974,
+      "grad_norm": 0.16087447106838226,
+      "learning_rate": 4.271435499866085e-06,
+      "loss": 0.8859,
+      "step": 100650
+    },
+    {
+      "epoch": 0.7286441254605601,
+      "grad_norm": 0.21453756093978882,
+      "learning_rate": 4.271363113205499e-06,
+      "loss": 0.8862,
+      "step": 100660
+    },
+    {
+      "epoch": 0.7287165121211463,
+      "grad_norm": 0.156850203871727,
+      "learning_rate": 4.271290726544913e-06,
+      "loss": 0.8844,
+      "step": 100670
+    },
+    {
+      "epoch": 0.7287888987817325,
+      "grad_norm": 0.16212424635887146,
+      "learning_rate": 4.2712183398843265e-06,
+      "loss": 0.8892,
+      "step": 100680
+    },
+    {
+      "epoch": 0.7288612854423187,
+      "grad_norm": 0.1557214856147766,
+      "learning_rate": 4.27114595322374e-06,
+      "loss": 0.8843,
+      "step": 100690
+    },
+    {
+      "epoch": 0.7289336721029048,
+      "grad_norm": 0.1972065418958664,
+      "learning_rate": 4.271073566563154e-06,
+      "loss": 0.8747,
+      "step": 100700
+    },
+    {
+      "epoch": 0.7290060587634911,
+      "grad_norm": 0.1524554044008255,
+      "learning_rate": 4.271001179902568e-06,
+      "loss": 0.8857,
+      "step": 100710
+    },
+    {
+      "epoch": 0.7290784454240773,
+      "grad_norm": 0.15759634971618652,
+      "learning_rate": 4.270928793241982e-06,
+      "loss": 0.8856,
+      "step": 100720
+    },
+    {
+      "epoch": 0.7291508320846635,
+      "grad_norm": 0.1459013819694519,
+      "learning_rate": 4.2708564065813955e-06,
+      "loss": 0.8737,
+      "step": 100730
+    },
+    {
+      "epoch": 0.7292232187452496,
+      "grad_norm": 0.15204448997974396,
+      "learning_rate": 4.270784019920809e-06,
+      "loss": 0.8969,
+      "step": 100740
+    },
+    {
+      "epoch": 0.7292956054058358,
+      "grad_norm": 0.16425342857837677,
+      "learning_rate": 4.2707116332602235e-06,
+      "loss": 0.8876,
+      "step": 100750
+    },
+    {
+      "epoch": 0.729367992066422,
+      "grad_norm": 0.16584093868732452,
+      "learning_rate": 4.270639246599637e-06,
+      "loss": 0.885,
+      "step": 100760
+    },
+    {
+      "epoch": 0.7294403787270082,
+      "grad_norm": 0.16400320827960968,
+      "learning_rate": 4.270566859939051e-06,
+      "loss": 0.8939,
+      "step": 100770
+    },
+    {
+      "epoch": 0.7295127653875944,
+      "grad_norm": 0.15366198122501373,
+      "learning_rate": 4.270494473278464e-06,
+      "loss": 0.8716,
+      "step": 100780
+    },
+    {
+      "epoch": 0.7295851520481805,
+      "grad_norm": 0.15150944888591766,
+      "learning_rate": 4.270422086617879e-06,
+      "loss": 0.89,
+      "step": 100790
+    },
+    {
+      "epoch": 0.7296575387087667,
+      "grad_norm": 0.1443614810705185,
+      "learning_rate": 4.2703496999572925e-06,
+      "loss": 0.8889,
+      "step": 100800
+    },
+    {
+      "epoch": 0.7297299253693529,
+      "grad_norm": 0.15036936104297638,
+      "learning_rate": 4.270277313296706e-06,
+      "loss": 0.9045,
+      "step": 100810
+    },
+    {
+      "epoch": 0.7298023120299392,
+      "grad_norm": 0.16487093269824982,
+      "learning_rate": 4.27020492663612e-06,
+      "loss": 0.8815,
+      "step": 100820
+    },
+    {
+      "epoch": 0.7298746986905253,
+      "grad_norm": 0.15365450084209442,
+      "learning_rate": 4.270132539975534e-06,
+      "loss": 0.894,
+      "step": 100830
+    },
+    {
+      "epoch": 0.7299470853511115,
+      "grad_norm": 0.1608169972896576,
+      "learning_rate": 4.270060153314948e-06,
+      "loss": 0.9013,
+      "step": 100840
+    },
+    {
+      "epoch": 0.7300194720116977,
+      "grad_norm": 0.16800221800804138,
+      "learning_rate": 4.269987766654361e-06,
+      "loss": 0.8778,
+      "step": 100850
+    },
+    {
+      "epoch": 0.7300918586722839,
+      "grad_norm": 0.15907716751098633,
+      "learning_rate": 4.269915379993775e-06,
+      "loss": 0.8859,
+      "step": 100860
+    },
+    {
+      "epoch": 0.73016424533287,
+      "grad_norm": 0.15400275588035583,
+      "learning_rate": 4.269842993333189e-06,
+      "loss": 0.8904,
+      "step": 100870
+    },
+    {
+      "epoch": 0.7302366319934562,
+      "grad_norm": 0.1577254682779312,
+      "learning_rate": 4.269770606672602e-06,
+      "loss": 0.8833,
+      "step": 100880
+    },
+    {
+      "epoch": 0.7303090186540424,
+      "grad_norm": 0.1583620309829712,
+      "learning_rate": 4.269698220012016e-06,
+      "loss": 0.8808,
+      "step": 100890
+    },
+    {
+      "epoch": 0.7303814053146286,
+      "grad_norm": 0.1512048840522766,
+      "learning_rate": 4.26962583335143e-06,
+      "loss": 0.892,
+      "step": 100900
+    },
+    {
+      "epoch": 0.7304537919752148,
+      "grad_norm": 0.1588258147239685,
+      "learning_rate": 4.269553446690844e-06,
+      "loss": 0.8959,
+      "step": 100910
+    },
+    {
+      "epoch": 0.730526178635801,
+      "grad_norm": 0.15298683941364288,
+      "learning_rate": 4.2694810600302576e-06,
+      "loss": 0.885,
+      "step": 100920
+    },
+    {
+      "epoch": 0.7305985652963872,
+      "grad_norm": 0.15516585111618042,
+      "learning_rate": 4.269408673369671e-06,
+      "loss": 0.8925,
+      "step": 100930
+    },
+    {
+      "epoch": 0.7306709519569734,
+      "grad_norm": 0.1673891842365265,
+      "learning_rate": 4.269336286709086e-06,
+      "loss": 0.9016,
+      "step": 100940
+    },
+    {
+      "epoch": 0.7307433386175596,
+      "grad_norm": 0.16964946687221527,
+      "learning_rate": 4.269263900048499e-06,
+      "loss": 0.9035,
+      "step": 100950
+    },
+    {
+      "epoch": 0.7308157252781458,
+      "grad_norm": 0.21331481635570526,
+      "learning_rate": 4.269191513387913e-06,
+      "loss": 0.8832,
+      "step": 100960
+    },
+    {
+      "epoch": 0.7308881119387319,
+      "grad_norm": 0.1504168063402176,
+      "learning_rate": 4.2691191267273265e-06,
+      "loss": 0.8759,
+      "step": 100970
+    },
+    {
+      "epoch": 0.7309604985993181,
+      "grad_norm": 0.14905038475990295,
+      "learning_rate": 4.269046740066741e-06,
+      "loss": 0.887,
+      "step": 100980
+    },
+    {
+      "epoch": 0.7310328852599043,
+      "grad_norm": 0.1641206294298172,
+      "learning_rate": 4.2689743534061546e-06,
+      "loss": 0.8813,
+      "step": 100990
+    },
+    {
+      "epoch": 0.7311052719204905,
+      "grad_norm": 0.3293008804321289,
+      "learning_rate": 4.268901966745568e-06,
+      "loss": 0.874,
+      "step": 101000
+    },
+    {
+      "epoch": 0.7311776585810766,
+      "grad_norm": 0.1844899207353592,
+      "learning_rate": 4.268829580084982e-06,
+      "loss": 0.8987,
+      "step": 101010
+    },
+    {
+      "epoch": 0.7312500452416628,
+      "grad_norm": 0.17655082046985626,
+      "learning_rate": 4.268757193424396e-06,
+      "loss": 0.8979,
+      "step": 101020
+    },
+    {
+      "epoch": 0.7313224319022491,
+      "grad_norm": 0.15127238631248474,
+      "learning_rate": 4.26868480676381e-06,
+      "loss": 0.8825,
+      "step": 101030
+    },
+    {
+      "epoch": 0.7313948185628353,
+      "grad_norm": 0.15606975555419922,
+      "learning_rate": 4.2686124201032235e-06,
+      "loss": 0.8874,
+      "step": 101040
+    },
+    {
+      "epoch": 0.7314672052234215,
+      "grad_norm": 0.1606336385011673,
+      "learning_rate": 4.268540033442637e-06,
+      "loss": 0.8774,
+      "step": 101050
+    },
+    {
+      "epoch": 0.7315395918840076,
+      "grad_norm": 0.16057442128658295,
+      "learning_rate": 4.268467646782052e-06,
+      "loss": 0.8981,
+      "step": 101060
+    },
+    {
+      "epoch": 0.7316119785445938,
+      "grad_norm": 0.1635402888059616,
+      "learning_rate": 4.268395260121465e-06,
+      "loss": 0.8952,
+      "step": 101070
+    },
+    {
+      "epoch": 0.73168436520518,
+      "grad_norm": 0.1442507654428482,
+      "learning_rate": 4.268322873460879e-06,
+      "loss": 0.8981,
+      "step": 101080
+    },
+    {
+      "epoch": 0.7317567518657662,
+      "grad_norm": 0.13882668316364288,
+      "learning_rate": 4.2682504868002924e-06,
+      "loss": 0.8905,
+      "step": 101090
+    },
+    {
+      "epoch": 0.7318291385263523,
+      "grad_norm": 0.15066657960414886,
+      "learning_rate": 4.268178100139707e-06,
+      "loss": 0.8987,
+      "step": 101100
+    },
+    {
+      "epoch": 0.7319015251869385,
+      "grad_norm": 0.1645701825618744,
+      "learning_rate": 4.2681057134791205e-06,
+      "loss": 0.8886,
+      "step": 101110
+    },
+    {
+      "epoch": 0.7319739118475247,
+      "grad_norm": 0.15226510167121887,
+      "learning_rate": 4.268033326818534e-06,
+      "loss": 0.8932,
+      "step": 101120
+    },
+    {
+      "epoch": 0.7320462985081109,
+      "grad_norm": 0.15741746127605438,
+      "learning_rate": 4.267960940157948e-06,
+      "loss": 0.8793,
+      "step": 101130
+    },
+    {
+      "epoch": 0.7321186851686972,
+      "grad_norm": 0.15548457205295563,
+      "learning_rate": 4.267888553497362e-06,
+      "loss": 0.881,
+      "step": 101140
+    },
+    {
+      "epoch": 0.7321910718292833,
+      "grad_norm": 0.2032490223646164,
+      "learning_rate": 4.267816166836776e-06,
+      "loss": 0.8916,
+      "step": 101150
+    },
+    {
+      "epoch": 0.7322634584898695,
+      "grad_norm": 0.2026352435350418,
+      "learning_rate": 4.2677437801761894e-06,
+      "loss": 0.8932,
+      "step": 101160
+    },
+    {
+      "epoch": 0.7323358451504557,
+      "grad_norm": 0.15487812459468842,
+      "learning_rate": 4.267671393515603e-06,
+      "loss": 0.8805,
+      "step": 101170
+    },
+    {
+      "epoch": 0.7324082318110419,
+      "grad_norm": 0.16896316409111023,
+      "learning_rate": 4.2675990068550175e-06,
+      "loss": 0.8811,
+      "step": 101180
+    },
+    {
+      "epoch": 0.732480618471628,
+      "grad_norm": 0.20788662135601044,
+      "learning_rate": 4.267526620194431e-06,
+      "loss": 0.8858,
+      "step": 101190
+    },
+    {
+      "epoch": 0.7325530051322142,
+      "grad_norm": 0.15774105489253998,
+      "learning_rate": 4.267454233533845e-06,
+      "loss": 0.9004,
+      "step": 101200
+    },
+    {
+      "epoch": 0.7326253917928004,
+      "grad_norm": 0.1576681286096573,
+      "learning_rate": 4.267381846873258e-06,
+      "loss": 0.9003,
+      "step": 101210
+    },
+    {
+      "epoch": 0.7326977784533866,
+      "grad_norm": 0.19305309653282166,
+      "learning_rate": 4.267309460212673e-06,
+      "loss": 0.894,
+      "step": 101220
+    },
+    {
+      "epoch": 0.7327701651139727,
+      "grad_norm": 0.16262653470039368,
+      "learning_rate": 4.2672370735520864e-06,
+      "loss": 0.8774,
+      "step": 101230
+    },
+    {
+      "epoch": 0.732842551774559,
+      "grad_norm": 0.14455868303775787,
+      "learning_rate": 4.2671646868915e-06,
+      "loss": 0.8963,
+      "step": 101240
+    },
+    {
+      "epoch": 0.7329149384351452,
+      "grad_norm": 0.19319617748260498,
+      "learning_rate": 4.267092300230914e-06,
+      "loss": 0.8832,
+      "step": 101250
+    },
+    {
+      "epoch": 0.7329873250957314,
+      "grad_norm": 0.2210075408220291,
+      "learning_rate": 4.267019913570328e-06,
+      "loss": 0.8884,
+      "step": 101260
+    },
+    {
+      "epoch": 0.7330597117563176,
+      "grad_norm": 0.16288064420223236,
+      "learning_rate": 4.266947526909742e-06,
+      "loss": 0.8762,
+      "step": 101270
+    },
+    {
+      "epoch": 0.7331320984169037,
+      "grad_norm": 0.16447566449642181,
+      "learning_rate": 4.266875140249155e-06,
+      "loss": 0.8895,
+      "step": 101280
+    },
+    {
+      "epoch": 0.7332044850774899,
+      "grad_norm": 0.15744997560977936,
+      "learning_rate": 4.266802753588569e-06,
+      "loss": 0.8948,
+      "step": 101290
+    },
+    {
+      "epoch": 0.7332768717380761,
+      "grad_norm": 0.1907864809036255,
+      "learning_rate": 4.2667303669279835e-06,
+      "loss": 0.8816,
+      "step": 101300
+    },
+    {
+      "epoch": 0.7333492583986623,
+      "grad_norm": 0.15406233072280884,
+      "learning_rate": 4.266657980267397e-06,
+      "loss": 0.8851,
+      "step": 101310
+    },
+    {
+      "epoch": 0.7334216450592485,
+      "grad_norm": 0.16773687303066254,
+      "learning_rate": 4.266585593606811e-06,
+      "loss": 0.8864,
+      "step": 101320
+    },
+    {
+      "epoch": 0.7334940317198346,
+      "grad_norm": 0.15521180629730225,
+      "learning_rate": 4.266513206946224e-06,
+      "loss": 0.8834,
+      "step": 101330
+    },
+    {
+      "epoch": 0.7335664183804208,
+      "grad_norm": 0.15735216438770294,
+      "learning_rate": 4.266440820285638e-06,
+      "loss": 0.8814,
+      "step": 101340
+    },
+    {
+      "epoch": 0.7336388050410071,
+      "grad_norm": 0.14961345493793488,
+      "learning_rate": 4.266368433625052e-06,
+      "loss": 0.8863,
+      "step": 101350
+    },
+    {
+      "epoch": 0.7337111917015933,
+      "grad_norm": 0.15976914763450623,
+      "learning_rate": 4.266296046964466e-06,
+      "loss": 0.8944,
+      "step": 101360
+    },
+    {
+      "epoch": 0.7337835783621794,
+      "grad_norm": 0.16493217647075653,
+      "learning_rate": 4.26622366030388e-06,
+      "loss": 0.8897,
+      "step": 101370
+    },
+    {
+      "epoch": 0.7338559650227656,
+      "grad_norm": 0.15834347903728485,
+      "learning_rate": 4.266151273643293e-06,
+      "loss": 0.8842,
+      "step": 101380
+    },
+    {
+      "epoch": 0.7339283516833518,
+      "grad_norm": 0.14904192090034485,
+      "learning_rate": 4.266078886982708e-06,
+      "loss": 0.8796,
+      "step": 101390
+    },
+    {
+      "epoch": 0.734000738343938,
+      "grad_norm": 0.16918309032917023,
+      "learning_rate": 4.2660065003221205e-06,
+      "loss": 0.8919,
+      "step": 101400
+    },
+    {
+      "epoch": 0.7340731250045242,
+      "grad_norm": 0.14453521370887756,
+      "learning_rate": 4.265934113661535e-06,
+      "loss": 0.888,
+      "step": 101410
+    },
+    {
+      "epoch": 0.7341455116651103,
+      "grad_norm": 0.15393070876598358,
+      "learning_rate": 4.2658617270009485e-06,
+      "loss": 0.8833,
+      "step": 101420
+    },
+    {
+      "epoch": 0.7342178983256965,
+      "grad_norm": 0.1721397191286087,
+      "learning_rate": 4.265789340340362e-06,
+      "loss": 0.8918,
+      "step": 101430
+    },
+    {
+      "epoch": 0.7342902849862827,
+      "grad_norm": 0.15190811455249786,
+      "learning_rate": 4.265716953679776e-06,
+      "loss": 0.8889,
+      "step": 101440
+    },
+    {
+      "epoch": 0.734362671646869,
+      "grad_norm": 0.14311714470386505,
+      "learning_rate": 4.26564456701919e-06,
+      "loss": 0.8907,
+      "step": 101450
+    },
+    {
+      "epoch": 0.7344350583074551,
+      "grad_norm": 0.17075756192207336,
+      "learning_rate": 4.265572180358604e-06,
+      "loss": 0.8994,
+      "step": 101460
+    },
+    {
+      "epoch": 0.7345074449680413,
+      "grad_norm": 0.1581360548734665,
+      "learning_rate": 4.2654997936980175e-06,
+      "loss": 0.8876,
+      "step": 101470
+    },
+    {
+      "epoch": 0.7345798316286275,
+      "grad_norm": 0.15147803723812103,
+      "learning_rate": 4.265427407037431e-06,
+      "loss": 0.9041,
+      "step": 101480
+    },
+    {
+      "epoch": 0.7346522182892137,
+      "grad_norm": 0.161724254488945,
+      "learning_rate": 4.265355020376845e-06,
+      "loss": 0.8929,
+      "step": 101490
+    },
+    {
+      "epoch": 0.7347246049497999,
+      "grad_norm": 0.1595461070537567,
+      "learning_rate": 4.265282633716259e-06,
+      "loss": 0.8882,
+      "step": 101500
+    },
+    {
+      "epoch": 0.734796991610386,
+      "grad_norm": 0.15513767302036285,
+      "learning_rate": 4.265210247055673e-06,
+      "loss": 0.8998,
+      "step": 101510
+    },
+    {
+      "epoch": 0.7348693782709722,
+      "grad_norm": 0.15473343431949615,
+      "learning_rate": 4.265137860395086e-06,
+      "loss": 0.9025,
+      "step": 101520
+    },
+    {
+      "epoch": 0.7349417649315584,
+      "grad_norm": 0.16891096532344818,
+      "learning_rate": 4.2650654737345e-06,
+      "loss": 0.8868,
+      "step": 101530
+    },
+    {
+      "epoch": 0.7350141515921446,
+      "grad_norm": 0.15746982395648956,
+      "learning_rate": 4.2649930870739145e-06,
+      "loss": 0.9042,
+      "step": 101540
+    },
+    {
+      "epoch": 0.7350865382527307,
+      "grad_norm": 0.16814929246902466,
+      "learning_rate": 4.264920700413328e-06,
+      "loss": 0.9029,
+      "step": 101550
+    },
+    {
+      "epoch": 0.735158924913317,
+      "grad_norm": 0.14766864478588104,
+      "learning_rate": 4.264848313752742e-06,
+      "loss": 0.8741,
+      "step": 101560
+    },
+    {
+      "epoch": 0.7352313115739032,
+      "grad_norm": 0.16124971210956573,
+      "learning_rate": 4.264775927092155e-06,
+      "loss": 0.8771,
+      "step": 101570
+    },
+    {
+      "epoch": 0.7353036982344894,
+      "grad_norm": 0.16348862648010254,
+      "learning_rate": 4.26470354043157e-06,
+      "loss": 0.8955,
+      "step": 101580
+    },
+    {
+      "epoch": 0.7353760848950756,
+      "grad_norm": 0.16436149179935455,
+      "learning_rate": 4.264631153770983e-06,
+      "loss": 0.8889,
+      "step": 101590
+    },
+    {
+      "epoch": 0.7354484715556617,
+      "grad_norm": 0.14332278072834015,
+      "learning_rate": 4.264558767110397e-06,
+      "loss": 0.8866,
+      "step": 101600
+    },
+    {
+      "epoch": 0.7355208582162479,
+      "grad_norm": 0.15618641674518585,
+      "learning_rate": 4.264486380449811e-06,
+      "loss": 0.8853,
+      "step": 101610
+    },
+    {
+      "epoch": 0.7355932448768341,
+      "grad_norm": 0.1528848558664322,
+      "learning_rate": 4.264413993789225e-06,
+      "loss": 0.8882,
+      "step": 101620
+    },
+    {
+      "epoch": 0.7356656315374203,
+      "grad_norm": 0.1625969558954239,
+      "learning_rate": 4.264341607128639e-06,
+      "loss": 0.8774,
+      "step": 101630
+    },
+    {
+      "epoch": 0.7357380181980064,
+      "grad_norm": 0.16516810655593872,
+      "learning_rate": 4.264269220468052e-06,
+      "loss": 0.887,
+      "step": 101640
+    },
+    {
+      "epoch": 0.7358104048585926,
+      "grad_norm": 0.1680750846862793,
+      "learning_rate": 4.264196833807466e-06,
+      "loss": 0.887,
+      "step": 101650
+    },
+    {
+      "epoch": 0.7358827915191788,
+      "grad_norm": 0.16871851682662964,
+      "learning_rate": 4.26412444714688e-06,
+      "loss": 0.8886,
+      "step": 101660
+    },
+    {
+      "epoch": 0.7359551781797651,
+      "grad_norm": 0.15388178825378418,
+      "learning_rate": 4.264052060486294e-06,
+      "loss": 0.8906,
+      "step": 101670
+    },
+    {
+      "epoch": 0.7360275648403513,
+      "grad_norm": 0.15053468942642212,
+      "learning_rate": 4.263979673825708e-06,
+      "loss": 0.8882,
+      "step": 101680
+    },
+    {
+      "epoch": 0.7360999515009374,
+      "grad_norm": 0.16372324526309967,
+      "learning_rate": 4.263907287165121e-06,
+      "loss": 0.8683,
+      "step": 101690
+    },
+    {
+      "epoch": 0.7361723381615236,
+      "grad_norm": 0.1610361635684967,
+      "learning_rate": 4.263834900504536e-06,
+      "loss": 0.8854,
+      "step": 101700
+    },
+    {
+      "epoch": 0.7362447248221098,
+      "grad_norm": 0.16573579609394073,
+      "learning_rate": 4.263762513843949e-06,
+      "loss": 0.9022,
+      "step": 101710
+    },
+    {
+      "epoch": 0.736317111482696,
+      "grad_norm": 0.16359181702136993,
+      "learning_rate": 4.263690127183363e-06,
+      "loss": 0.8873,
+      "step": 101720
+    },
+    {
+      "epoch": 0.7363894981432821,
+      "grad_norm": 0.296243280172348,
+      "learning_rate": 4.2636177405227766e-06,
+      "loss": 0.8911,
+      "step": 101730
+    },
+    {
+      "epoch": 0.7364618848038683,
+      "grad_norm": 0.16400732100009918,
+      "learning_rate": 4.263545353862191e-06,
+      "loss": 0.8859,
+      "step": 101740
+    },
+    {
+      "epoch": 0.7365342714644545,
+      "grad_norm": 0.15146301686763763,
+      "learning_rate": 4.263472967201605e-06,
+      "loss": 0.8912,
+      "step": 101750
+    },
+    {
+      "epoch": 0.7366066581250407,
+      "grad_norm": 0.1514797806739807,
+      "learning_rate": 4.263400580541018e-06,
+      "loss": 0.8805,
+      "step": 101760
+    },
+    {
+      "epoch": 0.736679044785627,
+      "grad_norm": 0.15184786915779114,
+      "learning_rate": 4.263328193880432e-06,
+      "loss": 0.8869,
+      "step": 101770
+    },
+    {
+      "epoch": 0.7367514314462131,
+      "grad_norm": 0.1572055220603943,
+      "learning_rate": 4.263255807219846e-06,
+      "loss": 0.8827,
+      "step": 101780
+    },
+    {
+      "epoch": 0.7368238181067993,
+      "grad_norm": 0.16821490228176117,
+      "learning_rate": 4.26318342055926e-06,
+      "loss": 0.9025,
+      "step": 101790
+    },
+    {
+      "epoch": 0.7368962047673855,
+      "grad_norm": 0.14948932826519012,
+      "learning_rate": 4.263111033898674e-06,
+      "loss": 0.8809,
+      "step": 101800
+    },
+    {
+      "epoch": 0.7369685914279717,
+      "grad_norm": 0.1437656432390213,
+      "learning_rate": 4.263038647238087e-06,
+      "loss": 0.894,
+      "step": 101810
+    },
+    {
+      "epoch": 0.7370409780885578,
+      "grad_norm": 0.16621340811252594,
+      "learning_rate": 4.262966260577502e-06,
+      "loss": 0.881,
+      "step": 101820
+    },
+    {
+      "epoch": 0.737113364749144,
+      "grad_norm": 0.14426559209823608,
+      "learning_rate": 4.262893873916915e-06,
+      "loss": 0.8927,
+      "step": 101830
+    },
+    {
+      "epoch": 0.7371857514097302,
+      "grad_norm": 0.16779235005378723,
+      "learning_rate": 4.262821487256329e-06,
+      "loss": 0.889,
+      "step": 101840
+    },
+    {
+      "epoch": 0.7372581380703164,
+      "grad_norm": 0.14878766238689423,
+      "learning_rate": 4.2627491005957425e-06,
+      "loss": 0.8903,
+      "step": 101850
+    },
+    {
+      "epoch": 0.7373305247309025,
+      "grad_norm": 0.15575487911701202,
+      "learning_rate": 4.262676713935157e-06,
+      "loss": 0.8779,
+      "step": 101860
+    },
+    {
+      "epoch": 0.7374029113914887,
+      "grad_norm": 0.1539398729801178,
+      "learning_rate": 4.262604327274571e-06,
+      "loss": 0.8843,
+      "step": 101870
+    },
+    {
+      "epoch": 0.737475298052075,
+      "grad_norm": 0.1627013087272644,
+      "learning_rate": 4.262531940613984e-06,
+      "loss": 0.8893,
+      "step": 101880
+    },
+    {
+      "epoch": 0.7375476847126612,
+      "grad_norm": 0.17876599729061127,
+      "learning_rate": 4.262459553953398e-06,
+      "loss": 0.8884,
+      "step": 101890
+    },
+    {
+      "epoch": 0.7376200713732474,
+      "grad_norm": 0.16160640120506287,
+      "learning_rate": 4.262387167292812e-06,
+      "loss": 0.8803,
+      "step": 101900
+    },
+    {
+      "epoch": 0.7376924580338335,
+      "grad_norm": 0.18297509849071503,
+      "learning_rate": 4.262314780632226e-06,
+      "loss": 0.8956,
+      "step": 101910
+    },
+    {
+      "epoch": 0.7377648446944197,
+      "grad_norm": 0.1643095761537552,
+      "learning_rate": 4.2622423939716395e-06,
+      "loss": 0.8855,
+      "step": 101920
+    },
+    {
+      "epoch": 0.7378372313550059,
+      "grad_norm": 0.17934858798980713,
+      "learning_rate": 4.262170007311053e-06,
+      "loss": 0.882,
+      "step": 101930
+    },
+    {
+      "epoch": 0.7379096180155921,
+      "grad_norm": 0.15763209760189056,
+      "learning_rate": 4.262097620650467e-06,
+      "loss": 0.8848,
+      "step": 101940
+    },
+    {
+      "epoch": 0.7379820046761782,
+      "grad_norm": 0.1518256813287735,
+      "learning_rate": 4.26202523398988e-06,
+      "loss": 0.8955,
+      "step": 101950
+    },
+    {
+      "epoch": 0.7380543913367644,
+      "grad_norm": 0.15647168457508087,
+      "learning_rate": 4.261952847329294e-06,
+      "loss": 0.9097,
+      "step": 101960
+    },
+    {
+      "epoch": 0.7381267779973506,
+      "grad_norm": 0.15015296638011932,
+      "learning_rate": 4.2618804606687084e-06,
+      "loss": 0.8752,
+      "step": 101970
+    },
+    {
+      "epoch": 0.7381991646579369,
+      "grad_norm": 0.15466606616973877,
+      "learning_rate": 4.261808074008122e-06,
+      "loss": 0.8849,
+      "step": 101980
+    },
+    {
+      "epoch": 0.7382715513185231,
+      "grad_norm": 0.17152267694473267,
+      "learning_rate": 4.261735687347536e-06,
+      "loss": 0.8728,
+      "step": 101990
+    },
+    {
+      "epoch": 0.7383439379791092,
+      "grad_norm": 0.14666572213172913,
+      "learning_rate": 4.261663300686949e-06,
+      "loss": 0.8757,
+      "step": 102000
+    },
+    {
+      "epoch": 0.7384163246396954,
+      "grad_norm": 0.1615079790353775,
+      "learning_rate": 4.261590914026364e-06,
+      "loss": 0.895,
+      "step": 102010
+    },
+    {
+      "epoch": 0.7384887113002816,
+      "grad_norm": 0.17585870623588562,
+      "learning_rate": 4.261518527365777e-06,
+      "loss": 0.8775,
+      "step": 102020
+    },
+    {
+      "epoch": 0.7385610979608678,
+      "grad_norm": 0.177878275513649,
+      "learning_rate": 4.261446140705191e-06,
+      "loss": 0.8664,
+      "step": 102030
+    },
+    {
+      "epoch": 0.738633484621454,
+      "grad_norm": 0.15934938192367554,
+      "learning_rate": 4.261373754044605e-06,
+      "loss": 0.8739,
+      "step": 102040
+    },
+    {
+      "epoch": 0.7387058712820401,
+      "grad_norm": 0.1503947675228119,
+      "learning_rate": 4.261301367384019e-06,
+      "loss": 0.8869,
+      "step": 102050
+    },
+    {
+      "epoch": 0.7387782579426263,
+      "grad_norm": 0.1672418862581253,
+      "learning_rate": 4.261228980723433e-06,
+      "loss": 0.8923,
+      "step": 102060
+    },
+    {
+      "epoch": 0.7388506446032125,
+      "grad_norm": 0.15783722698688507,
+      "learning_rate": 4.261156594062846e-06,
+      "loss": 0.9022,
+      "step": 102070
+    },
+    {
+      "epoch": 0.7389230312637987,
+      "grad_norm": 0.15607894957065582,
+      "learning_rate": 4.26108420740226e-06,
+      "loss": 0.8853,
+      "step": 102080
+    },
+    {
+      "epoch": 0.738995417924385,
+      "grad_norm": 0.14554698765277863,
+      "learning_rate": 4.261011820741674e-06,
+      "loss": 0.8896,
+      "step": 102090
+    },
+    {
+      "epoch": 0.7390678045849711,
+      "grad_norm": 0.15600426495075226,
+      "learning_rate": 4.260939434081088e-06,
+      "loss": 0.8755,
+      "step": 102100
+    },
+    {
+      "epoch": 0.7391401912455573,
+      "grad_norm": 0.14329098165035248,
+      "learning_rate": 4.260867047420502e-06,
+      "loss": 0.9022,
+      "step": 102110
+    },
+    {
+      "epoch": 0.7392125779061435,
+      "grad_norm": 0.1529379040002823,
+      "learning_rate": 4.260794660759915e-06,
+      "loss": 0.8871,
+      "step": 102120
+    },
+    {
+      "epoch": 0.7392849645667297,
+      "grad_norm": 0.15350185334682465,
+      "learning_rate": 4.260722274099329e-06,
+      "loss": 0.8908,
+      "step": 102130
+    },
+    {
+      "epoch": 0.7393573512273158,
+      "grad_norm": 0.17035044729709625,
+      "learning_rate": 4.260649887438743e-06,
+      "loss": 0.8809,
+      "step": 102140
+    },
+    {
+      "epoch": 0.739429737887902,
+      "grad_norm": 0.14859919250011444,
+      "learning_rate": 4.260577500778157e-06,
+      "loss": 0.901,
+      "step": 102150
+    },
+    {
+      "epoch": 0.7395021245484882,
+      "grad_norm": 0.15911567211151123,
+      "learning_rate": 4.2605051141175705e-06,
+      "loss": 0.8832,
+      "step": 102160
+    },
+    {
+      "epoch": 0.7395745112090744,
+      "grad_norm": 0.17544224858283997,
+      "learning_rate": 4.260432727456984e-06,
+      "loss": 0.884,
+      "step": 102170
+    },
+    {
+      "epoch": 0.7396468978696605,
+      "grad_norm": 0.15622149407863617,
+      "learning_rate": 4.260360340796399e-06,
+      "loss": 0.8888,
+      "step": 102180
+    },
+    {
+      "epoch": 0.7397192845302467,
+      "grad_norm": 0.20478832721710205,
+      "learning_rate": 4.260287954135812e-06,
+      "loss": 0.8916,
+      "step": 102190
+    },
+    {
+      "epoch": 0.739791671190833,
+      "grad_norm": 0.15085692703723907,
+      "learning_rate": 4.260215567475226e-06,
+      "loss": 0.8811,
+      "step": 102200
+    },
+    {
+      "epoch": 0.7398640578514192,
+      "grad_norm": 0.15468019247055054,
+      "learning_rate": 4.2601431808146395e-06,
+      "loss": 0.8901,
+      "step": 102210
+    },
+    {
+      "epoch": 0.7399364445120054,
+      "grad_norm": 0.15799905359745026,
+      "learning_rate": 4.260070794154054e-06,
+      "loss": 0.8872,
+      "step": 102220
+    },
+    {
+      "epoch": 0.7400088311725915,
+      "grad_norm": 0.14824278652668,
+      "learning_rate": 4.2599984074934675e-06,
+      "loss": 0.8929,
+      "step": 102230
+    },
+    {
+      "epoch": 0.7400812178331777,
+      "grad_norm": 0.15417352318763733,
+      "learning_rate": 4.259926020832881e-06,
+      "loss": 0.8917,
+      "step": 102240
+    },
+    {
+      "epoch": 0.7401536044937639,
+      "grad_norm": 0.16246527433395386,
+      "learning_rate": 4.259853634172295e-06,
+      "loss": 0.8828,
+      "step": 102250
+    },
+    {
+      "epoch": 0.7402259911543501,
+      "grad_norm": 0.14646829664707184,
+      "learning_rate": 4.259781247511709e-06,
+      "loss": 0.8966,
+      "step": 102260
+    },
+    {
+      "epoch": 0.7402983778149362,
+      "grad_norm": 0.15677151083946228,
+      "learning_rate": 4.259708860851123e-06,
+      "loss": 0.8897,
+      "step": 102270
+    },
+    {
+      "epoch": 0.7403707644755224,
+      "grad_norm": 0.16189172863960266,
+      "learning_rate": 4.2596364741905365e-06,
+      "loss": 0.8789,
+      "step": 102280
+    },
+    {
+      "epoch": 0.7404431511361086,
+      "grad_norm": 0.15842679142951965,
+      "learning_rate": 4.25956408752995e-06,
+      "loss": 0.8876,
+      "step": 102290
+    },
+    {
+      "epoch": 0.7405155377966949,
+      "grad_norm": 0.14995113015174866,
+      "learning_rate": 4.2594917008693646e-06,
+      "loss": 0.8881,
+      "step": 102300
+    },
+    {
+      "epoch": 0.740587924457281,
+      "grad_norm": 0.1539173424243927,
+      "learning_rate": 4.259419314208778e-06,
+      "loss": 0.8848,
+      "step": 102310
+    },
+    {
+      "epoch": 0.7406603111178672,
+      "grad_norm": 0.17772093415260315,
+      "learning_rate": 4.259346927548192e-06,
+      "loss": 0.8902,
+      "step": 102320
+    },
+    {
+      "epoch": 0.7407326977784534,
+      "grad_norm": 0.1554667055606842,
+      "learning_rate": 4.259274540887605e-06,
+      "loss": 0.8865,
+      "step": 102330
+    },
+    {
+      "epoch": 0.7408050844390396,
+      "grad_norm": 0.15641778707504272,
+      "learning_rate": 4.25920215422702e-06,
+      "loss": 0.9031,
+      "step": 102340
+    },
+    {
+      "epoch": 0.7408774710996258,
+      "grad_norm": 0.15458346903324127,
+      "learning_rate": 4.2591297675664335e-06,
+      "loss": 0.8843,
+      "step": 102350
+    },
+    {
+      "epoch": 0.7409498577602119,
+      "grad_norm": 0.15070627629756927,
+      "learning_rate": 4.259057380905847e-06,
+      "loss": 0.8975,
+      "step": 102360
+    },
+    {
+      "epoch": 0.7410222444207981,
+      "grad_norm": 0.15110282599925995,
+      "learning_rate": 4.258984994245261e-06,
+      "loss": 0.8928,
+      "step": 102370
+    },
+    {
+      "epoch": 0.7410946310813843,
+      "grad_norm": 0.1538836807012558,
+      "learning_rate": 4.258912607584675e-06,
+      "loss": 0.8946,
+      "step": 102380
+    },
+    {
+      "epoch": 0.7411670177419705,
+      "grad_norm": 0.15167924761772156,
+      "learning_rate": 4.258840220924089e-06,
+      "loss": 0.8839,
+      "step": 102390
+    },
+    {
+      "epoch": 0.7412394044025566,
+      "grad_norm": 0.15393514931201935,
+      "learning_rate": 4.258767834263502e-06,
+      "loss": 0.887,
+      "step": 102400
+    },
+    {
+      "epoch": 0.7413117910631429,
+      "grad_norm": 0.1549709439277649,
+      "learning_rate": 4.258695447602916e-06,
+      "loss": 0.8871,
+      "step": 102410
+    },
+    {
+      "epoch": 0.7413841777237291,
+      "grad_norm": 0.16167686879634857,
+      "learning_rate": 4.2586230609423305e-06,
+      "loss": 0.8783,
+      "step": 102420
+    },
+    {
+      "epoch": 0.7414565643843153,
+      "grad_norm": 0.15924765169620514,
+      "learning_rate": 4.258550674281744e-06,
+      "loss": 0.8769,
+      "step": 102430
+    },
+    {
+      "epoch": 0.7415289510449015,
+      "grad_norm": 0.15444104373455048,
+      "learning_rate": 4.258478287621158e-06,
+      "loss": 0.8888,
+      "step": 102440
+    },
+    {
+      "epoch": 0.7416013377054876,
+      "grad_norm": 0.14428524672985077,
+      "learning_rate": 4.258405900960571e-06,
+      "loss": 0.8683,
+      "step": 102450
+    },
+    {
+      "epoch": 0.7416737243660738,
+      "grad_norm": 0.16138869524002075,
+      "learning_rate": 4.258333514299985e-06,
+      "loss": 0.8838,
+      "step": 102460
+    },
+    {
+      "epoch": 0.74174611102666,
+      "grad_norm": 0.16185913980007172,
+      "learning_rate": 4.2582611276393986e-06,
+      "loss": 0.8833,
+      "step": 102470
+    },
+    {
+      "epoch": 0.7418184976872462,
+      "grad_norm": 0.1485951989889145,
+      "learning_rate": 4.258188740978812e-06,
+      "loss": 0.8956,
+      "step": 102480
+    },
+    {
+      "epoch": 0.7418908843478323,
+      "grad_norm": 0.16537578403949738,
+      "learning_rate": 4.258116354318227e-06,
+      "loss": 0.8927,
+      "step": 102490
+    },
+    {
+      "epoch": 0.7419632710084185,
+      "grad_norm": 0.14754781126976013,
+      "learning_rate": 4.25804396765764e-06,
+      "loss": 0.881,
+      "step": 102500
+    },
+    {
+      "epoch": 0.7420356576690048,
+      "grad_norm": 0.14885316789150238,
+      "learning_rate": 4.257971580997054e-06,
+      "loss": 0.8927,
+      "step": 102510
+    },
+    {
+      "epoch": 0.742108044329591,
+      "grad_norm": 0.17432914674282074,
+      "learning_rate": 4.2578991943364675e-06,
+      "loss": 0.8965,
+      "step": 102520
+    },
+    {
+      "epoch": 0.7421804309901772,
+      "grad_norm": 0.1665223240852356,
+      "learning_rate": 4.257826807675882e-06,
+      "loss": 0.8862,
+      "step": 102530
+    },
+    {
+      "epoch": 0.7422528176507633,
+      "grad_norm": 0.15308286249637604,
+      "learning_rate": 4.257754421015296e-06,
+      "loss": 0.8913,
+      "step": 102540
+    },
+    {
+      "epoch": 0.7423252043113495,
+      "grad_norm": 0.1635119467973709,
+      "learning_rate": 4.257682034354709e-06,
+      "loss": 0.8989,
+      "step": 102550
+    },
+    {
+      "epoch": 0.7423975909719357,
+      "grad_norm": 0.15072479844093323,
+      "learning_rate": 4.257609647694123e-06,
+      "loss": 0.8836,
+      "step": 102560
+    },
+    {
+      "epoch": 0.7424699776325219,
+      "grad_norm": 0.17337563633918762,
+      "learning_rate": 4.257537261033537e-06,
+      "loss": 0.8954,
+      "step": 102570
+    },
+    {
+      "epoch": 0.742542364293108,
+      "grad_norm": 0.14887109398841858,
+      "learning_rate": 4.257464874372951e-06,
+      "loss": 0.8954,
+      "step": 102580
+    },
+    {
+      "epoch": 0.7426147509536942,
+      "grad_norm": 0.16400760412216187,
+      "learning_rate": 4.2573924877123645e-06,
+      "loss": 0.8847,
+      "step": 102590
+    },
+    {
+      "epoch": 0.7426871376142804,
+      "grad_norm": 0.15407805144786835,
+      "learning_rate": 4.257320101051778e-06,
+      "loss": 0.8885,
+      "step": 102600
+    },
+    {
+      "epoch": 0.7427595242748666,
+      "grad_norm": 0.15420646965503693,
+      "learning_rate": 4.257247714391193e-06,
+      "loss": 0.8844,
+      "step": 102610
+    },
+    {
+      "epoch": 0.7428319109354529,
+      "grad_norm": 0.15996070206165314,
+      "learning_rate": 4.257175327730606e-06,
+      "loss": 0.8854,
+      "step": 102620
+    },
+    {
+      "epoch": 0.742904297596039,
+      "grad_norm": 0.16860343515872955,
+      "learning_rate": 4.25710294107002e-06,
+      "loss": 0.8879,
+      "step": 102630
+    },
+    {
+      "epoch": 0.7429766842566252,
+      "grad_norm": 0.15210293233394623,
+      "learning_rate": 4.2570305544094334e-06,
+      "loss": 0.8922,
+      "step": 102640
+    },
+    {
+      "epoch": 0.7430490709172114,
+      "grad_norm": 0.17375917732715607,
+      "learning_rate": 4.256958167748848e-06,
+      "loss": 0.8811,
+      "step": 102650
+    },
+    {
+      "epoch": 0.7431214575777976,
+      "grad_norm": 0.16159164905548096,
+      "learning_rate": 4.2568857810882615e-06,
+      "loss": 0.8704,
+      "step": 102660
+    },
+    {
+      "epoch": 0.7431938442383837,
+      "grad_norm": 0.1597278118133545,
+      "learning_rate": 4.256813394427675e-06,
+      "loss": 0.8723,
+      "step": 102670
+    },
+    {
+      "epoch": 0.7432662308989699,
+      "grad_norm": 0.16430126130580902,
+      "learning_rate": 4.256741007767089e-06,
+      "loss": 0.8836,
+      "step": 102680
+    },
+    {
+      "epoch": 0.7433386175595561,
+      "grad_norm": 0.15877282619476318,
+      "learning_rate": 4.256668621106503e-06,
+      "loss": 0.887,
+      "step": 102690
+    },
+    {
+      "epoch": 0.7434110042201423,
+      "grad_norm": 0.15375186502933502,
+      "learning_rate": 4.256596234445917e-06,
+      "loss": 0.8792,
+      "step": 102700
+    },
+    {
+      "epoch": 0.7434833908807285,
+      "grad_norm": 0.177475243806839,
+      "learning_rate": 4.2565238477853304e-06,
+      "loss": 0.8842,
+      "step": 102710
+    },
+    {
+      "epoch": 0.7435557775413146,
+      "grad_norm": 0.14837265014648438,
+      "learning_rate": 4.256451461124744e-06,
+      "loss": 0.8891,
+      "step": 102720
+    },
+    {
+      "epoch": 0.7436281642019009,
+      "grad_norm": 0.1412384957075119,
+      "learning_rate": 4.2563790744641585e-06,
+      "loss": 0.8849,
+      "step": 102730
+    },
+    {
+      "epoch": 0.7437005508624871,
+      "grad_norm": 0.15624116361141205,
+      "learning_rate": 4.256306687803572e-06,
+      "loss": 0.893,
+      "step": 102740
+    },
+    {
+      "epoch": 0.7437729375230733,
+      "grad_norm": 0.15126636624336243,
+      "learning_rate": 4.256234301142986e-06,
+      "loss": 0.8776,
+      "step": 102750
+    },
+    {
+      "epoch": 0.7438453241836595,
+      "grad_norm": 0.1584903746843338,
+      "learning_rate": 4.256161914482399e-06,
+      "loss": 0.8855,
+      "step": 102760
+    },
+    {
+      "epoch": 0.7439177108442456,
+      "grad_norm": 0.1627122163772583,
+      "learning_rate": 4.256089527821813e-06,
+      "loss": 0.89,
+      "step": 102770
+    },
+    {
+      "epoch": 0.7439900975048318,
+      "grad_norm": 0.15770822763442993,
+      "learning_rate": 4.2560171411612274e-06,
+      "loss": 0.8928,
+      "step": 102780
+    },
+    {
+      "epoch": 0.744062484165418,
+      "grad_norm": 0.1587304323911667,
+      "learning_rate": 4.255944754500641e-06,
+      "loss": 0.8867,
+      "step": 102790
+    },
+    {
+      "epoch": 0.7441348708260042,
+      "grad_norm": 0.15314078330993652,
+      "learning_rate": 4.255872367840055e-06,
+      "loss": 0.8957,
+      "step": 102800
+    },
+    {
+      "epoch": 0.7442072574865903,
+      "grad_norm": 0.14915263652801514,
+      "learning_rate": 4.255799981179468e-06,
+      "loss": 0.8805,
+      "step": 102810
+    },
+    {
+      "epoch": 0.7442796441471765,
+      "grad_norm": 0.14680497348308563,
+      "learning_rate": 4.255727594518883e-06,
+      "loss": 0.8882,
+      "step": 102820
+    },
+    {
+      "epoch": 0.7443520308077628,
+      "grad_norm": 0.15903522074222565,
+      "learning_rate": 4.255655207858296e-06,
+      "loss": 0.8976,
+      "step": 102830
+    },
+    {
+      "epoch": 0.744424417468349,
+      "grad_norm": 0.1559790074825287,
+      "learning_rate": 4.25558282119771e-06,
+      "loss": 0.8856,
+      "step": 102840
+    },
+    {
+      "epoch": 0.7444968041289352,
+      "grad_norm": 0.15489821135997772,
+      "learning_rate": 4.255510434537124e-06,
+      "loss": 0.8878,
+      "step": 102850
+    },
+    {
+      "epoch": 0.7445691907895213,
+      "grad_norm": 0.15823949873447418,
+      "learning_rate": 4.255438047876538e-06,
+      "loss": 0.8764,
+      "step": 102860
+    },
+    {
+      "epoch": 0.7446415774501075,
+      "grad_norm": 0.1716890037059784,
+      "learning_rate": 4.255365661215952e-06,
+      "loss": 0.8994,
+      "step": 102870
+    },
+    {
+      "epoch": 0.7447139641106937,
+      "grad_norm": 0.15007559955120087,
+      "learning_rate": 4.255293274555365e-06,
+      "loss": 0.8947,
+      "step": 102880
+    },
+    {
+      "epoch": 0.7447863507712799,
+      "grad_norm": 0.16091489791870117,
+      "learning_rate": 4.255220887894779e-06,
+      "loss": 0.8847,
+      "step": 102890
+    },
+    {
+      "epoch": 0.744858737431866,
+      "grad_norm": 0.1675153523683548,
+      "learning_rate": 4.255148501234193e-06,
+      "loss": 0.8804,
+      "step": 102900
+    },
+    {
+      "epoch": 0.7449311240924522,
+      "grad_norm": 0.15456238389015198,
+      "learning_rate": 4.255076114573607e-06,
+      "loss": 0.8986,
+      "step": 102910
+    },
+    {
+      "epoch": 0.7450035107530384,
+      "grad_norm": 0.15441204607486725,
+      "learning_rate": 4.255003727913021e-06,
+      "loss": 0.8742,
+      "step": 102920
+    },
+    {
+      "epoch": 0.7450758974136246,
+      "grad_norm": 0.16027353703975677,
+      "learning_rate": 4.254931341252434e-06,
+      "loss": 0.8856,
+      "step": 102930
+    },
+    {
+      "epoch": 0.7451482840742109,
+      "grad_norm": 0.16300475597381592,
+      "learning_rate": 4.254858954591849e-06,
+      "loss": 0.882,
+      "step": 102940
+    },
+    {
+      "epoch": 0.745220670734797,
+      "grad_norm": 0.15692980587482452,
+      "learning_rate": 4.254786567931262e-06,
+      "loss": 0.8875,
+      "step": 102950
+    },
+    {
+      "epoch": 0.7452930573953832,
+      "grad_norm": 0.16681601107120514,
+      "learning_rate": 4.254714181270676e-06,
+      "loss": 0.8866,
+      "step": 102960
+    },
+    {
+      "epoch": 0.7453654440559694,
+      "grad_norm": 0.16296398639678955,
+      "learning_rate": 4.2546417946100895e-06,
+      "loss": 0.8969,
+      "step": 102970
+    },
+    {
+      "epoch": 0.7454378307165556,
+      "grad_norm": 0.1497073471546173,
+      "learning_rate": 4.254569407949504e-06,
+      "loss": 0.8949,
+      "step": 102980
+    },
+    {
+      "epoch": 0.7455102173771417,
+      "grad_norm": 0.17505697906017303,
+      "learning_rate": 4.254497021288917e-06,
+      "loss": 0.8935,
+      "step": 102990
+    },
+    {
+      "epoch": 0.7455826040377279,
+      "grad_norm": 0.15840458869934082,
+      "learning_rate": 4.25442463462833e-06,
+      "loss": 0.8957,
+      "step": 103000
+    },
+    {
+      "epoch": 0.7456549906983141,
+      "grad_norm": 0.1545899212360382,
+      "learning_rate": 4.254352247967745e-06,
+      "loss": 0.8853,
+      "step": 103010
+    },
+    {
+      "epoch": 0.7457273773589003,
+      "grad_norm": 0.1966148018836975,
+      "learning_rate": 4.2542798613071585e-06,
+      "loss": 0.8883,
+      "step": 103020
+    },
+    {
+      "epoch": 0.7457997640194864,
+      "grad_norm": 0.1721884310245514,
+      "learning_rate": 4.254207474646572e-06,
+      "loss": 0.8887,
+      "step": 103030
+    },
+    {
+      "epoch": 0.7458721506800726,
+      "grad_norm": 0.19111919403076172,
+      "learning_rate": 4.254135087985986e-06,
+      "loss": 0.8829,
+      "step": 103040
+    },
+    {
+      "epoch": 0.7459445373406589,
+      "grad_norm": 0.15273134410381317,
+      "learning_rate": 4.2540627013254e-06,
+      "loss": 0.8952,
+      "step": 103050
+    },
+    {
+      "epoch": 0.7460169240012451,
+      "grad_norm": 0.16259609162807465,
+      "learning_rate": 4.253990314664814e-06,
+      "loss": 0.8866,
+      "step": 103060
+    },
+    {
+      "epoch": 0.7460893106618313,
+      "grad_norm": 0.15013474225997925,
+      "learning_rate": 4.253917928004227e-06,
+      "loss": 0.8792,
+      "step": 103070
+    },
+    {
+      "epoch": 0.7461616973224174,
+      "grad_norm": 0.17235596477985382,
+      "learning_rate": 4.253845541343641e-06,
+      "loss": 0.9079,
+      "step": 103080
+    },
+    {
+      "epoch": 0.7462340839830036,
+      "grad_norm": 0.16108207404613495,
+      "learning_rate": 4.2537731546830555e-06,
+      "loss": 0.8797,
+      "step": 103090
+    },
+    {
+      "epoch": 0.7463064706435898,
+      "grad_norm": 0.16072994470596313,
+      "learning_rate": 4.253700768022469e-06,
+      "loss": 0.8804,
+      "step": 103100
+    },
+    {
+      "epoch": 0.746378857304176,
+      "grad_norm": 0.1662396937608719,
+      "learning_rate": 4.253628381361883e-06,
+      "loss": 0.8866,
+      "step": 103110
+    },
+    {
+      "epoch": 0.7464512439647621,
+      "grad_norm": 0.16367369890213013,
+      "learning_rate": 4.253555994701296e-06,
+      "loss": 0.8975,
+      "step": 103120
+    },
+    {
+      "epoch": 0.7465236306253483,
+      "grad_norm": 0.15681147575378418,
+      "learning_rate": 4.253483608040711e-06,
+      "loss": 0.8995,
+      "step": 103130
+    },
+    {
+      "epoch": 0.7465960172859345,
+      "grad_norm": 0.1485588252544403,
+      "learning_rate": 4.253411221380124e-06,
+      "loss": 0.8863,
+      "step": 103140
+    },
+    {
+      "epoch": 0.7466684039465208,
+      "grad_norm": 0.1582004874944687,
+      "learning_rate": 4.253338834719538e-06,
+      "loss": 0.8895,
+      "step": 103150
+    },
+    {
+      "epoch": 0.746740790607107,
+      "grad_norm": 0.17631734907627106,
+      "learning_rate": 4.253266448058952e-06,
+      "loss": 0.8993,
+      "step": 103160
+    },
+    {
+      "epoch": 0.7468131772676931,
+      "grad_norm": 0.15485930442810059,
+      "learning_rate": 4.253194061398366e-06,
+      "loss": 0.8882,
+      "step": 103170
+    },
+    {
+      "epoch": 0.7468855639282793,
+      "grad_norm": 0.18264351785182953,
+      "learning_rate": 4.25312167473778e-06,
+      "loss": 0.8989,
+      "step": 103180
+    },
+    {
+      "epoch": 0.7469579505888655,
+      "grad_norm": 0.15158303081989288,
+      "learning_rate": 4.253049288077193e-06,
+      "loss": 0.9056,
+      "step": 103190
+    },
+    {
+      "epoch": 0.7470303372494517,
+      "grad_norm": 0.1658327281475067,
+      "learning_rate": 4.252976901416607e-06,
+      "loss": 0.8896,
+      "step": 103200
+    },
+    {
+      "epoch": 0.7471027239100378,
+      "grad_norm": 0.15178616344928741,
+      "learning_rate": 4.252904514756021e-06,
+      "loss": 0.8784,
+      "step": 103210
+    },
+    {
+      "epoch": 0.747175110570624,
+      "grad_norm": 0.15876880288124084,
+      "learning_rate": 4.252832128095435e-06,
+      "loss": 0.8802,
+      "step": 103220
+    },
+    {
+      "epoch": 0.7472474972312102,
+      "grad_norm": 0.15720294415950775,
+      "learning_rate": 4.252759741434849e-06,
+      "loss": 0.8837,
+      "step": 103230
+    },
+    {
+      "epoch": 0.7473198838917964,
+      "grad_norm": 0.16239741444587708,
+      "learning_rate": 4.252687354774262e-06,
+      "loss": 0.8917,
+      "step": 103240
+    },
+    {
+      "epoch": 0.7473922705523826,
+      "grad_norm": 0.19867613911628723,
+      "learning_rate": 4.252614968113677e-06,
+      "loss": 0.8822,
+      "step": 103250
+    },
+    {
+      "epoch": 0.7474646572129688,
+      "grad_norm": 0.15661980211734772,
+      "learning_rate": 4.25254258145309e-06,
+      "loss": 0.8789,
+      "step": 103260
+    },
+    {
+      "epoch": 0.747537043873555,
+      "grad_norm": 0.16807293891906738,
+      "learning_rate": 4.252470194792504e-06,
+      "loss": 0.9001,
+      "step": 103270
+    },
+    {
+      "epoch": 0.7476094305341412,
+      "grad_norm": 0.15940342843532562,
+      "learning_rate": 4.252397808131918e-06,
+      "loss": 0.8969,
+      "step": 103280
+    },
+    {
+      "epoch": 0.7476818171947274,
+      "grad_norm": 0.1640319973230362,
+      "learning_rate": 4.252325421471332e-06,
+      "loss": 0.8795,
+      "step": 103290
+    },
+    {
+      "epoch": 0.7477542038553135,
+      "grad_norm": 0.17689263820648193,
+      "learning_rate": 4.252253034810746e-06,
+      "loss": 0.891,
+      "step": 103300
+    },
+    {
+      "epoch": 0.7478265905158997,
+      "grad_norm": 0.14825983345508575,
+      "learning_rate": 4.252180648150159e-06,
+      "loss": 0.8791,
+      "step": 103310
+    },
+    {
+      "epoch": 0.7478989771764859,
+      "grad_norm": 0.162586972117424,
+      "learning_rate": 4.252108261489573e-06,
+      "loss": 0.8809,
+      "step": 103320
+    },
+    {
+      "epoch": 0.7479713638370721,
+      "grad_norm": 0.1549188792705536,
+      "learning_rate": 4.252035874828987e-06,
+      "loss": 0.8842,
+      "step": 103330
+    },
+    {
+      "epoch": 0.7480437504976583,
+      "grad_norm": 0.1467401683330536,
+      "learning_rate": 4.251963488168401e-06,
+      "loss": 0.8806,
+      "step": 103340
+    },
+    {
+      "epoch": 0.7481161371582444,
+      "grad_norm": 0.16220062971115112,
+      "learning_rate": 4.251891101507815e-06,
+      "loss": 0.8836,
+      "step": 103350
+    },
+    {
+      "epoch": 0.7481885238188307,
+      "grad_norm": 0.1533544659614563,
+      "learning_rate": 4.251818714847228e-06,
+      "loss": 0.8901,
+      "step": 103360
+    },
+    {
+      "epoch": 0.7482609104794169,
+      "grad_norm": 0.1560657024383545,
+      "learning_rate": 4.251746328186642e-06,
+      "loss": 0.8769,
+      "step": 103370
+    },
+    {
+      "epoch": 0.7483332971400031,
+      "grad_norm": 0.1645650416612625,
+      "learning_rate": 4.251673941526056e-06,
+      "loss": 0.8804,
+      "step": 103380
+    },
+    {
+      "epoch": 0.7484056838005892,
+      "grad_norm": 0.1500735580921173,
+      "learning_rate": 4.25160155486547e-06,
+      "loss": 0.8849,
+      "step": 103390
+    },
+    {
+      "epoch": 0.7484780704611754,
+      "grad_norm": 0.1593628227710724,
+      "learning_rate": 4.2515291682048835e-06,
+      "loss": 0.8793,
+      "step": 103400
+    },
+    {
+      "epoch": 0.7485504571217616,
+      "grad_norm": 0.16192486882209778,
+      "learning_rate": 4.251456781544297e-06,
+      "loss": 0.8873,
+      "step": 103410
+    },
+    {
+      "epoch": 0.7486228437823478,
+      "grad_norm": 0.147709459066391,
+      "learning_rate": 4.251384394883712e-06,
+      "loss": 0.8904,
+      "step": 103420
+    },
+    {
+      "epoch": 0.748695230442934,
+      "grad_norm": 0.16684521734714508,
+      "learning_rate": 4.251312008223125e-06,
+      "loss": 0.895,
+      "step": 103430
+    },
+    {
+      "epoch": 0.7487676171035201,
+      "grad_norm": 0.14957979321479797,
+      "learning_rate": 4.251239621562539e-06,
+      "loss": 0.8818,
+      "step": 103440
+    },
+    {
+      "epoch": 0.7488400037641063,
+      "grad_norm": 0.15316711366176605,
+      "learning_rate": 4.2511672349019524e-06,
+      "loss": 0.8816,
+      "step": 103450
+    },
+    {
+      "epoch": 0.7489123904246925,
+      "grad_norm": 0.15910635888576508,
+      "learning_rate": 4.251094848241367e-06,
+      "loss": 0.8801,
+      "step": 103460
+    },
+    {
+      "epoch": 0.7489847770852788,
+      "grad_norm": 0.1638316661119461,
+      "learning_rate": 4.2510224615807805e-06,
+      "loss": 0.8842,
+      "step": 103470
+    },
+    {
+      "epoch": 0.749057163745865,
+      "grad_norm": 0.15128594636917114,
+      "learning_rate": 4.250950074920194e-06,
+      "loss": 0.8978,
+      "step": 103480
+    },
+    {
+      "epoch": 0.7491295504064511,
+      "grad_norm": 0.15200836956501007,
+      "learning_rate": 4.250877688259608e-06,
+      "loss": 0.8843,
+      "step": 103490
+    },
+    {
+      "epoch": 0.7492019370670373,
+      "grad_norm": 0.19085319340229034,
+      "learning_rate": 4.250805301599022e-06,
+      "loss": 0.8784,
+      "step": 103500
+    },
+    {
+      "epoch": 0.7492743237276235,
+      "grad_norm": 0.15487466752529144,
+      "learning_rate": 4.250732914938436e-06,
+      "loss": 0.8954,
+      "step": 103510
+    },
+    {
+      "epoch": 0.7493467103882097,
+      "grad_norm": 0.15988904237747192,
+      "learning_rate": 4.2506605282778494e-06,
+      "loss": 0.8951,
+      "step": 103520
+    },
+    {
+      "epoch": 0.7494190970487958,
+      "grad_norm": 0.1502448171377182,
+      "learning_rate": 4.250588141617263e-06,
+      "loss": 0.8985,
+      "step": 103530
+    },
+    {
+      "epoch": 0.749491483709382,
+      "grad_norm": 0.153715580701828,
+      "learning_rate": 4.250515754956677e-06,
+      "loss": 0.8955,
+      "step": 103540
+    },
+    {
+      "epoch": 0.7495638703699682,
+      "grad_norm": 0.1608363538980484,
+      "learning_rate": 4.25044336829609e-06,
+      "loss": 0.8897,
+      "step": 103550
+    },
+    {
+      "epoch": 0.7496362570305544,
+      "grad_norm": 0.1529805064201355,
+      "learning_rate": 4.250370981635504e-06,
+      "loss": 0.8936,
+      "step": 103560
+    },
+    {
+      "epoch": 0.7497086436911405,
+      "grad_norm": 0.15650229156017303,
+      "learning_rate": 4.250298594974918e-06,
+      "loss": 0.8815,
+      "step": 103570
+    },
+    {
+      "epoch": 0.7497810303517268,
+      "grad_norm": 0.1722974181175232,
+      "learning_rate": 4.250226208314332e-06,
+      "loss": 0.9026,
+      "step": 103580
+    },
+    {
+      "epoch": 0.749853417012313,
+      "grad_norm": 0.16492018103599548,
+      "learning_rate": 4.250153821653746e-06,
+      "loss": 0.8814,
+      "step": 103590
+    },
+    {
+      "epoch": 0.7499258036728992,
+      "grad_norm": 0.15420758724212646,
+      "learning_rate": 4.250081434993159e-06,
+      "loss": 0.888,
+      "step": 103600
+    },
+    {
+      "epoch": 0.7499981903334854,
+      "grad_norm": 0.15207520127296448,
+      "learning_rate": 4.250009048332574e-06,
+      "loss": 0.8778,
+      "step": 103610
+    },
+    {
+      "epoch": 0.7500705769940715,
+      "grad_norm": 0.15864045917987823,
+      "learning_rate": 4.249936661671987e-06,
+      "loss": 0.885,
+      "step": 103620
+    },
+    {
+      "epoch": 0.7501429636546577,
+      "grad_norm": 0.16581547260284424,
+      "learning_rate": 4.249864275011401e-06,
+      "loss": 0.8845,
+      "step": 103630
+    },
+    {
+      "epoch": 0.7502153503152439,
+      "grad_norm": 0.16255062818527222,
+      "learning_rate": 4.2497918883508145e-06,
+      "loss": 0.8849,
+      "step": 103640
+    },
+    {
+      "epoch": 0.7502877369758301,
+      "grad_norm": 0.15730057656764984,
+      "learning_rate": 4.249719501690229e-06,
+      "loss": 0.8975,
+      "step": 103650
+    },
+    {
+      "epoch": 0.7503601236364162,
+      "grad_norm": 0.16164420545101166,
+      "learning_rate": 4.249647115029643e-06,
+      "loss": 0.8865,
+      "step": 103660
+    },
+    {
+      "epoch": 0.7504325102970024,
+      "grad_norm": 0.1671602427959442,
+      "learning_rate": 4.249574728369056e-06,
+      "loss": 0.8888,
+      "step": 103670
+    },
+    {
+      "epoch": 0.7505048969575887,
+      "grad_norm": 0.1699591726064682,
+      "learning_rate": 4.24950234170847e-06,
+      "loss": 0.87,
+      "step": 103680
+    },
+    {
+      "epoch": 0.7505772836181749,
+      "grad_norm": 0.16800114512443542,
+      "learning_rate": 4.249429955047884e-06,
+      "loss": 0.9011,
+      "step": 103690
+    },
+    {
+      "epoch": 0.7506496702787611,
+      "grad_norm": 0.18705791234970093,
+      "learning_rate": 4.249357568387298e-06,
+      "loss": 0.8766,
+      "step": 103700
+    },
+    {
+      "epoch": 0.7507220569393472,
+      "grad_norm": 0.1909557431936264,
+      "learning_rate": 4.2492851817267115e-06,
+      "loss": 0.9045,
+      "step": 103710
+    },
+    {
+      "epoch": 0.7507944435999334,
+      "grad_norm": 0.15778642892837524,
+      "learning_rate": 4.249212795066125e-06,
+      "loss": 0.8715,
+      "step": 103720
+    },
+    {
+      "epoch": 0.7508668302605196,
+      "grad_norm": 0.1488863229751587,
+      "learning_rate": 4.24914040840554e-06,
+      "loss": 0.8756,
+      "step": 103730
+    },
+    {
+      "epoch": 0.7509392169211058,
+      "grad_norm": 0.15307924151420593,
+      "learning_rate": 4.249068021744953e-06,
+      "loss": 0.8738,
+      "step": 103740
+    },
+    {
+      "epoch": 0.7510116035816919,
+      "grad_norm": 0.18600323796272278,
+      "learning_rate": 4.248995635084367e-06,
+      "loss": 0.8883,
+      "step": 103750
+    },
+    {
+      "epoch": 0.7510839902422781,
+      "grad_norm": 0.16874092817306519,
+      "learning_rate": 4.2489232484237805e-06,
+      "loss": 0.8835,
+      "step": 103760
+    },
+    {
+      "epoch": 0.7511563769028643,
+      "grad_norm": 0.1458248794078827,
+      "learning_rate": 4.248850861763195e-06,
+      "loss": 0.878,
+      "step": 103770
+    },
+    {
+      "epoch": 0.7512287635634505,
+      "grad_norm": 0.1547769457101822,
+      "learning_rate": 4.2487784751026086e-06,
+      "loss": 0.888,
+      "step": 103780
+    },
+    {
+      "epoch": 0.7513011502240368,
+      "grad_norm": 0.15332648158073425,
+      "learning_rate": 4.248706088442022e-06,
+      "loss": 0.8808,
+      "step": 103790
+    },
+    {
+      "epoch": 0.7513735368846229,
+      "grad_norm": 0.16886559128761292,
+      "learning_rate": 4.248633701781436e-06,
+      "loss": 0.884,
+      "step": 103800
+    },
+    {
+      "epoch": 0.7514459235452091,
+      "grad_norm": 0.16709497570991516,
+      "learning_rate": 4.24856131512085e-06,
+      "loss": 0.8972,
+      "step": 103810
+    },
+    {
+      "epoch": 0.7515183102057953,
+      "grad_norm": 0.15135438740253448,
+      "learning_rate": 4.248488928460264e-06,
+      "loss": 0.8969,
+      "step": 103820
+    },
+    {
+      "epoch": 0.7515906968663815,
+      "grad_norm": 0.15672151744365692,
+      "learning_rate": 4.2484165417996775e-06,
+      "loss": 0.893,
+      "step": 103830
+    },
+    {
+      "epoch": 0.7516630835269676,
+      "grad_norm": 0.14277760684490204,
+      "learning_rate": 4.248344155139091e-06,
+      "loss": 0.8874,
+      "step": 103840
+    },
+    {
+      "epoch": 0.7517354701875538,
+      "grad_norm": 0.15540048480033875,
+      "learning_rate": 4.2482717684785056e-06,
+      "loss": 0.8904,
+      "step": 103850
+    },
+    {
+      "epoch": 0.75180785684814,
+      "grad_norm": 0.14748062193393707,
+      "learning_rate": 4.248199381817919e-06,
+      "loss": 0.8929,
+      "step": 103860
+    },
+    {
+      "epoch": 0.7518802435087262,
+      "grad_norm": 0.154066801071167,
+      "learning_rate": 4.248126995157333e-06,
+      "loss": 0.8819,
+      "step": 103870
+    },
+    {
+      "epoch": 0.7519526301693124,
+      "grad_norm": 0.15664027631282806,
+      "learning_rate": 4.248054608496746e-06,
+      "loss": 0.8937,
+      "step": 103880
+    },
+    {
+      "epoch": 0.7520250168298986,
+      "grad_norm": 0.15651507675647736,
+      "learning_rate": 4.247982221836161e-06,
+      "loss": 0.8824,
+      "step": 103890
+    },
+    {
+      "epoch": 0.7520974034904848,
+      "grad_norm": 0.15763552486896515,
+      "learning_rate": 4.2479098351755745e-06,
+      "loss": 0.8743,
+      "step": 103900
+    },
+    {
+      "epoch": 0.752169790151071,
+      "grad_norm": 0.15285298228263855,
+      "learning_rate": 4.247837448514988e-06,
+      "loss": 0.8947,
+      "step": 103910
+    },
+    {
+      "epoch": 0.7522421768116572,
+      "grad_norm": 0.15118758380413055,
+      "learning_rate": 4.247765061854402e-06,
+      "loss": 0.8839,
+      "step": 103920
+    },
+    {
+      "epoch": 0.7523145634722433,
+      "grad_norm": 0.16061747074127197,
+      "learning_rate": 4.247692675193816e-06,
+      "loss": 0.8882,
+      "step": 103930
+    },
+    {
+      "epoch": 0.7523869501328295,
+      "grad_norm": 0.15958862006664276,
+      "learning_rate": 4.24762028853323e-06,
+      "loss": 0.8741,
+      "step": 103940
+    },
+    {
+      "epoch": 0.7524593367934157,
+      "grad_norm": 0.152791365981102,
+      "learning_rate": 4.247547901872643e-06,
+      "loss": 0.8921,
+      "step": 103950
+    },
+    {
+      "epoch": 0.7525317234540019,
+      "grad_norm": 0.15019190311431885,
+      "learning_rate": 4.247475515212057e-06,
+      "loss": 0.8939,
+      "step": 103960
+    },
+    {
+      "epoch": 0.752604110114588,
+      "grad_norm": 0.15106353163719177,
+      "learning_rate": 4.2474031285514715e-06,
+      "loss": 0.877,
+      "step": 103970
+    },
+    {
+      "epoch": 0.7526764967751742,
+      "grad_norm": 0.15266717970371246,
+      "learning_rate": 4.247330741890885e-06,
+      "loss": 0.8791,
+      "step": 103980
+    },
+    {
+      "epoch": 0.7527488834357604,
+      "grad_norm": 0.16882237792015076,
+      "learning_rate": 4.247258355230299e-06,
+      "loss": 0.8827,
+      "step": 103990
+    },
+    {
+      "epoch": 0.7528212700963467,
+      "grad_norm": 0.15489928424358368,
+      "learning_rate": 4.247185968569712e-06,
+      "loss": 0.8958,
+      "step": 104000
+    },
+    {
+      "epoch": 0.7528936567569329,
+      "grad_norm": 0.15682126581668854,
+      "learning_rate": 4.247113581909126e-06,
+      "loss": 0.8914,
+      "step": 104010
+    },
+    {
+      "epoch": 0.752966043417519,
+      "grad_norm": 0.1847575455904007,
+      "learning_rate": 4.24704119524854e-06,
+      "loss": 0.8897,
+      "step": 104020
+    },
+    {
+      "epoch": 0.7530384300781052,
+      "grad_norm": 0.15872369706630707,
+      "learning_rate": 4.246968808587954e-06,
+      "loss": 0.8898,
+      "step": 104030
+    },
+    {
+      "epoch": 0.7531108167386914,
+      "grad_norm": 0.16287031769752502,
+      "learning_rate": 4.246896421927368e-06,
+      "loss": 0.8857,
+      "step": 104040
+    },
+    {
+      "epoch": 0.7531832033992776,
+      "grad_norm": 0.14909902215003967,
+      "learning_rate": 4.246824035266781e-06,
+      "loss": 0.8943,
+      "step": 104050
+    },
+    {
+      "epoch": 0.7532555900598638,
+      "grad_norm": 0.1613544374704361,
+      "learning_rate": 4.246751648606195e-06,
+      "loss": 0.8926,
+      "step": 104060
+    },
+    {
+      "epoch": 0.7533279767204499,
+      "grad_norm": 0.14487162232398987,
+      "learning_rate": 4.2466792619456085e-06,
+      "loss": 0.8836,
+      "step": 104070
+    },
+    {
+      "epoch": 0.7534003633810361,
+      "grad_norm": 0.16244520246982574,
+      "learning_rate": 4.246606875285023e-06,
+      "loss": 0.8857,
+      "step": 104080
+    },
+    {
+      "epoch": 0.7534727500416223,
+      "grad_norm": 0.15103542804718018,
+      "learning_rate": 4.246534488624437e-06,
+      "loss": 0.8759,
+      "step": 104090
+    },
+    {
+      "epoch": 0.7535451367022085,
+      "grad_norm": 0.15711627900600433,
+      "learning_rate": 4.24646210196385e-06,
+      "loss": 0.8965,
+      "step": 104100
+    },
+    {
+      "epoch": 0.7536175233627948,
+      "grad_norm": 0.15874861180782318,
+      "learning_rate": 4.246389715303264e-06,
+      "loss": 0.9034,
+      "step": 104110
+    },
+    {
+      "epoch": 0.7536899100233809,
+      "grad_norm": 0.14917586743831635,
+      "learning_rate": 4.246317328642678e-06,
+      "loss": 0.8858,
+      "step": 104120
+    },
+    {
+      "epoch": 0.7537622966839671,
+      "grad_norm": 0.14904716610908508,
+      "learning_rate": 4.246244941982092e-06,
+      "loss": 0.8882,
+      "step": 104130
+    },
+    {
+      "epoch": 0.7538346833445533,
+      "grad_norm": 0.15895779430866241,
+      "learning_rate": 4.2461725553215055e-06,
+      "loss": 0.8892,
+      "step": 104140
+    },
+    {
+      "epoch": 0.7539070700051395,
+      "grad_norm": 0.2372187227010727,
+      "learning_rate": 4.246100168660919e-06,
+      "loss": 0.89,
+      "step": 104150
+    },
+    {
+      "epoch": 0.7539794566657256,
+      "grad_norm": 0.18698617815971375,
+      "learning_rate": 4.246027782000333e-06,
+      "loss": 0.9019,
+      "step": 104160
+    },
+    {
+      "epoch": 0.7540518433263118,
+      "grad_norm": 0.16438809037208557,
+      "learning_rate": 4.245955395339747e-06,
+      "loss": 0.8853,
+      "step": 104170
+    },
+    {
+      "epoch": 0.754124229986898,
+      "grad_norm": 0.17748364806175232,
+      "learning_rate": 4.245883008679161e-06,
+      "loss": 0.8724,
+      "step": 104180
+    },
+    {
+      "epoch": 0.7541966166474842,
+      "grad_norm": 0.14284367859363556,
+      "learning_rate": 4.2458106220185744e-06,
+      "loss": 0.889,
+      "step": 104190
+    },
+    {
+      "epoch": 0.7542690033080703,
+      "grad_norm": 0.15544088184833527,
+      "learning_rate": 4.245738235357988e-06,
+      "loss": 0.875,
+      "step": 104200
+    },
+    {
+      "epoch": 0.7543413899686566,
+      "grad_norm": 0.16390086710453033,
+      "learning_rate": 4.2456658486974025e-06,
+      "loss": 0.8853,
+      "step": 104210
+    },
+    {
+      "epoch": 0.7544137766292428,
+      "grad_norm": 0.15978281199932098,
+      "learning_rate": 4.245593462036816e-06,
+      "loss": 0.8848,
+      "step": 104220
+    },
+    {
+      "epoch": 0.754486163289829,
+      "grad_norm": 0.1489153504371643,
+      "learning_rate": 4.24552107537623e-06,
+      "loss": 0.8817,
+      "step": 104230
+    },
+    {
+      "epoch": 0.7545585499504152,
+      "grad_norm": 0.15438929200172424,
+      "learning_rate": 4.245448688715643e-06,
+      "loss": 0.8807,
+      "step": 104240
+    },
+    {
+      "epoch": 0.7546309366110013,
+      "grad_norm": 0.18427646160125732,
+      "learning_rate": 4.245376302055058e-06,
+      "loss": 0.89,
+      "step": 104250
+    },
+    {
+      "epoch": 0.7547033232715875,
+      "grad_norm": 0.15060459077358246,
+      "learning_rate": 4.2453039153944714e-06,
+      "loss": 0.8951,
+      "step": 104260
+    },
+    {
+      "epoch": 0.7547757099321737,
+      "grad_norm": 0.16309115290641785,
+      "learning_rate": 4.245231528733885e-06,
+      "loss": 0.8921,
+      "step": 104270
+    },
+    {
+      "epoch": 0.7548480965927599,
+      "grad_norm": 0.15942274034023285,
+      "learning_rate": 4.245159142073299e-06,
+      "loss": 0.8872,
+      "step": 104280
+    },
+    {
+      "epoch": 0.754920483253346,
+      "grad_norm": 0.1642897129058838,
+      "learning_rate": 4.245086755412713e-06,
+      "loss": 0.8889,
+      "step": 104290
+    },
+    {
+      "epoch": 0.7549928699139322,
+      "grad_norm": 0.15134483575820923,
+      "learning_rate": 4.245014368752127e-06,
+      "loss": 0.8769,
+      "step": 104300
+    },
+    {
+      "epoch": 0.7550652565745184,
+      "grad_norm": 0.16363342106342316,
+      "learning_rate": 4.24494198209154e-06,
+      "loss": 0.8751,
+      "step": 104310
+    },
+    {
+      "epoch": 0.7551376432351047,
+      "grad_norm": 0.1584373414516449,
+      "learning_rate": 4.244869595430954e-06,
+      "loss": 0.8846,
+      "step": 104320
+    },
+    {
+      "epoch": 0.7552100298956909,
+      "grad_norm": 0.15007972717285156,
+      "learning_rate": 4.2447972087703685e-06,
+      "loss": 0.8924,
+      "step": 104330
+    },
+    {
+      "epoch": 0.755282416556277,
+      "grad_norm": 0.16009405255317688,
+      "learning_rate": 4.244724822109782e-06,
+      "loss": 0.8788,
+      "step": 104340
+    },
+    {
+      "epoch": 0.7553548032168632,
+      "grad_norm": 0.14448218047618866,
+      "learning_rate": 4.244652435449196e-06,
+      "loss": 0.8778,
+      "step": 104350
+    },
+    {
+      "epoch": 0.7554271898774494,
+      "grad_norm": 0.1468927413225174,
+      "learning_rate": 4.244580048788609e-06,
+      "loss": 0.8914,
+      "step": 104360
+    },
+    {
+      "epoch": 0.7554995765380356,
+      "grad_norm": 0.14824378490447998,
+      "learning_rate": 4.244507662128024e-06,
+      "loss": 0.8868,
+      "step": 104370
+    },
+    {
+      "epoch": 0.7555719631986217,
+      "grad_norm": 0.15983250737190247,
+      "learning_rate": 4.244435275467437e-06,
+      "loss": 0.8888,
+      "step": 104380
+    },
+    {
+      "epoch": 0.7556443498592079,
+      "grad_norm": 0.15000095963478088,
+      "learning_rate": 4.244362888806851e-06,
+      "loss": 0.9001,
+      "step": 104390
+    },
+    {
+      "epoch": 0.7557167365197941,
+      "grad_norm": 0.15501455962657928,
+      "learning_rate": 4.244290502146265e-06,
+      "loss": 0.8791,
+      "step": 104400
+    },
+    {
+      "epoch": 0.7557891231803803,
+      "grad_norm": 0.14795927703380585,
+      "learning_rate": 4.244218115485679e-06,
+      "loss": 0.8769,
+      "step": 104410
+    },
+    {
+      "epoch": 0.7558615098409666,
+      "grad_norm": 0.16325373947620392,
+      "learning_rate": 4.244145728825093e-06,
+      "loss": 0.8827,
+      "step": 104420
+    },
+    {
+      "epoch": 0.7559338965015527,
+      "grad_norm": 0.15596868097782135,
+      "learning_rate": 4.244073342164506e-06,
+      "loss": 0.8761,
+      "step": 104430
+    },
+    {
+      "epoch": 0.7560062831621389,
+      "grad_norm": 0.14845876395702362,
+      "learning_rate": 4.24400095550392e-06,
+      "loss": 0.8798,
+      "step": 104440
+    },
+    {
+      "epoch": 0.7560786698227251,
+      "grad_norm": 0.1506635695695877,
+      "learning_rate": 4.243928568843334e-06,
+      "loss": 0.8831,
+      "step": 104450
+    },
+    {
+      "epoch": 0.7561510564833113,
+      "grad_norm": 0.14508011937141418,
+      "learning_rate": 4.243856182182748e-06,
+      "loss": 0.8923,
+      "step": 104460
+    },
+    {
+      "epoch": 0.7562234431438974,
+      "grad_norm": 0.16762535274028778,
+      "learning_rate": 4.243783795522162e-06,
+      "loss": 0.8903,
+      "step": 104470
+    },
+    {
+      "epoch": 0.7562958298044836,
+      "grad_norm": 0.15426135063171387,
+      "learning_rate": 4.243711408861575e-06,
+      "loss": 0.8721,
+      "step": 104480
+    },
+    {
+      "epoch": 0.7563682164650698,
+      "grad_norm": 0.15148262679576874,
+      "learning_rate": 4.24363902220099e-06,
+      "loss": 0.8973,
+      "step": 104490
+    },
+    {
+      "epoch": 0.756440603125656,
+      "grad_norm": 0.14398479461669922,
+      "learning_rate": 4.243566635540403e-06,
+      "loss": 0.8937,
+      "step": 104500
+    },
+    {
+      "epoch": 0.7565129897862422,
+      "grad_norm": 0.16111652553081512,
+      "learning_rate": 4.243494248879817e-06,
+      "loss": 0.8981,
+      "step": 104510
+    },
+    {
+      "epoch": 0.7565853764468283,
+      "grad_norm": 0.14975638687610626,
+      "learning_rate": 4.2434218622192306e-06,
+      "loss": 0.8924,
+      "step": 104520
+    },
+    {
+      "epoch": 0.7566577631074146,
+      "grad_norm": 0.1521756798028946,
+      "learning_rate": 4.243349475558645e-06,
+      "loss": 0.88,
+      "step": 104530
+    },
+    {
+      "epoch": 0.7567301497680008,
+      "grad_norm": 0.17231915891170502,
+      "learning_rate": 4.243277088898059e-06,
+      "loss": 0.8897,
+      "step": 104540
+    },
+    {
+      "epoch": 0.756802536428587,
+      "grad_norm": 0.16323207318782806,
+      "learning_rate": 4.243204702237472e-06,
+      "loss": 0.8935,
+      "step": 104550
+    },
+    {
+      "epoch": 0.7568749230891731,
+      "grad_norm": 0.14698879420757294,
+      "learning_rate": 4.243132315576886e-06,
+      "loss": 0.8892,
+      "step": 104560
+    },
+    {
+      "epoch": 0.7569473097497593,
+      "grad_norm": 0.15942469239234924,
+      "learning_rate": 4.2430599289163e-06,
+      "loss": 0.8943,
+      "step": 104570
+    },
+    {
+      "epoch": 0.7570196964103455,
+      "grad_norm": 0.16518035531044006,
+      "learning_rate": 4.242987542255713e-06,
+      "loss": 0.8947,
+      "step": 104580
+    },
+    {
+      "epoch": 0.7570920830709317,
+      "grad_norm": 0.14789389073848724,
+      "learning_rate": 4.242915155595127e-06,
+      "loss": 0.8858,
+      "step": 104590
+    },
+    {
+      "epoch": 0.7571644697315179,
+      "grad_norm": 0.1392948478460312,
+      "learning_rate": 4.242842768934541e-06,
+      "loss": 0.8891,
+      "step": 104600
+    },
+    {
+      "epoch": 0.757236856392104,
+      "grad_norm": 0.15600177645683289,
+      "learning_rate": 4.242770382273955e-06,
+      "loss": 0.9023,
+      "step": 104610
+    },
+    {
+      "epoch": 0.7573092430526902,
+      "grad_norm": 0.16960683465003967,
+      "learning_rate": 4.242697995613368e-06,
+      "loss": 0.879,
+      "step": 104620
+    },
+    {
+      "epoch": 0.7573816297132764,
+      "grad_norm": 0.15232223272323608,
+      "learning_rate": 4.242625608952782e-06,
+      "loss": 0.8809,
+      "step": 104630
+    },
+    {
+      "epoch": 0.7574540163738627,
+      "grad_norm": 0.1578724980354309,
+      "learning_rate": 4.2425532222921965e-06,
+      "loss": 0.9066,
+      "step": 104640
+    },
+    {
+      "epoch": 0.7575264030344488,
+      "grad_norm": 0.1544656753540039,
+      "learning_rate": 4.24248083563161e-06,
+      "loss": 0.8877,
+      "step": 104650
+    },
+    {
+      "epoch": 0.757598789695035,
+      "grad_norm": 0.154565691947937,
+      "learning_rate": 4.242408448971024e-06,
+      "loss": 0.8876,
+      "step": 104660
+    },
+    {
+      "epoch": 0.7576711763556212,
+      "grad_norm": 0.15601760149002075,
+      "learning_rate": 4.242336062310437e-06,
+      "loss": 0.8946,
+      "step": 104670
+    },
+    {
+      "epoch": 0.7577435630162074,
+      "grad_norm": 0.16208244860172272,
+      "learning_rate": 4.242263675649852e-06,
+      "loss": 0.8835,
+      "step": 104680
+    },
+    {
+      "epoch": 0.7578159496767936,
+      "grad_norm": 0.1622067540884018,
+      "learning_rate": 4.242191288989265e-06,
+      "loss": 0.8915,
+      "step": 104690
+    },
+    {
+      "epoch": 0.7578883363373797,
+      "grad_norm": 0.1567181497812271,
+      "learning_rate": 4.242118902328679e-06,
+      "loss": 0.8854,
+      "step": 104700
+    },
+    {
+      "epoch": 0.7579607229979659,
+      "grad_norm": 0.17146632075309753,
+      "learning_rate": 4.242046515668093e-06,
+      "loss": 0.8797,
+      "step": 104710
+    },
+    {
+      "epoch": 0.7580331096585521,
+      "grad_norm": 0.15992100536823273,
+      "learning_rate": 4.241974129007507e-06,
+      "loss": 0.8882,
+      "step": 104720
+    },
+    {
+      "epoch": 0.7581054963191383,
+      "grad_norm": 0.14678719639778137,
+      "learning_rate": 4.241901742346921e-06,
+      "loss": 0.8827,
+      "step": 104730
+    },
+    {
+      "epoch": 0.7581778829797245,
+      "grad_norm": 0.15735352039337158,
+      "learning_rate": 4.241829355686334e-06,
+      "loss": 0.8882,
+      "step": 104740
+    },
+    {
+      "epoch": 0.7582502696403107,
+      "grad_norm": 0.16120299696922302,
+      "learning_rate": 4.241756969025748e-06,
+      "loss": 0.8893,
+      "step": 104750
+    },
+    {
+      "epoch": 0.7583226563008969,
+      "grad_norm": 0.15673673152923584,
+      "learning_rate": 4.241684582365162e-06,
+      "loss": 0.8916,
+      "step": 104760
+    },
+    {
+      "epoch": 0.7583950429614831,
+      "grad_norm": 0.1934574395418167,
+      "learning_rate": 4.241612195704576e-06,
+      "loss": 0.8908,
+      "step": 104770
+    },
+    {
+      "epoch": 0.7584674296220693,
+      "grad_norm": 0.17176105082035065,
+      "learning_rate": 4.24153980904399e-06,
+      "loss": 0.8849,
+      "step": 104780
+    },
+    {
+      "epoch": 0.7585398162826554,
+      "grad_norm": 0.16662052273750305,
+      "learning_rate": 4.241467422383403e-06,
+      "loss": 0.8946,
+      "step": 104790
+    },
+    {
+      "epoch": 0.7586122029432416,
+      "grad_norm": 0.14775685966014862,
+      "learning_rate": 4.241395035722817e-06,
+      "loss": 0.8891,
+      "step": 104800
+    },
+    {
+      "epoch": 0.7586845896038278,
+      "grad_norm": 0.15374141931533813,
+      "learning_rate": 4.241322649062231e-06,
+      "loss": 0.8965,
+      "step": 104810
+    },
+    {
+      "epoch": 0.758756976264414,
+      "grad_norm": 0.15306535363197327,
+      "learning_rate": 4.241250262401645e-06,
+      "loss": 0.8876,
+      "step": 104820
+    },
+    {
+      "epoch": 0.7588293629250001,
+      "grad_norm": 0.14979711174964905,
+      "learning_rate": 4.241177875741059e-06,
+      "loss": 0.8802,
+      "step": 104830
+    },
+    {
+      "epoch": 0.7589017495855863,
+      "grad_norm": 0.15853987634181976,
+      "learning_rate": 4.241105489080472e-06,
+      "loss": 0.8827,
+      "step": 104840
+    },
+    {
+      "epoch": 0.7589741362461726,
+      "grad_norm": 0.15730775892734528,
+      "learning_rate": 4.241033102419887e-06,
+      "loss": 0.8964,
+      "step": 104850
+    },
+    {
+      "epoch": 0.7590465229067588,
+      "grad_norm": 0.1486206203699112,
+      "learning_rate": 4.2409607157593e-06,
+      "loss": 0.9028,
+      "step": 104860
+    },
+    {
+      "epoch": 0.759118909567345,
+      "grad_norm": 0.15792833268642426,
+      "learning_rate": 4.240888329098714e-06,
+      "loss": 0.8772,
+      "step": 104870
+    },
+    {
+      "epoch": 0.7591912962279311,
+      "grad_norm": 0.16453076899051666,
+      "learning_rate": 4.2408159424381275e-06,
+      "loss": 0.8917,
+      "step": 104880
+    },
+    {
+      "epoch": 0.7592636828885173,
+      "grad_norm": 0.1514229029417038,
+      "learning_rate": 4.240743555777542e-06,
+      "loss": 0.8884,
+      "step": 104890
+    },
+    {
+      "epoch": 0.7593360695491035,
+      "grad_norm": 0.15336290001869202,
+      "learning_rate": 4.240671169116956e-06,
+      "loss": 0.8828,
+      "step": 104900
+    },
+    {
+      "epoch": 0.7594084562096897,
+      "grad_norm": 0.1626693159341812,
+      "learning_rate": 4.240598782456369e-06,
+      "loss": 0.8948,
+      "step": 104910
+    },
+    {
+      "epoch": 0.7594808428702758,
+      "grad_norm": 0.15300175547599792,
+      "learning_rate": 4.240526395795783e-06,
+      "loss": 0.8984,
+      "step": 104920
+    },
+    {
+      "epoch": 0.759553229530862,
+      "grad_norm": 0.1569639891386032,
+      "learning_rate": 4.240454009135197e-06,
+      "loss": 0.869,
+      "step": 104930
+    },
+    {
+      "epoch": 0.7596256161914482,
+      "grad_norm": 0.15857458114624023,
+      "learning_rate": 4.240381622474611e-06,
+      "loss": 0.8881,
+      "step": 104940
+    },
+    {
+      "epoch": 0.7596980028520345,
+      "grad_norm": 0.15155275166034698,
+      "learning_rate": 4.2403092358140245e-06,
+      "loss": 0.8706,
+      "step": 104950
+    },
+    {
+      "epoch": 0.7597703895126207,
+      "grad_norm": 0.16041214764118195,
+      "learning_rate": 4.240236849153438e-06,
+      "loss": 0.8975,
+      "step": 104960
+    },
+    {
+      "epoch": 0.7598427761732068,
+      "grad_norm": 0.16059066355228424,
+      "learning_rate": 4.240164462492853e-06,
+      "loss": 0.8794,
+      "step": 104970
+    },
+    {
+      "epoch": 0.759915162833793,
+      "grad_norm": 0.15635879337787628,
+      "learning_rate": 4.240092075832266e-06,
+      "loss": 0.8938,
+      "step": 104980
+    },
+    {
+      "epoch": 0.7599875494943792,
+      "grad_norm": 0.14419741928577423,
+      "learning_rate": 4.24001968917168e-06,
+      "loss": 0.8759,
+      "step": 104990
+    },
+    {
+      "epoch": 0.7600599361549654,
+      "grad_norm": 0.18103042244911194,
+      "learning_rate": 4.2399473025110934e-06,
+      "loss": 0.8966,
+      "step": 105000
+    },
+    {
+      "epoch": 0.7601323228155515,
+      "grad_norm": 0.15891040861606598,
+      "learning_rate": 4.239874915850508e-06,
+      "loss": 0.8981,
+      "step": 105010
+    },
+    {
+      "epoch": 0.7602047094761377,
+      "grad_norm": 0.17460259795188904,
+      "learning_rate": 4.2398025291899215e-06,
+      "loss": 0.8862,
+      "step": 105020
+    },
+    {
+      "epoch": 0.7602770961367239,
+      "grad_norm": 0.17469242215156555,
+      "learning_rate": 4.239730142529335e-06,
+      "loss": 0.8902,
+      "step": 105030
+    },
+    {
+      "epoch": 0.7603494827973101,
+      "grad_norm": 0.1552739441394806,
+      "learning_rate": 4.239657755868749e-06,
+      "loss": 0.8872,
+      "step": 105040
+    },
+    {
+      "epoch": 0.7604218694578962,
+      "grad_norm": 0.16493083536624908,
+      "learning_rate": 4.239585369208163e-06,
+      "loss": 0.8876,
+      "step": 105050
+    },
+    {
+      "epoch": 0.7604942561184825,
+      "grad_norm": 0.1514502614736557,
+      "learning_rate": 4.239512982547577e-06,
+      "loss": 0.876,
+      "step": 105060
+    },
+    {
+      "epoch": 0.7605666427790687,
+      "grad_norm": 0.16472308337688446,
+      "learning_rate": 4.2394405958869905e-06,
+      "loss": 0.8893,
+      "step": 105070
+    },
+    {
+      "epoch": 0.7606390294396549,
+      "grad_norm": 0.15388494729995728,
+      "learning_rate": 4.239368209226404e-06,
+      "loss": 0.8934,
+      "step": 105080
+    },
+    {
+      "epoch": 0.7607114161002411,
+      "grad_norm": 0.1528458446264267,
+      "learning_rate": 4.2392958225658185e-06,
+      "loss": 0.8948,
+      "step": 105090
+    },
+    {
+      "epoch": 0.7607838027608272,
+      "grad_norm": 0.1468532532453537,
+      "learning_rate": 4.239223435905232e-06,
+      "loss": 0.8768,
+      "step": 105100
+    },
+    {
+      "epoch": 0.7608561894214134,
+      "grad_norm": 0.14190086722373962,
+      "learning_rate": 4.239151049244645e-06,
+      "loss": 0.8829,
+      "step": 105110
+    },
+    {
+      "epoch": 0.7609285760819996,
+      "grad_norm": 0.1714107096195221,
+      "learning_rate": 4.239078662584059e-06,
+      "loss": 0.8895,
+      "step": 105120
+    },
+    {
+      "epoch": 0.7610009627425858,
+      "grad_norm": 0.15026678144931793,
+      "learning_rate": 4.239006275923473e-06,
+      "loss": 0.895,
+      "step": 105130
+    },
+    {
+      "epoch": 0.761073349403172,
+      "grad_norm": 0.16360147297382355,
+      "learning_rate": 4.238933889262887e-06,
+      "loss": 0.8825,
+      "step": 105140
+    },
+    {
+      "epoch": 0.7611457360637581,
+      "grad_norm": 0.17714418470859528,
+      "learning_rate": 4.2388615026023e-06,
+      "loss": 0.8734,
+      "step": 105150
+    },
+    {
+      "epoch": 0.7612181227243443,
+      "grad_norm": 0.1679549664258957,
+      "learning_rate": 4.238789115941715e-06,
+      "loss": 0.8955,
+      "step": 105160
+    },
+    {
+      "epoch": 0.7612905093849306,
+      "grad_norm": 0.1560448259115219,
+      "learning_rate": 4.238716729281128e-06,
+      "loss": 0.893,
+      "step": 105170
+    },
+    {
+      "epoch": 0.7613628960455168,
+      "grad_norm": 0.15543989837169647,
+      "learning_rate": 4.238644342620542e-06,
+      "loss": 0.8896,
+      "step": 105180
+    },
+    {
+      "epoch": 0.761435282706103,
+      "grad_norm": 0.15329879522323608,
+      "learning_rate": 4.2385719559599555e-06,
+      "loss": 0.8894,
+      "step": 105190
+    },
+    {
+      "epoch": 0.7615076693666891,
+      "grad_norm": 0.15145981311798096,
+      "learning_rate": 4.23849956929937e-06,
+      "loss": 0.8888,
+      "step": 105200
+    },
+    {
+      "epoch": 0.7615800560272753,
+      "grad_norm": 0.14573727548122406,
+      "learning_rate": 4.238427182638784e-06,
+      "loss": 0.8841,
+      "step": 105210
+    },
+    {
+      "epoch": 0.7616524426878615,
+      "grad_norm": 0.15317559242248535,
+      "learning_rate": 4.238354795978197e-06,
+      "loss": 0.8869,
+      "step": 105220
+    },
+    {
+      "epoch": 0.7617248293484477,
+      "grad_norm": 0.17463774979114532,
+      "learning_rate": 4.238282409317611e-06,
+      "loss": 0.8971,
+      "step": 105230
+    },
+    {
+      "epoch": 0.7617972160090338,
+      "grad_norm": 0.14841240644454956,
+      "learning_rate": 4.238210022657025e-06,
+      "loss": 0.8845,
+      "step": 105240
+    },
+    {
+      "epoch": 0.76186960266962,
+      "grad_norm": 0.19012483954429626,
+      "learning_rate": 4.238137635996439e-06,
+      "loss": 0.8936,
+      "step": 105250
+    },
+    {
+      "epoch": 0.7619419893302062,
+      "grad_norm": 0.2200062870979309,
+      "learning_rate": 4.2380652493358525e-06,
+      "loss": 0.8938,
+      "step": 105260
+    },
+    {
+      "epoch": 0.7620143759907925,
+      "grad_norm": 0.16414809226989746,
+      "learning_rate": 4.237992862675266e-06,
+      "loss": 0.8925,
+      "step": 105270
+    },
+    {
+      "epoch": 0.7620867626513786,
+      "grad_norm": 0.16722236573696136,
+      "learning_rate": 4.237920476014681e-06,
+      "loss": 0.8895,
+      "step": 105280
+    },
+    {
+      "epoch": 0.7621591493119648,
+      "grad_norm": 0.17089565098285675,
+      "learning_rate": 4.237848089354094e-06,
+      "loss": 0.8922,
+      "step": 105290
+    },
+    {
+      "epoch": 0.762231535972551,
+      "grad_norm": 0.1593043953180313,
+      "learning_rate": 4.237775702693508e-06,
+      "loss": 0.8754,
+      "step": 105300
+    },
+    {
+      "epoch": 0.7623039226331372,
+      "grad_norm": 0.16244159638881683,
+      "learning_rate": 4.2377033160329215e-06,
+      "loss": 0.8918,
+      "step": 105310
+    },
+    {
+      "epoch": 0.7623763092937234,
+      "grad_norm": 0.155963733792305,
+      "learning_rate": 4.237630929372336e-06,
+      "loss": 0.8795,
+      "step": 105320
+    },
+    {
+      "epoch": 0.7624486959543095,
+      "grad_norm": 0.15185341238975525,
+      "learning_rate": 4.2375585427117496e-06,
+      "loss": 0.8946,
+      "step": 105330
+    },
+    {
+      "epoch": 0.7625210826148957,
+      "grad_norm": 0.16116154193878174,
+      "learning_rate": 4.237486156051163e-06,
+      "loss": 0.8874,
+      "step": 105340
+    },
+    {
+      "epoch": 0.7625934692754819,
+      "grad_norm": 0.15430016815662384,
+      "learning_rate": 4.237413769390577e-06,
+      "loss": 0.8861,
+      "step": 105350
+    },
+    {
+      "epoch": 0.7626658559360681,
+      "grad_norm": 0.15945422649383545,
+      "learning_rate": 4.237341382729991e-06,
+      "loss": 0.8794,
+      "step": 105360
+    },
+    {
+      "epoch": 0.7627382425966542,
+      "grad_norm": 0.1454370766878128,
+      "learning_rate": 4.237268996069405e-06,
+      "loss": 0.8782,
+      "step": 105370
+    },
+    {
+      "epoch": 0.7628106292572405,
+      "grad_norm": 0.14304585754871368,
+      "learning_rate": 4.2371966094088185e-06,
+      "loss": 0.8795,
+      "step": 105380
+    },
+    {
+      "epoch": 0.7628830159178267,
+      "grad_norm": 0.165323406457901,
+      "learning_rate": 4.237124222748232e-06,
+      "loss": 0.8894,
+      "step": 105390
+    },
+    {
+      "epoch": 0.7629554025784129,
+      "grad_norm": 0.1540648490190506,
+      "learning_rate": 4.2370518360876466e-06,
+      "loss": 0.8751,
+      "step": 105400
+    },
+    {
+      "epoch": 0.763027789238999,
+      "grad_norm": 0.1511969268321991,
+      "learning_rate": 4.23697944942706e-06,
+      "loss": 0.8812,
+      "step": 105410
+    },
+    {
+      "epoch": 0.7631001758995852,
+      "grad_norm": 0.15881909430027008,
+      "learning_rate": 4.236907062766474e-06,
+      "loss": 0.8935,
+      "step": 105420
+    },
+    {
+      "epoch": 0.7631725625601714,
+      "grad_norm": 0.1552649736404419,
+      "learning_rate": 4.236834676105887e-06,
+      "loss": 0.8867,
+      "step": 105430
+    },
+    {
+      "epoch": 0.7632449492207576,
+      "grad_norm": 0.16431519389152527,
+      "learning_rate": 4.236762289445301e-06,
+      "loss": 0.8761,
+      "step": 105440
+    },
+    {
+      "epoch": 0.7633173358813438,
+      "grad_norm": 0.17564263939857483,
+      "learning_rate": 4.2366899027847155e-06,
+      "loss": 0.8907,
+      "step": 105450
+    },
+    {
+      "epoch": 0.7633897225419299,
+      "grad_norm": 0.15047433972358704,
+      "learning_rate": 4.236617516124129e-06,
+      "loss": 0.8874,
+      "step": 105460
+    },
+    {
+      "epoch": 0.7634621092025161,
+      "grad_norm": 0.2053581178188324,
+      "learning_rate": 4.236545129463543e-06,
+      "loss": 0.8866,
+      "step": 105470
+    },
+    {
+      "epoch": 0.7635344958631024,
+      "grad_norm": 0.16297610104084015,
+      "learning_rate": 4.236472742802956e-06,
+      "loss": 0.8818,
+      "step": 105480
+    },
+    {
+      "epoch": 0.7636068825236886,
+      "grad_norm": 0.1815577745437622,
+      "learning_rate": 4.236400356142371e-06,
+      "loss": 0.8894,
+      "step": 105490
+    },
+    {
+      "epoch": 0.7636792691842748,
+      "grad_norm": 0.14771369099617004,
+      "learning_rate": 4.236327969481784e-06,
+      "loss": 0.8805,
+      "step": 105500
+    },
+    {
+      "epoch": 0.7637516558448609,
+      "grad_norm": 0.22556520998477936,
+      "learning_rate": 4.236255582821198e-06,
+      "loss": 0.8771,
+      "step": 105510
+    },
+    {
+      "epoch": 0.7638240425054471,
+      "grad_norm": 0.15500441193580627,
+      "learning_rate": 4.236183196160612e-06,
+      "loss": 0.8816,
+      "step": 105520
+    },
+    {
+      "epoch": 0.7638964291660333,
+      "grad_norm": 0.18774858117103577,
+      "learning_rate": 4.236110809500026e-06,
+      "loss": 0.8809,
+      "step": 105530
+    },
+    {
+      "epoch": 0.7639688158266195,
+      "grad_norm": 0.15794971585273743,
+      "learning_rate": 4.23603842283944e-06,
+      "loss": 0.886,
+      "step": 105540
+    },
+    {
+      "epoch": 0.7640412024872056,
+      "grad_norm": 0.16550391912460327,
+      "learning_rate": 4.235966036178853e-06,
+      "loss": 0.8963,
+      "step": 105550
+    },
+    {
+      "epoch": 0.7641135891477918,
+      "grad_norm": 0.15253488719463348,
+      "learning_rate": 4.235893649518267e-06,
+      "loss": 0.8904,
+      "step": 105560
+    },
+    {
+      "epoch": 0.764185975808378,
+      "grad_norm": 0.16290663182735443,
+      "learning_rate": 4.2358212628576814e-06,
+      "loss": 0.8908,
+      "step": 105570
+    },
+    {
+      "epoch": 0.7642583624689642,
+      "grad_norm": 0.16055020689964294,
+      "learning_rate": 4.235748876197095e-06,
+      "loss": 0.8807,
+      "step": 105580
+    },
+    {
+      "epoch": 0.7643307491295505,
+      "grad_norm": 0.1527642160654068,
+      "learning_rate": 4.235676489536509e-06,
+      "loss": 0.8882,
+      "step": 105590
+    },
+    {
+      "epoch": 0.7644031357901366,
+      "grad_norm": 0.1476249247789383,
+      "learning_rate": 4.235604102875922e-06,
+      "loss": 0.8715,
+      "step": 105600
+    },
+    {
+      "epoch": 0.7644755224507228,
+      "grad_norm": 0.15459424257278442,
+      "learning_rate": 4.235531716215337e-06,
+      "loss": 0.8853,
+      "step": 105610
+    },
+    {
+      "epoch": 0.764547909111309,
+      "grad_norm": 0.17739763855934143,
+      "learning_rate": 4.23545932955475e-06,
+      "loss": 0.8846,
+      "step": 105620
+    },
+    {
+      "epoch": 0.7646202957718952,
+      "grad_norm": 0.158504918217659,
+      "learning_rate": 4.235386942894164e-06,
+      "loss": 0.8892,
+      "step": 105630
+    },
+    {
+      "epoch": 0.7646926824324813,
+      "grad_norm": 0.16947391629219055,
+      "learning_rate": 4.235314556233578e-06,
+      "loss": 0.8886,
+      "step": 105640
+    },
+    {
+      "epoch": 0.7647650690930675,
+      "grad_norm": 0.15372443199157715,
+      "learning_rate": 4.235242169572991e-06,
+      "loss": 0.8757,
+      "step": 105650
+    },
+    {
+      "epoch": 0.7648374557536537,
+      "grad_norm": 0.16209058463573456,
+      "learning_rate": 4.235169782912405e-06,
+      "loss": 0.8973,
+      "step": 105660
+    },
+    {
+      "epoch": 0.7649098424142399,
+      "grad_norm": 0.15519694983959198,
+      "learning_rate": 4.2350973962518184e-06,
+      "loss": 0.8926,
+      "step": 105670
+    },
+    {
+      "epoch": 0.764982229074826,
+      "grad_norm": 0.15895330905914307,
+      "learning_rate": 4.235025009591233e-06,
+      "loss": 0.8882,
+      "step": 105680
+    },
+    {
+      "epoch": 0.7650546157354122,
+      "grad_norm": 0.14420805871486664,
+      "learning_rate": 4.2349526229306465e-06,
+      "loss": 0.8822,
+      "step": 105690
+    },
+    {
+      "epoch": 0.7651270023959985,
+      "grad_norm": 0.15106813609600067,
+      "learning_rate": 4.23488023627006e-06,
+      "loss": 0.8966,
+      "step": 105700
+    },
+    {
+      "epoch": 0.7651993890565847,
+      "grad_norm": 0.14976173639297485,
+      "learning_rate": 4.234807849609474e-06,
+      "loss": 0.8817,
+      "step": 105710
+    },
+    {
+      "epoch": 0.7652717757171709,
+      "grad_norm": 0.16191081702709198,
+      "learning_rate": 4.234735462948888e-06,
+      "loss": 0.8792,
+      "step": 105720
+    },
+    {
+      "epoch": 0.765344162377757,
+      "grad_norm": 0.15428221225738525,
+      "learning_rate": 4.234663076288302e-06,
+      "loss": 0.8836,
+      "step": 105730
+    },
+    {
+      "epoch": 0.7654165490383432,
+      "grad_norm": 0.15489520132541656,
+      "learning_rate": 4.2345906896277154e-06,
+      "loss": 0.8786,
+      "step": 105740
+    },
+    {
+      "epoch": 0.7654889356989294,
+      "grad_norm": 0.16444027423858643,
+      "learning_rate": 4.234518302967129e-06,
+      "loss": 0.883,
+      "step": 105750
+    },
+    {
+      "epoch": 0.7655613223595156,
+      "grad_norm": 0.1668035089969635,
+      "learning_rate": 4.2344459163065435e-06,
+      "loss": 0.893,
+      "step": 105760
+    },
+    {
+      "epoch": 0.7656337090201017,
+      "grad_norm": 0.15421725809574127,
+      "learning_rate": 4.234373529645957e-06,
+      "loss": 0.8888,
+      "step": 105770
+    },
+    {
+      "epoch": 0.7657060956806879,
+      "grad_norm": 0.15628722310066223,
+      "learning_rate": 4.234301142985371e-06,
+      "loss": 0.8877,
+      "step": 105780
+    },
+    {
+      "epoch": 0.7657784823412741,
+      "grad_norm": 0.16434313356876373,
+      "learning_rate": 4.234228756324784e-06,
+      "loss": 0.8856,
+      "step": 105790
+    },
+    {
+      "epoch": 0.7658508690018604,
+      "grad_norm": 0.14600999653339386,
+      "learning_rate": 4.234156369664199e-06,
+      "loss": 0.8916,
+      "step": 105800
+    },
+    {
+      "epoch": 0.7659232556624466,
+      "grad_norm": 0.23364056646823883,
+      "learning_rate": 4.2340839830036125e-06,
+      "loss": 0.8865,
+      "step": 105810
+    },
+    {
+      "epoch": 0.7659956423230327,
+      "grad_norm": 0.15147335827350616,
+      "learning_rate": 4.234011596343026e-06,
+      "loss": 0.875,
+      "step": 105820
+    },
+    {
+      "epoch": 0.7660680289836189,
+      "grad_norm": 0.16839240491390228,
+      "learning_rate": 4.23393920968244e-06,
+      "loss": 0.8807,
+      "step": 105830
+    },
+    {
+      "epoch": 0.7661404156442051,
+      "grad_norm": 0.1934797465801239,
+      "learning_rate": 4.233866823021854e-06,
+      "loss": 0.8983,
+      "step": 105840
+    },
+    {
+      "epoch": 0.7662128023047913,
+      "grad_norm": 0.15738998353481293,
+      "learning_rate": 4.233794436361268e-06,
+      "loss": 0.8784,
+      "step": 105850
+    },
+    {
+      "epoch": 0.7662851889653775,
+      "grad_norm": 0.14478467404842377,
+      "learning_rate": 4.233722049700681e-06,
+      "loss": 0.896,
+      "step": 105860
+    },
+    {
+      "epoch": 0.7663575756259636,
+      "grad_norm": 0.17345765233039856,
+      "learning_rate": 4.233649663040095e-06,
+      "loss": 0.8857,
+      "step": 105870
+    },
+    {
+      "epoch": 0.7664299622865498,
+      "grad_norm": 0.15138719975948334,
+      "learning_rate": 4.2335772763795095e-06,
+      "loss": 0.8979,
+      "step": 105880
+    },
+    {
+      "epoch": 0.766502348947136,
+      "grad_norm": 0.15025945007801056,
+      "learning_rate": 4.233504889718923e-06,
+      "loss": 0.8786,
+      "step": 105890
+    },
+    {
+      "epoch": 0.7665747356077222,
+      "grad_norm": 0.16647055745124817,
+      "learning_rate": 4.233432503058337e-06,
+      "loss": 0.897,
+      "step": 105900
+    },
+    {
+      "epoch": 0.7666471222683084,
+      "grad_norm": 0.16044731438159943,
+      "learning_rate": 4.23336011639775e-06,
+      "loss": 0.8921,
+      "step": 105910
+    },
+    {
+      "epoch": 0.7667195089288946,
+      "grad_norm": 0.1512782722711563,
+      "learning_rate": 4.233287729737165e-06,
+      "loss": 0.8839,
+      "step": 105920
+    },
+    {
+      "epoch": 0.7667918955894808,
+      "grad_norm": 0.16747620701789856,
+      "learning_rate": 4.233215343076578e-06,
+      "loss": 0.8857,
+      "step": 105930
+    },
+    {
+      "epoch": 0.766864282250067,
+      "grad_norm": 0.17304562032222748,
+      "learning_rate": 4.233142956415992e-06,
+      "loss": 0.8897,
+      "step": 105940
+    },
+    {
+      "epoch": 0.7669366689106532,
+      "grad_norm": 0.15581056475639343,
+      "learning_rate": 4.233070569755406e-06,
+      "loss": 0.8906,
+      "step": 105950
+    },
+    {
+      "epoch": 0.7670090555712393,
+      "grad_norm": 0.17081882059574127,
+      "learning_rate": 4.23299818309482e-06,
+      "loss": 0.8815,
+      "step": 105960
+    },
+    {
+      "epoch": 0.7670814422318255,
+      "grad_norm": 0.15507552027702332,
+      "learning_rate": 4.232925796434234e-06,
+      "loss": 0.8847,
+      "step": 105970
+    },
+    {
+      "epoch": 0.7671538288924117,
+      "grad_norm": 0.16203370690345764,
+      "learning_rate": 4.232853409773647e-06,
+      "loss": 0.8817,
+      "step": 105980
+    },
+    {
+      "epoch": 0.7672262155529979,
+      "grad_norm": 0.1585424691438675,
+      "learning_rate": 4.232781023113061e-06,
+      "loss": 0.8779,
+      "step": 105990
+    },
+    {
+      "epoch": 0.767298602213584,
+      "grad_norm": 0.15579846501350403,
+      "learning_rate": 4.232708636452475e-06,
+      "loss": 0.8773,
+      "step": 106000
+    },
+    {
+      "epoch": 0.7673709888741703,
+      "grad_norm": 0.1483168751001358,
+      "learning_rate": 4.232636249791889e-06,
+      "loss": 0.8916,
+      "step": 106010
+    },
+    {
+      "epoch": 0.7674433755347565,
+      "grad_norm": 0.18050970137119293,
+      "learning_rate": 4.232563863131303e-06,
+      "loss": 0.8731,
+      "step": 106020
+    },
+    {
+      "epoch": 0.7675157621953427,
+      "grad_norm": 0.17349663376808167,
+      "learning_rate": 4.232491476470716e-06,
+      "loss": 0.883,
+      "step": 106030
+    },
+    {
+      "epoch": 0.7675881488559289,
+      "grad_norm": 0.17154832184314728,
+      "learning_rate": 4.23241908981013e-06,
+      "loss": 0.8769,
+      "step": 106040
+    },
+    {
+      "epoch": 0.767660535516515,
+      "grad_norm": 0.15172946453094482,
+      "learning_rate": 4.232346703149544e-06,
+      "loss": 0.8715,
+      "step": 106050
+    },
+    {
+      "epoch": 0.7677329221771012,
+      "grad_norm": 0.16642993688583374,
+      "learning_rate": 4.232274316488958e-06,
+      "loss": 0.8857,
+      "step": 106060
+    },
+    {
+      "epoch": 0.7678053088376874,
+      "grad_norm": 0.15337328612804413,
+      "learning_rate": 4.2322019298283716e-06,
+      "loss": 0.872,
+      "step": 106070
+    },
+    {
+      "epoch": 0.7678776954982736,
+      "grad_norm": 0.15362398326396942,
+      "learning_rate": 4.232129543167785e-06,
+      "loss": 0.8852,
+      "step": 106080
+    },
+    {
+      "epoch": 0.7679500821588597,
+      "grad_norm": 0.16665634512901306,
+      "learning_rate": 4.2320571565072e-06,
+      "loss": 0.8877,
+      "step": 106090
+    },
+    {
+      "epoch": 0.7680224688194459,
+      "grad_norm": 0.14505624771118164,
+      "learning_rate": 4.231984769846613e-06,
+      "loss": 0.8936,
+      "step": 106100
+    },
+    {
+      "epoch": 0.7680948554800321,
+      "grad_norm": 0.16204003989696503,
+      "learning_rate": 4.231912383186027e-06,
+      "loss": 0.8903,
+      "step": 106110
+    },
+    {
+      "epoch": 0.7681672421406184,
+      "grad_norm": 0.16394628584384918,
+      "learning_rate": 4.2318399965254405e-06,
+      "loss": 0.8986,
+      "step": 106120
+    },
+    {
+      "epoch": 0.7682396288012046,
+      "grad_norm": 0.15969203412532806,
+      "learning_rate": 4.231767609864855e-06,
+      "loss": 0.8769,
+      "step": 106130
+    },
+    {
+      "epoch": 0.7683120154617907,
+      "grad_norm": 0.1562148928642273,
+      "learning_rate": 4.2316952232042686e-06,
+      "loss": 0.872,
+      "step": 106140
+    },
+    {
+      "epoch": 0.7683844021223769,
+      "grad_norm": 0.171428382396698,
+      "learning_rate": 4.231622836543682e-06,
+      "loss": 0.8811,
+      "step": 106150
+    },
+    {
+      "epoch": 0.7684567887829631,
+      "grad_norm": 0.16853584349155426,
+      "learning_rate": 4.231550449883096e-06,
+      "loss": 0.8836,
+      "step": 106160
+    },
+    {
+      "epoch": 0.7685291754435493,
+      "grad_norm": 0.15788713097572327,
+      "learning_rate": 4.23147806322251e-06,
+      "loss": 0.8912,
+      "step": 106170
+    },
+    {
+      "epoch": 0.7686015621041354,
+      "grad_norm": 0.15766339004039764,
+      "learning_rate": 4.231405676561923e-06,
+      "loss": 0.8992,
+      "step": 106180
+    },
+    {
+      "epoch": 0.7686739487647216,
+      "grad_norm": 0.14977054297924042,
+      "learning_rate": 4.2313332899013375e-06,
+      "loss": 0.8844,
+      "step": 106190
+    },
+    {
+      "epoch": 0.7687463354253078,
+      "grad_norm": 0.15274566411972046,
+      "learning_rate": 4.231260903240751e-06,
+      "loss": 0.8939,
+      "step": 106200
+    },
+    {
+      "epoch": 0.768818722085894,
+      "grad_norm": 0.17646168172359467,
+      "learning_rate": 4.231188516580165e-06,
+      "loss": 0.8851,
+      "step": 106210
+    },
+    {
+      "epoch": 0.7688911087464801,
+      "grad_norm": 0.16536547243595123,
+      "learning_rate": 4.231116129919578e-06,
+      "loss": 0.879,
+      "step": 106220
+    },
+    {
+      "epoch": 0.7689634954070664,
+      "grad_norm": 0.17476871609687805,
+      "learning_rate": 4.231043743258992e-06,
+      "loss": 0.8759,
+      "step": 106230
+    },
+    {
+      "epoch": 0.7690358820676526,
+      "grad_norm": 0.1473916620016098,
+      "learning_rate": 4.230971356598406e-06,
+      "loss": 0.8778,
+      "step": 106240
+    },
+    {
+      "epoch": 0.7691082687282388,
+      "grad_norm": 0.15702074766159058,
+      "learning_rate": 4.23089896993782e-06,
+      "loss": 0.8932,
+      "step": 106250
+    },
+    {
+      "epoch": 0.769180655388825,
+      "grad_norm": 0.14984412491321564,
+      "learning_rate": 4.230826583277234e-06,
+      "loss": 0.8931,
+      "step": 106260
+    },
+    {
+      "epoch": 0.7692530420494111,
+      "grad_norm": 0.15122853219509125,
+      "learning_rate": 4.230754196616647e-06,
+      "loss": 0.8831,
+      "step": 106270
+    },
+    {
+      "epoch": 0.7693254287099973,
+      "grad_norm": 0.16905367374420166,
+      "learning_rate": 4.230681809956062e-06,
+      "loss": 0.8764,
+      "step": 106280
+    },
+    {
+      "epoch": 0.7693978153705835,
+      "grad_norm": 0.15527789294719696,
+      "learning_rate": 4.230609423295475e-06,
+      "loss": 0.8804,
+      "step": 106290
+    },
+    {
+      "epoch": 0.7694702020311697,
+      "grad_norm": 0.16785407066345215,
+      "learning_rate": 4.230537036634889e-06,
+      "loss": 0.8924,
+      "step": 106300
+    },
+    {
+      "epoch": 0.7695425886917558,
+      "grad_norm": 0.15901590883731842,
+      "learning_rate": 4.230464649974303e-06,
+      "loss": 0.8922,
+      "step": 106310
+    },
+    {
+      "epoch": 0.769614975352342,
+      "grad_norm": 0.17190630733966827,
+      "learning_rate": 4.230392263313717e-06,
+      "loss": 0.8802,
+      "step": 106320
+    },
+    {
+      "epoch": 0.7696873620129283,
+      "grad_norm": 0.27805203199386597,
+      "learning_rate": 4.230319876653131e-06,
+      "loss": 0.8847,
+      "step": 106330
+    },
+    {
+      "epoch": 0.7697597486735145,
+      "grad_norm": 0.1641608327627182,
+      "learning_rate": 4.230247489992544e-06,
+      "loss": 0.8951,
+      "step": 106340
+    },
+    {
+      "epoch": 0.7698321353341007,
+      "grad_norm": 0.15645158290863037,
+      "learning_rate": 4.230175103331958e-06,
+      "loss": 0.8771,
+      "step": 106350
+    },
+    {
+      "epoch": 0.7699045219946868,
+      "grad_norm": 0.15807659924030304,
+      "learning_rate": 4.230102716671372e-06,
+      "loss": 0.8841,
+      "step": 106360
+    },
+    {
+      "epoch": 0.769976908655273,
+      "grad_norm": 0.15633799135684967,
+      "learning_rate": 4.230030330010786e-06,
+      "loss": 0.8917,
+      "step": 106370
+    },
+    {
+      "epoch": 0.7700492953158592,
+      "grad_norm": 0.15708708763122559,
+      "learning_rate": 4.2299579433502e-06,
+      "loss": 0.8789,
+      "step": 106380
+    },
+    {
+      "epoch": 0.7701216819764454,
+      "grad_norm": 0.15080325305461884,
+      "learning_rate": 4.229885556689613e-06,
+      "loss": 0.8861,
+      "step": 106390
+    },
+    {
+      "epoch": 0.7701940686370315,
+      "grad_norm": 0.14527060091495514,
+      "learning_rate": 4.229813170029028e-06,
+      "loss": 0.8835,
+      "step": 106400
+    },
+    {
+      "epoch": 0.7702664552976177,
+      "grad_norm": 0.1920609176158905,
+      "learning_rate": 4.229740783368441e-06,
+      "loss": 0.8916,
+      "step": 106410
+    },
+    {
+      "epoch": 0.7703388419582039,
+      "grad_norm": 0.14528389275074005,
+      "learning_rate": 4.229668396707855e-06,
+      "loss": 0.8716,
+      "step": 106420
+    },
+    {
+      "epoch": 0.7704112286187901,
+      "grad_norm": 0.1617295742034912,
+      "learning_rate": 4.2295960100472685e-06,
+      "loss": 0.8786,
+      "step": 106430
+    },
+    {
+      "epoch": 0.7704836152793764,
+      "grad_norm": 0.15510478615760803,
+      "learning_rate": 4.229523623386683e-06,
+      "loss": 0.8832,
+      "step": 106440
+    },
+    {
+      "epoch": 0.7705560019399625,
+      "grad_norm": 0.1545783132314682,
+      "learning_rate": 4.229451236726097e-06,
+      "loss": 0.8805,
+      "step": 106450
+    },
+    {
+      "epoch": 0.7706283886005487,
+      "grad_norm": 0.15689776837825775,
+      "learning_rate": 4.22937885006551e-06,
+      "loss": 0.8718,
+      "step": 106460
+    },
+    {
+      "epoch": 0.7707007752611349,
+      "grad_norm": 0.1707046777009964,
+      "learning_rate": 4.229306463404924e-06,
+      "loss": 0.8843,
+      "step": 106470
+    },
+    {
+      "epoch": 0.7707731619217211,
+      "grad_norm": 0.16164378821849823,
+      "learning_rate": 4.229234076744338e-06,
+      "loss": 0.8876,
+      "step": 106480
+    },
+    {
+      "epoch": 0.7708455485823072,
+      "grad_norm": 0.16344380378723145,
+      "learning_rate": 4.229161690083752e-06,
+      "loss": 0.8894,
+      "step": 106490
+    },
+    {
+      "epoch": 0.7709179352428934,
+      "grad_norm": 0.15482939779758453,
+      "learning_rate": 4.2290893034231655e-06,
+      "loss": 0.887,
+      "step": 106500
+    },
+    {
+      "epoch": 0.7709903219034796,
+      "grad_norm": 0.16493044793605804,
+      "learning_rate": 4.229016916762579e-06,
+      "loss": 0.8868,
+      "step": 106510
+    },
+    {
+      "epoch": 0.7710627085640658,
+      "grad_norm": 0.15701046586036682,
+      "learning_rate": 4.228944530101994e-06,
+      "loss": 0.8802,
+      "step": 106520
+    },
+    {
+      "epoch": 0.771135095224652,
+      "grad_norm": 0.15896005928516388,
+      "learning_rate": 4.228872143441407e-06,
+      "loss": 0.8828,
+      "step": 106530
+    },
+    {
+      "epoch": 0.7712074818852381,
+      "grad_norm": 0.15199479460716248,
+      "learning_rate": 4.228799756780821e-06,
+      "loss": 0.8855,
+      "step": 106540
+    },
+    {
+      "epoch": 0.7712798685458244,
+      "grad_norm": 0.14489343762397766,
+      "learning_rate": 4.2287273701202345e-06,
+      "loss": 0.8969,
+      "step": 106550
+    },
+    {
+      "epoch": 0.7713522552064106,
+      "grad_norm": 0.1755223423242569,
+      "learning_rate": 4.228654983459649e-06,
+      "loss": 0.8788,
+      "step": 106560
+    },
+    {
+      "epoch": 0.7714246418669968,
+      "grad_norm": 0.17546701431274414,
+      "learning_rate": 4.2285825967990625e-06,
+      "loss": 0.8877,
+      "step": 106570
+    },
+    {
+      "epoch": 0.771497028527583,
+      "grad_norm": 0.1525055468082428,
+      "learning_rate": 4.228510210138476e-06,
+      "loss": 0.8842,
+      "step": 106580
+    },
+    {
+      "epoch": 0.7715694151881691,
+      "grad_norm": 0.14916154742240906,
+      "learning_rate": 4.22843782347789e-06,
+      "loss": 0.8826,
+      "step": 106590
+    },
+    {
+      "epoch": 0.7716418018487553,
+      "grad_norm": 0.16282054781913757,
+      "learning_rate": 4.228365436817304e-06,
+      "loss": 0.8843,
+      "step": 106600
+    },
+    {
+      "epoch": 0.7717141885093415,
+      "grad_norm": 0.16194520890712738,
+      "learning_rate": 4.228293050156718e-06,
+      "loss": 0.8881,
+      "step": 106610
+    },
+    {
+      "epoch": 0.7717865751699277,
+      "grad_norm": 0.1717979609966278,
+      "learning_rate": 4.2282206634961315e-06,
+      "loss": 0.8785,
+      "step": 106620
+    },
+    {
+      "epoch": 0.7718589618305138,
+      "grad_norm": 0.24075248837471008,
+      "learning_rate": 4.228148276835545e-06,
+      "loss": 0.8907,
+      "step": 106630
+    },
+    {
+      "epoch": 0.7719313484911,
+      "grad_norm": 0.3043558597564697,
+      "learning_rate": 4.2280758901749595e-06,
+      "loss": 0.8813,
+      "step": 106640
+    },
+    {
+      "epoch": 0.7720037351516863,
+      "grad_norm": 0.17792464792728424,
+      "learning_rate": 4.228003503514373e-06,
+      "loss": 0.8881,
+      "step": 106650
+    },
+    {
+      "epoch": 0.7720761218122725,
+      "grad_norm": 0.16710121929645538,
+      "learning_rate": 4.227931116853787e-06,
+      "loss": 0.8809,
+      "step": 106660
+    },
+    {
+      "epoch": 0.7721485084728587,
+      "grad_norm": 0.14636112749576569,
+      "learning_rate": 4.2278587301932e-06,
+      "loss": 0.8869,
+      "step": 106670
+    },
+    {
+      "epoch": 0.7722208951334448,
+      "grad_norm": 0.16227993369102478,
+      "learning_rate": 4.227786343532614e-06,
+      "loss": 0.8696,
+      "step": 106680
+    },
+    {
+      "epoch": 0.772293281794031,
+      "grad_norm": 0.15392231941223145,
+      "learning_rate": 4.2277139568720285e-06,
+      "loss": 0.8921,
+      "step": 106690
+    },
+    {
+      "epoch": 0.7723656684546172,
+      "grad_norm": 0.1547156274318695,
+      "learning_rate": 4.227641570211442e-06,
+      "loss": 0.8892,
+      "step": 106700
+    },
+    {
+      "epoch": 0.7724380551152034,
+      "grad_norm": 0.15668383240699768,
+      "learning_rate": 4.227569183550856e-06,
+      "loss": 0.8795,
+      "step": 106710
+    },
+    {
+      "epoch": 0.7725104417757895,
+      "grad_norm": 0.15469415485858917,
+      "learning_rate": 4.227496796890269e-06,
+      "loss": 0.8863,
+      "step": 106720
+    },
+    {
+      "epoch": 0.7725828284363757,
+      "grad_norm": 0.16153530776500702,
+      "learning_rate": 4.227424410229683e-06,
+      "loss": 0.8813,
+      "step": 106730
+    },
+    {
+      "epoch": 0.7726552150969619,
+      "grad_norm": 0.15066730976104736,
+      "learning_rate": 4.2273520235690965e-06,
+      "loss": 0.8876,
+      "step": 106740
+    },
+    {
+      "epoch": 0.7727276017575481,
+      "grad_norm": 0.15237820148468018,
+      "learning_rate": 4.227279636908511e-06,
+      "loss": 0.8915,
+      "step": 106750
+    },
+    {
+      "epoch": 0.7727999884181344,
+      "grad_norm": 0.1508363038301468,
+      "learning_rate": 4.227207250247925e-06,
+      "loss": 0.8846,
+      "step": 106760
+    },
+    {
+      "epoch": 0.7728723750787205,
+      "grad_norm": 0.16356633603572845,
+      "learning_rate": 4.227134863587338e-06,
+      "loss": 0.8851,
+      "step": 106770
+    },
+    {
+      "epoch": 0.7729447617393067,
+      "grad_norm": 0.1548301726579666,
+      "learning_rate": 4.227062476926752e-06,
+      "loss": 0.8941,
+      "step": 106780
+    },
+    {
+      "epoch": 0.7730171483998929,
+      "grad_norm": 0.18178719282150269,
+      "learning_rate": 4.226990090266166e-06,
+      "loss": 0.8839,
+      "step": 106790
+    },
+    {
+      "epoch": 0.7730895350604791,
+      "grad_norm": 0.16349080204963684,
+      "learning_rate": 4.22691770360558e-06,
+      "loss": 0.8818,
+      "step": 106800
+    },
+    {
+      "epoch": 0.7731619217210652,
+      "grad_norm": 0.15550731122493744,
+      "learning_rate": 4.2268453169449936e-06,
+      "loss": 0.8791,
+      "step": 106810
+    },
+    {
+      "epoch": 0.7732343083816514,
+      "grad_norm": 0.15300001204013824,
+      "learning_rate": 4.226772930284407e-06,
+      "loss": 0.8669,
+      "step": 106820
+    },
+    {
+      "epoch": 0.7733066950422376,
+      "grad_norm": 0.15288126468658447,
+      "learning_rate": 4.226700543623821e-06,
+      "loss": 0.8753,
+      "step": 106830
+    },
+    {
+      "epoch": 0.7733790817028238,
+      "grad_norm": 0.17350119352340698,
+      "learning_rate": 4.226628156963235e-06,
+      "loss": 0.8931,
+      "step": 106840
+    },
+    {
+      "epoch": 0.7734514683634099,
+      "grad_norm": 0.144486203789711,
+      "learning_rate": 4.226555770302649e-06,
+      "loss": 0.8797,
+      "step": 106850
+    },
+    {
+      "epoch": 0.7735238550239962,
+      "grad_norm": 0.15180061757564545,
+      "learning_rate": 4.2264833836420625e-06,
+      "loss": 0.883,
+      "step": 106860
+    },
+    {
+      "epoch": 0.7735962416845824,
+      "grad_norm": 0.16879640519618988,
+      "learning_rate": 4.226410996981476e-06,
+      "loss": 0.8795,
+      "step": 106870
+    },
+    {
+      "epoch": 0.7736686283451686,
+      "grad_norm": 0.17711849510669708,
+      "learning_rate": 4.2263386103208906e-06,
+      "loss": 0.881,
+      "step": 106880
+    },
+    {
+      "epoch": 0.7737410150057548,
+      "grad_norm": 0.15794023871421814,
+      "learning_rate": 4.226266223660304e-06,
+      "loss": 0.883,
+      "step": 106890
+    },
+    {
+      "epoch": 0.7738134016663409,
+      "grad_norm": 0.16354835033416748,
+      "learning_rate": 4.226193836999718e-06,
+      "loss": 0.8808,
+      "step": 106900
+    },
+    {
+      "epoch": 0.7738857883269271,
+      "grad_norm": 0.16571597754955292,
+      "learning_rate": 4.226121450339131e-06,
+      "loss": 0.879,
+      "step": 106910
+    },
+    {
+      "epoch": 0.7739581749875133,
+      "grad_norm": 0.16619838774204254,
+      "learning_rate": 4.226049063678546e-06,
+      "loss": 0.8966,
+      "step": 106920
+    },
+    {
+      "epoch": 0.7740305616480995,
+      "grad_norm": 0.16915611922740936,
+      "learning_rate": 4.2259766770179595e-06,
+      "loss": 0.867,
+      "step": 106930
+    },
+    {
+      "epoch": 0.7741029483086856,
+      "grad_norm": 0.15059566497802734,
+      "learning_rate": 4.225904290357373e-06,
+      "loss": 0.8969,
+      "step": 106940
+    },
+    {
+      "epoch": 0.7741753349692718,
+      "grad_norm": 0.15447822213172913,
+      "learning_rate": 4.225831903696787e-06,
+      "loss": 0.8692,
+      "step": 106950
+    },
+    {
+      "epoch": 0.774247721629858,
+      "grad_norm": 0.15346454083919525,
+      "learning_rate": 4.225759517036201e-06,
+      "loss": 0.879,
+      "step": 106960
+    },
+    {
+      "epoch": 0.7743201082904443,
+      "grad_norm": 0.15431329607963562,
+      "learning_rate": 4.225687130375615e-06,
+      "loss": 0.8781,
+      "step": 106970
+    },
+    {
+      "epoch": 0.7743924949510305,
+      "grad_norm": 0.15542955696582794,
+      "learning_rate": 4.225614743715028e-06,
+      "loss": 0.8815,
+      "step": 106980
+    },
+    {
+      "epoch": 0.7744648816116166,
+      "grad_norm": 0.15447083115577698,
+      "learning_rate": 4.225542357054442e-06,
+      "loss": 0.8799,
+      "step": 106990
+    },
+    {
+      "epoch": 0.7745372682722028,
+      "grad_norm": 0.17327377200126648,
+      "learning_rate": 4.2254699703938565e-06,
+      "loss": 0.8878,
+      "step": 107000
+    },
+    {
+      "epoch": 0.774609654932789,
+      "grad_norm": 0.15756526589393616,
+      "learning_rate": 4.22539758373327e-06,
+      "loss": 0.8798,
+      "step": 107010
+    },
+    {
+      "epoch": 0.7746820415933752,
+      "grad_norm": 0.15248951315879822,
+      "learning_rate": 4.225325197072684e-06,
+      "loss": 0.8899,
+      "step": 107020
+    },
+    {
+      "epoch": 0.7747544282539613,
+      "grad_norm": 0.1477757692337036,
+      "learning_rate": 4.225252810412097e-06,
+      "loss": 0.8826,
+      "step": 107030
+    },
+    {
+      "epoch": 0.7748268149145475,
+      "grad_norm": 0.1861177235841751,
+      "learning_rate": 4.225180423751512e-06,
+      "loss": 0.8875,
+      "step": 107040
+    },
+    {
+      "epoch": 0.7748992015751337,
+      "grad_norm": 0.14727658033370972,
+      "learning_rate": 4.2251080370909254e-06,
+      "loss": 0.8786,
+      "step": 107050
+    },
+    {
+      "epoch": 0.7749715882357199,
+      "grad_norm": 0.15688657760620117,
+      "learning_rate": 4.225035650430339e-06,
+      "loss": 0.8887,
+      "step": 107060
+    },
+    {
+      "epoch": 0.775043974896306,
+      "grad_norm": 0.15266485512256622,
+      "learning_rate": 4.224963263769753e-06,
+      "loss": 0.8883,
+      "step": 107070
+    },
+    {
+      "epoch": 0.7751163615568923,
+      "grad_norm": 0.15053589642047882,
+      "learning_rate": 4.224890877109167e-06,
+      "loss": 0.8763,
+      "step": 107080
+    },
+    {
+      "epoch": 0.7751887482174785,
+      "grad_norm": 0.1497945785522461,
+      "learning_rate": 4.224818490448581e-06,
+      "loss": 0.8768,
+      "step": 107090
+    },
+    {
+      "epoch": 0.7752611348780647,
+      "grad_norm": 0.16105857491493225,
+      "learning_rate": 4.224746103787994e-06,
+      "loss": 0.8819,
+      "step": 107100
+    },
+    {
+      "epoch": 0.7753335215386509,
+      "grad_norm": 0.15862345695495605,
+      "learning_rate": 4.224673717127408e-06,
+      "loss": 0.8753,
+      "step": 107110
+    },
+    {
+      "epoch": 0.775405908199237,
+      "grad_norm": 0.15614831447601318,
+      "learning_rate": 4.2246013304668224e-06,
+      "loss": 0.8811,
+      "step": 107120
+    },
+    {
+      "epoch": 0.7754782948598232,
+      "grad_norm": 0.19445110857486725,
+      "learning_rate": 4.224528943806236e-06,
+      "loss": 0.8964,
+      "step": 107130
+    },
+    {
+      "epoch": 0.7755506815204094,
+      "grad_norm": 0.15942369401454926,
+      "learning_rate": 4.22445655714565e-06,
+      "loss": 0.8658,
+      "step": 107140
+    },
+    {
+      "epoch": 0.7756230681809956,
+      "grad_norm": 0.1634572595357895,
+      "learning_rate": 4.224384170485063e-06,
+      "loss": 0.8844,
+      "step": 107150
+    },
+    {
+      "epoch": 0.7756954548415818,
+      "grad_norm": 0.14864301681518555,
+      "learning_rate": 4.224311783824478e-06,
+      "loss": 0.8885,
+      "step": 107160
+    },
+    {
+      "epoch": 0.7757678415021679,
+      "grad_norm": 0.1603207290172577,
+      "learning_rate": 4.224239397163891e-06,
+      "loss": 0.8811,
+      "step": 107170
+    },
+    {
+      "epoch": 0.7758402281627542,
+      "grad_norm": 0.15209531784057617,
+      "learning_rate": 4.224167010503305e-06,
+      "loss": 0.8856,
+      "step": 107180
+    },
+    {
+      "epoch": 0.7759126148233404,
+      "grad_norm": 0.1611732542514801,
+      "learning_rate": 4.224094623842719e-06,
+      "loss": 0.8802,
+      "step": 107190
+    },
+    {
+      "epoch": 0.7759850014839266,
+      "grad_norm": 0.16331136226654053,
+      "learning_rate": 4.224022237182133e-06,
+      "loss": 0.8931,
+      "step": 107200
+    },
+    {
+      "epoch": 0.7760573881445127,
+      "grad_norm": 0.17706800997257233,
+      "learning_rate": 4.223949850521547e-06,
+      "loss": 0.8834,
+      "step": 107210
+    },
+    {
+      "epoch": 0.7761297748050989,
+      "grad_norm": 0.14724914729595184,
+      "learning_rate": 4.22387746386096e-06,
+      "loss": 0.8868,
+      "step": 107220
+    },
+    {
+      "epoch": 0.7762021614656851,
+      "grad_norm": 0.16513299942016602,
+      "learning_rate": 4.223805077200374e-06,
+      "loss": 0.8752,
+      "step": 107230
+    },
+    {
+      "epoch": 0.7762745481262713,
+      "grad_norm": 0.17045806348323822,
+      "learning_rate": 4.2237326905397875e-06,
+      "loss": 0.8909,
+      "step": 107240
+    },
+    {
+      "epoch": 0.7763469347868575,
+      "grad_norm": 0.189479261636734,
+      "learning_rate": 4.223660303879201e-06,
+      "loss": 0.884,
+      "step": 107250
+    },
+    {
+      "epoch": 0.7764193214474436,
+      "grad_norm": 0.15895524621009827,
+      "learning_rate": 4.223587917218615e-06,
+      "loss": 0.8878,
+      "step": 107260
+    },
+    {
+      "epoch": 0.7764917081080298,
+      "grad_norm": 0.16213759779930115,
+      "learning_rate": 4.223515530558029e-06,
+      "loss": 0.8825,
+      "step": 107270
+    },
+    {
+      "epoch": 0.776564094768616,
+      "grad_norm": 0.1568875014781952,
+      "learning_rate": 4.223443143897443e-06,
+      "loss": 0.8751,
+      "step": 107280
+    },
+    {
+      "epoch": 0.7766364814292023,
+      "grad_norm": 0.1578979790210724,
+      "learning_rate": 4.2233707572368565e-06,
+      "loss": 0.883,
+      "step": 107290
+    },
+    {
+      "epoch": 0.7767088680897885,
+      "grad_norm": 0.1542271375656128,
+      "learning_rate": 4.22329837057627e-06,
+      "loss": 0.8786,
+      "step": 107300
+    },
+    {
+      "epoch": 0.7767812547503746,
+      "grad_norm": 0.16791146993637085,
+      "learning_rate": 4.2232259839156845e-06,
+      "loss": 0.8914,
+      "step": 107310
+    },
+    {
+      "epoch": 0.7768536414109608,
+      "grad_norm": 0.1576090306043625,
+      "learning_rate": 4.223153597255098e-06,
+      "loss": 0.894,
+      "step": 107320
+    },
+    {
+      "epoch": 0.776926028071547,
+      "grad_norm": 0.15662311017513275,
+      "learning_rate": 4.223081210594512e-06,
+      "loss": 0.8998,
+      "step": 107330
+    },
+    {
+      "epoch": 0.7769984147321332,
+      "grad_norm": 0.14646415412425995,
+      "learning_rate": 4.223008823933925e-06,
+      "loss": 0.867,
+      "step": 107340
+    },
+    {
+      "epoch": 0.7770708013927193,
+      "grad_norm": 0.14766213297843933,
+      "learning_rate": 4.22293643727334e-06,
+      "loss": 0.8937,
+      "step": 107350
+    },
+    {
+      "epoch": 0.7771431880533055,
+      "grad_norm": 0.1753314882516861,
+      "learning_rate": 4.2228640506127535e-06,
+      "loss": 0.8736,
+      "step": 107360
+    },
+    {
+      "epoch": 0.7772155747138917,
+      "grad_norm": 0.15868641436100006,
+      "learning_rate": 4.222791663952167e-06,
+      "loss": 0.9,
+      "step": 107370
+    },
+    {
+      "epoch": 0.7772879613744779,
+      "grad_norm": 0.15650039911270142,
+      "learning_rate": 4.222719277291581e-06,
+      "loss": 0.886,
+      "step": 107380
+    },
+    {
+      "epoch": 0.7773603480350642,
+      "grad_norm": 0.17144428193569183,
+      "learning_rate": 4.222646890630995e-06,
+      "loss": 0.892,
+      "step": 107390
+    },
+    {
+      "epoch": 0.7774327346956503,
+      "grad_norm": 0.1921321600675583,
+      "learning_rate": 4.222574503970409e-06,
+      "loss": 0.8881,
+      "step": 107400
+    },
+    {
+      "epoch": 0.7775051213562365,
+      "grad_norm": 0.1584097146987915,
+      "learning_rate": 4.222502117309822e-06,
+      "loss": 0.8806,
+      "step": 107410
+    },
+    {
+      "epoch": 0.7775775080168227,
+      "grad_norm": 0.15249648690223694,
+      "learning_rate": 4.222429730649236e-06,
+      "loss": 0.8956,
+      "step": 107420
+    },
+    {
+      "epoch": 0.7776498946774089,
+      "grad_norm": 0.15863776206970215,
+      "learning_rate": 4.2223573439886505e-06,
+      "loss": 0.8773,
+      "step": 107430
+    },
+    {
+      "epoch": 0.777722281337995,
+      "grad_norm": 0.16656135022640228,
+      "learning_rate": 4.222284957328064e-06,
+      "loss": 0.8973,
+      "step": 107440
+    },
+    {
+      "epoch": 0.7777946679985812,
+      "grad_norm": 0.15339358150959015,
+      "learning_rate": 4.222212570667478e-06,
+      "loss": 0.8851,
+      "step": 107450
+    },
+    {
+      "epoch": 0.7778670546591674,
+      "grad_norm": 0.1573115736246109,
+      "learning_rate": 4.222140184006891e-06,
+      "loss": 0.8841,
+      "step": 107460
+    },
+    {
+      "epoch": 0.7779394413197536,
+      "grad_norm": 0.1542723923921585,
+      "learning_rate": 4.222067797346305e-06,
+      "loss": 0.8954,
+      "step": 107470
+    },
+    {
+      "epoch": 0.7780118279803397,
+      "grad_norm": 0.1786673665046692,
+      "learning_rate": 4.221995410685719e-06,
+      "loss": 0.8873,
+      "step": 107480
+    },
+    {
+      "epoch": 0.7780842146409259,
+      "grad_norm": 0.17204001545906067,
+      "learning_rate": 4.221923024025133e-06,
+      "loss": 0.8873,
+      "step": 107490
+    },
+    {
+      "epoch": 0.7781566013015122,
+      "grad_norm": 0.14830465614795685,
+      "learning_rate": 4.221850637364547e-06,
+      "loss": 0.8775,
+      "step": 107500
+    },
+    {
+      "epoch": 0.7782289879620984,
+      "grad_norm": 0.1740427315235138,
+      "learning_rate": 4.22177825070396e-06,
+      "loss": 0.8825,
+      "step": 107510
+    },
+    {
+      "epoch": 0.7783013746226846,
+      "grad_norm": 0.1551501452922821,
+      "learning_rate": 4.221705864043375e-06,
+      "loss": 0.9021,
+      "step": 107520
+    },
+    {
+      "epoch": 0.7783737612832707,
+      "grad_norm": 0.1553511619567871,
+      "learning_rate": 4.221633477382788e-06,
+      "loss": 0.8773,
+      "step": 107530
+    },
+    {
+      "epoch": 0.7784461479438569,
+      "grad_norm": 0.15967567265033722,
+      "learning_rate": 4.221561090722202e-06,
+      "loss": 0.8829,
+      "step": 107540
+    },
+    {
+      "epoch": 0.7785185346044431,
+      "grad_norm": 0.14767323434352875,
+      "learning_rate": 4.2214887040616156e-06,
+      "loss": 0.8687,
+      "step": 107550
+    },
+    {
+      "epoch": 0.7785909212650293,
+      "grad_norm": 0.14567668735980988,
+      "learning_rate": 4.22141631740103e-06,
+      "loss": 0.8757,
+      "step": 107560
+    },
+    {
+      "epoch": 0.7786633079256154,
+      "grad_norm": 0.14703209698200226,
+      "learning_rate": 4.221343930740444e-06,
+      "loss": 0.8835,
+      "step": 107570
+    },
+    {
+      "epoch": 0.7787356945862016,
+      "grad_norm": 0.15349993109703064,
+      "learning_rate": 4.221271544079857e-06,
+      "loss": 0.8685,
+      "step": 107580
+    },
+    {
+      "epoch": 0.7788080812467878,
+      "grad_norm": 0.16429787874221802,
+      "learning_rate": 4.221199157419271e-06,
+      "loss": 0.8891,
+      "step": 107590
+    },
+    {
+      "epoch": 0.778880467907374,
+      "grad_norm": 0.15371567010879517,
+      "learning_rate": 4.221126770758685e-06,
+      "loss": 0.8729,
+      "step": 107600
+    },
+    {
+      "epoch": 0.7789528545679603,
+      "grad_norm": 0.1502668410539627,
+      "learning_rate": 4.221054384098099e-06,
+      "loss": 0.8834,
+      "step": 107610
+    },
+    {
+      "epoch": 0.7790252412285464,
+      "grad_norm": 0.166198268532753,
+      "learning_rate": 4.2209819974375126e-06,
+      "loss": 0.876,
+      "step": 107620
+    },
+    {
+      "epoch": 0.7790976278891326,
+      "grad_norm": 0.1724855899810791,
+      "learning_rate": 4.220909610776926e-06,
+      "loss": 0.8752,
+      "step": 107630
+    },
+    {
+      "epoch": 0.7791700145497188,
+      "grad_norm": 0.15645526349544525,
+      "learning_rate": 4.220837224116341e-06,
+      "loss": 0.8787,
+      "step": 107640
+    },
+    {
+      "epoch": 0.779242401210305,
+      "grad_norm": 0.16536635160446167,
+      "learning_rate": 4.220764837455754e-06,
+      "loss": 0.8852,
+      "step": 107650
+    },
+    {
+      "epoch": 0.7793147878708911,
+      "grad_norm": 0.1508960872888565,
+      "learning_rate": 4.220692450795168e-06,
+      "loss": 0.8768,
+      "step": 107660
+    },
+    {
+      "epoch": 0.7793871745314773,
+      "grad_norm": 0.1536785364151001,
+      "learning_rate": 4.2206200641345815e-06,
+      "loss": 0.8825,
+      "step": 107670
+    },
+    {
+      "epoch": 0.7794595611920635,
+      "grad_norm": 0.16157682240009308,
+      "learning_rate": 4.220547677473996e-06,
+      "loss": 0.8861,
+      "step": 107680
+    },
+    {
+      "epoch": 0.7795319478526497,
+      "grad_norm": 0.15818680822849274,
+      "learning_rate": 4.2204752908134096e-06,
+      "loss": 0.8745,
+      "step": 107690
+    },
+    {
+      "epoch": 0.7796043345132359,
+      "grad_norm": 0.16059935092926025,
+      "learning_rate": 4.220402904152823e-06,
+      "loss": 0.8934,
+      "step": 107700
+    },
+    {
+      "epoch": 0.7796767211738221,
+      "grad_norm": 0.15591976046562195,
+      "learning_rate": 4.220330517492237e-06,
+      "loss": 0.8838,
+      "step": 107710
+    },
+    {
+      "epoch": 0.7797491078344083,
+      "grad_norm": 0.15530046820640564,
+      "learning_rate": 4.220258130831651e-06,
+      "loss": 0.8794,
+      "step": 107720
+    },
+    {
+      "epoch": 0.7798214944949945,
+      "grad_norm": 0.1756300926208496,
+      "learning_rate": 4.220185744171065e-06,
+      "loss": 0.8819,
+      "step": 107730
+    },
+    {
+      "epoch": 0.7798938811555807,
+      "grad_norm": 0.16055084764957428,
+      "learning_rate": 4.2201133575104785e-06,
+      "loss": 0.8594,
+      "step": 107740
+    },
+    {
+      "epoch": 0.7799662678161668,
+      "grad_norm": 0.1619112342596054,
+      "learning_rate": 4.220040970849892e-06,
+      "loss": 0.8905,
+      "step": 107750
+    },
+    {
+      "epoch": 0.780038654476753,
+      "grad_norm": 0.15275640785694122,
+      "learning_rate": 4.2199685841893066e-06,
+      "loss": 0.8799,
+      "step": 107760
+    },
+    {
+      "epoch": 0.7801110411373392,
+      "grad_norm": 0.1480998545885086,
+      "learning_rate": 4.219896197528719e-06,
+      "loss": 0.8672,
+      "step": 107770
+    },
+    {
+      "epoch": 0.7801834277979254,
+      "grad_norm": 0.15729357302188873,
+      "learning_rate": 4.219823810868133e-06,
+      "loss": 0.8956,
+      "step": 107780
+    },
+    {
+      "epoch": 0.7802558144585116,
+      "grad_norm": 0.1509513109922409,
+      "learning_rate": 4.2197514242075474e-06,
+      "loss": 0.8796,
+      "step": 107790
+    },
+    {
+      "epoch": 0.7803282011190977,
+      "grad_norm": 0.15310880541801453,
+      "learning_rate": 4.219679037546961e-06,
+      "loss": 0.8726,
+      "step": 107800
+    },
+    {
+      "epoch": 0.7804005877796839,
+      "grad_norm": 0.16130521893501282,
+      "learning_rate": 4.219606650886375e-06,
+      "loss": 0.8855,
+      "step": 107810
+    },
+    {
+      "epoch": 0.7804729744402702,
+      "grad_norm": 0.18404695391654968,
+      "learning_rate": 4.219534264225788e-06,
+      "loss": 0.8839,
+      "step": 107820
+    },
+    {
+      "epoch": 0.7805453611008564,
+      "grad_norm": 0.16077350080013275,
+      "learning_rate": 4.219461877565203e-06,
+      "loss": 0.8933,
+      "step": 107830
+    },
+    {
+      "epoch": 0.7806177477614425,
+      "grad_norm": 0.1624414175748825,
+      "learning_rate": 4.219389490904616e-06,
+      "loss": 0.871,
+      "step": 107840
+    },
+    {
+      "epoch": 0.7806901344220287,
+      "grad_norm": 0.16145184636116028,
+      "learning_rate": 4.21931710424403e-06,
+      "loss": 0.8888,
+      "step": 107850
+    },
+    {
+      "epoch": 0.7807625210826149,
+      "grad_norm": 0.17542187869548798,
+      "learning_rate": 4.219244717583444e-06,
+      "loss": 0.8951,
+      "step": 107860
+    },
+    {
+      "epoch": 0.7808349077432011,
+      "grad_norm": 0.14952528476715088,
+      "learning_rate": 4.219172330922858e-06,
+      "loss": 0.8886,
+      "step": 107870
+    },
+    {
+      "epoch": 0.7809072944037873,
+      "grad_norm": 0.1723993569612503,
+      "learning_rate": 4.219099944262272e-06,
+      "loss": 0.8807,
+      "step": 107880
+    },
+    {
+      "epoch": 0.7809796810643734,
+      "grad_norm": 0.1505005955696106,
+      "learning_rate": 4.219027557601685e-06,
+      "loss": 0.8711,
+      "step": 107890
+    },
+    {
+      "epoch": 0.7810520677249596,
+      "grad_norm": 0.15281620621681213,
+      "learning_rate": 4.218955170941099e-06,
+      "loss": 0.887,
+      "step": 107900
+    },
+    {
+      "epoch": 0.7811244543855458,
+      "grad_norm": 0.1522083282470703,
+      "learning_rate": 4.218882784280513e-06,
+      "loss": 0.8973,
+      "step": 107910
+    },
+    {
+      "epoch": 0.7811968410461321,
+      "grad_norm": 0.16179485619068146,
+      "learning_rate": 4.218810397619927e-06,
+      "loss": 0.8856,
+      "step": 107920
+    },
+    {
+      "epoch": 0.7812692277067183,
+      "grad_norm": 0.15184660255908966,
+      "learning_rate": 4.218738010959341e-06,
+      "loss": 0.8856,
+      "step": 107930
+    },
+    {
+      "epoch": 0.7813416143673044,
+      "grad_norm": 0.19640418887138367,
+      "learning_rate": 4.218665624298754e-06,
+      "loss": 0.8837,
+      "step": 107940
+    },
+    {
+      "epoch": 0.7814140010278906,
+      "grad_norm": 0.16016776859760284,
+      "learning_rate": 4.218593237638169e-06,
+      "loss": 0.887,
+      "step": 107950
+    },
+    {
+      "epoch": 0.7814863876884768,
+      "grad_norm": 0.1551421880722046,
+      "learning_rate": 4.218520850977582e-06,
+      "loss": 0.8665,
+      "step": 107960
+    },
+    {
+      "epoch": 0.781558774349063,
+      "grad_norm": 0.16472944617271423,
+      "learning_rate": 4.218448464316996e-06,
+      "loss": 0.8814,
+      "step": 107970
+    },
+    {
+      "epoch": 0.7816311610096491,
+      "grad_norm": 0.1580583155155182,
+      "learning_rate": 4.2183760776564095e-06,
+      "loss": 0.8823,
+      "step": 107980
+    },
+    {
+      "epoch": 0.7817035476702353,
+      "grad_norm": 0.1603555530309677,
+      "learning_rate": 4.218303690995824e-06,
+      "loss": 0.872,
+      "step": 107990
+    },
+    {
+      "epoch": 0.7817759343308215,
+      "grad_norm": 0.1556985229253769,
+      "learning_rate": 4.218231304335238e-06,
+      "loss": 0.8787,
+      "step": 108000
+    },
+    {
+      "epoch": 0.7818483209914077,
+      "grad_norm": 0.1534159928560257,
+      "learning_rate": 4.218158917674651e-06,
+      "loss": 0.9064,
+      "step": 108010
+    },
+    {
+      "epoch": 0.7819207076519938,
+      "grad_norm": 0.1716359406709671,
+      "learning_rate": 4.218086531014065e-06,
+      "loss": 0.8834,
+      "step": 108020
+    },
+    {
+      "epoch": 0.7819930943125801,
+      "grad_norm": 0.15171417593955994,
+      "learning_rate": 4.218014144353479e-06,
+      "loss": 0.8825,
+      "step": 108030
+    },
+    {
+      "epoch": 0.7820654809731663,
+      "grad_norm": 0.1647900491952896,
+      "learning_rate": 4.217941757692893e-06,
+      "loss": 0.8718,
+      "step": 108040
+    },
+    {
+      "epoch": 0.7821378676337525,
+      "grad_norm": 0.1467769294977188,
+      "learning_rate": 4.2178693710323065e-06,
+      "loss": 0.8832,
+      "step": 108050
+    },
+    {
+      "epoch": 0.7822102542943387,
+      "grad_norm": 0.15823620557785034,
+      "learning_rate": 4.21779698437172e-06,
+      "loss": 0.8914,
+      "step": 108060
+    },
+    {
+      "epoch": 0.7822826409549248,
+      "grad_norm": 0.17605745792388916,
+      "learning_rate": 4.217724597711134e-06,
+      "loss": 0.8871,
+      "step": 108070
+    },
+    {
+      "epoch": 0.782355027615511,
+      "grad_norm": 0.15321215987205505,
+      "learning_rate": 4.217652211050548e-06,
+      "loss": 0.8906,
+      "step": 108080
+    },
+    {
+      "epoch": 0.7824274142760972,
+      "grad_norm": 0.15181642770767212,
+      "learning_rate": 4.217579824389962e-06,
+      "loss": 0.8662,
+      "step": 108090
+    },
+    {
+      "epoch": 0.7824998009366834,
+      "grad_norm": 0.17298097908496857,
+      "learning_rate": 4.2175074377293755e-06,
+      "loss": 0.8795,
+      "step": 108100
+    },
+    {
+      "epoch": 0.7825721875972695,
+      "grad_norm": 0.1724858433008194,
+      "learning_rate": 4.217435051068789e-06,
+      "loss": 0.9011,
+      "step": 108110
+    },
+    {
+      "epoch": 0.7826445742578557,
+      "grad_norm": 0.1518513262271881,
+      "learning_rate": 4.2173626644082035e-06,
+      "loss": 0.8938,
+      "step": 108120
+    },
+    {
+      "epoch": 0.7827169609184419,
+      "grad_norm": 0.15819215774536133,
+      "learning_rate": 4.217290277747617e-06,
+      "loss": 0.8725,
+      "step": 108130
+    },
+    {
+      "epoch": 0.7827893475790282,
+      "grad_norm": 0.1756516396999359,
+      "learning_rate": 4.217217891087031e-06,
+      "loss": 0.8867,
+      "step": 108140
+    },
+    {
+      "epoch": 0.7828617342396144,
+      "grad_norm": 0.1819458156824112,
+      "learning_rate": 4.217145504426444e-06,
+      "loss": 0.8883,
+      "step": 108150
+    },
+    {
+      "epoch": 0.7829341209002005,
+      "grad_norm": 0.14967991411685944,
+      "learning_rate": 4.217073117765859e-06,
+      "loss": 0.8814,
+      "step": 108160
+    },
+    {
+      "epoch": 0.7830065075607867,
+      "grad_norm": 0.15189316868782043,
+      "learning_rate": 4.2170007311052725e-06,
+      "loss": 0.8822,
+      "step": 108170
+    },
+    {
+      "epoch": 0.7830788942213729,
+      "grad_norm": 0.159009650349617,
+      "learning_rate": 4.216928344444686e-06,
+      "loss": 0.8774,
+      "step": 108180
+    },
+    {
+      "epoch": 0.7831512808819591,
+      "grad_norm": 0.15388379991054535,
+      "learning_rate": 4.2168559577841e-06,
+      "loss": 0.8821,
+      "step": 108190
+    },
+    {
+      "epoch": 0.7832236675425452,
+      "grad_norm": 0.1746244877576828,
+      "learning_rate": 4.216783571123514e-06,
+      "loss": 0.8937,
+      "step": 108200
+    },
+    {
+      "epoch": 0.7832960542031314,
+      "grad_norm": 0.1544540673494339,
+      "learning_rate": 4.216711184462928e-06,
+      "loss": 0.8892,
+      "step": 108210
+    },
+    {
+      "epoch": 0.7833684408637176,
+      "grad_norm": 0.15060152113437653,
+      "learning_rate": 4.216638797802341e-06,
+      "loss": 0.8786,
+      "step": 108220
+    },
+    {
+      "epoch": 0.7834408275243038,
+      "grad_norm": 0.15450206398963928,
+      "learning_rate": 4.216566411141755e-06,
+      "loss": 0.8823,
+      "step": 108230
+    },
+    {
+      "epoch": 0.7835132141848901,
+      "grad_norm": 0.1713588386774063,
+      "learning_rate": 4.2164940244811695e-06,
+      "loss": 0.896,
+      "step": 108240
+    },
+    {
+      "epoch": 0.7835856008454762,
+      "grad_norm": 0.1668989658355713,
+      "learning_rate": 4.216421637820583e-06,
+      "loss": 0.8828,
+      "step": 108250
+    },
+    {
+      "epoch": 0.7836579875060624,
+      "grad_norm": 0.1604335904121399,
+      "learning_rate": 4.216349251159997e-06,
+      "loss": 0.8747,
+      "step": 108260
+    },
+    {
+      "epoch": 0.7837303741666486,
+      "grad_norm": 0.1643938422203064,
+      "learning_rate": 4.21627686449941e-06,
+      "loss": 0.8799,
+      "step": 108270
+    },
+    {
+      "epoch": 0.7838027608272348,
+      "grad_norm": 0.1590445637702942,
+      "learning_rate": 4.216204477838825e-06,
+      "loss": 0.8859,
+      "step": 108280
+    },
+    {
+      "epoch": 0.783875147487821,
+      "grad_norm": 0.16419881582260132,
+      "learning_rate": 4.216132091178238e-06,
+      "loss": 0.8834,
+      "step": 108290
+    },
+    {
+      "epoch": 0.7839475341484071,
+      "grad_norm": 0.14862844347953796,
+      "learning_rate": 4.216059704517651e-06,
+      "loss": 0.8793,
+      "step": 108300
+    },
+    {
+      "epoch": 0.7840199208089933,
+      "grad_norm": 0.1492660492658615,
+      "learning_rate": 4.215987317857066e-06,
+      "loss": 0.8808,
+      "step": 108310
+    },
+    {
+      "epoch": 0.7840923074695795,
+      "grad_norm": 0.14114709198474884,
+      "learning_rate": 4.215914931196479e-06,
+      "loss": 0.865,
+      "step": 108320
+    },
+    {
+      "epoch": 0.7841646941301657,
+      "grad_norm": 0.16671665012836456,
+      "learning_rate": 4.215842544535893e-06,
+      "loss": 0.8726,
+      "step": 108330
+    },
+    {
+      "epoch": 0.7842370807907518,
+      "grad_norm": 0.1602155566215515,
+      "learning_rate": 4.2157701578753065e-06,
+      "loss": 0.8726,
+      "step": 108340
+    },
+    {
+      "epoch": 0.7843094674513381,
+      "grad_norm": 0.1590067595243454,
+      "learning_rate": 4.215697771214721e-06,
+      "loss": 0.8723,
+      "step": 108350
+    },
+    {
+      "epoch": 0.7843818541119243,
+      "grad_norm": 0.16449514031410217,
+      "learning_rate": 4.2156253845541346e-06,
+      "loss": 0.8795,
+      "step": 108360
+    },
+    {
+      "epoch": 0.7844542407725105,
+      "grad_norm": 0.16773192584514618,
+      "learning_rate": 4.215552997893548e-06,
+      "loss": 0.8772,
+      "step": 108370
+    },
+    {
+      "epoch": 0.7845266274330966,
+      "grad_norm": 0.15345753729343414,
+      "learning_rate": 4.215480611232962e-06,
+      "loss": 0.8893,
+      "step": 108380
+    },
+    {
+      "epoch": 0.7845990140936828,
+      "grad_norm": 0.1570970118045807,
+      "learning_rate": 4.215408224572376e-06,
+      "loss": 0.8911,
+      "step": 108390
+    },
+    {
+      "epoch": 0.784671400754269,
+      "grad_norm": 0.480564683675766,
+      "learning_rate": 4.21533583791179e-06,
+      "loss": 0.881,
+      "step": 108400
+    },
+    {
+      "epoch": 0.7847437874148552,
+      "grad_norm": 0.1536029726266861,
+      "learning_rate": 4.2152634512512035e-06,
+      "loss": 0.8657,
+      "step": 108410
+    },
+    {
+      "epoch": 0.7848161740754414,
+      "grad_norm": 0.15583275258541107,
+      "learning_rate": 4.215191064590617e-06,
+      "loss": 0.8856,
+      "step": 108420
+    },
+    {
+      "epoch": 0.7848885607360275,
+      "grad_norm": 0.17179587483406067,
+      "learning_rate": 4.2151186779300316e-06,
+      "loss": 0.8915,
+      "step": 108430
+    },
+    {
+      "epoch": 0.7849609473966137,
+      "grad_norm": 0.16397567093372345,
+      "learning_rate": 4.215046291269445e-06,
+      "loss": 0.87,
+      "step": 108440
+    },
+    {
+      "epoch": 0.7850333340572,
+      "grad_norm": 0.1486414670944214,
+      "learning_rate": 4.214973904608859e-06,
+      "loss": 0.8715,
+      "step": 108450
+    },
+    {
+      "epoch": 0.7851057207177862,
+      "grad_norm": 0.14893367886543274,
+      "learning_rate": 4.214901517948272e-06,
+      "loss": 0.88,
+      "step": 108460
+    },
+    {
+      "epoch": 0.7851781073783723,
+      "grad_norm": 0.15935064852237701,
+      "learning_rate": 4.214829131287687e-06,
+      "loss": 0.8744,
+      "step": 108470
+    },
+    {
+      "epoch": 0.7852504940389585,
+      "grad_norm": 0.16078276932239532,
+      "learning_rate": 4.2147567446271005e-06,
+      "loss": 0.8863,
+      "step": 108480
+    },
+    {
+      "epoch": 0.7853228806995447,
+      "grad_norm": 0.16032767295837402,
+      "learning_rate": 4.214684357966514e-06,
+      "loss": 0.8813,
+      "step": 108490
+    },
+    {
+      "epoch": 0.7853952673601309,
+      "grad_norm": 0.2287614494562149,
+      "learning_rate": 4.214611971305928e-06,
+      "loss": 0.8818,
+      "step": 108500
+    },
+    {
+      "epoch": 0.785467654020717,
+      "grad_norm": 0.16381344199180603,
+      "learning_rate": 4.214539584645342e-06,
+      "loss": 0.884,
+      "step": 108510
+    },
+    {
+      "epoch": 0.7855400406813032,
+      "grad_norm": 0.16530375182628632,
+      "learning_rate": 4.214467197984756e-06,
+      "loss": 0.8746,
+      "step": 108520
+    },
+    {
+      "epoch": 0.7856124273418894,
+      "grad_norm": 0.14450208842754364,
+      "learning_rate": 4.2143948113241694e-06,
+      "loss": 0.8855,
+      "step": 108530
+    },
+    {
+      "epoch": 0.7856848140024756,
+      "grad_norm": 0.16523830592632294,
+      "learning_rate": 4.214322424663583e-06,
+      "loss": 0.8822,
+      "step": 108540
+    },
+    {
+      "epoch": 0.7857572006630618,
+      "grad_norm": 0.175279900431633,
+      "learning_rate": 4.2142500380029975e-06,
+      "loss": 0.8739,
+      "step": 108550
+    },
+    {
+      "epoch": 0.785829587323648,
+      "grad_norm": 0.16809509694576263,
+      "learning_rate": 4.214177651342411e-06,
+      "loss": 0.8929,
+      "step": 108560
+    },
+    {
+      "epoch": 0.7859019739842342,
+      "grad_norm": 0.14289383590221405,
+      "learning_rate": 4.214105264681825e-06,
+      "loss": 0.8728,
+      "step": 108570
+    },
+    {
+      "epoch": 0.7859743606448204,
+      "grad_norm": 0.14453014731407166,
+      "learning_rate": 4.214032878021238e-06,
+      "loss": 0.8962,
+      "step": 108580
+    },
+    {
+      "epoch": 0.7860467473054066,
+      "grad_norm": 0.1566169410943985,
+      "learning_rate": 4.213960491360653e-06,
+      "loss": 0.8758,
+      "step": 108590
+    },
+    {
+      "epoch": 0.7861191339659928,
+      "grad_norm": 0.1643407642841339,
+      "learning_rate": 4.2138881047000664e-06,
+      "loss": 0.882,
+      "step": 108600
+    },
+    {
+      "epoch": 0.7861915206265789,
+      "grad_norm": 0.16235338151454926,
+      "learning_rate": 4.21381571803948e-06,
+      "loss": 0.8701,
+      "step": 108610
+    },
+    {
+      "epoch": 0.7862639072871651,
+      "grad_norm": 0.153314471244812,
+      "learning_rate": 4.213743331378894e-06,
+      "loss": 0.8827,
+      "step": 108620
+    },
+    {
+      "epoch": 0.7863362939477513,
+      "grad_norm": 0.15935984253883362,
+      "learning_rate": 4.213670944718308e-06,
+      "loss": 0.8793,
+      "step": 108630
+    },
+    {
+      "epoch": 0.7864086806083375,
+      "grad_norm": 0.15511925518512726,
+      "learning_rate": 4.213598558057722e-06,
+      "loss": 0.8916,
+      "step": 108640
+    },
+    {
+      "epoch": 0.7864810672689236,
+      "grad_norm": 0.15609268844127655,
+      "learning_rate": 4.213526171397135e-06,
+      "loss": 0.8937,
+      "step": 108650
+    },
+    {
+      "epoch": 0.7865534539295098,
+      "grad_norm": 0.14593185484409332,
+      "learning_rate": 4.213453784736549e-06,
+      "loss": 0.8766,
+      "step": 108660
+    },
+    {
+      "epoch": 0.7866258405900961,
+      "grad_norm": 0.16142848134040833,
+      "learning_rate": 4.2133813980759634e-06,
+      "loss": 0.8711,
+      "step": 108670
+    },
+    {
+      "epoch": 0.7866982272506823,
+      "grad_norm": 0.15131999552249908,
+      "learning_rate": 4.213309011415377e-06,
+      "loss": 0.8837,
+      "step": 108680
+    },
+    {
+      "epoch": 0.7867706139112685,
+      "grad_norm": 0.16211840510368347,
+      "learning_rate": 4.213236624754791e-06,
+      "loss": 0.8916,
+      "step": 108690
+    },
+    {
+      "epoch": 0.7868430005718546,
+      "grad_norm": 0.16512945294380188,
+      "learning_rate": 4.213164238094204e-06,
+      "loss": 0.8844,
+      "step": 108700
+    },
+    {
+      "epoch": 0.7869153872324408,
+      "grad_norm": 0.15159179270267487,
+      "learning_rate": 4.213091851433618e-06,
+      "loss": 0.8814,
+      "step": 108710
+    },
+    {
+      "epoch": 0.786987773893027,
+      "grad_norm": 0.17902809381484985,
+      "learning_rate": 4.213019464773032e-06,
+      "loss": 0.8774,
+      "step": 108720
+    },
+    {
+      "epoch": 0.7870601605536132,
+      "grad_norm": 0.16519932448863983,
+      "learning_rate": 4.212947078112446e-06,
+      "loss": 0.8849,
+      "step": 108730
+    },
+    {
+      "epoch": 0.7871325472141993,
+      "grad_norm": 0.16074302792549133,
+      "learning_rate": 4.21287469145186e-06,
+      "loss": 0.8875,
+      "step": 108740
+    },
+    {
+      "epoch": 0.7872049338747855,
+      "grad_norm": 0.1652543991804123,
+      "learning_rate": 4.212802304791273e-06,
+      "loss": 0.8727,
+      "step": 108750
+    },
+    {
+      "epoch": 0.7872773205353717,
+      "grad_norm": 0.17259052395820618,
+      "learning_rate": 4.212729918130688e-06,
+      "loss": 0.8849,
+      "step": 108760
+    },
+    {
+      "epoch": 0.787349707195958,
+      "grad_norm": 0.17187197506427765,
+      "learning_rate": 4.212657531470101e-06,
+      "loss": 0.8853,
+      "step": 108770
+    },
+    {
+      "epoch": 0.7874220938565442,
+      "grad_norm": 0.1663118600845337,
+      "learning_rate": 4.212585144809515e-06,
+      "loss": 0.8764,
+      "step": 108780
+    },
+    {
+      "epoch": 0.7874944805171303,
+      "grad_norm": 0.15814106166362762,
+      "learning_rate": 4.2125127581489285e-06,
+      "loss": 0.8648,
+      "step": 108790
+    },
+    {
+      "epoch": 0.7875668671777165,
+      "grad_norm": 0.17825469374656677,
+      "learning_rate": 4.212440371488343e-06,
+      "loss": 0.8818,
+      "step": 108800
+    },
+    {
+      "epoch": 0.7876392538383027,
+      "grad_norm": 0.191095769405365,
+      "learning_rate": 4.212367984827757e-06,
+      "loss": 0.8861,
+      "step": 108810
+    },
+    {
+      "epoch": 0.7877116404988889,
+      "grad_norm": 0.16488607227802277,
+      "learning_rate": 4.21229559816717e-06,
+      "loss": 0.8915,
+      "step": 108820
+    },
+    {
+      "epoch": 0.787784027159475,
+      "grad_norm": 0.15883904695510864,
+      "learning_rate": 4.212223211506584e-06,
+      "loss": 0.8806,
+      "step": 108830
+    },
+    {
+      "epoch": 0.7878564138200612,
+      "grad_norm": 0.1884288191795349,
+      "learning_rate": 4.2121508248459975e-06,
+      "loss": 0.8955,
+      "step": 108840
+    },
+    {
+      "epoch": 0.7879288004806474,
+      "grad_norm": 0.15324948728084564,
+      "learning_rate": 4.212078438185411e-06,
+      "loss": 0.8832,
+      "step": 108850
+    },
+    {
+      "epoch": 0.7880011871412336,
+      "grad_norm": 0.16065652668476105,
+      "learning_rate": 4.2120060515248255e-06,
+      "loss": 0.8755,
+      "step": 108860
+    },
+    {
+      "epoch": 0.7880735738018197,
+      "grad_norm": 0.15885667502880096,
+      "learning_rate": 4.211933664864239e-06,
+      "loss": 0.8754,
+      "step": 108870
+    },
+    {
+      "epoch": 0.788145960462406,
+      "grad_norm": 0.1391364336013794,
+      "learning_rate": 4.211861278203653e-06,
+      "loss": 0.8806,
+      "step": 108880
+    },
+    {
+      "epoch": 0.7882183471229922,
+      "grad_norm": 0.15506839752197266,
+      "learning_rate": 4.211788891543066e-06,
+      "loss": 0.8831,
+      "step": 108890
+    },
+    {
+      "epoch": 0.7882907337835784,
+      "grad_norm": 0.15888062119483948,
+      "learning_rate": 4.21171650488248e-06,
+      "loss": 0.8819,
+      "step": 108900
+    },
+    {
+      "epoch": 0.7883631204441646,
+      "grad_norm": 0.1553460657596588,
+      "learning_rate": 4.2116441182218945e-06,
+      "loss": 0.8916,
+      "step": 108910
+    },
+    {
+      "epoch": 0.7884355071047507,
+      "grad_norm": 0.14751103520393372,
+      "learning_rate": 4.211571731561308e-06,
+      "loss": 0.876,
+      "step": 108920
+    },
+    {
+      "epoch": 0.7885078937653369,
+      "grad_norm": 0.17180080711841583,
+      "learning_rate": 4.211499344900722e-06,
+      "loss": 0.8759,
+      "step": 108930
+    },
+    {
+      "epoch": 0.7885802804259231,
+      "grad_norm": 0.15554192662239075,
+      "learning_rate": 4.211426958240135e-06,
+      "loss": 0.8813,
+      "step": 108940
+    },
+    {
+      "epoch": 0.7886526670865093,
+      "grad_norm": 0.1551918089389801,
+      "learning_rate": 4.21135457157955e-06,
+      "loss": 0.884,
+      "step": 108950
+    },
+    {
+      "epoch": 0.7887250537470955,
+      "grad_norm": 0.17789483070373535,
+      "learning_rate": 4.211282184918963e-06,
+      "loss": 0.8821,
+      "step": 108960
+    },
+    {
+      "epoch": 0.7887974404076816,
+      "grad_norm": 0.16537193953990936,
+      "learning_rate": 4.211209798258377e-06,
+      "loss": 0.8805,
+      "step": 108970
+    },
+    {
+      "epoch": 0.7888698270682679,
+      "grad_norm": 0.15402130782604218,
+      "learning_rate": 4.211137411597791e-06,
+      "loss": 0.8804,
+      "step": 108980
+    },
+    {
+      "epoch": 0.7889422137288541,
+      "grad_norm": 0.15392087399959564,
+      "learning_rate": 4.211065024937205e-06,
+      "loss": 0.8975,
+      "step": 108990
+    },
+    {
+      "epoch": 0.7890146003894403,
+      "grad_norm": 0.1604469269514084,
+      "learning_rate": 4.210992638276619e-06,
+      "loss": 0.8778,
+      "step": 109000
+    },
+    {
+      "epoch": 0.7890869870500264,
+      "grad_norm": 0.14916957914829254,
+      "learning_rate": 4.210920251616032e-06,
+      "loss": 0.8859,
+      "step": 109010
+    },
+    {
+      "epoch": 0.7891593737106126,
+      "grad_norm": 0.15784524381160736,
+      "learning_rate": 4.210847864955446e-06,
+      "loss": 0.8833,
+      "step": 109020
+    },
+    {
+      "epoch": 0.7892317603711988,
+      "grad_norm": 0.26077041029930115,
+      "learning_rate": 4.21077547829486e-06,
+      "loss": 0.8962,
+      "step": 109030
+    },
+    {
+      "epoch": 0.789304147031785,
+      "grad_norm": 0.16116316616535187,
+      "learning_rate": 4.210703091634274e-06,
+      "loss": 0.8888,
+      "step": 109040
+    },
+    {
+      "epoch": 0.7893765336923712,
+      "grad_norm": 0.24238155782222748,
+      "learning_rate": 4.210630704973688e-06,
+      "loss": 0.8827,
+      "step": 109050
+    },
+    {
+      "epoch": 0.7894489203529573,
+      "grad_norm": 0.14621655642986298,
+      "learning_rate": 4.210558318313101e-06,
+      "loss": 0.8893,
+      "step": 109060
+    },
+    {
+      "epoch": 0.7895213070135435,
+      "grad_norm": 0.2018691748380661,
+      "learning_rate": 4.210485931652516e-06,
+      "loss": 0.8831,
+      "step": 109070
+    },
+    {
+      "epoch": 0.7895936936741297,
+      "grad_norm": 0.148194819688797,
+      "learning_rate": 4.210413544991929e-06,
+      "loss": 0.876,
+      "step": 109080
+    },
+    {
+      "epoch": 0.789666080334716,
+      "grad_norm": 0.15002113580703735,
+      "learning_rate": 4.210341158331343e-06,
+      "loss": 0.8754,
+      "step": 109090
+    },
+    {
+      "epoch": 0.7897384669953021,
+      "grad_norm": 0.15236304700374603,
+      "learning_rate": 4.2102687716707566e-06,
+      "loss": 0.8849,
+      "step": 109100
+    },
+    {
+      "epoch": 0.7898108536558883,
+      "grad_norm": 0.17015591263771057,
+      "learning_rate": 4.210196385010171e-06,
+      "loss": 0.8818,
+      "step": 109110
+    },
+    {
+      "epoch": 0.7898832403164745,
+      "grad_norm": 0.16506314277648926,
+      "learning_rate": 4.210123998349585e-06,
+      "loss": 0.8789,
+      "step": 109120
+    },
+    {
+      "epoch": 0.7899556269770607,
+      "grad_norm": 0.16899830102920532,
+      "learning_rate": 4.210051611688998e-06,
+      "loss": 0.8763,
+      "step": 109130
+    },
+    {
+      "epoch": 0.7900280136376469,
+      "grad_norm": 0.16103564202785492,
+      "learning_rate": 4.209979225028412e-06,
+      "loss": 0.8816,
+      "step": 109140
+    },
+    {
+      "epoch": 0.790100400298233,
+      "grad_norm": 0.15541161596775055,
+      "learning_rate": 4.209906838367826e-06,
+      "loss": 0.8864,
+      "step": 109150
+    },
+    {
+      "epoch": 0.7901727869588192,
+      "grad_norm": 0.15353739261627197,
+      "learning_rate": 4.20983445170724e-06,
+      "loss": 0.8881,
+      "step": 109160
+    },
+    {
+      "epoch": 0.7902451736194054,
+      "grad_norm": 0.1619664430618286,
+      "learning_rate": 4.2097620650466536e-06,
+      "loss": 0.8927,
+      "step": 109170
+    },
+    {
+      "epoch": 0.7903175602799916,
+      "grad_norm": 0.15257126092910767,
+      "learning_rate": 4.209689678386067e-06,
+      "loss": 0.8725,
+      "step": 109180
+    },
+    {
+      "epoch": 0.7903899469405777,
+      "grad_norm": 0.1462344229221344,
+      "learning_rate": 4.209617291725482e-06,
+      "loss": 0.8893,
+      "step": 109190
+    },
+    {
+      "epoch": 0.790462333601164,
+      "grad_norm": 0.15440794825553894,
+      "learning_rate": 4.209544905064895e-06,
+      "loss": 0.8796,
+      "step": 109200
+    },
+    {
+      "epoch": 0.7905347202617502,
+      "grad_norm": 0.1578925997018814,
+      "learning_rate": 4.209472518404309e-06,
+      "loss": 0.8796,
+      "step": 109210
+    },
+    {
+      "epoch": 0.7906071069223364,
+      "grad_norm": 0.1685589849948883,
+      "learning_rate": 4.2094001317437225e-06,
+      "loss": 0.8761,
+      "step": 109220
+    },
+    {
+      "epoch": 0.7906794935829226,
+      "grad_norm": 0.16151396930217743,
+      "learning_rate": 4.209327745083137e-06,
+      "loss": 0.8849,
+      "step": 109230
+    },
+    {
+      "epoch": 0.7907518802435087,
+      "grad_norm": 0.16625353693962097,
+      "learning_rate": 4.2092553584225506e-06,
+      "loss": 0.8767,
+      "step": 109240
+    },
+    {
+      "epoch": 0.7908242669040949,
+      "grad_norm": 0.16339325904846191,
+      "learning_rate": 4.209182971761964e-06,
+      "loss": 0.8856,
+      "step": 109250
+    },
+    {
+      "epoch": 0.7908966535646811,
+      "grad_norm": 0.16669736802577972,
+      "learning_rate": 4.209110585101378e-06,
+      "loss": 0.887,
+      "step": 109260
+    },
+    {
+      "epoch": 0.7909690402252673,
+      "grad_norm": 0.15320435166358948,
+      "learning_rate": 4.209038198440792e-06,
+      "loss": 0.8849,
+      "step": 109270
+    },
+    {
+      "epoch": 0.7910414268858534,
+      "grad_norm": 0.1580091118812561,
+      "learning_rate": 4.208965811780206e-06,
+      "loss": 0.8786,
+      "step": 109280
+    },
+    {
+      "epoch": 0.7911138135464396,
+      "grad_norm": 0.15802930295467377,
+      "learning_rate": 4.2088934251196195e-06,
+      "loss": 0.8855,
+      "step": 109290
+    },
+    {
+      "epoch": 0.7911862002070259,
+      "grad_norm": 0.1613604873418808,
+      "learning_rate": 4.208821038459033e-06,
+      "loss": 0.8679,
+      "step": 109300
+    },
+    {
+      "epoch": 0.7912585868676121,
+      "grad_norm": 0.17970921099185944,
+      "learning_rate": 4.208748651798448e-06,
+      "loss": 0.8885,
+      "step": 109310
+    },
+    {
+      "epoch": 0.7913309735281983,
+      "grad_norm": 0.18362759053707123,
+      "learning_rate": 4.208676265137861e-06,
+      "loss": 0.8811,
+      "step": 109320
+    },
+    {
+      "epoch": 0.7914033601887844,
+      "grad_norm": 0.16677528619766235,
+      "learning_rate": 4.208603878477275e-06,
+      "loss": 0.8826,
+      "step": 109330
+    },
+    {
+      "epoch": 0.7914757468493706,
+      "grad_norm": 0.17493285238742828,
+      "learning_rate": 4.2085314918166884e-06,
+      "loss": 0.873,
+      "step": 109340
+    },
+    {
+      "epoch": 0.7915481335099568,
+      "grad_norm": 0.16501615941524506,
+      "learning_rate": 4.208459105156102e-06,
+      "loss": 0.8832,
+      "step": 109350
+    },
+    {
+      "epoch": 0.791620520170543,
+      "grad_norm": 0.15602637827396393,
+      "learning_rate": 4.208386718495516e-06,
+      "loss": 0.8867,
+      "step": 109360
+    },
+    {
+      "epoch": 0.7916929068311291,
+      "grad_norm": 0.1481214165687561,
+      "learning_rate": 4.208314331834929e-06,
+      "loss": 0.8815,
+      "step": 109370
+    },
+    {
+      "epoch": 0.7917652934917153,
+      "grad_norm": 0.1560535579919815,
+      "learning_rate": 4.208241945174344e-06,
+      "loss": 0.8858,
+      "step": 109380
+    },
+    {
+      "epoch": 0.7918376801523015,
+      "grad_norm": 0.21015238761901855,
+      "learning_rate": 4.208169558513757e-06,
+      "loss": 0.8827,
+      "step": 109390
+    },
+    {
+      "epoch": 0.7919100668128877,
+      "grad_norm": 0.16126051545143127,
+      "learning_rate": 4.208097171853171e-06,
+      "loss": 0.8978,
+      "step": 109400
+    },
+    {
+      "epoch": 0.791982453473474,
+      "grad_norm": 0.15194587409496307,
+      "learning_rate": 4.208024785192585e-06,
+      "loss": 0.8785,
+      "step": 109410
+    },
+    {
+      "epoch": 0.7920548401340601,
+      "grad_norm": 0.1646006554365158,
+      "learning_rate": 4.207952398531999e-06,
+      "loss": 0.88,
+      "step": 109420
+    },
+    {
+      "epoch": 0.7921272267946463,
+      "grad_norm": 0.21086090803146362,
+      "learning_rate": 4.207880011871413e-06,
+      "loss": 0.8909,
+      "step": 109430
+    },
+    {
+      "epoch": 0.7921996134552325,
+      "grad_norm": 0.1583826094865799,
+      "learning_rate": 4.207807625210826e-06,
+      "loss": 0.8872,
+      "step": 109440
+    },
+    {
+      "epoch": 0.7922720001158187,
+      "grad_norm": 0.15756238996982574,
+      "learning_rate": 4.20773523855024e-06,
+      "loss": 0.8844,
+      "step": 109450
+    },
+    {
+      "epoch": 0.7923443867764048,
+      "grad_norm": 0.1494230180978775,
+      "learning_rate": 4.207662851889654e-06,
+      "loss": 0.8814,
+      "step": 109460
+    },
+    {
+      "epoch": 0.792416773436991,
+      "grad_norm": 0.17796126008033752,
+      "learning_rate": 4.207590465229068e-06,
+      "loss": 0.8824,
+      "step": 109470
+    },
+    {
+      "epoch": 0.7924891600975772,
+      "grad_norm": 0.16645440459251404,
+      "learning_rate": 4.207518078568482e-06,
+      "loss": 0.892,
+      "step": 109480
+    },
+    {
+      "epoch": 0.7925615467581634,
+      "grad_norm": 0.16064368188381195,
+      "learning_rate": 4.207445691907895e-06,
+      "loss": 0.8843,
+      "step": 109490
+    },
+    {
+      "epoch": 0.7926339334187495,
+      "grad_norm": 0.18398964405059814,
+      "learning_rate": 4.207373305247309e-06,
+      "loss": 0.8878,
+      "step": 109500
+    },
+    {
+      "epoch": 0.7927063200793357,
+      "grad_norm": 0.14704552292823792,
+      "learning_rate": 4.207300918586723e-06,
+      "loss": 0.8858,
+      "step": 109510
+    },
+    {
+      "epoch": 0.792778706739922,
+      "grad_norm": 0.15785852074623108,
+      "learning_rate": 4.207228531926137e-06,
+      "loss": 0.8876,
+      "step": 109520
+    },
+    {
+      "epoch": 0.7928510934005082,
+      "grad_norm": 0.16036728024482727,
+      "learning_rate": 4.2071561452655505e-06,
+      "loss": 0.8926,
+      "step": 109530
+    },
+    {
+      "epoch": 0.7929234800610944,
+      "grad_norm": 0.1538962423801422,
+      "learning_rate": 4.207083758604964e-06,
+      "loss": 0.89,
+      "step": 109540
+    },
+    {
+      "epoch": 0.7929958667216805,
+      "grad_norm": 0.15451432764530182,
+      "learning_rate": 4.207011371944379e-06,
+      "loss": 0.8863,
+      "step": 109550
+    },
+    {
+      "epoch": 0.7930682533822667,
+      "grad_norm": 0.1714572012424469,
+      "learning_rate": 4.206938985283792e-06,
+      "loss": 0.8779,
+      "step": 109560
+    },
+    {
+      "epoch": 0.7931406400428529,
+      "grad_norm": 0.15013469755649567,
+      "learning_rate": 4.206866598623206e-06,
+      "loss": 0.8753,
+      "step": 109570
+    },
+    {
+      "epoch": 0.7932130267034391,
+      "grad_norm": 0.15864507853984833,
+      "learning_rate": 4.2067942119626195e-06,
+      "loss": 0.8817,
+      "step": 109580
+    },
+    {
+      "epoch": 0.7932854133640252,
+      "grad_norm": 0.18490475416183472,
+      "learning_rate": 4.206721825302034e-06,
+      "loss": 0.8838,
+      "step": 109590
+    },
+    {
+      "epoch": 0.7933578000246114,
+      "grad_norm": 0.15001821517944336,
+      "learning_rate": 4.2066494386414475e-06,
+      "loss": 0.8807,
+      "step": 109600
+    },
+    {
+      "epoch": 0.7934301866851976,
+      "grad_norm": 0.1856260895729065,
+      "learning_rate": 4.206577051980861e-06,
+      "loss": 0.8846,
+      "step": 109610
+    },
+    {
+      "epoch": 0.7935025733457839,
+      "grad_norm": 0.14746098220348358,
+      "learning_rate": 4.206504665320275e-06,
+      "loss": 0.8684,
+      "step": 109620
+    },
+    {
+      "epoch": 0.7935749600063701,
+      "grad_norm": 0.16342267394065857,
+      "learning_rate": 4.206432278659689e-06,
+      "loss": 0.8689,
+      "step": 109630
+    },
+    {
+      "epoch": 0.7936473466669562,
+      "grad_norm": 0.15328651666641235,
+      "learning_rate": 4.206359891999103e-06,
+      "loss": 0.8678,
+      "step": 109640
+    },
+    {
+      "epoch": 0.7937197333275424,
+      "grad_norm": 0.164528951048851,
+      "learning_rate": 4.2062875053385165e-06,
+      "loss": 0.8783,
+      "step": 109650
+    },
+    {
+      "epoch": 0.7937921199881286,
+      "grad_norm": 0.1499549150466919,
+      "learning_rate": 4.20621511867793e-06,
+      "loss": 0.8813,
+      "step": 109660
+    },
+    {
+      "epoch": 0.7938645066487148,
+      "grad_norm": 0.14804977178573608,
+      "learning_rate": 4.2061427320173445e-06,
+      "loss": 0.8748,
+      "step": 109670
+    },
+    {
+      "epoch": 0.793936893309301,
+      "grad_norm": 0.1729278266429901,
+      "learning_rate": 4.206070345356758e-06,
+      "loss": 0.8744,
+      "step": 109680
+    },
+    {
+      "epoch": 0.7940092799698871,
+      "grad_norm": 0.15878424048423767,
+      "learning_rate": 4.205997958696172e-06,
+      "loss": 0.8887,
+      "step": 109690
+    },
+    {
+      "epoch": 0.7940816666304733,
+      "grad_norm": 0.15942630171775818,
+      "learning_rate": 4.205925572035585e-06,
+      "loss": 0.8915,
+      "step": 109700
+    },
+    {
+      "epoch": 0.7941540532910595,
+      "grad_norm": 0.16669262945652008,
+      "learning_rate": 4.205853185375e-06,
+      "loss": 0.8915,
+      "step": 109710
+    },
+    {
+      "epoch": 0.7942264399516457,
+      "grad_norm": 0.1515069305896759,
+      "learning_rate": 4.2057807987144135e-06,
+      "loss": 0.8784,
+      "step": 109720
+    },
+    {
+      "epoch": 0.794298826612232,
+      "grad_norm": 0.1678621619939804,
+      "learning_rate": 4.205708412053827e-06,
+      "loss": 0.8901,
+      "step": 109730
+    },
+    {
+      "epoch": 0.7943712132728181,
+      "grad_norm": 0.17552782595157623,
+      "learning_rate": 4.205636025393241e-06,
+      "loss": 0.8664,
+      "step": 109740
+    },
+    {
+      "epoch": 0.7944435999334043,
+      "grad_norm": 0.1632026731967926,
+      "learning_rate": 4.205563638732655e-06,
+      "loss": 0.8854,
+      "step": 109750
+    },
+    {
+      "epoch": 0.7945159865939905,
+      "grad_norm": 0.15322089195251465,
+      "learning_rate": 4.205491252072069e-06,
+      "loss": 0.8788,
+      "step": 109760
+    },
+    {
+      "epoch": 0.7945883732545767,
+      "grad_norm": 0.1526355892419815,
+      "learning_rate": 4.205418865411482e-06,
+      "loss": 0.8773,
+      "step": 109770
+    },
+    {
+      "epoch": 0.7946607599151628,
+      "grad_norm": 0.1596207320690155,
+      "learning_rate": 4.205346478750896e-06,
+      "loss": 0.8807,
+      "step": 109780
+    },
+    {
+      "epoch": 0.794733146575749,
+      "grad_norm": 0.16599130630493164,
+      "learning_rate": 4.2052740920903105e-06,
+      "loss": 0.8716,
+      "step": 109790
+    },
+    {
+      "epoch": 0.7948055332363352,
+      "grad_norm": 0.1544189453125,
+      "learning_rate": 4.205201705429724e-06,
+      "loss": 0.8769,
+      "step": 109800
+    },
+    {
+      "epoch": 0.7948779198969214,
+      "grad_norm": 0.1597534716129303,
+      "learning_rate": 4.205129318769138e-06,
+      "loss": 0.8723,
+      "step": 109810
+    },
+    {
+      "epoch": 0.7949503065575075,
+      "grad_norm": 0.15642835199832916,
+      "learning_rate": 4.205056932108551e-06,
+      "loss": 0.8835,
+      "step": 109820
+    },
+    {
+      "epoch": 0.7950226932180938,
+      "grad_norm": 0.16617697477340698,
+      "learning_rate": 4.204984545447966e-06,
+      "loss": 0.8745,
+      "step": 109830
+    },
+    {
+      "epoch": 0.79509507987868,
+      "grad_norm": 0.15826547145843506,
+      "learning_rate": 4.204912158787379e-06,
+      "loss": 0.8885,
+      "step": 109840
+    },
+    {
+      "epoch": 0.7951674665392662,
+      "grad_norm": 0.16026552021503448,
+      "learning_rate": 4.204839772126793e-06,
+      "loss": 0.8901,
+      "step": 109850
+    },
+    {
+      "epoch": 0.7952398531998524,
+      "grad_norm": 0.16128748655319214,
+      "learning_rate": 4.204767385466207e-06,
+      "loss": 0.8934,
+      "step": 109860
+    },
+    {
+      "epoch": 0.7953122398604385,
+      "grad_norm": 0.1773352324962616,
+      "learning_rate": 4.204694998805621e-06,
+      "loss": 0.8806,
+      "step": 109870
+    },
+    {
+      "epoch": 0.7953846265210247,
+      "grad_norm": 0.16145089268684387,
+      "learning_rate": 4.204622612145035e-06,
+      "loss": 0.8779,
+      "step": 109880
+    },
+    {
+      "epoch": 0.7954570131816109,
+      "grad_norm": 0.14754821360111237,
+      "learning_rate": 4.2045502254844475e-06,
+      "loss": 0.8763,
+      "step": 109890
+    },
+    {
+      "epoch": 0.7955293998421971,
+      "grad_norm": 0.15587793290615082,
+      "learning_rate": 4.204477838823862e-06,
+      "loss": 0.8794,
+      "step": 109900
+    },
+    {
+      "epoch": 0.7956017865027832,
+      "grad_norm": 0.14746980369091034,
+      "learning_rate": 4.2044054521632756e-06,
+      "loss": 0.887,
+      "step": 109910
+    },
+    {
+      "epoch": 0.7956741731633694,
+      "grad_norm": 0.15477393567562103,
+      "learning_rate": 4.204333065502689e-06,
+      "loss": 0.8808,
+      "step": 109920
+    },
+    {
+      "epoch": 0.7957465598239556,
+      "grad_norm": 0.14774759113788605,
+      "learning_rate": 4.204260678842103e-06,
+      "loss": 0.8813,
+      "step": 109930
+    },
+    {
+      "epoch": 0.7958189464845419,
+      "grad_norm": 0.1507684290409088,
+      "learning_rate": 4.204188292181517e-06,
+      "loss": 0.8908,
+      "step": 109940
+    },
+    {
+      "epoch": 0.795891333145128,
+      "grad_norm": 0.15101462602615356,
+      "learning_rate": 4.204115905520931e-06,
+      "loss": 0.8836,
+      "step": 109950
+    },
+    {
+      "epoch": 0.7959637198057142,
+      "grad_norm": 0.14433503150939941,
+      "learning_rate": 4.2040435188603445e-06,
+      "loss": 0.8769,
+      "step": 109960
+    },
+    {
+      "epoch": 0.7960361064663004,
+      "grad_norm": 0.17462539672851562,
+      "learning_rate": 4.203971132199758e-06,
+      "loss": 0.8851,
+      "step": 109970
+    },
+    {
+      "epoch": 0.7961084931268866,
+      "grad_norm": 0.1587734818458557,
+      "learning_rate": 4.2038987455391726e-06,
+      "loss": 0.8821,
+      "step": 109980
+    },
+    {
+      "epoch": 0.7961808797874728,
+      "grad_norm": 0.14876559376716614,
+      "learning_rate": 4.203826358878586e-06,
+      "loss": 0.8717,
+      "step": 109990
+    },
+    {
+      "epoch": 0.7962532664480589,
+      "grad_norm": 0.1661464273929596,
+      "learning_rate": 4.203753972218e-06,
+      "loss": 0.8884,
+      "step": 110000
+    },
+    {
+      "epoch": 0.7963256531086451,
+      "grad_norm": 0.21970897912979126,
+      "learning_rate": 4.203681585557413e-06,
+      "loss": 0.8812,
+      "step": 110010
+    },
+    {
+      "epoch": 0.7963980397692313,
+      "grad_norm": 0.15922003984451294,
+      "learning_rate": 4.203609198896828e-06,
+      "loss": 0.8924,
+      "step": 110020
+    },
+    {
+      "epoch": 0.7964704264298175,
+      "grad_norm": 0.1996181309223175,
+      "learning_rate": 4.2035368122362415e-06,
+      "loss": 0.8815,
+      "step": 110030
+    },
+    {
+      "epoch": 0.7965428130904036,
+      "grad_norm": 0.14531823992729187,
+      "learning_rate": 4.203464425575655e-06,
+      "loss": 0.8849,
+      "step": 110040
+    },
+    {
+      "epoch": 0.7966151997509899,
+      "grad_norm": 0.15441548824310303,
+      "learning_rate": 4.203392038915069e-06,
+      "loss": 0.8793,
+      "step": 110050
+    },
+    {
+      "epoch": 0.7966875864115761,
+      "grad_norm": 0.15467193722724915,
+      "learning_rate": 4.203319652254483e-06,
+      "loss": 0.8773,
+      "step": 110060
+    },
+    {
+      "epoch": 0.7967599730721623,
+      "grad_norm": 0.15833494067192078,
+      "learning_rate": 4.203247265593897e-06,
+      "loss": 0.8748,
+      "step": 110070
+    },
+    {
+      "epoch": 0.7968323597327485,
+      "grad_norm": 0.1572531759738922,
+      "learning_rate": 4.2031748789333104e-06,
+      "loss": 0.8769,
+      "step": 110080
+    },
+    {
+      "epoch": 0.7969047463933346,
+      "grad_norm": 0.16923896968364716,
+      "learning_rate": 4.203102492272724e-06,
+      "loss": 0.8714,
+      "step": 110090
+    },
+    {
+      "epoch": 0.7969771330539208,
+      "grad_norm": 0.15771649777889252,
+      "learning_rate": 4.2030301056121385e-06,
+      "loss": 0.8861,
+      "step": 110100
+    },
+    {
+      "epoch": 0.797049519714507,
+      "grad_norm": 0.1439589411020279,
+      "learning_rate": 4.202957718951552e-06,
+      "loss": 0.8705,
+      "step": 110110
+    },
+    {
+      "epoch": 0.7971219063750932,
+      "grad_norm": 0.14609919488430023,
+      "learning_rate": 4.202885332290966e-06,
+      "loss": 0.883,
+      "step": 110120
+    },
+    {
+      "epoch": 0.7971942930356793,
+      "grad_norm": 0.15220913290977478,
+      "learning_rate": 4.202812945630379e-06,
+      "loss": 0.889,
+      "step": 110130
+    },
+    {
+      "epoch": 0.7972666796962655,
+      "grad_norm": 0.16518516838550568,
+      "learning_rate": 4.202740558969793e-06,
+      "loss": 0.8745,
+      "step": 110140
+    },
+    {
+      "epoch": 0.7973390663568518,
+      "grad_norm": 0.15049389004707336,
+      "learning_rate": 4.2026681723092074e-06,
+      "loss": 0.8761,
+      "step": 110150
+    },
+    {
+      "epoch": 0.797411453017438,
+      "grad_norm": 0.13817547261714935,
+      "learning_rate": 4.202595785648621e-06,
+      "loss": 0.8687,
+      "step": 110160
+    },
+    {
+      "epoch": 0.7974838396780242,
+      "grad_norm": 0.15669967234134674,
+      "learning_rate": 4.202523398988035e-06,
+      "loss": 0.8928,
+      "step": 110170
+    },
+    {
+      "epoch": 0.7975562263386103,
+      "grad_norm": 0.15119336545467377,
+      "learning_rate": 4.202451012327448e-06,
+      "loss": 0.8827,
+      "step": 110180
+    },
+    {
+      "epoch": 0.7976286129991965,
+      "grad_norm": 0.14785991609096527,
+      "learning_rate": 4.202378625666863e-06,
+      "loss": 0.8917,
+      "step": 110190
+    },
+    {
+      "epoch": 0.7977009996597827,
+      "grad_norm": 0.15478040277957916,
+      "learning_rate": 4.202306239006276e-06,
+      "loss": 0.873,
+      "step": 110200
+    },
+    {
+      "epoch": 0.7977733863203689,
+      "grad_norm": 0.17014062404632568,
+      "learning_rate": 4.20223385234569e-06,
+      "loss": 0.885,
+      "step": 110210
+    },
+    {
+      "epoch": 0.797845772980955,
+      "grad_norm": 0.1526460498571396,
+      "learning_rate": 4.202161465685104e-06,
+      "loss": 0.8793,
+      "step": 110220
+    },
+    {
+      "epoch": 0.7979181596415412,
+      "grad_norm": 0.15299007296562195,
+      "learning_rate": 4.202089079024518e-06,
+      "loss": 0.8705,
+      "step": 110230
+    },
+    {
+      "epoch": 0.7979905463021274,
+      "grad_norm": 0.1601126492023468,
+      "learning_rate": 4.202016692363932e-06,
+      "loss": 0.8854,
+      "step": 110240
+    },
+    {
+      "epoch": 0.7980629329627136,
+      "grad_norm": 0.14891035854816437,
+      "learning_rate": 4.201944305703345e-06,
+      "loss": 0.8736,
+      "step": 110250
+    },
+    {
+      "epoch": 0.7981353196232999,
+      "grad_norm": 0.17252424359321594,
+      "learning_rate": 4.201871919042759e-06,
+      "loss": 0.8858,
+      "step": 110260
+    },
+    {
+      "epoch": 0.798207706283886,
+      "grad_norm": 0.16554145514965057,
+      "learning_rate": 4.201799532382173e-06,
+      "loss": 0.8679,
+      "step": 110270
+    },
+    {
+      "epoch": 0.7982800929444722,
+      "grad_norm": 0.17143869400024414,
+      "learning_rate": 4.201727145721587e-06,
+      "loss": 0.8895,
+      "step": 110280
+    },
+    {
+      "epoch": 0.7983524796050584,
+      "grad_norm": 0.14797469973564148,
+      "learning_rate": 4.201654759061001e-06,
+      "loss": 0.8776,
+      "step": 110290
+    },
+    {
+      "epoch": 0.7984248662656446,
+      "grad_norm": 0.15571478009223938,
+      "learning_rate": 4.201582372400414e-06,
+      "loss": 0.8765,
+      "step": 110300
+    },
+    {
+      "epoch": 0.7984972529262307,
+      "grad_norm": 0.1487504541873932,
+      "learning_rate": 4.201509985739829e-06,
+      "loss": 0.8822,
+      "step": 110310
+    },
+    {
+      "epoch": 0.7985696395868169,
+      "grad_norm": 0.15721207857131958,
+      "learning_rate": 4.201437599079242e-06,
+      "loss": 0.8609,
+      "step": 110320
+    },
+    {
+      "epoch": 0.7986420262474031,
+      "grad_norm": 0.14724335074424744,
+      "learning_rate": 4.201365212418656e-06,
+      "loss": 0.8808,
+      "step": 110330
+    },
+    {
+      "epoch": 0.7987144129079893,
+      "grad_norm": 0.1642737239599228,
+      "learning_rate": 4.2012928257580695e-06,
+      "loss": 0.8882,
+      "step": 110340
+    },
+    {
+      "epoch": 0.7987867995685755,
+      "grad_norm": 0.1529485136270523,
+      "learning_rate": 4.201220439097484e-06,
+      "loss": 0.8793,
+      "step": 110350
+    },
+    {
+      "epoch": 0.7988591862291617,
+      "grad_norm": 0.14472009241580963,
+      "learning_rate": 4.201148052436898e-06,
+      "loss": 0.8705,
+      "step": 110360
+    },
+    {
+      "epoch": 0.7989315728897479,
+      "grad_norm": 0.16149692237377167,
+      "learning_rate": 4.201075665776311e-06,
+      "loss": 0.8788,
+      "step": 110370
+    },
+    {
+      "epoch": 0.7990039595503341,
+      "grad_norm": 0.16890741884708405,
+      "learning_rate": 4.201003279115725e-06,
+      "loss": 0.8714,
+      "step": 110380
+    },
+    {
+      "epoch": 0.7990763462109203,
+      "grad_norm": 0.16988790035247803,
+      "learning_rate": 4.200930892455139e-06,
+      "loss": 0.8946,
+      "step": 110390
+    },
+    {
+      "epoch": 0.7991487328715065,
+      "grad_norm": 0.14879997074604034,
+      "learning_rate": 4.200858505794553e-06,
+      "loss": 0.8776,
+      "step": 110400
+    },
+    {
+      "epoch": 0.7992211195320926,
+      "grad_norm": 0.38371655344963074,
+      "learning_rate": 4.2007861191339665e-06,
+      "loss": 0.892,
+      "step": 110410
+    },
+    {
+      "epoch": 0.7992935061926788,
+      "grad_norm": 0.14874762296676636,
+      "learning_rate": 4.20071373247338e-06,
+      "loss": 0.8787,
+      "step": 110420
+    },
+    {
+      "epoch": 0.799365892853265,
+      "grad_norm": 0.1458701342344284,
+      "learning_rate": 4.200641345812794e-06,
+      "loss": 0.8788,
+      "step": 110430
+    },
+    {
+      "epoch": 0.7994382795138512,
+      "grad_norm": 0.1612052172422409,
+      "learning_rate": 4.200568959152207e-06,
+      "loss": 0.8816,
+      "step": 110440
+    },
+    {
+      "epoch": 0.7995106661744373,
+      "grad_norm": 0.1589481681585312,
+      "learning_rate": 4.200496572491621e-06,
+      "loss": 0.8847,
+      "step": 110450
+    },
+    {
+      "epoch": 0.7995830528350235,
+      "grad_norm": 0.15650621056556702,
+      "learning_rate": 4.2004241858310355e-06,
+      "loss": 0.8862,
+      "step": 110460
+    },
+    {
+      "epoch": 0.7996554394956098,
+      "grad_norm": 0.1470274180173874,
+      "learning_rate": 4.200351799170449e-06,
+      "loss": 0.8822,
+      "step": 110470
+    },
+    {
+      "epoch": 0.799727826156196,
+      "grad_norm": 0.15329478681087494,
+      "learning_rate": 4.200279412509863e-06,
+      "loss": 0.8777,
+      "step": 110480
+    },
+    {
+      "epoch": 0.7998002128167822,
+      "grad_norm": 0.16613824665546417,
+      "learning_rate": 4.200207025849276e-06,
+      "loss": 0.8915,
+      "step": 110490
+    },
+    {
+      "epoch": 0.7998725994773683,
+      "grad_norm": 0.15577919781208038,
+      "learning_rate": 4.200134639188691e-06,
+      "loss": 0.8799,
+      "step": 110500
+    },
+    {
+      "epoch": 0.7999449861379545,
+      "grad_norm": 0.16274091601371765,
+      "learning_rate": 4.200062252528104e-06,
+      "loss": 0.8822,
+      "step": 110510
+    },
+    {
+      "epoch": 0.8000173727985407,
+      "grad_norm": 0.16509264707565308,
+      "learning_rate": 4.199989865867518e-06,
+      "loss": 0.8921,
+      "step": 110520
+    },
+    {
+      "epoch": 0.8000897594591269,
+      "grad_norm": 0.16428282856941223,
+      "learning_rate": 4.199917479206932e-06,
+      "loss": 0.861,
+      "step": 110530
+    },
+    {
+      "epoch": 0.800162146119713,
+      "grad_norm": 0.15787632763385773,
+      "learning_rate": 4.199845092546346e-06,
+      "loss": 0.8679,
+      "step": 110540
+    },
+    {
+      "epoch": 0.8002345327802992,
+      "grad_norm": 0.21818390488624573,
+      "learning_rate": 4.19977270588576e-06,
+      "loss": 0.8813,
+      "step": 110550
+    },
+    {
+      "epoch": 0.8003069194408854,
+      "grad_norm": 0.16257944703102112,
+      "learning_rate": 4.199700319225173e-06,
+      "loss": 0.8811,
+      "step": 110560
+    },
+    {
+      "epoch": 0.8003793061014716,
+      "grad_norm": 0.16678005456924438,
+      "learning_rate": 4.199627932564587e-06,
+      "loss": 0.8873,
+      "step": 110570
+    },
+    {
+      "epoch": 0.8004516927620579,
+      "grad_norm": 0.15862931311130524,
+      "learning_rate": 4.199555545904001e-06,
+      "loss": 0.8825,
+      "step": 110580
+    },
+    {
+      "epoch": 0.800524079422644,
+      "grad_norm": 0.18531429767608643,
+      "learning_rate": 4.199483159243415e-06,
+      "loss": 0.8642,
+      "step": 110590
+    },
+    {
+      "epoch": 0.8005964660832302,
+      "grad_norm": 0.15238986909389496,
+      "learning_rate": 4.199410772582829e-06,
+      "loss": 0.882,
+      "step": 110600
+    },
+    {
+      "epoch": 0.8006688527438164,
+      "grad_norm": 0.15185806155204773,
+      "learning_rate": 4.199338385922242e-06,
+      "loss": 0.8812,
+      "step": 110610
+    },
+    {
+      "epoch": 0.8007412394044026,
+      "grad_norm": 0.14736582338809967,
+      "learning_rate": 4.199265999261657e-06,
+      "loss": 0.8764,
+      "step": 110620
+    },
+    {
+      "epoch": 0.8008136260649887,
+      "grad_norm": 0.191138356924057,
+      "learning_rate": 4.19919361260107e-06,
+      "loss": 0.8821,
+      "step": 110630
+    },
+    {
+      "epoch": 0.8008860127255749,
+      "grad_norm": 0.1546011120080948,
+      "learning_rate": 4.199121225940484e-06,
+      "loss": 0.8797,
+      "step": 110640
+    },
+    {
+      "epoch": 0.8009583993861611,
+      "grad_norm": 0.14093956351280212,
+      "learning_rate": 4.1990488392798976e-06,
+      "loss": 0.8744,
+      "step": 110650
+    },
+    {
+      "epoch": 0.8010307860467473,
+      "grad_norm": 0.14860482513904572,
+      "learning_rate": 4.198976452619312e-06,
+      "loss": 0.8811,
+      "step": 110660
+    },
+    {
+      "epoch": 0.8011031727073334,
+      "grad_norm": 0.15203125774860382,
+      "learning_rate": 4.198904065958726e-06,
+      "loss": 0.8726,
+      "step": 110670
+    },
+    {
+      "epoch": 0.8011755593679197,
+      "grad_norm": 0.1518326997756958,
+      "learning_rate": 4.198831679298139e-06,
+      "loss": 0.8747,
+      "step": 110680
+    },
+    {
+      "epoch": 0.8012479460285059,
+      "grad_norm": 0.16520649194717407,
+      "learning_rate": 4.198759292637553e-06,
+      "loss": 0.8794,
+      "step": 110690
+    },
+    {
+      "epoch": 0.8013203326890921,
+      "grad_norm": 0.16305768489837646,
+      "learning_rate": 4.198686905976967e-06,
+      "loss": 0.8866,
+      "step": 110700
+    },
+    {
+      "epoch": 0.8013927193496783,
+      "grad_norm": 0.15262234210968018,
+      "learning_rate": 4.198614519316381e-06,
+      "loss": 0.8865,
+      "step": 110710
+    },
+    {
+      "epoch": 0.8014651060102644,
+      "grad_norm": 0.163343608379364,
+      "learning_rate": 4.1985421326557946e-06,
+      "loss": 0.8781,
+      "step": 110720
+    },
+    {
+      "epoch": 0.8015374926708506,
+      "grad_norm": 0.15089716017246246,
+      "learning_rate": 4.198469745995208e-06,
+      "loss": 0.883,
+      "step": 110730
+    },
+    {
+      "epoch": 0.8016098793314368,
+      "grad_norm": 0.1560911238193512,
+      "learning_rate": 4.198397359334622e-06,
+      "loss": 0.8744,
+      "step": 110740
+    },
+    {
+      "epoch": 0.801682265992023,
+      "grad_norm": 1.2095776796340942,
+      "learning_rate": 4.198324972674036e-06,
+      "loss": 0.8845,
+      "step": 110750
+    },
+    {
+      "epoch": 0.8017546526526091,
+      "grad_norm": 0.15118081867694855,
+      "learning_rate": 4.19825258601345e-06,
+      "loss": 0.8761,
+      "step": 110760
+    },
+    {
+      "epoch": 0.8018270393131953,
+      "grad_norm": 0.15144529938697815,
+      "learning_rate": 4.1981801993528635e-06,
+      "loss": 0.879,
+      "step": 110770
+    },
+    {
+      "epoch": 0.8018994259737815,
+      "grad_norm": 0.16330844163894653,
+      "learning_rate": 4.198107812692277e-06,
+      "loss": 0.8809,
+      "step": 110780
+    },
+    {
+      "epoch": 0.8019718126343678,
+      "grad_norm": 0.14975833892822266,
+      "learning_rate": 4.198035426031692e-06,
+      "loss": 0.8788,
+      "step": 110790
+    },
+    {
+      "epoch": 0.802044199294954,
+      "grad_norm": 0.15585237741470337,
+      "learning_rate": 4.197963039371105e-06,
+      "loss": 0.882,
+      "step": 110800
+    },
+    {
+      "epoch": 0.8021165859555401,
+      "grad_norm": 0.14996300637722015,
+      "learning_rate": 4.197890652710519e-06,
+      "loss": 0.8803,
+      "step": 110810
+    },
+    {
+      "epoch": 0.8021889726161263,
+      "grad_norm": 0.1632111817598343,
+      "learning_rate": 4.1978182660499324e-06,
+      "loss": 0.8709,
+      "step": 110820
+    },
+    {
+      "epoch": 0.8022613592767125,
+      "grad_norm": 0.1543271541595459,
+      "learning_rate": 4.197745879389347e-06,
+      "loss": 0.8823,
+      "step": 110830
+    },
+    {
+      "epoch": 0.8023337459372987,
+      "grad_norm": 0.25150734186172485,
+      "learning_rate": 4.1976734927287605e-06,
+      "loss": 0.8912,
+      "step": 110840
+    },
+    {
+      "epoch": 0.8024061325978848,
+      "grad_norm": 0.14821644127368927,
+      "learning_rate": 4.197601106068174e-06,
+      "loss": 0.8951,
+      "step": 110850
+    },
+    {
+      "epoch": 0.802478519258471,
+      "grad_norm": 0.5984487533569336,
+      "learning_rate": 4.197528719407588e-06,
+      "loss": 0.8829,
+      "step": 110860
+    },
+    {
+      "epoch": 0.8025509059190572,
+      "grad_norm": 0.1596580147743225,
+      "learning_rate": 4.197456332747002e-06,
+      "loss": 0.8795,
+      "step": 110870
+    },
+    {
+      "epoch": 0.8026232925796434,
+      "grad_norm": 0.1528296023607254,
+      "learning_rate": 4.197383946086416e-06,
+      "loss": 0.8802,
+      "step": 110880
+    },
+    {
+      "epoch": 0.8026956792402297,
+      "grad_norm": 0.15530245006084442,
+      "learning_rate": 4.1973115594258294e-06,
+      "loss": 0.8859,
+      "step": 110890
+    },
+    {
+      "epoch": 0.8027680659008158,
+      "grad_norm": 0.1550951898097992,
+      "learning_rate": 4.197239172765243e-06,
+      "loss": 0.8788,
+      "step": 110900
+    },
+    {
+      "epoch": 0.802840452561402,
+      "grad_norm": 0.17830154299736023,
+      "learning_rate": 4.1971667861046575e-06,
+      "loss": 0.8782,
+      "step": 110910
+    },
+    {
+      "epoch": 0.8029128392219882,
+      "grad_norm": 0.1467135101556778,
+      "learning_rate": 4.197094399444071e-06,
+      "loss": 0.8776,
+      "step": 110920
+    },
+    {
+      "epoch": 0.8029852258825744,
+      "grad_norm": 0.1502387672662735,
+      "learning_rate": 4.197022012783485e-06,
+      "loss": 0.8782,
+      "step": 110930
+    },
+    {
+      "epoch": 0.8030576125431605,
+      "grad_norm": 0.14576847851276398,
+      "learning_rate": 4.196949626122898e-06,
+      "loss": 0.8663,
+      "step": 110940
+    },
+    {
+      "epoch": 0.8031299992037467,
+      "grad_norm": 0.14512509107589722,
+      "learning_rate": 4.196877239462312e-06,
+      "loss": 0.878,
+      "step": 110950
+    },
+    {
+      "epoch": 0.8032023858643329,
+      "grad_norm": 0.1578640192747116,
+      "learning_rate": 4.196804852801726e-06,
+      "loss": 0.8812,
+      "step": 110960
+    },
+    {
+      "epoch": 0.8032747725249191,
+      "grad_norm": 0.1540568768978119,
+      "learning_rate": 4.196732466141139e-06,
+      "loss": 0.8948,
+      "step": 110970
+    },
+    {
+      "epoch": 0.8033471591855053,
+      "grad_norm": 0.16598433256149292,
+      "learning_rate": 4.196660079480554e-06,
+      "loss": 0.8928,
+      "step": 110980
+    },
+    {
+      "epoch": 0.8034195458460914,
+      "grad_norm": 0.15447190403938293,
+      "learning_rate": 4.196587692819967e-06,
+      "loss": 0.8822,
+      "step": 110990
+    },
+    {
+      "epoch": 0.8034919325066777,
+      "grad_norm": 0.15389618277549744,
+      "learning_rate": 4.196515306159381e-06,
+      "loss": 0.872,
+      "step": 111000
+    },
+    {
+      "epoch": 0.8035643191672639,
+      "grad_norm": 0.1530335396528244,
+      "learning_rate": 4.1964429194987945e-06,
+      "loss": 0.871,
+      "step": 111010
+    },
+    {
+      "epoch": 0.8036367058278501,
+      "grad_norm": 0.15146398544311523,
+      "learning_rate": 4.196370532838209e-06,
+      "loss": 0.8834,
+      "step": 111020
+    },
+    {
+      "epoch": 0.8037090924884362,
+      "grad_norm": 0.15704941749572754,
+      "learning_rate": 4.196298146177623e-06,
+      "loss": 0.9041,
+      "step": 111030
+    },
+    {
+      "epoch": 0.8037814791490224,
+      "grad_norm": 0.15479826927185059,
+      "learning_rate": 4.196225759517036e-06,
+      "loss": 0.8853,
+      "step": 111040
+    },
+    {
+      "epoch": 0.8038538658096086,
+      "grad_norm": 0.1650967299938202,
+      "learning_rate": 4.19615337285645e-06,
+      "loss": 0.8816,
+      "step": 111050
+    },
+    {
+      "epoch": 0.8039262524701948,
+      "grad_norm": 0.15710927546024323,
+      "learning_rate": 4.196080986195864e-06,
+      "loss": 0.8873,
+      "step": 111060
+    },
+    {
+      "epoch": 0.803998639130781,
+      "grad_norm": 0.15104107558727264,
+      "learning_rate": 4.196008599535278e-06,
+      "loss": 0.8729,
+      "step": 111070
+    },
+    {
+      "epoch": 0.8040710257913671,
+      "grad_norm": 0.15065807104110718,
+      "learning_rate": 4.1959362128746915e-06,
+      "loss": 0.8849,
+      "step": 111080
+    },
+    {
+      "epoch": 0.8041434124519533,
+      "grad_norm": 0.16708062589168549,
+      "learning_rate": 4.195863826214105e-06,
+      "loss": 0.8839,
+      "step": 111090
+    },
+    {
+      "epoch": 0.8042157991125395,
+      "grad_norm": 0.155837744474411,
+      "learning_rate": 4.19579143955352e-06,
+      "loss": 0.8737,
+      "step": 111100
+    },
+    {
+      "epoch": 0.8042881857731258,
+      "grad_norm": 0.1683686077594757,
+      "learning_rate": 4.195719052892933e-06,
+      "loss": 0.8767,
+      "step": 111110
+    },
+    {
+      "epoch": 0.804360572433712,
+      "grad_norm": 0.15654446184635162,
+      "learning_rate": 4.195646666232347e-06,
+      "loss": 0.8707,
+      "step": 111120
+    },
+    {
+      "epoch": 0.8044329590942981,
+      "grad_norm": 0.15169669687747955,
+      "learning_rate": 4.1955742795717605e-06,
+      "loss": 0.883,
+      "step": 111130
+    },
+    {
+      "epoch": 0.8045053457548843,
+      "grad_norm": 0.15075841546058655,
+      "learning_rate": 4.195501892911175e-06,
+      "loss": 0.8914,
+      "step": 111140
+    },
+    {
+      "epoch": 0.8045777324154705,
+      "grad_norm": 0.16259559988975525,
+      "learning_rate": 4.1954295062505885e-06,
+      "loss": 0.8853,
+      "step": 111150
+    },
+    {
+      "epoch": 0.8046501190760567,
+      "grad_norm": 0.2971009910106659,
+      "learning_rate": 4.195357119590002e-06,
+      "loss": 0.8743,
+      "step": 111160
+    },
+    {
+      "epoch": 0.8047225057366428,
+      "grad_norm": 0.17599447071552277,
+      "learning_rate": 4.195284732929416e-06,
+      "loss": 0.8837,
+      "step": 111170
+    },
+    {
+      "epoch": 0.804794892397229,
+      "grad_norm": 0.1559291034936905,
+      "learning_rate": 4.19521234626883e-06,
+      "loss": 0.8789,
+      "step": 111180
+    },
+    {
+      "epoch": 0.8048672790578152,
+      "grad_norm": 0.14445562660694122,
+      "learning_rate": 4.195139959608244e-06,
+      "loss": 0.8738,
+      "step": 111190
+    },
+    {
+      "epoch": 0.8049396657184014,
+      "grad_norm": 0.1521482765674591,
+      "learning_rate": 4.1950675729476575e-06,
+      "loss": 0.8897,
+      "step": 111200
+    },
+    {
+      "epoch": 0.8050120523789877,
+      "grad_norm": 0.15571485459804535,
+      "learning_rate": 4.194995186287071e-06,
+      "loss": 0.8981,
+      "step": 111210
+    },
+    {
+      "epoch": 0.8050844390395738,
+      "grad_norm": 0.1559680998325348,
+      "learning_rate": 4.1949227996264855e-06,
+      "loss": 0.89,
+      "step": 111220
+    },
+    {
+      "epoch": 0.80515682570016,
+      "grad_norm": 0.1483583301305771,
+      "learning_rate": 4.194850412965899e-06,
+      "loss": 0.8838,
+      "step": 111230
+    },
+    {
+      "epoch": 0.8052292123607462,
+      "grad_norm": 0.15938538312911987,
+      "learning_rate": 4.194778026305313e-06,
+      "loss": 0.8759,
+      "step": 111240
+    },
+    {
+      "epoch": 0.8053015990213324,
+      "grad_norm": 0.15668749809265137,
+      "learning_rate": 4.194705639644726e-06,
+      "loss": 0.8753,
+      "step": 111250
+    },
+    {
+      "epoch": 0.8053739856819185,
+      "grad_norm": 0.1507354974746704,
+      "learning_rate": 4.194633252984141e-06,
+      "loss": 0.8911,
+      "step": 111260
+    },
+    {
+      "epoch": 0.8054463723425047,
+      "grad_norm": 0.15574651956558228,
+      "learning_rate": 4.1945608663235545e-06,
+      "loss": 0.879,
+      "step": 111270
+    },
+    {
+      "epoch": 0.8055187590030909,
+      "grad_norm": 0.2068503350019455,
+      "learning_rate": 4.194488479662968e-06,
+      "loss": 0.8897,
+      "step": 111280
+    },
+    {
+      "epoch": 0.8055911456636771,
+      "grad_norm": 0.1596960425376892,
+      "learning_rate": 4.194416093002382e-06,
+      "loss": 0.8782,
+      "step": 111290
+    },
+    {
+      "epoch": 0.8056635323242632,
+      "grad_norm": 0.1368640810251236,
+      "learning_rate": 4.194343706341796e-06,
+      "loss": 0.8767,
+      "step": 111300
+    },
+    {
+      "epoch": 0.8057359189848494,
+      "grad_norm": 0.18080635368824005,
+      "learning_rate": 4.19427131968121e-06,
+      "loss": 0.8798,
+      "step": 111310
+    },
+    {
+      "epoch": 0.8058083056454357,
+      "grad_norm": 0.1573970466852188,
+      "learning_rate": 4.194198933020623e-06,
+      "loss": 0.8708,
+      "step": 111320
+    },
+    {
+      "epoch": 0.8058806923060219,
+      "grad_norm": 0.1502685248851776,
+      "learning_rate": 4.194126546360037e-06,
+      "loss": 0.8842,
+      "step": 111330
+    },
+    {
+      "epoch": 0.8059530789666081,
+      "grad_norm": 0.15076951682567596,
+      "learning_rate": 4.1940541596994515e-06,
+      "loss": 0.8821,
+      "step": 111340
+    },
+    {
+      "epoch": 0.8060254656271942,
+      "grad_norm": 0.15020160377025604,
+      "learning_rate": 4.193981773038865e-06,
+      "loss": 0.886,
+      "step": 111350
+    },
+    {
+      "epoch": 0.8060978522877804,
+      "grad_norm": 0.356048047542572,
+      "learning_rate": 4.193909386378279e-06,
+      "loss": 0.8936,
+      "step": 111360
+    },
+    {
+      "epoch": 0.8061702389483666,
+      "grad_norm": 0.15093164145946503,
+      "learning_rate": 4.193836999717692e-06,
+      "loss": 0.8714,
+      "step": 111370
+    },
+    {
+      "epoch": 0.8062426256089528,
+      "grad_norm": 0.16395637392997742,
+      "learning_rate": 4.193764613057106e-06,
+      "loss": 0.876,
+      "step": 111380
+    },
+    {
+      "epoch": 0.8063150122695389,
+      "grad_norm": 0.15686947107315063,
+      "learning_rate": 4.19369222639652e-06,
+      "loss": 0.8807,
+      "step": 111390
+    },
+    {
+      "epoch": 0.8063873989301251,
+      "grad_norm": 0.15679305791854858,
+      "learning_rate": 4.193619839735934e-06,
+      "loss": 0.8945,
+      "step": 111400
+    },
+    {
+      "epoch": 0.8064597855907113,
+      "grad_norm": 0.15202337503433228,
+      "learning_rate": 4.193547453075348e-06,
+      "loss": 0.8781,
+      "step": 111410
+    },
+    {
+      "epoch": 0.8065321722512976,
+      "grad_norm": 0.1549692004919052,
+      "learning_rate": 4.193475066414761e-06,
+      "loss": 0.8694,
+      "step": 111420
+    },
+    {
+      "epoch": 0.8066045589118838,
+      "grad_norm": 0.15202471613883972,
+      "learning_rate": 4.193402679754176e-06,
+      "loss": 0.8718,
+      "step": 111430
+    },
+    {
+      "epoch": 0.8066769455724699,
+      "grad_norm": 0.15716615319252014,
+      "learning_rate": 4.193330293093589e-06,
+      "loss": 0.888,
+      "step": 111440
+    },
+    {
+      "epoch": 0.8067493322330561,
+      "grad_norm": 2.5106468200683594,
+      "learning_rate": 4.193257906433003e-06,
+      "loss": 0.87,
+      "step": 111450
+    },
+    {
+      "epoch": 0.8068217188936423,
+      "grad_norm": 0.15499623119831085,
+      "learning_rate": 4.1931855197724166e-06,
+      "loss": 0.8882,
+      "step": 111460
+    },
+    {
+      "epoch": 0.8068941055542285,
+      "grad_norm": 0.14410343766212463,
+      "learning_rate": 4.193113133111831e-06,
+      "loss": 0.8796,
+      "step": 111470
+    },
+    {
+      "epoch": 0.8069664922148146,
+      "grad_norm": 0.14761362969875336,
+      "learning_rate": 4.193040746451244e-06,
+      "loss": 0.8715,
+      "step": 111480
+    },
+    {
+      "epoch": 0.8070388788754008,
+      "grad_norm": 0.1514003425836563,
+      "learning_rate": 4.192968359790658e-06,
+      "loss": 0.8761,
+      "step": 111490
+    },
+    {
+      "epoch": 0.807111265535987,
+      "grad_norm": 0.15471474826335907,
+      "learning_rate": 4.192895973130072e-06,
+      "loss": 0.877,
+      "step": 111500
+    },
+    {
+      "epoch": 0.8071836521965732,
+      "grad_norm": 0.17473427951335907,
+      "learning_rate": 4.1928235864694855e-06,
+      "loss": 0.8767,
+      "step": 111510
+    },
+    {
+      "epoch": 0.8072560388571594,
+      "grad_norm": 0.14548401534557343,
+      "learning_rate": 4.192751199808899e-06,
+      "loss": 0.8961,
+      "step": 111520
+    },
+    {
+      "epoch": 0.8073284255177456,
+      "grad_norm": 0.15971076488494873,
+      "learning_rate": 4.192678813148313e-06,
+      "loss": 0.8899,
+      "step": 111530
+    },
+    {
+      "epoch": 0.8074008121783318,
+      "grad_norm": 0.16011503338813782,
+      "learning_rate": 4.192606426487727e-06,
+      "loss": 0.8803,
+      "step": 111540
+    },
+    {
+      "epoch": 0.807473198838918,
+      "grad_norm": 0.15592749416828156,
+      "learning_rate": 4.192534039827141e-06,
+      "loss": 0.878,
+      "step": 111550
+    },
+    {
+      "epoch": 0.8075455854995042,
+      "grad_norm": 0.14613130688667297,
+      "learning_rate": 4.1924616531665544e-06,
+      "loss": 0.8678,
+      "step": 111560
+    },
+    {
+      "epoch": 0.8076179721600903,
+      "grad_norm": 0.15929226577281952,
+      "learning_rate": 4.192389266505968e-06,
+      "loss": 0.8775,
+      "step": 111570
+    },
+    {
+      "epoch": 0.8076903588206765,
+      "grad_norm": 0.1534443497657776,
+      "learning_rate": 4.1923168798453825e-06,
+      "loss": 0.8827,
+      "step": 111580
+    },
+    {
+      "epoch": 0.8077627454812627,
+      "grad_norm": 0.145346000790596,
+      "learning_rate": 4.192244493184796e-06,
+      "loss": 0.8658,
+      "step": 111590
+    },
+    {
+      "epoch": 0.8078351321418489,
+      "grad_norm": 0.14903073012828827,
+      "learning_rate": 4.19217210652421e-06,
+      "loss": 0.8829,
+      "step": 111600
+    },
+    {
+      "epoch": 0.807907518802435,
+      "grad_norm": 0.14224445819854736,
+      "learning_rate": 4.192099719863623e-06,
+      "loss": 0.8843,
+      "step": 111610
+    },
+    {
+      "epoch": 0.8079799054630212,
+      "grad_norm": 0.3385683596134186,
+      "learning_rate": 4.192027333203038e-06,
+      "loss": 0.8851,
+      "step": 111620
+    },
+    {
+      "epoch": 0.8080522921236074,
+      "grad_norm": 0.14634902775287628,
+      "learning_rate": 4.1919549465424514e-06,
+      "loss": 0.8848,
+      "step": 111630
+    },
+    {
+      "epoch": 0.8081246787841937,
+      "grad_norm": 0.16326604783535004,
+      "learning_rate": 4.191882559881865e-06,
+      "loss": 0.8758,
+      "step": 111640
+    },
+    {
+      "epoch": 0.8081970654447799,
+      "grad_norm": 0.15901891887187958,
+      "learning_rate": 4.191810173221279e-06,
+      "loss": 0.8679,
+      "step": 111650
+    },
+    {
+      "epoch": 0.808269452105366,
+      "grad_norm": 0.15597860515117645,
+      "learning_rate": 4.191737786560693e-06,
+      "loss": 0.8848,
+      "step": 111660
+    },
+    {
+      "epoch": 0.8083418387659522,
+      "grad_norm": 0.15394975244998932,
+      "learning_rate": 4.191665399900107e-06,
+      "loss": 0.8767,
+      "step": 111670
+    },
+    {
+      "epoch": 0.8084142254265384,
+      "grad_norm": 0.1595843881368637,
+      "learning_rate": 4.19159301323952e-06,
+      "loss": 0.8717,
+      "step": 111680
+    },
+    {
+      "epoch": 0.8084866120871246,
+      "grad_norm": 0.15284503996372223,
+      "learning_rate": 4.191520626578934e-06,
+      "loss": 0.8774,
+      "step": 111690
+    },
+    {
+      "epoch": 0.8085589987477108,
+      "grad_norm": 0.16651558876037598,
+      "learning_rate": 4.1914482399183484e-06,
+      "loss": 0.8686,
+      "step": 111700
+    },
+    {
+      "epoch": 0.8086313854082969,
+      "grad_norm": 0.16672812402248383,
+      "learning_rate": 4.191375853257762e-06,
+      "loss": 0.8755,
+      "step": 111710
+    },
+    {
+      "epoch": 0.8087037720688831,
+      "grad_norm": 0.1647217869758606,
+      "learning_rate": 4.191303466597176e-06,
+      "loss": 0.8887,
+      "step": 111720
+    },
+    {
+      "epoch": 0.8087761587294693,
+      "grad_norm": 0.16164937615394592,
+      "learning_rate": 4.191231079936589e-06,
+      "loss": 0.8733,
+      "step": 111730
+    },
+    {
+      "epoch": 0.8088485453900556,
+      "grad_norm": 0.1532217264175415,
+      "learning_rate": 4.191158693276004e-06,
+      "loss": 0.877,
+      "step": 111740
+    },
+    {
+      "epoch": 0.8089209320506417,
+      "grad_norm": 0.16159088909626007,
+      "learning_rate": 4.191086306615417e-06,
+      "loss": 0.8977,
+      "step": 111750
+    },
+    {
+      "epoch": 0.8089933187112279,
+      "grad_norm": 0.1658451408147812,
+      "learning_rate": 4.191013919954831e-06,
+      "loss": 0.8906,
+      "step": 111760
+    },
+    {
+      "epoch": 0.8090657053718141,
+      "grad_norm": 0.43993157148361206,
+      "learning_rate": 4.190941533294245e-06,
+      "loss": 0.8782,
+      "step": 111770
+    },
+    {
+      "epoch": 0.8091380920324003,
+      "grad_norm": 0.1686534732580185,
+      "learning_rate": 4.190869146633659e-06,
+      "loss": 0.8833,
+      "step": 111780
+    },
+    {
+      "epoch": 0.8092104786929865,
+      "grad_norm": 0.16583459079265594,
+      "learning_rate": 4.190796759973073e-06,
+      "loss": 0.8893,
+      "step": 111790
+    },
+    {
+      "epoch": 0.8092828653535726,
+      "grad_norm": 0.21636223793029785,
+      "learning_rate": 4.190724373312486e-06,
+      "loss": 0.89,
+      "step": 111800
+    },
+    {
+      "epoch": 0.8093552520141588,
+      "grad_norm": 0.22253377735614777,
+      "learning_rate": 4.1906519866519e-06,
+      "loss": 0.8847,
+      "step": 111810
+    },
+    {
+      "epoch": 0.809427638674745,
+      "grad_norm": 0.15331260859966278,
+      "learning_rate": 4.190579599991314e-06,
+      "loss": 0.875,
+      "step": 111820
+    },
+    {
+      "epoch": 0.8095000253353312,
+      "grad_norm": 0.1543157696723938,
+      "learning_rate": 4.190507213330728e-06,
+      "loss": 0.8843,
+      "step": 111830
+    },
+    {
+      "epoch": 0.8095724119959173,
+      "grad_norm": 0.17324155569076538,
+      "learning_rate": 4.190434826670142e-06,
+      "loss": 0.8831,
+      "step": 111840
+    },
+    {
+      "epoch": 0.8096447986565036,
+      "grad_norm": 0.16567941009998322,
+      "learning_rate": 4.190362440009555e-06,
+      "loss": 0.8866,
+      "step": 111850
+    },
+    {
+      "epoch": 0.8097171853170898,
+      "grad_norm": 0.15467321872711182,
+      "learning_rate": 4.19029005334897e-06,
+      "loss": 0.8789,
+      "step": 111860
+    },
+    {
+      "epoch": 0.809789571977676,
+      "grad_norm": 0.1545843631029129,
+      "learning_rate": 4.190217666688383e-06,
+      "loss": 0.8849,
+      "step": 111870
+    },
+    {
+      "epoch": 0.8098619586382622,
+      "grad_norm": 0.15270915627479553,
+      "learning_rate": 4.190145280027797e-06,
+      "loss": 0.8768,
+      "step": 111880
+    },
+    {
+      "epoch": 0.8099343452988483,
+      "grad_norm": 0.14915820956230164,
+      "learning_rate": 4.1900728933672105e-06,
+      "loss": 0.878,
+      "step": 111890
+    },
+    {
+      "epoch": 0.8100067319594345,
+      "grad_norm": 0.15363983809947968,
+      "learning_rate": 4.190000506706625e-06,
+      "loss": 0.8841,
+      "step": 111900
+    },
+    {
+      "epoch": 0.8100791186200207,
+      "grad_norm": 0.1456158310174942,
+      "learning_rate": 4.189928120046039e-06,
+      "loss": 0.8803,
+      "step": 111910
+    },
+    {
+      "epoch": 0.8101515052806069,
+      "grad_norm": 0.16374152898788452,
+      "learning_rate": 4.189855733385452e-06,
+      "loss": 0.8793,
+      "step": 111920
+    },
+    {
+      "epoch": 0.810223891941193,
+      "grad_norm": 0.15842416882514954,
+      "learning_rate": 4.189783346724866e-06,
+      "loss": 0.8783,
+      "step": 111930
+    },
+    {
+      "epoch": 0.8102962786017792,
+      "grad_norm": 0.15646661818027496,
+      "learning_rate": 4.18971096006428e-06,
+      "loss": 0.8745,
+      "step": 111940
+    },
+    {
+      "epoch": 0.8103686652623655,
+      "grad_norm": 0.16457587480545044,
+      "learning_rate": 4.189638573403694e-06,
+      "loss": 0.8782,
+      "step": 111950
+    },
+    {
+      "epoch": 0.8104410519229517,
+      "grad_norm": 0.14522863924503326,
+      "learning_rate": 4.1895661867431075e-06,
+      "loss": 0.874,
+      "step": 111960
+    },
+    {
+      "epoch": 0.8105134385835379,
+      "grad_norm": 0.15910083055496216,
+      "learning_rate": 4.189493800082521e-06,
+      "loss": 0.8912,
+      "step": 111970
+    },
+    {
+      "epoch": 0.810585825244124,
+      "grad_norm": 0.164401113986969,
+      "learning_rate": 4.189421413421935e-06,
+      "loss": 0.8799,
+      "step": 111980
+    },
+    {
+      "epoch": 0.8106582119047102,
+      "grad_norm": 0.15682414174079895,
+      "learning_rate": 4.189349026761349e-06,
+      "loss": 0.8782,
+      "step": 111990
+    },
+    {
+      "epoch": 0.8107305985652964,
+      "grad_norm": 0.17602966725826263,
+      "learning_rate": 4.189276640100763e-06,
+      "loss": 0.8776,
+      "step": 112000
+    },
+    {
+      "epoch": 0.8108029852258826,
+      "grad_norm": 0.15074023604393005,
+      "learning_rate": 4.1892042534401765e-06,
+      "loss": 0.8892,
+      "step": 112010
+    },
+    {
+      "epoch": 0.8108753718864687,
+      "grad_norm": 0.1462225317955017,
+      "learning_rate": 4.18913186677959e-06,
+      "loss": 0.8541,
+      "step": 112020
+    },
+    {
+      "epoch": 0.8109477585470549,
+      "grad_norm": 0.15102677047252655,
+      "learning_rate": 4.189059480119004e-06,
+      "loss": 0.8776,
+      "step": 112030
+    },
+    {
+      "epoch": 0.8110201452076411,
+      "grad_norm": 0.40401482582092285,
+      "learning_rate": 4.188987093458417e-06,
+      "loss": 0.8831,
+      "step": 112040
+    },
+    {
+      "epoch": 0.8110925318682273,
+      "grad_norm": 0.3952445983886719,
+      "learning_rate": 4.188914706797832e-06,
+      "loss": 0.878,
+      "step": 112050
+    },
+    {
+      "epoch": 0.8111649185288136,
+      "grad_norm": 0.15976792573928833,
+      "learning_rate": 4.188842320137245e-06,
+      "loss": 0.8804,
+      "step": 112060
+    },
+    {
+      "epoch": 0.8112373051893997,
+      "grad_norm": 0.14807958900928497,
+      "learning_rate": 4.188769933476659e-06,
+      "loss": 0.8783,
+      "step": 112070
+    },
+    {
+      "epoch": 0.8113096918499859,
+      "grad_norm": 0.15941010415554047,
+      "learning_rate": 4.188697546816073e-06,
+      "loss": 0.8738,
+      "step": 112080
+    },
+    {
+      "epoch": 0.8113820785105721,
+      "grad_norm": 0.14998309314250946,
+      "learning_rate": 4.188625160155487e-06,
+      "loss": 0.8743,
+      "step": 112090
+    },
+    {
+      "epoch": 0.8114544651711583,
+      "grad_norm": 0.15989407896995544,
+      "learning_rate": 4.188552773494901e-06,
+      "loss": 0.8853,
+      "step": 112100
+    },
+    {
+      "epoch": 0.8115268518317444,
+      "grad_norm": 0.15547819435596466,
+      "learning_rate": 4.188480386834314e-06,
+      "loss": 0.8755,
+      "step": 112110
+    },
+    {
+      "epoch": 0.8115992384923306,
+      "grad_norm": 0.15554924309253693,
+      "learning_rate": 4.188408000173728e-06,
+      "loss": 0.8842,
+      "step": 112120
+    },
+    {
+      "epoch": 0.8116716251529168,
+      "grad_norm": 0.15123188495635986,
+      "learning_rate": 4.188335613513142e-06,
+      "loss": 0.8748,
+      "step": 112130
+    },
+    {
+      "epoch": 0.811744011813503,
+      "grad_norm": 0.16083499789237976,
+      "learning_rate": 4.188263226852556e-06,
+      "loss": 0.8765,
+      "step": 112140
+    },
+    {
+      "epoch": 0.8118163984740892,
+      "grad_norm": 0.14771822094917297,
+      "learning_rate": 4.18819084019197e-06,
+      "loss": 0.875,
+      "step": 112150
+    },
+    {
+      "epoch": 0.8118887851346753,
+      "grad_norm": 0.1459280550479889,
+      "learning_rate": 4.188118453531383e-06,
+      "loss": 0.8924,
+      "step": 112160
+    },
+    {
+      "epoch": 0.8119611717952616,
+      "grad_norm": 0.1660243421792984,
+      "learning_rate": 4.188046066870797e-06,
+      "loss": 0.8802,
+      "step": 112170
+    },
+    {
+      "epoch": 0.8120335584558478,
+      "grad_norm": 0.17147773504257202,
+      "learning_rate": 4.187973680210211e-06,
+      "loss": 0.883,
+      "step": 112180
+    },
+    {
+      "epoch": 0.812105945116434,
+      "grad_norm": 0.16224630177021027,
+      "learning_rate": 4.187901293549625e-06,
+      "loss": 0.8721,
+      "step": 112190
+    },
+    {
+      "epoch": 0.8121783317770201,
+      "grad_norm": 0.1489465981721878,
+      "learning_rate": 4.1878289068890386e-06,
+      "loss": 0.8739,
+      "step": 112200
+    },
+    {
+      "epoch": 0.8122507184376063,
+      "grad_norm": 0.14807015657424927,
+      "learning_rate": 4.187756520228452e-06,
+      "loss": 0.8842,
+      "step": 112210
+    },
+    {
+      "epoch": 0.8123231050981925,
+      "grad_norm": 0.15409156680107117,
+      "learning_rate": 4.187684133567867e-06,
+      "loss": 0.8738,
+      "step": 112220
+    },
+    {
+      "epoch": 0.8123954917587787,
+      "grad_norm": 0.16499954462051392,
+      "learning_rate": 4.18761174690728e-06,
+      "loss": 0.8848,
+      "step": 112230
+    },
+    {
+      "epoch": 0.8124678784193649,
+      "grad_norm": 0.15704099833965302,
+      "learning_rate": 4.187539360246694e-06,
+      "loss": 0.8767,
+      "step": 112240
+    },
+    {
+      "epoch": 0.812540265079951,
+      "grad_norm": 0.1506892442703247,
+      "learning_rate": 4.1874669735861075e-06,
+      "loss": 0.8802,
+      "step": 112250
+    },
+    {
+      "epoch": 0.8126126517405372,
+      "grad_norm": 0.1446923166513443,
+      "learning_rate": 4.187394586925522e-06,
+      "loss": 0.8753,
+      "step": 112260
+    },
+    {
+      "epoch": 0.8126850384011235,
+      "grad_norm": 0.16159547865390778,
+      "learning_rate": 4.187322200264936e-06,
+      "loss": 0.8751,
+      "step": 112270
+    },
+    {
+      "epoch": 0.8127574250617097,
+      "grad_norm": 0.17383888363838196,
+      "learning_rate": 4.187249813604349e-06,
+      "loss": 0.8793,
+      "step": 112280
+    },
+    {
+      "epoch": 0.8128298117222958,
+      "grad_norm": 0.1629524528980255,
+      "learning_rate": 4.187177426943763e-06,
+      "loss": 0.8804,
+      "step": 112290
+    },
+    {
+      "epoch": 0.812902198382882,
+      "grad_norm": 0.17006246745586395,
+      "learning_rate": 4.187105040283177e-06,
+      "loss": 0.8731,
+      "step": 112300
+    },
+    {
+      "epoch": 0.8129745850434682,
+      "grad_norm": 0.16095757484436035,
+      "learning_rate": 4.187032653622591e-06,
+      "loss": 0.8835,
+      "step": 112310
+    },
+    {
+      "epoch": 0.8130469717040544,
+      "grad_norm": 0.16775591671466827,
+      "learning_rate": 4.1869602669620045e-06,
+      "loss": 0.8859,
+      "step": 112320
+    },
+    {
+      "epoch": 0.8131193583646406,
+      "grad_norm": 0.16004826128482819,
+      "learning_rate": 4.186887880301418e-06,
+      "loss": 0.865,
+      "step": 112330
+    },
+    {
+      "epoch": 0.8131917450252267,
+      "grad_norm": 0.15483014285564423,
+      "learning_rate": 4.186815493640833e-06,
+      "loss": 0.8848,
+      "step": 112340
+    },
+    {
+      "epoch": 0.8132641316858129,
+      "grad_norm": 0.16190212965011597,
+      "learning_rate": 4.186743106980246e-06,
+      "loss": 0.8848,
+      "step": 112350
+    },
+    {
+      "epoch": 0.8133365183463991,
+      "grad_norm": 0.14579403400421143,
+      "learning_rate": 4.18667072031966e-06,
+      "loss": 0.8696,
+      "step": 112360
+    },
+    {
+      "epoch": 0.8134089050069853,
+      "grad_norm": 0.16750074923038483,
+      "learning_rate": 4.1865983336590734e-06,
+      "loss": 0.8826,
+      "step": 112370
+    },
+    {
+      "epoch": 0.8134812916675715,
+      "grad_norm": 0.1532454639673233,
+      "learning_rate": 4.186525946998488e-06,
+      "loss": 0.8722,
+      "step": 112380
+    },
+    {
+      "epoch": 0.8135536783281577,
+      "grad_norm": 0.15037626028060913,
+      "learning_rate": 4.1864535603379015e-06,
+      "loss": 0.8787,
+      "step": 112390
+    },
+    {
+      "epoch": 0.8136260649887439,
+      "grad_norm": 0.16376477479934692,
+      "learning_rate": 4.186381173677315e-06,
+      "loss": 0.8689,
+      "step": 112400
+    },
+    {
+      "epoch": 0.8136984516493301,
+      "grad_norm": 0.15625233948230743,
+      "learning_rate": 4.186308787016729e-06,
+      "loss": 0.8647,
+      "step": 112410
+    },
+    {
+      "epoch": 0.8137708383099163,
+      "grad_norm": 0.1553266942501068,
+      "learning_rate": 4.186236400356143e-06,
+      "loss": 0.8792,
+      "step": 112420
+    },
+    {
+      "epoch": 0.8138432249705024,
+      "grad_norm": 0.1567511111497879,
+      "learning_rate": 4.186164013695557e-06,
+      "loss": 0.8802,
+      "step": 112430
+    },
+    {
+      "epoch": 0.8139156116310886,
+      "grad_norm": 0.15767507255077362,
+      "learning_rate": 4.1860916270349704e-06,
+      "loss": 0.8783,
+      "step": 112440
+    },
+    {
+      "epoch": 0.8139879982916748,
+      "grad_norm": 0.1529453545808792,
+      "learning_rate": 4.186019240374384e-06,
+      "loss": 0.8856,
+      "step": 112450
+    },
+    {
+      "epoch": 0.814060384952261,
+      "grad_norm": 0.14311917126178741,
+      "learning_rate": 4.1859468537137985e-06,
+      "loss": 0.8824,
+      "step": 112460
+    },
+    {
+      "epoch": 0.8141327716128471,
+      "grad_norm": 0.1506299078464508,
+      "learning_rate": 4.185874467053212e-06,
+      "loss": 0.8952,
+      "step": 112470
+    },
+    {
+      "epoch": 0.8142051582734334,
+      "grad_norm": 0.14917585253715515,
+      "learning_rate": 4.185802080392626e-06,
+      "loss": 0.8784,
+      "step": 112480
+    },
+    {
+      "epoch": 0.8142775449340196,
+      "grad_norm": 0.16028118133544922,
+      "learning_rate": 4.185729693732039e-06,
+      "loss": 0.8708,
+      "step": 112490
+    },
+    {
+      "epoch": 0.8143499315946058,
+      "grad_norm": 0.18150749802589417,
+      "learning_rate": 4.185657307071454e-06,
+      "loss": 0.8822,
+      "step": 112500
+    },
+    {
+      "epoch": 0.814422318255192,
+      "grad_norm": 0.1553160399198532,
+      "learning_rate": 4.1855849204108674e-06,
+      "loss": 0.8889,
+      "step": 112510
+    },
+    {
+      "epoch": 0.8144947049157781,
+      "grad_norm": 0.15606872737407684,
+      "learning_rate": 4.185512533750281e-06,
+      "loss": 0.8843,
+      "step": 112520
+    },
+    {
+      "epoch": 0.8145670915763643,
+      "grad_norm": 0.1746978759765625,
+      "learning_rate": 4.185440147089695e-06,
+      "loss": 0.8768,
+      "step": 112530
+    },
+    {
+      "epoch": 0.8146394782369505,
+      "grad_norm": 0.20451655983924866,
+      "learning_rate": 4.185367760429108e-06,
+      "loss": 0.8694,
+      "step": 112540
+    },
+    {
+      "epoch": 0.8147118648975367,
+      "grad_norm": 0.15979830920696259,
+      "learning_rate": 4.185295373768522e-06,
+      "loss": 0.8778,
+      "step": 112550
+    },
+    {
+      "epoch": 0.8147842515581228,
+      "grad_norm": 0.15347421169281006,
+      "learning_rate": 4.1852229871079355e-06,
+      "loss": 0.8757,
+      "step": 112560
+    },
+    {
+      "epoch": 0.814856638218709,
+      "grad_norm": 0.1704648733139038,
+      "learning_rate": 4.18515060044735e-06,
+      "loss": 0.8759,
+      "step": 112570
+    },
+    {
+      "epoch": 0.8149290248792952,
+      "grad_norm": 0.14998449385166168,
+      "learning_rate": 4.185078213786764e-06,
+      "loss": 0.8819,
+      "step": 112580
+    },
+    {
+      "epoch": 0.8150014115398815,
+      "grad_norm": 0.1555691659450531,
+      "learning_rate": 4.185005827126177e-06,
+      "loss": 0.8858,
+      "step": 112590
+    },
+    {
+      "epoch": 0.8150737982004677,
+      "grad_norm": 0.15047410130500793,
+      "learning_rate": 4.184933440465591e-06,
+      "loss": 0.8789,
+      "step": 112600
+    },
+    {
+      "epoch": 0.8151461848610538,
+      "grad_norm": 0.14893095195293427,
+      "learning_rate": 4.184861053805005e-06,
+      "loss": 0.8807,
+      "step": 112610
+    },
+    {
+      "epoch": 0.81521857152164,
+      "grad_norm": 0.14945295453071594,
+      "learning_rate": 4.184788667144419e-06,
+      "loss": 0.8863,
+      "step": 112620
+    },
+    {
+      "epoch": 0.8152909581822262,
+      "grad_norm": 0.1564187854528427,
+      "learning_rate": 4.1847162804838325e-06,
+      "loss": 0.8864,
+      "step": 112630
+    },
+    {
+      "epoch": 0.8153633448428124,
+      "grad_norm": 0.15783673524856567,
+      "learning_rate": 4.184643893823246e-06,
+      "loss": 0.8753,
+      "step": 112640
+    },
+    {
+      "epoch": 0.8154357315033985,
+      "grad_norm": 0.16287674009799957,
+      "learning_rate": 4.184571507162661e-06,
+      "loss": 0.8739,
+      "step": 112650
+    },
+    {
+      "epoch": 0.8155081181639847,
+      "grad_norm": 0.1520308554172516,
+      "learning_rate": 4.184499120502074e-06,
+      "loss": 0.8993,
+      "step": 112660
+    },
+    {
+      "epoch": 0.8155805048245709,
+      "grad_norm": 0.14746682345867157,
+      "learning_rate": 4.184426733841488e-06,
+      "loss": 0.8857,
+      "step": 112670
+    },
+    {
+      "epoch": 0.8156528914851571,
+      "grad_norm": 0.15371263027191162,
+      "learning_rate": 4.1843543471809015e-06,
+      "loss": 0.869,
+      "step": 112680
+    },
+    {
+      "epoch": 0.8157252781457432,
+      "grad_norm": 0.1501394808292389,
+      "learning_rate": 4.184281960520316e-06,
+      "loss": 0.8704,
+      "step": 112690
+    },
+    {
+      "epoch": 0.8157976648063295,
+      "grad_norm": 0.18657702207565308,
+      "learning_rate": 4.1842095738597295e-06,
+      "loss": 0.8919,
+      "step": 112700
+    },
+    {
+      "epoch": 0.8158700514669157,
+      "grad_norm": 0.15628504753112793,
+      "learning_rate": 4.184137187199143e-06,
+      "loss": 0.8848,
+      "step": 112710
+    },
+    {
+      "epoch": 0.8159424381275019,
+      "grad_norm": 0.14718109369277954,
+      "learning_rate": 4.184064800538557e-06,
+      "loss": 0.8754,
+      "step": 112720
+    },
+    {
+      "epoch": 0.8160148247880881,
+      "grad_norm": 0.15502269566059113,
+      "learning_rate": 4.183992413877971e-06,
+      "loss": 0.8684,
+      "step": 112730
+    },
+    {
+      "epoch": 0.8160872114486742,
+      "grad_norm": 0.15216968953609467,
+      "learning_rate": 4.183920027217385e-06,
+      "loss": 0.8889,
+      "step": 112740
+    },
+    {
+      "epoch": 0.8161595981092604,
+      "grad_norm": 0.23924054205417633,
+      "learning_rate": 4.1838476405567985e-06,
+      "loss": 0.8764,
+      "step": 112750
+    },
+    {
+      "epoch": 0.8162319847698466,
+      "grad_norm": 0.1529170572757721,
+      "learning_rate": 4.183775253896212e-06,
+      "loss": 0.8778,
+      "step": 112760
+    },
+    {
+      "epoch": 0.8163043714304328,
+      "grad_norm": 0.15385037660598755,
+      "learning_rate": 4.1837028672356266e-06,
+      "loss": 0.8746,
+      "step": 112770
+    },
+    {
+      "epoch": 0.816376758091019,
+      "grad_norm": 0.15637922286987305,
+      "learning_rate": 4.18363048057504e-06,
+      "loss": 0.8715,
+      "step": 112780
+    },
+    {
+      "epoch": 0.8164491447516051,
+      "grad_norm": 0.1572360396385193,
+      "learning_rate": 4.183558093914454e-06,
+      "loss": 0.8725,
+      "step": 112790
+    },
+    {
+      "epoch": 0.8165215314121914,
+      "grad_norm": 0.14621710777282715,
+      "learning_rate": 4.183485707253867e-06,
+      "loss": 0.8683,
+      "step": 112800
+    },
+    {
+      "epoch": 0.8165939180727776,
+      "grad_norm": 0.15248066186904907,
+      "learning_rate": 4.183413320593281e-06,
+      "loss": 0.8836,
+      "step": 112810
+    },
+    {
+      "epoch": 0.8166663047333638,
+      "grad_norm": 0.16547314822673798,
+      "learning_rate": 4.1833409339326955e-06,
+      "loss": 0.8836,
+      "step": 112820
+    },
+    {
+      "epoch": 0.81673869139395,
+      "grad_norm": 0.15647459030151367,
+      "learning_rate": 4.183268547272109e-06,
+      "loss": 0.88,
+      "step": 112830
+    },
+    {
+      "epoch": 0.8168110780545361,
+      "grad_norm": 0.16002123057842255,
+      "learning_rate": 4.183196160611523e-06,
+      "loss": 0.887,
+      "step": 112840
+    },
+    {
+      "epoch": 0.8168834647151223,
+      "grad_norm": 0.14930887520313263,
+      "learning_rate": 4.183123773950936e-06,
+      "loss": 0.8787,
+      "step": 112850
+    },
+    {
+      "epoch": 0.8169558513757085,
+      "grad_norm": 0.15242883563041687,
+      "learning_rate": 4.183051387290351e-06,
+      "loss": 0.8712,
+      "step": 112860
+    },
+    {
+      "epoch": 0.8170282380362947,
+      "grad_norm": 0.1501031219959259,
+      "learning_rate": 4.182979000629764e-06,
+      "loss": 0.8827,
+      "step": 112870
+    },
+    {
+      "epoch": 0.8171006246968808,
+      "grad_norm": 0.14793910086154938,
+      "learning_rate": 4.182906613969178e-06,
+      "loss": 0.8773,
+      "step": 112880
+    },
+    {
+      "epoch": 0.817173011357467,
+      "grad_norm": 0.14584796130657196,
+      "learning_rate": 4.182834227308592e-06,
+      "loss": 0.8774,
+      "step": 112890
+    },
+    {
+      "epoch": 0.8172453980180532,
+      "grad_norm": 0.16166086494922638,
+      "learning_rate": 4.182761840648006e-06,
+      "loss": 0.8875,
+      "step": 112900
+    },
+    {
+      "epoch": 0.8173177846786395,
+      "grad_norm": 0.16169136762619019,
+      "learning_rate": 4.18268945398742e-06,
+      "loss": 0.876,
+      "step": 112910
+    },
+    {
+      "epoch": 0.8173901713392256,
+      "grad_norm": 0.15373431146144867,
+      "learning_rate": 4.182617067326833e-06,
+      "loss": 0.8835,
+      "step": 112920
+    },
+    {
+      "epoch": 0.8174625579998118,
+      "grad_norm": 0.15151308476924896,
+      "learning_rate": 4.182544680666247e-06,
+      "loss": 0.8961,
+      "step": 112930
+    },
+    {
+      "epoch": 0.817534944660398,
+      "grad_norm": 0.150752454996109,
+      "learning_rate": 4.182472294005661e-06,
+      "loss": 0.8785,
+      "step": 112940
+    },
+    {
+      "epoch": 0.8176073313209842,
+      "grad_norm": 0.15306270122528076,
+      "learning_rate": 4.182399907345075e-06,
+      "loss": 0.8856,
+      "step": 112950
+    },
+    {
+      "epoch": 0.8176797179815704,
+      "grad_norm": 0.15430527925491333,
+      "learning_rate": 4.182327520684489e-06,
+      "loss": 0.8826,
+      "step": 112960
+    },
+    {
+      "epoch": 0.8177521046421565,
+      "grad_norm": 0.1908227801322937,
+      "learning_rate": 4.182255134023902e-06,
+      "loss": 0.8828,
+      "step": 112970
+    },
+    {
+      "epoch": 0.8178244913027427,
+      "grad_norm": 0.15534339845180511,
+      "learning_rate": 4.182182747363317e-06,
+      "loss": 0.8882,
+      "step": 112980
+    },
+    {
+      "epoch": 0.8178968779633289,
+      "grad_norm": 0.1552184671163559,
+      "learning_rate": 4.18211036070273e-06,
+      "loss": 0.875,
+      "step": 112990
+    },
+    {
+      "epoch": 0.8179692646239151,
+      "grad_norm": 0.16356508433818817,
+      "learning_rate": 4.182037974042144e-06,
+      "loss": 0.8764,
+      "step": 113000
+    },
+    {
+      "epoch": 0.8180416512845012,
+      "grad_norm": 0.14462150633335114,
+      "learning_rate": 4.1819655873815576e-06,
+      "loss": 0.8844,
+      "step": 113010
+    },
+    {
+      "epoch": 0.8181140379450875,
+      "grad_norm": 0.15308737754821777,
+      "learning_rate": 4.181893200720972e-06,
+      "loss": 0.8758,
+      "step": 113020
+    },
+    {
+      "epoch": 0.8181864246056737,
+      "grad_norm": 0.1581387221813202,
+      "learning_rate": 4.181820814060386e-06,
+      "loss": 0.886,
+      "step": 113030
+    },
+    {
+      "epoch": 0.8182588112662599,
+      "grad_norm": 0.16034311056137085,
+      "learning_rate": 4.181748427399799e-06,
+      "loss": 0.8668,
+      "step": 113040
+    },
+    {
+      "epoch": 0.818331197926846,
+      "grad_norm": 0.15236437320709229,
+      "learning_rate": 4.181676040739213e-06,
+      "loss": 0.8861,
+      "step": 113050
+    },
+    {
+      "epoch": 0.8184035845874322,
+      "grad_norm": 0.20184694230556488,
+      "learning_rate": 4.181603654078627e-06,
+      "loss": 0.8921,
+      "step": 113060
+    },
+    {
+      "epoch": 0.8184759712480184,
+      "grad_norm": 0.15360352396965027,
+      "learning_rate": 4.18153126741804e-06,
+      "loss": 0.8988,
+      "step": 113070
+    },
+    {
+      "epoch": 0.8185483579086046,
+      "grad_norm": 0.1674332171678543,
+      "learning_rate": 4.181458880757454e-06,
+      "loss": 0.8742,
+      "step": 113080
+    },
+    {
+      "epoch": 0.8186207445691908,
+      "grad_norm": 0.16225215792655945,
+      "learning_rate": 4.181386494096868e-06,
+      "loss": 0.877,
+      "step": 113090
+    },
+    {
+      "epoch": 0.8186931312297769,
+      "grad_norm": 0.1606452614068985,
+      "learning_rate": 4.181314107436282e-06,
+      "loss": 0.8686,
+      "step": 113100
+    },
+    {
+      "epoch": 0.8187655178903631,
+      "grad_norm": 0.1443140208721161,
+      "learning_rate": 4.1812417207756954e-06,
+      "loss": 0.9018,
+      "step": 113110
+    },
+    {
+      "epoch": 0.8188379045509494,
+      "grad_norm": 0.16372881829738617,
+      "learning_rate": 4.181169334115109e-06,
+      "loss": 0.8783,
+      "step": 113120
+    },
+    {
+      "epoch": 0.8189102912115356,
+      "grad_norm": 0.1642576903104782,
+      "learning_rate": 4.1810969474545235e-06,
+      "loss": 0.8701,
+      "step": 113130
+    },
+    {
+      "epoch": 0.8189826778721218,
+      "grad_norm": 0.15896013379096985,
+      "learning_rate": 4.181024560793937e-06,
+      "loss": 0.8781,
+      "step": 113140
+    },
+    {
+      "epoch": 0.8190550645327079,
+      "grad_norm": 0.17695783078670502,
+      "learning_rate": 4.180952174133351e-06,
+      "loss": 0.8797,
+      "step": 113150
+    },
+    {
+      "epoch": 0.8191274511932941,
+      "grad_norm": 0.15274715423583984,
+      "learning_rate": 4.180879787472764e-06,
+      "loss": 0.8736,
+      "step": 113160
+    },
+    {
+      "epoch": 0.8191998378538803,
+      "grad_norm": 0.14954541623592377,
+      "learning_rate": 4.180807400812179e-06,
+      "loss": 0.8771,
+      "step": 113170
+    },
+    {
+      "epoch": 0.8192722245144665,
+      "grad_norm": 0.14884912967681885,
+      "learning_rate": 4.1807350141515924e-06,
+      "loss": 0.8703,
+      "step": 113180
+    },
+    {
+      "epoch": 0.8193446111750526,
+      "grad_norm": 0.18822531402111053,
+      "learning_rate": 4.180662627491006e-06,
+      "loss": 0.8762,
+      "step": 113190
+    },
+    {
+      "epoch": 0.8194169978356388,
+      "grad_norm": 0.16255638003349304,
+      "learning_rate": 4.18059024083042e-06,
+      "loss": 0.8758,
+      "step": 113200
+    },
+    {
+      "epoch": 0.819489384496225,
+      "grad_norm": 0.16536031663417816,
+      "learning_rate": 4.180517854169834e-06,
+      "loss": 0.8704,
+      "step": 113210
+    },
+    {
+      "epoch": 0.8195617711568112,
+      "grad_norm": 0.16202905774116516,
+      "learning_rate": 4.180445467509248e-06,
+      "loss": 0.8886,
+      "step": 113220
+    },
+    {
+      "epoch": 0.8196341578173975,
+      "grad_norm": 0.2304823398590088,
+      "learning_rate": 4.180373080848661e-06,
+      "loss": 0.8815,
+      "step": 113230
+    },
+    {
+      "epoch": 0.8197065444779836,
+      "grad_norm": 0.15564051270484924,
+      "learning_rate": 4.180300694188075e-06,
+      "loss": 0.8844,
+      "step": 113240
+    },
+    {
+      "epoch": 0.8197789311385698,
+      "grad_norm": 0.1432119905948639,
+      "learning_rate": 4.1802283075274894e-06,
+      "loss": 0.8713,
+      "step": 113250
+    },
+    {
+      "epoch": 0.819851317799156,
+      "grad_norm": 0.16350963711738586,
+      "learning_rate": 4.180155920866903e-06,
+      "loss": 0.8831,
+      "step": 113260
+    },
+    {
+      "epoch": 0.8199237044597422,
+      "grad_norm": 0.16729813814163208,
+      "learning_rate": 4.180083534206317e-06,
+      "loss": 0.879,
+      "step": 113270
+    },
+    {
+      "epoch": 0.8199960911203283,
+      "grad_norm": 0.1555587649345398,
+      "learning_rate": 4.18001114754573e-06,
+      "loss": 0.8793,
+      "step": 113280
+    },
+    {
+      "epoch": 0.8200684777809145,
+      "grad_norm": 0.1441824585199356,
+      "learning_rate": 4.179938760885145e-06,
+      "loss": 0.8703,
+      "step": 113290
+    },
+    {
+      "epoch": 0.8201408644415007,
+      "grad_norm": 0.1523018628358841,
+      "learning_rate": 4.179866374224558e-06,
+      "loss": 0.8796,
+      "step": 113300
+    },
+    {
+      "epoch": 0.8202132511020869,
+      "grad_norm": 0.15191130340099335,
+      "learning_rate": 4.179793987563972e-06,
+      "loss": 0.8681,
+      "step": 113310
+    },
+    {
+      "epoch": 0.820285637762673,
+      "grad_norm": 0.14424999058246613,
+      "learning_rate": 4.179721600903386e-06,
+      "loss": 0.8875,
+      "step": 113320
+    },
+    {
+      "epoch": 0.8203580244232593,
+      "grad_norm": 0.1524539440870285,
+      "learning_rate": 4.1796492142428e-06,
+      "loss": 0.8752,
+      "step": 113330
+    },
+    {
+      "epoch": 0.8204304110838455,
+      "grad_norm": 0.14882370829582214,
+      "learning_rate": 4.179576827582214e-06,
+      "loss": 0.8885,
+      "step": 113340
+    },
+    {
+      "epoch": 0.8205027977444317,
+      "grad_norm": 0.15619249641895294,
+      "learning_rate": 4.179504440921627e-06,
+      "loss": 0.8795,
+      "step": 113350
+    },
+    {
+      "epoch": 0.8205751844050179,
+      "grad_norm": 0.16282761096954346,
+      "learning_rate": 4.179432054261041e-06,
+      "loss": 0.8792,
+      "step": 113360
+    },
+    {
+      "epoch": 0.820647571065604,
+      "grad_norm": 0.15634344518184662,
+      "learning_rate": 4.179359667600455e-06,
+      "loss": 0.8811,
+      "step": 113370
+    },
+    {
+      "epoch": 0.8207199577261902,
+      "grad_norm": 0.15985414385795593,
+      "learning_rate": 4.179287280939869e-06,
+      "loss": 0.8803,
+      "step": 113380
+    },
+    {
+      "epoch": 0.8207923443867764,
+      "grad_norm": 0.1588483601808548,
+      "learning_rate": 4.179214894279283e-06,
+      "loss": 0.8858,
+      "step": 113390
+    },
+    {
+      "epoch": 0.8208647310473626,
+      "grad_norm": 0.1567647010087967,
+      "learning_rate": 4.179142507618696e-06,
+      "loss": 0.8786,
+      "step": 113400
+    },
+    {
+      "epoch": 0.8209371177079487,
+      "grad_norm": 0.18194611370563507,
+      "learning_rate": 4.17907012095811e-06,
+      "loss": 0.8917,
+      "step": 113410
+    },
+    {
+      "epoch": 0.8210095043685349,
+      "grad_norm": 0.1811460703611374,
+      "learning_rate": 4.178997734297524e-06,
+      "loss": 0.8777,
+      "step": 113420
+    },
+    {
+      "epoch": 0.8210818910291211,
+      "grad_norm": 0.15702205896377563,
+      "learning_rate": 4.178925347636938e-06,
+      "loss": 0.8673,
+      "step": 113430
+    },
+    {
+      "epoch": 0.8211542776897074,
+      "grad_norm": 0.1575247198343277,
+      "learning_rate": 4.1788529609763515e-06,
+      "loss": 0.8919,
+      "step": 113440
+    },
+    {
+      "epoch": 0.8212266643502936,
+      "grad_norm": 0.16069388389587402,
+      "learning_rate": 4.178780574315765e-06,
+      "loss": 0.8805,
+      "step": 113450
+    },
+    {
+      "epoch": 0.8212990510108797,
+      "grad_norm": 0.15721113979816437,
+      "learning_rate": 4.17870818765518e-06,
+      "loss": 0.8744,
+      "step": 113460
+    },
+    {
+      "epoch": 0.8213714376714659,
+      "grad_norm": 0.18812179565429688,
+      "learning_rate": 4.178635800994593e-06,
+      "loss": 0.884,
+      "step": 113470
+    },
+    {
+      "epoch": 0.8214438243320521,
+      "grad_norm": 0.1596520096063614,
+      "learning_rate": 4.178563414334007e-06,
+      "loss": 0.878,
+      "step": 113480
+    },
+    {
+      "epoch": 0.8215162109926383,
+      "grad_norm": 0.15708090364933014,
+      "learning_rate": 4.1784910276734205e-06,
+      "loss": 0.8749,
+      "step": 113490
+    },
+    {
+      "epoch": 0.8215885976532245,
+      "grad_norm": 0.1581474244594574,
+      "learning_rate": 4.178418641012835e-06,
+      "loss": 0.8897,
+      "step": 113500
+    },
+    {
+      "epoch": 0.8216609843138106,
+      "grad_norm": 0.15259206295013428,
+      "learning_rate": 4.1783462543522486e-06,
+      "loss": 0.8834,
+      "step": 113510
+    },
+    {
+      "epoch": 0.8217333709743968,
+      "grad_norm": 0.15696731209754944,
+      "learning_rate": 4.178273867691662e-06,
+      "loss": 0.8793,
+      "step": 113520
+    },
+    {
+      "epoch": 0.821805757634983,
+      "grad_norm": 0.15300579369068146,
+      "learning_rate": 4.178201481031076e-06,
+      "loss": 0.878,
+      "step": 113530
+    },
+    {
+      "epoch": 0.8218781442955692,
+      "grad_norm": 0.15157224237918854,
+      "learning_rate": 4.17812909437049e-06,
+      "loss": 0.8733,
+      "step": 113540
+    },
+    {
+      "epoch": 0.8219505309561554,
+      "grad_norm": 0.15784545242786407,
+      "learning_rate": 4.178056707709904e-06,
+      "loss": 0.879,
+      "step": 113550
+    },
+    {
+      "epoch": 0.8220229176167416,
+      "grad_norm": 0.1615738868713379,
+      "learning_rate": 4.1779843210493175e-06,
+      "loss": 0.8809,
+      "step": 113560
+    },
+    {
+      "epoch": 0.8220953042773278,
+      "grad_norm": 0.16567973792552948,
+      "learning_rate": 4.177911934388731e-06,
+      "loss": 0.8967,
+      "step": 113570
+    },
+    {
+      "epoch": 0.822167690937914,
+      "grad_norm": 0.15392477810382843,
+      "learning_rate": 4.1778395477281456e-06,
+      "loss": 0.8712,
+      "step": 113580
+    },
+    {
+      "epoch": 0.8222400775985002,
+      "grad_norm": 0.17302165925502777,
+      "learning_rate": 4.177767161067559e-06,
+      "loss": 0.8868,
+      "step": 113590
+    },
+    {
+      "epoch": 0.8223124642590863,
+      "grad_norm": 0.16119389235973358,
+      "learning_rate": 4.177694774406972e-06,
+      "loss": 0.8805,
+      "step": 113600
+    },
+    {
+      "epoch": 0.8223848509196725,
+      "grad_norm": 0.16274970769882202,
+      "learning_rate": 4.177622387746386e-06,
+      "loss": 0.8818,
+      "step": 113610
+    },
+    {
+      "epoch": 0.8224572375802587,
+      "grad_norm": 0.1551283448934555,
+      "learning_rate": 4.1775500010858e-06,
+      "loss": 0.8802,
+      "step": 113620
+    },
+    {
+      "epoch": 0.8225296242408449,
+      "grad_norm": 0.1489950567483902,
+      "learning_rate": 4.177477614425214e-06,
+      "loss": 0.8803,
+      "step": 113630
+    },
+    {
+      "epoch": 0.822602010901431,
+      "grad_norm": 0.15743660926818848,
+      "learning_rate": 4.177405227764627e-06,
+      "loss": 0.8856,
+      "step": 113640
+    },
+    {
+      "epoch": 0.8226743975620173,
+      "grad_norm": 0.16645437479019165,
+      "learning_rate": 4.177332841104042e-06,
+      "loss": 0.8809,
+      "step": 113650
+    },
+    {
+      "epoch": 0.8227467842226035,
+      "grad_norm": 0.14299704134464264,
+      "learning_rate": 4.177260454443455e-06,
+      "loss": 0.8884,
+      "step": 113660
+    },
+    {
+      "epoch": 0.8228191708831897,
+      "grad_norm": 0.1666107475757599,
+      "learning_rate": 4.177188067782869e-06,
+      "loss": 0.8872,
+      "step": 113670
+    },
+    {
+      "epoch": 0.8228915575437759,
+      "grad_norm": 0.1572570949792862,
+      "learning_rate": 4.1771156811222826e-06,
+      "loss": 0.8743,
+      "step": 113680
+    },
+    {
+      "epoch": 0.822963944204362,
+      "grad_norm": 0.15440626442432404,
+      "learning_rate": 4.177043294461697e-06,
+      "loss": 0.8864,
+      "step": 113690
+    },
+    {
+      "epoch": 0.8230363308649482,
+      "grad_norm": 0.14731813967227936,
+      "learning_rate": 4.176970907801111e-06,
+      "loss": 0.8812,
+      "step": 113700
+    },
+    {
+      "epoch": 0.8231087175255344,
+      "grad_norm": 0.15947289764881134,
+      "learning_rate": 4.176898521140524e-06,
+      "loss": 0.8917,
+      "step": 113710
+    },
+    {
+      "epoch": 0.8231811041861206,
+      "grad_norm": 0.15511277318000793,
+      "learning_rate": 4.176826134479938e-06,
+      "loss": 0.8687,
+      "step": 113720
+    },
+    {
+      "epoch": 0.8232534908467067,
+      "grad_norm": 0.14976176619529724,
+      "learning_rate": 4.176753747819352e-06,
+      "loss": 0.8789,
+      "step": 113730
+    },
+    {
+      "epoch": 0.8233258775072929,
+      "grad_norm": 0.1585080772638321,
+      "learning_rate": 4.176681361158766e-06,
+      "loss": 0.8886,
+      "step": 113740
+    },
+    {
+      "epoch": 0.8233982641678791,
+      "grad_norm": 0.1489742547273636,
+      "learning_rate": 4.1766089744981796e-06,
+      "loss": 0.8814,
+      "step": 113750
+    },
+    {
+      "epoch": 0.8234706508284654,
+      "grad_norm": 0.15232981741428375,
+      "learning_rate": 4.176536587837593e-06,
+      "loss": 0.8869,
+      "step": 113760
+    },
+    {
+      "epoch": 0.8235430374890516,
+      "grad_norm": 0.15081308782100677,
+      "learning_rate": 4.176464201177008e-06,
+      "loss": 0.8853,
+      "step": 113770
+    },
+    {
+      "epoch": 0.8236154241496377,
+      "grad_norm": 0.1606776863336563,
+      "learning_rate": 4.176391814516421e-06,
+      "loss": 0.8917,
+      "step": 113780
+    },
+    {
+      "epoch": 0.8236878108102239,
+      "grad_norm": 0.15332511067390442,
+      "learning_rate": 4.176319427855835e-06,
+      "loss": 0.8662,
+      "step": 113790
+    },
+    {
+      "epoch": 0.8237601974708101,
+      "grad_norm": 0.15498222410678864,
+      "learning_rate": 4.1762470411952485e-06,
+      "loss": 0.8738,
+      "step": 113800
+    },
+    {
+      "epoch": 0.8238325841313963,
+      "grad_norm": 0.174130380153656,
+      "learning_rate": 4.176174654534663e-06,
+      "loss": 0.8872,
+      "step": 113810
+    },
+    {
+      "epoch": 0.8239049707919824,
+      "grad_norm": 0.15380537509918213,
+      "learning_rate": 4.176102267874077e-06,
+      "loss": 0.8666,
+      "step": 113820
+    },
+    {
+      "epoch": 0.8239773574525686,
+      "grad_norm": 0.15310508012771606,
+      "learning_rate": 4.17602988121349e-06,
+      "loss": 0.8979,
+      "step": 113830
+    },
+    {
+      "epoch": 0.8240497441131548,
+      "grad_norm": 0.1675233244895935,
+      "learning_rate": 4.175957494552904e-06,
+      "loss": 0.8791,
+      "step": 113840
+    },
+    {
+      "epoch": 0.824122130773741,
+      "grad_norm": 0.14090007543563843,
+      "learning_rate": 4.175885107892318e-06,
+      "loss": 0.8752,
+      "step": 113850
+    },
+    {
+      "epoch": 0.8241945174343273,
+      "grad_norm": 0.15238140523433685,
+      "learning_rate": 4.175812721231732e-06,
+      "loss": 0.868,
+      "step": 113860
+    },
+    {
+      "epoch": 0.8242669040949134,
+      "grad_norm": 0.16735051572322845,
+      "learning_rate": 4.1757403345711455e-06,
+      "loss": 0.8843,
+      "step": 113870
+    },
+    {
+      "epoch": 0.8243392907554996,
+      "grad_norm": 0.18536004424095154,
+      "learning_rate": 4.175667947910559e-06,
+      "loss": 0.8736,
+      "step": 113880
+    },
+    {
+      "epoch": 0.8244116774160858,
+      "grad_norm": 0.15690474212169647,
+      "learning_rate": 4.175595561249974e-06,
+      "loss": 0.8772,
+      "step": 113890
+    },
+    {
+      "epoch": 0.824484064076672,
+      "grad_norm": 0.17251384258270264,
+      "learning_rate": 4.175523174589387e-06,
+      "loss": 0.878,
+      "step": 113900
+    },
+    {
+      "epoch": 0.8245564507372581,
+      "grad_norm": 0.15271571278572083,
+      "learning_rate": 4.175450787928801e-06,
+      "loss": 0.8737,
+      "step": 113910
+    },
+    {
+      "epoch": 0.8246288373978443,
+      "grad_norm": 0.1521008163690567,
+      "learning_rate": 4.1753784012682144e-06,
+      "loss": 0.89,
+      "step": 113920
+    },
+    {
+      "epoch": 0.8247012240584305,
+      "grad_norm": 0.15803080797195435,
+      "learning_rate": 4.175306014607629e-06,
+      "loss": 0.8795,
+      "step": 113930
+    },
+    {
+      "epoch": 0.8247736107190167,
+      "grad_norm": 0.14416874945163727,
+      "learning_rate": 4.1752336279470425e-06,
+      "loss": 0.8853,
+      "step": 113940
+    },
+    {
+      "epoch": 0.8248459973796028,
+      "grad_norm": 0.16256214678287506,
+      "learning_rate": 4.175161241286456e-06,
+      "loss": 0.8969,
+      "step": 113950
+    },
+    {
+      "epoch": 0.824918384040189,
+      "grad_norm": 0.15503908693790436,
+      "learning_rate": 4.17508885462587e-06,
+      "loss": 0.8903,
+      "step": 113960
+    },
+    {
+      "epoch": 0.8249907707007753,
+      "grad_norm": 0.23007570207118988,
+      "learning_rate": 4.175016467965284e-06,
+      "loss": 0.8737,
+      "step": 113970
+    },
+    {
+      "epoch": 0.8250631573613615,
+      "grad_norm": 0.15308153629302979,
+      "learning_rate": 4.174944081304698e-06,
+      "loss": 0.8723,
+      "step": 113980
+    },
+    {
+      "epoch": 0.8251355440219477,
+      "grad_norm": 0.14756432175636292,
+      "learning_rate": 4.1748716946441114e-06,
+      "loss": 0.8749,
+      "step": 113990
+    },
+    {
+      "epoch": 0.8252079306825338,
+      "grad_norm": 0.17486760020256042,
+      "learning_rate": 4.174799307983525e-06,
+      "loss": 0.8632,
+      "step": 114000
+    },
+    {
+      "epoch": 0.82528031734312,
+      "grad_norm": 0.1607871949672699,
+      "learning_rate": 4.1747269213229395e-06,
+      "loss": 0.8716,
+      "step": 114010
+    },
+    {
+      "epoch": 0.8253527040037062,
+      "grad_norm": 0.1484338939189911,
+      "learning_rate": 4.174654534662353e-06,
+      "loss": 0.8746,
+      "step": 114020
+    },
+    {
+      "epoch": 0.8254250906642924,
+      "grad_norm": 0.21677958965301514,
+      "learning_rate": 4.174582148001767e-06,
+      "loss": 0.8857,
+      "step": 114030
+    },
+    {
+      "epoch": 0.8254974773248785,
+      "grad_norm": 0.16989293694496155,
+      "learning_rate": 4.17450976134118e-06,
+      "loss": 0.8781,
+      "step": 114040
+    },
+    {
+      "epoch": 0.8255698639854647,
+      "grad_norm": 0.1774820238351822,
+      "learning_rate": 4.174437374680594e-06,
+      "loss": 0.8694,
+      "step": 114050
+    },
+    {
+      "epoch": 0.8256422506460509,
+      "grad_norm": 0.15557865798473358,
+      "learning_rate": 4.1743649880200085e-06,
+      "loss": 0.8745,
+      "step": 114060
+    },
+    {
+      "epoch": 0.8257146373066371,
+      "grad_norm": 0.15898308157920837,
+      "learning_rate": 4.174292601359422e-06,
+      "loss": 0.8887,
+      "step": 114070
+    },
+    {
+      "epoch": 0.8257870239672234,
+      "grad_norm": 0.17024999856948853,
+      "learning_rate": 4.174220214698836e-06,
+      "loss": 0.8772,
+      "step": 114080
+    },
+    {
+      "epoch": 0.8258594106278095,
+      "grad_norm": 0.1510993391275406,
+      "learning_rate": 4.174147828038249e-06,
+      "loss": 0.8678,
+      "step": 114090
+    },
+    {
+      "epoch": 0.8259317972883957,
+      "grad_norm": 0.15980158746242523,
+      "learning_rate": 4.174075441377664e-06,
+      "loss": 0.8756,
+      "step": 114100
+    },
+    {
+      "epoch": 0.8260041839489819,
+      "grad_norm": 0.15219931304454803,
+      "learning_rate": 4.174003054717077e-06,
+      "loss": 0.8959,
+      "step": 114110
+    },
+    {
+      "epoch": 0.8260765706095681,
+      "grad_norm": 0.14390040934085846,
+      "learning_rate": 4.173930668056491e-06,
+      "loss": 0.8794,
+      "step": 114120
+    },
+    {
+      "epoch": 0.8261489572701542,
+      "grad_norm": 0.14578518271446228,
+      "learning_rate": 4.173858281395905e-06,
+      "loss": 0.8719,
+      "step": 114130
+    },
+    {
+      "epoch": 0.8262213439307404,
+      "grad_norm": 0.1575828641653061,
+      "learning_rate": 4.173785894735318e-06,
+      "loss": 0.8812,
+      "step": 114140
+    },
+    {
+      "epoch": 0.8262937305913266,
+      "grad_norm": 0.21511991322040558,
+      "learning_rate": 4.173713508074732e-06,
+      "loss": 0.8674,
+      "step": 114150
+    },
+    {
+      "epoch": 0.8263661172519128,
+      "grad_norm": 0.16267074644565582,
+      "learning_rate": 4.173641121414146e-06,
+      "loss": 0.8874,
+      "step": 114160
+    },
+    {
+      "epoch": 0.826438503912499,
+      "grad_norm": 0.14810219407081604,
+      "learning_rate": 4.17356873475356e-06,
+      "loss": 0.8732,
+      "step": 114170
+    },
+    {
+      "epoch": 0.8265108905730852,
+      "grad_norm": 0.14335951209068298,
+      "learning_rate": 4.1734963480929735e-06,
+      "loss": 0.8814,
+      "step": 114180
+    },
+    {
+      "epoch": 0.8265832772336714,
+      "grad_norm": 0.14947238564491272,
+      "learning_rate": 4.173423961432387e-06,
+      "loss": 0.8886,
+      "step": 114190
+    },
+    {
+      "epoch": 0.8266556638942576,
+      "grad_norm": 0.15292997658252716,
+      "learning_rate": 4.173351574771801e-06,
+      "loss": 0.8808,
+      "step": 114200
+    },
+    {
+      "epoch": 0.8267280505548438,
+      "grad_norm": 0.16210046410560608,
+      "learning_rate": 4.173279188111215e-06,
+      "loss": 0.8807,
+      "step": 114210
+    },
+    {
+      "epoch": 0.82680043721543,
+      "grad_norm": 0.1552553027868271,
+      "learning_rate": 4.173206801450629e-06,
+      "loss": 0.9006,
+      "step": 114220
+    },
+    {
+      "epoch": 0.8268728238760161,
+      "grad_norm": 0.15879151225090027,
+      "learning_rate": 4.1731344147900425e-06,
+      "loss": 0.8699,
+      "step": 114230
+    },
+    {
+      "epoch": 0.8269452105366023,
+      "grad_norm": 0.1656922549009323,
+      "learning_rate": 4.173062028129456e-06,
+      "loss": 0.8742,
+      "step": 114240
+    },
+    {
+      "epoch": 0.8270175971971885,
+      "grad_norm": 0.18779303133487701,
+      "learning_rate": 4.1729896414688705e-06,
+      "loss": 0.876,
+      "step": 114250
+    },
+    {
+      "epoch": 0.8270899838577747,
+      "grad_norm": 0.15145491063594818,
+      "learning_rate": 4.172917254808284e-06,
+      "loss": 0.8818,
+      "step": 114260
+    },
+    {
+      "epoch": 0.8271623705183608,
+      "grad_norm": 0.1533774733543396,
+      "learning_rate": 4.172844868147698e-06,
+      "loss": 0.8762,
+      "step": 114270
+    },
+    {
+      "epoch": 0.827234757178947,
+      "grad_norm": 0.15074725449085236,
+      "learning_rate": 4.172772481487111e-06,
+      "loss": 0.8766,
+      "step": 114280
+    },
+    {
+      "epoch": 0.8273071438395333,
+      "grad_norm": 0.15934348106384277,
+      "learning_rate": 4.172700094826526e-06,
+      "loss": 0.8789,
+      "step": 114290
+    },
+    {
+      "epoch": 0.8273795305001195,
+      "grad_norm": 0.1573365181684494,
+      "learning_rate": 4.1726277081659395e-06,
+      "loss": 0.8783,
+      "step": 114300
+    },
+    {
+      "epoch": 0.8274519171607057,
+      "grad_norm": 0.15947797894477844,
+      "learning_rate": 4.172555321505353e-06,
+      "loss": 0.8645,
+      "step": 114310
+    },
+    {
+      "epoch": 0.8275243038212918,
+      "grad_norm": 0.1520025134086609,
+      "learning_rate": 4.172482934844767e-06,
+      "loss": 0.8753,
+      "step": 114320
+    },
+    {
+      "epoch": 0.827596690481878,
+      "grad_norm": 0.15688809752464294,
+      "learning_rate": 4.172410548184181e-06,
+      "loss": 0.8712,
+      "step": 114330
+    },
+    {
+      "epoch": 0.8276690771424642,
+      "grad_norm": 0.23521029949188232,
+      "learning_rate": 4.172338161523595e-06,
+      "loss": 0.8903,
+      "step": 114340
+    },
+    {
+      "epoch": 0.8277414638030504,
+      "grad_norm": 0.15197953581809998,
+      "learning_rate": 4.172265774863008e-06,
+      "loss": 0.8847,
+      "step": 114350
+    },
+    {
+      "epoch": 0.8278138504636365,
+      "grad_norm": 0.151088684797287,
+      "learning_rate": 4.172193388202422e-06,
+      "loss": 0.8656,
+      "step": 114360
+    },
+    {
+      "epoch": 0.8278862371242227,
+      "grad_norm": 0.14569281041622162,
+      "learning_rate": 4.1721210015418365e-06,
+      "loss": 0.8814,
+      "step": 114370
+    },
+    {
+      "epoch": 0.8279586237848089,
+      "grad_norm": 0.1544409692287445,
+      "learning_rate": 4.17204861488125e-06,
+      "loss": 0.8876,
+      "step": 114380
+    },
+    {
+      "epoch": 0.8280310104453952,
+      "grad_norm": 0.17135676741600037,
+      "learning_rate": 4.171976228220664e-06,
+      "loss": 0.8809,
+      "step": 114390
+    },
+    {
+      "epoch": 0.8281033971059814,
+      "grad_norm": 0.15658897161483765,
+      "learning_rate": 4.171903841560077e-06,
+      "loss": 0.8838,
+      "step": 114400
+    },
+    {
+      "epoch": 0.8281757837665675,
+      "grad_norm": 0.14812763035297394,
+      "learning_rate": 4.171831454899492e-06,
+      "loss": 0.8784,
+      "step": 114410
+    },
+    {
+      "epoch": 0.8282481704271537,
+      "grad_norm": 0.15863005816936493,
+      "learning_rate": 4.171759068238905e-06,
+      "loss": 0.89,
+      "step": 114420
+    },
+    {
+      "epoch": 0.8283205570877399,
+      "grad_norm": 0.15445846319198608,
+      "learning_rate": 4.171686681578319e-06,
+      "loss": 0.8813,
+      "step": 114430
+    },
+    {
+      "epoch": 0.8283929437483261,
+      "grad_norm": 0.15649083256721497,
+      "learning_rate": 4.171614294917733e-06,
+      "loss": 0.8706,
+      "step": 114440
+    },
+    {
+      "epoch": 0.8284653304089122,
+      "grad_norm": 0.15367910265922546,
+      "learning_rate": 4.171541908257147e-06,
+      "loss": 0.8636,
+      "step": 114450
+    },
+    {
+      "epoch": 0.8285377170694984,
+      "grad_norm": 0.16421107947826385,
+      "learning_rate": 4.171469521596561e-06,
+      "loss": 0.8749,
+      "step": 114460
+    },
+    {
+      "epoch": 0.8286101037300846,
+      "grad_norm": 0.1638004332780838,
+      "learning_rate": 4.171397134935974e-06,
+      "loss": 0.8675,
+      "step": 114470
+    },
+    {
+      "epoch": 0.8286824903906708,
+      "grad_norm": 0.15033425390720367,
+      "learning_rate": 4.171324748275388e-06,
+      "loss": 0.8858,
+      "step": 114480
+    },
+    {
+      "epoch": 0.8287548770512569,
+      "grad_norm": 0.16598929464817047,
+      "learning_rate": 4.171252361614802e-06,
+      "loss": 0.8906,
+      "step": 114490
+    },
+    {
+      "epoch": 0.8288272637118432,
+      "grad_norm": 0.17145119607448578,
+      "learning_rate": 4.171179974954216e-06,
+      "loss": 0.8716,
+      "step": 114500
+    },
+    {
+      "epoch": 0.8288996503724294,
+      "grad_norm": 0.20420026779174805,
+      "learning_rate": 4.17110758829363e-06,
+      "loss": 0.8753,
+      "step": 114510
+    },
+    {
+      "epoch": 0.8289720370330156,
+      "grad_norm": 0.15310752391815186,
+      "learning_rate": 4.171035201633043e-06,
+      "loss": 0.8767,
+      "step": 114520
+    },
+    {
+      "epoch": 0.8290444236936018,
+      "grad_norm": 0.15935362875461578,
+      "learning_rate": 4.170962814972458e-06,
+      "loss": 0.8764,
+      "step": 114530
+    },
+    {
+      "epoch": 0.8291168103541879,
+      "grad_norm": 0.16182856261730194,
+      "learning_rate": 4.170890428311871e-06,
+      "loss": 0.8739,
+      "step": 114540
+    },
+    {
+      "epoch": 0.8291891970147741,
+      "grad_norm": 0.15881557762622833,
+      "learning_rate": 4.170818041651285e-06,
+      "loss": 0.8855,
+      "step": 114550
+    },
+    {
+      "epoch": 0.8292615836753603,
+      "grad_norm": 0.1685194969177246,
+      "learning_rate": 4.170745654990699e-06,
+      "loss": 0.8758,
+      "step": 114560
+    },
+    {
+      "epoch": 0.8293339703359465,
+      "grad_norm": 0.14720526337623596,
+      "learning_rate": 4.170673268330113e-06,
+      "loss": 0.873,
+      "step": 114570
+    },
+    {
+      "epoch": 0.8294063569965326,
+      "grad_norm": 0.16784626245498657,
+      "learning_rate": 4.170600881669527e-06,
+      "loss": 0.8853,
+      "step": 114580
+    },
+    {
+      "epoch": 0.8294787436571188,
+      "grad_norm": 0.1470102220773697,
+      "learning_rate": 4.17052849500894e-06,
+      "loss": 0.8832,
+      "step": 114590
+    },
+    {
+      "epoch": 0.829551130317705,
+      "grad_norm": 0.16039815545082092,
+      "learning_rate": 4.170456108348354e-06,
+      "loss": 0.8632,
+      "step": 114600
+    },
+    {
+      "epoch": 0.8296235169782913,
+      "grad_norm": 0.1540583372116089,
+      "learning_rate": 4.170383721687768e-06,
+      "loss": 0.8813,
+      "step": 114610
+    },
+    {
+      "epoch": 0.8296959036388775,
+      "grad_norm": 0.15052036941051483,
+      "learning_rate": 4.170311335027182e-06,
+      "loss": 0.8849,
+      "step": 114620
+    },
+    {
+      "epoch": 0.8297682902994636,
+      "grad_norm": 0.1492500752210617,
+      "learning_rate": 4.170238948366596e-06,
+      "loss": 0.885,
+      "step": 114630
+    },
+    {
+      "epoch": 0.8298406769600498,
+      "grad_norm": 0.15592175722122192,
+      "learning_rate": 4.170166561706009e-06,
+      "loss": 0.8654,
+      "step": 114640
+    },
+    {
+      "epoch": 0.829913063620636,
+      "grad_norm": 0.15170545876026154,
+      "learning_rate": 4.170094175045423e-06,
+      "loss": 0.8728,
+      "step": 114650
+    },
+    {
+      "epoch": 0.8299854502812222,
+      "grad_norm": 0.45990192890167236,
+      "learning_rate": 4.1700217883848364e-06,
+      "loss": 0.8802,
+      "step": 114660
+    },
+    {
+      "epoch": 0.8300578369418083,
+      "grad_norm": 0.1639576107263565,
+      "learning_rate": 4.16994940172425e-06,
+      "loss": 0.8959,
+      "step": 114670
+    },
+    {
+      "epoch": 0.8301302236023945,
+      "grad_norm": 0.15342135727405548,
+      "learning_rate": 4.1698770150636645e-06,
+      "loss": 0.8907,
+      "step": 114680
+    },
+    {
+      "epoch": 0.8302026102629807,
+      "grad_norm": 0.15938065946102142,
+      "learning_rate": 4.169804628403078e-06,
+      "loss": 0.873,
+      "step": 114690
+    },
+    {
+      "epoch": 0.8302749969235669,
+      "grad_norm": 0.15566521883010864,
+      "learning_rate": 4.169732241742492e-06,
+      "loss": 0.8848,
+      "step": 114700
+    },
+    {
+      "epoch": 0.8303473835841532,
+      "grad_norm": 0.1593281626701355,
+      "learning_rate": 4.169659855081905e-06,
+      "loss": 0.8775,
+      "step": 114710
+    },
+    {
+      "epoch": 0.8304197702447393,
+      "grad_norm": 0.15656022727489471,
+      "learning_rate": 4.16958746842132e-06,
+      "loss": 0.8735,
+      "step": 114720
+    },
+    {
+      "epoch": 0.8304921569053255,
+      "grad_norm": 0.15564171969890594,
+      "learning_rate": 4.1695150817607334e-06,
+      "loss": 0.8776,
+      "step": 114730
+    },
+    {
+      "epoch": 0.8305645435659117,
+      "grad_norm": 0.15010464191436768,
+      "learning_rate": 4.169442695100147e-06,
+      "loss": 0.8914,
+      "step": 114740
+    },
+    {
+      "epoch": 0.8306369302264979,
+      "grad_norm": 0.15904924273490906,
+      "learning_rate": 4.169370308439561e-06,
+      "loss": 0.8804,
+      "step": 114750
+    },
+    {
+      "epoch": 0.830709316887084,
+      "grad_norm": 0.161235973238945,
+      "learning_rate": 4.169297921778975e-06,
+      "loss": 0.8811,
+      "step": 114760
+    },
+    {
+      "epoch": 0.8307817035476702,
+      "grad_norm": 0.15518715977668762,
+      "learning_rate": 4.169225535118389e-06,
+      "loss": 0.8712,
+      "step": 114770
+    },
+    {
+      "epoch": 0.8308540902082564,
+      "grad_norm": 0.15426094830036163,
+      "learning_rate": 4.169153148457802e-06,
+      "loss": 0.881,
+      "step": 114780
+    },
+    {
+      "epoch": 0.8309264768688426,
+      "grad_norm": 0.15717415511608124,
+      "learning_rate": 4.169080761797216e-06,
+      "loss": 0.8656,
+      "step": 114790
+    },
+    {
+      "epoch": 0.8309988635294288,
+      "grad_norm": 0.19086652994155884,
+      "learning_rate": 4.1690083751366305e-06,
+      "loss": 0.8722,
+      "step": 114800
+    },
+    {
+      "epoch": 0.8310712501900149,
+      "grad_norm": 0.18602590262889862,
+      "learning_rate": 4.168935988476044e-06,
+      "loss": 0.8677,
+      "step": 114810
+    },
+    {
+      "epoch": 0.8311436368506012,
+      "grad_norm": 0.16731368005275726,
+      "learning_rate": 4.168863601815458e-06,
+      "loss": 0.8814,
+      "step": 114820
+    },
+    {
+      "epoch": 0.8312160235111874,
+      "grad_norm": 0.162908136844635,
+      "learning_rate": 4.168791215154871e-06,
+      "loss": 0.8795,
+      "step": 114830
+    },
+    {
+      "epoch": 0.8312884101717736,
+      "grad_norm": 0.15628595650196075,
+      "learning_rate": 4.168718828494285e-06,
+      "loss": 0.8818,
+      "step": 114840
+    },
+    {
+      "epoch": 0.8313607968323597,
+      "grad_norm": 0.15655048191547394,
+      "learning_rate": 4.168646441833699e-06,
+      "loss": 0.8835,
+      "step": 114850
+    },
+    {
+      "epoch": 0.8314331834929459,
+      "grad_norm": 0.15166090428829193,
+      "learning_rate": 4.168574055173113e-06,
+      "loss": 0.888,
+      "step": 114860
+    },
+    {
+      "epoch": 0.8315055701535321,
+      "grad_norm": 0.19675646722316742,
+      "learning_rate": 4.168501668512527e-06,
+      "loss": 0.8821,
+      "step": 114870
+    },
+    {
+      "epoch": 0.8315779568141183,
+      "grad_norm": 0.15683965384960175,
+      "learning_rate": 4.16842928185194e-06,
+      "loss": 0.8894,
+      "step": 114880
+    },
+    {
+      "epoch": 0.8316503434747045,
+      "grad_norm": 0.15531839430332184,
+      "learning_rate": 4.168356895191355e-06,
+      "loss": 0.8811,
+      "step": 114890
+    },
+    {
+      "epoch": 0.8317227301352906,
+      "grad_norm": 0.1496214121580124,
+      "learning_rate": 4.168284508530768e-06,
+      "loss": 0.873,
+      "step": 114900
+    },
+    {
+      "epoch": 0.8317951167958768,
+      "grad_norm": 0.16152338683605194,
+      "learning_rate": 4.168212121870182e-06,
+      "loss": 0.8725,
+      "step": 114910
+    },
+    {
+      "epoch": 0.8318675034564631,
+      "grad_norm": 0.17109687626361847,
+      "learning_rate": 4.1681397352095955e-06,
+      "loss": 0.8678,
+      "step": 114920
+    },
+    {
+      "epoch": 0.8319398901170493,
+      "grad_norm": 0.15160439908504486,
+      "learning_rate": 4.16806734854901e-06,
+      "loss": 0.8881,
+      "step": 114930
+    },
+    {
+      "epoch": 0.8320122767776355,
+      "grad_norm": 0.15334677696228027,
+      "learning_rate": 4.167994961888424e-06,
+      "loss": 0.882,
+      "step": 114940
+    },
+    {
+      "epoch": 0.8320846634382216,
+      "grad_norm": 0.16867488622665405,
+      "learning_rate": 4.167922575227837e-06,
+      "loss": 0.8815,
+      "step": 114950
+    },
+    {
+      "epoch": 0.8321570500988078,
+      "grad_norm": 0.16307266056537628,
+      "learning_rate": 4.167850188567251e-06,
+      "loss": 0.8737,
+      "step": 114960
+    },
+    {
+      "epoch": 0.832229436759394,
+      "grad_norm": 0.17095071077346802,
+      "learning_rate": 4.167777801906665e-06,
+      "loss": 0.868,
+      "step": 114970
+    },
+    {
+      "epoch": 0.8323018234199802,
+      "grad_norm": 0.1598438024520874,
+      "learning_rate": 4.167705415246079e-06,
+      "loss": 0.8826,
+      "step": 114980
+    },
+    {
+      "epoch": 0.8323742100805663,
+      "grad_norm": 0.15811651945114136,
+      "learning_rate": 4.1676330285854925e-06,
+      "loss": 0.8842,
+      "step": 114990
+    },
+    {
+      "epoch": 0.8324465967411525,
+      "grad_norm": 0.14370112121105194,
+      "learning_rate": 4.167560641924906e-06,
+      "loss": 0.8631,
+      "step": 115000
+    },
+    {
+      "epoch": 0.8325189834017387,
+      "grad_norm": 0.15269970893859863,
+      "learning_rate": 4.167488255264321e-06,
+      "loss": 0.8808,
+      "step": 115010
+    },
+    {
+      "epoch": 0.8325913700623249,
+      "grad_norm": 0.15361453592777252,
+      "learning_rate": 4.167415868603734e-06,
+      "loss": 0.87,
+      "step": 115020
+    },
+    {
+      "epoch": 0.8326637567229112,
+      "grad_norm": 0.1495579481124878,
+      "learning_rate": 4.167343481943148e-06,
+      "loss": 0.8709,
+      "step": 115030
+    },
+    {
+      "epoch": 0.8327361433834973,
+      "grad_norm": 0.14331714808940887,
+      "learning_rate": 4.1672710952825615e-06,
+      "loss": 0.8837,
+      "step": 115040
+    },
+    {
+      "epoch": 0.8328085300440835,
+      "grad_norm": 0.2528897225856781,
+      "learning_rate": 4.167198708621976e-06,
+      "loss": 0.8815,
+      "step": 115050
+    },
+    {
+      "epoch": 0.8328809167046697,
+      "grad_norm": 0.15450528264045715,
+      "learning_rate": 4.1671263219613896e-06,
+      "loss": 0.8856,
+      "step": 115060
+    },
+    {
+      "epoch": 0.8329533033652559,
+      "grad_norm": 0.1559913456439972,
+      "learning_rate": 4.167053935300803e-06,
+      "loss": 0.8908,
+      "step": 115070
+    },
+    {
+      "epoch": 0.833025690025842,
+      "grad_norm": 0.1556912660598755,
+      "learning_rate": 4.166981548640217e-06,
+      "loss": 0.8851,
+      "step": 115080
+    },
+    {
+      "epoch": 0.8330980766864282,
+      "grad_norm": 0.15410785377025604,
+      "learning_rate": 4.166909161979631e-06,
+      "loss": 0.8828,
+      "step": 115090
+    },
+    {
+      "epoch": 0.8331704633470144,
+      "grad_norm": 0.20597724616527557,
+      "learning_rate": 4.166836775319045e-06,
+      "loss": 0.8725,
+      "step": 115100
+    },
+    {
+      "epoch": 0.8332428500076006,
+      "grad_norm": 0.1543237566947937,
+      "learning_rate": 4.1667643886584585e-06,
+      "loss": 0.8656,
+      "step": 115110
+    },
+    {
+      "epoch": 0.8333152366681867,
+      "grad_norm": 0.14461509883403778,
+      "learning_rate": 4.166692001997872e-06,
+      "loss": 0.8763,
+      "step": 115120
+    },
+    {
+      "epoch": 0.8333876233287729,
+      "grad_norm": 0.14833268523216248,
+      "learning_rate": 4.1666196153372866e-06,
+      "loss": 0.8725,
+      "step": 115130
+    },
+    {
+      "epoch": 0.8334600099893592,
+      "grad_norm": 0.16073955595493317,
+      "learning_rate": 4.1665472286767e-06,
+      "loss": 0.8727,
+      "step": 115140
+    },
+    {
+      "epoch": 0.8335323966499454,
+      "grad_norm": 0.13932476937770844,
+      "learning_rate": 4.166474842016114e-06,
+      "loss": 0.8792,
+      "step": 115150
+    },
+    {
+      "epoch": 0.8336047833105316,
+      "grad_norm": 0.1595870852470398,
+      "learning_rate": 4.166402455355527e-06,
+      "loss": 0.8769,
+      "step": 115160
+    },
+    {
+      "epoch": 0.8336771699711177,
+      "grad_norm": 0.14972224831581116,
+      "learning_rate": 4.166330068694942e-06,
+      "loss": 0.8873,
+      "step": 115170
+    },
+    {
+      "epoch": 0.8337495566317039,
+      "grad_norm": 0.14646044373512268,
+      "learning_rate": 4.1662576820343555e-06,
+      "loss": 0.8845,
+      "step": 115180
+    },
+    {
+      "epoch": 0.8338219432922901,
+      "grad_norm": 0.1509612500667572,
+      "learning_rate": 4.166185295373768e-06,
+      "loss": 0.8673,
+      "step": 115190
+    },
+    {
+      "epoch": 0.8338943299528763,
+      "grad_norm": 0.1453905701637268,
+      "learning_rate": 4.166112908713183e-06,
+      "loss": 0.8754,
+      "step": 115200
+    },
+    {
+      "epoch": 0.8339667166134624,
+      "grad_norm": 0.14863267540931702,
+      "learning_rate": 4.166040522052596e-06,
+      "loss": 0.869,
+      "step": 115210
+    },
+    {
+      "epoch": 0.8340391032740486,
+      "grad_norm": 0.1638236790895462,
+      "learning_rate": 4.16596813539201e-06,
+      "loss": 0.8878,
+      "step": 115220
+    },
+    {
+      "epoch": 0.8341114899346348,
+      "grad_norm": 0.17066480219364166,
+      "learning_rate": 4.1658957487314236e-06,
+      "loss": 0.8719,
+      "step": 115230
+    },
+    {
+      "epoch": 0.8341838765952211,
+      "grad_norm": 0.16603226959705353,
+      "learning_rate": 4.165823362070838e-06,
+      "loss": 0.8733,
+      "step": 115240
+    },
+    {
+      "epoch": 0.8342562632558073,
+      "grad_norm": 0.141592338681221,
+      "learning_rate": 4.165750975410252e-06,
+      "loss": 0.8729,
+      "step": 115250
+    },
+    {
+      "epoch": 0.8343286499163934,
+      "grad_norm": 0.15747976303100586,
+      "learning_rate": 4.165678588749665e-06,
+      "loss": 0.8633,
+      "step": 115260
+    },
+    {
+      "epoch": 0.8344010365769796,
+      "grad_norm": 0.1882728487253189,
+      "learning_rate": 4.165606202089079e-06,
+      "loss": 0.8736,
+      "step": 115270
+    },
+    {
+      "epoch": 0.8344734232375658,
+      "grad_norm": 0.16340763866901398,
+      "learning_rate": 4.165533815428493e-06,
+      "loss": 0.8885,
+      "step": 115280
+    },
+    {
+      "epoch": 0.834545809898152,
+      "grad_norm": 0.1481885462999344,
+      "learning_rate": 4.165461428767907e-06,
+      "loss": 0.8885,
+      "step": 115290
+    },
+    {
+      "epoch": 0.8346181965587381,
+      "grad_norm": 0.15606743097305298,
+      "learning_rate": 4.165389042107321e-06,
+      "loss": 0.8767,
+      "step": 115300
+    },
+    {
+      "epoch": 0.8346905832193243,
+      "grad_norm": 0.14937707781791687,
+      "learning_rate": 4.165316655446734e-06,
+      "loss": 0.875,
+      "step": 115310
+    },
+    {
+      "epoch": 0.8347629698799105,
+      "grad_norm": 0.16030281782150269,
+      "learning_rate": 4.165244268786149e-06,
+      "loss": 0.8741,
+      "step": 115320
+    },
+    {
+      "epoch": 0.8348353565404967,
+      "grad_norm": 0.15192563831806183,
+      "learning_rate": 4.165171882125562e-06,
+      "loss": 0.8686,
+      "step": 115330
+    },
+    {
+      "epoch": 0.8349077432010829,
+      "grad_norm": 0.15150035917758942,
+      "learning_rate": 4.165099495464976e-06,
+      "loss": 0.8875,
+      "step": 115340
+    },
+    {
+      "epoch": 0.8349801298616691,
+      "grad_norm": 0.1519050896167755,
+      "learning_rate": 4.1650271088043895e-06,
+      "loss": 0.8844,
+      "step": 115350
+    },
+    {
+      "epoch": 0.8350525165222553,
+      "grad_norm": 0.15744005143642426,
+      "learning_rate": 4.164954722143804e-06,
+      "loss": 0.8619,
+      "step": 115360
+    },
+    {
+      "epoch": 0.8351249031828415,
+      "grad_norm": 0.16231480240821838,
+      "learning_rate": 4.164882335483218e-06,
+      "loss": 0.8722,
+      "step": 115370
+    },
+    {
+      "epoch": 0.8351972898434277,
+      "grad_norm": 0.15285032987594604,
+      "learning_rate": 4.164809948822631e-06,
+      "loss": 0.8796,
+      "step": 115380
+    },
+    {
+      "epoch": 0.8352696765040138,
+      "grad_norm": 0.16190151870250702,
+      "learning_rate": 4.164737562162045e-06,
+      "loss": 0.8804,
+      "step": 115390
+    },
+    {
+      "epoch": 0.8353420631646,
+      "grad_norm": 0.16439662873744965,
+      "learning_rate": 4.164665175501459e-06,
+      "loss": 0.8944,
+      "step": 115400
+    },
+    {
+      "epoch": 0.8354144498251862,
+      "grad_norm": 0.14439240097999573,
+      "learning_rate": 4.164592788840873e-06,
+      "loss": 0.8802,
+      "step": 115410
+    },
+    {
+      "epoch": 0.8354868364857724,
+      "grad_norm": 0.14769050478935242,
+      "learning_rate": 4.1645204021802865e-06,
+      "loss": 0.8807,
+      "step": 115420
+    },
+    {
+      "epoch": 0.8355592231463586,
+      "grad_norm": 0.15976214408874512,
+      "learning_rate": 4.1644480155197e-06,
+      "loss": 0.8796,
+      "step": 115430
+    },
+    {
+      "epoch": 0.8356316098069447,
+      "grad_norm": 0.15138176083564758,
+      "learning_rate": 4.164375628859115e-06,
+      "loss": 0.8711,
+      "step": 115440
+    },
+    {
+      "epoch": 0.835703996467531,
+      "grad_norm": 0.15986670553684235,
+      "learning_rate": 4.164303242198528e-06,
+      "loss": 0.889,
+      "step": 115450
+    },
+    {
+      "epoch": 0.8357763831281172,
+      "grad_norm": 0.1555429995059967,
+      "learning_rate": 4.164230855537942e-06,
+      "loss": 0.865,
+      "step": 115460
+    },
+    {
+      "epoch": 0.8358487697887034,
+      "grad_norm": 0.15238524973392487,
+      "learning_rate": 4.1641584688773554e-06,
+      "loss": 0.8672,
+      "step": 115470
+    },
+    {
+      "epoch": 0.8359211564492895,
+      "grad_norm": 0.16362081468105316,
+      "learning_rate": 4.164086082216769e-06,
+      "loss": 0.8658,
+      "step": 115480
+    },
+    {
+      "epoch": 0.8359935431098757,
+      "grad_norm": 0.16536161303520203,
+      "learning_rate": 4.1640136955561835e-06,
+      "loss": 0.8911,
+      "step": 115490
+    },
+    {
+      "epoch": 0.8360659297704619,
+      "grad_norm": 0.15160474181175232,
+      "learning_rate": 4.163941308895597e-06,
+      "loss": 0.8678,
+      "step": 115500
+    },
+    {
+      "epoch": 0.8361383164310481,
+      "grad_norm": 0.16615518927574158,
+      "learning_rate": 4.163868922235011e-06,
+      "loss": 0.8866,
+      "step": 115510
+    },
+    {
+      "epoch": 0.8362107030916343,
+      "grad_norm": 0.15683193504810333,
+      "learning_rate": 4.163796535574424e-06,
+      "loss": 0.866,
+      "step": 115520
+    },
+    {
+      "epoch": 0.8362830897522204,
+      "grad_norm": 0.149534672498703,
+      "learning_rate": 4.163724148913839e-06,
+      "loss": 0.8748,
+      "step": 115530
+    },
+    {
+      "epoch": 0.8363554764128066,
+      "grad_norm": 0.15053287148475647,
+      "learning_rate": 4.1636517622532525e-06,
+      "loss": 0.8713,
+      "step": 115540
+    },
+    {
+      "epoch": 0.8364278630733928,
+      "grad_norm": 0.16546009480953217,
+      "learning_rate": 4.163579375592666e-06,
+      "loss": 0.8776,
+      "step": 115550
+    },
+    {
+      "epoch": 0.8365002497339791,
+      "grad_norm": 0.1460997462272644,
+      "learning_rate": 4.16350698893208e-06,
+      "loss": 0.8818,
+      "step": 115560
+    },
+    {
+      "epoch": 0.8365726363945652,
+      "grad_norm": 0.15319761633872986,
+      "learning_rate": 4.163434602271494e-06,
+      "loss": 0.8797,
+      "step": 115570
+    },
+    {
+      "epoch": 0.8366450230551514,
+      "grad_norm": 0.234059140086174,
+      "learning_rate": 4.163362215610908e-06,
+      "loss": 0.8659,
+      "step": 115580
+    },
+    {
+      "epoch": 0.8367174097157376,
+      "grad_norm": 0.147258460521698,
+      "learning_rate": 4.163289828950321e-06,
+      "loss": 0.8754,
+      "step": 115590
+    },
+    {
+      "epoch": 0.8367897963763238,
+      "grad_norm": 0.15908755362033844,
+      "learning_rate": 4.163217442289735e-06,
+      "loss": 0.872,
+      "step": 115600
+    },
+    {
+      "epoch": 0.83686218303691,
+      "grad_norm": 0.15638591349124908,
+      "learning_rate": 4.1631450556291495e-06,
+      "loss": 0.8971,
+      "step": 115610
+    },
+    {
+      "epoch": 0.8369345696974961,
+      "grad_norm": 0.1701616644859314,
+      "learning_rate": 4.163072668968563e-06,
+      "loss": 0.8726,
+      "step": 115620
+    },
+    {
+      "epoch": 0.8370069563580823,
+      "grad_norm": 0.15191833674907684,
+      "learning_rate": 4.163000282307977e-06,
+      "loss": 0.8767,
+      "step": 115630
+    },
+    {
+      "epoch": 0.8370793430186685,
+      "grad_norm": 0.15473678708076477,
+      "learning_rate": 4.16292789564739e-06,
+      "loss": 0.8773,
+      "step": 115640
+    },
+    {
+      "epoch": 0.8371517296792547,
+      "grad_norm": 0.15959081053733826,
+      "learning_rate": 4.162855508986805e-06,
+      "loss": 0.8745,
+      "step": 115650
+    },
+    {
+      "epoch": 0.8372241163398408,
+      "grad_norm": 0.1590321958065033,
+      "learning_rate": 4.162783122326218e-06,
+      "loss": 0.8818,
+      "step": 115660
+    },
+    {
+      "epoch": 0.8372965030004271,
+      "grad_norm": 0.16567932069301605,
+      "learning_rate": 4.162710735665632e-06,
+      "loss": 0.8798,
+      "step": 115670
+    },
+    {
+      "epoch": 0.8373688896610133,
+      "grad_norm": 0.15759676694869995,
+      "learning_rate": 4.162638349005046e-06,
+      "loss": 0.8689,
+      "step": 115680
+    },
+    {
+      "epoch": 0.8374412763215995,
+      "grad_norm": 0.15198834240436554,
+      "learning_rate": 4.16256596234446e-06,
+      "loss": 0.8867,
+      "step": 115690
+    },
+    {
+      "epoch": 0.8375136629821857,
+      "grad_norm": 0.1595429629087448,
+      "learning_rate": 4.162493575683874e-06,
+      "loss": 0.8792,
+      "step": 115700
+    },
+    {
+      "epoch": 0.8375860496427718,
+      "grad_norm": 0.15620209276676178,
+      "learning_rate": 4.162421189023287e-06,
+      "loss": 0.8751,
+      "step": 115710
+    },
+    {
+      "epoch": 0.837658436303358,
+      "grad_norm": 0.18191352486610413,
+      "learning_rate": 4.162348802362701e-06,
+      "loss": 0.8682,
+      "step": 115720
+    },
+    {
+      "epoch": 0.8377308229639442,
+      "grad_norm": 0.1610066294670105,
+      "learning_rate": 4.1622764157021145e-06,
+      "loss": 0.8813,
+      "step": 115730
+    },
+    {
+      "epoch": 0.8378032096245304,
+      "grad_norm": 0.16806887090206146,
+      "learning_rate": 4.162204029041528e-06,
+      "loss": 0.8786,
+      "step": 115740
+    },
+    {
+      "epoch": 0.8378755962851165,
+      "grad_norm": 0.1434309184551239,
+      "learning_rate": 4.162131642380942e-06,
+      "loss": 0.8844,
+      "step": 115750
+    },
+    {
+      "epoch": 0.8379479829457027,
+      "grad_norm": 0.17224647104740143,
+      "learning_rate": 4.162059255720356e-06,
+      "loss": 0.8848,
+      "step": 115760
+    },
+    {
+      "epoch": 0.838020369606289,
+      "grad_norm": 0.1476389765739441,
+      "learning_rate": 4.16198686905977e-06,
+      "loss": 0.8813,
+      "step": 115770
+    },
+    {
+      "epoch": 0.8380927562668752,
+      "grad_norm": 0.1633952558040619,
+      "learning_rate": 4.1619144823991835e-06,
+      "loss": 0.8842,
+      "step": 115780
+    },
+    {
+      "epoch": 0.8381651429274614,
+      "grad_norm": 0.16024617850780487,
+      "learning_rate": 4.161842095738597e-06,
+      "loss": 0.8829,
+      "step": 115790
+    },
+    {
+      "epoch": 0.8382375295880475,
+      "grad_norm": 0.15520702302455902,
+      "learning_rate": 4.1617697090780116e-06,
+      "loss": 0.8744,
+      "step": 115800
+    },
+    {
+      "epoch": 0.8383099162486337,
+      "grad_norm": 0.161240816116333,
+      "learning_rate": 4.161697322417425e-06,
+      "loss": 0.8912,
+      "step": 115810
+    },
+    {
+      "epoch": 0.8383823029092199,
+      "grad_norm": 0.14296600222587585,
+      "learning_rate": 4.161624935756839e-06,
+      "loss": 0.8727,
+      "step": 115820
+    },
+    {
+      "epoch": 0.8384546895698061,
+      "grad_norm": 0.16847427189350128,
+      "learning_rate": 4.161552549096252e-06,
+      "loss": 0.8812,
+      "step": 115830
+    },
+    {
+      "epoch": 0.8385270762303922,
+      "grad_norm": 0.14801815152168274,
+      "learning_rate": 4.161480162435667e-06,
+      "loss": 0.8822,
+      "step": 115840
+    },
+    {
+      "epoch": 0.8385994628909784,
+      "grad_norm": 0.15415239334106445,
+      "learning_rate": 4.1614077757750805e-06,
+      "loss": 0.8754,
+      "step": 115850
+    },
+    {
+      "epoch": 0.8386718495515646,
+      "grad_norm": 0.1483432799577713,
+      "learning_rate": 4.161335389114494e-06,
+      "loss": 0.8831,
+      "step": 115860
+    },
+    {
+      "epoch": 0.8387442362121508,
+      "grad_norm": 0.1458449810743332,
+      "learning_rate": 4.161263002453908e-06,
+      "loss": 0.8847,
+      "step": 115870
+    },
+    {
+      "epoch": 0.8388166228727371,
+      "grad_norm": 0.15387925505638123,
+      "learning_rate": 4.161190615793322e-06,
+      "loss": 0.8719,
+      "step": 115880
+    },
+    {
+      "epoch": 0.8388890095333232,
+      "grad_norm": 0.14547225832939148,
+      "learning_rate": 4.161118229132736e-06,
+      "loss": 0.8866,
+      "step": 115890
+    },
+    {
+      "epoch": 0.8389613961939094,
+      "grad_norm": 0.15607166290283203,
+      "learning_rate": 4.161045842472149e-06,
+      "loss": 0.8811,
+      "step": 115900
+    },
+    {
+      "epoch": 0.8390337828544956,
+      "grad_norm": 0.14948482811450958,
+      "learning_rate": 4.160973455811563e-06,
+      "loss": 0.8838,
+      "step": 115910
+    },
+    {
+      "epoch": 0.8391061695150818,
+      "grad_norm": 0.16144073009490967,
+      "learning_rate": 4.1609010691509775e-06,
+      "loss": 0.8786,
+      "step": 115920
+    },
+    {
+      "epoch": 0.839178556175668,
+      "grad_norm": 0.15592895448207855,
+      "learning_rate": 4.160828682490391e-06,
+      "loss": 0.8812,
+      "step": 115930
+    },
+    {
+      "epoch": 0.8392509428362541,
+      "grad_norm": 0.18400424718856812,
+      "learning_rate": 4.160756295829805e-06,
+      "loss": 0.87,
+      "step": 115940
+    },
+    {
+      "epoch": 0.8393233294968403,
+      "grad_norm": 0.1547369509935379,
+      "learning_rate": 4.160683909169218e-06,
+      "loss": 0.8918,
+      "step": 115950
+    },
+    {
+      "epoch": 0.8393957161574265,
+      "grad_norm": 0.16612297296524048,
+      "learning_rate": 4.160611522508633e-06,
+      "loss": 0.8735,
+      "step": 115960
+    },
+    {
+      "epoch": 0.8394681028180127,
+      "grad_norm": 0.14709250628948212,
+      "learning_rate": 4.160539135848046e-06,
+      "loss": 0.8752,
+      "step": 115970
+    },
+    {
+      "epoch": 0.8395404894785989,
+      "grad_norm": 0.15569816529750824,
+      "learning_rate": 4.16046674918746e-06,
+      "loss": 0.8725,
+      "step": 115980
+    },
+    {
+      "epoch": 0.8396128761391851,
+      "grad_norm": 0.16147053241729736,
+      "learning_rate": 4.160394362526874e-06,
+      "loss": 0.8772,
+      "step": 115990
+    },
+    {
+      "epoch": 0.8396852627997713,
+      "grad_norm": 0.15154972672462463,
+      "learning_rate": 4.160321975866288e-06,
+      "loss": 0.881,
+      "step": 116000
+    },
+    {
+      "epoch": 0.8397576494603575,
+      "grad_norm": 0.15269294381141663,
+      "learning_rate": 4.160249589205702e-06,
+      "loss": 0.883,
+      "step": 116010
+    },
+    {
+      "epoch": 0.8398300361209436,
+      "grad_norm": 0.17886610329151154,
+      "learning_rate": 4.160177202545115e-06,
+      "loss": 0.8735,
+      "step": 116020
+    },
+    {
+      "epoch": 0.8399024227815298,
+      "grad_norm": 0.15143749117851257,
+      "learning_rate": 4.160104815884529e-06,
+      "loss": 0.8752,
+      "step": 116030
+    },
+    {
+      "epoch": 0.839974809442116,
+      "grad_norm": 0.15021578967571259,
+      "learning_rate": 4.1600324292239434e-06,
+      "loss": 0.8707,
+      "step": 116040
+    },
+    {
+      "epoch": 0.8400471961027022,
+      "grad_norm": 0.15974397957324982,
+      "learning_rate": 4.159960042563357e-06,
+      "loss": 0.8743,
+      "step": 116050
+    },
+    {
+      "epoch": 0.8401195827632884,
+      "grad_norm": 0.15328460931777954,
+      "learning_rate": 4.159887655902771e-06,
+      "loss": 0.878,
+      "step": 116060
+    },
+    {
+      "epoch": 0.8401919694238745,
+      "grad_norm": 0.16074135899543762,
+      "learning_rate": 4.159815269242184e-06,
+      "loss": 0.8775,
+      "step": 116070
+    },
+    {
+      "epoch": 0.8402643560844607,
+      "grad_norm": 0.14766909182071686,
+      "learning_rate": 4.159742882581598e-06,
+      "loss": 0.8734,
+      "step": 116080
+    },
+    {
+      "epoch": 0.840336742745047,
+      "grad_norm": 0.1620090901851654,
+      "learning_rate": 4.159670495921012e-06,
+      "loss": 0.884,
+      "step": 116090
+    },
+    {
+      "epoch": 0.8404091294056332,
+      "grad_norm": 0.1521306186914444,
+      "learning_rate": 4.159598109260426e-06,
+      "loss": 0.8904,
+      "step": 116100
+    },
+    {
+      "epoch": 0.8404815160662193,
+      "grad_norm": 0.15059085190296173,
+      "learning_rate": 4.15952572259984e-06,
+      "loss": 0.8783,
+      "step": 116110
+    },
+    {
+      "epoch": 0.8405539027268055,
+      "grad_norm": 0.159867063164711,
+      "learning_rate": 4.159453335939253e-06,
+      "loss": 0.8777,
+      "step": 116120
+    },
+    {
+      "epoch": 0.8406262893873917,
+      "grad_norm": 0.15567173063755035,
+      "learning_rate": 4.159380949278668e-06,
+      "loss": 0.8751,
+      "step": 116130
+    },
+    {
+      "epoch": 0.8406986760479779,
+      "grad_norm": 0.16036923229694366,
+      "learning_rate": 4.159308562618081e-06,
+      "loss": 0.8946,
+      "step": 116140
+    },
+    {
+      "epoch": 0.840771062708564,
+      "grad_norm": 0.151793971657753,
+      "learning_rate": 4.159236175957495e-06,
+      "loss": 0.8785,
+      "step": 116150
+    },
+    {
+      "epoch": 0.8408434493691502,
+      "grad_norm": 0.17096100747585297,
+      "learning_rate": 4.1591637892969085e-06,
+      "loss": 0.8795,
+      "step": 116160
+    },
+    {
+      "epoch": 0.8409158360297364,
+      "grad_norm": 0.1503993719816208,
+      "learning_rate": 4.159091402636323e-06,
+      "loss": 0.8877,
+      "step": 116170
+    },
+    {
+      "epoch": 0.8409882226903226,
+      "grad_norm": 0.1676565408706665,
+      "learning_rate": 4.159019015975737e-06,
+      "loss": 0.8669,
+      "step": 116180
+    },
+    {
+      "epoch": 0.8410606093509088,
+      "grad_norm": 0.14891086518764496,
+      "learning_rate": 4.15894662931515e-06,
+      "loss": 0.885,
+      "step": 116190
+    },
+    {
+      "epoch": 0.841132996011495,
+      "grad_norm": 0.15103653073310852,
+      "learning_rate": 4.158874242654564e-06,
+      "loss": 0.8782,
+      "step": 116200
+    },
+    {
+      "epoch": 0.8412053826720812,
+      "grad_norm": 0.15480820834636688,
+      "learning_rate": 4.158801855993978e-06,
+      "loss": 0.8741,
+      "step": 116210
+    },
+    {
+      "epoch": 0.8412777693326674,
+      "grad_norm": 0.17282061278820038,
+      "learning_rate": 4.158729469333392e-06,
+      "loss": 0.8843,
+      "step": 116220
+    },
+    {
+      "epoch": 0.8413501559932536,
+      "grad_norm": 0.15457768738269806,
+      "learning_rate": 4.1586570826728055e-06,
+      "loss": 0.8808,
+      "step": 116230
+    },
+    {
+      "epoch": 0.8414225426538398,
+      "grad_norm": 0.16778331995010376,
+      "learning_rate": 4.158584696012219e-06,
+      "loss": 0.8677,
+      "step": 116240
+    },
+    {
+      "epoch": 0.8414949293144259,
+      "grad_norm": 0.15565545856952667,
+      "learning_rate": 4.158512309351633e-06,
+      "loss": 0.8793,
+      "step": 116250
+    },
+    {
+      "epoch": 0.8415673159750121,
+      "grad_norm": 0.14847850799560547,
+      "learning_rate": 4.158439922691046e-06,
+      "loss": 0.8677,
+      "step": 116260
+    },
+    {
+      "epoch": 0.8416397026355983,
+      "grad_norm": 0.15020453929901123,
+      "learning_rate": 4.15836753603046e-06,
+      "loss": 0.8863,
+      "step": 116270
+    },
+    {
+      "epoch": 0.8417120892961845,
+      "grad_norm": 0.16893228888511658,
+      "learning_rate": 4.1582951493698745e-06,
+      "loss": 0.8815,
+      "step": 116280
+    },
+    {
+      "epoch": 0.8417844759567706,
+      "grad_norm": 0.17535464465618134,
+      "learning_rate": 4.158222762709288e-06,
+      "loss": 0.8735,
+      "step": 116290
+    },
+    {
+      "epoch": 0.8418568626173569,
+      "grad_norm": 0.15316392481327057,
+      "learning_rate": 4.158150376048702e-06,
+      "loss": 0.876,
+      "step": 116300
+    },
+    {
+      "epoch": 0.8419292492779431,
+      "grad_norm": 0.16685836017131805,
+      "learning_rate": 4.158077989388115e-06,
+      "loss": 0.873,
+      "step": 116310
+    },
+    {
+      "epoch": 0.8420016359385293,
+      "grad_norm": 0.16232609748840332,
+      "learning_rate": 4.15800560272753e-06,
+      "loss": 0.8792,
+      "step": 116320
+    },
+    {
+      "epoch": 0.8420740225991155,
+      "grad_norm": 0.1534125953912735,
+      "learning_rate": 4.157933216066943e-06,
+      "loss": 0.8802,
+      "step": 116330
+    },
+    {
+      "epoch": 0.8421464092597016,
+      "grad_norm": 0.16465212404727936,
+      "learning_rate": 4.157860829406357e-06,
+      "loss": 0.8792,
+      "step": 116340
+    },
+    {
+      "epoch": 0.8422187959202878,
+      "grad_norm": 0.1542976200580597,
+      "learning_rate": 4.157788442745771e-06,
+      "loss": 0.8803,
+      "step": 116350
+    },
+    {
+      "epoch": 0.842291182580874,
+      "grad_norm": 0.1525125801563263,
+      "learning_rate": 4.157716056085185e-06,
+      "loss": 0.8787,
+      "step": 116360
+    },
+    {
+      "epoch": 0.8423635692414602,
+      "grad_norm": 0.16965779662132263,
+      "learning_rate": 4.157643669424599e-06,
+      "loss": 0.8799,
+      "step": 116370
+    },
+    {
+      "epoch": 0.8424359559020463,
+      "grad_norm": 0.14481857419013977,
+      "learning_rate": 4.157571282764012e-06,
+      "loss": 0.8803,
+      "step": 116380
+    },
+    {
+      "epoch": 0.8425083425626325,
+      "grad_norm": 0.15240401029586792,
+      "learning_rate": 4.157498896103426e-06,
+      "loss": 0.8813,
+      "step": 116390
+    },
+    {
+      "epoch": 0.8425807292232187,
+      "grad_norm": 0.17337167263031006,
+      "learning_rate": 4.15742650944284e-06,
+      "loss": 0.8711,
+      "step": 116400
+    },
+    {
+      "epoch": 0.842653115883805,
+      "grad_norm": 0.14756740629673004,
+      "learning_rate": 4.157354122782254e-06,
+      "loss": 0.8615,
+      "step": 116410
+    },
+    {
+      "epoch": 0.8427255025443912,
+      "grad_norm": 0.16462619602680206,
+      "learning_rate": 4.157281736121668e-06,
+      "loss": 0.8717,
+      "step": 116420
+    },
+    {
+      "epoch": 0.8427978892049773,
+      "grad_norm": 0.1649799644947052,
+      "learning_rate": 4.157209349461081e-06,
+      "loss": 0.8778,
+      "step": 116430
+    },
+    {
+      "epoch": 0.8428702758655635,
+      "grad_norm": 0.1735529750585556,
+      "learning_rate": 4.157136962800496e-06,
+      "loss": 0.8802,
+      "step": 116440
+    },
+    {
+      "epoch": 0.8429426625261497,
+      "grad_norm": 0.19218334555625916,
+      "learning_rate": 4.157064576139909e-06,
+      "loss": 0.8801,
+      "step": 116450
+    },
+    {
+      "epoch": 0.8430150491867359,
+      "grad_norm": 0.15204769372940063,
+      "learning_rate": 4.156992189479323e-06,
+      "loss": 0.8728,
+      "step": 116460
+    },
+    {
+      "epoch": 0.843087435847322,
+      "grad_norm": 0.15448568761348724,
+      "learning_rate": 4.1569198028187365e-06,
+      "loss": 0.8687,
+      "step": 116470
+    },
+    {
+      "epoch": 0.8431598225079082,
+      "grad_norm": 0.15800927579402924,
+      "learning_rate": 4.156847416158151e-06,
+      "loss": 0.8742,
+      "step": 116480
+    },
+    {
+      "epoch": 0.8432322091684944,
+      "grad_norm": 0.1503394991159439,
+      "learning_rate": 4.156775029497565e-06,
+      "loss": 0.875,
+      "step": 116490
+    },
+    {
+      "epoch": 0.8433045958290806,
+      "grad_norm": 0.5347150564193726,
+      "learning_rate": 4.156702642836978e-06,
+      "loss": 0.8686,
+      "step": 116500
+    },
+    {
+      "epoch": 0.8433769824896667,
+      "grad_norm": 0.1692754626274109,
+      "learning_rate": 4.156630256176392e-06,
+      "loss": 0.8775,
+      "step": 116510
+    },
+    {
+      "epoch": 0.843449369150253,
+      "grad_norm": 0.15633852779865265,
+      "learning_rate": 4.156557869515806e-06,
+      "loss": 0.8743,
+      "step": 116520
+    },
+    {
+      "epoch": 0.8435217558108392,
+      "grad_norm": 0.15997891128063202,
+      "learning_rate": 4.15648548285522e-06,
+      "loss": 0.8746,
+      "step": 116530
+    },
+    {
+      "epoch": 0.8435941424714254,
+      "grad_norm": 0.15815088152885437,
+      "learning_rate": 4.1564130961946336e-06,
+      "loss": 0.8701,
+      "step": 116540
+    },
+    {
+      "epoch": 0.8436665291320116,
+      "grad_norm": 0.1581558734178543,
+      "learning_rate": 4.156340709534047e-06,
+      "loss": 0.8803,
+      "step": 116550
+    },
+    {
+      "epoch": 0.8437389157925977,
+      "grad_norm": 0.15266630053520203,
+      "learning_rate": 4.156268322873462e-06,
+      "loss": 0.8638,
+      "step": 116560
+    },
+    {
+      "epoch": 0.8438113024531839,
+      "grad_norm": 0.15142551064491272,
+      "learning_rate": 4.156195936212875e-06,
+      "loss": 0.8699,
+      "step": 116570
+    },
+    {
+      "epoch": 0.8438836891137701,
+      "grad_norm": 0.1561920940876007,
+      "learning_rate": 4.156123549552289e-06,
+      "loss": 0.8756,
+      "step": 116580
+    },
+    {
+      "epoch": 0.8439560757743563,
+      "grad_norm": 0.16398534178733826,
+      "learning_rate": 4.1560511628917025e-06,
+      "loss": 0.8767,
+      "step": 116590
+    },
+    {
+      "epoch": 0.8440284624349425,
+      "grad_norm": 0.1780395209789276,
+      "learning_rate": 4.155978776231117e-06,
+      "loss": 0.8656,
+      "step": 116600
+    },
+    {
+      "epoch": 0.8441008490955286,
+      "grad_norm": 0.16467039287090302,
+      "learning_rate": 4.1559063895705306e-06,
+      "loss": 0.8734,
+      "step": 116610
+    },
+    {
+      "epoch": 0.8441732357561149,
+      "grad_norm": 0.16053692996501923,
+      "learning_rate": 4.155834002909944e-06,
+      "loss": 0.8777,
+      "step": 116620
+    },
+    {
+      "epoch": 0.8442456224167011,
+      "grad_norm": 0.15318801999092102,
+      "learning_rate": 4.155761616249358e-06,
+      "loss": 0.8797,
+      "step": 116630
+    },
+    {
+      "epoch": 0.8443180090772873,
+      "grad_norm": 0.1649479866027832,
+      "learning_rate": 4.155689229588772e-06,
+      "loss": 0.886,
+      "step": 116640
+    },
+    {
+      "epoch": 0.8443903957378734,
+      "grad_norm": 0.15145239233970642,
+      "learning_rate": 4.155616842928186e-06,
+      "loss": 0.8599,
+      "step": 116650
+    },
+    {
+      "epoch": 0.8444627823984596,
+      "grad_norm": 0.15609456598758698,
+      "learning_rate": 4.1555444562675995e-06,
+      "loss": 0.8787,
+      "step": 116660
+    },
+    {
+      "epoch": 0.8445351690590458,
+      "grad_norm": 0.14667485654354095,
+      "learning_rate": 4.155472069607013e-06,
+      "loss": 0.8736,
+      "step": 116670
+    },
+    {
+      "epoch": 0.844607555719632,
+      "grad_norm": 0.1538429707288742,
+      "learning_rate": 4.1553996829464276e-06,
+      "loss": 0.8859,
+      "step": 116680
+    },
+    {
+      "epoch": 0.8446799423802182,
+      "grad_norm": 0.1470920741558075,
+      "learning_rate": 4.155327296285841e-06,
+      "loss": 0.8801,
+      "step": 116690
+    },
+    {
+      "epoch": 0.8447523290408043,
+      "grad_norm": 0.2034630924463272,
+      "learning_rate": 4.155254909625255e-06,
+      "loss": 0.8741,
+      "step": 116700
+    },
+    {
+      "epoch": 0.8448247157013905,
+      "grad_norm": 0.14927048981189728,
+      "learning_rate": 4.155182522964668e-06,
+      "loss": 0.876,
+      "step": 116710
+    },
+    {
+      "epoch": 0.8448971023619767,
+      "grad_norm": 0.16222061216831207,
+      "learning_rate": 4.155110136304082e-06,
+      "loss": 0.8748,
+      "step": 116720
+    },
+    {
+      "epoch": 0.844969489022563,
+      "grad_norm": 0.15882587432861328,
+      "learning_rate": 4.1550377496434965e-06,
+      "loss": 0.885,
+      "step": 116730
+    },
+    {
+      "epoch": 0.8450418756831491,
+      "grad_norm": 0.1530759185552597,
+      "learning_rate": 4.15496536298291e-06,
+      "loss": 0.8758,
+      "step": 116740
+    },
+    {
+      "epoch": 0.8451142623437353,
+      "grad_norm": 0.15886208415031433,
+      "learning_rate": 4.154892976322324e-06,
+      "loss": 0.8756,
+      "step": 116750
+    },
+    {
+      "epoch": 0.8451866490043215,
+      "grad_norm": 0.1658525913953781,
+      "learning_rate": 4.154820589661737e-06,
+      "loss": 0.8642,
+      "step": 116760
+    },
+    {
+      "epoch": 0.8452590356649077,
+      "grad_norm": 0.15931586921215057,
+      "learning_rate": 4.154748203001152e-06,
+      "loss": 0.8726,
+      "step": 116770
+    },
+    {
+      "epoch": 0.8453314223254939,
+      "grad_norm": 0.16602514684200287,
+      "learning_rate": 4.154675816340565e-06,
+      "loss": 0.8694,
+      "step": 116780
+    },
+    {
+      "epoch": 0.84540380898608,
+      "grad_norm": 0.14427199959754944,
+      "learning_rate": 4.154603429679979e-06,
+      "loss": 0.8666,
+      "step": 116790
+    },
+    {
+      "epoch": 0.8454761956466662,
+      "grad_norm": 0.15520738065242767,
+      "learning_rate": 4.154531043019393e-06,
+      "loss": 0.8778,
+      "step": 116800
+    },
+    {
+      "epoch": 0.8455485823072524,
+      "grad_norm": 0.15710857510566711,
+      "learning_rate": 4.154458656358806e-06,
+      "loss": 0.8681,
+      "step": 116810
+    },
+    {
+      "epoch": 0.8456209689678386,
+      "grad_norm": 0.1453404724597931,
+      "learning_rate": 4.15438626969822e-06,
+      "loss": 0.8734,
+      "step": 116820
+    },
+    {
+      "epoch": 0.8456933556284248,
+      "grad_norm": 0.16362370550632477,
+      "learning_rate": 4.154313883037634e-06,
+      "loss": 0.8842,
+      "step": 116830
+    },
+    {
+      "epoch": 0.845765742289011,
+      "grad_norm": 0.14579138159751892,
+      "learning_rate": 4.154241496377048e-06,
+      "loss": 0.8635,
+      "step": 116840
+    },
+    {
+      "epoch": 0.8458381289495972,
+      "grad_norm": 0.1520829200744629,
+      "learning_rate": 4.154169109716462e-06,
+      "loss": 0.8801,
+      "step": 116850
+    },
+    {
+      "epoch": 0.8459105156101834,
+      "grad_norm": 0.1613858938217163,
+      "learning_rate": 4.154096723055875e-06,
+      "loss": 0.8858,
+      "step": 116860
+    },
+    {
+      "epoch": 0.8459829022707696,
+      "grad_norm": 0.14903327822685242,
+      "learning_rate": 4.154024336395289e-06,
+      "loss": 0.8801,
+      "step": 116870
+    },
+    {
+      "epoch": 0.8460552889313557,
+      "grad_norm": 0.2931915819644928,
+      "learning_rate": 4.153951949734703e-06,
+      "loss": 0.8663,
+      "step": 116880
+    },
+    {
+      "epoch": 0.8461276755919419,
+      "grad_norm": 0.1518881767988205,
+      "learning_rate": 4.153879563074117e-06,
+      "loss": 0.8729,
+      "step": 116890
+    },
+    {
+      "epoch": 0.8462000622525281,
+      "grad_norm": 0.15249276161193848,
+      "learning_rate": 4.1538071764135305e-06,
+      "loss": 0.8809,
+      "step": 116900
+    },
+    {
+      "epoch": 0.8462724489131143,
+      "grad_norm": 0.1551447957754135,
+      "learning_rate": 4.153734789752944e-06,
+      "loss": 0.8857,
+      "step": 116910
+    },
+    {
+      "epoch": 0.8463448355737004,
+      "grad_norm": 0.17283110320568085,
+      "learning_rate": 4.153662403092359e-06,
+      "loss": 0.8796,
+      "step": 116920
+    },
+    {
+      "epoch": 0.8464172222342866,
+      "grad_norm": 0.1588117629289627,
+      "learning_rate": 4.153590016431772e-06,
+      "loss": 0.8738,
+      "step": 116930
+    },
+    {
+      "epoch": 0.8464896088948729,
+      "grad_norm": 0.15109075605869293,
+      "learning_rate": 4.153517629771186e-06,
+      "loss": 0.8842,
+      "step": 116940
+    },
+    {
+      "epoch": 0.8465619955554591,
+      "grad_norm": 0.15133486688137054,
+      "learning_rate": 4.1534452431105994e-06,
+      "loss": 0.8784,
+      "step": 116950
+    },
+    {
+      "epoch": 0.8466343822160453,
+      "grad_norm": 0.1452382504940033,
+      "learning_rate": 4.153372856450014e-06,
+      "loss": 0.8816,
+      "step": 116960
+    },
+    {
+      "epoch": 0.8467067688766314,
+      "grad_norm": 0.14603115618228912,
+      "learning_rate": 4.1533004697894275e-06,
+      "loss": 0.89,
+      "step": 116970
+    },
+    {
+      "epoch": 0.8467791555372176,
+      "grad_norm": 0.15353167057037354,
+      "learning_rate": 4.153228083128841e-06,
+      "loss": 0.8786,
+      "step": 116980
+    },
+    {
+      "epoch": 0.8468515421978038,
+      "grad_norm": 0.14847798645496368,
+      "learning_rate": 4.153155696468255e-06,
+      "loss": 0.8782,
+      "step": 116990
+    },
+    {
+      "epoch": 0.84692392885839,
+      "grad_norm": 0.15351562201976776,
+      "learning_rate": 4.153083309807669e-06,
+      "loss": 0.8894,
+      "step": 117000
+    },
+    {
+      "epoch": 0.8469963155189761,
+      "grad_norm": 0.14607466757297516,
+      "learning_rate": 4.153010923147083e-06,
+      "loss": 0.8665,
+      "step": 117010
+    },
+    {
+      "epoch": 0.8470687021795623,
+      "grad_norm": 0.1430024653673172,
+      "learning_rate": 4.1529385364864964e-06,
+      "loss": 0.8724,
+      "step": 117020
+    },
+    {
+      "epoch": 0.8471410888401485,
+      "grad_norm": 0.15325815975666046,
+      "learning_rate": 4.15286614982591e-06,
+      "loss": 0.863,
+      "step": 117030
+    },
+    {
+      "epoch": 0.8472134755007347,
+      "grad_norm": 0.14200498163700104,
+      "learning_rate": 4.1527937631653245e-06,
+      "loss": 0.8753,
+      "step": 117040
+    },
+    {
+      "epoch": 0.847285862161321,
+      "grad_norm": 0.17369510233402252,
+      "learning_rate": 4.152721376504738e-06,
+      "loss": 0.8768,
+      "step": 117050
+    },
+    {
+      "epoch": 0.8473582488219071,
+      "grad_norm": 0.15766370296478271,
+      "learning_rate": 4.152648989844152e-06,
+      "loss": 0.877,
+      "step": 117060
+    },
+    {
+      "epoch": 0.8474306354824933,
+      "grad_norm": 0.16461563110351562,
+      "learning_rate": 4.152576603183565e-06,
+      "loss": 0.8692,
+      "step": 117070
+    },
+    {
+      "epoch": 0.8475030221430795,
+      "grad_norm": 0.1531338095664978,
+      "learning_rate": 4.15250421652298e-06,
+      "loss": 0.8897,
+      "step": 117080
+    },
+    {
+      "epoch": 0.8475754088036657,
+      "grad_norm": 0.15825554728507996,
+      "learning_rate": 4.1524318298623935e-06,
+      "loss": 0.8883,
+      "step": 117090
+    },
+    {
+      "epoch": 0.8476477954642518,
+      "grad_norm": 0.1514403074979782,
+      "learning_rate": 4.152359443201807e-06,
+      "loss": 0.886,
+      "step": 117100
+    },
+    {
+      "epoch": 0.847720182124838,
+      "grad_norm": 0.1458583027124405,
+      "learning_rate": 4.152287056541221e-06,
+      "loss": 0.8762,
+      "step": 117110
+    },
+    {
+      "epoch": 0.8477925687854242,
+      "grad_norm": 0.18490411341190338,
+      "learning_rate": 4.152214669880635e-06,
+      "loss": 0.8753,
+      "step": 117120
+    },
+    {
+      "epoch": 0.8478649554460104,
+      "grad_norm": 0.16968803107738495,
+      "learning_rate": 4.152142283220049e-06,
+      "loss": 0.8885,
+      "step": 117130
+    },
+    {
+      "epoch": 0.8479373421065965,
+      "grad_norm": 0.15850697457790375,
+      "learning_rate": 4.152069896559462e-06,
+      "loss": 0.8867,
+      "step": 117140
+    },
+    {
+      "epoch": 0.8480097287671828,
+      "grad_norm": 0.15073266625404358,
+      "learning_rate": 4.151997509898876e-06,
+      "loss": 0.8838,
+      "step": 117150
+    },
+    {
+      "epoch": 0.848082115427769,
+      "grad_norm": 0.1633184403181076,
+      "learning_rate": 4.1519251232382905e-06,
+      "loss": 0.8646,
+      "step": 117160
+    },
+    {
+      "epoch": 0.8481545020883552,
+      "grad_norm": 0.1570475846529007,
+      "learning_rate": 4.151852736577704e-06,
+      "loss": 0.8869,
+      "step": 117170
+    },
+    {
+      "epoch": 0.8482268887489414,
+      "grad_norm": 0.152681365609169,
+      "learning_rate": 4.151780349917118e-06,
+      "loss": 0.874,
+      "step": 117180
+    },
+    {
+      "epoch": 0.8482992754095275,
+      "grad_norm": 0.15066421031951904,
+      "learning_rate": 4.151707963256531e-06,
+      "loss": 0.8739,
+      "step": 117190
+    },
+    {
+      "epoch": 0.8483716620701137,
+      "grad_norm": 0.14346186816692352,
+      "learning_rate": 4.151635576595946e-06,
+      "loss": 0.8754,
+      "step": 117200
+    },
+    {
+      "epoch": 0.8484440487306999,
+      "grad_norm": 0.14868153631687164,
+      "learning_rate": 4.151563189935359e-06,
+      "loss": 0.8744,
+      "step": 117210
+    },
+    {
+      "epoch": 0.8485164353912861,
+      "grad_norm": 0.16770720481872559,
+      "learning_rate": 4.151490803274773e-06,
+      "loss": 0.8894,
+      "step": 117220
+    },
+    {
+      "epoch": 0.8485888220518722,
+      "grad_norm": 0.15046921372413635,
+      "learning_rate": 4.151418416614187e-06,
+      "loss": 0.8617,
+      "step": 117230
+    },
+    {
+      "epoch": 0.8486612087124584,
+      "grad_norm": 0.596076488494873,
+      "learning_rate": 4.151346029953601e-06,
+      "loss": 0.8723,
+      "step": 117240
+    },
+    {
+      "epoch": 0.8487335953730446,
+      "grad_norm": 0.15184585750102997,
+      "learning_rate": 4.151273643293015e-06,
+      "loss": 0.8841,
+      "step": 117250
+    },
+    {
+      "epoch": 0.8488059820336309,
+      "grad_norm": 0.14896786212921143,
+      "learning_rate": 4.151201256632428e-06,
+      "loss": 0.8806,
+      "step": 117260
+    },
+    {
+      "epoch": 0.8488783686942171,
+      "grad_norm": 0.15315642952919006,
+      "learning_rate": 4.151128869971842e-06,
+      "loss": 0.8845,
+      "step": 117270
+    },
+    {
+      "epoch": 0.8489507553548032,
+      "grad_norm": 0.15430568158626556,
+      "learning_rate": 4.151056483311256e-06,
+      "loss": 0.8775,
+      "step": 117280
+    },
+    {
+      "epoch": 0.8490231420153894,
+      "grad_norm": 0.170336052775383,
+      "learning_rate": 4.15098409665067e-06,
+      "loss": 0.8547,
+      "step": 117290
+    },
+    {
+      "epoch": 0.8490955286759756,
+      "grad_norm": 0.14343459904193878,
+      "learning_rate": 4.150911709990084e-06,
+      "loss": 0.873,
+      "step": 117300
+    },
+    {
+      "epoch": 0.8491679153365618,
+      "grad_norm": 0.20575720071792603,
+      "learning_rate": 4.150839323329497e-06,
+      "loss": 0.873,
+      "step": 117310
+    },
+    {
+      "epoch": 0.849240301997148,
+      "grad_norm": 0.15282517671585083,
+      "learning_rate": 4.150766936668911e-06,
+      "loss": 0.8895,
+      "step": 117320
+    },
+    {
+      "epoch": 0.8493126886577341,
+      "grad_norm": 0.1475241631269455,
+      "learning_rate": 4.1506945500083245e-06,
+      "loss": 0.8727,
+      "step": 117330
+    },
+    {
+      "epoch": 0.8493850753183203,
+      "grad_norm": 0.1757318675518036,
+      "learning_rate": 4.150622163347738e-06,
+      "loss": 0.8726,
+      "step": 117340
+    },
+    {
+      "epoch": 0.8494574619789065,
+      "grad_norm": 0.1519278883934021,
+      "learning_rate": 4.1505497766871526e-06,
+      "loss": 0.8614,
+      "step": 117350
+    },
+    {
+      "epoch": 0.8495298486394928,
+      "grad_norm": 0.14336329698562622,
+      "learning_rate": 4.150477390026566e-06,
+      "loss": 0.879,
+      "step": 117360
+    },
+    {
+      "epoch": 0.849602235300079,
+      "grad_norm": 0.15254177153110504,
+      "learning_rate": 4.15040500336598e-06,
+      "loss": 0.8768,
+      "step": 117370
+    },
+    {
+      "epoch": 0.8496746219606651,
+      "grad_norm": 0.15974296629428864,
+      "learning_rate": 4.150332616705393e-06,
+      "loss": 0.8659,
+      "step": 117380
+    },
+    {
+      "epoch": 0.8497470086212513,
+      "grad_norm": 0.15729989111423492,
+      "learning_rate": 4.150260230044808e-06,
+      "loss": 0.8687,
+      "step": 117390
+    },
+    {
+      "epoch": 0.8498193952818375,
+      "grad_norm": 0.15651953220367432,
+      "learning_rate": 4.1501878433842215e-06,
+      "loss": 0.8749,
+      "step": 117400
+    },
+    {
+      "epoch": 0.8498917819424237,
+      "grad_norm": 0.14588718116283417,
+      "learning_rate": 4.150115456723635e-06,
+      "loss": 0.8758,
+      "step": 117410
+    },
+    {
+      "epoch": 0.8499641686030098,
+      "grad_norm": 0.15261155366897583,
+      "learning_rate": 4.150043070063049e-06,
+      "loss": 0.8791,
+      "step": 117420
+    },
+    {
+      "epoch": 0.850036555263596,
+      "grad_norm": 0.16166460514068604,
+      "learning_rate": 4.149970683402463e-06,
+      "loss": 0.8829,
+      "step": 117430
+    },
+    {
+      "epoch": 0.8501089419241822,
+      "grad_norm": 0.1595904678106308,
+      "learning_rate": 4.149898296741877e-06,
+      "loss": 0.8713,
+      "step": 117440
+    },
+    {
+      "epoch": 0.8501813285847684,
+      "grad_norm": 0.1586706042289734,
+      "learning_rate": 4.14982591008129e-06,
+      "loss": 0.8689,
+      "step": 117450
+    },
+    {
+      "epoch": 0.8502537152453545,
+      "grad_norm": 0.17200522124767303,
+      "learning_rate": 4.149753523420704e-06,
+      "loss": 0.88,
+      "step": 117460
+    },
+    {
+      "epoch": 0.8503261019059408,
+      "grad_norm": 0.17994257807731628,
+      "learning_rate": 4.1496811367601185e-06,
+      "loss": 0.8885,
+      "step": 117470
+    },
+    {
+      "epoch": 0.850398488566527,
+      "grad_norm": 0.1691487431526184,
+      "learning_rate": 4.149608750099532e-06,
+      "loss": 0.8759,
+      "step": 117480
+    },
+    {
+      "epoch": 0.8504708752271132,
+      "grad_norm": 0.1492878645658493,
+      "learning_rate": 4.149536363438946e-06,
+      "loss": 0.8862,
+      "step": 117490
+    },
+    {
+      "epoch": 0.8505432618876994,
+      "grad_norm": 0.15258543193340302,
+      "learning_rate": 4.149463976778359e-06,
+      "loss": 0.8684,
+      "step": 117500
+    },
+    {
+      "epoch": 0.8506156485482855,
+      "grad_norm": 0.14704549312591553,
+      "learning_rate": 4.149391590117773e-06,
+      "loss": 0.8682,
+      "step": 117510
+    },
+    {
+      "epoch": 0.8506880352088717,
+      "grad_norm": 0.1570146381855011,
+      "learning_rate": 4.1493192034571874e-06,
+      "loss": 0.8638,
+      "step": 117520
+    },
+    {
+      "epoch": 0.8507604218694579,
+      "grad_norm": 0.1731637567281723,
+      "learning_rate": 4.149246816796601e-06,
+      "loss": 0.8875,
+      "step": 117530
+    },
+    {
+      "epoch": 0.8508328085300441,
+      "grad_norm": 0.15617217123508453,
+      "learning_rate": 4.149174430136015e-06,
+      "loss": 0.8981,
+      "step": 117540
+    },
+    {
+      "epoch": 0.8509051951906302,
+      "grad_norm": 0.15860138833522797,
+      "learning_rate": 4.149102043475428e-06,
+      "loss": 0.8783,
+      "step": 117550
+    },
+    {
+      "epoch": 0.8509775818512164,
+      "grad_norm": 0.15152351558208466,
+      "learning_rate": 4.149029656814843e-06,
+      "loss": 0.8784,
+      "step": 117560
+    },
+    {
+      "epoch": 0.8510499685118026,
+      "grad_norm": 0.15892478823661804,
+      "learning_rate": 4.148957270154256e-06,
+      "loss": 0.8943,
+      "step": 117570
+    },
+    {
+      "epoch": 0.8511223551723889,
+      "grad_norm": 0.13866350054740906,
+      "learning_rate": 4.14888488349367e-06,
+      "loss": 0.8671,
+      "step": 117580
+    },
+    {
+      "epoch": 0.851194741832975,
+      "grad_norm": 0.22664818167686462,
+      "learning_rate": 4.148812496833084e-06,
+      "loss": 0.87,
+      "step": 117590
+    },
+    {
+      "epoch": 0.8512671284935612,
+      "grad_norm": 0.1927105337381363,
+      "learning_rate": 4.148740110172498e-06,
+      "loss": 0.8713,
+      "step": 117600
+    },
+    {
+      "epoch": 0.8513395151541474,
+      "grad_norm": 0.19467182457447052,
+      "learning_rate": 4.148667723511912e-06,
+      "loss": 0.879,
+      "step": 117610
+    },
+    {
+      "epoch": 0.8514119018147336,
+      "grad_norm": 0.1516137421131134,
+      "learning_rate": 4.148595336851325e-06,
+      "loss": 0.8693,
+      "step": 117620
+    },
+    {
+      "epoch": 0.8514842884753198,
+      "grad_norm": 0.15127213299274445,
+      "learning_rate": 4.148522950190739e-06,
+      "loss": 0.8648,
+      "step": 117630
+    },
+    {
+      "epoch": 0.8515566751359059,
+      "grad_norm": 0.16966821253299713,
+      "learning_rate": 4.148450563530153e-06,
+      "loss": 0.8681,
+      "step": 117640
+    },
+    {
+      "epoch": 0.8516290617964921,
+      "grad_norm": 0.17517682909965515,
+      "learning_rate": 4.148378176869567e-06,
+      "loss": 0.8782,
+      "step": 117650
+    },
+    {
+      "epoch": 0.8517014484570783,
+      "grad_norm": 0.154166117310524,
+      "learning_rate": 4.148305790208981e-06,
+      "loss": 0.8838,
+      "step": 117660
+    },
+    {
+      "epoch": 0.8517738351176645,
+      "grad_norm": 0.16335949301719666,
+      "learning_rate": 4.148233403548394e-06,
+      "loss": 0.88,
+      "step": 117670
+    },
+    {
+      "epoch": 0.8518462217782508,
+      "grad_norm": 0.15556417405605316,
+      "learning_rate": 4.148161016887809e-06,
+      "loss": 0.8693,
+      "step": 117680
+    },
+    {
+      "epoch": 0.8519186084388369,
+      "grad_norm": 0.14646877348423004,
+      "learning_rate": 4.148088630227222e-06,
+      "loss": 0.8817,
+      "step": 117690
+    },
+    {
+      "epoch": 0.8519909950994231,
+      "grad_norm": 0.15498358011245728,
+      "learning_rate": 4.148016243566636e-06,
+      "loss": 0.875,
+      "step": 117700
+    },
+    {
+      "epoch": 0.8520633817600093,
+      "grad_norm": 0.17575348913669586,
+      "learning_rate": 4.1479438569060495e-06,
+      "loss": 0.8792,
+      "step": 117710
+    },
+    {
+      "epoch": 0.8521357684205955,
+      "grad_norm": 0.14134739339351654,
+      "learning_rate": 4.147871470245464e-06,
+      "loss": 0.8778,
+      "step": 117720
+    },
+    {
+      "epoch": 0.8522081550811816,
+      "grad_norm": 0.18208590149879456,
+      "learning_rate": 4.147799083584878e-06,
+      "loss": 0.8799,
+      "step": 117730
+    },
+    {
+      "epoch": 0.8522805417417678,
+      "grad_norm": 0.15532927215099335,
+      "learning_rate": 4.147726696924291e-06,
+      "loss": 0.8852,
+      "step": 117740
+    },
+    {
+      "epoch": 0.852352928402354,
+      "grad_norm": 0.15121373534202576,
+      "learning_rate": 4.147654310263705e-06,
+      "loss": 0.8639,
+      "step": 117750
+    },
+    {
+      "epoch": 0.8524253150629402,
+      "grad_norm": 0.14457499980926514,
+      "learning_rate": 4.147581923603119e-06,
+      "loss": 0.8829,
+      "step": 117760
+    },
+    {
+      "epoch": 0.8524977017235263,
+      "grad_norm": 0.1544269621372223,
+      "learning_rate": 4.147509536942533e-06,
+      "loss": 0.8777,
+      "step": 117770
+    },
+    {
+      "epoch": 0.8525700883841125,
+      "grad_norm": 0.15642063319683075,
+      "learning_rate": 4.1474371502819465e-06,
+      "loss": 0.879,
+      "step": 117780
+    },
+    {
+      "epoch": 0.8526424750446988,
+      "grad_norm": 0.3743261694908142,
+      "learning_rate": 4.14736476362136e-06,
+      "loss": 0.8811,
+      "step": 117790
+    },
+    {
+      "epoch": 0.852714861705285,
+      "grad_norm": 0.14951781928539276,
+      "learning_rate": 4.147292376960775e-06,
+      "loss": 0.8695,
+      "step": 117800
+    },
+    {
+      "epoch": 0.8527872483658712,
+      "grad_norm": 0.153753861784935,
+      "learning_rate": 4.147219990300188e-06,
+      "loss": 0.8706,
+      "step": 117810
+    },
+    {
+      "epoch": 0.8528596350264573,
+      "grad_norm": 0.15326263010501862,
+      "learning_rate": 4.147147603639602e-06,
+      "loss": 0.8746,
+      "step": 117820
+    },
+    {
+      "epoch": 0.8529320216870435,
+      "grad_norm": 0.16046510636806488,
+      "learning_rate": 4.1470752169790155e-06,
+      "loss": 0.8765,
+      "step": 117830
+    },
+    {
+      "epoch": 0.8530044083476297,
+      "grad_norm": 0.15566720068454742,
+      "learning_rate": 4.147002830318429e-06,
+      "loss": 0.8806,
+      "step": 117840
+    },
+    {
+      "epoch": 0.8530767950082159,
+      "grad_norm": 0.16101893782615662,
+      "learning_rate": 4.146930443657843e-06,
+      "loss": 0.8745,
+      "step": 117850
+    },
+    {
+      "epoch": 0.853149181668802,
+      "grad_norm": 0.15617725253105164,
+      "learning_rate": 4.146858056997256e-06,
+      "loss": 0.8773,
+      "step": 117860
+    },
+    {
+      "epoch": 0.8532215683293882,
+      "grad_norm": 0.19228564202785492,
+      "learning_rate": 4.146785670336671e-06,
+      "loss": 0.8827,
+      "step": 117870
+    },
+    {
+      "epoch": 0.8532939549899744,
+      "grad_norm": 0.17179349064826965,
+      "learning_rate": 4.146713283676084e-06,
+      "loss": 0.8657,
+      "step": 117880
+    },
+    {
+      "epoch": 0.8533663416505607,
+      "grad_norm": 0.1750807911157608,
+      "learning_rate": 4.146640897015498e-06,
+      "loss": 0.8613,
+      "step": 117890
+    },
+    {
+      "epoch": 0.8534387283111469,
+      "grad_norm": 0.17412197589874268,
+      "learning_rate": 4.146568510354912e-06,
+      "loss": 0.8641,
+      "step": 117900
+    },
+    {
+      "epoch": 0.853511114971733,
+      "grad_norm": 0.154750794172287,
+      "learning_rate": 4.146496123694326e-06,
+      "loss": 0.8799,
+      "step": 117910
+    },
+    {
+      "epoch": 0.8535835016323192,
+      "grad_norm": 0.17946046590805054,
+      "learning_rate": 4.14642373703374e-06,
+      "loss": 0.8661,
+      "step": 117920
+    },
+    {
+      "epoch": 0.8536558882929054,
+      "grad_norm": 0.17114438116550446,
+      "learning_rate": 4.146351350373153e-06,
+      "loss": 0.8755,
+      "step": 117930
+    },
+    {
+      "epoch": 0.8537282749534916,
+      "grad_norm": 0.15185768902301788,
+      "learning_rate": 4.146278963712567e-06,
+      "loss": 0.8746,
+      "step": 117940
+    },
+    {
+      "epoch": 0.8538006616140777,
+      "grad_norm": 0.17097008228302002,
+      "learning_rate": 4.146206577051981e-06,
+      "loss": 0.8732,
+      "step": 117950
+    },
+    {
+      "epoch": 0.8538730482746639,
+      "grad_norm": 0.154009148478508,
+      "learning_rate": 4.146134190391395e-06,
+      "loss": 0.8758,
+      "step": 117960
+    },
+    {
+      "epoch": 0.8539454349352501,
+      "grad_norm": 0.15704505145549774,
+      "learning_rate": 4.146061803730809e-06,
+      "loss": 0.8888,
+      "step": 117970
+    },
+    {
+      "epoch": 0.8540178215958363,
+      "grad_norm": 0.19270089268684387,
+      "learning_rate": 4.145989417070222e-06,
+      "loss": 0.8657,
+      "step": 117980
+    },
+    {
+      "epoch": 0.8540902082564225,
+      "grad_norm": 0.16790702939033508,
+      "learning_rate": 4.145917030409637e-06,
+      "loss": 0.8659,
+      "step": 117990
+    },
+    {
+      "epoch": 0.8541625949170087,
+      "grad_norm": 0.15922467410564423,
+      "learning_rate": 4.14584464374905e-06,
+      "loss": 0.8762,
+      "step": 118000
+    },
+    {
+      "epoch": 0.8542349815775949,
+      "grad_norm": 0.14866970479488373,
+      "learning_rate": 4.145772257088464e-06,
+      "loss": 0.881,
+      "step": 118010
+    },
+    {
+      "epoch": 0.8543073682381811,
+      "grad_norm": 0.15993216633796692,
+      "learning_rate": 4.1456998704278776e-06,
+      "loss": 0.8748,
+      "step": 118020
+    },
+    {
+      "epoch": 0.8543797548987673,
+      "grad_norm": 0.1770089566707611,
+      "learning_rate": 4.145627483767292e-06,
+      "loss": 0.8689,
+      "step": 118030
+    },
+    {
+      "epoch": 0.8544521415593535,
+      "grad_norm": 0.1457049697637558,
+      "learning_rate": 4.145555097106706e-06,
+      "loss": 0.8777,
+      "step": 118040
+    },
+    {
+      "epoch": 0.8545245282199396,
+      "grad_norm": 0.1655452698469162,
+      "learning_rate": 4.145482710446119e-06,
+      "loss": 0.8742,
+      "step": 118050
+    },
+    {
+      "epoch": 0.8545969148805258,
+      "grad_norm": 0.15786688029766083,
+      "learning_rate": 4.145410323785533e-06,
+      "loss": 0.8852,
+      "step": 118060
+    },
+    {
+      "epoch": 0.854669301541112,
+      "grad_norm": 0.14149673283100128,
+      "learning_rate": 4.145337937124947e-06,
+      "loss": 0.8803,
+      "step": 118070
+    },
+    {
+      "epoch": 0.8547416882016982,
+      "grad_norm": 0.19031822681427002,
+      "learning_rate": 4.145265550464361e-06,
+      "loss": 0.8798,
+      "step": 118080
+    },
+    {
+      "epoch": 0.8548140748622843,
+      "grad_norm": 0.1616319864988327,
+      "learning_rate": 4.1451931638037746e-06,
+      "loss": 0.8837,
+      "step": 118090
+    },
+    {
+      "epoch": 0.8548864615228705,
+      "grad_norm": 0.16045629978179932,
+      "learning_rate": 4.145120777143188e-06,
+      "loss": 0.8746,
+      "step": 118100
+    },
+    {
+      "epoch": 0.8549588481834568,
+      "grad_norm": 0.14599646627902985,
+      "learning_rate": 4.145048390482602e-06,
+      "loss": 0.878,
+      "step": 118110
+    },
+    {
+      "epoch": 0.855031234844043,
+      "grad_norm": 0.1440078318119049,
+      "learning_rate": 4.144976003822016e-06,
+      "loss": 0.8766,
+      "step": 118120
+    },
+    {
+      "epoch": 0.8551036215046292,
+      "grad_norm": 0.15769658982753754,
+      "learning_rate": 4.14490361716143e-06,
+      "loss": 0.8753,
+      "step": 118130
+    },
+    {
+      "epoch": 0.8551760081652153,
+      "grad_norm": 0.15050294995307922,
+      "learning_rate": 4.1448312305008435e-06,
+      "loss": 0.8737,
+      "step": 118140
+    },
+    {
+      "epoch": 0.8552483948258015,
+      "grad_norm": 0.15845508873462677,
+      "learning_rate": 4.144758843840257e-06,
+      "loss": 0.8665,
+      "step": 118150
+    },
+    {
+      "epoch": 0.8553207814863877,
+      "grad_norm": 0.16100691258907318,
+      "learning_rate": 4.1446864571796716e-06,
+      "loss": 0.8818,
+      "step": 118160
+    },
+    {
+      "epoch": 0.8553931681469739,
+      "grad_norm": 0.1659129559993744,
+      "learning_rate": 4.144614070519085e-06,
+      "loss": 0.874,
+      "step": 118170
+    },
+    {
+      "epoch": 0.85546555480756,
+      "grad_norm": 0.15678264200687408,
+      "learning_rate": 4.144541683858499e-06,
+      "loss": 0.8712,
+      "step": 118180
+    },
+    {
+      "epoch": 0.8555379414681462,
+      "grad_norm": 0.16013620793819427,
+      "learning_rate": 4.144469297197912e-06,
+      "loss": 0.8695,
+      "step": 118190
+    },
+    {
+      "epoch": 0.8556103281287324,
+      "grad_norm": 0.16265377402305603,
+      "learning_rate": 4.144396910537327e-06,
+      "loss": 0.8779,
+      "step": 118200
+    },
+    {
+      "epoch": 0.8556827147893187,
+      "grad_norm": 0.15654130280017853,
+      "learning_rate": 4.1443245238767405e-06,
+      "loss": 0.8683,
+      "step": 118210
+    },
+    {
+      "epoch": 0.8557551014499049,
+      "grad_norm": 0.17270372807979584,
+      "learning_rate": 4.144252137216154e-06,
+      "loss": 0.8814,
+      "step": 118220
+    },
+    {
+      "epoch": 0.855827488110491,
+      "grad_norm": 0.15407659113407135,
+      "learning_rate": 4.144179750555568e-06,
+      "loss": 0.8727,
+      "step": 118230
+    },
+    {
+      "epoch": 0.8558998747710772,
+      "grad_norm": 0.14619654417037964,
+      "learning_rate": 4.144107363894982e-06,
+      "loss": 0.8825,
+      "step": 118240
+    },
+    {
+      "epoch": 0.8559722614316634,
+      "grad_norm": 0.15296125411987305,
+      "learning_rate": 4.144034977234396e-06,
+      "loss": 0.8842,
+      "step": 118250
+    },
+    {
+      "epoch": 0.8560446480922496,
+      "grad_norm": 0.14285480976104736,
+      "learning_rate": 4.143962590573809e-06,
+      "loss": 0.8817,
+      "step": 118260
+    },
+    {
+      "epoch": 0.8561170347528357,
+      "grad_norm": 0.15438437461853027,
+      "learning_rate": 4.143890203913223e-06,
+      "loss": 0.8721,
+      "step": 118270
+    },
+    {
+      "epoch": 0.8561894214134219,
+      "grad_norm": 0.14615854620933533,
+      "learning_rate": 4.1438178172526375e-06,
+      "loss": 0.8703,
+      "step": 118280
+    },
+    {
+      "epoch": 0.8562618080740081,
+      "grad_norm": 0.1698932945728302,
+      "learning_rate": 4.143745430592051e-06,
+      "loss": 0.8623,
+      "step": 118290
+    },
+    {
+      "epoch": 0.8563341947345943,
+      "grad_norm": 0.14447180926799774,
+      "learning_rate": 4.143673043931465e-06,
+      "loss": 0.8622,
+      "step": 118300
+    },
+    {
+      "epoch": 0.8564065813951804,
+      "grad_norm": 0.15869450569152832,
+      "learning_rate": 4.143600657270878e-06,
+      "loss": 0.8625,
+      "step": 118310
+    },
+    {
+      "epoch": 0.8564789680557667,
+      "grad_norm": 0.16322019696235657,
+      "learning_rate": 4.143528270610293e-06,
+      "loss": 0.8714,
+      "step": 118320
+    },
+    {
+      "epoch": 0.8565513547163529,
+      "grad_norm": 0.15476880967617035,
+      "learning_rate": 4.1434558839497064e-06,
+      "loss": 0.8786,
+      "step": 118330
+    },
+    {
+      "epoch": 0.8566237413769391,
+      "grad_norm": 0.17096053063869476,
+      "learning_rate": 4.14338349728912e-06,
+      "loss": 0.8831,
+      "step": 118340
+    },
+    {
+      "epoch": 0.8566961280375253,
+      "grad_norm": 0.15711814165115356,
+      "learning_rate": 4.143311110628534e-06,
+      "loss": 0.8487,
+      "step": 118350
+    },
+    {
+      "epoch": 0.8567685146981114,
+      "grad_norm": 0.15190854668617249,
+      "learning_rate": 4.143238723967948e-06,
+      "loss": 0.872,
+      "step": 118360
+    },
+    {
+      "epoch": 0.8568409013586976,
+      "grad_norm": 0.15767568349838257,
+      "learning_rate": 4.143166337307361e-06,
+      "loss": 0.8787,
+      "step": 118370
+    },
+    {
+      "epoch": 0.8569132880192838,
+      "grad_norm": 0.13986888527870178,
+      "learning_rate": 4.1430939506467745e-06,
+      "loss": 0.8605,
+      "step": 118380
+    },
+    {
+      "epoch": 0.85698567467987,
+      "grad_norm": 0.1628597527742386,
+      "learning_rate": 4.143021563986189e-06,
+      "loss": 0.8824,
+      "step": 118390
+    },
+    {
+      "epoch": 0.8570580613404561,
+      "grad_norm": 0.14581960439682007,
+      "learning_rate": 4.142949177325603e-06,
+      "loss": 0.8637,
+      "step": 118400
+    },
+    {
+      "epoch": 0.8571304480010423,
+      "grad_norm": 0.15314272046089172,
+      "learning_rate": 4.142876790665016e-06,
+      "loss": 0.871,
+      "step": 118410
+    },
+    {
+      "epoch": 0.8572028346616286,
+      "grad_norm": 0.13960693776607513,
+      "learning_rate": 4.14280440400443e-06,
+      "loss": 0.8791,
+      "step": 118420
+    },
+    {
+      "epoch": 0.8572752213222148,
+      "grad_norm": 0.31276777386665344,
+      "learning_rate": 4.142732017343844e-06,
+      "loss": 0.8804,
+      "step": 118430
+    },
+    {
+      "epoch": 0.857347607982801,
+      "grad_norm": 0.15629543364048004,
+      "learning_rate": 4.142659630683258e-06,
+      "loss": 0.8891,
+      "step": 118440
+    },
+    {
+      "epoch": 0.8574199946433871,
+      "grad_norm": 0.15342466533184052,
+      "learning_rate": 4.1425872440226715e-06,
+      "loss": 0.883,
+      "step": 118450
+    },
+    {
+      "epoch": 0.8574923813039733,
+      "grad_norm": 0.14505906403064728,
+      "learning_rate": 4.142514857362085e-06,
+      "loss": 0.8773,
+      "step": 118460
+    },
+    {
+      "epoch": 0.8575647679645595,
+      "grad_norm": 0.17075826227664948,
+      "learning_rate": 4.1424424707015e-06,
+      "loss": 0.8876,
+      "step": 118470
+    },
+    {
+      "epoch": 0.8576371546251457,
+      "grad_norm": 0.14717842638492584,
+      "learning_rate": 4.142370084040913e-06,
+      "loss": 0.875,
+      "step": 118480
+    },
+    {
+      "epoch": 0.8577095412857318,
+      "grad_norm": 0.1587383896112442,
+      "learning_rate": 4.142297697380327e-06,
+      "loss": 0.878,
+      "step": 118490
+    },
+    {
+      "epoch": 0.857781927946318,
+      "grad_norm": 0.15151575207710266,
+      "learning_rate": 4.1422253107197404e-06,
+      "loss": 0.8881,
+      "step": 118500
+    },
+    {
+      "epoch": 0.8578543146069042,
+      "grad_norm": 0.17024588584899902,
+      "learning_rate": 4.142152924059155e-06,
+      "loss": 0.8746,
+      "step": 118510
+    },
+    {
+      "epoch": 0.8579267012674904,
+      "grad_norm": 0.16007643938064575,
+      "learning_rate": 4.1420805373985685e-06,
+      "loss": 0.8731,
+      "step": 118520
+    },
+    {
+      "epoch": 0.8579990879280767,
+      "grad_norm": 0.1677238941192627,
+      "learning_rate": 4.142008150737982e-06,
+      "loss": 0.87,
+      "step": 118530
+    },
+    {
+      "epoch": 0.8580714745886628,
+      "grad_norm": 0.1518600583076477,
+      "learning_rate": 4.141935764077396e-06,
+      "loss": 0.8786,
+      "step": 118540
+    },
+    {
+      "epoch": 0.858143861249249,
+      "grad_norm": 0.15245488286018372,
+      "learning_rate": 4.14186337741681e-06,
+      "loss": 0.8882,
+      "step": 118550
+    },
+    {
+      "epoch": 0.8582162479098352,
+      "grad_norm": 0.16277900338172913,
+      "learning_rate": 4.141790990756224e-06,
+      "loss": 0.8778,
+      "step": 118560
+    },
+    {
+      "epoch": 0.8582886345704214,
+      "grad_norm": 0.1517462432384491,
+      "learning_rate": 4.1417186040956375e-06,
+      "loss": 0.8821,
+      "step": 118570
+    },
+    {
+      "epoch": 0.8583610212310075,
+      "grad_norm": 0.1491396576166153,
+      "learning_rate": 4.141646217435051e-06,
+      "loss": 0.8738,
+      "step": 118580
+    },
+    {
+      "epoch": 0.8584334078915937,
+      "grad_norm": 0.19470219314098358,
+      "learning_rate": 4.1415738307744655e-06,
+      "loss": 0.8708,
+      "step": 118590
+    },
+    {
+      "epoch": 0.8585057945521799,
+      "grad_norm": 0.30775144696235657,
+      "learning_rate": 4.141501444113879e-06,
+      "loss": 0.8803,
+      "step": 118600
+    },
+    {
+      "epoch": 0.8585781812127661,
+      "grad_norm": 0.154507115483284,
+      "learning_rate": 4.141429057453293e-06,
+      "loss": 0.8731,
+      "step": 118610
+    },
+    {
+      "epoch": 0.8586505678733523,
+      "grad_norm": 0.17000029981136322,
+      "learning_rate": 4.141356670792706e-06,
+      "loss": 0.8884,
+      "step": 118620
+    },
+    {
+      "epoch": 0.8587229545339384,
+      "grad_norm": 0.15258820354938507,
+      "learning_rate": 4.141284284132121e-06,
+      "loss": 0.8584,
+      "step": 118630
+    },
+    {
+      "epoch": 0.8587953411945247,
+      "grad_norm": 0.15323597192764282,
+      "learning_rate": 4.1412118974715345e-06,
+      "loss": 0.8639,
+      "step": 118640
+    },
+    {
+      "epoch": 0.8588677278551109,
+      "grad_norm": 0.1578139066696167,
+      "learning_rate": 4.141139510810948e-06,
+      "loss": 0.8929,
+      "step": 118650
+    },
+    {
+      "epoch": 0.8589401145156971,
+      "grad_norm": 0.16174130141735077,
+      "learning_rate": 4.141067124150362e-06,
+      "loss": 0.887,
+      "step": 118660
+    },
+    {
+      "epoch": 0.8590125011762832,
+      "grad_norm": 0.15842093527317047,
+      "learning_rate": 4.140994737489776e-06,
+      "loss": 0.8724,
+      "step": 118670
+    },
+    {
+      "epoch": 0.8590848878368694,
+      "grad_norm": 0.1600816696882248,
+      "learning_rate": 4.14092235082919e-06,
+      "loss": 0.8744,
+      "step": 118680
+    },
+    {
+      "epoch": 0.8591572744974556,
+      "grad_norm": 0.17777082324028015,
+      "learning_rate": 4.140849964168603e-06,
+      "loss": 0.8832,
+      "step": 118690
+    },
+    {
+      "epoch": 0.8592296611580418,
+      "grad_norm": 0.15657740831375122,
+      "learning_rate": 4.140777577508017e-06,
+      "loss": 0.8791,
+      "step": 118700
+    },
+    {
+      "epoch": 0.859302047818628,
+      "grad_norm": 0.16433501243591309,
+      "learning_rate": 4.1407051908474315e-06,
+      "loss": 0.879,
+      "step": 118710
+    },
+    {
+      "epoch": 0.8593744344792141,
+      "grad_norm": 0.148828387260437,
+      "learning_rate": 4.140632804186845e-06,
+      "loss": 0.8825,
+      "step": 118720
+    },
+    {
+      "epoch": 0.8594468211398003,
+      "grad_norm": 0.1595117151737213,
+      "learning_rate": 4.140560417526259e-06,
+      "loss": 0.8729,
+      "step": 118730
+    },
+    {
+      "epoch": 0.8595192078003866,
+      "grad_norm": 0.15022175014019012,
+      "learning_rate": 4.140488030865672e-06,
+      "loss": 0.8778,
+      "step": 118740
+    },
+    {
+      "epoch": 0.8595915944609728,
+      "grad_norm": 0.16430038213729858,
+      "learning_rate": 4.140415644205086e-06,
+      "loss": 0.8535,
+      "step": 118750
+    },
+    {
+      "epoch": 0.859663981121559,
+      "grad_norm": 0.1567695438861847,
+      "learning_rate": 4.1403432575445e-06,
+      "loss": 0.8784,
+      "step": 118760
+    },
+    {
+      "epoch": 0.8597363677821451,
+      "grad_norm": 0.15546707808971405,
+      "learning_rate": 4.140270870883914e-06,
+      "loss": 0.874,
+      "step": 118770
+    },
+    {
+      "epoch": 0.8598087544427313,
+      "grad_norm": 0.16317854821681976,
+      "learning_rate": 4.140198484223328e-06,
+      "loss": 0.8694,
+      "step": 118780
+    },
+    {
+      "epoch": 0.8598811411033175,
+      "grad_norm": 0.15389098227024078,
+      "learning_rate": 4.140126097562741e-06,
+      "loss": 0.8778,
+      "step": 118790
+    },
+    {
+      "epoch": 0.8599535277639037,
+      "grad_norm": 0.17565035820007324,
+      "learning_rate": 4.140053710902156e-06,
+      "loss": 0.8855,
+      "step": 118800
+    },
+    {
+      "epoch": 0.8600259144244898,
+      "grad_norm": 0.16407647728919983,
+      "learning_rate": 4.139981324241569e-06,
+      "loss": 0.8706,
+      "step": 118810
+    },
+    {
+      "epoch": 0.860098301085076,
+      "grad_norm": 0.15503662824630737,
+      "learning_rate": 4.139908937580983e-06,
+      "loss": 0.8748,
+      "step": 118820
+    },
+    {
+      "epoch": 0.8601706877456622,
+      "grad_norm": 0.15762311220169067,
+      "learning_rate": 4.1398365509203966e-06,
+      "loss": 0.8765,
+      "step": 118830
+    },
+    {
+      "epoch": 0.8602430744062484,
+      "grad_norm": 0.16244664788246155,
+      "learning_rate": 4.139764164259811e-06,
+      "loss": 0.8762,
+      "step": 118840
+    },
+    {
+      "epoch": 0.8603154610668347,
+      "grad_norm": 0.15617172420024872,
+      "learning_rate": 4.139691777599225e-06,
+      "loss": 0.8736,
+      "step": 118850
+    },
+    {
+      "epoch": 0.8603878477274208,
+      "grad_norm": 0.1622217446565628,
+      "learning_rate": 4.139619390938638e-06,
+      "loss": 0.8735,
+      "step": 118860
+    },
+    {
+      "epoch": 0.860460234388007,
+      "grad_norm": 0.15860269963741302,
+      "learning_rate": 4.139547004278052e-06,
+      "loss": 0.8614,
+      "step": 118870
+    },
+    {
+      "epoch": 0.8605326210485932,
+      "grad_norm": 0.1792127788066864,
+      "learning_rate": 4.139474617617466e-06,
+      "loss": 0.8649,
+      "step": 118880
+    },
+    {
+      "epoch": 0.8606050077091794,
+      "grad_norm": 0.14843524992465973,
+      "learning_rate": 4.13940223095688e-06,
+      "loss": 0.8749,
+      "step": 118890
+    },
+    {
+      "epoch": 0.8606773943697655,
+      "grad_norm": 0.15046702325344086,
+      "learning_rate": 4.1393298442962936e-06,
+      "loss": 0.8697,
+      "step": 118900
+    },
+    {
+      "epoch": 0.8607497810303517,
+      "grad_norm": 0.15024615824222565,
+      "learning_rate": 4.139257457635707e-06,
+      "loss": 0.8616,
+      "step": 118910
+    },
+    {
+      "epoch": 0.8608221676909379,
+      "grad_norm": 0.16122505068778992,
+      "learning_rate": 4.139185070975121e-06,
+      "loss": 0.8701,
+      "step": 118920
+    },
+    {
+      "epoch": 0.8608945543515241,
+      "grad_norm": 0.16019709408283234,
+      "learning_rate": 4.139112684314534e-06,
+      "loss": 0.8733,
+      "step": 118930
+    },
+    {
+      "epoch": 0.8609669410121102,
+      "grad_norm": 0.21425500512123108,
+      "learning_rate": 4.139040297653948e-06,
+      "loss": 0.8828,
+      "step": 118940
+    },
+    {
+      "epoch": 0.8610393276726965,
+      "grad_norm": 0.15902172029018402,
+      "learning_rate": 4.1389679109933625e-06,
+      "loss": 0.8829,
+      "step": 118950
+    },
+    {
+      "epoch": 0.8611117143332827,
+      "grad_norm": 0.15424564480781555,
+      "learning_rate": 4.138895524332776e-06,
+      "loss": 0.8682,
+      "step": 118960
+    },
+    {
+      "epoch": 0.8611841009938689,
+      "grad_norm": 0.14803998172283173,
+      "learning_rate": 4.13882313767219e-06,
+      "loss": 0.8536,
+      "step": 118970
+    },
+    {
+      "epoch": 0.8612564876544551,
+      "grad_norm": 0.17855031788349152,
+      "learning_rate": 4.138750751011603e-06,
+      "loss": 0.8756,
+      "step": 118980
+    },
+    {
+      "epoch": 0.8613288743150412,
+      "grad_norm": 0.19807036221027374,
+      "learning_rate": 4.138678364351018e-06,
+      "loss": 0.8816,
+      "step": 118990
+    },
+    {
+      "epoch": 0.8614012609756274,
+      "grad_norm": 0.15931545197963715,
+      "learning_rate": 4.138605977690431e-06,
+      "loss": 0.8715,
+      "step": 119000
+    },
+    {
+      "epoch": 0.8614736476362136,
+      "grad_norm": 0.14279036223888397,
+      "learning_rate": 4.138533591029845e-06,
+      "loss": 0.8736,
+      "step": 119010
+    },
+    {
+      "epoch": 0.8615460342967998,
+      "grad_norm": 0.14687687158584595,
+      "learning_rate": 4.138461204369259e-06,
+      "loss": 0.8761,
+      "step": 119020
+    },
+    {
+      "epoch": 0.8616184209573859,
+      "grad_norm": 0.15899838507175446,
+      "learning_rate": 4.138388817708673e-06,
+      "loss": 0.8743,
+      "step": 119030
+    },
+    {
+      "epoch": 0.8616908076179721,
+      "grad_norm": 0.18828865885734558,
+      "learning_rate": 4.138316431048087e-06,
+      "loss": 0.8716,
+      "step": 119040
+    },
+    {
+      "epoch": 0.8617631942785583,
+      "grad_norm": 0.17458096146583557,
+      "learning_rate": 4.1382440443875e-06,
+      "loss": 0.8581,
+      "step": 119050
+    },
+    {
+      "epoch": 0.8618355809391446,
+      "grad_norm": 0.1492881029844284,
+      "learning_rate": 4.138171657726914e-06,
+      "loss": 0.8767,
+      "step": 119060
+    },
+    {
+      "epoch": 0.8619079675997308,
+      "grad_norm": 0.1459776908159256,
+      "learning_rate": 4.1380992710663284e-06,
+      "loss": 0.8777,
+      "step": 119070
+    },
+    {
+      "epoch": 0.8619803542603169,
+      "grad_norm": 0.15040789544582367,
+      "learning_rate": 4.138026884405742e-06,
+      "loss": 0.8845,
+      "step": 119080
+    },
+    {
+      "epoch": 0.8620527409209031,
+      "grad_norm": 0.14359919726848602,
+      "learning_rate": 4.137954497745156e-06,
+      "loss": 0.8878,
+      "step": 119090
+    },
+    {
+      "epoch": 0.8621251275814893,
+      "grad_norm": 0.1604946255683899,
+      "learning_rate": 4.137882111084569e-06,
+      "loss": 0.875,
+      "step": 119100
+    },
+    {
+      "epoch": 0.8621975142420755,
+      "grad_norm": 0.15070165693759918,
+      "learning_rate": 4.137809724423984e-06,
+      "loss": 0.8829,
+      "step": 119110
+    },
+    {
+      "epoch": 0.8622699009026616,
+      "grad_norm": 0.15339529514312744,
+      "learning_rate": 4.137737337763397e-06,
+      "loss": 0.881,
+      "step": 119120
+    },
+    {
+      "epoch": 0.8623422875632478,
+      "grad_norm": 0.15151731669902802,
+      "learning_rate": 4.137664951102811e-06,
+      "loss": 0.8773,
+      "step": 119130
+    },
+    {
+      "epoch": 0.862414674223834,
+      "grad_norm": 0.17294517159461975,
+      "learning_rate": 4.137592564442225e-06,
+      "loss": 0.8631,
+      "step": 119140
+    },
+    {
+      "epoch": 0.8624870608844202,
+      "grad_norm": 0.1720060110092163,
+      "learning_rate": 4.137520177781639e-06,
+      "loss": 0.8931,
+      "step": 119150
+    },
+    {
+      "epoch": 0.8625594475450064,
+      "grad_norm": 0.15391665697097778,
+      "learning_rate": 4.137447791121053e-06,
+      "loss": 0.8624,
+      "step": 119160
+    },
+    {
+      "epoch": 0.8626318342055926,
+      "grad_norm": 0.14894205331802368,
+      "learning_rate": 4.137375404460466e-06,
+      "loss": 0.8758,
+      "step": 119170
+    },
+    {
+      "epoch": 0.8627042208661788,
+      "grad_norm": 0.16286589205265045,
+      "learning_rate": 4.13730301779988e-06,
+      "loss": 0.8836,
+      "step": 119180
+    },
+    {
+      "epoch": 0.862776607526765,
+      "grad_norm": 0.1608862280845642,
+      "learning_rate": 4.137230631139294e-06,
+      "loss": 0.8801,
+      "step": 119190
+    },
+    {
+      "epoch": 0.8628489941873512,
+      "grad_norm": 0.1520557999610901,
+      "learning_rate": 4.137158244478708e-06,
+      "loss": 0.8725,
+      "step": 119200
+    },
+    {
+      "epoch": 0.8629213808479373,
+      "grad_norm": 0.17940600216388702,
+      "learning_rate": 4.137085857818122e-06,
+      "loss": 0.8713,
+      "step": 119210
+    },
+    {
+      "epoch": 0.8629937675085235,
+      "grad_norm": 0.14350375533103943,
+      "learning_rate": 4.137013471157535e-06,
+      "loss": 0.8842,
+      "step": 119220
+    },
+    {
+      "epoch": 0.8630661541691097,
+      "grad_norm": 0.15040133893489838,
+      "learning_rate": 4.13694108449695e-06,
+      "loss": 0.8723,
+      "step": 119230
+    },
+    {
+      "epoch": 0.8631385408296959,
+      "grad_norm": 0.16370560228824615,
+      "learning_rate": 4.136868697836363e-06,
+      "loss": 0.8713,
+      "step": 119240
+    },
+    {
+      "epoch": 0.863210927490282,
+      "grad_norm": 0.16433537006378174,
+      "learning_rate": 4.136796311175777e-06,
+      "loss": 0.8691,
+      "step": 119250
+    },
+    {
+      "epoch": 0.8632833141508682,
+      "grad_norm": 0.15914808213710785,
+      "learning_rate": 4.1367239245151905e-06,
+      "loss": 0.8863,
+      "step": 119260
+    },
+    {
+      "epoch": 0.8633557008114545,
+      "grad_norm": 0.16615451872348785,
+      "learning_rate": 4.136651537854605e-06,
+      "loss": 0.8857,
+      "step": 119270
+    },
+    {
+      "epoch": 0.8634280874720407,
+      "grad_norm": 0.157662034034729,
+      "learning_rate": 4.136579151194019e-06,
+      "loss": 0.887,
+      "step": 119280
+    },
+    {
+      "epoch": 0.8635004741326269,
+      "grad_norm": 0.5979766249656677,
+      "learning_rate": 4.136506764533432e-06,
+      "loss": 0.8737,
+      "step": 119290
+    },
+    {
+      "epoch": 0.863572860793213,
+      "grad_norm": 0.15033407509326935,
+      "learning_rate": 4.136434377872846e-06,
+      "loss": 0.8869,
+      "step": 119300
+    },
+    {
+      "epoch": 0.8636452474537992,
+      "grad_norm": 0.1554785966873169,
+      "learning_rate": 4.13636199121226e-06,
+      "loss": 0.8854,
+      "step": 119310
+    },
+    {
+      "epoch": 0.8637176341143854,
+      "grad_norm": 0.17543356120586395,
+      "learning_rate": 4.136289604551674e-06,
+      "loss": 0.8879,
+      "step": 119320
+    },
+    {
+      "epoch": 0.8637900207749716,
+      "grad_norm": 0.14748305082321167,
+      "learning_rate": 4.1362172178910875e-06,
+      "loss": 0.8729,
+      "step": 119330
+    },
+    {
+      "epoch": 0.8638624074355578,
+      "grad_norm": 0.15673819184303284,
+      "learning_rate": 4.136144831230501e-06,
+      "loss": 0.8585,
+      "step": 119340
+    },
+    {
+      "epoch": 0.8639347940961439,
+      "grad_norm": 0.1523292362689972,
+      "learning_rate": 4.136072444569916e-06,
+      "loss": 0.8832,
+      "step": 119350
+    },
+    {
+      "epoch": 0.8640071807567301,
+      "grad_norm": 0.2544263005256653,
+      "learning_rate": 4.136000057909329e-06,
+      "loss": 0.8712,
+      "step": 119360
+    },
+    {
+      "epoch": 0.8640795674173163,
+      "grad_norm": 0.15314503014087677,
+      "learning_rate": 4.135927671248743e-06,
+      "loss": 0.8739,
+      "step": 119370
+    },
+    {
+      "epoch": 0.8641519540779026,
+      "grad_norm": 0.15046413242816925,
+      "learning_rate": 4.1358552845881565e-06,
+      "loss": 0.8834,
+      "step": 119380
+    },
+    {
+      "epoch": 0.8642243407384887,
+      "grad_norm": 0.4037892520427704,
+      "learning_rate": 4.13578289792757e-06,
+      "loss": 0.8753,
+      "step": 119390
+    },
+    {
+      "epoch": 0.8642967273990749,
+      "grad_norm": 0.15684808790683746,
+      "learning_rate": 4.1357105112669845e-06,
+      "loss": 0.8746,
+      "step": 119400
+    },
+    {
+      "epoch": 0.8643691140596611,
+      "grad_norm": 0.15342099964618683,
+      "learning_rate": 4.135638124606398e-06,
+      "loss": 0.8746,
+      "step": 119410
+    },
+    {
+      "epoch": 0.8644415007202473,
+      "grad_norm": 0.15408830344676971,
+      "learning_rate": 4.135565737945812e-06,
+      "loss": 0.8748,
+      "step": 119420
+    },
+    {
+      "epoch": 0.8645138873808335,
+      "grad_norm": 0.16404138505458832,
+      "learning_rate": 4.135493351285225e-06,
+      "loss": 0.8657,
+      "step": 119430
+    },
+    {
+      "epoch": 0.8645862740414196,
+      "grad_norm": 0.14688260853290558,
+      "learning_rate": 4.135420964624639e-06,
+      "loss": 0.8836,
+      "step": 119440
+    },
+    {
+      "epoch": 0.8646586607020058,
+      "grad_norm": 0.16032230854034424,
+      "learning_rate": 4.135348577964053e-06,
+      "loss": 0.8735,
+      "step": 119450
+    },
+    {
+      "epoch": 0.864731047362592,
+      "grad_norm": 0.14813324809074402,
+      "learning_rate": 4.135276191303467e-06,
+      "loss": 0.8774,
+      "step": 119460
+    },
+    {
+      "epoch": 0.8648034340231782,
+      "grad_norm": 0.1528783142566681,
+      "learning_rate": 4.135203804642881e-06,
+      "loss": 0.8737,
+      "step": 119470
+    },
+    {
+      "epoch": 0.8648758206837645,
+      "grad_norm": 0.15664593875408173,
+      "learning_rate": 4.135131417982294e-06,
+      "loss": 0.88,
+      "step": 119480
+    },
+    {
+      "epoch": 0.8649482073443506,
+      "grad_norm": 0.1477356255054474,
+      "learning_rate": 4.135059031321708e-06,
+      "loss": 0.8672,
+      "step": 119490
+    },
+    {
+      "epoch": 0.8650205940049368,
+      "grad_norm": 0.1740371733903885,
+      "learning_rate": 4.134986644661122e-06,
+      "loss": 0.8781,
+      "step": 119500
+    },
+    {
+      "epoch": 0.865092980665523,
+      "grad_norm": 0.15743084251880646,
+      "learning_rate": 4.134914258000536e-06,
+      "loss": 0.8829,
+      "step": 119510
+    },
+    {
+      "epoch": 0.8651653673261092,
+      "grad_norm": 0.14371630549430847,
+      "learning_rate": 4.13484187133995e-06,
+      "loss": 0.8789,
+      "step": 119520
+    },
+    {
+      "epoch": 0.8652377539866953,
+      "grad_norm": 0.1605149209499359,
+      "learning_rate": 4.134769484679363e-06,
+      "loss": 0.8863,
+      "step": 119530
+    },
+    {
+      "epoch": 0.8653101406472815,
+      "grad_norm": 0.16169756650924683,
+      "learning_rate": 4.134697098018777e-06,
+      "loss": 0.8839,
+      "step": 119540
+    },
+    {
+      "epoch": 0.8653825273078677,
+      "grad_norm": 0.15459372103214264,
+      "learning_rate": 4.134624711358191e-06,
+      "loss": 0.8706,
+      "step": 119550
+    },
+    {
+      "epoch": 0.8654549139684539,
+      "grad_norm": 0.2985534369945526,
+      "learning_rate": 4.134552324697605e-06,
+      "loss": 0.8732,
+      "step": 119560
+    },
+    {
+      "epoch": 0.86552730062904,
+      "grad_norm": 0.1482762098312378,
+      "learning_rate": 4.1344799380370186e-06,
+      "loss": 0.8806,
+      "step": 119570
+    },
+    {
+      "epoch": 0.8655996872896262,
+      "grad_norm": 0.15104518830776215,
+      "learning_rate": 4.134407551376432e-06,
+      "loss": 0.884,
+      "step": 119580
+    },
+    {
+      "epoch": 0.8656720739502125,
+      "grad_norm": 0.15297864377498627,
+      "learning_rate": 4.134335164715847e-06,
+      "loss": 0.8779,
+      "step": 119590
+    },
+    {
+      "epoch": 0.8657444606107987,
+      "grad_norm": 0.1583460420370102,
+      "learning_rate": 4.13426277805526e-06,
+      "loss": 0.8681,
+      "step": 119600
+    },
+    {
+      "epoch": 0.8658168472713849,
+      "grad_norm": 0.15396851301193237,
+      "learning_rate": 4.134190391394674e-06,
+      "loss": 0.8802,
+      "step": 119610
+    },
+    {
+      "epoch": 0.865889233931971,
+      "grad_norm": 0.16072429716587067,
+      "learning_rate": 4.1341180047340875e-06,
+      "loss": 0.8759,
+      "step": 119620
+    },
+    {
+      "epoch": 0.8659616205925572,
+      "grad_norm": 0.14890296757221222,
+      "learning_rate": 4.134045618073502e-06,
+      "loss": 0.8725,
+      "step": 119630
+    },
+    {
+      "epoch": 0.8660340072531434,
+      "grad_norm": 0.1567048579454422,
+      "learning_rate": 4.1339732314129156e-06,
+      "loss": 0.8839,
+      "step": 119640
+    },
+    {
+      "epoch": 0.8661063939137296,
+      "grad_norm": 0.15268154442310333,
+      "learning_rate": 4.133900844752329e-06,
+      "loss": 0.8609,
+      "step": 119650
+    },
+    {
+      "epoch": 0.8661787805743157,
+      "grad_norm": 0.15023602545261383,
+      "learning_rate": 4.133828458091743e-06,
+      "loss": 0.8758,
+      "step": 119660
+    },
+    {
+      "epoch": 0.8662511672349019,
+      "grad_norm": 0.15714387595653534,
+      "learning_rate": 4.133756071431157e-06,
+      "loss": 0.873,
+      "step": 119670
+    },
+    {
+      "epoch": 0.8663235538954881,
+      "grad_norm": 0.15032929182052612,
+      "learning_rate": 4.133683684770571e-06,
+      "loss": 0.8631,
+      "step": 119680
+    },
+    {
+      "epoch": 0.8663959405560743,
+      "grad_norm": 0.14457829296588898,
+      "learning_rate": 4.1336112981099845e-06,
+      "loss": 0.8716,
+      "step": 119690
+    },
+    {
+      "epoch": 0.8664683272166606,
+      "grad_norm": 0.17244747281074524,
+      "learning_rate": 4.133538911449398e-06,
+      "loss": 0.8842,
+      "step": 119700
+    },
+    {
+      "epoch": 0.8665407138772467,
+      "grad_norm": 0.15346111357212067,
+      "learning_rate": 4.1334665247888126e-06,
+      "loss": 0.884,
+      "step": 119710
+    },
+    {
+      "epoch": 0.8666131005378329,
+      "grad_norm": 0.1665041148662567,
+      "learning_rate": 4.133394138128226e-06,
+      "loss": 0.8807,
+      "step": 119720
+    },
+    {
+      "epoch": 0.8666854871984191,
+      "grad_norm": 0.1579190045595169,
+      "learning_rate": 4.13332175146764e-06,
+      "loss": 0.882,
+      "step": 119730
+    },
+    {
+      "epoch": 0.8667578738590053,
+      "grad_norm": 1.4202749729156494,
+      "learning_rate": 4.133249364807053e-06,
+      "loss": 0.8748,
+      "step": 119740
+    },
+    {
+      "epoch": 0.8668302605195914,
+      "grad_norm": 0.1646961271762848,
+      "learning_rate": 4.133176978146468e-06,
+      "loss": 0.8626,
+      "step": 119750
+    },
+    {
+      "epoch": 0.8669026471801776,
+      "grad_norm": 0.1607542634010315,
+      "learning_rate": 4.1331045914858815e-06,
+      "loss": 0.8768,
+      "step": 119760
+    },
+    {
+      "epoch": 0.8669750338407638,
+      "grad_norm": 0.1496744155883789,
+      "learning_rate": 4.133032204825295e-06,
+      "loss": 0.8702,
+      "step": 119770
+    },
+    {
+      "epoch": 0.86704742050135,
+      "grad_norm": 0.14871948957443237,
+      "learning_rate": 4.132959818164709e-06,
+      "loss": 0.8846,
+      "step": 119780
+    },
+    {
+      "epoch": 0.8671198071619362,
+      "grad_norm": 0.14933007955551147,
+      "learning_rate": 4.132887431504123e-06,
+      "loss": 0.8641,
+      "step": 119790
+    },
+    {
+      "epoch": 0.8671921938225224,
+      "grad_norm": 0.16172604262828827,
+      "learning_rate": 4.132815044843537e-06,
+      "loss": 0.8764,
+      "step": 119800
+    },
+    {
+      "epoch": 0.8672645804831086,
+      "grad_norm": 0.1538577377796173,
+      "learning_rate": 4.1327426581829504e-06,
+      "loss": 0.8734,
+      "step": 119810
+    },
+    {
+      "epoch": 0.8673369671436948,
+      "grad_norm": 0.14605380594730377,
+      "learning_rate": 4.132670271522364e-06,
+      "loss": 0.8692,
+      "step": 119820
+    },
+    {
+      "epoch": 0.867409353804281,
+      "grad_norm": 0.16161808371543884,
+      "learning_rate": 4.1325978848617785e-06,
+      "loss": 0.8729,
+      "step": 119830
+    },
+    {
+      "epoch": 0.8674817404648671,
+      "grad_norm": 0.2714543044567108,
+      "learning_rate": 4.132525498201192e-06,
+      "loss": 0.8787,
+      "step": 119840
+    },
+    {
+      "epoch": 0.8675541271254533,
+      "grad_norm": 0.1541425883769989,
+      "learning_rate": 4.132453111540606e-06,
+      "loss": 0.853,
+      "step": 119850
+    },
+    {
+      "epoch": 0.8676265137860395,
+      "grad_norm": 0.15290531516075134,
+      "learning_rate": 4.132380724880019e-06,
+      "loss": 0.8684,
+      "step": 119860
+    },
+    {
+      "epoch": 0.8676989004466257,
+      "grad_norm": 0.1534222811460495,
+      "learning_rate": 4.132308338219434e-06,
+      "loss": 0.8673,
+      "step": 119870
+    },
+    {
+      "epoch": 0.8677712871072119,
+      "grad_norm": 0.16768594086170197,
+      "learning_rate": 4.1322359515588474e-06,
+      "loss": 0.8653,
+      "step": 119880
+    },
+    {
+      "epoch": 0.867843673767798,
+      "grad_norm": 0.1455199420452118,
+      "learning_rate": 4.132163564898261e-06,
+      "loss": 0.8762,
+      "step": 119890
+    },
+    {
+      "epoch": 0.8679160604283842,
+      "grad_norm": 0.15997451543807983,
+      "learning_rate": 4.132091178237675e-06,
+      "loss": 0.884,
+      "step": 119900
+    },
+    {
+      "epoch": 0.8679884470889705,
+      "grad_norm": 0.15855015814304352,
+      "learning_rate": 4.132018791577089e-06,
+      "loss": 0.8886,
+      "step": 119910
+    },
+    {
+      "epoch": 0.8680608337495567,
+      "grad_norm": 0.1565515547990799,
+      "learning_rate": 4.131946404916503e-06,
+      "loss": 0.8721,
+      "step": 119920
+    },
+    {
+      "epoch": 0.8681332204101428,
+      "grad_norm": 0.1573457270860672,
+      "learning_rate": 4.131874018255916e-06,
+      "loss": 0.8682,
+      "step": 119930
+    },
+    {
+      "epoch": 0.868205607070729,
+      "grad_norm": 0.1630558967590332,
+      "learning_rate": 4.13180163159533e-06,
+      "loss": 0.8654,
+      "step": 119940
+    },
+    {
+      "epoch": 0.8682779937313152,
+      "grad_norm": 0.1974228322505951,
+      "learning_rate": 4.1317292449347444e-06,
+      "loss": 0.8581,
+      "step": 119950
+    },
+    {
+      "epoch": 0.8683503803919014,
+      "grad_norm": 0.14881473779678345,
+      "learning_rate": 4.131656858274157e-06,
+      "loss": 0.8827,
+      "step": 119960
+    },
+    {
+      "epoch": 0.8684227670524876,
+      "grad_norm": 0.16378778219223022,
+      "learning_rate": 4.131584471613571e-06,
+      "loss": 0.8598,
+      "step": 119970
+    },
+    {
+      "epoch": 0.8684951537130737,
+      "grad_norm": 0.1859802007675171,
+      "learning_rate": 4.131512084952985e-06,
+      "loss": 0.8768,
+      "step": 119980
+    },
+    {
+      "epoch": 0.8685675403736599,
+      "grad_norm": 0.16006912291049957,
+      "learning_rate": 4.131439698292399e-06,
+      "loss": 0.8718,
+      "step": 119990
+    },
+    {
+      "epoch": 0.8686399270342461,
+      "grad_norm": 0.25071021914482117,
+      "learning_rate": 4.1313673116318125e-06,
+      "loss": 0.8708,
+      "step": 120000
+    },
+    {
+      "epoch": 0.8687123136948323,
+      "grad_norm": 0.15014329552650452,
+      "learning_rate": 4.131294924971226e-06,
+      "loss": 0.8719,
+      "step": 120010
+    },
+    {
+      "epoch": 0.8687847003554185,
+      "grad_norm": 0.1454050987958908,
+      "learning_rate": 4.131222538310641e-06,
+      "loss": 0.87,
+      "step": 120020
+    },
+    {
+      "epoch": 0.8688570870160047,
+      "grad_norm": 0.1666693240404129,
+      "learning_rate": 4.131150151650054e-06,
+      "loss": 0.8761,
+      "step": 120030
+    },
+    {
+      "epoch": 0.8689294736765909,
+      "grad_norm": 0.16259945929050446,
+      "learning_rate": 4.131077764989468e-06,
+      "loss": 0.8679,
+      "step": 120040
+    },
+    {
+      "epoch": 0.8690018603371771,
+      "grad_norm": 0.16060245037078857,
+      "learning_rate": 4.1310053783288815e-06,
+      "loss": 0.872,
+      "step": 120050
+    },
+    {
+      "epoch": 0.8690742469977633,
+      "grad_norm": 0.3217458724975586,
+      "learning_rate": 4.130932991668296e-06,
+      "loss": 0.8638,
+      "step": 120060
+    },
+    {
+      "epoch": 0.8691466336583494,
+      "grad_norm": 0.14817246794700623,
+      "learning_rate": 4.1308606050077095e-06,
+      "loss": 0.8703,
+      "step": 120070
+    },
+    {
+      "epoch": 0.8692190203189356,
+      "grad_norm": 0.1485157608985901,
+      "learning_rate": 4.130788218347123e-06,
+      "loss": 0.8652,
+      "step": 120080
+    },
+    {
+      "epoch": 0.8692914069795218,
+      "grad_norm": 0.16461946070194244,
+      "learning_rate": 4.130715831686537e-06,
+      "loss": 0.858,
+      "step": 120090
+    },
+    {
+      "epoch": 0.869363793640108,
+      "grad_norm": 0.15408070385456085,
+      "learning_rate": 4.130643445025951e-06,
+      "loss": 0.8723,
+      "step": 120100
+    },
+    {
+      "epoch": 0.8694361803006941,
+      "grad_norm": 0.15531690418720245,
+      "learning_rate": 4.130571058365365e-06,
+      "loss": 0.8761,
+      "step": 120110
+    },
+    {
+      "epoch": 0.8695085669612804,
+      "grad_norm": 0.15016768872737885,
+      "learning_rate": 4.1304986717047785e-06,
+      "loss": 0.8732,
+      "step": 120120
+    },
+    {
+      "epoch": 0.8695809536218666,
+      "grad_norm": 0.1652849316596985,
+      "learning_rate": 4.130426285044192e-06,
+      "loss": 0.8677,
+      "step": 120130
+    },
+    {
+      "epoch": 0.8696533402824528,
+      "grad_norm": 0.15715721249580383,
+      "learning_rate": 4.1303538983836065e-06,
+      "loss": 0.8773,
+      "step": 120140
+    },
+    {
+      "epoch": 0.869725726943039,
+      "grad_norm": 0.2709084749221802,
+      "learning_rate": 4.13028151172302e-06,
+      "loss": 0.8724,
+      "step": 120150
+    },
+    {
+      "epoch": 0.8697981136036251,
+      "grad_norm": 0.147217258810997,
+      "learning_rate": 4.130209125062434e-06,
+      "loss": 0.8757,
+      "step": 120160
+    },
+    {
+      "epoch": 0.8698705002642113,
+      "grad_norm": 0.15031427145004272,
+      "learning_rate": 4.130136738401847e-06,
+      "loss": 0.8757,
+      "step": 120170
+    },
+    {
+      "epoch": 0.8699428869247975,
+      "grad_norm": 0.1699194461107254,
+      "learning_rate": 4.130064351741261e-06,
+      "loss": 0.8655,
+      "step": 120180
+    },
+    {
+      "epoch": 0.8700152735853837,
+      "grad_norm": 0.17292308807373047,
+      "learning_rate": 4.1299919650806755e-06,
+      "loss": 0.8879,
+      "step": 120190
+    },
+    {
+      "epoch": 0.8700876602459698,
+      "grad_norm": 0.15235485136508942,
+      "learning_rate": 4.129919578420089e-06,
+      "loss": 0.8707,
+      "step": 120200
+    },
+    {
+      "epoch": 0.870160046906556,
+      "grad_norm": 0.15500420331954956,
+      "learning_rate": 4.129847191759503e-06,
+      "loss": 0.8743,
+      "step": 120210
+    },
+    {
+      "epoch": 0.8702324335671422,
+      "grad_norm": 0.1642414629459381,
+      "learning_rate": 4.129774805098916e-06,
+      "loss": 0.8661,
+      "step": 120220
+    },
+    {
+      "epoch": 0.8703048202277285,
+      "grad_norm": 0.16339725255966187,
+      "learning_rate": 4.129702418438331e-06,
+      "loss": 0.8755,
+      "step": 120230
+    },
+    {
+      "epoch": 0.8703772068883147,
+      "grad_norm": 0.1617591381072998,
+      "learning_rate": 4.129630031777744e-06,
+      "loss": 0.8659,
+      "step": 120240
+    },
+    {
+      "epoch": 0.8704495935489008,
+      "grad_norm": 0.14315065741539001,
+      "learning_rate": 4.129557645117158e-06,
+      "loss": 0.8774,
+      "step": 120250
+    },
+    {
+      "epoch": 0.870521980209487,
+      "grad_norm": 0.16367150843143463,
+      "learning_rate": 4.129485258456572e-06,
+      "loss": 0.8774,
+      "step": 120260
+    },
+    {
+      "epoch": 0.8705943668700732,
+      "grad_norm": 0.1460050344467163,
+      "learning_rate": 4.129412871795986e-06,
+      "loss": 0.8779,
+      "step": 120270
+    },
+    {
+      "epoch": 0.8706667535306594,
+      "grad_norm": 0.17741134762763977,
+      "learning_rate": 4.1293404851354e-06,
+      "loss": 0.8806,
+      "step": 120280
+    },
+    {
+      "epoch": 0.8707391401912455,
+      "grad_norm": 0.1527169644832611,
+      "learning_rate": 4.129268098474813e-06,
+      "loss": 0.8783,
+      "step": 120290
+    },
+    {
+      "epoch": 0.8708115268518317,
+      "grad_norm": 0.1656719595193863,
+      "learning_rate": 4.129195711814227e-06,
+      "loss": 0.874,
+      "step": 120300
+    },
+    {
+      "epoch": 0.8708839135124179,
+      "grad_norm": 0.15612590312957764,
+      "learning_rate": 4.129123325153641e-06,
+      "loss": 0.8766,
+      "step": 120310
+    },
+    {
+      "epoch": 0.8709563001730041,
+      "grad_norm": 0.182773619890213,
+      "learning_rate": 4.129050938493055e-06,
+      "loss": 0.863,
+      "step": 120320
+    },
+    {
+      "epoch": 0.8710286868335904,
+      "grad_norm": 0.1543162316083908,
+      "learning_rate": 4.128978551832469e-06,
+      "loss": 0.8738,
+      "step": 120330
+    },
+    {
+      "epoch": 0.8711010734941765,
+      "grad_norm": 0.18090280890464783,
+      "learning_rate": 4.128906165171882e-06,
+      "loss": 0.8703,
+      "step": 120340
+    },
+    {
+      "epoch": 0.8711734601547627,
+      "grad_norm": 0.15451836585998535,
+      "learning_rate": 4.128833778511297e-06,
+      "loss": 0.8816,
+      "step": 120350
+    },
+    {
+      "epoch": 0.8712458468153489,
+      "grad_norm": 0.146149680018425,
+      "learning_rate": 4.12876139185071e-06,
+      "loss": 0.8759,
+      "step": 120360
+    },
+    {
+      "epoch": 0.8713182334759351,
+      "grad_norm": 0.14919118583202362,
+      "learning_rate": 4.128689005190124e-06,
+      "loss": 0.8645,
+      "step": 120370
+    },
+    {
+      "epoch": 0.8713906201365212,
+      "grad_norm": 0.6079381704330444,
+      "learning_rate": 4.1286166185295376e-06,
+      "loss": 0.8763,
+      "step": 120380
+    },
+    {
+      "epoch": 0.8714630067971074,
+      "grad_norm": 0.162376269698143,
+      "learning_rate": 4.128544231868952e-06,
+      "loss": 0.8675,
+      "step": 120390
+    },
+    {
+      "epoch": 0.8715353934576936,
+      "grad_norm": 0.16012080013751984,
+      "learning_rate": 4.128471845208366e-06,
+      "loss": 0.874,
+      "step": 120400
+    },
+    {
+      "epoch": 0.8716077801182798,
+      "grad_norm": 0.16266874969005585,
+      "learning_rate": 4.128399458547779e-06,
+      "loss": 0.868,
+      "step": 120410
+    },
+    {
+      "epoch": 0.871680166778866,
+      "grad_norm": 0.15482492744922638,
+      "learning_rate": 4.128327071887193e-06,
+      "loss": 0.8837,
+      "step": 120420
+    },
+    {
+      "epoch": 0.8717525534394521,
+      "grad_norm": 0.15119405090808868,
+      "learning_rate": 4.128254685226607e-06,
+      "loss": 0.8738,
+      "step": 120430
+    },
+    {
+      "epoch": 0.8718249401000384,
+      "grad_norm": 0.1444867104291916,
+      "learning_rate": 4.128182298566021e-06,
+      "loss": 0.8617,
+      "step": 120440
+    },
+    {
+      "epoch": 0.8718973267606246,
+      "grad_norm": 0.15251514315605164,
+      "learning_rate": 4.1281099119054346e-06,
+      "loss": 0.8706,
+      "step": 120450
+    },
+    {
+      "epoch": 0.8719697134212108,
+      "grad_norm": 0.16033178567886353,
+      "learning_rate": 4.128037525244848e-06,
+      "loss": 0.8756,
+      "step": 120460
+    },
+    {
+      "epoch": 0.872042100081797,
+      "grad_norm": 0.1551380306482315,
+      "learning_rate": 4.127965138584263e-06,
+      "loss": 0.8642,
+      "step": 120470
+    },
+    {
+      "epoch": 0.8721144867423831,
+      "grad_norm": 0.14764218032360077,
+      "learning_rate": 4.127892751923676e-06,
+      "loss": 0.8646,
+      "step": 120480
+    },
+    {
+      "epoch": 0.8721868734029693,
+      "grad_norm": 0.14466458559036255,
+      "learning_rate": 4.127820365263089e-06,
+      "loss": 0.8727,
+      "step": 120490
+    },
+    {
+      "epoch": 0.8722592600635555,
+      "grad_norm": 0.15080726146697998,
+      "learning_rate": 4.1277479786025035e-06,
+      "loss": 0.8934,
+      "step": 120500
+    },
+    {
+      "epoch": 0.8723316467241417,
+      "grad_norm": 0.16469760239124298,
+      "learning_rate": 4.127675591941917e-06,
+      "loss": 0.8752,
+      "step": 120510
+    },
+    {
+      "epoch": 0.8724040333847278,
+      "grad_norm": 0.1857549250125885,
+      "learning_rate": 4.127603205281331e-06,
+      "loss": 0.866,
+      "step": 120520
+    },
+    {
+      "epoch": 0.872476420045314,
+      "grad_norm": 0.18460987508296967,
+      "learning_rate": 4.127530818620744e-06,
+      "loss": 0.8576,
+      "step": 120530
+    },
+    {
+      "epoch": 0.8725488067059002,
+      "grad_norm": 0.15353545546531677,
+      "learning_rate": 4.127458431960159e-06,
+      "loss": 0.8828,
+      "step": 120540
+    },
+    {
+      "epoch": 0.8726211933664865,
+      "grad_norm": 0.14069363474845886,
+      "learning_rate": 4.1273860452995724e-06,
+      "loss": 0.868,
+      "step": 120550
+    },
+    {
+      "epoch": 0.8726935800270726,
+      "grad_norm": 0.1678398996591568,
+      "learning_rate": 4.127313658638986e-06,
+      "loss": 0.8802,
+      "step": 120560
+    },
+    {
+      "epoch": 0.8727659666876588,
+      "grad_norm": 0.15249879658222198,
+      "learning_rate": 4.1272412719784e-06,
+      "loss": 0.8821,
+      "step": 120570
+    },
+    {
+      "epoch": 0.872838353348245,
+      "grad_norm": 0.154356449842453,
+      "learning_rate": 4.127168885317814e-06,
+      "loss": 0.8736,
+      "step": 120580
+    },
+    {
+      "epoch": 0.8729107400088312,
+      "grad_norm": 0.16177409887313843,
+      "learning_rate": 4.127096498657228e-06,
+      "loss": 0.8793,
+      "step": 120590
+    },
+    {
+      "epoch": 0.8729831266694174,
+      "grad_norm": 0.14435283839702606,
+      "learning_rate": 4.127024111996641e-06,
+      "loss": 0.8678,
+      "step": 120600
+    },
+    {
+      "epoch": 0.8730555133300035,
+      "grad_norm": 0.1692737489938736,
+      "learning_rate": 4.126951725336055e-06,
+      "loss": 0.8701,
+      "step": 120610
+    },
+    {
+      "epoch": 0.8731278999905897,
+      "grad_norm": 0.14820191264152527,
+      "learning_rate": 4.1268793386754694e-06,
+      "loss": 0.8674,
+      "step": 120620
+    },
+    {
+      "epoch": 0.8732002866511759,
+      "grad_norm": 0.15043896436691284,
+      "learning_rate": 4.126806952014883e-06,
+      "loss": 0.8733,
+      "step": 120630
+    },
+    {
+      "epoch": 0.8732726733117621,
+      "grad_norm": 0.1535714864730835,
+      "learning_rate": 4.126734565354297e-06,
+      "loss": 0.8787,
+      "step": 120640
+    },
+    {
+      "epoch": 0.8733450599723483,
+      "grad_norm": 0.1565728783607483,
+      "learning_rate": 4.12666217869371e-06,
+      "loss": 0.8652,
+      "step": 120650
+    },
+    {
+      "epoch": 0.8734174466329345,
+      "grad_norm": 0.15889082849025726,
+      "learning_rate": 4.126589792033125e-06,
+      "loss": 0.8727,
+      "step": 120660
+    },
+    {
+      "epoch": 0.8734898332935207,
+      "grad_norm": 0.15119129419326782,
+      "learning_rate": 4.126517405372538e-06,
+      "loss": 0.8723,
+      "step": 120670
+    },
+    {
+      "epoch": 0.8735622199541069,
+      "grad_norm": 0.14388339221477509,
+      "learning_rate": 4.126445018711952e-06,
+      "loss": 0.8686,
+      "step": 120680
+    },
+    {
+      "epoch": 0.873634606614693,
+      "grad_norm": 0.14522388577461243,
+      "learning_rate": 4.126372632051366e-06,
+      "loss": 0.8696,
+      "step": 120690
+    },
+    {
+      "epoch": 0.8737069932752792,
+      "grad_norm": 0.15283215045928955,
+      "learning_rate": 4.12630024539078e-06,
+      "loss": 0.8731,
+      "step": 120700
+    },
+    {
+      "epoch": 0.8737793799358654,
+      "grad_norm": 0.15034788846969604,
+      "learning_rate": 4.126227858730194e-06,
+      "loss": 0.8837,
+      "step": 120710
+    },
+    {
+      "epoch": 0.8738517665964516,
+      "grad_norm": 0.15683099627494812,
+      "learning_rate": 4.126155472069607e-06,
+      "loss": 0.8686,
+      "step": 120720
+    },
+    {
+      "epoch": 0.8739241532570378,
+      "grad_norm": 0.19069434702396393,
+      "learning_rate": 4.126083085409021e-06,
+      "loss": 0.8651,
+      "step": 120730
+    },
+    {
+      "epoch": 0.8739965399176239,
+      "grad_norm": 0.15311843156814575,
+      "learning_rate": 4.126010698748435e-06,
+      "loss": 0.8822,
+      "step": 120740
+    },
+    {
+      "epoch": 0.8740689265782101,
+      "grad_norm": 0.1613323837518692,
+      "learning_rate": 4.125938312087849e-06,
+      "loss": 0.8744,
+      "step": 120750
+    },
+    {
+      "epoch": 0.8741413132387964,
+      "grad_norm": 0.1653766632080078,
+      "learning_rate": 4.125865925427263e-06,
+      "loss": 0.8658,
+      "step": 120760
+    },
+    {
+      "epoch": 0.8742136998993826,
+      "grad_norm": 0.20106831192970276,
+      "learning_rate": 4.125793538766676e-06,
+      "loss": 0.867,
+      "step": 120770
+    },
+    {
+      "epoch": 0.8742860865599688,
+      "grad_norm": 0.1640590876340866,
+      "learning_rate": 4.12572115210609e-06,
+      "loss": 0.8731,
+      "step": 120780
+    },
+    {
+      "epoch": 0.8743584732205549,
+      "grad_norm": 0.16234862804412842,
+      "learning_rate": 4.125648765445504e-06,
+      "loss": 0.8798,
+      "step": 120790
+    },
+    {
+      "epoch": 0.8744308598811411,
+      "grad_norm": 0.16816940903663635,
+      "learning_rate": 4.125576378784918e-06,
+      "loss": 0.8721,
+      "step": 120800
+    },
+    {
+      "epoch": 0.8745032465417273,
+      "grad_norm": 0.15265819430351257,
+      "learning_rate": 4.1255039921243315e-06,
+      "loss": 0.8658,
+      "step": 120810
+    },
+    {
+      "epoch": 0.8745756332023135,
+      "grad_norm": 0.15596769750118256,
+      "learning_rate": 4.125431605463745e-06,
+      "loss": 0.8784,
+      "step": 120820
+    },
+    {
+      "epoch": 0.8746480198628996,
+      "grad_norm": 0.15130725502967834,
+      "learning_rate": 4.12535921880316e-06,
+      "loss": 0.8847,
+      "step": 120830
+    },
+    {
+      "epoch": 0.8747204065234858,
+      "grad_norm": 0.22041760385036469,
+      "learning_rate": 4.125286832142573e-06,
+      "loss": 0.8765,
+      "step": 120840
+    },
+    {
+      "epoch": 0.874792793184072,
+      "grad_norm": 0.14767880737781525,
+      "learning_rate": 4.125214445481987e-06,
+      "loss": 0.8817,
+      "step": 120850
+    },
+    {
+      "epoch": 0.8748651798446583,
+      "grad_norm": 0.21697768568992615,
+      "learning_rate": 4.1251420588214005e-06,
+      "loss": 0.8704,
+      "step": 120860
+    },
+    {
+      "epoch": 0.8749375665052445,
+      "grad_norm": 0.1474272608757019,
+      "learning_rate": 4.125069672160815e-06,
+      "loss": 0.8718,
+      "step": 120870
+    },
+    {
+      "epoch": 0.8750099531658306,
+      "grad_norm": 0.18004824221134186,
+      "learning_rate": 4.1249972855002285e-06,
+      "loss": 0.8832,
+      "step": 120880
+    },
+    {
+      "epoch": 0.8750823398264168,
+      "grad_norm": 0.1615532785654068,
+      "learning_rate": 4.124924898839642e-06,
+      "loss": 0.869,
+      "step": 120890
+    },
+    {
+      "epoch": 0.875154726487003,
+      "grad_norm": 0.14899741113185883,
+      "learning_rate": 4.124852512179056e-06,
+      "loss": 0.8693,
+      "step": 120900
+    },
+    {
+      "epoch": 0.8752271131475892,
+      "grad_norm": 0.16730321943759918,
+      "learning_rate": 4.12478012551847e-06,
+      "loss": 0.8847,
+      "step": 120910
+    },
+    {
+      "epoch": 0.8752994998081753,
+      "grad_norm": 0.15044957399368286,
+      "learning_rate": 4.124707738857884e-06,
+      "loss": 0.8684,
+      "step": 120920
+    },
+    {
+      "epoch": 0.8753718864687615,
+      "grad_norm": 0.17763184010982513,
+      "learning_rate": 4.1246353521972975e-06,
+      "loss": 0.8853,
+      "step": 120930
+    },
+    {
+      "epoch": 0.8754442731293477,
+      "grad_norm": 0.16393078863620758,
+      "learning_rate": 4.124562965536711e-06,
+      "loss": 0.8696,
+      "step": 120940
+    },
+    {
+      "epoch": 0.8755166597899339,
+      "grad_norm": 0.15573619306087494,
+      "learning_rate": 4.1244905788761255e-06,
+      "loss": 0.8609,
+      "step": 120950
+    },
+    {
+      "epoch": 0.87558904645052,
+      "grad_norm": 0.15625816583633423,
+      "learning_rate": 4.124418192215539e-06,
+      "loss": 0.8735,
+      "step": 120960
+    },
+    {
+      "epoch": 0.8756614331111063,
+      "grad_norm": 0.14690211415290833,
+      "learning_rate": 4.124345805554953e-06,
+      "loss": 0.8688,
+      "step": 120970
+    },
+    {
+      "epoch": 0.8757338197716925,
+      "grad_norm": 0.17025691270828247,
+      "learning_rate": 4.124273418894366e-06,
+      "loss": 0.8582,
+      "step": 120980
+    },
+    {
+      "epoch": 0.8758062064322787,
+      "grad_norm": 0.159259632229805,
+      "learning_rate": 4.124201032233781e-06,
+      "loss": 0.8892,
+      "step": 120990
+    },
+    {
+      "epoch": 0.8758785930928649,
+      "grad_norm": 0.16585783660411835,
+      "learning_rate": 4.1241286455731945e-06,
+      "loss": 0.8723,
+      "step": 121000
+    },
+    {
+      "epoch": 0.875950979753451,
+      "grad_norm": 0.15400756895542145,
+      "learning_rate": 4.124056258912608e-06,
+      "loss": 0.8656,
+      "step": 121010
+    },
+    {
+      "epoch": 0.8760233664140372,
+      "grad_norm": 0.16609486937522888,
+      "learning_rate": 4.123983872252022e-06,
+      "loss": 0.8726,
+      "step": 121020
+    },
+    {
+      "epoch": 0.8760957530746234,
+      "grad_norm": 0.16967037320137024,
+      "learning_rate": 4.123911485591435e-06,
+      "loss": 0.8606,
+      "step": 121030
+    },
+    {
+      "epoch": 0.8761681397352096,
+      "grad_norm": 0.1568365842103958,
+      "learning_rate": 4.123839098930849e-06,
+      "loss": 0.8669,
+      "step": 121040
+    },
+    {
+      "epoch": 0.8762405263957957,
+      "grad_norm": 0.16394557058811188,
+      "learning_rate": 4.1237667122702626e-06,
+      "loss": 0.8723,
+      "step": 121050
+    },
+    {
+      "epoch": 0.8763129130563819,
+      "grad_norm": 0.15548640489578247,
+      "learning_rate": 4.123694325609677e-06,
+      "loss": 0.8712,
+      "step": 121060
+    },
+    {
+      "epoch": 0.8763852997169681,
+      "grad_norm": 0.1536964625120163,
+      "learning_rate": 4.123621938949091e-06,
+      "loss": 0.8664,
+      "step": 121070
+    },
+    {
+      "epoch": 0.8764576863775544,
+      "grad_norm": 0.14496274292469025,
+      "learning_rate": 4.123549552288504e-06,
+      "loss": 0.8656,
+      "step": 121080
+    },
+    {
+      "epoch": 0.8765300730381406,
+      "grad_norm": 0.16302034258842468,
+      "learning_rate": 4.123477165627918e-06,
+      "loss": 0.8763,
+      "step": 121090
+    },
+    {
+      "epoch": 0.8766024596987267,
+      "grad_norm": 0.15400786697864532,
+      "learning_rate": 4.123404778967332e-06,
+      "loss": 0.8673,
+      "step": 121100
+    },
+    {
+      "epoch": 0.8766748463593129,
+      "grad_norm": 0.1585310846567154,
+      "learning_rate": 4.123332392306746e-06,
+      "loss": 0.874,
+      "step": 121110
+    },
+    {
+      "epoch": 0.8767472330198991,
+      "grad_norm": 0.1589457243680954,
+      "learning_rate": 4.1232600056461596e-06,
+      "loss": 0.8741,
+      "step": 121120
+    },
+    {
+      "epoch": 0.8768196196804853,
+      "grad_norm": 0.16599847376346588,
+      "learning_rate": 4.123187618985573e-06,
+      "loss": 0.8783,
+      "step": 121130
+    },
+    {
+      "epoch": 0.8768920063410715,
+      "grad_norm": 0.1981654316186905,
+      "learning_rate": 4.123115232324988e-06,
+      "loss": 0.8759,
+      "step": 121140
+    },
+    {
+      "epoch": 0.8769643930016576,
+      "grad_norm": 0.19641701877117157,
+      "learning_rate": 4.123042845664401e-06,
+      "loss": 0.8701,
+      "step": 121150
+    },
+    {
+      "epoch": 0.8770367796622438,
+      "grad_norm": 0.16379612684249878,
+      "learning_rate": 4.122970459003815e-06,
+      "loss": 0.8603,
+      "step": 121160
+    },
+    {
+      "epoch": 0.87710916632283,
+      "grad_norm": 0.16375406086444855,
+      "learning_rate": 4.1228980723432285e-06,
+      "loss": 0.8772,
+      "step": 121170
+    },
+    {
+      "epoch": 0.8771815529834163,
+      "grad_norm": 0.16766558587551117,
+      "learning_rate": 4.122825685682643e-06,
+      "loss": 0.887,
+      "step": 121180
+    },
+    {
+      "epoch": 0.8772539396440024,
+      "grad_norm": 0.14811073243618011,
+      "learning_rate": 4.1227532990220566e-06,
+      "loss": 0.8706,
+      "step": 121190
+    },
+    {
+      "epoch": 0.8773263263045886,
+      "grad_norm": 0.16084469854831696,
+      "learning_rate": 4.12268091236147e-06,
+      "loss": 0.873,
+      "step": 121200
+    },
+    {
+      "epoch": 0.8773987129651748,
+      "grad_norm": 0.17017702758312225,
+      "learning_rate": 4.122608525700884e-06,
+      "loss": 0.873,
+      "step": 121210
+    },
+    {
+      "epoch": 0.877471099625761,
+      "grad_norm": 0.1788460612297058,
+      "learning_rate": 4.122536139040298e-06,
+      "loss": 0.875,
+      "step": 121220
+    },
+    {
+      "epoch": 0.8775434862863472,
+      "grad_norm": 0.1578466147184372,
+      "learning_rate": 4.122463752379712e-06,
+      "loss": 0.8784,
+      "step": 121230
+    },
+    {
+      "epoch": 0.8776158729469333,
+      "grad_norm": 0.14853112399578094,
+      "learning_rate": 4.1223913657191255e-06,
+      "loss": 0.8869,
+      "step": 121240
+    },
+    {
+      "epoch": 0.8776882596075195,
+      "grad_norm": 0.15817275643348694,
+      "learning_rate": 4.122318979058539e-06,
+      "loss": 0.871,
+      "step": 121250
+    },
+    {
+      "epoch": 0.8777606462681057,
+      "grad_norm": 0.1494825780391693,
+      "learning_rate": 4.122246592397954e-06,
+      "loss": 0.8697,
+      "step": 121260
+    },
+    {
+      "epoch": 0.8778330329286919,
+      "grad_norm": 0.1578221619129181,
+      "learning_rate": 4.122174205737367e-06,
+      "loss": 0.8748,
+      "step": 121270
+    },
+    {
+      "epoch": 0.877905419589278,
+      "grad_norm": 0.14515846967697144,
+      "learning_rate": 4.122101819076781e-06,
+      "loss": 0.8718,
+      "step": 121280
+    },
+    {
+      "epoch": 0.8779778062498643,
+      "grad_norm": 0.15049944818019867,
+      "learning_rate": 4.1220294324161944e-06,
+      "loss": 0.8619,
+      "step": 121290
+    },
+    {
+      "epoch": 0.8780501929104505,
+      "grad_norm": 0.15272891521453857,
+      "learning_rate": 4.121957045755609e-06,
+      "loss": 0.8802,
+      "step": 121300
+    },
+    {
+      "epoch": 0.8781225795710367,
+      "grad_norm": 0.16228324174880981,
+      "learning_rate": 4.1218846590950225e-06,
+      "loss": 0.8673,
+      "step": 121310
+    },
+    {
+      "epoch": 0.8781949662316229,
+      "grad_norm": 0.15620221197605133,
+      "learning_rate": 4.121812272434436e-06,
+      "loss": 0.8788,
+      "step": 121320
+    },
+    {
+      "epoch": 0.878267352892209,
+      "grad_norm": 0.16147710382938385,
+      "learning_rate": 4.12173988577385e-06,
+      "loss": 0.887,
+      "step": 121330
+    },
+    {
+      "epoch": 0.8783397395527952,
+      "grad_norm": 0.15584251284599304,
+      "learning_rate": 4.121667499113264e-06,
+      "loss": 0.8716,
+      "step": 121340
+    },
+    {
+      "epoch": 0.8784121262133814,
+      "grad_norm": 0.14494866132736206,
+      "learning_rate": 4.121595112452678e-06,
+      "loss": 0.8705,
+      "step": 121350
+    },
+    {
+      "epoch": 0.8784845128739676,
+      "grad_norm": 0.15706604719161987,
+      "learning_rate": 4.1215227257920914e-06,
+      "loss": 0.8711,
+      "step": 121360
+    },
+    {
+      "epoch": 0.8785568995345537,
+      "grad_norm": 0.15880705416202545,
+      "learning_rate": 4.121450339131505e-06,
+      "loss": 0.8671,
+      "step": 121370
+    },
+    {
+      "epoch": 0.8786292861951399,
+      "grad_norm": 0.1646871417760849,
+      "learning_rate": 4.1213779524709195e-06,
+      "loss": 0.8686,
+      "step": 121380
+    },
+    {
+      "epoch": 0.8787016728557262,
+      "grad_norm": 0.14550799131393433,
+      "learning_rate": 4.121305565810333e-06,
+      "loss": 0.8771,
+      "step": 121390
+    },
+    {
+      "epoch": 0.8787740595163124,
+      "grad_norm": 0.16066545248031616,
+      "learning_rate": 4.121233179149747e-06,
+      "loss": 0.8774,
+      "step": 121400
+    },
+    {
+      "epoch": 0.8788464461768986,
+      "grad_norm": 0.16294661164283752,
+      "learning_rate": 4.12116079248916e-06,
+      "loss": 0.889,
+      "step": 121410
+    },
+    {
+      "epoch": 0.8789188328374847,
+      "grad_norm": 0.16425122320652008,
+      "learning_rate": 4.121088405828574e-06,
+      "loss": 0.8681,
+      "step": 121420
+    },
+    {
+      "epoch": 0.8789912194980709,
+      "grad_norm": 0.1627512127161026,
+      "learning_rate": 4.1210160191679884e-06,
+      "loss": 0.8666,
+      "step": 121430
+    },
+    {
+      "epoch": 0.8790636061586571,
+      "grad_norm": 0.13960957527160645,
+      "learning_rate": 4.120943632507402e-06,
+      "loss": 0.8701,
+      "step": 121440
+    },
+    {
+      "epoch": 0.8791359928192433,
+      "grad_norm": 0.16151773929595947,
+      "learning_rate": 4.120871245846816e-06,
+      "loss": 0.8845,
+      "step": 121450
+    },
+    {
+      "epoch": 0.8792083794798294,
+      "grad_norm": 0.14862598478794098,
+      "learning_rate": 4.120798859186229e-06,
+      "loss": 0.8767,
+      "step": 121460
+    },
+    {
+      "epoch": 0.8792807661404156,
+      "grad_norm": 0.15480075776576996,
+      "learning_rate": 4.120726472525644e-06,
+      "loss": 0.8746,
+      "step": 121470
+    },
+    {
+      "epoch": 0.8793531528010018,
+      "grad_norm": 0.15683260560035706,
+      "learning_rate": 4.120654085865057e-06,
+      "loss": 0.8714,
+      "step": 121480
+    },
+    {
+      "epoch": 0.879425539461588,
+      "grad_norm": 0.15848079323768616,
+      "learning_rate": 4.120581699204471e-06,
+      "loss": 0.8831,
+      "step": 121490
+    },
+    {
+      "epoch": 0.8794979261221743,
+      "grad_norm": 0.16250035166740417,
+      "learning_rate": 4.120509312543885e-06,
+      "loss": 0.8825,
+      "step": 121500
+    },
+    {
+      "epoch": 0.8795703127827604,
+      "grad_norm": 0.21030132472515106,
+      "learning_rate": 4.120436925883299e-06,
+      "loss": 0.8641,
+      "step": 121510
+    },
+    {
+      "epoch": 0.8796426994433466,
+      "grad_norm": 0.1805466115474701,
+      "learning_rate": 4.120364539222713e-06,
+      "loss": 0.872,
+      "step": 121520
+    },
+    {
+      "epoch": 0.8797150861039328,
+      "grad_norm": 0.14784234762191772,
+      "learning_rate": 4.120292152562126e-06,
+      "loss": 0.8725,
+      "step": 121530
+    },
+    {
+      "epoch": 0.879787472764519,
+      "grad_norm": 0.16358047723770142,
+      "learning_rate": 4.12021976590154e-06,
+      "loss": 0.8708,
+      "step": 121540
+    },
+    {
+      "epoch": 0.8798598594251051,
+      "grad_norm": 0.16846664249897003,
+      "learning_rate": 4.1201473792409535e-06,
+      "loss": 0.874,
+      "step": 121550
+    },
+    {
+      "epoch": 0.8799322460856913,
+      "grad_norm": 0.15270091593265533,
+      "learning_rate": 4.120074992580367e-06,
+      "loss": 0.8719,
+      "step": 121560
+    },
+    {
+      "epoch": 0.8800046327462775,
+      "grad_norm": 0.1571713536977768,
+      "learning_rate": 4.120002605919781e-06,
+      "loss": 0.8754,
+      "step": 121570
+    },
+    {
+      "epoch": 0.8800770194068637,
+      "grad_norm": 0.1490432471036911,
+      "learning_rate": 4.119930219259195e-06,
+      "loss": 0.8702,
+      "step": 121580
+    },
+    {
+      "epoch": 0.8801494060674498,
+      "grad_norm": 0.14698143303394318,
+      "learning_rate": 4.119857832598609e-06,
+      "loss": 0.8756,
+      "step": 121590
+    },
+    {
+      "epoch": 0.880221792728036,
+      "grad_norm": 0.14973314106464386,
+      "learning_rate": 4.1197854459380225e-06,
+      "loss": 0.8748,
+      "step": 121600
+    },
+    {
+      "epoch": 0.8802941793886223,
+      "grad_norm": 0.20785781741142273,
+      "learning_rate": 4.119713059277436e-06,
+      "loss": 0.8852,
+      "step": 121610
+    },
+    {
+      "epoch": 0.8803665660492085,
+      "grad_norm": 0.16038250923156738,
+      "learning_rate": 4.1196406726168505e-06,
+      "loss": 0.8608,
+      "step": 121620
+    },
+    {
+      "epoch": 0.8804389527097947,
+      "grad_norm": 0.15908242762088776,
+      "learning_rate": 4.119568285956264e-06,
+      "loss": 0.8659,
+      "step": 121630
+    },
+    {
+      "epoch": 0.8805113393703808,
+      "grad_norm": 0.15829870104789734,
+      "learning_rate": 4.119495899295678e-06,
+      "loss": 0.869,
+      "step": 121640
+    },
+    {
+      "epoch": 0.880583726030967,
+      "grad_norm": 0.16468794643878937,
+      "learning_rate": 4.119423512635091e-06,
+      "loss": 0.8803,
+      "step": 121650
+    },
+    {
+      "epoch": 0.8806561126915532,
+      "grad_norm": 0.15394102036952972,
+      "learning_rate": 4.119351125974506e-06,
+      "loss": 0.8783,
+      "step": 121660
+    },
+    {
+      "epoch": 0.8807284993521394,
+      "grad_norm": 0.15285733342170715,
+      "learning_rate": 4.1192787393139195e-06,
+      "loss": 0.8683,
+      "step": 121670
+    },
+    {
+      "epoch": 0.8808008860127255,
+      "grad_norm": 0.16268157958984375,
+      "learning_rate": 4.119206352653333e-06,
+      "loss": 0.8668,
+      "step": 121680
+    },
+    {
+      "epoch": 0.8808732726733117,
+      "grad_norm": 0.15528251230716705,
+      "learning_rate": 4.119133965992747e-06,
+      "loss": 0.8625,
+      "step": 121690
+    },
+    {
+      "epoch": 0.8809456593338979,
+      "grad_norm": 0.15688292682170868,
+      "learning_rate": 4.119061579332161e-06,
+      "loss": 0.8734,
+      "step": 121700
+    },
+    {
+      "epoch": 0.8810180459944842,
+      "grad_norm": 0.15693509578704834,
+      "learning_rate": 4.118989192671575e-06,
+      "loss": 0.8748,
+      "step": 121710
+    },
+    {
+      "epoch": 0.8810904326550704,
+      "grad_norm": 0.16579411923885345,
+      "learning_rate": 4.118916806010988e-06,
+      "loss": 0.8751,
+      "step": 121720
+    },
+    {
+      "epoch": 0.8811628193156565,
+      "grad_norm": 0.16935928165912628,
+      "learning_rate": 4.118844419350402e-06,
+      "loss": 0.8638,
+      "step": 121730
+    },
+    {
+      "epoch": 0.8812352059762427,
+      "grad_norm": 0.194980189204216,
+      "learning_rate": 4.1187720326898165e-06,
+      "loss": 0.8801,
+      "step": 121740
+    },
+    {
+      "epoch": 0.8813075926368289,
+      "grad_norm": 0.16746121644973755,
+      "learning_rate": 4.11869964602923e-06,
+      "loss": 0.8841,
+      "step": 121750
+    },
+    {
+      "epoch": 0.8813799792974151,
+      "grad_norm": 0.1961222141981125,
+      "learning_rate": 4.118627259368644e-06,
+      "loss": 0.8761,
+      "step": 121760
+    },
+    {
+      "epoch": 0.8814523659580012,
+      "grad_norm": 0.15616725385189056,
+      "learning_rate": 4.118554872708057e-06,
+      "loss": 0.869,
+      "step": 121770
+    },
+    {
+      "epoch": 0.8815247526185874,
+      "grad_norm": 0.15364345908164978,
+      "learning_rate": 4.118482486047472e-06,
+      "loss": 0.8766,
+      "step": 121780
+    },
+    {
+      "epoch": 0.8815971392791736,
+      "grad_norm": 0.15156231820583344,
+      "learning_rate": 4.118410099386885e-06,
+      "loss": 0.8673,
+      "step": 121790
+    },
+    {
+      "epoch": 0.8816695259397598,
+      "grad_norm": 0.2237502783536911,
+      "learning_rate": 4.118337712726299e-06,
+      "loss": 0.8718,
+      "step": 121800
+    },
+    {
+      "epoch": 0.881741912600346,
+      "grad_norm": 0.14461927115917206,
+      "learning_rate": 4.118265326065713e-06,
+      "loss": 0.872,
+      "step": 121810
+    },
+    {
+      "epoch": 0.8818142992609322,
+      "grad_norm": 0.17010053992271423,
+      "learning_rate": 4.118192939405127e-06,
+      "loss": 0.8745,
+      "step": 121820
+    },
+    {
+      "epoch": 0.8818866859215184,
+      "grad_norm": 0.16288408637046814,
+      "learning_rate": 4.118120552744541e-06,
+      "loss": 0.8762,
+      "step": 121830
+    },
+    {
+      "epoch": 0.8819590725821046,
+      "grad_norm": 0.1665002405643463,
+      "learning_rate": 4.118048166083954e-06,
+      "loss": 0.8678,
+      "step": 121840
+    },
+    {
+      "epoch": 0.8820314592426908,
+      "grad_norm": 0.16218537092208862,
+      "learning_rate": 4.117975779423368e-06,
+      "loss": 0.8829,
+      "step": 121850
+    },
+    {
+      "epoch": 0.882103845903277,
+      "grad_norm": 0.14851604402065277,
+      "learning_rate": 4.117903392762782e-06,
+      "loss": 0.874,
+      "step": 121860
+    },
+    {
+      "epoch": 0.8821762325638631,
+      "grad_norm": 0.1481228917837143,
+      "learning_rate": 4.117831006102196e-06,
+      "loss": 0.868,
+      "step": 121870
+    },
+    {
+      "epoch": 0.8822486192244493,
+      "grad_norm": 0.16877874732017517,
+      "learning_rate": 4.11775861944161e-06,
+      "loss": 0.8754,
+      "step": 121880
+    },
+    {
+      "epoch": 0.8823210058850355,
+      "grad_norm": 0.15102118253707886,
+      "learning_rate": 4.117686232781023e-06,
+      "loss": 0.8721,
+      "step": 121890
+    },
+    {
+      "epoch": 0.8823933925456217,
+      "grad_norm": 0.15276771783828735,
+      "learning_rate": 4.117613846120438e-06,
+      "loss": 0.8833,
+      "step": 121900
+    },
+    {
+      "epoch": 0.8824657792062078,
+      "grad_norm": 0.14909601211547852,
+      "learning_rate": 4.117541459459851e-06,
+      "loss": 0.8613,
+      "step": 121910
+    },
+    {
+      "epoch": 0.8825381658667941,
+      "grad_norm": 0.15736235678195953,
+      "learning_rate": 4.117469072799265e-06,
+      "loss": 0.8756,
+      "step": 121920
+    },
+    {
+      "epoch": 0.8826105525273803,
+      "grad_norm": 0.1750316172838211,
+      "learning_rate": 4.1173966861386786e-06,
+      "loss": 0.8851,
+      "step": 121930
+    },
+    {
+      "epoch": 0.8826829391879665,
+      "grad_norm": 0.14961692690849304,
+      "learning_rate": 4.117324299478093e-06,
+      "loss": 0.871,
+      "step": 121940
+    },
+    {
+      "epoch": 0.8827553258485527,
+      "grad_norm": 0.13990993797779083,
+      "learning_rate": 4.117251912817507e-06,
+      "loss": 0.8573,
+      "step": 121950
+    },
+    {
+      "epoch": 0.8828277125091388,
+      "grad_norm": 0.1395825743675232,
+      "learning_rate": 4.11717952615692e-06,
+      "loss": 0.8589,
+      "step": 121960
+    },
+    {
+      "epoch": 0.882900099169725,
+      "grad_norm": 0.15364444255828857,
+      "learning_rate": 4.117107139496334e-06,
+      "loss": 0.867,
+      "step": 121970
+    },
+    {
+      "epoch": 0.8829724858303112,
+      "grad_norm": 3.056195020675659,
+      "learning_rate": 4.117034752835748e-06,
+      "loss": 0.882,
+      "step": 121980
+    },
+    {
+      "epoch": 0.8830448724908974,
+      "grad_norm": 0.14895865321159363,
+      "learning_rate": 4.116962366175162e-06,
+      "loss": 0.8752,
+      "step": 121990
+    },
+    {
+      "epoch": 0.8831172591514835,
+      "grad_norm": 0.1695874035358429,
+      "learning_rate": 4.116889979514576e-06,
+      "loss": 0.877,
+      "step": 122000
+    },
+    {
+      "epoch": 0.8831896458120697,
+      "grad_norm": 0.1436607986688614,
+      "learning_rate": 4.116817592853989e-06,
+      "loss": 0.8785,
+      "step": 122010
+    },
+    {
+      "epoch": 0.8832620324726559,
+      "grad_norm": 0.15994448959827423,
+      "learning_rate": 4.116745206193403e-06,
+      "loss": 0.8697,
+      "step": 122020
+    },
+    {
+      "epoch": 0.8833344191332422,
+      "grad_norm": 0.15391020476818085,
+      "learning_rate": 4.116672819532817e-06,
+      "loss": 0.8702,
+      "step": 122030
+    },
+    {
+      "epoch": 0.8834068057938284,
+      "grad_norm": 0.1519947201013565,
+      "learning_rate": 4.116600432872231e-06,
+      "loss": 0.8728,
+      "step": 122040
+    },
+    {
+      "epoch": 0.8834791924544145,
+      "grad_norm": 0.14958973228931427,
+      "learning_rate": 4.1165280462116445e-06,
+      "loss": 0.8603,
+      "step": 122050
+    },
+    {
+      "epoch": 0.8835515791150007,
+      "grad_norm": 0.15004867315292358,
+      "learning_rate": 4.116455659551058e-06,
+      "loss": 0.878,
+      "step": 122060
+    },
+    {
+      "epoch": 0.8836239657755869,
+      "grad_norm": 0.16854992508888245,
+      "learning_rate": 4.116383272890473e-06,
+      "loss": 0.8622,
+      "step": 122070
+    },
+    {
+      "epoch": 0.8836963524361731,
+      "grad_norm": 0.14855122566223145,
+      "learning_rate": 4.116310886229885e-06,
+      "loss": 0.8679,
+      "step": 122080
+    },
+    {
+      "epoch": 0.8837687390967592,
+      "grad_norm": 0.17604343593120575,
+      "learning_rate": 4.1162384995693e-06,
+      "loss": 0.8664,
+      "step": 122090
+    },
+    {
+      "epoch": 0.8838411257573454,
+      "grad_norm": 0.15962982177734375,
+      "learning_rate": 4.1161661129087134e-06,
+      "loss": 0.8831,
+      "step": 122100
+    },
+    {
+      "epoch": 0.8839135124179316,
+      "grad_norm": 0.15554451942443848,
+      "learning_rate": 4.116093726248127e-06,
+      "loss": 0.8701,
+      "step": 122110
+    },
+    {
+      "epoch": 0.8839858990785178,
+      "grad_norm": 0.15746499598026276,
+      "learning_rate": 4.116021339587541e-06,
+      "loss": 0.8682,
+      "step": 122120
+    },
+    {
+      "epoch": 0.8840582857391039,
+      "grad_norm": 0.1579037606716156,
+      "learning_rate": 4.115948952926955e-06,
+      "loss": 0.8738,
+      "step": 122130
+    },
+    {
+      "epoch": 0.8841306723996902,
+      "grad_norm": 0.16659896075725555,
+      "learning_rate": 4.115876566266369e-06,
+      "loss": 0.874,
+      "step": 122140
+    },
+    {
+      "epoch": 0.8842030590602764,
+      "grad_norm": 0.183345228433609,
+      "learning_rate": 4.115804179605782e-06,
+      "loss": 0.8697,
+      "step": 122150
+    },
+    {
+      "epoch": 0.8842754457208626,
+      "grad_norm": 0.15425361692905426,
+      "learning_rate": 4.115731792945196e-06,
+      "loss": 0.8637,
+      "step": 122160
+    },
+    {
+      "epoch": 0.8843478323814488,
+      "grad_norm": 0.1493588238954544,
+      "learning_rate": 4.1156594062846104e-06,
+      "loss": 0.8512,
+      "step": 122170
+    },
+    {
+      "epoch": 0.8844202190420349,
+      "grad_norm": 0.14579205214977264,
+      "learning_rate": 4.115587019624024e-06,
+      "loss": 0.8685,
+      "step": 122180
+    },
+    {
+      "epoch": 0.8844926057026211,
+      "grad_norm": 0.14941352605819702,
+      "learning_rate": 4.115514632963438e-06,
+      "loss": 0.8795,
+      "step": 122190
+    },
+    {
+      "epoch": 0.8845649923632073,
+      "grad_norm": 0.16284583508968353,
+      "learning_rate": 4.115442246302851e-06,
+      "loss": 0.8814,
+      "step": 122200
+    },
+    {
+      "epoch": 0.8846373790237935,
+      "grad_norm": 0.15186572074890137,
+      "learning_rate": 4.115369859642265e-06,
+      "loss": 0.8693,
+      "step": 122210
+    },
+    {
+      "epoch": 0.8847097656843796,
+      "grad_norm": 0.15203732252120972,
+      "learning_rate": 4.115297472981679e-06,
+      "loss": 0.879,
+      "step": 122220
+    },
+    {
+      "epoch": 0.8847821523449658,
+      "grad_norm": 0.15315940976142883,
+      "learning_rate": 4.115225086321093e-06,
+      "loss": 0.8679,
+      "step": 122230
+    },
+    {
+      "epoch": 0.8848545390055521,
+      "grad_norm": 0.15874212980270386,
+      "learning_rate": 4.115152699660507e-06,
+      "loss": 0.8919,
+      "step": 122240
+    },
+    {
+      "epoch": 0.8849269256661383,
+      "grad_norm": 0.15111921727657318,
+      "learning_rate": 4.11508031299992e-06,
+      "loss": 0.876,
+      "step": 122250
+    },
+    {
+      "epoch": 0.8849993123267245,
+      "grad_norm": 0.16939522325992584,
+      "learning_rate": 4.115007926339335e-06,
+      "loss": 0.8728,
+      "step": 122260
+    },
+    {
+      "epoch": 0.8850716989873106,
+      "grad_norm": 0.14914065599441528,
+      "learning_rate": 4.114935539678748e-06,
+      "loss": 0.8666,
+      "step": 122270
+    },
+    {
+      "epoch": 0.8851440856478968,
+      "grad_norm": 0.17796002328395844,
+      "learning_rate": 4.114863153018162e-06,
+      "loss": 0.86,
+      "step": 122280
+    },
+    {
+      "epoch": 0.885216472308483,
+      "grad_norm": 0.14601661264896393,
+      "learning_rate": 4.1147907663575755e-06,
+      "loss": 0.8799,
+      "step": 122290
+    },
+    {
+      "epoch": 0.8852888589690692,
+      "grad_norm": 0.14311879873275757,
+      "learning_rate": 4.11471837969699e-06,
+      "loss": 0.872,
+      "step": 122300
+    },
+    {
+      "epoch": 0.8853612456296553,
+      "grad_norm": 0.14968661963939667,
+      "learning_rate": 4.114645993036404e-06,
+      "loss": 0.8701,
+      "step": 122310
+    },
+    {
+      "epoch": 0.8854336322902415,
+      "grad_norm": 0.15330290794372559,
+      "learning_rate": 4.114573606375817e-06,
+      "loss": 0.8773,
+      "step": 122320
+    },
+    {
+      "epoch": 0.8855060189508277,
+      "grad_norm": 0.14958740770816803,
+      "learning_rate": 4.114501219715231e-06,
+      "loss": 0.8634,
+      "step": 122330
+    },
+    {
+      "epoch": 0.8855784056114139,
+      "grad_norm": 0.15254485607147217,
+      "learning_rate": 4.114428833054645e-06,
+      "loss": 0.863,
+      "step": 122340
+    },
+    {
+      "epoch": 0.8856507922720002,
+      "grad_norm": 0.18143147230148315,
+      "learning_rate": 4.114356446394059e-06,
+      "loss": 0.862,
+      "step": 122350
+    },
+    {
+      "epoch": 0.8857231789325863,
+      "grad_norm": 0.1426178365945816,
+      "learning_rate": 4.1142840597334725e-06,
+      "loss": 0.8654,
+      "step": 122360
+    },
+    {
+      "epoch": 0.8857955655931725,
+      "grad_norm": 0.15494516491889954,
+      "learning_rate": 4.114211673072886e-06,
+      "loss": 0.8658,
+      "step": 122370
+    },
+    {
+      "epoch": 0.8858679522537587,
+      "grad_norm": 0.15428562462329865,
+      "learning_rate": 4.114139286412301e-06,
+      "loss": 0.8713,
+      "step": 122380
+    },
+    {
+      "epoch": 0.8859403389143449,
+      "grad_norm": 0.14963556826114655,
+      "learning_rate": 4.114066899751714e-06,
+      "loss": 0.8621,
+      "step": 122390
+    },
+    {
+      "epoch": 0.886012725574931,
+      "grad_norm": 0.16026413440704346,
+      "learning_rate": 4.113994513091128e-06,
+      "loss": 0.8767,
+      "step": 122400
+    },
+    {
+      "epoch": 0.8860851122355172,
+      "grad_norm": 0.14730548858642578,
+      "learning_rate": 4.1139221264305415e-06,
+      "loss": 0.866,
+      "step": 122410
+    },
+    {
+      "epoch": 0.8861574988961034,
+      "grad_norm": 0.14938196539878845,
+      "learning_rate": 4.113849739769956e-06,
+      "loss": 0.8596,
+      "step": 122420
+    },
+    {
+      "epoch": 0.8862298855566896,
+      "grad_norm": 0.16436588764190674,
+      "learning_rate": 4.1137773531093695e-06,
+      "loss": 0.8769,
+      "step": 122430
+    },
+    {
+      "epoch": 0.8863022722172758,
+      "grad_norm": 0.15649719536304474,
+      "learning_rate": 4.113704966448783e-06,
+      "loss": 0.872,
+      "step": 122440
+    },
+    {
+      "epoch": 0.886374658877862,
+      "grad_norm": 0.1726832240819931,
+      "learning_rate": 4.113632579788197e-06,
+      "loss": 0.8779,
+      "step": 122450
+    },
+    {
+      "epoch": 0.8864470455384482,
+      "grad_norm": 0.14762026071548462,
+      "learning_rate": 4.113560193127611e-06,
+      "loss": 0.8725,
+      "step": 122460
+    },
+    {
+      "epoch": 0.8865194321990344,
+      "grad_norm": 0.14345690608024597,
+      "learning_rate": 4.113487806467025e-06,
+      "loss": 0.8574,
+      "step": 122470
+    },
+    {
+      "epoch": 0.8865918188596206,
+      "grad_norm": 0.16033095121383667,
+      "learning_rate": 4.1134154198064385e-06,
+      "loss": 0.8639,
+      "step": 122480
+    },
+    {
+      "epoch": 0.8866642055202067,
+      "grad_norm": 0.14597992599010468,
+      "learning_rate": 4.113343033145852e-06,
+      "loss": 0.8729,
+      "step": 122490
+    },
+    {
+      "epoch": 0.8867365921807929,
+      "grad_norm": 0.16098947823047638,
+      "learning_rate": 4.1132706464852666e-06,
+      "loss": 0.8659,
+      "step": 122500
+    },
+    {
+      "epoch": 0.8868089788413791,
+      "grad_norm": 0.14861907064914703,
+      "learning_rate": 4.11319825982468e-06,
+      "loss": 0.8686,
+      "step": 122510
+    },
+    {
+      "epoch": 0.8868813655019653,
+      "grad_norm": 0.15192726254463196,
+      "learning_rate": 4.113125873164094e-06,
+      "loss": 0.873,
+      "step": 122520
+    },
+    {
+      "epoch": 0.8869537521625515,
+      "grad_norm": 0.16257117688655853,
+      "learning_rate": 4.113053486503507e-06,
+      "loss": 0.8881,
+      "step": 122530
+    },
+    {
+      "epoch": 0.8870261388231376,
+      "grad_norm": 0.15565155446529388,
+      "learning_rate": 4.112981099842922e-06,
+      "loss": 0.8813,
+      "step": 122540
+    },
+    {
+      "epoch": 0.8870985254837238,
+      "grad_norm": 0.16683930158615112,
+      "learning_rate": 4.1129087131823355e-06,
+      "loss": 0.8844,
+      "step": 122550
+    },
+    {
+      "epoch": 0.8871709121443101,
+      "grad_norm": 0.16827575862407684,
+      "learning_rate": 4.112836326521749e-06,
+      "loss": 0.8845,
+      "step": 122560
+    },
+    {
+      "epoch": 0.8872432988048963,
+      "grad_norm": 0.14905016124248505,
+      "learning_rate": 4.112763939861163e-06,
+      "loss": 0.8852,
+      "step": 122570
+    },
+    {
+      "epoch": 0.8873156854654825,
+      "grad_norm": 0.15825332701206207,
+      "learning_rate": 4.112691553200577e-06,
+      "loss": 0.8702,
+      "step": 122580
+    },
+    {
+      "epoch": 0.8873880721260686,
+      "grad_norm": 0.15095843374729156,
+      "learning_rate": 4.112619166539991e-06,
+      "loss": 0.8667,
+      "step": 122590
+    },
+    {
+      "epoch": 0.8874604587866548,
+      "grad_norm": 0.15348178148269653,
+      "learning_rate": 4.112546779879404e-06,
+      "loss": 0.8696,
+      "step": 122600
+    },
+    {
+      "epoch": 0.887532845447241,
+      "grad_norm": 0.14814035594463348,
+      "learning_rate": 4.112474393218818e-06,
+      "loss": 0.8742,
+      "step": 122610
+    },
+    {
+      "epoch": 0.8876052321078272,
+      "grad_norm": 0.15248076617717743,
+      "learning_rate": 4.112402006558232e-06,
+      "loss": 0.8855,
+      "step": 122620
+    },
+    {
+      "epoch": 0.8876776187684133,
+      "grad_norm": 0.15458175539970398,
+      "learning_rate": 4.112329619897645e-06,
+      "loss": 0.8722,
+      "step": 122630
+    },
+    {
+      "epoch": 0.8877500054289995,
+      "grad_norm": 0.15646861493587494,
+      "learning_rate": 4.112257233237059e-06,
+      "loss": 0.87,
+      "step": 122640
+    },
+    {
+      "epoch": 0.8878223920895857,
+      "grad_norm": 0.14675036072731018,
+      "learning_rate": 4.112184846576473e-06,
+      "loss": 0.8763,
+      "step": 122650
+    },
+    {
+      "epoch": 0.8878947787501719,
+      "grad_norm": 0.16464652121067047,
+      "learning_rate": 4.112112459915887e-06,
+      "loss": 0.8817,
+      "step": 122660
+    },
+    {
+      "epoch": 0.8879671654107582,
+      "grad_norm": 0.15563102066516876,
+      "learning_rate": 4.1120400732553006e-06,
+      "loss": 0.8728,
+      "step": 122670
+    },
+    {
+      "epoch": 0.8880395520713443,
+      "grad_norm": 0.15346914529800415,
+      "learning_rate": 4.111967686594714e-06,
+      "loss": 0.8636,
+      "step": 122680
+    },
+    {
+      "epoch": 0.8881119387319305,
+      "grad_norm": 0.15571348369121552,
+      "learning_rate": 4.111895299934129e-06,
+      "loss": 0.8643,
+      "step": 122690
+    },
+    {
+      "epoch": 0.8881843253925167,
+      "grad_norm": 0.16230016946792603,
+      "learning_rate": 4.111822913273542e-06,
+      "loss": 0.8615,
+      "step": 122700
+    },
+    {
+      "epoch": 0.8882567120531029,
+      "grad_norm": 0.1600496470928192,
+      "learning_rate": 4.111750526612956e-06,
+      "loss": 0.8824,
+      "step": 122710
+    },
+    {
+      "epoch": 0.888329098713689,
+      "grad_norm": 0.14802932739257812,
+      "learning_rate": 4.1116781399523695e-06,
+      "loss": 0.8607,
+      "step": 122720
+    },
+    {
+      "epoch": 0.8884014853742752,
+      "grad_norm": 0.14707952737808228,
+      "learning_rate": 4.111605753291784e-06,
+      "loss": 0.8645,
+      "step": 122730
+    },
+    {
+      "epoch": 0.8884738720348614,
+      "grad_norm": 0.15505805611610413,
+      "learning_rate": 4.1115333666311976e-06,
+      "loss": 0.874,
+      "step": 122740
+    },
+    {
+      "epoch": 0.8885462586954476,
+      "grad_norm": 0.1629432737827301,
+      "learning_rate": 4.111460979970611e-06,
+      "loss": 0.8765,
+      "step": 122750
+    },
+    {
+      "epoch": 0.8886186453560337,
+      "grad_norm": 0.15821245312690735,
+      "learning_rate": 4.111388593310025e-06,
+      "loss": 0.8836,
+      "step": 122760
+    },
+    {
+      "epoch": 0.88869103201662,
+      "grad_norm": 0.1640791893005371,
+      "learning_rate": 4.111316206649439e-06,
+      "loss": 0.8718,
+      "step": 122770
+    },
+    {
+      "epoch": 0.8887634186772062,
+      "grad_norm": 0.1496168076992035,
+      "learning_rate": 4.111243819988853e-06,
+      "loss": 0.8776,
+      "step": 122780
+    },
+    {
+      "epoch": 0.8888358053377924,
+      "grad_norm": 0.15713010728359222,
+      "learning_rate": 4.1111714333282665e-06,
+      "loss": 0.8767,
+      "step": 122790
+    },
+    {
+      "epoch": 0.8889081919983786,
+      "grad_norm": 0.15316422283649445,
+      "learning_rate": 4.11109904666768e-06,
+      "loss": 0.8752,
+      "step": 122800
+    },
+    {
+      "epoch": 0.8889805786589647,
+      "grad_norm": 0.17414002120494843,
+      "learning_rate": 4.111026660007095e-06,
+      "loss": 0.8743,
+      "step": 122810
+    },
+    {
+      "epoch": 0.8890529653195509,
+      "grad_norm": 0.1568707376718521,
+      "learning_rate": 4.110954273346508e-06,
+      "loss": 0.8779,
+      "step": 122820
+    },
+    {
+      "epoch": 0.8891253519801371,
+      "grad_norm": 0.15081565082073212,
+      "learning_rate": 4.110881886685922e-06,
+      "loss": 0.8679,
+      "step": 122830
+    },
+    {
+      "epoch": 0.8891977386407233,
+      "grad_norm": 0.14779068529605865,
+      "learning_rate": 4.1108095000253354e-06,
+      "loss": 0.8696,
+      "step": 122840
+    },
+    {
+      "epoch": 0.8892701253013094,
+      "grad_norm": 0.17924214899539948,
+      "learning_rate": 4.110737113364749e-06,
+      "loss": 0.8915,
+      "step": 122850
+    },
+    {
+      "epoch": 0.8893425119618956,
+      "grad_norm": 0.1565881222486496,
+      "learning_rate": 4.1106647267041635e-06,
+      "loss": 0.8798,
+      "step": 122860
+    },
+    {
+      "epoch": 0.8894148986224818,
+      "grad_norm": 0.16054075956344604,
+      "learning_rate": 4.110592340043577e-06,
+      "loss": 0.8543,
+      "step": 122870
+    },
+    {
+      "epoch": 0.8894872852830681,
+      "grad_norm": 0.1694210320711136,
+      "learning_rate": 4.110519953382991e-06,
+      "loss": 0.877,
+      "step": 122880
+    },
+    {
+      "epoch": 0.8895596719436543,
+      "grad_norm": 0.15159261226654053,
+      "learning_rate": 4.110447566722404e-06,
+      "loss": 0.8787,
+      "step": 122890
+    },
+    {
+      "epoch": 0.8896320586042404,
+      "grad_norm": 0.1697186380624771,
+      "learning_rate": 4.110375180061819e-06,
+      "loss": 0.8796,
+      "step": 122900
+    },
+    {
+      "epoch": 0.8897044452648266,
+      "grad_norm": 0.1503906100988388,
+      "learning_rate": 4.1103027934012324e-06,
+      "loss": 0.873,
+      "step": 122910
+    },
+    {
+      "epoch": 0.8897768319254128,
+      "grad_norm": 0.14617328345775604,
+      "learning_rate": 4.110230406740646e-06,
+      "loss": 0.8758,
+      "step": 122920
+    },
+    {
+      "epoch": 0.889849218585999,
+      "grad_norm": 0.16019441187381744,
+      "learning_rate": 4.11015802008006e-06,
+      "loss": 0.8714,
+      "step": 122930
+    },
+    {
+      "epoch": 0.8899216052465851,
+      "grad_norm": 0.14964258670806885,
+      "learning_rate": 4.110085633419474e-06,
+      "loss": 0.8641,
+      "step": 122940
+    },
+    {
+      "epoch": 0.8899939919071713,
+      "grad_norm": 0.15857601165771484,
+      "learning_rate": 4.110013246758888e-06,
+      "loss": 0.8818,
+      "step": 122950
+    },
+    {
+      "epoch": 0.8900663785677575,
+      "grad_norm": 0.1659708172082901,
+      "learning_rate": 4.109940860098301e-06,
+      "loss": 0.8809,
+      "step": 122960
+    },
+    {
+      "epoch": 0.8901387652283437,
+      "grad_norm": 0.15709270536899567,
+      "learning_rate": 4.109868473437715e-06,
+      "loss": 0.878,
+      "step": 122970
+    },
+    {
+      "epoch": 0.89021115188893,
+      "grad_norm": 0.15638253092765808,
+      "learning_rate": 4.1097960867771294e-06,
+      "loss": 0.88,
+      "step": 122980
+    },
+    {
+      "epoch": 0.8902835385495161,
+      "grad_norm": 0.2203262895345688,
+      "learning_rate": 4.109723700116543e-06,
+      "loss": 0.872,
+      "step": 122990
+    },
+    {
+      "epoch": 0.8903559252101023,
+      "grad_norm": 0.1518968641757965,
+      "learning_rate": 4.109651313455957e-06,
+      "loss": 0.8815,
+      "step": 123000
+    },
+    {
+      "epoch": 0.8904283118706885,
+      "grad_norm": 0.1435840129852295,
+      "learning_rate": 4.10957892679537e-06,
+      "loss": 0.8743,
+      "step": 123010
+    },
+    {
+      "epoch": 0.8905006985312747,
+      "grad_norm": 0.15603044629096985,
+      "learning_rate": 4.109506540134785e-06,
+      "loss": 0.8781,
+      "step": 123020
+    },
+    {
+      "epoch": 0.8905730851918608,
+      "grad_norm": 0.15523672103881836,
+      "learning_rate": 4.109434153474198e-06,
+      "loss": 0.8721,
+      "step": 123030
+    },
+    {
+      "epoch": 0.890645471852447,
+      "grad_norm": 0.15901529788970947,
+      "learning_rate": 4.109361766813612e-06,
+      "loss": 0.8793,
+      "step": 123040
+    },
+    {
+      "epoch": 0.8907178585130332,
+      "grad_norm": 0.14832401275634766,
+      "learning_rate": 4.109289380153026e-06,
+      "loss": 0.8713,
+      "step": 123050
+    },
+    {
+      "epoch": 0.8907902451736194,
+      "grad_norm": 0.14594705402851105,
+      "learning_rate": 4.10921699349244e-06,
+      "loss": 0.8588,
+      "step": 123060
+    },
+    {
+      "epoch": 0.8908626318342056,
+      "grad_norm": 0.21617244184017181,
+      "learning_rate": 4.109144606831854e-06,
+      "loss": 0.8714,
+      "step": 123070
+    },
+    {
+      "epoch": 0.8909350184947917,
+      "grad_norm": 0.15614116191864014,
+      "learning_rate": 4.109072220171267e-06,
+      "loss": 0.8754,
+      "step": 123080
+    },
+    {
+      "epoch": 0.891007405155378,
+      "grad_norm": 0.17854410409927368,
+      "learning_rate": 4.108999833510681e-06,
+      "loss": 0.8709,
+      "step": 123090
+    },
+    {
+      "epoch": 0.8910797918159642,
+      "grad_norm": 0.146412193775177,
+      "learning_rate": 4.108927446850095e-06,
+      "loss": 0.8688,
+      "step": 123100
+    },
+    {
+      "epoch": 0.8911521784765504,
+      "grad_norm": 0.16947603225708008,
+      "learning_rate": 4.108855060189509e-06,
+      "loss": 0.8774,
+      "step": 123110
+    },
+    {
+      "epoch": 0.8912245651371365,
+      "grad_norm": 0.16192446649074554,
+      "learning_rate": 4.108782673528923e-06,
+      "loss": 0.8858,
+      "step": 123120
+    },
+    {
+      "epoch": 0.8912969517977227,
+      "grad_norm": 0.15947166085243225,
+      "learning_rate": 4.108710286868336e-06,
+      "loss": 0.8668,
+      "step": 123130
+    },
+    {
+      "epoch": 0.8913693384583089,
+      "grad_norm": 0.14823932945728302,
+      "learning_rate": 4.108637900207751e-06,
+      "loss": 0.8724,
+      "step": 123140
+    },
+    {
+      "epoch": 0.8914417251188951,
+      "grad_norm": 0.15632663667201996,
+      "learning_rate": 4.1085655135471635e-06,
+      "loss": 0.8697,
+      "step": 123150
+    },
+    {
+      "epoch": 0.8915141117794813,
+      "grad_norm": 0.18125022947788239,
+      "learning_rate": 4.108493126886577e-06,
+      "loss": 0.8844,
+      "step": 123160
+    },
+    {
+      "epoch": 0.8915864984400674,
+      "grad_norm": 0.1472315490245819,
+      "learning_rate": 4.1084207402259915e-06,
+      "loss": 0.8656,
+      "step": 123170
+    },
+    {
+      "epoch": 0.8916588851006536,
+      "grad_norm": 0.16403494775295258,
+      "learning_rate": 4.108348353565405e-06,
+      "loss": 0.8732,
+      "step": 123180
+    },
+    {
+      "epoch": 0.8917312717612398,
+      "grad_norm": 0.17012234032154083,
+      "learning_rate": 4.108275966904819e-06,
+      "loss": 0.8715,
+      "step": 123190
+    },
+    {
+      "epoch": 0.8918036584218261,
+      "grad_norm": 0.16016776859760284,
+      "learning_rate": 4.108203580244232e-06,
+      "loss": 0.8698,
+      "step": 123200
+    },
+    {
+      "epoch": 0.8918760450824122,
+      "grad_norm": 0.1564156860113144,
+      "learning_rate": 4.108131193583647e-06,
+      "loss": 0.8808,
+      "step": 123210
+    },
+    {
+      "epoch": 0.8919484317429984,
+      "grad_norm": 0.14456337690353394,
+      "learning_rate": 4.1080588069230605e-06,
+      "loss": 0.8633,
+      "step": 123220
+    },
+    {
+      "epoch": 0.8920208184035846,
+      "grad_norm": 0.1573518067598343,
+      "learning_rate": 4.107986420262474e-06,
+      "loss": 0.873,
+      "step": 123230
+    },
+    {
+      "epoch": 0.8920932050641708,
+      "grad_norm": 0.14703720808029175,
+      "learning_rate": 4.107914033601888e-06,
+      "loss": 0.8733,
+      "step": 123240
+    },
+    {
+      "epoch": 0.892165591724757,
+      "grad_norm": 0.14381085336208344,
+      "learning_rate": 4.107841646941302e-06,
+      "loss": 0.8629,
+      "step": 123250
+    },
+    {
+      "epoch": 0.8922379783853431,
+      "grad_norm": 0.15476030111312866,
+      "learning_rate": 4.107769260280716e-06,
+      "loss": 0.8765,
+      "step": 123260
+    },
+    {
+      "epoch": 0.8923103650459293,
+      "grad_norm": 0.1462145894765854,
+      "learning_rate": 4.107696873620129e-06,
+      "loss": 0.8677,
+      "step": 123270
+    },
+    {
+      "epoch": 0.8923827517065155,
+      "grad_norm": 0.18984442949295044,
+      "learning_rate": 4.107624486959543e-06,
+      "loss": 0.8851,
+      "step": 123280
+    },
+    {
+      "epoch": 0.8924551383671017,
+      "grad_norm": 0.15868966281414032,
+      "learning_rate": 4.1075521002989575e-06,
+      "loss": 0.864,
+      "step": 123290
+    },
+    {
+      "epoch": 0.892527525027688,
+      "grad_norm": 0.1640649437904358,
+      "learning_rate": 4.107479713638371e-06,
+      "loss": 0.8773,
+      "step": 123300
+    },
+    {
+      "epoch": 0.8925999116882741,
+      "grad_norm": 0.16619038581848145,
+      "learning_rate": 4.107407326977785e-06,
+      "loss": 0.8643,
+      "step": 123310
+    },
+    {
+      "epoch": 0.8926722983488603,
+      "grad_norm": 0.14695781469345093,
+      "learning_rate": 4.107334940317198e-06,
+      "loss": 0.8755,
+      "step": 123320
+    },
+    {
+      "epoch": 0.8927446850094465,
+      "grad_norm": 0.14879372715950012,
+      "learning_rate": 4.107262553656613e-06,
+      "loss": 0.8826,
+      "step": 123330
+    },
+    {
+      "epoch": 0.8928170716700327,
+      "grad_norm": 0.16833491623401642,
+      "learning_rate": 4.107190166996026e-06,
+      "loss": 0.8893,
+      "step": 123340
+    },
+    {
+      "epoch": 0.8928894583306188,
+      "grad_norm": 0.15543371438980103,
+      "learning_rate": 4.10711778033544e-06,
+      "loss": 0.8869,
+      "step": 123350
+    },
+    {
+      "epoch": 0.892961844991205,
+      "grad_norm": 0.15344348549842834,
+      "learning_rate": 4.107045393674854e-06,
+      "loss": 0.8791,
+      "step": 123360
+    },
+    {
+      "epoch": 0.8930342316517912,
+      "grad_norm": 0.15514753758907318,
+      "learning_rate": 4.106973007014268e-06,
+      "loss": 0.8844,
+      "step": 123370
+    },
+    {
+      "epoch": 0.8931066183123774,
+      "grad_norm": 0.16524678468704224,
+      "learning_rate": 4.106900620353682e-06,
+      "loss": 0.867,
+      "step": 123380
+    },
+    {
+      "epoch": 0.8931790049729635,
+      "grad_norm": 0.16478219628334045,
+      "learning_rate": 4.106828233693095e-06,
+      "loss": 0.8817,
+      "step": 123390
+    },
+    {
+      "epoch": 0.8932513916335497,
+      "grad_norm": 0.15810853242874146,
+      "learning_rate": 4.106755847032509e-06,
+      "loss": 0.8733,
+      "step": 123400
+    },
+    {
+      "epoch": 0.893323778294136,
+      "grad_norm": 0.1545628160238266,
+      "learning_rate": 4.106683460371923e-06,
+      "loss": 0.8761,
+      "step": 123410
+    },
+    {
+      "epoch": 0.8933961649547222,
+      "grad_norm": 0.176656574010849,
+      "learning_rate": 4.106611073711337e-06,
+      "loss": 0.8779,
+      "step": 123420
+    },
+    {
+      "epoch": 0.8934685516153084,
+      "grad_norm": 0.15695984661579132,
+      "learning_rate": 4.106538687050751e-06,
+      "loss": 0.8856,
+      "step": 123430
+    },
+    {
+      "epoch": 0.8935409382758945,
+      "grad_norm": 0.1536366492509842,
+      "learning_rate": 4.106466300390164e-06,
+      "loss": 0.8712,
+      "step": 123440
+    },
+    {
+      "epoch": 0.8936133249364807,
+      "grad_norm": 0.2222108393907547,
+      "learning_rate": 4.106393913729578e-06,
+      "loss": 0.8721,
+      "step": 123450
+    },
+    {
+      "epoch": 0.8936857115970669,
+      "grad_norm": 0.15123595297336578,
+      "learning_rate": 4.106321527068992e-06,
+      "loss": 0.8617,
+      "step": 123460
+    },
+    {
+      "epoch": 0.8937580982576531,
+      "grad_norm": 0.14488713443279266,
+      "learning_rate": 4.106249140408406e-06,
+      "loss": 0.8686,
+      "step": 123470
+    },
+    {
+      "epoch": 0.8938304849182392,
+      "grad_norm": 0.18304714560508728,
+      "learning_rate": 4.1061767537478196e-06,
+      "loss": 0.8768,
+      "step": 123480
+    },
+    {
+      "epoch": 0.8939028715788254,
+      "grad_norm": 0.16296693682670593,
+      "learning_rate": 4.106104367087233e-06,
+      "loss": 0.8661,
+      "step": 123490
+    },
+    {
+      "epoch": 0.8939752582394116,
+      "grad_norm": 0.14711807668209076,
+      "learning_rate": 4.106031980426648e-06,
+      "loss": 0.8769,
+      "step": 123500
+    },
+    {
+      "epoch": 0.8940476448999978,
+      "grad_norm": 0.1431189477443695,
+      "learning_rate": 4.105959593766061e-06,
+      "loss": 0.8894,
+      "step": 123510
+    },
+    {
+      "epoch": 0.8941200315605841,
+      "grad_norm": 0.1538097858428955,
+      "learning_rate": 4.105887207105475e-06,
+      "loss": 0.8737,
+      "step": 123520
+    },
+    {
+      "epoch": 0.8941924182211702,
+      "grad_norm": 0.15925562381744385,
+      "learning_rate": 4.1058148204448885e-06,
+      "loss": 0.8678,
+      "step": 123530
+    },
+    {
+      "epoch": 0.8942648048817564,
+      "grad_norm": 0.15328852832317352,
+      "learning_rate": 4.105742433784303e-06,
+      "loss": 0.8692,
+      "step": 123540
+    },
+    {
+      "epoch": 0.8943371915423426,
+      "grad_norm": 0.17088687419891357,
+      "learning_rate": 4.105670047123717e-06,
+      "loss": 0.8799,
+      "step": 123550
+    },
+    {
+      "epoch": 0.8944095782029288,
+      "grad_norm": 0.16045989096164703,
+      "learning_rate": 4.10559766046313e-06,
+      "loss": 0.8686,
+      "step": 123560
+    },
+    {
+      "epoch": 0.894481964863515,
+      "grad_norm": 0.17168870568275452,
+      "learning_rate": 4.105525273802544e-06,
+      "loss": 0.8733,
+      "step": 123570
+    },
+    {
+      "epoch": 0.8945543515241011,
+      "grad_norm": 0.16414594650268555,
+      "learning_rate": 4.105452887141958e-06,
+      "loss": 0.8873,
+      "step": 123580
+    },
+    {
+      "epoch": 0.8946267381846873,
+      "grad_norm": 0.15379905700683594,
+      "learning_rate": 4.105380500481372e-06,
+      "loss": 0.8712,
+      "step": 123590
+    },
+    {
+      "epoch": 0.8946991248452735,
+      "grad_norm": 0.1613912731409073,
+      "learning_rate": 4.1053081138207855e-06,
+      "loss": 0.8756,
+      "step": 123600
+    },
+    {
+      "epoch": 0.8947715115058597,
+      "grad_norm": 0.1442667692899704,
+      "learning_rate": 4.105235727160199e-06,
+      "loss": 0.8774,
+      "step": 123610
+    },
+    {
+      "epoch": 0.8948438981664459,
+      "grad_norm": 0.14742349088191986,
+      "learning_rate": 4.105163340499614e-06,
+      "loss": 0.8749,
+      "step": 123620
+    },
+    {
+      "epoch": 0.8949162848270321,
+      "grad_norm": 0.15726658701896667,
+      "learning_rate": 4.105090953839027e-06,
+      "loss": 0.8714,
+      "step": 123630
+    },
+    {
+      "epoch": 0.8949886714876183,
+      "grad_norm": 0.1479179859161377,
+      "learning_rate": 4.105018567178441e-06,
+      "loss": 0.8764,
+      "step": 123640
+    },
+    {
+      "epoch": 0.8950610581482045,
+      "grad_norm": 0.14696402847766876,
+      "learning_rate": 4.1049461805178544e-06,
+      "loss": 0.8743,
+      "step": 123650
+    },
+    {
+      "epoch": 0.8951334448087906,
+      "grad_norm": 0.19279779493808746,
+      "learning_rate": 4.104873793857269e-06,
+      "loss": 0.876,
+      "step": 123660
+    },
+    {
+      "epoch": 0.8952058314693768,
+      "grad_norm": 0.1497318148612976,
+      "learning_rate": 4.1048014071966825e-06,
+      "loss": 0.8704,
+      "step": 123670
+    },
+    {
+      "epoch": 0.895278218129963,
+      "grad_norm": 0.16485776007175446,
+      "learning_rate": 4.104729020536095e-06,
+      "loss": 0.882,
+      "step": 123680
+    },
+    {
+      "epoch": 0.8953506047905492,
+      "grad_norm": 0.18929380178451538,
+      "learning_rate": 4.10465663387551e-06,
+      "loss": 0.8752,
+      "step": 123690
+    },
+    {
+      "epoch": 0.8954229914511354,
+      "grad_norm": 0.1474197655916214,
+      "learning_rate": 4.104584247214923e-06,
+      "loss": 0.8665,
+      "step": 123700
+    },
+    {
+      "epoch": 0.8954953781117215,
+      "grad_norm": 0.1479775607585907,
+      "learning_rate": 4.104511860554337e-06,
+      "loss": 0.8665,
+      "step": 123710
+    },
+    {
+      "epoch": 0.8955677647723077,
+      "grad_norm": 0.26325809955596924,
+      "learning_rate": 4.104439473893751e-06,
+      "loss": 0.8675,
+      "step": 123720
+    },
+    {
+      "epoch": 0.895640151432894,
+      "grad_norm": 0.16428937017917633,
+      "learning_rate": 4.104367087233165e-06,
+      "loss": 0.8761,
+      "step": 123730
+    },
+    {
+      "epoch": 0.8957125380934802,
+      "grad_norm": 0.14757543802261353,
+      "learning_rate": 4.104294700572579e-06,
+      "loss": 0.8641,
+      "step": 123740
+    },
+    {
+      "epoch": 0.8957849247540663,
+      "grad_norm": 0.15158720314502716,
+      "learning_rate": 4.104222313911992e-06,
+      "loss": 0.8716,
+      "step": 123750
+    },
+    {
+      "epoch": 0.8958573114146525,
+      "grad_norm": 0.15300332009792328,
+      "learning_rate": 4.104149927251406e-06,
+      "loss": 0.8746,
+      "step": 123760
+    },
+    {
+      "epoch": 0.8959296980752387,
+      "grad_norm": 0.14569737017154694,
+      "learning_rate": 4.10407754059082e-06,
+      "loss": 0.8769,
+      "step": 123770
+    },
+    {
+      "epoch": 0.8960020847358249,
+      "grad_norm": 0.1715804487466812,
+      "learning_rate": 4.104005153930234e-06,
+      "loss": 0.871,
+      "step": 123780
+    },
+    {
+      "epoch": 0.896074471396411,
+      "grad_norm": 0.149086594581604,
+      "learning_rate": 4.103932767269648e-06,
+      "loss": 0.8719,
+      "step": 123790
+    },
+    {
+      "epoch": 0.8961468580569972,
+      "grad_norm": 0.14175446331501007,
+      "learning_rate": 4.103860380609061e-06,
+      "loss": 0.8699,
+      "step": 123800
+    },
+    {
+      "epoch": 0.8962192447175834,
+      "grad_norm": 0.15866614878177643,
+      "learning_rate": 4.103787993948476e-06,
+      "loss": 0.8739,
+      "step": 123810
+    },
+    {
+      "epoch": 0.8962916313781696,
+      "grad_norm": 0.15101435780525208,
+      "learning_rate": 4.103715607287889e-06,
+      "loss": 0.8812,
+      "step": 123820
+    },
+    {
+      "epoch": 0.8963640180387559,
+      "grad_norm": 0.16456730663776398,
+      "learning_rate": 4.103643220627303e-06,
+      "loss": 0.8643,
+      "step": 123830
+    },
+    {
+      "epoch": 0.896436404699342,
+      "grad_norm": 0.15458862483501434,
+      "learning_rate": 4.1035708339667165e-06,
+      "loss": 0.8746,
+      "step": 123840
+    },
+    {
+      "epoch": 0.8965087913599282,
+      "grad_norm": 0.15625207126140594,
+      "learning_rate": 4.103498447306131e-06,
+      "loss": 0.8852,
+      "step": 123850
+    },
+    {
+      "epoch": 0.8965811780205144,
+      "grad_norm": 0.1557016670703888,
+      "learning_rate": 4.103426060645545e-06,
+      "loss": 0.8615,
+      "step": 123860
+    },
+    {
+      "epoch": 0.8966535646811006,
+      "grad_norm": 0.15922847390174866,
+      "learning_rate": 4.103353673984958e-06,
+      "loss": 0.8678,
+      "step": 123870
+    },
+    {
+      "epoch": 0.8967259513416868,
+      "grad_norm": 0.15058737993240356,
+      "learning_rate": 4.103281287324372e-06,
+      "loss": 0.8657,
+      "step": 123880
+    },
+    {
+      "epoch": 0.8967983380022729,
+      "grad_norm": 0.15507520735263824,
+      "learning_rate": 4.103208900663786e-06,
+      "loss": 0.8639,
+      "step": 123890
+    },
+    {
+      "epoch": 0.8968707246628591,
+      "grad_norm": 0.1978985220193863,
+      "learning_rate": 4.1031365140032e-06,
+      "loss": 0.8783,
+      "step": 123900
+    },
+    {
+      "epoch": 0.8969431113234453,
+      "grad_norm": 0.1425454467535019,
+      "learning_rate": 4.1030641273426135e-06,
+      "loss": 0.8741,
+      "step": 123910
+    },
+    {
+      "epoch": 0.8970154979840315,
+      "grad_norm": 0.14902305603027344,
+      "learning_rate": 4.102991740682027e-06,
+      "loss": 0.8698,
+      "step": 123920
+    },
+    {
+      "epoch": 0.8970878846446176,
+      "grad_norm": 0.1497499793767929,
+      "learning_rate": 4.102919354021442e-06,
+      "loss": 0.8657,
+      "step": 123930
+    },
+    {
+      "epoch": 0.8971602713052039,
+      "grad_norm": 0.16021916270256042,
+      "learning_rate": 4.102846967360855e-06,
+      "loss": 0.8808,
+      "step": 123940
+    },
+    {
+      "epoch": 0.8972326579657901,
+      "grad_norm": 0.14823076128959656,
+      "learning_rate": 4.102774580700269e-06,
+      "loss": 0.8701,
+      "step": 123950
+    },
+    {
+      "epoch": 0.8973050446263763,
+      "grad_norm": 0.16842210292816162,
+      "learning_rate": 4.1027021940396825e-06,
+      "loss": 0.8782,
+      "step": 123960
+    },
+    {
+      "epoch": 0.8973774312869625,
+      "grad_norm": 0.14396853744983673,
+      "learning_rate": 4.102629807379097e-06,
+      "loss": 0.8804,
+      "step": 123970
+    },
+    {
+      "epoch": 0.8974498179475486,
+      "grad_norm": 0.1455616056919098,
+      "learning_rate": 4.1025574207185105e-06,
+      "loss": 0.869,
+      "step": 123980
+    },
+    {
+      "epoch": 0.8975222046081348,
+      "grad_norm": 0.14303292334079742,
+      "learning_rate": 4.102485034057924e-06,
+      "loss": 0.8642,
+      "step": 123990
+    },
+    {
+      "epoch": 0.897594591268721,
+      "grad_norm": 0.15414045751094818,
+      "learning_rate": 4.102412647397338e-06,
+      "loss": 0.8639,
+      "step": 124000
+    },
+    {
+      "epoch": 0.8976669779293072,
+      "grad_norm": 0.15850482881069183,
+      "learning_rate": 4.102340260736752e-06,
+      "loss": 0.8639,
+      "step": 124010
+    },
+    {
+      "epoch": 0.8977393645898933,
+      "grad_norm": 0.15761762857437134,
+      "learning_rate": 4.102267874076166e-06,
+      "loss": 0.8722,
+      "step": 124020
+    },
+    {
+      "epoch": 0.8978117512504795,
+      "grad_norm": 0.1510854810476303,
+      "learning_rate": 4.1021954874155795e-06,
+      "loss": 0.8801,
+      "step": 124030
+    },
+    {
+      "epoch": 0.8978841379110657,
+      "grad_norm": 0.14972640573978424,
+      "learning_rate": 4.102123100754993e-06,
+      "loss": 0.8731,
+      "step": 124040
+    },
+    {
+      "epoch": 0.897956524571652,
+      "grad_norm": 0.15360723435878754,
+      "learning_rate": 4.1020507140944076e-06,
+      "loss": 0.8713,
+      "step": 124050
+    },
+    {
+      "epoch": 0.8980289112322382,
+      "grad_norm": 0.16010282933712006,
+      "learning_rate": 4.101978327433821e-06,
+      "loss": 0.8626,
+      "step": 124060
+    },
+    {
+      "epoch": 0.8981012978928243,
+      "grad_norm": 0.15653282403945923,
+      "learning_rate": 4.101905940773235e-06,
+      "loss": 0.8747,
+      "step": 124070
+    },
+    {
+      "epoch": 0.8981736845534105,
+      "grad_norm": 0.14662018418312073,
+      "learning_rate": 4.101833554112648e-06,
+      "loss": 0.8779,
+      "step": 124080
+    },
+    {
+      "epoch": 0.8982460712139967,
+      "grad_norm": 0.1570512056350708,
+      "learning_rate": 4.101761167452062e-06,
+      "loss": 0.8845,
+      "step": 124090
+    },
+    {
+      "epoch": 0.8983184578745829,
+      "grad_norm": 0.16235743463039398,
+      "learning_rate": 4.1016887807914765e-06,
+      "loss": 0.8702,
+      "step": 124100
+    },
+    {
+      "epoch": 0.898390844535169,
+      "grad_norm": 0.1602431684732437,
+      "learning_rate": 4.10161639413089e-06,
+      "loss": 0.8693,
+      "step": 124110
+    },
+    {
+      "epoch": 0.8984632311957552,
+      "grad_norm": 0.15021947026252747,
+      "learning_rate": 4.101544007470304e-06,
+      "loss": 0.8704,
+      "step": 124120
+    },
+    {
+      "epoch": 0.8985356178563414,
+      "grad_norm": 0.17709344625473022,
+      "learning_rate": 4.101471620809717e-06,
+      "loss": 0.8725,
+      "step": 124130
+    },
+    {
+      "epoch": 0.8986080045169276,
+      "grad_norm": 0.3144311308860779,
+      "learning_rate": 4.101399234149132e-06,
+      "loss": 0.8644,
+      "step": 124140
+    },
+    {
+      "epoch": 0.8986803911775139,
+      "grad_norm": 0.1658957153558731,
+      "learning_rate": 4.101326847488545e-06,
+      "loss": 0.8645,
+      "step": 124150
+    },
+    {
+      "epoch": 0.8987527778381,
+      "grad_norm": 0.1864367425441742,
+      "learning_rate": 4.101254460827959e-06,
+      "loss": 0.8769,
+      "step": 124160
+    },
+    {
+      "epoch": 0.8988251644986862,
+      "grad_norm": 0.1641472429037094,
+      "learning_rate": 4.101182074167373e-06,
+      "loss": 0.8743,
+      "step": 124170
+    },
+    {
+      "epoch": 0.8988975511592724,
+      "grad_norm": 0.156347393989563,
+      "learning_rate": 4.101109687506787e-06,
+      "loss": 0.863,
+      "step": 124180
+    },
+    {
+      "epoch": 0.8989699378198586,
+      "grad_norm": 0.14861541986465454,
+      "learning_rate": 4.101037300846201e-06,
+      "loss": 0.8728,
+      "step": 124190
+    },
+    {
+      "epoch": 0.8990423244804447,
+      "grad_norm": 0.15074244141578674,
+      "learning_rate": 4.100964914185614e-06,
+      "loss": 0.8771,
+      "step": 124200
+    },
+    {
+      "epoch": 0.8991147111410309,
+      "grad_norm": 0.14890417456626892,
+      "learning_rate": 4.100892527525028e-06,
+      "loss": 0.8639,
+      "step": 124210
+    },
+    {
+      "epoch": 0.8991870978016171,
+      "grad_norm": 0.1627875566482544,
+      "learning_rate": 4.1008201408644416e-06,
+      "loss": 0.8691,
+      "step": 124220
+    },
+    {
+      "epoch": 0.8992594844622033,
+      "grad_norm": 0.15136182308197021,
+      "learning_rate": 4.100747754203855e-06,
+      "loss": 0.8836,
+      "step": 124230
+    },
+    {
+      "epoch": 0.8993318711227895,
+      "grad_norm": 0.14906705915927887,
+      "learning_rate": 4.100675367543269e-06,
+      "loss": 0.8721,
+      "step": 124240
+    },
+    {
+      "epoch": 0.8994042577833756,
+      "grad_norm": 0.14865903556346893,
+      "learning_rate": 4.100602980882683e-06,
+      "loss": 0.873,
+      "step": 124250
+    },
+    {
+      "epoch": 0.8994766444439619,
+      "grad_norm": 0.19429634511470795,
+      "learning_rate": 4.100530594222097e-06,
+      "loss": 0.8822,
+      "step": 124260
+    },
+    {
+      "epoch": 0.8995490311045481,
+      "grad_norm": 0.16122131049633026,
+      "learning_rate": 4.1004582075615105e-06,
+      "loss": 0.874,
+      "step": 124270
+    },
+    {
+      "epoch": 0.8996214177651343,
+      "grad_norm": 0.14246490597724915,
+      "learning_rate": 4.100385820900924e-06,
+      "loss": 0.8657,
+      "step": 124280
+    },
+    {
+      "epoch": 0.8996938044257204,
+      "grad_norm": 0.15221592783927917,
+      "learning_rate": 4.100313434240339e-06,
+      "loss": 0.8718,
+      "step": 124290
+    },
+    {
+      "epoch": 0.8997661910863066,
+      "grad_norm": 0.1513119488954544,
+      "learning_rate": 4.100241047579752e-06,
+      "loss": 0.8552,
+      "step": 124300
+    },
+    {
+      "epoch": 0.8998385777468928,
+      "grad_norm": 0.1578131765127182,
+      "learning_rate": 4.100168660919166e-06,
+      "loss": 0.873,
+      "step": 124310
+    },
+    {
+      "epoch": 0.899910964407479,
+      "grad_norm": 0.1699373573064804,
+      "learning_rate": 4.1000962742585794e-06,
+      "loss": 0.883,
+      "step": 124320
+    },
+    {
+      "epoch": 0.8999833510680652,
+      "grad_norm": 0.16044111549854279,
+      "learning_rate": 4.100023887597994e-06,
+      "loss": 0.8759,
+      "step": 124330
+    },
+    {
+      "epoch": 0.9000557377286513,
+      "grad_norm": 0.16332702338695526,
+      "learning_rate": 4.0999515009374075e-06,
+      "loss": 0.8708,
+      "step": 124340
+    },
+    {
+      "epoch": 0.9001281243892375,
+      "grad_norm": 0.15864451229572296,
+      "learning_rate": 4.099879114276821e-06,
+      "loss": 0.8728,
+      "step": 124350
+    },
+    {
+      "epoch": 0.9002005110498238,
+      "grad_norm": 0.16290296614170074,
+      "learning_rate": 4.099806727616235e-06,
+      "loss": 0.8597,
+      "step": 124360
+    },
+    {
+      "epoch": 0.90027289771041,
+      "grad_norm": 0.16254830360412598,
+      "learning_rate": 4.099734340955649e-06,
+      "loss": 0.8672,
+      "step": 124370
+    },
+    {
+      "epoch": 0.9003452843709961,
+      "grad_norm": 0.15096203982830048,
+      "learning_rate": 4.099661954295063e-06,
+      "loss": 0.884,
+      "step": 124380
+    },
+    {
+      "epoch": 0.9004176710315823,
+      "grad_norm": 0.1599651724100113,
+      "learning_rate": 4.0995895676344764e-06,
+      "loss": 0.8764,
+      "step": 124390
+    },
+    {
+      "epoch": 0.9004900576921685,
+      "grad_norm": 0.1575443595647812,
+      "learning_rate": 4.09951718097389e-06,
+      "loss": 0.8833,
+      "step": 124400
+    },
+    {
+      "epoch": 0.9005624443527547,
+      "grad_norm": 0.15898630023002625,
+      "learning_rate": 4.0994447943133045e-06,
+      "loss": 0.8664,
+      "step": 124410
+    },
+    {
+      "epoch": 0.9006348310133409,
+      "grad_norm": 0.14191798865795135,
+      "learning_rate": 4.099372407652718e-06,
+      "loss": 0.8758,
+      "step": 124420
+    },
+    {
+      "epoch": 0.900707217673927,
+      "grad_norm": 0.15920232236385345,
+      "learning_rate": 4.099300020992132e-06,
+      "loss": 0.8628,
+      "step": 124430
+    },
+    {
+      "epoch": 0.9007796043345132,
+      "grad_norm": 0.15842455625534058,
+      "learning_rate": 4.099227634331545e-06,
+      "loss": 0.8698,
+      "step": 124440
+    },
+    {
+      "epoch": 0.9008519909950994,
+      "grad_norm": 0.1533687710762024,
+      "learning_rate": 4.09915524767096e-06,
+      "loss": 0.8723,
+      "step": 124450
+    },
+    {
+      "epoch": 0.9009243776556856,
+      "grad_norm": 0.16112767159938812,
+      "learning_rate": 4.0990828610103734e-06,
+      "loss": 0.8559,
+      "step": 124460
+    },
+    {
+      "epoch": 0.9009967643162718,
+      "grad_norm": 0.15855056047439575,
+      "learning_rate": 4.099010474349787e-06,
+      "loss": 0.8793,
+      "step": 124470
+    },
+    {
+      "epoch": 0.901069150976858,
+      "grad_norm": 0.15525835752487183,
+      "learning_rate": 4.098938087689201e-06,
+      "loss": 0.8706,
+      "step": 124480
+    },
+    {
+      "epoch": 0.9011415376374442,
+      "grad_norm": 0.14787094295024872,
+      "learning_rate": 4.098865701028615e-06,
+      "loss": 0.8685,
+      "step": 124490
+    },
+    {
+      "epoch": 0.9012139242980304,
+      "grad_norm": 0.17414870858192444,
+      "learning_rate": 4.098793314368029e-06,
+      "loss": 0.8715,
+      "step": 124500
+    },
+    {
+      "epoch": 0.9012863109586166,
+      "grad_norm": 0.15117833018302917,
+      "learning_rate": 4.098720927707442e-06,
+      "loss": 0.8646,
+      "step": 124510
+    },
+    {
+      "epoch": 0.9013586976192027,
+      "grad_norm": 0.22992941737174988,
+      "learning_rate": 4.098648541046856e-06,
+      "loss": 0.8764,
+      "step": 124520
+    },
+    {
+      "epoch": 0.9014310842797889,
+      "grad_norm": 0.15416789054870605,
+      "learning_rate": 4.0985761543862705e-06,
+      "loss": 0.881,
+      "step": 124530
+    },
+    {
+      "epoch": 0.9015034709403751,
+      "grad_norm": 0.17136387526988983,
+      "learning_rate": 4.098503767725684e-06,
+      "loss": 0.8801,
+      "step": 124540
+    },
+    {
+      "epoch": 0.9015758576009613,
+      "grad_norm": 0.1540147364139557,
+      "learning_rate": 4.098431381065098e-06,
+      "loss": 0.8798,
+      "step": 124550
+    },
+    {
+      "epoch": 0.9016482442615474,
+      "grad_norm": 0.15776564180850983,
+      "learning_rate": 4.098358994404511e-06,
+      "loss": 0.8718,
+      "step": 124560
+    },
+    {
+      "epoch": 0.9017206309221336,
+      "grad_norm": 0.15084987878799438,
+      "learning_rate": 4.098286607743926e-06,
+      "loss": 0.8809,
+      "step": 124570
+    },
+    {
+      "epoch": 0.9017930175827199,
+      "grad_norm": 0.15924064815044403,
+      "learning_rate": 4.098214221083339e-06,
+      "loss": 0.8815,
+      "step": 124580
+    },
+    {
+      "epoch": 0.9018654042433061,
+      "grad_norm": 0.14319369196891785,
+      "learning_rate": 4.098141834422753e-06,
+      "loss": 0.8619,
+      "step": 124590
+    },
+    {
+      "epoch": 0.9019377909038923,
+      "grad_norm": 0.1575179100036621,
+      "learning_rate": 4.098069447762167e-06,
+      "loss": 0.8651,
+      "step": 124600
+    },
+    {
+      "epoch": 0.9020101775644784,
+      "grad_norm": 0.1562921404838562,
+      "learning_rate": 4.097997061101581e-06,
+      "loss": 0.8687,
+      "step": 124610
+    },
+    {
+      "epoch": 0.9020825642250646,
+      "grad_norm": 0.1615796834230423,
+      "learning_rate": 4.097924674440995e-06,
+      "loss": 0.8539,
+      "step": 124620
+    },
+    {
+      "epoch": 0.9021549508856508,
+      "grad_norm": 0.14882075786590576,
+      "learning_rate": 4.097852287780408e-06,
+      "loss": 0.8667,
+      "step": 124630
+    },
+    {
+      "epoch": 0.902227337546237,
+      "grad_norm": 0.15828663110733032,
+      "learning_rate": 4.097779901119822e-06,
+      "loss": 0.8749,
+      "step": 124640
+    },
+    {
+      "epoch": 0.9022997242068231,
+      "grad_norm": 0.15658116340637207,
+      "learning_rate": 4.097707514459236e-06,
+      "loss": 0.8891,
+      "step": 124650
+    },
+    {
+      "epoch": 0.9023721108674093,
+      "grad_norm": 0.15174657106399536,
+      "learning_rate": 4.09763512779865e-06,
+      "loss": 0.8691,
+      "step": 124660
+    },
+    {
+      "epoch": 0.9024444975279955,
+      "grad_norm": 0.14801457524299622,
+      "learning_rate": 4.097562741138064e-06,
+      "loss": 0.8774,
+      "step": 124670
+    },
+    {
+      "epoch": 0.9025168841885818,
+      "grad_norm": 0.17308656871318817,
+      "learning_rate": 4.097490354477477e-06,
+      "loss": 0.8792,
+      "step": 124680
+    },
+    {
+      "epoch": 0.902589270849168,
+      "grad_norm": 0.13914184272289276,
+      "learning_rate": 4.097417967816891e-06,
+      "loss": 0.8703,
+      "step": 124690
+    },
+    {
+      "epoch": 0.9026616575097541,
+      "grad_norm": 0.16653895378112793,
+      "learning_rate": 4.097345581156305e-06,
+      "loss": 0.8584,
+      "step": 124700
+    },
+    {
+      "epoch": 0.9027340441703403,
+      "grad_norm": 0.15545400977134705,
+      "learning_rate": 4.097273194495719e-06,
+      "loss": 0.8746,
+      "step": 124710
+    },
+    {
+      "epoch": 0.9028064308309265,
+      "grad_norm": 0.15247924625873566,
+      "learning_rate": 4.0972008078351325e-06,
+      "loss": 0.8756,
+      "step": 124720
+    },
+    {
+      "epoch": 0.9028788174915127,
+      "grad_norm": 0.15128053724765778,
+      "learning_rate": 4.097128421174546e-06,
+      "loss": 0.8757,
+      "step": 124730
+    },
+    {
+      "epoch": 0.9029512041520988,
+      "grad_norm": 0.1656970977783203,
+      "learning_rate": 4.09705603451396e-06,
+      "loss": 0.864,
+      "step": 124740
+    },
+    {
+      "epoch": 0.903023590812685,
+      "grad_norm": 0.15432408452033997,
+      "learning_rate": 4.096983647853373e-06,
+      "loss": 0.8673,
+      "step": 124750
+    },
+    {
+      "epoch": 0.9030959774732712,
+      "grad_norm": 0.14504657685756683,
+      "learning_rate": 4.096911261192788e-06,
+      "loss": 0.8722,
+      "step": 124760
+    },
+    {
+      "epoch": 0.9031683641338574,
+      "grad_norm": 0.181019127368927,
+      "learning_rate": 4.0968388745322015e-06,
+      "loss": 0.8777,
+      "step": 124770
+    },
+    {
+      "epoch": 0.9032407507944435,
+      "grad_norm": 0.1841985583305359,
+      "learning_rate": 4.096766487871615e-06,
+      "loss": 0.8764,
+      "step": 124780
+    },
+    {
+      "epoch": 0.9033131374550298,
+      "grad_norm": 0.14566998183727264,
+      "learning_rate": 4.096694101211029e-06,
+      "loss": 0.8788,
+      "step": 124790
+    },
+    {
+      "epoch": 0.903385524115616,
+      "grad_norm": 0.15821722149848938,
+      "learning_rate": 4.096621714550443e-06,
+      "loss": 0.8675,
+      "step": 124800
+    },
+    {
+      "epoch": 0.9034579107762022,
+      "grad_norm": 0.14918698370456696,
+      "learning_rate": 4.096549327889857e-06,
+      "loss": 0.8612,
+      "step": 124810
+    },
+    {
+      "epoch": 0.9035302974367884,
+      "grad_norm": 0.16349859535694122,
+      "learning_rate": 4.09647694122927e-06,
+      "loss": 0.8707,
+      "step": 124820
+    },
+    {
+      "epoch": 0.9036026840973745,
+      "grad_norm": 0.15375465154647827,
+      "learning_rate": 4.096404554568684e-06,
+      "loss": 0.8806,
+      "step": 124830
+    },
+    {
+      "epoch": 0.9036750707579607,
+      "grad_norm": 0.152941033244133,
+      "learning_rate": 4.0963321679080985e-06,
+      "loss": 0.881,
+      "step": 124840
+    },
+    {
+      "epoch": 0.9037474574185469,
+      "grad_norm": 0.22806885838508606,
+      "learning_rate": 4.096259781247512e-06,
+      "loss": 0.8626,
+      "step": 124850
+    },
+    {
+      "epoch": 0.9038198440791331,
+      "grad_norm": 0.15706026554107666,
+      "learning_rate": 4.096187394586926e-06,
+      "loss": 0.8711,
+      "step": 124860
+    },
+    {
+      "epoch": 0.9038922307397192,
+      "grad_norm": 0.1524626761674881,
+      "learning_rate": 4.096115007926339e-06,
+      "loss": 0.8779,
+      "step": 124870
+    },
+    {
+      "epoch": 0.9039646174003054,
+      "grad_norm": 0.16618412733078003,
+      "learning_rate": 4.096042621265753e-06,
+      "loss": 0.8715,
+      "step": 124880
+    },
+    {
+      "epoch": 0.9040370040608917,
+      "grad_norm": 0.1479671746492386,
+      "learning_rate": 4.095970234605167e-06,
+      "loss": 0.8798,
+      "step": 124890
+    },
+    {
+      "epoch": 0.9041093907214779,
+      "grad_norm": 0.22761158645153046,
+      "learning_rate": 4.095897847944581e-06,
+      "loss": 0.8708,
+      "step": 124900
+    },
+    {
+      "epoch": 0.9041817773820641,
+      "grad_norm": 0.15918904542922974,
+      "learning_rate": 4.095825461283995e-06,
+      "loss": 0.8679,
+      "step": 124910
+    },
+    {
+      "epoch": 0.9042541640426502,
+      "grad_norm": 0.16984111070632935,
+      "learning_rate": 4.095753074623408e-06,
+      "loss": 0.8686,
+      "step": 124920
+    },
+    {
+      "epoch": 0.9043265507032364,
+      "grad_norm": 0.15530432760715485,
+      "learning_rate": 4.095680687962823e-06,
+      "loss": 0.8733,
+      "step": 124930
+    },
+    {
+      "epoch": 0.9043989373638226,
+      "grad_norm": 0.15321482717990875,
+      "learning_rate": 4.095608301302236e-06,
+      "loss": 0.868,
+      "step": 124940
+    },
+    {
+      "epoch": 0.9044713240244088,
+      "grad_norm": 0.15977735817432404,
+      "learning_rate": 4.09553591464165e-06,
+      "loss": 0.8704,
+      "step": 124950
+    },
+    {
+      "epoch": 0.904543710684995,
+      "grad_norm": 0.14585714042186737,
+      "learning_rate": 4.0954635279810636e-06,
+      "loss": 0.8688,
+      "step": 124960
+    },
+    {
+      "epoch": 0.9046160973455811,
+      "grad_norm": 0.16191209852695465,
+      "learning_rate": 4.095391141320478e-06,
+      "loss": 0.8638,
+      "step": 124970
+    },
+    {
+      "epoch": 0.9046884840061673,
+      "grad_norm": 0.27564188838005066,
+      "learning_rate": 4.095318754659892e-06,
+      "loss": 0.8708,
+      "step": 124980
+    },
+    {
+      "epoch": 0.9047608706667535,
+      "grad_norm": 0.14919497072696686,
+      "learning_rate": 4.095246367999305e-06,
+      "loss": 0.8662,
+      "step": 124990
+    },
+    {
+      "epoch": 0.9048332573273398,
+      "grad_norm": 0.1528129130601883,
+      "learning_rate": 4.095173981338719e-06,
+      "loss": 0.8759,
+      "step": 125000
+    },
+    {
+      "epoch": 0.904905643987926,
+      "grad_norm": 0.15867576003074646,
+      "learning_rate": 4.095101594678133e-06,
+      "loss": 0.888,
+      "step": 125010
+    },
+    {
+      "epoch": 0.9049780306485121,
+      "grad_norm": 0.16656945645809174,
+      "learning_rate": 4.095029208017547e-06,
+      "loss": 0.8661,
+      "step": 125020
+    },
+    {
+      "epoch": 0.9050504173090983,
+      "grad_norm": 0.1601569950580597,
+      "learning_rate": 4.094956821356961e-06,
+      "loss": 0.8645,
+      "step": 125030
+    },
+    {
+      "epoch": 0.9051228039696845,
+      "grad_norm": 0.1524617075920105,
+      "learning_rate": 4.094884434696374e-06,
+      "loss": 0.8727,
+      "step": 125040
+    },
+    {
+      "epoch": 0.9051951906302707,
+      "grad_norm": 0.15600845217704773,
+      "learning_rate": 4.094812048035789e-06,
+      "loss": 0.8757,
+      "step": 125050
+    },
+    {
+      "epoch": 0.9052675772908568,
+      "grad_norm": 0.15215645730495453,
+      "learning_rate": 4.094739661375202e-06,
+      "loss": 0.8556,
+      "step": 125060
+    },
+    {
+      "epoch": 0.905339963951443,
+      "grad_norm": 0.14857865869998932,
+      "learning_rate": 4.094667274714616e-06,
+      "loss": 0.8629,
+      "step": 125070
+    },
+    {
+      "epoch": 0.9054123506120292,
+      "grad_norm": 0.15590626001358032,
+      "learning_rate": 4.0945948880540295e-06,
+      "loss": 0.8733,
+      "step": 125080
+    },
+    {
+      "epoch": 0.9054847372726154,
+      "grad_norm": 0.15361756086349487,
+      "learning_rate": 4.094522501393444e-06,
+      "loss": 0.8775,
+      "step": 125090
+    },
+    {
+      "epoch": 0.9055571239332015,
+      "grad_norm": 0.14809496700763702,
+      "learning_rate": 4.094450114732858e-06,
+      "loss": 0.8752,
+      "step": 125100
+    },
+    {
+      "epoch": 0.9056295105937878,
+      "grad_norm": 0.1699819564819336,
+      "learning_rate": 4.094377728072271e-06,
+      "loss": 0.8768,
+      "step": 125110
+    },
+    {
+      "epoch": 0.905701897254374,
+      "grad_norm": 0.3279818892478943,
+      "learning_rate": 4.094305341411685e-06,
+      "loss": 0.8726,
+      "step": 125120
+    },
+    {
+      "epoch": 0.9057742839149602,
+      "grad_norm": 0.14810748398303986,
+      "learning_rate": 4.094232954751099e-06,
+      "loss": 0.874,
+      "step": 125130
+    },
+    {
+      "epoch": 0.9058466705755464,
+      "grad_norm": 0.14194615185260773,
+      "learning_rate": 4.094160568090513e-06,
+      "loss": 0.8608,
+      "step": 125140
+    },
+    {
+      "epoch": 0.9059190572361325,
+      "grad_norm": 0.1805509328842163,
+      "learning_rate": 4.0940881814299265e-06,
+      "loss": 0.8603,
+      "step": 125150
+    },
+    {
+      "epoch": 0.9059914438967187,
+      "grad_norm": 0.15983721613883972,
+      "learning_rate": 4.09401579476934e-06,
+      "loss": 0.8724,
+      "step": 125160
+    },
+    {
+      "epoch": 0.9060638305573049,
+      "grad_norm": 0.15340054035186768,
+      "learning_rate": 4.093943408108755e-06,
+      "loss": 0.8659,
+      "step": 125170
+    },
+    {
+      "epoch": 0.9061362172178911,
+      "grad_norm": 0.14701798558235168,
+      "learning_rate": 4.093871021448168e-06,
+      "loss": 0.8699,
+      "step": 125180
+    },
+    {
+      "epoch": 0.9062086038784772,
+      "grad_norm": 0.14838139712810516,
+      "learning_rate": 4.093798634787582e-06,
+      "loss": 0.8712,
+      "step": 125190
+    },
+    {
+      "epoch": 0.9062809905390634,
+      "grad_norm": 0.1553456038236618,
+      "learning_rate": 4.0937262481269954e-06,
+      "loss": 0.8742,
+      "step": 125200
+    },
+    {
+      "epoch": 0.9063533771996497,
+      "grad_norm": 0.15716488659381866,
+      "learning_rate": 4.09365386146641e-06,
+      "loss": 0.8726,
+      "step": 125210
+    },
+    {
+      "epoch": 0.9064257638602359,
+      "grad_norm": 0.15702371299266815,
+      "learning_rate": 4.0935814748058235e-06,
+      "loss": 0.8847,
+      "step": 125220
+    },
+    {
+      "epoch": 0.906498150520822,
+      "grad_norm": 0.16816936433315277,
+      "learning_rate": 4.093509088145237e-06,
+      "loss": 0.8732,
+      "step": 125230
+    },
+    {
+      "epoch": 0.9065705371814082,
+      "grad_norm": 0.14697718620300293,
+      "learning_rate": 4.093436701484651e-06,
+      "loss": 0.8725,
+      "step": 125240
+    },
+    {
+      "epoch": 0.9066429238419944,
+      "grad_norm": 0.1436859369277954,
+      "learning_rate": 4.093364314824065e-06,
+      "loss": 0.8681,
+      "step": 125250
+    },
+    {
+      "epoch": 0.9067153105025806,
+      "grad_norm": 0.1546768695116043,
+      "learning_rate": 4.093291928163479e-06,
+      "loss": 0.8904,
+      "step": 125260
+    },
+    {
+      "epoch": 0.9067876971631668,
+      "grad_norm": 0.16405488550662994,
+      "learning_rate": 4.093219541502892e-06,
+      "loss": 0.8736,
+      "step": 125270
+    },
+    {
+      "epoch": 0.9068600838237529,
+      "grad_norm": 0.15231408178806305,
+      "learning_rate": 4.093147154842306e-06,
+      "loss": 0.8566,
+      "step": 125280
+    },
+    {
+      "epoch": 0.9069324704843391,
+      "grad_norm": 0.15230032801628113,
+      "learning_rate": 4.09307476818172e-06,
+      "loss": 0.8713,
+      "step": 125290
+    },
+    {
+      "epoch": 0.9070048571449253,
+      "grad_norm": 0.15494680404663086,
+      "learning_rate": 4.093002381521133e-06,
+      "loss": 0.8725,
+      "step": 125300
+    },
+    {
+      "epoch": 0.9070772438055115,
+      "grad_norm": 0.17046596109867096,
+      "learning_rate": 4.092929994860547e-06,
+      "loss": 0.8698,
+      "step": 125310
+    },
+    {
+      "epoch": 0.9071496304660978,
+      "grad_norm": 0.16179198026657104,
+      "learning_rate": 4.092857608199961e-06,
+      "loss": 0.8697,
+      "step": 125320
+    },
+    {
+      "epoch": 0.9072220171266839,
+      "grad_norm": 0.15248672664165497,
+      "learning_rate": 4.092785221539375e-06,
+      "loss": 0.8528,
+      "step": 125330
+    },
+    {
+      "epoch": 0.9072944037872701,
+      "grad_norm": 0.1498788446187973,
+      "learning_rate": 4.092712834878789e-06,
+      "loss": 0.8753,
+      "step": 125340
+    },
+    {
+      "epoch": 0.9073667904478563,
+      "grad_norm": 0.1529855877161026,
+      "learning_rate": 4.092640448218202e-06,
+      "loss": 0.8749,
+      "step": 125350
+    },
+    {
+      "epoch": 0.9074391771084425,
+      "grad_norm": 0.154828742146492,
+      "learning_rate": 4.092568061557617e-06,
+      "loss": 0.8729,
+      "step": 125360
+    },
+    {
+      "epoch": 0.9075115637690286,
+      "grad_norm": 0.15640468895435333,
+      "learning_rate": 4.09249567489703e-06,
+      "loss": 0.8613,
+      "step": 125370
+    },
+    {
+      "epoch": 0.9075839504296148,
+      "grad_norm": 0.16197548806667328,
+      "learning_rate": 4.092423288236444e-06,
+      "loss": 0.8664,
+      "step": 125380
+    },
+    {
+      "epoch": 0.907656337090201,
+      "grad_norm": 0.15120939910411835,
+      "learning_rate": 4.0923509015758575e-06,
+      "loss": 0.8811,
+      "step": 125390
+    },
+    {
+      "epoch": 0.9077287237507872,
+      "grad_norm": 0.14062069356441498,
+      "learning_rate": 4.092278514915272e-06,
+      "loss": 0.8531,
+      "step": 125400
+    },
+    {
+      "epoch": 0.9078011104113733,
+      "grad_norm": 0.15563814342021942,
+      "learning_rate": 4.092206128254686e-06,
+      "loss": 0.8861,
+      "step": 125410
+    },
+    {
+      "epoch": 0.9078734970719596,
+      "grad_norm": 0.16093406081199646,
+      "learning_rate": 4.092133741594099e-06,
+      "loss": 0.8706,
+      "step": 125420
+    },
+    {
+      "epoch": 0.9079458837325458,
+      "grad_norm": 0.15003357827663422,
+      "learning_rate": 4.092061354933513e-06,
+      "loss": 0.8709,
+      "step": 125430
+    },
+    {
+      "epoch": 0.908018270393132,
+      "grad_norm": 0.16230422258377075,
+      "learning_rate": 4.091988968272927e-06,
+      "loss": 0.8653,
+      "step": 125440
+    },
+    {
+      "epoch": 0.9080906570537182,
+      "grad_norm": 0.14883364737033844,
+      "learning_rate": 4.091916581612341e-06,
+      "loss": 0.8632,
+      "step": 125450
+    },
+    {
+      "epoch": 0.9081630437143043,
+      "grad_norm": 0.15273550152778625,
+      "learning_rate": 4.0918441949517545e-06,
+      "loss": 0.8826,
+      "step": 125460
+    },
+    {
+      "epoch": 0.9082354303748905,
+      "grad_norm": 0.1566764861345291,
+      "learning_rate": 4.091771808291168e-06,
+      "loss": 0.8754,
+      "step": 125470
+    },
+    {
+      "epoch": 0.9083078170354767,
+      "grad_norm": 0.17311665415763855,
+      "learning_rate": 4.091699421630582e-06,
+      "loss": 0.8671,
+      "step": 125480
+    },
+    {
+      "epoch": 0.9083802036960629,
+      "grad_norm": 0.15010863542556763,
+      "learning_rate": 4.091627034969996e-06,
+      "loss": 0.8626,
+      "step": 125490
+    },
+    {
+      "epoch": 0.908452590356649,
+      "grad_norm": 0.1590012162923813,
+      "learning_rate": 4.09155464830941e-06,
+      "loss": 0.8547,
+      "step": 125500
+    },
+    {
+      "epoch": 0.9085249770172352,
+      "grad_norm": 0.6873774528503418,
+      "learning_rate": 4.0914822616488235e-06,
+      "loss": 0.8869,
+      "step": 125510
+    },
+    {
+      "epoch": 0.9085973636778214,
+      "grad_norm": 0.1598593294620514,
+      "learning_rate": 4.091409874988237e-06,
+      "loss": 0.8807,
+      "step": 125520
+    },
+    {
+      "epoch": 0.9086697503384077,
+      "grad_norm": 0.1523318886756897,
+      "learning_rate": 4.0913374883276516e-06,
+      "loss": 0.8734,
+      "step": 125530
+    },
+    {
+      "epoch": 0.9087421369989939,
+      "grad_norm": 0.15472367405891418,
+      "learning_rate": 4.091265101667065e-06,
+      "loss": 0.8717,
+      "step": 125540
+    },
+    {
+      "epoch": 0.90881452365958,
+      "grad_norm": 0.15248602628707886,
+      "learning_rate": 4.091192715006479e-06,
+      "loss": 0.8718,
+      "step": 125550
+    },
+    {
+      "epoch": 0.9088869103201662,
+      "grad_norm": 0.15343599021434784,
+      "learning_rate": 4.091120328345892e-06,
+      "loss": 0.878,
+      "step": 125560
+    },
+    {
+      "epoch": 0.9089592969807524,
+      "grad_norm": 0.16457286477088928,
+      "learning_rate": 4.091047941685307e-06,
+      "loss": 0.8691,
+      "step": 125570
+    },
+    {
+      "epoch": 0.9090316836413386,
+      "grad_norm": 0.15641097724437714,
+      "learning_rate": 4.0909755550247205e-06,
+      "loss": 0.8795,
+      "step": 125580
+    },
+    {
+      "epoch": 0.9091040703019247,
+      "grad_norm": 0.18575336039066315,
+      "learning_rate": 4.090903168364134e-06,
+      "loss": 0.8739,
+      "step": 125590
+    },
+    {
+      "epoch": 0.9091764569625109,
+      "grad_norm": 0.1508977711200714,
+      "learning_rate": 4.090830781703548e-06,
+      "loss": 0.8898,
+      "step": 125600
+    },
+    {
+      "epoch": 0.9092488436230971,
+      "grad_norm": 0.14920225739479065,
+      "learning_rate": 4.090758395042962e-06,
+      "loss": 0.8587,
+      "step": 125610
+    },
+    {
+      "epoch": 0.9093212302836833,
+      "grad_norm": 0.15833619236946106,
+      "learning_rate": 4.090686008382376e-06,
+      "loss": 0.8692,
+      "step": 125620
+    },
+    {
+      "epoch": 0.9093936169442695,
+      "grad_norm": 0.15184055268764496,
+      "learning_rate": 4.090613621721789e-06,
+      "loss": 0.8775,
+      "step": 125630
+    },
+    {
+      "epoch": 0.9094660036048557,
+      "grad_norm": 0.14898662269115448,
+      "learning_rate": 4.090541235061203e-06,
+      "loss": 0.8754,
+      "step": 125640
+    },
+    {
+      "epoch": 0.9095383902654419,
+      "grad_norm": 0.16337692737579346,
+      "learning_rate": 4.0904688484006175e-06,
+      "loss": 0.875,
+      "step": 125650
+    },
+    {
+      "epoch": 0.9096107769260281,
+      "grad_norm": 0.15499453246593475,
+      "learning_rate": 4.090396461740031e-06,
+      "loss": 0.8584,
+      "step": 125660
+    },
+    {
+      "epoch": 0.9096831635866143,
+      "grad_norm": 0.15772931277751923,
+      "learning_rate": 4.090324075079445e-06,
+      "loss": 0.8766,
+      "step": 125670
+    },
+    {
+      "epoch": 0.9097555502472005,
+      "grad_norm": 0.14884252846240997,
+      "learning_rate": 4.090251688418858e-06,
+      "loss": 0.8841,
+      "step": 125680
+    },
+    {
+      "epoch": 0.9098279369077866,
+      "grad_norm": 0.14398325979709625,
+      "learning_rate": 4.090179301758273e-06,
+      "loss": 0.876,
+      "step": 125690
+    },
+    {
+      "epoch": 0.9099003235683728,
+      "grad_norm": 0.1522645205259323,
+      "learning_rate": 4.090106915097686e-06,
+      "loss": 0.8766,
+      "step": 125700
+    },
+    {
+      "epoch": 0.909972710228959,
+      "grad_norm": 0.18375910818576813,
+      "learning_rate": 4.0900345284371e-06,
+      "loss": 0.8715,
+      "step": 125710
+    },
+    {
+      "epoch": 0.9100450968895452,
+      "grad_norm": 0.1611146479845047,
+      "learning_rate": 4.089962141776514e-06,
+      "loss": 0.8661,
+      "step": 125720
+    },
+    {
+      "epoch": 0.9101174835501313,
+      "grad_norm": 0.2902297079563141,
+      "learning_rate": 4.089889755115928e-06,
+      "loss": 0.8631,
+      "step": 125730
+    },
+    {
+      "epoch": 0.9101898702107176,
+      "grad_norm": 0.18060381710529327,
+      "learning_rate": 4.089817368455342e-06,
+      "loss": 0.8654,
+      "step": 125740
+    },
+    {
+      "epoch": 0.9102622568713038,
+      "grad_norm": 0.15165190398693085,
+      "learning_rate": 4.089744981794755e-06,
+      "loss": 0.8734,
+      "step": 125750
+    },
+    {
+      "epoch": 0.91033464353189,
+      "grad_norm": 0.16775788366794586,
+      "learning_rate": 4.089672595134169e-06,
+      "loss": 0.8643,
+      "step": 125760
+    },
+    {
+      "epoch": 0.9104070301924762,
+      "grad_norm": 0.1607564091682434,
+      "learning_rate": 4.0896002084735834e-06,
+      "loss": 0.8653,
+      "step": 125770
+    },
+    {
+      "epoch": 0.9104794168530623,
+      "grad_norm": 0.15660177171230316,
+      "learning_rate": 4.089527821812997e-06,
+      "loss": 0.8727,
+      "step": 125780
+    },
+    {
+      "epoch": 0.9105518035136485,
+      "grad_norm": 0.14147257804870605,
+      "learning_rate": 4.089455435152411e-06,
+      "loss": 0.8755,
+      "step": 125790
+    },
+    {
+      "epoch": 0.9106241901742347,
+      "grad_norm": 0.1649179309606552,
+      "learning_rate": 4.089383048491824e-06,
+      "loss": 0.87,
+      "step": 125800
+    },
+    {
+      "epoch": 0.9106965768348209,
+      "grad_norm": 0.15854406356811523,
+      "learning_rate": 4.089310661831238e-06,
+      "loss": 0.8649,
+      "step": 125810
+    },
+    {
+      "epoch": 0.910768963495407,
+      "grad_norm": 0.16019472479820251,
+      "learning_rate": 4.0892382751706515e-06,
+      "loss": 0.8806,
+      "step": 125820
+    },
+    {
+      "epoch": 0.9108413501559932,
+      "grad_norm": 0.16454534232616425,
+      "learning_rate": 4.089165888510065e-06,
+      "loss": 0.874,
+      "step": 125830
+    },
+    {
+      "epoch": 0.9109137368165794,
+      "grad_norm": 0.14616376161575317,
+      "learning_rate": 4.08909350184948e-06,
+      "loss": 0.8781,
+      "step": 125840
+    },
+    {
+      "epoch": 0.9109861234771657,
+      "grad_norm": 0.16948239505290985,
+      "learning_rate": 4.089021115188893e-06,
+      "loss": 0.8746,
+      "step": 125850
+    },
+    {
+      "epoch": 0.9110585101377519,
+      "grad_norm": 0.1634904146194458,
+      "learning_rate": 4.088948728528307e-06,
+      "loss": 0.8729,
+      "step": 125860
+    },
+    {
+      "epoch": 0.911130896798338,
+      "grad_norm": 0.16780352592468262,
+      "learning_rate": 4.0888763418677204e-06,
+      "loss": 0.8731,
+      "step": 125870
+    },
+    {
+      "epoch": 0.9112032834589242,
+      "grad_norm": 0.1554594188928604,
+      "learning_rate": 4.088803955207135e-06,
+      "loss": 0.8751,
+      "step": 125880
+    },
+    {
+      "epoch": 0.9112756701195104,
+      "grad_norm": 0.1543656289577484,
+      "learning_rate": 4.0887315685465485e-06,
+      "loss": 0.8811,
+      "step": 125890
+    },
+    {
+      "epoch": 0.9113480567800966,
+      "grad_norm": 0.14425186812877655,
+      "learning_rate": 4.088659181885962e-06,
+      "loss": 0.8709,
+      "step": 125900
+    },
+    {
+      "epoch": 0.9114204434406827,
+      "grad_norm": 0.15529589354991913,
+      "learning_rate": 4.088586795225376e-06,
+      "loss": 0.8594,
+      "step": 125910
+    },
+    {
+      "epoch": 0.9114928301012689,
+      "grad_norm": 0.1789928823709488,
+      "learning_rate": 4.08851440856479e-06,
+      "loss": 0.867,
+      "step": 125920
+    },
+    {
+      "epoch": 0.9115652167618551,
+      "grad_norm": 0.15948614478111267,
+      "learning_rate": 4.088442021904204e-06,
+      "loss": 0.8622,
+      "step": 125930
+    },
+    {
+      "epoch": 0.9116376034224413,
+      "grad_norm": 0.18289946019649506,
+      "learning_rate": 4.0883696352436174e-06,
+      "loss": 0.8776,
+      "step": 125940
+    },
+    {
+      "epoch": 0.9117099900830276,
+      "grad_norm": 0.16194994747638702,
+      "learning_rate": 4.088297248583031e-06,
+      "loss": 0.8787,
+      "step": 125950
+    },
+    {
+      "epoch": 0.9117823767436137,
+      "grad_norm": 0.15668463706970215,
+      "learning_rate": 4.0882248619224455e-06,
+      "loss": 0.8665,
+      "step": 125960
+    },
+    {
+      "epoch": 0.9118547634041999,
+      "grad_norm": 0.14782801270484924,
+      "learning_rate": 4.088152475261859e-06,
+      "loss": 0.8671,
+      "step": 125970
+    },
+    {
+      "epoch": 0.9119271500647861,
+      "grad_norm": 0.14951111376285553,
+      "learning_rate": 4.088080088601273e-06,
+      "loss": 0.8714,
+      "step": 125980
+    },
+    {
+      "epoch": 0.9119995367253723,
+      "grad_norm": 0.15753866732120514,
+      "learning_rate": 4.088007701940686e-06,
+      "loss": 0.8804,
+      "step": 125990
+    },
+    {
+      "epoch": 0.9120719233859584,
+      "grad_norm": 0.1543973982334137,
+      "learning_rate": 4.087935315280101e-06,
+      "loss": 0.8618,
+      "step": 126000
+    },
+    {
+      "epoch": 0.9121443100465446,
+      "grad_norm": 0.16389751434326172,
+      "learning_rate": 4.0878629286195145e-06,
+      "loss": 0.865,
+      "step": 126010
+    },
+    {
+      "epoch": 0.9122166967071308,
+      "grad_norm": 0.15064431726932526,
+      "learning_rate": 4.087790541958928e-06,
+      "loss": 0.8691,
+      "step": 126020
+    },
+    {
+      "epoch": 0.912289083367717,
+      "grad_norm": 0.15450918674468994,
+      "learning_rate": 4.087718155298342e-06,
+      "loss": 0.8802,
+      "step": 126030
+    },
+    {
+      "epoch": 0.9123614700283031,
+      "grad_norm": 0.15472367405891418,
+      "learning_rate": 4.087645768637756e-06,
+      "loss": 0.8812,
+      "step": 126040
+    },
+    {
+      "epoch": 0.9124338566888893,
+      "grad_norm": 0.16118258237838745,
+      "learning_rate": 4.08757338197717e-06,
+      "loss": 0.8693,
+      "step": 126050
+    },
+    {
+      "epoch": 0.9125062433494756,
+      "grad_norm": 0.21667267382144928,
+      "learning_rate": 4.087500995316583e-06,
+      "loss": 0.8782,
+      "step": 126060
+    },
+    {
+      "epoch": 0.9125786300100618,
+      "grad_norm": 0.1615353375673294,
+      "learning_rate": 4.087428608655997e-06,
+      "loss": 0.8707,
+      "step": 126070
+    },
+    {
+      "epoch": 0.912651016670648,
+      "grad_norm": 0.15932497382164001,
+      "learning_rate": 4.0873562219954115e-06,
+      "loss": 0.8674,
+      "step": 126080
+    },
+    {
+      "epoch": 0.9127234033312341,
+      "grad_norm": 0.14429889619350433,
+      "learning_rate": 4.087283835334825e-06,
+      "loss": 0.8676,
+      "step": 126090
+    },
+    {
+      "epoch": 0.9127957899918203,
+      "grad_norm": 0.1498817354440689,
+      "learning_rate": 4.087211448674239e-06,
+      "loss": 0.8839,
+      "step": 126100
+    },
+    {
+      "epoch": 0.9128681766524065,
+      "grad_norm": 0.1536143273115158,
+      "learning_rate": 4.087139062013652e-06,
+      "loss": 0.866,
+      "step": 126110
+    },
+    {
+      "epoch": 0.9129405633129927,
+      "grad_norm": 0.14300012588500977,
+      "learning_rate": 4.087066675353066e-06,
+      "loss": 0.8694,
+      "step": 126120
+    },
+    {
+      "epoch": 0.9130129499735788,
+      "grad_norm": 0.15428438782691956,
+      "learning_rate": 4.08699428869248e-06,
+      "loss": 0.8779,
+      "step": 126130
+    },
+    {
+      "epoch": 0.913085336634165,
+      "grad_norm": 0.17595326900482178,
+      "learning_rate": 4.086921902031894e-06,
+      "loss": 0.8634,
+      "step": 126140
+    },
+    {
+      "epoch": 0.9131577232947512,
+      "grad_norm": 0.15784429013729095,
+      "learning_rate": 4.086849515371308e-06,
+      "loss": 0.8764,
+      "step": 126150
+    },
+    {
+      "epoch": 0.9132301099553374,
+      "grad_norm": 0.15594114363193512,
+      "learning_rate": 4.086777128710721e-06,
+      "loss": 0.8722,
+      "step": 126160
+    },
+    {
+      "epoch": 0.9133024966159237,
+      "grad_norm": 0.14309413731098175,
+      "learning_rate": 4.086704742050136e-06,
+      "loss": 0.8726,
+      "step": 126170
+    },
+    {
+      "epoch": 0.9133748832765098,
+      "grad_norm": 0.15265756845474243,
+      "learning_rate": 4.086632355389549e-06,
+      "loss": 0.8784,
+      "step": 126180
+    },
+    {
+      "epoch": 0.913447269937096,
+      "grad_norm": 0.20048628747463226,
+      "learning_rate": 4.086559968728963e-06,
+      "loss": 0.8717,
+      "step": 126190
+    },
+    {
+      "epoch": 0.9135196565976822,
+      "grad_norm": 0.15623369812965393,
+      "learning_rate": 4.0864875820683765e-06,
+      "loss": 0.8691,
+      "step": 126200
+    },
+    {
+      "epoch": 0.9135920432582684,
+      "grad_norm": 0.1530992090702057,
+      "learning_rate": 4.086415195407791e-06,
+      "loss": 0.8678,
+      "step": 126210
+    },
+    {
+      "epoch": 0.9136644299188545,
+      "grad_norm": 0.15348680317401886,
+      "learning_rate": 4.086342808747205e-06,
+      "loss": 0.8609,
+      "step": 126220
+    },
+    {
+      "epoch": 0.9137368165794407,
+      "grad_norm": 0.1623658686876297,
+      "learning_rate": 4.086270422086618e-06,
+      "loss": 0.874,
+      "step": 126230
+    },
+    {
+      "epoch": 0.9138092032400269,
+      "grad_norm": 0.16523189842700958,
+      "learning_rate": 4.086198035426032e-06,
+      "loss": 0.8834,
+      "step": 126240
+    },
+    {
+      "epoch": 0.9138815899006131,
+      "grad_norm": 0.148997500538826,
+      "learning_rate": 4.086125648765446e-06,
+      "loss": 0.8689,
+      "step": 126250
+    },
+    {
+      "epoch": 0.9139539765611993,
+      "grad_norm": 0.15621952712535858,
+      "learning_rate": 4.08605326210486e-06,
+      "loss": 0.8786,
+      "step": 126260
+    },
+    {
+      "epoch": 0.9140263632217855,
+      "grad_norm": 0.14902496337890625,
+      "learning_rate": 4.0859808754442736e-06,
+      "loss": 0.87,
+      "step": 126270
+    },
+    {
+      "epoch": 0.9140987498823717,
+      "grad_norm": 0.15417183935642242,
+      "learning_rate": 4.085908488783687e-06,
+      "loss": 0.8674,
+      "step": 126280
+    },
+    {
+      "epoch": 0.9141711365429579,
+      "grad_norm": 0.19318684935569763,
+      "learning_rate": 4.085836102123102e-06,
+      "loss": 0.858,
+      "step": 126290
+    },
+    {
+      "epoch": 0.9142435232035441,
+      "grad_norm": 0.15379764139652252,
+      "learning_rate": 4.085763715462515e-06,
+      "loss": 0.8635,
+      "step": 126300
+    },
+    {
+      "epoch": 0.9143159098641302,
+      "grad_norm": 0.37275516986846924,
+      "learning_rate": 4.085691328801929e-06,
+      "loss": 0.8751,
+      "step": 126310
+    },
+    {
+      "epoch": 0.9143882965247164,
+      "grad_norm": 0.1587948054075241,
+      "learning_rate": 4.0856189421413425e-06,
+      "loss": 0.8655,
+      "step": 126320
+    },
+    {
+      "epoch": 0.9144606831853026,
+      "grad_norm": 0.15148475766181946,
+      "learning_rate": 4.085546555480756e-06,
+      "loss": 0.863,
+      "step": 126330
+    },
+    {
+      "epoch": 0.9145330698458888,
+      "grad_norm": 0.16496942937374115,
+      "learning_rate": 4.08547416882017e-06,
+      "loss": 0.8457,
+      "step": 126340
+    },
+    {
+      "epoch": 0.914605456506475,
+      "grad_norm": 0.15391036868095398,
+      "learning_rate": 4.085401782159583e-06,
+      "loss": 0.8728,
+      "step": 126350
+    },
+    {
+      "epoch": 0.9146778431670611,
+      "grad_norm": 0.15899689495563507,
+      "learning_rate": 4.085329395498998e-06,
+      "loss": 0.8644,
+      "step": 126360
+    },
+    {
+      "epoch": 0.9147502298276473,
+      "grad_norm": 0.1451529711484909,
+      "learning_rate": 4.085257008838411e-06,
+      "loss": 0.871,
+      "step": 126370
+    },
+    {
+      "epoch": 0.9148226164882336,
+      "grad_norm": 0.14316491782665253,
+      "learning_rate": 4.085184622177825e-06,
+      "loss": 0.8741,
+      "step": 126380
+    },
+    {
+      "epoch": 0.9148950031488198,
+      "grad_norm": 0.1564132571220398,
+      "learning_rate": 4.085112235517239e-06,
+      "loss": 0.8763,
+      "step": 126390
+    },
+    {
+      "epoch": 0.914967389809406,
+      "grad_norm": 0.16010133922100067,
+      "learning_rate": 4.085039848856653e-06,
+      "loss": 0.8729,
+      "step": 126400
+    },
+    {
+      "epoch": 0.9150397764699921,
+      "grad_norm": 0.15824563801288605,
+      "learning_rate": 4.084967462196067e-06,
+      "loss": 0.8694,
+      "step": 126410
+    },
+    {
+      "epoch": 0.9151121631305783,
+      "grad_norm": 0.16496235132217407,
+      "learning_rate": 4.08489507553548e-06,
+      "loss": 0.886,
+      "step": 126420
+    },
+    {
+      "epoch": 0.9151845497911645,
+      "grad_norm": 0.15750516951084137,
+      "learning_rate": 4.084822688874894e-06,
+      "loss": 0.8668,
+      "step": 126430
+    },
+    {
+      "epoch": 0.9152569364517507,
+      "grad_norm": 0.1529252976179123,
+      "learning_rate": 4.084750302214308e-06,
+      "loss": 0.8663,
+      "step": 126440
+    },
+    {
+      "epoch": 0.9153293231123368,
+      "grad_norm": 0.15626297891139984,
+      "learning_rate": 4.084677915553722e-06,
+      "loss": 0.8725,
+      "step": 126450
+    },
+    {
+      "epoch": 0.915401709772923,
+      "grad_norm": 0.15462850034236908,
+      "learning_rate": 4.084605528893136e-06,
+      "loss": 0.8787,
+      "step": 126460
+    },
+    {
+      "epoch": 0.9154740964335092,
+      "grad_norm": 0.15536385774612427,
+      "learning_rate": 4.084533142232549e-06,
+      "loss": 0.8707,
+      "step": 126470
+    },
+    {
+      "epoch": 0.9155464830940954,
+      "grad_norm": 0.1599668562412262,
+      "learning_rate": 4.084460755571964e-06,
+      "loss": 0.8703,
+      "step": 126480
+    },
+    {
+      "epoch": 0.9156188697546817,
+      "grad_norm": 0.1455414891242981,
+      "learning_rate": 4.084388368911377e-06,
+      "loss": 0.8686,
+      "step": 126490
+    },
+    {
+      "epoch": 0.9156912564152678,
+      "grad_norm": 0.15002378821372986,
+      "learning_rate": 4.084315982250791e-06,
+      "loss": 0.8834,
+      "step": 126500
+    },
+    {
+      "epoch": 0.915763643075854,
+      "grad_norm": 0.16886621713638306,
+      "learning_rate": 4.084243595590205e-06,
+      "loss": 0.8663,
+      "step": 126510
+    },
+    {
+      "epoch": 0.9158360297364402,
+      "grad_norm": 0.15072478353977203,
+      "learning_rate": 4.084171208929619e-06,
+      "loss": 0.8776,
+      "step": 126520
+    },
+    {
+      "epoch": 0.9159084163970264,
+      "grad_norm": 0.18889220058918,
+      "learning_rate": 4.084098822269033e-06,
+      "loss": 0.8762,
+      "step": 126530
+    },
+    {
+      "epoch": 0.9159808030576125,
+      "grad_norm": 0.1520804464817047,
+      "learning_rate": 4.084026435608446e-06,
+      "loss": 0.8669,
+      "step": 126540
+    },
+    {
+      "epoch": 0.9160531897181987,
+      "grad_norm": 0.16211113333702087,
+      "learning_rate": 4.08395404894786e-06,
+      "loss": 0.8734,
+      "step": 126550
+    },
+    {
+      "epoch": 0.9161255763787849,
+      "grad_norm": 0.21941564977169037,
+      "learning_rate": 4.083881662287274e-06,
+      "loss": 0.876,
+      "step": 126560
+    },
+    {
+      "epoch": 0.9161979630393711,
+      "grad_norm": 0.14378584921360016,
+      "learning_rate": 4.083809275626688e-06,
+      "loss": 0.8728,
+      "step": 126570
+    },
+    {
+      "epoch": 0.9162703496999572,
+      "grad_norm": 0.14332708716392517,
+      "learning_rate": 4.083736888966102e-06,
+      "loss": 0.8764,
+      "step": 126580
+    },
+    {
+      "epoch": 0.9163427363605435,
+      "grad_norm": 0.14835390448570251,
+      "learning_rate": 4.083664502305515e-06,
+      "loss": 0.8651,
+      "step": 126590
+    },
+    {
+      "epoch": 0.9164151230211297,
+      "grad_norm": 0.1542014479637146,
+      "learning_rate": 4.08359211564493e-06,
+      "loss": 0.8678,
+      "step": 126600
+    },
+    {
+      "epoch": 0.9164875096817159,
+      "grad_norm": 0.1536477953195572,
+      "learning_rate": 4.083519728984343e-06,
+      "loss": 0.88,
+      "step": 126610
+    },
+    {
+      "epoch": 0.9165598963423021,
+      "grad_norm": 0.16406361758708954,
+      "learning_rate": 4.083447342323757e-06,
+      "loss": 0.8792,
+      "step": 126620
+    },
+    {
+      "epoch": 0.9166322830028882,
+      "grad_norm": 0.14225560426712036,
+      "learning_rate": 4.0833749556631705e-06,
+      "loss": 0.878,
+      "step": 126630
+    },
+    {
+      "epoch": 0.9167046696634744,
+      "grad_norm": 0.1528317928314209,
+      "learning_rate": 4.083302569002585e-06,
+      "loss": 0.8706,
+      "step": 126640
+    },
+    {
+      "epoch": 0.9167770563240606,
+      "grad_norm": 0.1531572937965393,
+      "learning_rate": 4.083230182341999e-06,
+      "loss": 0.8681,
+      "step": 126650
+    },
+    {
+      "epoch": 0.9168494429846468,
+      "grad_norm": 0.163166806101799,
+      "learning_rate": 4.083157795681412e-06,
+      "loss": 0.8694,
+      "step": 126660
+    },
+    {
+      "epoch": 0.9169218296452329,
+      "grad_norm": 0.1512100100517273,
+      "learning_rate": 4.083085409020826e-06,
+      "loss": 0.8655,
+      "step": 126670
+    },
+    {
+      "epoch": 0.9169942163058191,
+      "grad_norm": 0.14937952160835266,
+      "learning_rate": 4.08301302236024e-06,
+      "loss": 0.8743,
+      "step": 126680
+    },
+    {
+      "epoch": 0.9170666029664053,
+      "grad_norm": 0.16589467227458954,
+      "learning_rate": 4.082940635699654e-06,
+      "loss": 0.8736,
+      "step": 126690
+    },
+    {
+      "epoch": 0.9171389896269916,
+      "grad_norm": 0.15574873983860016,
+      "learning_rate": 4.0828682490390675e-06,
+      "loss": 0.8724,
+      "step": 126700
+    },
+    {
+      "epoch": 0.9172113762875778,
+      "grad_norm": 0.15169155597686768,
+      "learning_rate": 4.082795862378481e-06,
+      "loss": 0.8803,
+      "step": 126710
+    },
+    {
+      "epoch": 0.9172837629481639,
+      "grad_norm": 0.15095655620098114,
+      "learning_rate": 4.082723475717896e-06,
+      "loss": 0.8695,
+      "step": 126720
+    },
+    {
+      "epoch": 0.9173561496087501,
+      "grad_norm": 0.144984170794487,
+      "learning_rate": 4.082651089057309e-06,
+      "loss": 0.8814,
+      "step": 126730
+    },
+    {
+      "epoch": 0.9174285362693363,
+      "grad_norm": 0.15031631290912628,
+      "learning_rate": 4.082578702396723e-06,
+      "loss": 0.8549,
+      "step": 126740
+    },
+    {
+      "epoch": 0.9175009229299225,
+      "grad_norm": 0.15027374029159546,
+      "learning_rate": 4.0825063157361364e-06,
+      "loss": 0.8781,
+      "step": 126750
+    },
+    {
+      "epoch": 0.9175733095905086,
+      "grad_norm": 0.1555909812450409,
+      "learning_rate": 4.08243392907555e-06,
+      "loss": 0.8563,
+      "step": 126760
+    },
+    {
+      "epoch": 0.9176456962510948,
+      "grad_norm": 0.1572585105895996,
+      "learning_rate": 4.0823615424149645e-06,
+      "loss": 0.88,
+      "step": 126770
+    },
+    {
+      "epoch": 0.917718082911681,
+      "grad_norm": 0.15121687948703766,
+      "learning_rate": 4.082289155754378e-06,
+      "loss": 0.864,
+      "step": 126780
+    },
+    {
+      "epoch": 0.9177904695722672,
+      "grad_norm": 0.14316906034946442,
+      "learning_rate": 4.082216769093792e-06,
+      "loss": 0.877,
+      "step": 126790
+    },
+    {
+      "epoch": 0.9178628562328535,
+      "grad_norm": 0.1634669452905655,
+      "learning_rate": 4.082144382433205e-06,
+      "loss": 0.8763,
+      "step": 126800
+    },
+    {
+      "epoch": 0.9179352428934396,
+      "grad_norm": 0.1625238060951233,
+      "learning_rate": 4.08207199577262e-06,
+      "loss": 0.8826,
+      "step": 126810
+    },
+    {
+      "epoch": 0.9180076295540258,
+      "grad_norm": 0.1542353332042694,
+      "learning_rate": 4.0819996091120335e-06,
+      "loss": 0.8715,
+      "step": 126820
+    },
+    {
+      "epoch": 0.918080016214612,
+      "grad_norm": 0.156948059797287,
+      "learning_rate": 4.081927222451447e-06,
+      "loss": 0.8703,
+      "step": 126830
+    },
+    {
+      "epoch": 0.9181524028751982,
+      "grad_norm": 0.17528340220451355,
+      "learning_rate": 4.081854835790861e-06,
+      "loss": 0.8721,
+      "step": 126840
+    },
+    {
+      "epoch": 0.9182247895357843,
+      "grad_norm": 0.15125548839569092,
+      "learning_rate": 4.081782449130275e-06,
+      "loss": 0.8732,
+      "step": 126850
+    },
+    {
+      "epoch": 0.9182971761963705,
+      "grad_norm": 0.1517082154750824,
+      "learning_rate": 4.081710062469688e-06,
+      "loss": 0.8781,
+      "step": 126860
+    },
+    {
+      "epoch": 0.9183695628569567,
+      "grad_norm": 0.16170631349086761,
+      "learning_rate": 4.081637675809102e-06,
+      "loss": 0.8683,
+      "step": 126870
+    },
+    {
+      "epoch": 0.9184419495175429,
+      "grad_norm": 0.15255410969257355,
+      "learning_rate": 4.081565289148516e-06,
+      "loss": 0.8665,
+      "step": 126880
+    },
+    {
+      "epoch": 0.918514336178129,
+      "grad_norm": 0.158643901348114,
+      "learning_rate": 4.08149290248793e-06,
+      "loss": 0.8795,
+      "step": 126890
+    },
+    {
+      "epoch": 0.9185867228387152,
+      "grad_norm": 0.15852761268615723,
+      "learning_rate": 4.081420515827343e-06,
+      "loss": 0.87,
+      "step": 126900
+    },
+    {
+      "epoch": 0.9186591094993015,
+      "grad_norm": 0.2482609897851944,
+      "learning_rate": 4.081348129166757e-06,
+      "loss": 0.8684,
+      "step": 126910
+    },
+    {
+      "epoch": 0.9187314961598877,
+      "grad_norm": 0.16423265635967255,
+      "learning_rate": 4.081275742506171e-06,
+      "loss": 0.8796,
+      "step": 126920
+    },
+    {
+      "epoch": 0.9188038828204739,
+      "grad_norm": 0.1586569845676422,
+      "learning_rate": 4.081203355845585e-06,
+      "loss": 0.8862,
+      "step": 126930
+    },
+    {
+      "epoch": 0.91887626948106,
+      "grad_norm": 0.1470942199230194,
+      "learning_rate": 4.0811309691849985e-06,
+      "loss": 0.8605,
+      "step": 126940
+    },
+    {
+      "epoch": 0.9189486561416462,
+      "grad_norm": 0.17311806976795197,
+      "learning_rate": 4.081058582524412e-06,
+      "loss": 0.8771,
+      "step": 126950
+    },
+    {
+      "epoch": 0.9190210428022324,
+      "grad_norm": 0.16579419374465942,
+      "learning_rate": 4.080986195863827e-06,
+      "loss": 0.8782,
+      "step": 126960
+    },
+    {
+      "epoch": 0.9190934294628186,
+      "grad_norm": 0.1688244491815567,
+      "learning_rate": 4.08091380920324e-06,
+      "loss": 0.8752,
+      "step": 126970
+    },
+    {
+      "epoch": 0.9191658161234048,
+      "grad_norm": 0.14942839741706848,
+      "learning_rate": 4.080841422542654e-06,
+      "loss": 0.8588,
+      "step": 126980
+    },
+    {
+      "epoch": 0.9192382027839909,
+      "grad_norm": 0.18559300899505615,
+      "learning_rate": 4.0807690358820675e-06,
+      "loss": 0.8744,
+      "step": 126990
+    },
+    {
+      "epoch": 0.9193105894445771,
+      "grad_norm": 0.15597479045391083,
+      "learning_rate": 4.080696649221482e-06,
+      "loss": 0.8656,
+      "step": 127000
+    },
+    {
+      "epoch": 0.9193829761051633,
+      "grad_norm": 0.15126492083072662,
+      "learning_rate": 4.0806242625608956e-06,
+      "loss": 0.87,
+      "step": 127010
+    },
+    {
+      "epoch": 0.9194553627657496,
+      "grad_norm": 0.1641504019498825,
+      "learning_rate": 4.080551875900309e-06,
+      "loss": 0.8796,
+      "step": 127020
+    },
+    {
+      "epoch": 0.9195277494263357,
+      "grad_norm": 0.15217795968055725,
+      "learning_rate": 4.080479489239723e-06,
+      "loss": 0.8741,
+      "step": 127030
+    },
+    {
+      "epoch": 0.9196001360869219,
+      "grad_norm": 0.16015304625034332,
+      "learning_rate": 4.080407102579137e-06,
+      "loss": 0.8903,
+      "step": 127040
+    },
+    {
+      "epoch": 0.9196725227475081,
+      "grad_norm": 0.15204580128192902,
+      "learning_rate": 4.080334715918551e-06,
+      "loss": 0.8619,
+      "step": 127050
+    },
+    {
+      "epoch": 0.9197449094080943,
+      "grad_norm": 0.14324235916137695,
+      "learning_rate": 4.0802623292579645e-06,
+      "loss": 0.8609,
+      "step": 127060
+    },
+    {
+      "epoch": 0.9198172960686805,
+      "grad_norm": 0.1518729031085968,
+      "learning_rate": 4.080189942597378e-06,
+      "loss": 0.8761,
+      "step": 127070
+    },
+    {
+      "epoch": 0.9198896827292666,
+      "grad_norm": 0.15307843685150146,
+      "learning_rate": 4.0801175559367926e-06,
+      "loss": 0.8673,
+      "step": 127080
+    },
+    {
+      "epoch": 0.9199620693898528,
+      "grad_norm": 0.14242909848690033,
+      "learning_rate": 4.080045169276206e-06,
+      "loss": 0.8756,
+      "step": 127090
+    },
+    {
+      "epoch": 0.920034456050439,
+      "grad_norm": 0.16166019439697266,
+      "learning_rate": 4.07997278261562e-06,
+      "loss": 0.8742,
+      "step": 127100
+    },
+    {
+      "epoch": 0.9201068427110252,
+      "grad_norm": 0.15393179655075073,
+      "learning_rate": 4.079900395955033e-06,
+      "loss": 0.8783,
+      "step": 127110
+    },
+    {
+      "epoch": 0.9201792293716115,
+      "grad_norm": 0.16669778525829315,
+      "learning_rate": 4.079828009294448e-06,
+      "loss": 0.8653,
+      "step": 127120
+    },
+    {
+      "epoch": 0.9202516160321976,
+      "grad_norm": 0.15323293209075928,
+      "learning_rate": 4.0797556226338615e-06,
+      "loss": 0.8681,
+      "step": 127130
+    },
+    {
+      "epoch": 0.9203240026927838,
+      "grad_norm": 0.1925155520439148,
+      "learning_rate": 4.079683235973275e-06,
+      "loss": 0.8669,
+      "step": 127140
+    },
+    {
+      "epoch": 0.92039638935337,
+      "grad_norm": 0.16045774519443512,
+      "learning_rate": 4.079610849312689e-06,
+      "loss": 0.8688,
+      "step": 127150
+    },
+    {
+      "epoch": 0.9204687760139562,
+      "grad_norm": 0.16183599829673767,
+      "learning_rate": 4.079538462652103e-06,
+      "loss": 0.8602,
+      "step": 127160
+    },
+    {
+      "epoch": 0.9205411626745423,
+      "grad_norm": 0.15572576224803925,
+      "learning_rate": 4.079466075991517e-06,
+      "loss": 0.8722,
+      "step": 127170
+    },
+    {
+      "epoch": 0.9206135493351285,
+      "grad_norm": 0.15877743065357208,
+      "learning_rate": 4.07939368933093e-06,
+      "loss": 0.8783,
+      "step": 127180
+    },
+    {
+      "epoch": 0.9206859359957147,
+      "grad_norm": 0.14954781532287598,
+      "learning_rate": 4.079321302670344e-06,
+      "loss": 0.8597,
+      "step": 127190
+    },
+    {
+      "epoch": 0.9207583226563009,
+      "grad_norm": 0.1656097173690796,
+      "learning_rate": 4.0792489160097585e-06,
+      "loss": 0.8725,
+      "step": 127200
+    },
+    {
+      "epoch": 0.920830709316887,
+      "grad_norm": 0.15802302956581116,
+      "learning_rate": 4.079176529349172e-06,
+      "loss": 0.8739,
+      "step": 127210
+    },
+    {
+      "epoch": 0.9209030959774732,
+      "grad_norm": 0.14850866794586182,
+      "learning_rate": 4.079104142688586e-06,
+      "loss": 0.854,
+      "step": 127220
+    },
+    {
+      "epoch": 0.9209754826380595,
+      "grad_norm": 0.15718205273151398,
+      "learning_rate": 4.079031756027999e-06,
+      "loss": 0.876,
+      "step": 127230
+    },
+    {
+      "epoch": 0.9210478692986457,
+      "grad_norm": 0.15593793988227844,
+      "learning_rate": 4.078959369367414e-06,
+      "loss": 0.8773,
+      "step": 127240
+    },
+    {
+      "epoch": 0.9211202559592319,
+      "grad_norm": 0.14483477175235748,
+      "learning_rate": 4.0788869827068274e-06,
+      "loss": 0.8722,
+      "step": 127250
+    },
+    {
+      "epoch": 0.921192642619818,
+      "grad_norm": 0.16658411920070648,
+      "learning_rate": 4.078814596046241e-06,
+      "loss": 0.8677,
+      "step": 127260
+    },
+    {
+      "epoch": 0.9212650292804042,
+      "grad_norm": 0.1465682089328766,
+      "learning_rate": 4.078742209385655e-06,
+      "loss": 0.8688,
+      "step": 127270
+    },
+    {
+      "epoch": 0.9213374159409904,
+      "grad_norm": 0.16307535767555237,
+      "learning_rate": 4.078669822725069e-06,
+      "loss": 0.8704,
+      "step": 127280
+    },
+    {
+      "epoch": 0.9214098026015766,
+      "grad_norm": 0.303844153881073,
+      "learning_rate": 4.078597436064483e-06,
+      "loss": 0.8708,
+      "step": 127290
+    },
+    {
+      "epoch": 0.9214821892621627,
+      "grad_norm": 0.14299264550209045,
+      "learning_rate": 4.078525049403896e-06,
+      "loss": 0.8653,
+      "step": 127300
+    },
+    {
+      "epoch": 0.9215545759227489,
+      "grad_norm": 0.14078418910503387,
+      "learning_rate": 4.07845266274331e-06,
+      "loss": 0.8736,
+      "step": 127310
+    },
+    {
+      "epoch": 0.9216269625833351,
+      "grad_norm": 0.14355109632015228,
+      "learning_rate": 4.0783802760827244e-06,
+      "loss": 0.8658,
+      "step": 127320
+    },
+    {
+      "epoch": 0.9216993492439214,
+      "grad_norm": 0.18450787663459778,
+      "learning_rate": 4.078307889422138e-06,
+      "loss": 0.8626,
+      "step": 127330
+    },
+    {
+      "epoch": 0.9217717359045076,
+      "grad_norm": 0.16404852271080017,
+      "learning_rate": 4.078235502761552e-06,
+      "loss": 0.8701,
+      "step": 127340
+    },
+    {
+      "epoch": 0.9218441225650937,
+      "grad_norm": 0.15658390522003174,
+      "learning_rate": 4.078163116100965e-06,
+      "loss": 0.8798,
+      "step": 127350
+    },
+    {
+      "epoch": 0.9219165092256799,
+      "grad_norm": 0.2226731926202774,
+      "learning_rate": 4.078090729440379e-06,
+      "loss": 0.8801,
+      "step": 127360
+    },
+    {
+      "epoch": 0.9219888958862661,
+      "grad_norm": 0.19657254219055176,
+      "learning_rate": 4.078018342779793e-06,
+      "loss": 0.8749,
+      "step": 127370
+    },
+    {
+      "epoch": 0.9220612825468523,
+      "grad_norm": 0.15708957612514496,
+      "learning_rate": 4.077945956119207e-06,
+      "loss": 0.8697,
+      "step": 127380
+    },
+    {
+      "epoch": 0.9221336692074384,
+      "grad_norm": 0.16604341566562653,
+      "learning_rate": 4.077873569458621e-06,
+      "loss": 0.8535,
+      "step": 127390
+    },
+    {
+      "epoch": 0.9222060558680246,
+      "grad_norm": 0.14287860691547394,
+      "learning_rate": 4.077801182798034e-06,
+      "loss": 0.8556,
+      "step": 127400
+    },
+    {
+      "epoch": 0.9222784425286108,
+      "grad_norm": 0.1693638414144516,
+      "learning_rate": 4.077728796137448e-06,
+      "loss": 0.8789,
+      "step": 127410
+    },
+    {
+      "epoch": 0.922350829189197,
+      "grad_norm": 0.1491190791130066,
+      "learning_rate": 4.0776564094768614e-06,
+      "loss": 0.863,
+      "step": 127420
+    },
+    {
+      "epoch": 0.9224232158497832,
+      "grad_norm": 0.14451493322849274,
+      "learning_rate": 4.077584022816276e-06,
+      "loss": 0.8768,
+      "step": 127430
+    },
+    {
+      "epoch": 0.9224956025103694,
+      "grad_norm": 0.155923992395401,
+      "learning_rate": 4.0775116361556895e-06,
+      "loss": 0.8745,
+      "step": 127440
+    },
+    {
+      "epoch": 0.9225679891709556,
+      "grad_norm": 0.1764858216047287,
+      "learning_rate": 4.077439249495103e-06,
+      "loss": 0.8792,
+      "step": 127450
+    },
+    {
+      "epoch": 0.9226403758315418,
+      "grad_norm": 0.16587840020656586,
+      "learning_rate": 4.077366862834517e-06,
+      "loss": 0.8666,
+      "step": 127460
+    },
+    {
+      "epoch": 0.922712762492128,
+      "grad_norm": 0.16507935523986816,
+      "learning_rate": 4.077294476173931e-06,
+      "loss": 0.8765,
+      "step": 127470
+    },
+    {
+      "epoch": 0.9227851491527141,
+      "grad_norm": 0.15929803252220154,
+      "learning_rate": 4.077222089513345e-06,
+      "loss": 0.8667,
+      "step": 127480
+    },
+    {
+      "epoch": 0.9228575358133003,
+      "grad_norm": 0.15834636986255646,
+      "learning_rate": 4.0771497028527584e-06,
+      "loss": 0.8802,
+      "step": 127490
+    },
+    {
+      "epoch": 0.9229299224738865,
+      "grad_norm": 0.1508595049381256,
+      "learning_rate": 4.077077316192172e-06,
+      "loss": 0.8584,
+      "step": 127500
+    },
+    {
+      "epoch": 0.9230023091344727,
+      "grad_norm": 0.14436863362789154,
+      "learning_rate": 4.0770049295315865e-06,
+      "loss": 0.8725,
+      "step": 127510
+    },
+    {
+      "epoch": 0.9230746957950589,
+      "grad_norm": 0.17195680737495422,
+      "learning_rate": 4.076932542871e-06,
+      "loss": 0.8717,
+      "step": 127520
+    },
+    {
+      "epoch": 0.923147082455645,
+      "grad_norm": 0.16129696369171143,
+      "learning_rate": 4.076860156210414e-06,
+      "loss": 0.8821,
+      "step": 127530
+    },
+    {
+      "epoch": 0.9232194691162312,
+      "grad_norm": 0.15177470445632935,
+      "learning_rate": 4.076787769549827e-06,
+      "loss": 0.8636,
+      "step": 127540
+    },
+    {
+      "epoch": 0.9232918557768175,
+      "grad_norm": 0.16349272429943085,
+      "learning_rate": 4.076715382889241e-06,
+      "loss": 0.8705,
+      "step": 127550
+    },
+    {
+      "epoch": 0.9233642424374037,
+      "grad_norm": 0.14663663506507874,
+      "learning_rate": 4.0766429962286555e-06,
+      "loss": 0.8571,
+      "step": 127560
+    },
+    {
+      "epoch": 0.9234366290979898,
+      "grad_norm": 0.16342896223068237,
+      "learning_rate": 4.076570609568069e-06,
+      "loss": 0.8687,
+      "step": 127570
+    },
+    {
+      "epoch": 0.923509015758576,
+      "grad_norm": 0.15321224927902222,
+      "learning_rate": 4.076498222907483e-06,
+      "loss": 0.8622,
+      "step": 127580
+    },
+    {
+      "epoch": 0.9235814024191622,
+      "grad_norm": 0.1598549485206604,
+      "learning_rate": 4.076425836246896e-06,
+      "loss": 0.8786,
+      "step": 127590
+    },
+    {
+      "epoch": 0.9236537890797484,
+      "grad_norm": 0.16147591173648834,
+      "learning_rate": 4.076353449586311e-06,
+      "loss": 0.872,
+      "step": 127600
+    },
+    {
+      "epoch": 0.9237261757403346,
+      "grad_norm": 0.14562930166721344,
+      "learning_rate": 4.076281062925724e-06,
+      "loss": 0.8682,
+      "step": 127610
+    },
+    {
+      "epoch": 0.9237985624009207,
+      "grad_norm": 0.1647154986858368,
+      "learning_rate": 4.076208676265138e-06,
+      "loss": 0.8815,
+      "step": 127620
+    },
+    {
+      "epoch": 0.9238709490615069,
+      "grad_norm": 0.1684718132019043,
+      "learning_rate": 4.076136289604552e-06,
+      "loss": 0.8714,
+      "step": 127630
+    },
+    {
+      "epoch": 0.9239433357220931,
+      "grad_norm": 0.16165629029273987,
+      "learning_rate": 4.076063902943966e-06,
+      "loss": 0.8705,
+      "step": 127640
+    },
+    {
+      "epoch": 0.9240157223826794,
+      "grad_norm": 0.15720994770526886,
+      "learning_rate": 4.07599151628338e-06,
+      "loss": 0.8629,
+      "step": 127650
+    },
+    {
+      "epoch": 0.9240881090432655,
+      "grad_norm": 0.16634640097618103,
+      "learning_rate": 4.075919129622793e-06,
+      "loss": 0.883,
+      "step": 127660
+    },
+    {
+      "epoch": 0.9241604957038517,
+      "grad_norm": 0.1512623280286789,
+      "learning_rate": 4.075846742962207e-06,
+      "loss": 0.8628,
+      "step": 127670
+    },
+    {
+      "epoch": 0.9242328823644379,
+      "grad_norm": 0.14942419528961182,
+      "learning_rate": 4.075774356301621e-06,
+      "loss": 0.8543,
+      "step": 127680
+    },
+    {
+      "epoch": 0.9243052690250241,
+      "grad_norm": 0.15057924389839172,
+      "learning_rate": 4.075701969641035e-06,
+      "loss": 0.8695,
+      "step": 127690
+    },
+    {
+      "epoch": 0.9243776556856103,
+      "grad_norm": 0.15089672803878784,
+      "learning_rate": 4.075629582980449e-06,
+      "loss": 0.8924,
+      "step": 127700
+    },
+    {
+      "epoch": 0.9244500423461964,
+      "grad_norm": 0.15450294315814972,
+      "learning_rate": 4.075557196319862e-06,
+      "loss": 0.8642,
+      "step": 127710
+    },
+    {
+      "epoch": 0.9245224290067826,
+      "grad_norm": 0.14952602982521057,
+      "learning_rate": 4.075484809659277e-06,
+      "loss": 0.8694,
+      "step": 127720
+    },
+    {
+      "epoch": 0.9245948156673688,
+      "grad_norm": 0.16396868228912354,
+      "learning_rate": 4.07541242299869e-06,
+      "loss": 0.8905,
+      "step": 127730
+    },
+    {
+      "epoch": 0.924667202327955,
+      "grad_norm": 0.1575574278831482,
+      "learning_rate": 4.075340036338104e-06,
+      "loss": 0.8772,
+      "step": 127740
+    },
+    {
+      "epoch": 0.9247395889885411,
+      "grad_norm": 0.16081462800502777,
+      "learning_rate": 4.0752676496775176e-06,
+      "loss": 0.8768,
+      "step": 127750
+    },
+    {
+      "epoch": 0.9248119756491274,
+      "grad_norm": 0.14505434036254883,
+      "learning_rate": 4.075195263016932e-06,
+      "loss": 0.8524,
+      "step": 127760
+    },
+    {
+      "epoch": 0.9248843623097136,
+      "grad_norm": 0.15242122113704681,
+      "learning_rate": 4.075122876356346e-06,
+      "loss": 0.8701,
+      "step": 127770
+    },
+    {
+      "epoch": 0.9249567489702998,
+      "grad_norm": 0.14712516963481903,
+      "learning_rate": 4.075050489695759e-06,
+      "loss": 0.8619,
+      "step": 127780
+    },
+    {
+      "epoch": 0.925029135630886,
+      "grad_norm": 0.15457618236541748,
+      "learning_rate": 4.074978103035173e-06,
+      "loss": 0.8745,
+      "step": 127790
+    },
+    {
+      "epoch": 0.9251015222914721,
+      "grad_norm": 0.15267246961593628,
+      "learning_rate": 4.074905716374587e-06,
+      "loss": 0.8584,
+      "step": 127800
+    },
+    {
+      "epoch": 0.9251739089520583,
+      "grad_norm": 0.15300290286540985,
+      "learning_rate": 4.074833329714001e-06,
+      "loss": 0.8626,
+      "step": 127810
+    },
+    {
+      "epoch": 0.9252462956126445,
+      "grad_norm": 0.14224673807621002,
+      "learning_rate": 4.0747609430534146e-06,
+      "loss": 0.8753,
+      "step": 127820
+    },
+    {
+      "epoch": 0.9253186822732307,
+      "grad_norm": 0.15373513102531433,
+      "learning_rate": 4.074688556392828e-06,
+      "loss": 0.8811,
+      "step": 127830
+    },
+    {
+      "epoch": 0.9253910689338168,
+      "grad_norm": 0.15759536623954773,
+      "learning_rate": 4.074616169732243e-06,
+      "loss": 0.8691,
+      "step": 127840
+    },
+    {
+      "epoch": 0.925463455594403,
+      "grad_norm": 0.16300243139266968,
+      "learning_rate": 4.074543783071656e-06,
+      "loss": 0.8709,
+      "step": 127850
+    },
+    {
+      "epoch": 0.9255358422549893,
+      "grad_norm": 0.16847741603851318,
+      "learning_rate": 4.07447139641107e-06,
+      "loss": 0.8689,
+      "step": 127860
+    },
+    {
+      "epoch": 0.9256082289155755,
+      "grad_norm": 0.16154174506664276,
+      "learning_rate": 4.0743990097504835e-06,
+      "loss": 0.8533,
+      "step": 127870
+    },
+    {
+      "epoch": 0.9256806155761617,
+      "grad_norm": 0.15363070368766785,
+      "learning_rate": 4.074326623089898e-06,
+      "loss": 0.8616,
+      "step": 127880
+    },
+    {
+      "epoch": 0.9257530022367478,
+      "grad_norm": 0.14643029868602753,
+      "learning_rate": 4.0742542364293116e-06,
+      "loss": 0.8536,
+      "step": 127890
+    },
+    {
+      "epoch": 0.925825388897334,
+      "grad_norm": 0.14773331582546234,
+      "learning_rate": 4.074181849768725e-06,
+      "loss": 0.8606,
+      "step": 127900
+    },
+    {
+      "epoch": 0.9258977755579202,
+      "grad_norm": 0.1448742300271988,
+      "learning_rate": 4.074109463108139e-06,
+      "loss": 0.8676,
+      "step": 127910
+    },
+    {
+      "epoch": 0.9259701622185064,
+      "grad_norm": 0.18105871975421906,
+      "learning_rate": 4.074037076447552e-06,
+      "loss": 0.8664,
+      "step": 127920
+    },
+    {
+      "epoch": 0.9260425488790925,
+      "grad_norm": 0.1581772118806839,
+      "learning_rate": 4.073964689786966e-06,
+      "loss": 0.8803,
+      "step": 127930
+    },
+    {
+      "epoch": 0.9261149355396787,
+      "grad_norm": 0.1465241014957428,
+      "learning_rate": 4.07389230312638e-06,
+      "loss": 0.8762,
+      "step": 127940
+    },
+    {
+      "epoch": 0.9261873222002649,
+      "grad_norm": 0.16206571459770203,
+      "learning_rate": 4.073819916465794e-06,
+      "loss": 0.8781,
+      "step": 127950
+    },
+    {
+      "epoch": 0.9262597088608511,
+      "grad_norm": 0.1485998034477234,
+      "learning_rate": 4.073747529805208e-06,
+      "loss": 0.8712,
+      "step": 127960
+    },
+    {
+      "epoch": 0.9263320955214374,
+      "grad_norm": 0.15023688971996307,
+      "learning_rate": 4.073675143144621e-06,
+      "loss": 0.8626,
+      "step": 127970
+    },
+    {
+      "epoch": 0.9264044821820235,
+      "grad_norm": 0.14640280604362488,
+      "learning_rate": 4.073602756484035e-06,
+      "loss": 0.8683,
+      "step": 127980
+    },
+    {
+      "epoch": 0.9264768688426097,
+      "grad_norm": 0.15513427555561066,
+      "learning_rate": 4.073530369823449e-06,
+      "loss": 0.8722,
+      "step": 127990
+    },
+    {
+      "epoch": 0.9265492555031959,
+      "grad_norm": 0.14883030951023102,
+      "learning_rate": 4.073457983162863e-06,
+      "loss": 0.8616,
+      "step": 128000
+    },
+    {
+      "epoch": 0.9266216421637821,
+      "grad_norm": 0.17207932472229004,
+      "learning_rate": 4.073385596502277e-06,
+      "loss": 0.8649,
+      "step": 128010
+    },
+    {
+      "epoch": 0.9266940288243682,
+      "grad_norm": 0.14062434434890747,
+      "learning_rate": 4.07331320984169e-06,
+      "loss": 0.8589,
+      "step": 128020
+    },
+    {
+      "epoch": 0.9267664154849544,
+      "grad_norm": 0.14233818650245667,
+      "learning_rate": 4.073240823181105e-06,
+      "loss": 0.8628,
+      "step": 128030
+    },
+    {
+      "epoch": 0.9268388021455406,
+      "grad_norm": 0.15095971524715424,
+      "learning_rate": 4.073168436520518e-06,
+      "loss": 0.8647,
+      "step": 128040
+    },
+    {
+      "epoch": 0.9269111888061268,
+      "grad_norm": 0.164928138256073,
+      "learning_rate": 4.073096049859932e-06,
+      "loss": 0.8679,
+      "step": 128050
+    },
+    {
+      "epoch": 0.926983575466713,
+      "grad_norm": 0.15687911212444305,
+      "learning_rate": 4.073023663199346e-06,
+      "loss": 0.8691,
+      "step": 128060
+    },
+    {
+      "epoch": 0.9270559621272991,
+      "grad_norm": 0.16557860374450684,
+      "learning_rate": 4.07295127653876e-06,
+      "loss": 0.8834,
+      "step": 128070
+    },
+    {
+      "epoch": 0.9271283487878854,
+      "grad_norm": 0.15757247805595398,
+      "learning_rate": 4.072878889878174e-06,
+      "loss": 0.8759,
+      "step": 128080
+    },
+    {
+      "epoch": 0.9272007354484716,
+      "grad_norm": 0.15857826173305511,
+      "learning_rate": 4.072806503217587e-06,
+      "loss": 0.8754,
+      "step": 128090
+    },
+    {
+      "epoch": 0.9272731221090578,
+      "grad_norm": 0.15188080072402954,
+      "learning_rate": 4.072734116557001e-06,
+      "loss": 0.8669,
+      "step": 128100
+    },
+    {
+      "epoch": 0.927345508769644,
+      "grad_norm": 0.15522122383117676,
+      "learning_rate": 4.072661729896415e-06,
+      "loss": 0.8643,
+      "step": 128110
+    },
+    {
+      "epoch": 0.9274178954302301,
+      "grad_norm": 0.14088667929172516,
+      "learning_rate": 4.072589343235829e-06,
+      "loss": 0.8784,
+      "step": 128120
+    },
+    {
+      "epoch": 0.9274902820908163,
+      "grad_norm": 0.1532406508922577,
+      "learning_rate": 4.072516956575243e-06,
+      "loss": 0.8637,
+      "step": 128130
+    },
+    {
+      "epoch": 0.9275626687514025,
+      "grad_norm": 0.14696896076202393,
+      "learning_rate": 4.072444569914656e-06,
+      "loss": 0.8601,
+      "step": 128140
+    },
+    {
+      "epoch": 0.9276350554119887,
+      "grad_norm": 0.1489928513765335,
+      "learning_rate": 4.07237218325407e-06,
+      "loss": 0.8698,
+      "step": 128150
+    },
+    {
+      "epoch": 0.9277074420725748,
+      "grad_norm": 0.1538485586643219,
+      "learning_rate": 4.072299796593484e-06,
+      "loss": 0.8817,
+      "step": 128160
+    },
+    {
+      "epoch": 0.927779828733161,
+      "grad_norm": 0.14813606441020966,
+      "learning_rate": 4.072227409932898e-06,
+      "loss": 0.8636,
+      "step": 128170
+    },
+    {
+      "epoch": 0.9278522153937473,
+      "grad_norm": 0.1586228609085083,
+      "learning_rate": 4.0721550232723115e-06,
+      "loss": 0.868,
+      "step": 128180
+    },
+    {
+      "epoch": 0.9279246020543335,
+      "grad_norm": 0.21530722081661224,
+      "learning_rate": 4.072082636611725e-06,
+      "loss": 0.8735,
+      "step": 128190
+    },
+    {
+      "epoch": 0.9279969887149196,
+      "grad_norm": 0.18316511809825897,
+      "learning_rate": 4.07201024995114e-06,
+      "loss": 0.8693,
+      "step": 128200
+    },
+    {
+      "epoch": 0.9280693753755058,
+      "grad_norm": 0.15171143412590027,
+      "learning_rate": 4.071937863290553e-06,
+      "loss": 0.8617,
+      "step": 128210
+    },
+    {
+      "epoch": 0.928141762036092,
+      "grad_norm": 0.15644243359565735,
+      "learning_rate": 4.071865476629967e-06,
+      "loss": 0.8728,
+      "step": 128220
+    },
+    {
+      "epoch": 0.9282141486966782,
+      "grad_norm": 0.16042940318584442,
+      "learning_rate": 4.0717930899693804e-06,
+      "loss": 0.8778,
+      "step": 128230
+    },
+    {
+      "epoch": 0.9282865353572644,
+      "grad_norm": 0.15380287170410156,
+      "learning_rate": 4.071720703308795e-06,
+      "loss": 0.8594,
+      "step": 128240
+    },
+    {
+      "epoch": 0.9283589220178505,
+      "grad_norm": 0.15428663790225983,
+      "learning_rate": 4.0716483166482085e-06,
+      "loss": 0.8817,
+      "step": 128250
+    },
+    {
+      "epoch": 0.9284313086784367,
+      "grad_norm": 0.14984562993049622,
+      "learning_rate": 4.071575929987622e-06,
+      "loss": 0.8781,
+      "step": 128260
+    },
+    {
+      "epoch": 0.9285036953390229,
+      "grad_norm": 0.16001556813716888,
+      "learning_rate": 4.071503543327036e-06,
+      "loss": 0.8666,
+      "step": 128270
+    },
+    {
+      "epoch": 0.9285760819996091,
+      "grad_norm": 0.1445271223783493,
+      "learning_rate": 4.07143115666645e-06,
+      "loss": 0.8772,
+      "step": 128280
+    },
+    {
+      "epoch": 0.9286484686601953,
+      "grad_norm": 0.1755678355693817,
+      "learning_rate": 4.071358770005864e-06,
+      "loss": 0.8805,
+      "step": 128290
+    },
+    {
+      "epoch": 0.9287208553207815,
+      "grad_norm": 0.16216261684894562,
+      "learning_rate": 4.0712863833452775e-06,
+      "loss": 0.8694,
+      "step": 128300
+    },
+    {
+      "epoch": 0.9287932419813677,
+      "grad_norm": 0.15545831620693207,
+      "learning_rate": 4.071213996684691e-06,
+      "loss": 0.8681,
+      "step": 128310
+    },
+    {
+      "epoch": 0.9288656286419539,
+      "grad_norm": 0.15312260389328003,
+      "learning_rate": 4.0711416100241055e-06,
+      "loss": 0.8721,
+      "step": 128320
+    },
+    {
+      "epoch": 0.92893801530254,
+      "grad_norm": 0.1504194438457489,
+      "learning_rate": 4.071069223363519e-06,
+      "loss": 0.8624,
+      "step": 128330
+    },
+    {
+      "epoch": 0.9290104019631262,
+      "grad_norm": 0.15465429425239563,
+      "learning_rate": 4.070996836702933e-06,
+      "loss": 0.8799,
+      "step": 128340
+    },
+    {
+      "epoch": 0.9290827886237124,
+      "grad_norm": 0.14774566888809204,
+      "learning_rate": 4.070924450042346e-06,
+      "loss": 0.8717,
+      "step": 128350
+    },
+    {
+      "epoch": 0.9291551752842986,
+      "grad_norm": 0.16979655623435974,
+      "learning_rate": 4.070852063381761e-06,
+      "loss": 0.8552,
+      "step": 128360
+    },
+    {
+      "epoch": 0.9292275619448848,
+      "grad_norm": 0.15105919539928436,
+      "learning_rate": 4.0707796767211745e-06,
+      "loss": 0.8624,
+      "step": 128370
+    },
+    {
+      "epoch": 0.9292999486054709,
+      "grad_norm": 0.155971959233284,
+      "learning_rate": 4.070707290060588e-06,
+      "loss": 0.8592,
+      "step": 128380
+    },
+    {
+      "epoch": 0.9293723352660572,
+      "grad_norm": 0.162667378783226,
+      "learning_rate": 4.070634903400002e-06,
+      "loss": 0.8771,
+      "step": 128390
+    },
+    {
+      "epoch": 0.9294447219266434,
+      "grad_norm": 0.1585136353969574,
+      "learning_rate": 4.070562516739416e-06,
+      "loss": 0.8714,
+      "step": 128400
+    },
+    {
+      "epoch": 0.9295171085872296,
+      "grad_norm": 0.14089208841323853,
+      "learning_rate": 4.07049013007883e-06,
+      "loss": 0.8691,
+      "step": 128410
+    },
+    {
+      "epoch": 0.9295894952478158,
+      "grad_norm": 0.14816270768642426,
+      "learning_rate": 4.070417743418243e-06,
+      "loss": 0.8754,
+      "step": 128420
+    },
+    {
+      "epoch": 0.9296618819084019,
+      "grad_norm": 0.14089035987854004,
+      "learning_rate": 4.070345356757657e-06,
+      "loss": 0.869,
+      "step": 128430
+    },
+    {
+      "epoch": 0.9297342685689881,
+      "grad_norm": 0.15903660655021667,
+      "learning_rate": 4.0702729700970715e-06,
+      "loss": 0.8669,
+      "step": 128440
+    },
+    {
+      "epoch": 0.9298066552295743,
+      "grad_norm": 0.15594375133514404,
+      "learning_rate": 4.070200583436484e-06,
+      "loss": 0.8594,
+      "step": 128450
+    },
+    {
+      "epoch": 0.9298790418901605,
+      "grad_norm": 0.15002770721912384,
+      "learning_rate": 4.070128196775898e-06,
+      "loss": 0.858,
+      "step": 128460
+    },
+    {
+      "epoch": 0.9299514285507466,
+      "grad_norm": 0.1600421667098999,
+      "learning_rate": 4.070055810115312e-06,
+      "loss": 0.8613,
+      "step": 128470
+    },
+    {
+      "epoch": 0.9300238152113328,
+      "grad_norm": 0.16615080833435059,
+      "learning_rate": 4.069983423454726e-06,
+      "loss": 0.8763,
+      "step": 128480
+    },
+    {
+      "epoch": 0.930096201871919,
+      "grad_norm": 0.15148785710334778,
+      "learning_rate": 4.0699110367941396e-06,
+      "loss": 0.8695,
+      "step": 128490
+    },
+    {
+      "epoch": 0.9301685885325053,
+      "grad_norm": 0.14840847253799438,
+      "learning_rate": 4.069838650133553e-06,
+      "loss": 0.8739,
+      "step": 128500
+    },
+    {
+      "epoch": 0.9302409751930915,
+      "grad_norm": 0.13647682964801788,
+      "learning_rate": 4.069766263472968e-06,
+      "loss": 0.8672,
+      "step": 128510
+    },
+    {
+      "epoch": 0.9303133618536776,
+      "grad_norm": 0.1561964750289917,
+      "learning_rate": 4.069693876812381e-06,
+      "loss": 0.8637,
+      "step": 128520
+    },
+    {
+      "epoch": 0.9303857485142638,
+      "grad_norm": 0.22199493646621704,
+      "learning_rate": 4.069621490151795e-06,
+      "loss": 0.8707,
+      "step": 128530
+    },
+    {
+      "epoch": 0.93045813517485,
+      "grad_norm": 0.1517588496208191,
+      "learning_rate": 4.0695491034912085e-06,
+      "loss": 0.8663,
+      "step": 128540
+    },
+    {
+      "epoch": 0.9305305218354362,
+      "grad_norm": 0.15259967744350433,
+      "learning_rate": 4.069476716830623e-06,
+      "loss": 0.8684,
+      "step": 128550
+    },
+    {
+      "epoch": 0.9306029084960223,
+      "grad_norm": 0.15910792350769043,
+      "learning_rate": 4.0694043301700366e-06,
+      "loss": 0.8676,
+      "step": 128560
+    },
+    {
+      "epoch": 0.9306752951566085,
+      "grad_norm": 0.16429047286510468,
+      "learning_rate": 4.06933194350945e-06,
+      "loss": 0.8783,
+      "step": 128570
+    },
+    {
+      "epoch": 0.9307476818171947,
+      "grad_norm": 0.1512821912765503,
+      "learning_rate": 4.069259556848864e-06,
+      "loss": 0.8691,
+      "step": 128580
+    },
+    {
+      "epoch": 0.9308200684777809,
+      "grad_norm": 0.154913067817688,
+      "learning_rate": 4.069187170188278e-06,
+      "loss": 0.8761,
+      "step": 128590
+    },
+    {
+      "epoch": 0.930892455138367,
+      "grad_norm": 0.14742477238178253,
+      "learning_rate": 4.069114783527692e-06,
+      "loss": 0.8685,
+      "step": 128600
+    },
+    {
+      "epoch": 0.9309648417989533,
+      "grad_norm": 0.15343081951141357,
+      "learning_rate": 4.0690423968671055e-06,
+      "loss": 0.8675,
+      "step": 128610
+    },
+    {
+      "epoch": 0.9310372284595395,
+      "grad_norm": 0.18212413787841797,
+      "learning_rate": 4.068970010206519e-06,
+      "loss": 0.8671,
+      "step": 128620
+    },
+    {
+      "epoch": 0.9311096151201257,
+      "grad_norm": 0.14590677618980408,
+      "learning_rate": 4.0688976235459336e-06,
+      "loss": 0.8758,
+      "step": 128630
+    },
+    {
+      "epoch": 0.9311820017807119,
+      "grad_norm": 0.15685223042964935,
+      "learning_rate": 4.068825236885347e-06,
+      "loss": 0.8634,
+      "step": 128640
+    },
+    {
+      "epoch": 0.931254388441298,
+      "grad_norm": 0.1610851287841797,
+      "learning_rate": 4.068752850224761e-06,
+      "loss": 0.8678,
+      "step": 128650
+    },
+    {
+      "epoch": 0.9313267751018842,
+      "grad_norm": 0.1528211534023285,
+      "learning_rate": 4.068680463564174e-06,
+      "loss": 0.8797,
+      "step": 128660
+    },
+    {
+      "epoch": 0.9313991617624704,
+      "grad_norm": 0.1573326289653778,
+      "learning_rate": 4.068608076903589e-06,
+      "loss": 0.8785,
+      "step": 128670
+    },
+    {
+      "epoch": 0.9314715484230566,
+      "grad_norm": 0.15934540331363678,
+      "learning_rate": 4.0685356902430025e-06,
+      "loss": 0.8626,
+      "step": 128680
+    },
+    {
+      "epoch": 0.9315439350836427,
+      "grad_norm": 0.14272333681583405,
+      "learning_rate": 4.068463303582416e-06,
+      "loss": 0.8685,
+      "step": 128690
+    },
+    {
+      "epoch": 0.9316163217442289,
+      "grad_norm": 0.1473149210214615,
+      "learning_rate": 4.06839091692183e-06,
+      "loss": 0.8655,
+      "step": 128700
+    },
+    {
+      "epoch": 0.9316887084048152,
+      "grad_norm": 0.1499226689338684,
+      "learning_rate": 4.068318530261244e-06,
+      "loss": 0.8633,
+      "step": 128710
+    },
+    {
+      "epoch": 0.9317610950654014,
+      "grad_norm": 0.14663182199001312,
+      "learning_rate": 4.068246143600658e-06,
+      "loss": 0.852,
+      "step": 128720
+    },
+    {
+      "epoch": 0.9318334817259876,
+      "grad_norm": 0.15294471383094788,
+      "learning_rate": 4.068173756940071e-06,
+      "loss": 0.8769,
+      "step": 128730
+    },
+    {
+      "epoch": 0.9319058683865737,
+      "grad_norm": 0.16538985073566437,
+      "learning_rate": 4.068101370279485e-06,
+      "loss": 0.8754,
+      "step": 128740
+    },
+    {
+      "epoch": 0.9319782550471599,
+      "grad_norm": 0.14424091577529907,
+      "learning_rate": 4.0680289836188995e-06,
+      "loss": 0.8745,
+      "step": 128750
+    },
+    {
+      "epoch": 0.9320506417077461,
+      "grad_norm": 0.18575291335582733,
+      "learning_rate": 4.067956596958313e-06,
+      "loss": 0.8542,
+      "step": 128760
+    },
+    {
+      "epoch": 0.9321230283683323,
+      "grad_norm": 0.156608447432518,
+      "learning_rate": 4.067884210297727e-06,
+      "loss": 0.8703,
+      "step": 128770
+    },
+    {
+      "epoch": 0.9321954150289185,
+      "grad_norm": 0.16681206226348877,
+      "learning_rate": 4.06781182363714e-06,
+      "loss": 0.8691,
+      "step": 128780
+    },
+    {
+      "epoch": 0.9322678016895046,
+      "grad_norm": 0.18856379389762878,
+      "learning_rate": 4.067739436976554e-06,
+      "loss": 0.8727,
+      "step": 128790
+    },
+    {
+      "epoch": 0.9323401883500908,
+      "grad_norm": 0.15221746265888214,
+      "learning_rate": 4.0676670503159684e-06,
+      "loss": 0.8771,
+      "step": 128800
+    },
+    {
+      "epoch": 0.932412575010677,
+      "grad_norm": 0.15797759592533112,
+      "learning_rate": 4.067594663655382e-06,
+      "loss": 0.8682,
+      "step": 128810
+    },
+    {
+      "epoch": 0.9324849616712633,
+      "grad_norm": 0.1687854826450348,
+      "learning_rate": 4.067522276994796e-06,
+      "loss": 0.8743,
+      "step": 128820
+    },
+    {
+      "epoch": 0.9325573483318494,
+      "grad_norm": 0.15136481821537018,
+      "learning_rate": 4.067449890334209e-06,
+      "loss": 0.8712,
+      "step": 128830
+    },
+    {
+      "epoch": 0.9326297349924356,
+      "grad_norm": 0.1667127013206482,
+      "learning_rate": 4.067377503673624e-06,
+      "loss": 0.862,
+      "step": 128840
+    },
+    {
+      "epoch": 0.9327021216530218,
+      "grad_norm": 0.1545359492301941,
+      "learning_rate": 4.067305117013037e-06,
+      "loss": 0.8564,
+      "step": 128850
+    },
+    {
+      "epoch": 0.932774508313608,
+      "grad_norm": 0.18718428909778595,
+      "learning_rate": 4.067232730352451e-06,
+      "loss": 0.8584,
+      "step": 128860
+    },
+    {
+      "epoch": 0.9328468949741942,
+      "grad_norm": 0.14906947314739227,
+      "learning_rate": 4.067160343691865e-06,
+      "loss": 0.8737,
+      "step": 128870
+    },
+    {
+      "epoch": 0.9329192816347803,
+      "grad_norm": 0.1455237865447998,
+      "learning_rate": 4.067087957031279e-06,
+      "loss": 0.872,
+      "step": 128880
+    },
+    {
+      "epoch": 0.9329916682953665,
+      "grad_norm": 0.1723933070898056,
+      "learning_rate": 4.067015570370693e-06,
+      "loss": 0.8785,
+      "step": 128890
+    },
+    {
+      "epoch": 0.9330640549559527,
+      "grad_norm": 0.1570318043231964,
+      "learning_rate": 4.066943183710106e-06,
+      "loss": 0.8693,
+      "step": 128900
+    },
+    {
+      "epoch": 0.9331364416165389,
+      "grad_norm": 0.16334332525730133,
+      "learning_rate": 4.06687079704952e-06,
+      "loss": 0.8744,
+      "step": 128910
+    },
+    {
+      "epoch": 0.9332088282771251,
+      "grad_norm": 0.15056326985359192,
+      "learning_rate": 4.066798410388934e-06,
+      "loss": 0.8666,
+      "step": 128920
+    },
+    {
+      "epoch": 0.9332812149377113,
+      "grad_norm": 0.15324917435646057,
+      "learning_rate": 4.066726023728348e-06,
+      "loss": 0.8814,
+      "step": 128930
+    },
+    {
+      "epoch": 0.9333536015982975,
+      "grad_norm": 0.1748197376728058,
+      "learning_rate": 4.066653637067762e-06,
+      "loss": 0.869,
+      "step": 128940
+    },
+    {
+      "epoch": 0.9334259882588837,
+      "grad_norm": 0.1483113169670105,
+      "learning_rate": 4.066581250407175e-06,
+      "loss": 0.8726,
+      "step": 128950
+    },
+    {
+      "epoch": 0.9334983749194699,
+      "grad_norm": 0.14639945328235626,
+      "learning_rate": 4.06650886374659e-06,
+      "loss": 0.8641,
+      "step": 128960
+    },
+    {
+      "epoch": 0.933570761580056,
+      "grad_norm": 0.1470886915922165,
+      "learning_rate": 4.066436477086003e-06,
+      "loss": 0.8671,
+      "step": 128970
+    },
+    {
+      "epoch": 0.9336431482406422,
+      "grad_norm": 0.15743905305862427,
+      "learning_rate": 4.066364090425416e-06,
+      "loss": 0.8751,
+      "step": 128980
+    },
+    {
+      "epoch": 0.9337155349012284,
+      "grad_norm": 0.14820967614650726,
+      "learning_rate": 4.0662917037648305e-06,
+      "loss": 0.8643,
+      "step": 128990
+    },
+    {
+      "epoch": 0.9337879215618146,
+      "grad_norm": 0.15819282829761505,
+      "learning_rate": 4.066219317104244e-06,
+      "loss": 0.8715,
+      "step": 129000
+    },
+    {
+      "epoch": 0.9338603082224007,
+      "grad_norm": 0.16119031608104706,
+      "learning_rate": 4.066146930443658e-06,
+      "loss": 0.885,
+      "step": 129010
+    },
+    {
+      "epoch": 0.9339326948829869,
+      "grad_norm": 0.1570100635290146,
+      "learning_rate": 4.066074543783071e-06,
+      "loss": 0.863,
+      "step": 129020
+    },
+    {
+      "epoch": 0.9340050815435732,
+      "grad_norm": 0.15377597510814667,
+      "learning_rate": 4.066002157122486e-06,
+      "loss": 0.8705,
+      "step": 129030
+    },
+    {
+      "epoch": 0.9340774682041594,
+      "grad_norm": 0.1488153040409088,
+      "learning_rate": 4.0659297704618995e-06,
+      "loss": 0.8709,
+      "step": 129040
+    },
+    {
+      "epoch": 0.9341498548647456,
+      "grad_norm": 0.1504792720079422,
+      "learning_rate": 4.065857383801313e-06,
+      "loss": 0.8645,
+      "step": 129050
+    },
+    {
+      "epoch": 0.9342222415253317,
+      "grad_norm": 0.15076880156993866,
+      "learning_rate": 4.065784997140727e-06,
+      "loss": 0.8786,
+      "step": 129060
+    },
+    {
+      "epoch": 0.9342946281859179,
+      "grad_norm": 0.15644735097885132,
+      "learning_rate": 4.065712610480141e-06,
+      "loss": 0.8497,
+      "step": 129070
+    },
+    {
+      "epoch": 0.9343670148465041,
+      "grad_norm": 0.15488304197788239,
+      "learning_rate": 4.065640223819555e-06,
+      "loss": 0.8716,
+      "step": 129080
+    },
+    {
+      "epoch": 0.9344394015070903,
+      "grad_norm": 0.15287454426288605,
+      "learning_rate": 4.065567837158968e-06,
+      "loss": 0.8699,
+      "step": 129090
+    },
+    {
+      "epoch": 0.9345117881676764,
+      "grad_norm": 0.15674547851085663,
+      "learning_rate": 4.065495450498382e-06,
+      "loss": 0.8544,
+      "step": 129100
+    },
+    {
+      "epoch": 0.9345841748282626,
+      "grad_norm": 0.14883847534656525,
+      "learning_rate": 4.0654230638377965e-06,
+      "loss": 0.8748,
+      "step": 129110
+    },
+    {
+      "epoch": 0.9346565614888488,
+      "grad_norm": 0.1638948619365692,
+      "learning_rate": 4.06535067717721e-06,
+      "loss": 0.8734,
+      "step": 129120
+    },
+    {
+      "epoch": 0.934728948149435,
+      "grad_norm": 0.1496628224849701,
+      "learning_rate": 4.065278290516624e-06,
+      "loss": 0.8622,
+      "step": 129130
+    },
+    {
+      "epoch": 0.9348013348100213,
+      "grad_norm": 0.15044906735420227,
+      "learning_rate": 4.065205903856037e-06,
+      "loss": 0.8845,
+      "step": 129140
+    },
+    {
+      "epoch": 0.9348737214706074,
+      "grad_norm": 0.1459684520959854,
+      "learning_rate": 4.065133517195452e-06,
+      "loss": 0.8753,
+      "step": 129150
+    },
+    {
+      "epoch": 0.9349461081311936,
+      "grad_norm": 0.14683480560779572,
+      "learning_rate": 4.065061130534865e-06,
+      "loss": 0.8597,
+      "step": 129160
+    },
+    {
+      "epoch": 0.9350184947917798,
+      "grad_norm": 0.17215889692306519,
+      "learning_rate": 4.064988743874279e-06,
+      "loss": 0.8613,
+      "step": 129170
+    },
+    {
+      "epoch": 0.935090881452366,
+      "grad_norm": 0.15645846724510193,
+      "learning_rate": 4.064916357213693e-06,
+      "loss": 0.8649,
+      "step": 129180
+    },
+    {
+      "epoch": 0.9351632681129521,
+      "grad_norm": 0.15681461989879608,
+      "learning_rate": 4.064843970553107e-06,
+      "loss": 0.8675,
+      "step": 129190
+    },
+    {
+      "epoch": 0.9352356547735383,
+      "grad_norm": 0.15553514659404755,
+      "learning_rate": 4.064771583892521e-06,
+      "loss": 0.865,
+      "step": 129200
+    },
+    {
+      "epoch": 0.9353080414341245,
+      "grad_norm": 0.1579878181219101,
+      "learning_rate": 4.064699197231934e-06,
+      "loss": 0.8564,
+      "step": 129210
+    },
+    {
+      "epoch": 0.9353804280947107,
+      "grad_norm": 0.1529635339975357,
+      "learning_rate": 4.064626810571348e-06,
+      "loss": 0.8701,
+      "step": 129220
+    },
+    {
+      "epoch": 0.9354528147552968,
+      "grad_norm": 0.17497135698795319,
+      "learning_rate": 4.064554423910762e-06,
+      "loss": 0.858,
+      "step": 129230
+    },
+    {
+      "epoch": 0.9355252014158831,
+      "grad_norm": 0.1518746018409729,
+      "learning_rate": 4.064482037250176e-06,
+      "loss": 0.8595,
+      "step": 129240
+    },
+    {
+      "epoch": 0.9355975880764693,
+      "grad_norm": 0.15486803650856018,
+      "learning_rate": 4.06440965058959e-06,
+      "loss": 0.8762,
+      "step": 129250
+    },
+    {
+      "epoch": 0.9356699747370555,
+      "grad_norm": 0.16987523436546326,
+      "learning_rate": 4.064337263929003e-06,
+      "loss": 0.8698,
+      "step": 129260
+    },
+    {
+      "epoch": 0.9357423613976417,
+      "grad_norm": 0.15838555991649628,
+      "learning_rate": 4.064264877268418e-06,
+      "loss": 0.8614,
+      "step": 129270
+    },
+    {
+      "epoch": 0.9358147480582278,
+      "grad_norm": 0.15161006152629852,
+      "learning_rate": 4.064192490607831e-06,
+      "loss": 0.8582,
+      "step": 129280
+    },
+    {
+      "epoch": 0.935887134718814,
+      "grad_norm": 0.16381420195102692,
+      "learning_rate": 4.064120103947245e-06,
+      "loss": 0.8612,
+      "step": 129290
+    },
+    {
+      "epoch": 0.9359595213794002,
+      "grad_norm": 0.15253403782844543,
+      "learning_rate": 4.0640477172866586e-06,
+      "loss": 0.8672,
+      "step": 129300
+    },
+    {
+      "epoch": 0.9360319080399864,
+      "grad_norm": 0.14792008697986603,
+      "learning_rate": 4.063975330626073e-06,
+      "loss": 0.8663,
+      "step": 129310
+    },
+    {
+      "epoch": 0.9361042947005725,
+      "grad_norm": 0.15635430812835693,
+      "learning_rate": 4.063902943965487e-06,
+      "loss": 0.8706,
+      "step": 129320
+    },
+    {
+      "epoch": 0.9361766813611587,
+      "grad_norm": 0.16064366698265076,
+      "learning_rate": 4.0638305573049e-06,
+      "loss": 0.8765,
+      "step": 129330
+    },
+    {
+      "epoch": 0.9362490680217449,
+      "grad_norm": 0.156932070851326,
+      "learning_rate": 4.063758170644314e-06,
+      "loss": 0.881,
+      "step": 129340
+    },
+    {
+      "epoch": 0.9363214546823312,
+      "grad_norm": 0.1553388088941574,
+      "learning_rate": 4.063685783983728e-06,
+      "loss": 0.8881,
+      "step": 129350
+    },
+    {
+      "epoch": 0.9363938413429174,
+      "grad_norm": 0.13980628550052643,
+      "learning_rate": 4.063613397323142e-06,
+      "loss": 0.8828,
+      "step": 129360
+    },
+    {
+      "epoch": 0.9364662280035035,
+      "grad_norm": 0.14694495499134064,
+      "learning_rate": 4.0635410106625556e-06,
+      "loss": 0.8716,
+      "step": 129370
+    },
+    {
+      "epoch": 0.9365386146640897,
+      "grad_norm": 0.15779894590377808,
+      "learning_rate": 4.063468624001969e-06,
+      "loss": 0.8684,
+      "step": 129380
+    },
+    {
+      "epoch": 0.9366110013246759,
+      "grad_norm": 0.15921930968761444,
+      "learning_rate": 4.063396237341384e-06,
+      "loss": 0.868,
+      "step": 129390
+    },
+    {
+      "epoch": 0.9366833879852621,
+      "grad_norm": 0.15195946395397186,
+      "learning_rate": 4.063323850680797e-06,
+      "loss": 0.8714,
+      "step": 129400
+    },
+    {
+      "epoch": 0.9367557746458482,
+      "grad_norm": 0.1530548632144928,
+      "learning_rate": 4.063251464020211e-06,
+      "loss": 0.8691,
+      "step": 129410
+    },
+    {
+      "epoch": 0.9368281613064344,
+      "grad_norm": 0.15170645713806152,
+      "learning_rate": 4.0631790773596245e-06,
+      "loss": 0.8762,
+      "step": 129420
+    },
+    {
+      "epoch": 0.9369005479670206,
+      "grad_norm": 0.15962299704551697,
+      "learning_rate": 4.063106690699038e-06,
+      "loss": 0.871,
+      "step": 129430
+    },
+    {
+      "epoch": 0.9369729346276068,
+      "grad_norm": 0.16227169334888458,
+      "learning_rate": 4.0630343040384526e-06,
+      "loss": 0.8512,
+      "step": 129440
+    },
+    {
+      "epoch": 0.9370453212881931,
+      "grad_norm": 0.1503174751996994,
+      "learning_rate": 4.062961917377866e-06,
+      "loss": 0.8683,
+      "step": 129450
+    },
+    {
+      "epoch": 0.9371177079487792,
+      "grad_norm": 0.1538417786359787,
+      "learning_rate": 4.06288953071728e-06,
+      "loss": 0.8637,
+      "step": 129460
+    },
+    {
+      "epoch": 0.9371900946093654,
+      "grad_norm": 0.14984267950057983,
+      "learning_rate": 4.062817144056693e-06,
+      "loss": 0.8632,
+      "step": 129470
+    },
+    {
+      "epoch": 0.9372624812699516,
+      "grad_norm": 0.1558646708726883,
+      "learning_rate": 4.062744757396108e-06,
+      "loss": 0.8677,
+      "step": 129480
+    },
+    {
+      "epoch": 0.9373348679305378,
+      "grad_norm": 0.16419059038162231,
+      "learning_rate": 4.0626723707355215e-06,
+      "loss": 0.8634,
+      "step": 129490
+    },
+    {
+      "epoch": 0.937407254591124,
+      "grad_norm": 0.15692633390426636,
+      "learning_rate": 4.062599984074935e-06,
+      "loss": 0.8717,
+      "step": 129500
+    },
+    {
+      "epoch": 0.9374796412517101,
+      "grad_norm": 0.15984247624874115,
+      "learning_rate": 4.062527597414349e-06,
+      "loss": 0.8783,
+      "step": 129510
+    },
+    {
+      "epoch": 0.9375520279122963,
+      "grad_norm": 0.15747103095054626,
+      "learning_rate": 4.062455210753762e-06,
+      "loss": 0.8726,
+      "step": 129520
+    },
+    {
+      "epoch": 0.9376244145728825,
+      "grad_norm": 0.15458211302757263,
+      "learning_rate": 4.062382824093176e-06,
+      "loss": 0.8605,
+      "step": 129530
+    },
+    {
+      "epoch": 0.9376968012334687,
+      "grad_norm": 0.14359694719314575,
+      "learning_rate": 4.0623104374325904e-06,
+      "loss": 0.8852,
+      "step": 129540
+    },
+    {
+      "epoch": 0.9377691878940548,
+      "grad_norm": 0.16120558977127075,
+      "learning_rate": 4.062238050772004e-06,
+      "loss": 0.8745,
+      "step": 129550
+    },
+    {
+      "epoch": 0.9378415745546411,
+      "grad_norm": 0.15251369774341583,
+      "learning_rate": 4.062165664111418e-06,
+      "loss": 0.8723,
+      "step": 129560
+    },
+    {
+      "epoch": 0.9379139612152273,
+      "grad_norm": 0.15283681452274323,
+      "learning_rate": 4.062093277450831e-06,
+      "loss": 0.8648,
+      "step": 129570
+    },
+    {
+      "epoch": 0.9379863478758135,
+      "grad_norm": 0.15063002705574036,
+      "learning_rate": 4.062020890790245e-06,
+      "loss": 0.8652,
+      "step": 129580
+    },
+    {
+      "epoch": 0.9380587345363997,
+      "grad_norm": 0.1525861620903015,
+      "learning_rate": 4.061948504129659e-06,
+      "loss": 0.8836,
+      "step": 129590
+    },
+    {
+      "epoch": 0.9381311211969858,
+      "grad_norm": 0.1666775345802307,
+      "learning_rate": 4.061876117469073e-06,
+      "loss": 0.8762,
+      "step": 129600
+    },
+    {
+      "epoch": 0.938203507857572,
+      "grad_norm": 0.15316708385944366,
+      "learning_rate": 4.061803730808487e-06,
+      "loss": 0.8482,
+      "step": 129610
+    },
+    {
+      "epoch": 0.9382758945181582,
+      "grad_norm": 0.15615612268447876,
+      "learning_rate": 4.0617313441479e-06,
+      "loss": 0.8639,
+      "step": 129620
+    },
+    {
+      "epoch": 0.9383482811787444,
+      "grad_norm": 0.13969020545482635,
+      "learning_rate": 4.061658957487315e-06,
+      "loss": 0.8829,
+      "step": 129630
+    },
+    {
+      "epoch": 0.9384206678393305,
+      "grad_norm": 0.1508314609527588,
+      "learning_rate": 4.061586570826728e-06,
+      "loss": 0.8611,
+      "step": 129640
+    },
+    {
+      "epoch": 0.9384930544999167,
+      "grad_norm": 0.16280554234981537,
+      "learning_rate": 4.061514184166142e-06,
+      "loss": 0.883,
+      "step": 129650
+    },
+    {
+      "epoch": 0.9385654411605029,
+      "grad_norm": 0.15639394521713257,
+      "learning_rate": 4.0614417975055555e-06,
+      "loss": 0.8711,
+      "step": 129660
+    },
+    {
+      "epoch": 0.9386378278210892,
+      "grad_norm": 0.1565892994403839,
+      "learning_rate": 4.06136941084497e-06,
+      "loss": 0.8667,
+      "step": 129670
+    },
+    {
+      "epoch": 0.9387102144816754,
+      "grad_norm": 0.16035987436771393,
+      "learning_rate": 4.061297024184384e-06,
+      "loss": 0.8624,
+      "step": 129680
+    },
+    {
+      "epoch": 0.9387826011422615,
+      "grad_norm": 0.15157711505889893,
+      "learning_rate": 4.061224637523797e-06,
+      "loss": 0.8751,
+      "step": 129690
+    },
+    {
+      "epoch": 0.9388549878028477,
+      "grad_norm": 0.15233691036701202,
+      "learning_rate": 4.061152250863211e-06,
+      "loss": 0.8743,
+      "step": 129700
+    },
+    {
+      "epoch": 0.9389273744634339,
+      "grad_norm": 0.15194357931613922,
+      "learning_rate": 4.061079864202625e-06,
+      "loss": 0.875,
+      "step": 129710
+    },
+    {
+      "epoch": 0.9389997611240201,
+      "grad_norm": 0.1659833937883377,
+      "learning_rate": 4.061007477542039e-06,
+      "loss": 0.8714,
+      "step": 129720
+    },
+    {
+      "epoch": 0.9390721477846062,
+      "grad_norm": 0.15151038765907288,
+      "learning_rate": 4.0609350908814525e-06,
+      "loss": 0.863,
+      "step": 129730
+    },
+    {
+      "epoch": 0.9391445344451924,
+      "grad_norm": 0.16103753447532654,
+      "learning_rate": 4.060862704220866e-06,
+      "loss": 0.8715,
+      "step": 129740
+    },
+    {
+      "epoch": 0.9392169211057786,
+      "grad_norm": 0.1853325515985489,
+      "learning_rate": 4.060790317560281e-06,
+      "loss": 0.8664,
+      "step": 129750
+    },
+    {
+      "epoch": 0.9392893077663648,
+      "grad_norm": 0.15947386622428894,
+      "learning_rate": 4.060717930899694e-06,
+      "loss": 0.8698,
+      "step": 129760
+    },
+    {
+      "epoch": 0.939361694426951,
+      "grad_norm": 0.14465877413749695,
+      "learning_rate": 4.060645544239108e-06,
+      "loss": 0.8654,
+      "step": 129770
+    },
+    {
+      "epoch": 0.9394340810875372,
+      "grad_norm": 0.15076486766338348,
+      "learning_rate": 4.0605731575785215e-06,
+      "loss": 0.8733,
+      "step": 129780
+    },
+    {
+      "epoch": 0.9395064677481234,
+      "grad_norm": 0.14492474496364594,
+      "learning_rate": 4.060500770917936e-06,
+      "loss": 0.8629,
+      "step": 129790
+    },
+    {
+      "epoch": 0.9395788544087096,
+      "grad_norm": 0.15159480273723602,
+      "learning_rate": 4.0604283842573495e-06,
+      "loss": 0.875,
+      "step": 129800
+    },
+    {
+      "epoch": 0.9396512410692958,
+      "grad_norm": 0.14613956212997437,
+      "learning_rate": 4.060355997596763e-06,
+      "loss": 0.8781,
+      "step": 129810
+    },
+    {
+      "epoch": 0.9397236277298819,
+      "grad_norm": 0.16877029836177826,
+      "learning_rate": 4.060283610936177e-06,
+      "loss": 0.8768,
+      "step": 129820
+    },
+    {
+      "epoch": 0.9397960143904681,
+      "grad_norm": 0.15022648870944977,
+      "learning_rate": 4.060211224275591e-06,
+      "loss": 0.8699,
+      "step": 129830
+    },
+    {
+      "epoch": 0.9398684010510543,
+      "grad_norm": 0.1675875037908554,
+      "learning_rate": 4.060138837615005e-06,
+      "loss": 0.8626,
+      "step": 129840
+    },
+    {
+      "epoch": 0.9399407877116405,
+      "grad_norm": 0.15652434527873993,
+      "learning_rate": 4.0600664509544185e-06,
+      "loss": 0.8588,
+      "step": 129850
+    },
+    {
+      "epoch": 0.9400131743722266,
+      "grad_norm": 0.16846764087677002,
+      "learning_rate": 4.059994064293832e-06,
+      "loss": 0.859,
+      "step": 129860
+    },
+    {
+      "epoch": 0.9400855610328128,
+      "grad_norm": 0.1613306701183319,
+      "learning_rate": 4.0599216776332465e-06,
+      "loss": 0.8593,
+      "step": 129870
+    },
+    {
+      "epoch": 0.9401579476933991,
+      "grad_norm": 0.14691142737865448,
+      "learning_rate": 4.05984929097266e-06,
+      "loss": 0.8657,
+      "step": 129880
+    },
+    {
+      "epoch": 0.9402303343539853,
+      "grad_norm": 0.20344461500644684,
+      "learning_rate": 4.059776904312074e-06,
+      "loss": 0.8688,
+      "step": 129890
+    },
+    {
+      "epoch": 0.9403027210145715,
+      "grad_norm": 0.1496155560016632,
+      "learning_rate": 4.059704517651487e-06,
+      "loss": 0.862,
+      "step": 129900
+    },
+    {
+      "epoch": 0.9403751076751576,
+      "grad_norm": 0.14814414083957672,
+      "learning_rate": 4.059632130990902e-06,
+      "loss": 0.8479,
+      "step": 129910
+    },
+    {
+      "epoch": 0.9404474943357438,
+      "grad_norm": 0.1548141986131668,
+      "learning_rate": 4.0595597443303155e-06,
+      "loss": 0.8626,
+      "step": 129920
+    },
+    {
+      "epoch": 0.94051988099633,
+      "grad_norm": 0.14892597496509552,
+      "learning_rate": 4.059487357669729e-06,
+      "loss": 0.8761,
+      "step": 129930
+    },
+    {
+      "epoch": 0.9405922676569162,
+      "grad_norm": 0.14724701642990112,
+      "learning_rate": 4.059414971009143e-06,
+      "loss": 0.8737,
+      "step": 129940
+    },
+    {
+      "epoch": 0.9406646543175023,
+      "grad_norm": 0.1490176022052765,
+      "learning_rate": 4.059342584348557e-06,
+      "loss": 0.8762,
+      "step": 129950
+    },
+    {
+      "epoch": 0.9407370409780885,
+      "grad_norm": 0.1549617052078247,
+      "learning_rate": 4.059270197687971e-06,
+      "loss": 0.876,
+      "step": 129960
+    },
+    {
+      "epoch": 0.9408094276386747,
+      "grad_norm": 0.1603880673646927,
+      "learning_rate": 4.059197811027384e-06,
+      "loss": 0.8737,
+      "step": 129970
+    },
+    {
+      "epoch": 0.9408818142992609,
+      "grad_norm": 0.14581482112407684,
+      "learning_rate": 4.059125424366798e-06,
+      "loss": 0.8723,
+      "step": 129980
+    },
+    {
+      "epoch": 0.9409542009598472,
+      "grad_norm": 0.14896036684513092,
+      "learning_rate": 4.0590530377062125e-06,
+      "loss": 0.8674,
+      "step": 129990
+    },
+    {
+      "epoch": 0.9410265876204333,
+      "grad_norm": 0.1456650048494339,
+      "learning_rate": 4.058980651045626e-06,
+      "loss": 0.8697,
+      "step": 130000
+    },
+    {
+      "epoch": 0.9410989742810195,
+      "grad_norm": 0.16549433767795563,
+      "learning_rate": 4.05890826438504e-06,
+      "loss": 0.8668,
+      "step": 130010
+    },
+    {
+      "epoch": 0.9411713609416057,
+      "grad_norm": 0.15788212418556213,
+      "learning_rate": 4.058835877724453e-06,
+      "loss": 0.8884,
+      "step": 130020
+    },
+    {
+      "epoch": 0.9412437476021919,
+      "grad_norm": 0.20761068165302277,
+      "learning_rate": 4.058763491063867e-06,
+      "loss": 0.8641,
+      "step": 130030
+    },
+    {
+      "epoch": 0.941316134262778,
+      "grad_norm": 0.1683483123779297,
+      "learning_rate": 4.0586911044032806e-06,
+      "loss": 0.852,
+      "step": 130040
+    },
+    {
+      "epoch": 0.9413885209233642,
+      "grad_norm": 0.16688086092472076,
+      "learning_rate": 4.058618717742694e-06,
+      "loss": 0.8703,
+      "step": 130050
+    },
+    {
+      "epoch": 0.9414609075839504,
+      "grad_norm": 0.15116585791110992,
+      "learning_rate": 4.058546331082109e-06,
+      "loss": 0.8814,
+      "step": 130060
+    },
+    {
+      "epoch": 0.9415332942445366,
+      "grad_norm": 0.1498250663280487,
+      "learning_rate": 4.058473944421522e-06,
+      "loss": 0.8785,
+      "step": 130070
+    },
+    {
+      "epoch": 0.9416056809051228,
+      "grad_norm": 0.14295758306980133,
+      "learning_rate": 4.058401557760936e-06,
+      "loss": 0.8625,
+      "step": 130080
+    },
+    {
+      "epoch": 0.941678067565709,
+      "grad_norm": 0.16779643297195435,
+      "learning_rate": 4.0583291711003495e-06,
+      "loss": 0.8589,
+      "step": 130090
+    },
+    {
+      "epoch": 0.9417504542262952,
+      "grad_norm": 0.14837302267551422,
+      "learning_rate": 4.058256784439764e-06,
+      "loss": 0.8672,
+      "step": 130100
+    },
+    {
+      "epoch": 0.9418228408868814,
+      "grad_norm": 0.15054334700107574,
+      "learning_rate": 4.0581843977791776e-06,
+      "loss": 0.856,
+      "step": 130110
+    },
+    {
+      "epoch": 0.9418952275474676,
+      "grad_norm": 0.1450866162776947,
+      "learning_rate": 4.058112011118591e-06,
+      "loss": 0.8576,
+      "step": 130120
+    },
+    {
+      "epoch": 0.9419676142080537,
+      "grad_norm": 0.1485680341720581,
+      "learning_rate": 4.058039624458005e-06,
+      "loss": 0.8712,
+      "step": 130130
+    },
+    {
+      "epoch": 0.9420400008686399,
+      "grad_norm": 0.14476777613162994,
+      "learning_rate": 4.057967237797419e-06,
+      "loss": 0.8852,
+      "step": 130140
+    },
+    {
+      "epoch": 0.9421123875292261,
+      "grad_norm": 0.14453761279582977,
+      "learning_rate": 4.057894851136833e-06,
+      "loss": 0.8608,
+      "step": 130150
+    },
+    {
+      "epoch": 0.9421847741898123,
+      "grad_norm": 0.1491234302520752,
+      "learning_rate": 4.0578224644762465e-06,
+      "loss": 0.8662,
+      "step": 130160
+    },
+    {
+      "epoch": 0.9422571608503985,
+      "grad_norm": 0.1594741940498352,
+      "learning_rate": 4.05775007781566e-06,
+      "loss": 0.8748,
+      "step": 130170
+    },
+    {
+      "epoch": 0.9423295475109846,
+      "grad_norm": 0.15636101365089417,
+      "learning_rate": 4.0576776911550746e-06,
+      "loss": 0.8674,
+      "step": 130180
+    },
+    {
+      "epoch": 0.9424019341715708,
+      "grad_norm": 0.1590276062488556,
+      "learning_rate": 4.057605304494488e-06,
+      "loss": 0.8547,
+      "step": 130190
+    },
+    {
+      "epoch": 0.9424743208321571,
+      "grad_norm": 0.15569545328617096,
+      "learning_rate": 4.057532917833902e-06,
+      "loss": 0.8607,
+      "step": 130200
+    },
+    {
+      "epoch": 0.9425467074927433,
+      "grad_norm": 0.15706123411655426,
+      "learning_rate": 4.057460531173315e-06,
+      "loss": 0.8647,
+      "step": 130210
+    },
+    {
+      "epoch": 0.9426190941533295,
+      "grad_norm": 0.17118285596370697,
+      "learning_rate": 4.057388144512729e-06,
+      "loss": 0.8688,
+      "step": 130220
+    },
+    {
+      "epoch": 0.9426914808139156,
+      "grad_norm": 0.1491112858057022,
+      "learning_rate": 4.0573157578521435e-06,
+      "loss": 0.8635,
+      "step": 130230
+    },
+    {
+      "epoch": 0.9427638674745018,
+      "grad_norm": 0.15699629485607147,
+      "learning_rate": 4.057243371191557e-06,
+      "loss": 0.8717,
+      "step": 130240
+    },
+    {
+      "epoch": 0.942836254135088,
+      "grad_norm": 0.1544734388589859,
+      "learning_rate": 4.057170984530971e-06,
+      "loss": 0.8608,
+      "step": 130250
+    },
+    {
+      "epoch": 0.9429086407956742,
+      "grad_norm": 0.15462660789489746,
+      "learning_rate": 4.057098597870384e-06,
+      "loss": 0.8624,
+      "step": 130260
+    },
+    {
+      "epoch": 0.9429810274562603,
+      "grad_norm": 0.1702507883310318,
+      "learning_rate": 4.057026211209799e-06,
+      "loss": 0.8809,
+      "step": 130270
+    },
+    {
+      "epoch": 0.9430534141168465,
+      "grad_norm": 0.1495058238506317,
+      "learning_rate": 4.0569538245492124e-06,
+      "loss": 0.8564,
+      "step": 130280
+    },
+    {
+      "epoch": 0.9431258007774327,
+      "grad_norm": 0.1570599526166916,
+      "learning_rate": 4.056881437888626e-06,
+      "loss": 0.8738,
+      "step": 130290
+    },
+    {
+      "epoch": 0.943198187438019,
+      "grad_norm": 0.14418993890285492,
+      "learning_rate": 4.05680905122804e-06,
+      "loss": 0.8863,
+      "step": 130300
+    },
+    {
+      "epoch": 0.9432705740986052,
+      "grad_norm": 0.1544889211654663,
+      "learning_rate": 4.056736664567454e-06,
+      "loss": 0.878,
+      "step": 130310
+    },
+    {
+      "epoch": 0.9433429607591913,
+      "grad_norm": 0.151493638753891,
+      "learning_rate": 4.056664277906868e-06,
+      "loss": 0.8675,
+      "step": 130320
+    },
+    {
+      "epoch": 0.9434153474197775,
+      "grad_norm": 0.15243935585021973,
+      "learning_rate": 4.056591891246281e-06,
+      "loss": 0.8659,
+      "step": 130330
+    },
+    {
+      "epoch": 0.9434877340803637,
+      "grad_norm": 0.15640780329704285,
+      "learning_rate": 4.056519504585695e-06,
+      "loss": 0.8789,
+      "step": 130340
+    },
+    {
+      "epoch": 0.9435601207409499,
+      "grad_norm": 0.16906537115573883,
+      "learning_rate": 4.0564471179251094e-06,
+      "loss": 0.8718,
+      "step": 130350
+    },
+    {
+      "epoch": 0.943632507401536,
+      "grad_norm": 0.13774484395980835,
+      "learning_rate": 4.056374731264523e-06,
+      "loss": 0.8684,
+      "step": 130360
+    },
+    {
+      "epoch": 0.9437048940621222,
+      "grad_norm": 0.18474681675434113,
+      "learning_rate": 4.056302344603937e-06,
+      "loss": 0.8737,
+      "step": 130370
+    },
+    {
+      "epoch": 0.9437772807227084,
+      "grad_norm": 0.15171298384666443,
+      "learning_rate": 4.05622995794335e-06,
+      "loss": 0.8742,
+      "step": 130380
+    },
+    {
+      "epoch": 0.9438496673832946,
+      "grad_norm": 0.1634998321533203,
+      "learning_rate": 4.056157571282765e-06,
+      "loss": 0.8671,
+      "step": 130390
+    },
+    {
+      "epoch": 0.9439220540438807,
+      "grad_norm": 0.15286368131637573,
+      "learning_rate": 4.056085184622178e-06,
+      "loss": 0.874,
+      "step": 130400
+    },
+    {
+      "epoch": 0.943994440704467,
+      "grad_norm": 0.17988328635692596,
+      "learning_rate": 4.056012797961592e-06,
+      "loss": 0.8764,
+      "step": 130410
+    },
+    {
+      "epoch": 0.9440668273650532,
+      "grad_norm": 0.1676689237356186,
+      "learning_rate": 4.055940411301006e-06,
+      "loss": 0.8714,
+      "step": 130420
+    },
+    {
+      "epoch": 0.9441392140256394,
+      "grad_norm": 0.15611009299755096,
+      "learning_rate": 4.05586802464042e-06,
+      "loss": 0.8603,
+      "step": 130430
+    },
+    {
+      "epoch": 0.9442116006862256,
+      "grad_norm": 0.1466500610113144,
+      "learning_rate": 4.055795637979834e-06,
+      "loss": 0.8706,
+      "step": 130440
+    },
+    {
+      "epoch": 0.9442839873468117,
+      "grad_norm": 0.1896083503961563,
+      "learning_rate": 4.055723251319247e-06,
+      "loss": 0.8681,
+      "step": 130450
+    },
+    {
+      "epoch": 0.9443563740073979,
+      "grad_norm": 0.15304064750671387,
+      "learning_rate": 4.055650864658661e-06,
+      "loss": 0.8754,
+      "step": 130460
+    },
+    {
+      "epoch": 0.9444287606679841,
+      "grad_norm": 0.15458323061466217,
+      "learning_rate": 4.055578477998075e-06,
+      "loss": 0.8714,
+      "step": 130470
+    },
+    {
+      "epoch": 0.9445011473285703,
+      "grad_norm": 0.15180428326129913,
+      "learning_rate": 4.055506091337489e-06,
+      "loss": 0.87,
+      "step": 130480
+    },
+    {
+      "epoch": 0.9445735339891564,
+      "grad_norm": 0.1695541888475418,
+      "learning_rate": 4.055433704676903e-06,
+      "loss": 0.862,
+      "step": 130490
+    },
+    {
+      "epoch": 0.9446459206497426,
+      "grad_norm": 0.1498434692621231,
+      "learning_rate": 4.055361318016316e-06,
+      "loss": 0.8605,
+      "step": 130500
+    },
+    {
+      "epoch": 0.9447183073103288,
+      "grad_norm": 0.1435045748949051,
+      "learning_rate": 4.055288931355731e-06,
+      "loss": 0.8729,
+      "step": 130510
+    },
+    {
+      "epoch": 0.9447906939709151,
+      "grad_norm": 0.14459870755672455,
+      "learning_rate": 4.055216544695144e-06,
+      "loss": 0.8589,
+      "step": 130520
+    },
+    {
+      "epoch": 0.9448630806315013,
+      "grad_norm": 0.1518053263425827,
+      "learning_rate": 4.055144158034558e-06,
+      "loss": 0.8549,
+      "step": 130530
+    },
+    {
+      "epoch": 0.9449354672920874,
+      "grad_norm": 0.1755157858133316,
+      "learning_rate": 4.0550717713739715e-06,
+      "loss": 0.8627,
+      "step": 130540
+    },
+    {
+      "epoch": 0.9450078539526736,
+      "grad_norm": 0.16715474426746368,
+      "learning_rate": 4.054999384713386e-06,
+      "loss": 0.8732,
+      "step": 130550
+    },
+    {
+      "epoch": 0.9450802406132598,
+      "grad_norm": 0.16011476516723633,
+      "learning_rate": 4.0549269980528e-06,
+      "loss": 0.8697,
+      "step": 130560
+    },
+    {
+      "epoch": 0.945152627273846,
+      "grad_norm": 0.1612127423286438,
+      "learning_rate": 4.054854611392212e-06,
+      "loss": 0.8733,
+      "step": 130570
+    },
+    {
+      "epoch": 0.9452250139344321,
+      "grad_norm": 0.15211746096611023,
+      "learning_rate": 4.054782224731627e-06,
+      "loss": 0.8724,
+      "step": 130580
+    },
+    {
+      "epoch": 0.9452974005950183,
+      "grad_norm": 0.1653570979833603,
+      "learning_rate": 4.0547098380710405e-06,
+      "loss": 0.8681,
+      "step": 130590
+    },
+    {
+      "epoch": 0.9453697872556045,
+      "grad_norm": 0.14813409745693207,
+      "learning_rate": 4.054637451410454e-06,
+      "loss": 0.8693,
+      "step": 130600
+    },
+    {
+      "epoch": 0.9454421739161907,
+      "grad_norm": 0.15209084749221802,
+      "learning_rate": 4.054565064749868e-06,
+      "loss": 0.8623,
+      "step": 130610
+    },
+    {
+      "epoch": 0.945514560576777,
+      "grad_norm": 0.15478059649467468,
+      "learning_rate": 4.054492678089282e-06,
+      "loss": 0.874,
+      "step": 130620
+    },
+    {
+      "epoch": 0.9455869472373631,
+      "grad_norm": 0.3934093713760376,
+      "learning_rate": 4.054420291428696e-06,
+      "loss": 0.8708,
+      "step": 130630
+    },
+    {
+      "epoch": 0.9456593338979493,
+      "grad_norm": 0.1564350724220276,
+      "learning_rate": 4.054347904768109e-06,
+      "loss": 0.8693,
+      "step": 130640
+    },
+    {
+      "epoch": 0.9457317205585355,
+      "grad_norm": 0.14715883135795593,
+      "learning_rate": 4.054275518107523e-06,
+      "loss": 0.8701,
+      "step": 130650
+    },
+    {
+      "epoch": 0.9458041072191217,
+      "grad_norm": 0.1694241613149643,
+      "learning_rate": 4.0542031314469375e-06,
+      "loss": 0.8761,
+      "step": 130660
+    },
+    {
+      "epoch": 0.9458764938797078,
+      "grad_norm": 0.16506516933441162,
+      "learning_rate": 4.054130744786351e-06,
+      "loss": 0.8759,
+      "step": 130670
+    },
+    {
+      "epoch": 0.945948880540294,
+      "grad_norm": 0.14801914989948273,
+      "learning_rate": 4.054058358125765e-06,
+      "loss": 0.8667,
+      "step": 130680
+    },
+    {
+      "epoch": 0.9460212672008802,
+      "grad_norm": 0.15450334548950195,
+      "learning_rate": 4.053985971465178e-06,
+      "loss": 0.8587,
+      "step": 130690
+    },
+    {
+      "epoch": 0.9460936538614664,
+      "grad_norm": 0.1489882618188858,
+      "learning_rate": 4.053913584804593e-06,
+      "loss": 0.8715,
+      "step": 130700
+    },
+    {
+      "epoch": 0.9461660405220526,
+      "grad_norm": 0.1521863490343094,
+      "learning_rate": 4.053841198144006e-06,
+      "loss": 0.8708,
+      "step": 130710
+    },
+    {
+      "epoch": 0.9462384271826387,
+      "grad_norm": 0.15034830570220947,
+      "learning_rate": 4.05376881148342e-06,
+      "loss": 0.8874,
+      "step": 130720
+    },
+    {
+      "epoch": 0.946310813843225,
+      "grad_norm": 0.1745532602071762,
+      "learning_rate": 4.053696424822834e-06,
+      "loss": 0.8755,
+      "step": 130730
+    },
+    {
+      "epoch": 0.9463832005038112,
+      "grad_norm": 0.15382978320121765,
+      "learning_rate": 4.053624038162248e-06,
+      "loss": 0.8506,
+      "step": 130740
+    },
+    {
+      "epoch": 0.9464555871643974,
+      "grad_norm": 0.16145150363445282,
+      "learning_rate": 4.053551651501662e-06,
+      "loss": 0.8639,
+      "step": 130750
+    },
+    {
+      "epoch": 0.9465279738249835,
+      "grad_norm": 0.14968262612819672,
+      "learning_rate": 4.053479264841075e-06,
+      "loss": 0.8719,
+      "step": 130760
+    },
+    {
+      "epoch": 0.9466003604855697,
+      "grad_norm": 0.17151197791099548,
+      "learning_rate": 4.053406878180489e-06,
+      "loss": 0.8707,
+      "step": 130770
+    },
+    {
+      "epoch": 0.9466727471461559,
+      "grad_norm": 0.149154931306839,
+      "learning_rate": 4.053334491519903e-06,
+      "loss": 0.8683,
+      "step": 130780
+    },
+    {
+      "epoch": 0.9467451338067421,
+      "grad_norm": 0.19101570546627045,
+      "learning_rate": 4.053262104859317e-06,
+      "loss": 0.8669,
+      "step": 130790
+    },
+    {
+      "epoch": 0.9468175204673283,
+      "grad_norm": 0.14117294549942017,
+      "learning_rate": 4.053189718198731e-06,
+      "loss": 0.8595,
+      "step": 130800
+    },
+    {
+      "epoch": 0.9468899071279144,
+      "grad_norm": 0.14957231283187866,
+      "learning_rate": 4.053117331538144e-06,
+      "loss": 0.863,
+      "step": 130810
+    },
+    {
+      "epoch": 0.9469622937885006,
+      "grad_norm": 0.1667657345533371,
+      "learning_rate": 4.053044944877558e-06,
+      "loss": 0.8785,
+      "step": 130820
+    },
+    {
+      "epoch": 0.9470346804490869,
+      "grad_norm": 0.1488480567932129,
+      "learning_rate": 4.052972558216972e-06,
+      "loss": 0.8482,
+      "step": 130830
+    },
+    {
+      "epoch": 0.9471070671096731,
+      "grad_norm": 0.16403774917125702,
+      "learning_rate": 4.052900171556386e-06,
+      "loss": 0.8699,
+      "step": 130840
+    },
+    {
+      "epoch": 0.9471794537702592,
+      "grad_norm": 0.14479616284370422,
+      "learning_rate": 4.0528277848957996e-06,
+      "loss": 0.8675,
+      "step": 130850
+    },
+    {
+      "epoch": 0.9472518404308454,
+      "grad_norm": 0.157504141330719,
+      "learning_rate": 4.052755398235213e-06,
+      "loss": 0.862,
+      "step": 130860
+    },
+    {
+      "epoch": 0.9473242270914316,
+      "grad_norm": 0.15569128096103668,
+      "learning_rate": 4.052683011574628e-06,
+      "loss": 0.8629,
+      "step": 130870
+    },
+    {
+      "epoch": 0.9473966137520178,
+      "grad_norm": 0.15487146377563477,
+      "learning_rate": 4.052610624914041e-06,
+      "loss": 0.87,
+      "step": 130880
+    },
+    {
+      "epoch": 0.947469000412604,
+      "grad_norm": 0.152805894613266,
+      "learning_rate": 4.052538238253455e-06,
+      "loss": 0.8711,
+      "step": 130890
+    },
+    {
+      "epoch": 0.9475413870731901,
+      "grad_norm": 0.15563584864139557,
+      "learning_rate": 4.0524658515928685e-06,
+      "loss": 0.8485,
+      "step": 130900
+    },
+    {
+      "epoch": 0.9476137737337763,
+      "grad_norm": 0.1475389450788498,
+      "learning_rate": 4.052393464932283e-06,
+      "loss": 0.8782,
+      "step": 130910
+    },
+    {
+      "epoch": 0.9476861603943625,
+      "grad_norm": 0.160760760307312,
+      "learning_rate": 4.0523210782716966e-06,
+      "loss": 0.8786,
+      "step": 130920
+    },
+    {
+      "epoch": 0.9477585470549487,
+      "grad_norm": 0.1502762883901596,
+      "learning_rate": 4.05224869161111e-06,
+      "loss": 0.8633,
+      "step": 130930
+    },
+    {
+      "epoch": 0.947830933715535,
+      "grad_norm": 0.14664162695407867,
+      "learning_rate": 4.052176304950524e-06,
+      "loss": 0.878,
+      "step": 130940
+    },
+    {
+      "epoch": 0.9479033203761211,
+      "grad_norm": 0.15366783738136292,
+      "learning_rate": 4.052103918289938e-06,
+      "loss": 0.8672,
+      "step": 130950
+    },
+    {
+      "epoch": 0.9479757070367073,
+      "grad_norm": 0.1406678408384323,
+      "learning_rate": 4.052031531629352e-06,
+      "loss": 0.8716,
+      "step": 130960
+    },
+    {
+      "epoch": 0.9480480936972935,
+      "grad_norm": 0.14894571900367737,
+      "learning_rate": 4.0519591449687655e-06,
+      "loss": 0.8601,
+      "step": 130970
+    },
+    {
+      "epoch": 0.9481204803578797,
+      "grad_norm": 0.14522773027420044,
+      "learning_rate": 4.051886758308179e-06,
+      "loss": 0.8797,
+      "step": 130980
+    },
+    {
+      "epoch": 0.9481928670184658,
+      "grad_norm": 0.1636369526386261,
+      "learning_rate": 4.051814371647594e-06,
+      "loss": 0.87,
+      "step": 130990
+    },
+    {
+      "epoch": 0.948265253679052,
+      "grad_norm": 0.19674119353294373,
+      "learning_rate": 4.051741984987007e-06,
+      "loss": 0.8732,
+      "step": 131000
+    },
+    {
+      "epoch": 0.9483376403396382,
+      "grad_norm": 0.16425974667072296,
+      "learning_rate": 4.051669598326421e-06,
+      "loss": 0.8701,
+      "step": 131010
+    },
+    {
+      "epoch": 0.9484100270002244,
+      "grad_norm": 0.15251362323760986,
+      "learning_rate": 4.0515972116658344e-06,
+      "loss": 0.8686,
+      "step": 131020
+    },
+    {
+      "epoch": 0.9484824136608105,
+      "grad_norm": 0.15562012791633606,
+      "learning_rate": 4.051524825005249e-06,
+      "loss": 0.8516,
+      "step": 131030
+    },
+    {
+      "epoch": 0.9485548003213967,
+      "grad_norm": 0.14685966074466705,
+      "learning_rate": 4.0514524383446625e-06,
+      "loss": 0.8677,
+      "step": 131040
+    },
+    {
+      "epoch": 0.948627186981983,
+      "grad_norm": 0.15165336430072784,
+      "learning_rate": 4.051380051684076e-06,
+      "loss": 0.8594,
+      "step": 131050
+    },
+    {
+      "epoch": 0.9486995736425692,
+      "grad_norm": 0.13826829195022583,
+      "learning_rate": 4.05130766502349e-06,
+      "loss": 0.8599,
+      "step": 131060
+    },
+    {
+      "epoch": 0.9487719603031554,
+      "grad_norm": 0.19004106521606445,
+      "learning_rate": 4.051235278362904e-06,
+      "loss": 0.8743,
+      "step": 131070
+    },
+    {
+      "epoch": 0.9488443469637415,
+      "grad_norm": 0.1457131803035736,
+      "learning_rate": 4.051162891702318e-06,
+      "loss": 0.8736,
+      "step": 131080
+    },
+    {
+      "epoch": 0.9489167336243277,
+      "grad_norm": 0.16379250586032867,
+      "learning_rate": 4.0510905050417314e-06,
+      "loss": 0.86,
+      "step": 131090
+    },
+    {
+      "epoch": 0.9489891202849139,
+      "grad_norm": 0.14875370264053345,
+      "learning_rate": 4.051018118381145e-06,
+      "loss": 0.8776,
+      "step": 131100
+    },
+    {
+      "epoch": 0.9490615069455001,
+      "grad_norm": 0.14659325778484344,
+      "learning_rate": 4.050945731720559e-06,
+      "loss": 0.8758,
+      "step": 131110
+    },
+    {
+      "epoch": 0.9491338936060862,
+      "grad_norm": 0.15121561288833618,
+      "learning_rate": 4.050873345059972e-06,
+      "loss": 0.8544,
+      "step": 131120
+    },
+    {
+      "epoch": 0.9492062802666724,
+      "grad_norm": 0.14305542409420013,
+      "learning_rate": 4.050800958399386e-06,
+      "loss": 0.8744,
+      "step": 131130
+    },
+    {
+      "epoch": 0.9492786669272586,
+      "grad_norm": 0.15504080057144165,
+      "learning_rate": 4.0507285717388e-06,
+      "loss": 0.8705,
+      "step": 131140
+    },
+    {
+      "epoch": 0.9493510535878449,
+      "grad_norm": 0.14514702558517456,
+      "learning_rate": 4.050656185078214e-06,
+      "loss": 0.8657,
+      "step": 131150
+    },
+    {
+      "epoch": 0.9494234402484311,
+      "grad_norm": 0.15616527199745178,
+      "learning_rate": 4.050583798417628e-06,
+      "loss": 0.8626,
+      "step": 131160
+    },
+    {
+      "epoch": 0.9494958269090172,
+      "grad_norm": 0.15663272142410278,
+      "learning_rate": 4.050511411757041e-06,
+      "loss": 0.8605,
+      "step": 131170
+    },
+    {
+      "epoch": 0.9495682135696034,
+      "grad_norm": 0.1529805064201355,
+      "learning_rate": 4.050439025096456e-06,
+      "loss": 0.8636,
+      "step": 131180
+    },
+    {
+      "epoch": 0.9496406002301896,
+      "grad_norm": 0.16288772225379944,
+      "learning_rate": 4.050366638435869e-06,
+      "loss": 0.8525,
+      "step": 131190
+    },
+    {
+      "epoch": 0.9497129868907758,
+      "grad_norm": 0.14723965525627136,
+      "learning_rate": 4.050294251775283e-06,
+      "loss": 0.8596,
+      "step": 131200
+    },
+    {
+      "epoch": 0.9497853735513619,
+      "grad_norm": 0.14269182085990906,
+      "learning_rate": 4.0502218651146965e-06,
+      "loss": 0.8667,
+      "step": 131210
+    },
+    {
+      "epoch": 0.9498577602119481,
+      "grad_norm": 0.16036245226860046,
+      "learning_rate": 4.050149478454111e-06,
+      "loss": 0.8724,
+      "step": 131220
+    },
+    {
+      "epoch": 0.9499301468725343,
+      "grad_norm": 0.15207916498184204,
+      "learning_rate": 4.050077091793525e-06,
+      "loss": 0.8688,
+      "step": 131230
+    },
+    {
+      "epoch": 0.9500025335331205,
+      "grad_norm": 0.1509644240140915,
+      "learning_rate": 4.050004705132938e-06,
+      "loss": 0.8553,
+      "step": 131240
+    },
+    {
+      "epoch": 0.9500749201937067,
+      "grad_norm": 0.14878825843334198,
+      "learning_rate": 4.049932318472352e-06,
+      "loss": 0.8579,
+      "step": 131250
+    },
+    {
+      "epoch": 0.9501473068542929,
+      "grad_norm": 0.1610831767320633,
+      "learning_rate": 4.049859931811766e-06,
+      "loss": 0.8732,
+      "step": 131260
+    },
+    {
+      "epoch": 0.9502196935148791,
+      "grad_norm": 0.1690327674150467,
+      "learning_rate": 4.04978754515118e-06,
+      "loss": 0.8738,
+      "step": 131270
+    },
+    {
+      "epoch": 0.9502920801754653,
+      "grad_norm": 0.16976772248744965,
+      "learning_rate": 4.0497151584905935e-06,
+      "loss": 0.8626,
+      "step": 131280
+    },
+    {
+      "epoch": 0.9503644668360515,
+      "grad_norm": 0.1506384015083313,
+      "learning_rate": 4.049642771830007e-06,
+      "loss": 0.8642,
+      "step": 131290
+    },
+    {
+      "epoch": 0.9504368534966376,
+      "grad_norm": 0.1462264209985733,
+      "learning_rate": 4.049570385169422e-06,
+      "loss": 0.8646,
+      "step": 131300
+    },
+    {
+      "epoch": 0.9505092401572238,
+      "grad_norm": 0.14484186470508575,
+      "learning_rate": 4.049497998508835e-06,
+      "loss": 0.8771,
+      "step": 131310
+    },
+    {
+      "epoch": 0.95058162681781,
+      "grad_norm": 0.15840958058834076,
+      "learning_rate": 4.049425611848249e-06,
+      "loss": 0.8607,
+      "step": 131320
+    },
+    {
+      "epoch": 0.9506540134783962,
+      "grad_norm": 0.15120358765125275,
+      "learning_rate": 4.0493532251876625e-06,
+      "loss": 0.8766,
+      "step": 131330
+    },
+    {
+      "epoch": 0.9507264001389824,
+      "grad_norm": 0.15043680369853973,
+      "learning_rate": 4.049280838527077e-06,
+      "loss": 0.871,
+      "step": 131340
+    },
+    {
+      "epoch": 0.9507987867995685,
+      "grad_norm": 0.15595075488090515,
+      "learning_rate": 4.0492084518664905e-06,
+      "loss": 0.8668,
+      "step": 131350
+    },
+    {
+      "epoch": 0.9508711734601548,
+      "grad_norm": 0.15385489165782928,
+      "learning_rate": 4.049136065205904e-06,
+      "loss": 0.8748,
+      "step": 131360
+    },
+    {
+      "epoch": 0.950943560120741,
+      "grad_norm": 0.19055695831775665,
+      "learning_rate": 4.049063678545318e-06,
+      "loss": 0.8791,
+      "step": 131370
+    },
+    {
+      "epoch": 0.9510159467813272,
+      "grad_norm": 0.1542825996875763,
+      "learning_rate": 4.048991291884732e-06,
+      "loss": 0.8667,
+      "step": 131380
+    },
+    {
+      "epoch": 0.9510883334419133,
+      "grad_norm": 0.16472730040550232,
+      "learning_rate": 4.048918905224146e-06,
+      "loss": 0.8755,
+      "step": 131390
+    },
+    {
+      "epoch": 0.9511607201024995,
+      "grad_norm": 0.17082463204860687,
+      "learning_rate": 4.0488465185635595e-06,
+      "loss": 0.8434,
+      "step": 131400
+    },
+    {
+      "epoch": 0.9512331067630857,
+      "grad_norm": 0.1629323959350586,
+      "learning_rate": 4.048774131902973e-06,
+      "loss": 0.8712,
+      "step": 131410
+    },
+    {
+      "epoch": 0.9513054934236719,
+      "grad_norm": 0.1467832624912262,
+      "learning_rate": 4.0487017452423875e-06,
+      "loss": 0.8609,
+      "step": 131420
+    },
+    {
+      "epoch": 0.951377880084258,
+      "grad_norm": 0.15191030502319336,
+      "learning_rate": 4.048629358581801e-06,
+      "loss": 0.8718,
+      "step": 131430
+    },
+    {
+      "epoch": 0.9514502667448442,
+      "grad_norm": 0.17958003282546997,
+      "learning_rate": 4.048556971921215e-06,
+      "loss": 0.8696,
+      "step": 131440
+    },
+    {
+      "epoch": 0.9515226534054304,
+      "grad_norm": 0.16682404279708862,
+      "learning_rate": 4.048484585260628e-06,
+      "loss": 0.8676,
+      "step": 131450
+    },
+    {
+      "epoch": 0.9515950400660166,
+      "grad_norm": 0.1475543975830078,
+      "learning_rate": 4.048412198600042e-06,
+      "loss": 0.87,
+      "step": 131460
+    },
+    {
+      "epoch": 0.9516674267266029,
+      "grad_norm": 0.15572024881839752,
+      "learning_rate": 4.0483398119394565e-06,
+      "loss": 0.8793,
+      "step": 131470
+    },
+    {
+      "epoch": 0.951739813387189,
+      "grad_norm": 0.1579056978225708,
+      "learning_rate": 4.04826742527887e-06,
+      "loss": 0.8578,
+      "step": 131480
+    },
+    {
+      "epoch": 0.9518122000477752,
+      "grad_norm": 0.1470668464899063,
+      "learning_rate": 4.048195038618284e-06,
+      "loss": 0.8634,
+      "step": 131490
+    },
+    {
+      "epoch": 0.9518845867083614,
+      "grad_norm": 0.1533782035112381,
+      "learning_rate": 4.048122651957697e-06,
+      "loss": 0.852,
+      "step": 131500
+    },
+    {
+      "epoch": 0.9519569733689476,
+      "grad_norm": 0.16424772143363953,
+      "learning_rate": 4.048050265297112e-06,
+      "loss": 0.8566,
+      "step": 131510
+    },
+    {
+      "epoch": 0.9520293600295338,
+      "grad_norm": 0.17163357138633728,
+      "learning_rate": 4.047977878636525e-06,
+      "loss": 0.8657,
+      "step": 131520
+    },
+    {
+      "epoch": 0.9521017466901199,
+      "grad_norm": 0.15258784592151642,
+      "learning_rate": 4.047905491975939e-06,
+      "loss": 0.8786,
+      "step": 131530
+    },
+    {
+      "epoch": 0.9521741333507061,
+      "grad_norm": 0.15149055421352386,
+      "learning_rate": 4.047833105315353e-06,
+      "loss": 0.8627,
+      "step": 131540
+    },
+    {
+      "epoch": 0.9522465200112923,
+      "grad_norm": 0.15331679582595825,
+      "learning_rate": 4.047760718654767e-06,
+      "loss": 0.8613,
+      "step": 131550
+    },
+    {
+      "epoch": 0.9523189066718785,
+      "grad_norm": 0.15409405529499054,
+      "learning_rate": 4.047688331994181e-06,
+      "loss": 0.8754,
+      "step": 131560
+    },
+    {
+      "epoch": 0.9523912933324646,
+      "grad_norm": 0.16300641000270844,
+      "learning_rate": 4.047615945333594e-06,
+      "loss": 0.8851,
+      "step": 131570
+    },
+    {
+      "epoch": 0.9524636799930509,
+      "grad_norm": 0.1475706696510315,
+      "learning_rate": 4.047543558673008e-06,
+      "loss": 0.8686,
+      "step": 131580
+    },
+    {
+      "epoch": 0.9525360666536371,
+      "grad_norm": 0.14005255699157715,
+      "learning_rate": 4.047471172012422e-06,
+      "loss": 0.8771,
+      "step": 131590
+    },
+    {
+      "epoch": 0.9526084533142233,
+      "grad_norm": 0.152145653963089,
+      "learning_rate": 4.047398785351836e-06,
+      "loss": 0.8652,
+      "step": 131600
+    },
+    {
+      "epoch": 0.9526808399748095,
+      "grad_norm": 0.15361985564231873,
+      "learning_rate": 4.04732639869125e-06,
+      "loss": 0.8609,
+      "step": 131610
+    },
+    {
+      "epoch": 0.9527532266353956,
+      "grad_norm": 0.15738271176815033,
+      "learning_rate": 4.047254012030663e-06,
+      "loss": 0.8731,
+      "step": 131620
+    },
+    {
+      "epoch": 0.9528256132959818,
+      "grad_norm": 0.15454953908920288,
+      "learning_rate": 4.047181625370077e-06,
+      "loss": 0.8549,
+      "step": 131630
+    },
+    {
+      "epoch": 0.952897999956568,
+      "grad_norm": 0.17715829610824585,
+      "learning_rate": 4.0471092387094905e-06,
+      "loss": 0.864,
+      "step": 131640
+    },
+    {
+      "epoch": 0.9529703866171542,
+      "grad_norm": 0.16041779518127441,
+      "learning_rate": 4.047036852048904e-06,
+      "loss": 0.8823,
+      "step": 131650
+    },
+    {
+      "epoch": 0.9530427732777403,
+      "grad_norm": 0.1558634638786316,
+      "learning_rate": 4.0469644653883186e-06,
+      "loss": 0.8816,
+      "step": 131660
+    },
+    {
+      "epoch": 0.9531151599383265,
+      "grad_norm": 0.16207058727741241,
+      "learning_rate": 4.046892078727732e-06,
+      "loss": 0.8568,
+      "step": 131670
+    },
+    {
+      "epoch": 0.9531875465989128,
+      "grad_norm": 0.16530075669288635,
+      "learning_rate": 4.046819692067146e-06,
+      "loss": 0.8695,
+      "step": 131680
+    },
+    {
+      "epoch": 0.953259933259499,
+      "grad_norm": 0.16175314784049988,
+      "learning_rate": 4.046747305406559e-06,
+      "loss": 0.8662,
+      "step": 131690
+    },
+    {
+      "epoch": 0.9533323199200852,
+      "grad_norm": 0.1511378288269043,
+      "learning_rate": 4.046674918745974e-06,
+      "loss": 0.8655,
+      "step": 131700
+    },
+    {
+      "epoch": 0.9534047065806713,
+      "grad_norm": 0.15488837659358978,
+      "learning_rate": 4.0466025320853875e-06,
+      "loss": 0.8634,
+      "step": 131710
+    },
+    {
+      "epoch": 0.9534770932412575,
+      "grad_norm": 0.15649256110191345,
+      "learning_rate": 4.046530145424801e-06,
+      "loss": 0.8821,
+      "step": 131720
+    },
+    {
+      "epoch": 0.9535494799018437,
+      "grad_norm": 0.14642643928527832,
+      "learning_rate": 4.046457758764215e-06,
+      "loss": 0.8704,
+      "step": 131730
+    },
+    {
+      "epoch": 0.9536218665624299,
+      "grad_norm": 0.15652616322040558,
+      "learning_rate": 4.046385372103629e-06,
+      "loss": 0.8604,
+      "step": 131740
+    },
+    {
+      "epoch": 0.953694253223016,
+      "grad_norm": 0.15622968971729279,
+      "learning_rate": 4.046312985443043e-06,
+      "loss": 0.8704,
+      "step": 131750
+    },
+    {
+      "epoch": 0.9537666398836022,
+      "grad_norm": 0.1728099137544632,
+      "learning_rate": 4.0462405987824564e-06,
+      "loss": 0.8708,
+      "step": 131760
+    },
+    {
+      "epoch": 0.9538390265441884,
+      "grad_norm": 0.15809345245361328,
+      "learning_rate": 4.04616821212187e-06,
+      "loss": 0.8614,
+      "step": 131770
+    },
+    {
+      "epoch": 0.9539114132047746,
+      "grad_norm": 0.1720665991306305,
+      "learning_rate": 4.0460958254612845e-06,
+      "loss": 0.861,
+      "step": 131780
+    },
+    {
+      "epoch": 0.9539837998653609,
+      "grad_norm": 0.15099675953388214,
+      "learning_rate": 4.046023438800698e-06,
+      "loss": 0.8784,
+      "step": 131790
+    },
+    {
+      "epoch": 0.954056186525947,
+      "grad_norm": 0.14246854186058044,
+      "learning_rate": 4.045951052140112e-06,
+      "loss": 0.8659,
+      "step": 131800
+    },
+    {
+      "epoch": 0.9541285731865332,
+      "grad_norm": 0.15169619023799896,
+      "learning_rate": 4.045878665479525e-06,
+      "loss": 0.8672,
+      "step": 131810
+    },
+    {
+      "epoch": 0.9542009598471194,
+      "grad_norm": 0.15939058363437653,
+      "learning_rate": 4.04580627881894e-06,
+      "loss": 0.8608,
+      "step": 131820
+    },
+    {
+      "epoch": 0.9542733465077056,
+      "grad_norm": 0.16501100361347198,
+      "learning_rate": 4.0457338921583534e-06,
+      "loss": 0.857,
+      "step": 131830
+    },
+    {
+      "epoch": 0.9543457331682917,
+      "grad_norm": 1.6813664436340332,
+      "learning_rate": 4.045661505497767e-06,
+      "loss": 0.8666,
+      "step": 131840
+    },
+    {
+      "epoch": 0.9544181198288779,
+      "grad_norm": 0.14207109808921814,
+      "learning_rate": 4.045589118837181e-06,
+      "loss": 0.8631,
+      "step": 131850
+    },
+    {
+      "epoch": 0.9544905064894641,
+      "grad_norm": 0.14771947264671326,
+      "learning_rate": 4.045516732176595e-06,
+      "loss": 0.8668,
+      "step": 131860
+    },
+    {
+      "epoch": 0.9545628931500503,
+      "grad_norm": 0.15695475041866302,
+      "learning_rate": 4.045444345516009e-06,
+      "loss": 0.8629,
+      "step": 131870
+    },
+    {
+      "epoch": 0.9546352798106365,
+      "grad_norm": 0.15353751182556152,
+      "learning_rate": 4.045371958855422e-06,
+      "loss": 0.858,
+      "step": 131880
+    },
+    {
+      "epoch": 0.9547076664712227,
+      "grad_norm": 0.15183933079242706,
+      "learning_rate": 4.045299572194836e-06,
+      "loss": 0.8694,
+      "step": 131890
+    },
+    {
+      "epoch": 0.9547800531318089,
+      "grad_norm": 0.16019755601882935,
+      "learning_rate": 4.0452271855342504e-06,
+      "loss": 0.8774,
+      "step": 131900
+    },
+    {
+      "epoch": 0.9548524397923951,
+      "grad_norm": 0.15118776261806488,
+      "learning_rate": 4.045154798873664e-06,
+      "loss": 0.8733,
+      "step": 131910
+    },
+    {
+      "epoch": 0.9549248264529813,
+      "grad_norm": 0.15187504887580872,
+      "learning_rate": 4.045082412213078e-06,
+      "loss": 0.8695,
+      "step": 131920
+    },
+    {
+      "epoch": 0.9549972131135674,
+      "grad_norm": 0.15506109595298767,
+      "learning_rate": 4.045010025552491e-06,
+      "loss": 0.8637,
+      "step": 131930
+    },
+    {
+      "epoch": 0.9550695997741536,
+      "grad_norm": 0.14858941733837128,
+      "learning_rate": 4.044937638891906e-06,
+      "loss": 0.8727,
+      "step": 131940
+    },
+    {
+      "epoch": 0.9551419864347398,
+      "grad_norm": 0.15059681236743927,
+      "learning_rate": 4.044865252231319e-06,
+      "loss": 0.8503,
+      "step": 131950
+    },
+    {
+      "epoch": 0.955214373095326,
+      "grad_norm": 0.14260925352573395,
+      "learning_rate": 4.044792865570733e-06,
+      "loss": 0.8507,
+      "step": 131960
+    },
+    {
+      "epoch": 0.9552867597559122,
+      "grad_norm": 0.16660186648368835,
+      "learning_rate": 4.044720478910147e-06,
+      "loss": 0.8809,
+      "step": 131970
+    },
+    {
+      "epoch": 0.9553591464164983,
+      "grad_norm": 0.1452503353357315,
+      "learning_rate": 4.044648092249561e-06,
+      "loss": 0.8595,
+      "step": 131980
+    },
+    {
+      "epoch": 0.9554315330770845,
+      "grad_norm": 0.14808453619480133,
+      "learning_rate": 4.044575705588975e-06,
+      "loss": 0.8595,
+      "step": 131990
+    },
+    {
+      "epoch": 0.9555039197376708,
+      "grad_norm": 0.14562194049358368,
+      "learning_rate": 4.044503318928388e-06,
+      "loss": 0.8679,
+      "step": 132000
+    },
+    {
+      "epoch": 0.955576306398257,
+      "grad_norm": 0.1467532068490982,
+      "learning_rate": 4.044430932267802e-06,
+      "loss": 0.8722,
+      "step": 132010
+    },
+    {
+      "epoch": 0.9556486930588431,
+      "grad_norm": 0.15089406073093414,
+      "learning_rate": 4.044358545607216e-06,
+      "loss": 0.8647,
+      "step": 132020
+    },
+    {
+      "epoch": 0.9557210797194293,
+      "grad_norm": 0.1548631340265274,
+      "learning_rate": 4.04428615894663e-06,
+      "loss": 0.8675,
+      "step": 132030
+    },
+    {
+      "epoch": 0.9557934663800155,
+      "grad_norm": 0.15700556337833405,
+      "learning_rate": 4.044213772286044e-06,
+      "loss": 0.8643,
+      "step": 132040
+    },
+    {
+      "epoch": 0.9558658530406017,
+      "grad_norm": 0.1600215882062912,
+      "learning_rate": 4.044141385625457e-06,
+      "loss": 0.8749,
+      "step": 132050
+    },
+    {
+      "epoch": 0.9559382397011879,
+      "grad_norm": 0.1461002379655838,
+      "learning_rate": 4.044068998964871e-06,
+      "loss": 0.8759,
+      "step": 132060
+    },
+    {
+      "epoch": 0.956010626361774,
+      "grad_norm": 0.1531788557767868,
+      "learning_rate": 4.043996612304285e-06,
+      "loss": 0.8736,
+      "step": 132070
+    },
+    {
+      "epoch": 0.9560830130223602,
+      "grad_norm": 0.15142254531383514,
+      "learning_rate": 4.043924225643699e-06,
+      "loss": 0.857,
+      "step": 132080
+    },
+    {
+      "epoch": 0.9561553996829464,
+      "grad_norm": 0.1594085544347763,
+      "learning_rate": 4.0438518389831125e-06,
+      "loss": 0.87,
+      "step": 132090
+    },
+    {
+      "epoch": 0.9562277863435326,
+      "grad_norm": 0.16034048795700073,
+      "learning_rate": 4.043779452322526e-06,
+      "loss": 0.8755,
+      "step": 132100
+    },
+    {
+      "epoch": 0.9563001730041188,
+      "grad_norm": 0.1539410948753357,
+      "learning_rate": 4.043707065661941e-06,
+      "loss": 0.874,
+      "step": 132110
+    },
+    {
+      "epoch": 0.956372559664705,
+      "grad_norm": 0.15013930201530457,
+      "learning_rate": 4.043634679001354e-06,
+      "loss": 0.8666,
+      "step": 132120
+    },
+    {
+      "epoch": 0.9564449463252912,
+      "grad_norm": 0.15976254642009735,
+      "learning_rate": 4.043562292340768e-06,
+      "loss": 0.8697,
+      "step": 132130
+    },
+    {
+      "epoch": 0.9565173329858774,
+      "grad_norm": 0.16351239383220673,
+      "learning_rate": 4.0434899056801815e-06,
+      "loss": 0.8731,
+      "step": 132140
+    },
+    {
+      "epoch": 0.9565897196464636,
+      "grad_norm": 0.15584561228752136,
+      "learning_rate": 4.043417519019596e-06,
+      "loss": 0.8833,
+      "step": 132150
+    },
+    {
+      "epoch": 0.9566621063070497,
+      "grad_norm": 0.1562098264694214,
+      "learning_rate": 4.043345132359009e-06,
+      "loss": 0.872,
+      "step": 132160
+    },
+    {
+      "epoch": 0.9567344929676359,
+      "grad_norm": 0.156274676322937,
+      "learning_rate": 4.043272745698423e-06,
+      "loss": 0.8708,
+      "step": 132170
+    },
+    {
+      "epoch": 0.9568068796282221,
+      "grad_norm": 0.22564736008644104,
+      "learning_rate": 4.043200359037837e-06,
+      "loss": 0.8636,
+      "step": 132180
+    },
+    {
+      "epoch": 0.9568792662888083,
+      "grad_norm": 0.1602523922920227,
+      "learning_rate": 4.04312797237725e-06,
+      "loss": 0.8775,
+      "step": 132190
+    },
+    {
+      "epoch": 0.9569516529493944,
+      "grad_norm": 0.15143248438835144,
+      "learning_rate": 4.043055585716664e-06,
+      "loss": 0.8635,
+      "step": 132200
+    },
+    {
+      "epoch": 0.9570240396099807,
+      "grad_norm": 0.14714032411575317,
+      "learning_rate": 4.0429831990560785e-06,
+      "loss": 0.8642,
+      "step": 132210
+    },
+    {
+      "epoch": 0.9570964262705669,
+      "grad_norm": 0.1613510251045227,
+      "learning_rate": 4.042910812395492e-06,
+      "loss": 0.8645,
+      "step": 132220
+    },
+    {
+      "epoch": 0.9571688129311531,
+      "grad_norm": 0.17830723524093628,
+      "learning_rate": 4.042838425734906e-06,
+      "loss": 0.8672,
+      "step": 132230
+    },
+    {
+      "epoch": 0.9572411995917393,
+      "grad_norm": 0.15344306826591492,
+      "learning_rate": 4.042766039074319e-06,
+      "loss": 0.8679,
+      "step": 132240
+    },
+    {
+      "epoch": 0.9573135862523254,
+      "grad_norm": 0.15479233860969543,
+      "learning_rate": 4.042693652413733e-06,
+      "loss": 0.8674,
+      "step": 132250
+    },
+    {
+      "epoch": 0.9573859729129116,
+      "grad_norm": 0.1529737114906311,
+      "learning_rate": 4.042621265753147e-06,
+      "loss": 0.8539,
+      "step": 132260
+    },
+    {
+      "epoch": 0.9574583595734978,
+      "grad_norm": 0.14475677907466888,
+      "learning_rate": 4.042548879092561e-06,
+      "loss": 0.8693,
+      "step": 132270
+    },
+    {
+      "epoch": 0.957530746234084,
+      "grad_norm": 0.167455792427063,
+      "learning_rate": 4.042476492431975e-06,
+      "loss": 0.8576,
+      "step": 132280
+    },
+    {
+      "epoch": 0.9576031328946701,
+      "grad_norm": 0.14674553275108337,
+      "learning_rate": 4.042404105771388e-06,
+      "loss": 0.8753,
+      "step": 132290
+    },
+    {
+      "epoch": 0.9576755195552563,
+      "grad_norm": 0.20534364879131317,
+      "learning_rate": 4.042331719110803e-06,
+      "loss": 0.8661,
+      "step": 132300
+    },
+    {
+      "epoch": 0.9577479062158425,
+      "grad_norm": 0.16672727465629578,
+      "learning_rate": 4.042259332450216e-06,
+      "loss": 0.8645,
+      "step": 132310
+    },
+    {
+      "epoch": 0.9578202928764288,
+      "grad_norm": 0.14698807895183563,
+      "learning_rate": 4.04218694578963e-06,
+      "loss": 0.8645,
+      "step": 132320
+    },
+    {
+      "epoch": 0.957892679537015,
+      "grad_norm": 0.15321475267410278,
+      "learning_rate": 4.0421145591290436e-06,
+      "loss": 0.8692,
+      "step": 132330
+    },
+    {
+      "epoch": 0.9579650661976011,
+      "grad_norm": 0.15178924798965454,
+      "learning_rate": 4.042042172468458e-06,
+      "loss": 0.8705,
+      "step": 132340
+    },
+    {
+      "epoch": 0.9580374528581873,
+      "grad_norm": 0.701418936252594,
+      "learning_rate": 4.041969785807872e-06,
+      "loss": 0.8685,
+      "step": 132350
+    },
+    {
+      "epoch": 0.9581098395187735,
+      "grad_norm": 0.15373194217681885,
+      "learning_rate": 4.041897399147285e-06,
+      "loss": 0.867,
+      "step": 132360
+    },
+    {
+      "epoch": 0.9581822261793597,
+      "grad_norm": 0.15666644275188446,
+      "learning_rate": 4.041825012486699e-06,
+      "loss": 0.8715,
+      "step": 132370
+    },
+    {
+      "epoch": 0.9582546128399458,
+      "grad_norm": 0.14554648101329803,
+      "learning_rate": 4.041752625826113e-06,
+      "loss": 0.8612,
+      "step": 132380
+    },
+    {
+      "epoch": 0.958326999500532,
+      "grad_norm": 0.15833979845046997,
+      "learning_rate": 4.041680239165527e-06,
+      "loss": 0.8694,
+      "step": 132390
+    },
+    {
+      "epoch": 0.9583993861611182,
+      "grad_norm": 0.1463756412267685,
+      "learning_rate": 4.0416078525049406e-06,
+      "loss": 0.8691,
+      "step": 132400
+    },
+    {
+      "epoch": 0.9584717728217044,
+      "grad_norm": 0.14751847088336945,
+      "learning_rate": 4.041535465844354e-06,
+      "loss": 0.8666,
+      "step": 132410
+    },
+    {
+      "epoch": 0.9585441594822907,
+      "grad_norm": 0.16007982194423676,
+      "learning_rate": 4.041463079183769e-06,
+      "loss": 0.8766,
+      "step": 132420
+    },
+    {
+      "epoch": 0.9586165461428768,
+      "grad_norm": 0.1559884399175644,
+      "learning_rate": 4.041390692523182e-06,
+      "loss": 0.857,
+      "step": 132430
+    },
+    {
+      "epoch": 0.958688932803463,
+      "grad_norm": 0.16903673112392426,
+      "learning_rate": 4.041318305862596e-06,
+      "loss": 0.8728,
+      "step": 132440
+    },
+    {
+      "epoch": 0.9587613194640492,
+      "grad_norm": 0.15044556558132172,
+      "learning_rate": 4.0412459192020095e-06,
+      "loss": 0.8705,
+      "step": 132450
+    },
+    {
+      "epoch": 0.9588337061246354,
+      "grad_norm": 0.1696968823671341,
+      "learning_rate": 4.041173532541424e-06,
+      "loss": 0.8593,
+      "step": 132460
+    },
+    {
+      "epoch": 0.9589060927852215,
+      "grad_norm": 0.1612185686826706,
+      "learning_rate": 4.0411011458808376e-06,
+      "loss": 0.8674,
+      "step": 132470
+    },
+    {
+      "epoch": 0.9589784794458077,
+      "grad_norm": 0.1500626504421234,
+      "learning_rate": 4.041028759220251e-06,
+      "loss": 0.873,
+      "step": 132480
+    },
+    {
+      "epoch": 0.9590508661063939,
+      "grad_norm": 0.145673006772995,
+      "learning_rate": 4.040956372559665e-06,
+      "loss": 0.8659,
+      "step": 132490
+    },
+    {
+      "epoch": 0.9591232527669801,
+      "grad_norm": 0.15620875358581543,
+      "learning_rate": 4.040883985899079e-06,
+      "loss": 0.8645,
+      "step": 132500
+    },
+    {
+      "epoch": 0.9591956394275662,
+      "grad_norm": 0.16961318254470825,
+      "learning_rate": 4.040811599238493e-06,
+      "loss": 0.8722,
+      "step": 132510
+    },
+    {
+      "epoch": 0.9592680260881524,
+      "grad_norm": 0.14998477697372437,
+      "learning_rate": 4.0407392125779065e-06,
+      "loss": 0.8499,
+      "step": 132520
+    },
+    {
+      "epoch": 0.9593404127487387,
+      "grad_norm": 0.1547980010509491,
+      "learning_rate": 4.04066682591732e-06,
+      "loss": 0.8705,
+      "step": 132530
+    },
+    {
+      "epoch": 0.9594127994093249,
+      "grad_norm": 0.14744049310684204,
+      "learning_rate": 4.040594439256735e-06,
+      "loss": 0.8626,
+      "step": 132540
+    },
+    {
+      "epoch": 0.9594851860699111,
+      "grad_norm": 0.1580532044172287,
+      "learning_rate": 4.040522052596148e-06,
+      "loss": 0.8651,
+      "step": 132550
+    },
+    {
+      "epoch": 0.9595575727304972,
+      "grad_norm": 0.14146533608436584,
+      "learning_rate": 4.040449665935562e-06,
+      "loss": 0.8675,
+      "step": 132560
+    },
+    {
+      "epoch": 0.9596299593910834,
+      "grad_norm": 0.15694855153560638,
+      "learning_rate": 4.0403772792749754e-06,
+      "loss": 0.8776,
+      "step": 132570
+    },
+    {
+      "epoch": 0.9597023460516696,
+      "grad_norm": 0.15668827295303345,
+      "learning_rate": 4.04030489261439e-06,
+      "loss": 0.8643,
+      "step": 132580
+    },
+    {
+      "epoch": 0.9597747327122558,
+      "grad_norm": 0.15416784584522247,
+      "learning_rate": 4.0402325059538035e-06,
+      "loss": 0.8595,
+      "step": 132590
+    },
+    {
+      "epoch": 0.959847119372842,
+      "grad_norm": 0.15903611481189728,
+      "learning_rate": 4.040160119293217e-06,
+      "loss": 0.8715,
+      "step": 132600
+    },
+    {
+      "epoch": 0.9599195060334281,
+      "grad_norm": 0.14831218123435974,
+      "learning_rate": 4.040087732632631e-06,
+      "loss": 0.8719,
+      "step": 132610
+    },
+    {
+      "epoch": 0.9599918926940143,
+      "grad_norm": 0.16342884302139282,
+      "learning_rate": 4.040015345972045e-06,
+      "loss": 0.8718,
+      "step": 132620
+    },
+    {
+      "epoch": 0.9600642793546005,
+      "grad_norm": 0.16396649181842804,
+      "learning_rate": 4.039942959311459e-06,
+      "loss": 0.8556,
+      "step": 132630
+    },
+    {
+      "epoch": 0.9601366660151868,
+      "grad_norm": 0.16489477455615997,
+      "learning_rate": 4.0398705726508724e-06,
+      "loss": 0.8738,
+      "step": 132640
+    },
+    {
+      "epoch": 0.960209052675773,
+      "grad_norm": 0.14648328721523285,
+      "learning_rate": 4.039798185990286e-06,
+      "loss": 0.882,
+      "step": 132650
+    },
+    {
+      "epoch": 0.9602814393363591,
+      "grad_norm": 0.15724149346351624,
+      "learning_rate": 4.0397257993297005e-06,
+      "loss": 0.8638,
+      "step": 132660
+    },
+    {
+      "epoch": 0.9603538259969453,
+      "grad_norm": 0.15346843004226685,
+      "learning_rate": 4.039653412669114e-06,
+      "loss": 0.8824,
+      "step": 132670
+    },
+    {
+      "epoch": 0.9604262126575315,
+      "grad_norm": 0.15470701456069946,
+      "learning_rate": 4.039581026008528e-06,
+      "loss": 0.8663,
+      "step": 132680
+    },
+    {
+      "epoch": 0.9604985993181177,
+      "grad_norm": 0.14876993000507355,
+      "learning_rate": 4.039508639347941e-06,
+      "loss": 0.8618,
+      "step": 132690
+    },
+    {
+      "epoch": 0.9605709859787038,
+      "grad_norm": 0.14748069643974304,
+      "learning_rate": 4.039436252687355e-06,
+      "loss": 0.8742,
+      "step": 132700
+    },
+    {
+      "epoch": 0.96064337263929,
+      "grad_norm": 0.14472724497318268,
+      "learning_rate": 4.039363866026769e-06,
+      "loss": 0.8591,
+      "step": 132710
+    },
+    {
+      "epoch": 0.9607157592998762,
+      "grad_norm": 0.16464345157146454,
+      "learning_rate": 4.039291479366182e-06,
+      "loss": 0.868,
+      "step": 132720
+    },
+    {
+      "epoch": 0.9607881459604624,
+      "grad_norm": 0.15630069375038147,
+      "learning_rate": 4.039219092705597e-06,
+      "loss": 0.8675,
+      "step": 132730
+    },
+    {
+      "epoch": 0.9608605326210486,
+      "grad_norm": 0.1542377769947052,
+      "learning_rate": 4.03914670604501e-06,
+      "loss": 0.8643,
+      "step": 132740
+    },
+    {
+      "epoch": 0.9609329192816348,
+      "grad_norm": 0.15142075717449188,
+      "learning_rate": 4.039074319384424e-06,
+      "loss": 0.8583,
+      "step": 132750
+    },
+    {
+      "epoch": 0.961005305942221,
+      "grad_norm": 0.14970579743385315,
+      "learning_rate": 4.0390019327238375e-06,
+      "loss": 0.8726,
+      "step": 132760
+    },
+    {
+      "epoch": 0.9610776926028072,
+      "grad_norm": 0.1593393087387085,
+      "learning_rate": 4.038929546063252e-06,
+      "loss": 0.8746,
+      "step": 132770
+    },
+    {
+      "epoch": 0.9611500792633934,
+      "grad_norm": 0.154205784201622,
+      "learning_rate": 4.038857159402666e-06,
+      "loss": 0.8675,
+      "step": 132780
+    },
+    {
+      "epoch": 0.9612224659239795,
+      "grad_norm": 0.15986141562461853,
+      "learning_rate": 4.038784772742079e-06,
+      "loss": 0.8697,
+      "step": 132790
+    },
+    {
+      "epoch": 0.9612948525845657,
+      "grad_norm": 0.1516096591949463,
+      "learning_rate": 4.038712386081493e-06,
+      "loss": 0.8603,
+      "step": 132800
+    },
+    {
+      "epoch": 0.9613672392451519,
+      "grad_norm": 0.14772486686706543,
+      "learning_rate": 4.038639999420907e-06,
+      "loss": 0.8557,
+      "step": 132810
+    },
+    {
+      "epoch": 0.9614396259057381,
+      "grad_norm": 0.152225062251091,
+      "learning_rate": 4.038567612760321e-06,
+      "loss": 0.8603,
+      "step": 132820
+    },
+    {
+      "epoch": 0.9615120125663242,
+      "grad_norm": 0.1596631109714508,
+      "learning_rate": 4.0384952260997345e-06,
+      "loss": 0.8656,
+      "step": 132830
+    },
+    {
+      "epoch": 0.9615843992269104,
+      "grad_norm": 0.15353646874427795,
+      "learning_rate": 4.038422839439148e-06,
+      "loss": 0.8606,
+      "step": 132840
+    },
+    {
+      "epoch": 0.9616567858874967,
+      "grad_norm": 0.15093198418617249,
+      "learning_rate": 4.038350452778563e-06,
+      "loss": 0.8651,
+      "step": 132850
+    },
+    {
+      "epoch": 0.9617291725480829,
+      "grad_norm": 0.15121009945869446,
+      "learning_rate": 4.038278066117976e-06,
+      "loss": 0.8747,
+      "step": 132860
+    },
+    {
+      "epoch": 0.961801559208669,
+      "grad_norm": 0.142518550157547,
+      "learning_rate": 4.03820567945739e-06,
+      "loss": 0.8627,
+      "step": 132870
+    },
+    {
+      "epoch": 0.9618739458692552,
+      "grad_norm": 0.14168381690979004,
+      "learning_rate": 4.0381332927968035e-06,
+      "loss": 0.872,
+      "step": 132880
+    },
+    {
+      "epoch": 0.9619463325298414,
+      "grad_norm": 0.16068200767040253,
+      "learning_rate": 4.038060906136217e-06,
+      "loss": 0.8666,
+      "step": 132890
+    },
+    {
+      "epoch": 0.9620187191904276,
+      "grad_norm": 0.15603065490722656,
+      "learning_rate": 4.0379885194756315e-06,
+      "loss": 0.8768,
+      "step": 132900
+    },
+    {
+      "epoch": 0.9620911058510138,
+      "grad_norm": 0.16330550611019135,
+      "learning_rate": 4.037916132815045e-06,
+      "loss": 0.8713,
+      "step": 132910
+    },
+    {
+      "epoch": 0.9621634925115999,
+      "grad_norm": 0.15388086438179016,
+      "learning_rate": 4.037843746154459e-06,
+      "loss": 0.8824,
+      "step": 132920
+    },
+    {
+      "epoch": 0.9622358791721861,
+      "grad_norm": 0.1486055701971054,
+      "learning_rate": 4.037771359493872e-06,
+      "loss": 0.8652,
+      "step": 132930
+    },
+    {
+      "epoch": 0.9623082658327723,
+      "grad_norm": 0.16068662703037262,
+      "learning_rate": 4.037698972833287e-06,
+      "loss": 0.8553,
+      "step": 132940
+    },
+    {
+      "epoch": 0.9623806524933586,
+      "grad_norm": 0.14844128489494324,
+      "learning_rate": 4.0376265861727005e-06,
+      "loss": 0.8692,
+      "step": 132950
+    },
+    {
+      "epoch": 0.9624530391539448,
+      "grad_norm": 0.1547292321920395,
+      "learning_rate": 4.037554199512114e-06,
+      "loss": 0.8598,
+      "step": 132960
+    },
+    {
+      "epoch": 0.9625254258145309,
+      "grad_norm": 0.14296914637088776,
+      "learning_rate": 4.037481812851528e-06,
+      "loss": 0.8745,
+      "step": 132970
+    },
+    {
+      "epoch": 0.9625978124751171,
+      "grad_norm": 0.1457308977842331,
+      "learning_rate": 4.037409426190942e-06,
+      "loss": 0.8685,
+      "step": 132980
+    },
+    {
+      "epoch": 0.9626701991357033,
+      "grad_norm": 0.16108804941177368,
+      "learning_rate": 4.037337039530356e-06,
+      "loss": 0.8675,
+      "step": 132990
+    },
+    {
+      "epoch": 0.9627425857962895,
+      "grad_norm": 0.1842719167470932,
+      "learning_rate": 4.037264652869769e-06,
+      "loss": 0.8723,
+      "step": 133000
+    },
+    {
+      "epoch": 0.9628149724568756,
+      "grad_norm": 0.1494869738817215,
+      "learning_rate": 4.037192266209183e-06,
+      "loss": 0.8629,
+      "step": 133010
+    },
+    {
+      "epoch": 0.9628873591174618,
+      "grad_norm": 0.1545685976743698,
+      "learning_rate": 4.0371198795485975e-06,
+      "loss": 0.8732,
+      "step": 133020
+    },
+    {
+      "epoch": 0.962959745778048,
+      "grad_norm": 0.1837606430053711,
+      "learning_rate": 4.037047492888011e-06,
+      "loss": 0.8715,
+      "step": 133030
+    },
+    {
+      "epoch": 0.9630321324386342,
+      "grad_norm": 0.15043866634368896,
+      "learning_rate": 4.036975106227425e-06,
+      "loss": 0.8642,
+      "step": 133040
+    },
+    {
+      "epoch": 0.9631045190992203,
+      "grad_norm": 0.1640947014093399,
+      "learning_rate": 4.036902719566838e-06,
+      "loss": 0.8633,
+      "step": 133050
+    },
+    {
+      "epoch": 0.9631769057598066,
+      "grad_norm": 0.16168645024299622,
+      "learning_rate": 4.036830332906253e-06,
+      "loss": 0.8725,
+      "step": 133060
+    },
+    {
+      "epoch": 0.9632492924203928,
+      "grad_norm": 0.16159774363040924,
+      "learning_rate": 4.036757946245666e-06,
+      "loss": 0.8642,
+      "step": 133070
+    },
+    {
+      "epoch": 0.963321679080979,
+      "grad_norm": 0.19764064252376556,
+      "learning_rate": 4.03668555958508e-06,
+      "loss": 0.8708,
+      "step": 133080
+    },
+    {
+      "epoch": 0.9633940657415652,
+      "grad_norm": 0.16055376827716827,
+      "learning_rate": 4.036613172924494e-06,
+      "loss": 0.8577,
+      "step": 133090
+    },
+    {
+      "epoch": 0.9634664524021513,
+      "grad_norm": 0.14640270173549652,
+      "learning_rate": 4.036540786263908e-06,
+      "loss": 0.8696,
+      "step": 133100
+    },
+    {
+      "epoch": 0.9635388390627375,
+      "grad_norm": 0.18850268423557281,
+      "learning_rate": 4.036468399603322e-06,
+      "loss": 0.8629,
+      "step": 133110
+    },
+    {
+      "epoch": 0.9636112257233237,
+      "grad_norm": 0.16357038915157318,
+      "learning_rate": 4.036396012942735e-06,
+      "loss": 0.8767,
+      "step": 133120
+    },
+    {
+      "epoch": 0.9636836123839099,
+      "grad_norm": 0.17244000732898712,
+      "learning_rate": 4.036323626282149e-06,
+      "loss": 0.8624,
+      "step": 133130
+    },
+    {
+      "epoch": 0.963755999044496,
+      "grad_norm": 0.15212216973304749,
+      "learning_rate": 4.036251239621563e-06,
+      "loss": 0.8561,
+      "step": 133140
+    },
+    {
+      "epoch": 0.9638283857050822,
+      "grad_norm": 0.1720954179763794,
+      "learning_rate": 4.036178852960977e-06,
+      "loss": 0.8683,
+      "step": 133150
+    },
+    {
+      "epoch": 0.9639007723656684,
+      "grad_norm": 0.21422089636325836,
+      "learning_rate": 4.036106466300391e-06,
+      "loss": 0.8845,
+      "step": 133160
+    },
+    {
+      "epoch": 0.9639731590262547,
+      "grad_norm": 0.15769261121749878,
+      "learning_rate": 4.036034079639804e-06,
+      "loss": 0.8543,
+      "step": 133170
+    },
+    {
+      "epoch": 0.9640455456868409,
+      "grad_norm": 0.15512368083000183,
+      "learning_rate": 4.035961692979219e-06,
+      "loss": 0.8571,
+      "step": 133180
+    },
+    {
+      "epoch": 0.964117932347427,
+      "grad_norm": 0.16007642447948456,
+      "learning_rate": 4.035889306318632e-06,
+      "loss": 0.8621,
+      "step": 133190
+    },
+    {
+      "epoch": 0.9641903190080132,
+      "grad_norm": 0.16909575462341309,
+      "learning_rate": 4.035816919658046e-06,
+      "loss": 0.8751,
+      "step": 133200
+    },
+    {
+      "epoch": 0.9642627056685994,
+      "grad_norm": 0.158931702375412,
+      "learning_rate": 4.0357445329974596e-06,
+      "loss": 0.8655,
+      "step": 133210
+    },
+    {
+      "epoch": 0.9643350923291856,
+      "grad_norm": 0.16062036156654358,
+      "learning_rate": 4.035672146336873e-06,
+      "loss": 0.8705,
+      "step": 133220
+    },
+    {
+      "epoch": 0.9644074789897717,
+      "grad_norm": 0.15879908204078674,
+      "learning_rate": 4.035599759676287e-06,
+      "loss": 0.8609,
+      "step": 133230
+    },
+    {
+      "epoch": 0.9644798656503579,
+      "grad_norm": 0.15936613082885742,
+      "learning_rate": 4.0355273730157e-06,
+      "loss": 0.8663,
+      "step": 133240
+    },
+    {
+      "epoch": 0.9645522523109441,
+      "grad_norm": 0.159668430685997,
+      "learning_rate": 4.035454986355115e-06,
+      "loss": 0.8778,
+      "step": 133250
+    },
+    {
+      "epoch": 0.9646246389715303,
+      "grad_norm": 0.15538303554058075,
+      "learning_rate": 4.0353825996945285e-06,
+      "loss": 0.8709,
+      "step": 133260
+    },
+    {
+      "epoch": 0.9646970256321166,
+      "grad_norm": 0.19383670389652252,
+      "learning_rate": 4.035310213033942e-06,
+      "loss": 0.8635,
+      "step": 133270
+    },
+    {
+      "epoch": 0.9647694122927027,
+      "grad_norm": 0.14985784888267517,
+      "learning_rate": 4.035237826373356e-06,
+      "loss": 0.8639,
+      "step": 133280
+    },
+    {
+      "epoch": 0.9648417989532889,
+      "grad_norm": 0.15540927648544312,
+      "learning_rate": 4.03516543971277e-06,
+      "loss": 0.8737,
+      "step": 133290
+    },
+    {
+      "epoch": 0.9649141856138751,
+      "grad_norm": 0.16484995186328888,
+      "learning_rate": 4.035093053052184e-06,
+      "loss": 0.8722,
+      "step": 133300
+    },
+    {
+      "epoch": 0.9649865722744613,
+      "grad_norm": 0.15824109315872192,
+      "learning_rate": 4.0350206663915974e-06,
+      "loss": 0.8558,
+      "step": 133310
+    },
+    {
+      "epoch": 0.9650589589350475,
+      "grad_norm": 0.16974005103111267,
+      "learning_rate": 4.034948279731011e-06,
+      "loss": 0.8753,
+      "step": 133320
+    },
+    {
+      "epoch": 0.9651313455956336,
+      "grad_norm": 0.1548565924167633,
+      "learning_rate": 4.0348758930704255e-06,
+      "loss": 0.872,
+      "step": 133330
+    },
+    {
+      "epoch": 0.9652037322562198,
+      "grad_norm": 0.16429255902767181,
+      "learning_rate": 4.034803506409839e-06,
+      "loss": 0.8556,
+      "step": 133340
+    },
+    {
+      "epoch": 0.965276118916806,
+      "grad_norm": 0.16265404224395752,
+      "learning_rate": 4.034731119749253e-06,
+      "loss": 0.8644,
+      "step": 133350
+    },
+    {
+      "epoch": 0.9653485055773922,
+      "grad_norm": 0.15244075655937195,
+      "learning_rate": 4.034658733088666e-06,
+      "loss": 0.8821,
+      "step": 133360
+    },
+    {
+      "epoch": 0.9654208922379783,
+      "grad_norm": 0.14360937476158142,
+      "learning_rate": 4.034586346428081e-06,
+      "loss": 0.8896,
+      "step": 133370
+    },
+    {
+      "epoch": 0.9654932788985646,
+      "grad_norm": 0.22526615858078003,
+      "learning_rate": 4.0345139597674944e-06,
+      "loss": 0.8681,
+      "step": 133380
+    },
+    {
+      "epoch": 0.9655656655591508,
+      "grad_norm": 0.14955149590969086,
+      "learning_rate": 4.034441573106908e-06,
+      "loss": 0.8433,
+      "step": 133390
+    },
+    {
+      "epoch": 0.965638052219737,
+      "grad_norm": 0.1530320793390274,
+      "learning_rate": 4.034369186446322e-06,
+      "loss": 0.8705,
+      "step": 133400
+    },
+    {
+      "epoch": 0.9657104388803232,
+      "grad_norm": 0.15504902601242065,
+      "learning_rate": 4.034296799785736e-06,
+      "loss": 0.8548,
+      "step": 133410
+    },
+    {
+      "epoch": 0.9657828255409093,
+      "grad_norm": 0.1458277404308319,
+      "learning_rate": 4.03422441312515e-06,
+      "loss": 0.8773,
+      "step": 133420
+    },
+    {
+      "epoch": 0.9658552122014955,
+      "grad_norm": 0.14981764554977417,
+      "learning_rate": 4.034152026464563e-06,
+      "loss": 0.8849,
+      "step": 133430
+    },
+    {
+      "epoch": 0.9659275988620817,
+      "grad_norm": 0.15874448418617249,
+      "learning_rate": 4.034079639803977e-06,
+      "loss": 0.8573,
+      "step": 133440
+    },
+    {
+      "epoch": 0.9659999855226679,
+      "grad_norm": 0.1638738363981247,
+      "learning_rate": 4.0340072531433914e-06,
+      "loss": 0.859,
+      "step": 133450
+    },
+    {
+      "epoch": 0.966072372183254,
+      "grad_norm": 0.15449586510658264,
+      "learning_rate": 4.033934866482805e-06,
+      "loss": 0.8728,
+      "step": 133460
+    },
+    {
+      "epoch": 0.9661447588438402,
+      "grad_norm": 0.148344486951828,
+      "learning_rate": 4.033862479822219e-06,
+      "loss": 0.8585,
+      "step": 133470
+    },
+    {
+      "epoch": 0.9662171455044264,
+      "grad_norm": 0.14319495856761932,
+      "learning_rate": 4.033790093161632e-06,
+      "loss": 0.8753,
+      "step": 133480
+    },
+    {
+      "epoch": 0.9662895321650127,
+      "grad_norm": 0.16567128896713257,
+      "learning_rate": 4.033717706501046e-06,
+      "loss": 0.8751,
+      "step": 133490
+    },
+    {
+      "epoch": 0.9663619188255989,
+      "grad_norm": 0.14397595822811127,
+      "learning_rate": 4.03364531984046e-06,
+      "loss": 0.8629,
+      "step": 133500
+    },
+    {
+      "epoch": 0.966434305486185,
+      "grad_norm": 0.15650872886180878,
+      "learning_rate": 4.033572933179874e-06,
+      "loss": 0.8647,
+      "step": 133510
+    },
+    {
+      "epoch": 0.9665066921467712,
+      "grad_norm": 0.15653972327709198,
+      "learning_rate": 4.033500546519288e-06,
+      "loss": 0.8722,
+      "step": 133520
+    },
+    {
+      "epoch": 0.9665790788073574,
+      "grad_norm": 0.15006305277347565,
+      "learning_rate": 4.033428159858701e-06,
+      "loss": 0.8495,
+      "step": 133530
+    },
+    {
+      "epoch": 0.9666514654679436,
+      "grad_norm": 0.14864414930343628,
+      "learning_rate": 4.033355773198116e-06,
+      "loss": 0.8673,
+      "step": 133540
+    },
+    {
+      "epoch": 0.9667238521285297,
+      "grad_norm": 0.14984916150569916,
+      "learning_rate": 4.033283386537529e-06,
+      "loss": 0.8704,
+      "step": 133550
+    },
+    {
+      "epoch": 0.9667962387891159,
+      "grad_norm": 0.14670242369174957,
+      "learning_rate": 4.033210999876943e-06,
+      "loss": 0.8673,
+      "step": 133560
+    },
+    {
+      "epoch": 0.9668686254497021,
+      "grad_norm": 0.1624784618616104,
+      "learning_rate": 4.0331386132163565e-06,
+      "loss": 0.863,
+      "step": 133570
+    },
+    {
+      "epoch": 0.9669410121102883,
+      "grad_norm": 0.1424523890018463,
+      "learning_rate": 4.033066226555771e-06,
+      "loss": 0.8602,
+      "step": 133580
+    },
+    {
+      "epoch": 0.9670133987708746,
+      "grad_norm": 0.17853686213493347,
+      "learning_rate": 4.032993839895185e-06,
+      "loss": 0.8608,
+      "step": 133590
+    },
+    {
+      "epoch": 0.9670857854314607,
+      "grad_norm": 0.15653735399246216,
+      "learning_rate": 4.032921453234598e-06,
+      "loss": 0.8705,
+      "step": 133600
+    },
+    {
+      "epoch": 0.9671581720920469,
+      "grad_norm": 0.14804360270500183,
+      "learning_rate": 4.032849066574012e-06,
+      "loss": 0.8701,
+      "step": 133610
+    },
+    {
+      "epoch": 0.9672305587526331,
+      "grad_norm": 0.1479026824235916,
+      "learning_rate": 4.032776679913426e-06,
+      "loss": 0.8601,
+      "step": 133620
+    },
+    {
+      "epoch": 0.9673029454132193,
+      "grad_norm": 0.15589579939842224,
+      "learning_rate": 4.03270429325284e-06,
+      "loss": 0.8698,
+      "step": 133630
+    },
+    {
+      "epoch": 0.9673753320738054,
+      "grad_norm": 0.14816348254680634,
+      "learning_rate": 4.0326319065922535e-06,
+      "loss": 0.8706,
+      "step": 133640
+    },
+    {
+      "epoch": 0.9674477187343916,
+      "grad_norm": 0.15198203921318054,
+      "learning_rate": 4.032559519931667e-06,
+      "loss": 0.8615,
+      "step": 133650
+    },
+    {
+      "epoch": 0.9675201053949778,
+      "grad_norm": 0.16135896742343903,
+      "learning_rate": 4.032487133271082e-06,
+      "loss": 0.8692,
+      "step": 133660
+    },
+    {
+      "epoch": 0.967592492055564,
+      "grad_norm": 0.1466723084449768,
+      "learning_rate": 4.032414746610495e-06,
+      "loss": 0.8766,
+      "step": 133670
+    },
+    {
+      "epoch": 0.9676648787161501,
+      "grad_norm": 0.1564931869506836,
+      "learning_rate": 4.032342359949909e-06,
+      "loss": 0.8609,
+      "step": 133680
+    },
+    {
+      "epoch": 0.9677372653767363,
+      "grad_norm": 0.18006184697151184,
+      "learning_rate": 4.0322699732893225e-06,
+      "loss": 0.8721,
+      "step": 133690
+    },
+    {
+      "epoch": 0.9678096520373226,
+      "grad_norm": 0.1543225795030594,
+      "learning_rate": 4.032197586628737e-06,
+      "loss": 0.8594,
+      "step": 133700
+    },
+    {
+      "epoch": 0.9678820386979088,
+      "grad_norm": 0.14942817389965057,
+      "learning_rate": 4.0321251999681505e-06,
+      "loss": 0.8686,
+      "step": 133710
+    },
+    {
+      "epoch": 0.967954425358495,
+      "grad_norm": 0.1509200930595398,
+      "learning_rate": 4.032052813307564e-06,
+      "loss": 0.8603,
+      "step": 133720
+    },
+    {
+      "epoch": 0.9680268120190811,
+      "grad_norm": 0.20642779767513275,
+      "learning_rate": 4.031980426646978e-06,
+      "loss": 0.8635,
+      "step": 133730
+    },
+    {
+      "epoch": 0.9680991986796673,
+      "grad_norm": 0.1582535356283188,
+      "learning_rate": 4.031908039986392e-06,
+      "loss": 0.8653,
+      "step": 133740
+    },
+    {
+      "epoch": 0.9681715853402535,
+      "grad_norm": 0.15433159470558167,
+      "learning_rate": 4.031835653325805e-06,
+      "loss": 0.866,
+      "step": 133750
+    },
+    {
+      "epoch": 0.9682439720008397,
+      "grad_norm": 0.14288122951984406,
+      "learning_rate": 4.031763266665219e-06,
+      "loss": 0.8631,
+      "step": 133760
+    },
+    {
+      "epoch": 0.9683163586614258,
+      "grad_norm": 0.14594586193561554,
+      "learning_rate": 4.031690880004633e-06,
+      "loss": 0.8562,
+      "step": 133770
+    },
+    {
+      "epoch": 0.968388745322012,
+      "grad_norm": 0.1446864902973175,
+      "learning_rate": 4.031618493344047e-06,
+      "loss": 0.8636,
+      "step": 133780
+    },
+    {
+      "epoch": 0.9684611319825982,
+      "grad_norm": 0.15992321074008942,
+      "learning_rate": 4.03154610668346e-06,
+      "loss": 0.8636,
+      "step": 133790
+    },
+    {
+      "epoch": 0.9685335186431845,
+      "grad_norm": 0.15092787146568298,
+      "learning_rate": 4.031473720022874e-06,
+      "loss": 0.8564,
+      "step": 133800
+    },
+    {
+      "epoch": 0.9686059053037707,
+      "grad_norm": 0.1624569296836853,
+      "learning_rate": 4.031401333362288e-06,
+      "loss": 0.8754,
+      "step": 133810
+    },
+    {
+      "epoch": 0.9686782919643568,
+      "grad_norm": 0.14972303807735443,
+      "learning_rate": 4.031328946701702e-06,
+      "loss": 0.852,
+      "step": 133820
+    },
+    {
+      "epoch": 0.968750678624943,
+      "grad_norm": 0.1559324413537979,
+      "learning_rate": 4.031256560041116e-06,
+      "loss": 0.8814,
+      "step": 133830
+    },
+    {
+      "epoch": 0.9688230652855292,
+      "grad_norm": 0.14922378957271576,
+      "learning_rate": 4.031184173380529e-06,
+      "loss": 0.8717,
+      "step": 133840
+    },
+    {
+      "epoch": 0.9688954519461154,
+      "grad_norm": 0.15362094342708588,
+      "learning_rate": 4.031111786719944e-06,
+      "loss": 0.8554,
+      "step": 133850
+    },
+    {
+      "epoch": 0.9689678386067015,
+      "grad_norm": 0.16382701694965363,
+      "learning_rate": 4.031039400059357e-06,
+      "loss": 0.8724,
+      "step": 133860
+    },
+    {
+      "epoch": 0.9690402252672877,
+      "grad_norm": 0.15398184955120087,
+      "learning_rate": 4.030967013398771e-06,
+      "loss": 0.8605,
+      "step": 133870
+    },
+    {
+      "epoch": 0.9691126119278739,
+      "grad_norm": 0.16252100467681885,
+      "learning_rate": 4.0308946267381846e-06,
+      "loss": 0.8688,
+      "step": 133880
+    },
+    {
+      "epoch": 0.9691849985884601,
+      "grad_norm": 0.15495038032531738,
+      "learning_rate": 4.030822240077599e-06,
+      "loss": 0.8642,
+      "step": 133890
+    },
+    {
+      "epoch": 0.9692573852490463,
+      "grad_norm": 0.15822090208530426,
+      "learning_rate": 4.030749853417013e-06,
+      "loss": 0.8567,
+      "step": 133900
+    },
+    {
+      "epoch": 0.9693297719096325,
+      "grad_norm": 0.1519637256860733,
+      "learning_rate": 4.030677466756426e-06,
+      "loss": 0.8769,
+      "step": 133910
+    },
+    {
+      "epoch": 0.9694021585702187,
+      "grad_norm": 0.1706445962190628,
+      "learning_rate": 4.03060508009584e-06,
+      "loss": 0.861,
+      "step": 133920
+    },
+    {
+      "epoch": 0.9694745452308049,
+      "grad_norm": 0.15888677537441254,
+      "learning_rate": 4.030532693435254e-06,
+      "loss": 0.8584,
+      "step": 133930
+    },
+    {
+      "epoch": 0.9695469318913911,
+      "grad_norm": 0.18062569200992584,
+      "learning_rate": 4.030460306774668e-06,
+      "loss": 0.8679,
+      "step": 133940
+    },
+    {
+      "epoch": 0.9696193185519772,
+      "grad_norm": 0.1512601226568222,
+      "learning_rate": 4.0303879201140816e-06,
+      "loss": 0.8515,
+      "step": 133950
+    },
+    {
+      "epoch": 0.9696917052125634,
+      "grad_norm": 0.1415674239397049,
+      "learning_rate": 4.030315533453495e-06,
+      "loss": 0.8646,
+      "step": 133960
+    },
+    {
+      "epoch": 0.9697640918731496,
+      "grad_norm": 0.1439589112997055,
+      "learning_rate": 4.03024314679291e-06,
+      "loss": 0.8688,
+      "step": 133970
+    },
+    {
+      "epoch": 0.9698364785337358,
+      "grad_norm": 0.17175428569316864,
+      "learning_rate": 4.030170760132323e-06,
+      "loss": 0.8702,
+      "step": 133980
+    },
+    {
+      "epoch": 0.969908865194322,
+      "grad_norm": 0.15791040658950806,
+      "learning_rate": 4.030098373471737e-06,
+      "loss": 0.8602,
+      "step": 133990
+    },
+    {
+      "epoch": 0.9699812518549081,
+      "grad_norm": 0.163300022482872,
+      "learning_rate": 4.0300259868111505e-06,
+      "loss": 0.8759,
+      "step": 134000
+    },
+    {
+      "epoch": 0.9700536385154943,
+      "grad_norm": 0.15517699718475342,
+      "learning_rate": 4.029953600150565e-06,
+      "loss": 0.8717,
+      "step": 134010
+    },
+    {
+      "epoch": 0.9701260251760806,
+      "grad_norm": 0.15258009731769562,
+      "learning_rate": 4.029881213489979e-06,
+      "loss": 0.8838,
+      "step": 134020
+    },
+    {
+      "epoch": 0.9701984118366668,
+      "grad_norm": 0.16164755821228027,
+      "learning_rate": 4.029808826829392e-06,
+      "loss": 0.8635,
+      "step": 134030
+    },
+    {
+      "epoch": 0.970270798497253,
+      "grad_norm": 0.1693074256181717,
+      "learning_rate": 4.029736440168806e-06,
+      "loss": 0.8551,
+      "step": 134040
+    },
+    {
+      "epoch": 0.9703431851578391,
+      "grad_norm": 0.15383243560791016,
+      "learning_rate": 4.02966405350822e-06,
+      "loss": 0.87,
+      "step": 134050
+    },
+    {
+      "epoch": 0.9704155718184253,
+      "grad_norm": 0.1397397667169571,
+      "learning_rate": 4.029591666847634e-06,
+      "loss": 0.8602,
+      "step": 134060
+    },
+    {
+      "epoch": 0.9704879584790115,
+      "grad_norm": 0.15033358335494995,
+      "learning_rate": 4.0295192801870475e-06,
+      "loss": 0.858,
+      "step": 134070
+    },
+    {
+      "epoch": 0.9705603451395977,
+      "grad_norm": 0.1515614092350006,
+      "learning_rate": 4.029446893526461e-06,
+      "loss": 0.8869,
+      "step": 134080
+    },
+    {
+      "epoch": 0.9706327318001838,
+      "grad_norm": 0.1568913608789444,
+      "learning_rate": 4.029374506865876e-06,
+      "loss": 0.8814,
+      "step": 134090
+    },
+    {
+      "epoch": 0.97070511846077,
+      "grad_norm": 0.1536082923412323,
+      "learning_rate": 4.029302120205289e-06,
+      "loss": 0.8603,
+      "step": 134100
+    },
+    {
+      "epoch": 0.9707775051213562,
+      "grad_norm": 0.16278186440467834,
+      "learning_rate": 4.029229733544703e-06,
+      "loss": 0.8659,
+      "step": 134110
+    },
+    {
+      "epoch": 0.9708498917819425,
+      "grad_norm": 0.15788668394088745,
+      "learning_rate": 4.0291573468841164e-06,
+      "loss": 0.8711,
+      "step": 134120
+    },
+    {
+      "epoch": 0.9709222784425287,
+      "grad_norm": 0.14995813369750977,
+      "learning_rate": 4.02908496022353e-06,
+      "loss": 0.8567,
+      "step": 134130
+    },
+    {
+      "epoch": 0.9709946651031148,
+      "grad_norm": 0.1554499715566635,
+      "learning_rate": 4.0290125735629445e-06,
+      "loss": 0.8612,
+      "step": 134140
+    },
+    {
+      "epoch": 0.971067051763701,
+      "grad_norm": 0.15231886506080627,
+      "learning_rate": 4.028940186902358e-06,
+      "loss": 0.8668,
+      "step": 134150
+    },
+    {
+      "epoch": 0.9711394384242872,
+      "grad_norm": 0.16602960228919983,
+      "learning_rate": 4.028867800241772e-06,
+      "loss": 0.8603,
+      "step": 134160
+    },
+    {
+      "epoch": 0.9712118250848734,
+      "grad_norm": 0.15559101104736328,
+      "learning_rate": 4.028795413581185e-06,
+      "loss": 0.8724,
+      "step": 134170
+    },
+    {
+      "epoch": 0.9712842117454595,
+      "grad_norm": 0.15899375081062317,
+      "learning_rate": 4.0287230269206e-06,
+      "loss": 0.8588,
+      "step": 134180
+    },
+    {
+      "epoch": 0.9713565984060457,
+      "grad_norm": 0.1541290581226349,
+      "learning_rate": 4.0286506402600134e-06,
+      "loss": 0.8554,
+      "step": 134190
+    },
+    {
+      "epoch": 0.9714289850666319,
+      "grad_norm": 0.16547095775604248,
+      "learning_rate": 4.028578253599427e-06,
+      "loss": 0.8729,
+      "step": 134200
+    },
+    {
+      "epoch": 0.9715013717272181,
+      "grad_norm": 0.14189139008522034,
+      "learning_rate": 4.028505866938841e-06,
+      "loss": 0.8666,
+      "step": 134210
+    },
+    {
+      "epoch": 0.9715737583878042,
+      "grad_norm": 0.16001036763191223,
+      "learning_rate": 4.028433480278255e-06,
+      "loss": 0.87,
+      "step": 134220
+    },
+    {
+      "epoch": 0.9716461450483905,
+      "grad_norm": 0.163381889462471,
+      "learning_rate": 4.028361093617669e-06,
+      "loss": 0.8479,
+      "step": 134230
+    },
+    {
+      "epoch": 0.9717185317089767,
+      "grad_norm": 0.15338028967380524,
+      "learning_rate": 4.028288706957082e-06,
+      "loss": 0.8646,
+      "step": 134240
+    },
+    {
+      "epoch": 0.9717909183695629,
+      "grad_norm": 0.152128666639328,
+      "learning_rate": 4.028216320296496e-06,
+      "loss": 0.8672,
+      "step": 134250
+    },
+    {
+      "epoch": 0.9718633050301491,
+      "grad_norm": 0.1537809669971466,
+      "learning_rate": 4.0281439336359105e-06,
+      "loss": 0.8555,
+      "step": 134260
+    },
+    {
+      "epoch": 0.9719356916907352,
+      "grad_norm": 0.14656396210193634,
+      "learning_rate": 4.028071546975324e-06,
+      "loss": 0.8636,
+      "step": 134270
+    },
+    {
+      "epoch": 0.9720080783513214,
+      "grad_norm": 0.15450994670391083,
+      "learning_rate": 4.027999160314737e-06,
+      "loss": 0.8658,
+      "step": 134280
+    },
+    {
+      "epoch": 0.9720804650119076,
+      "grad_norm": 0.15121851861476898,
+      "learning_rate": 4.027926773654151e-06,
+      "loss": 0.8681,
+      "step": 134290
+    },
+    {
+      "epoch": 0.9721528516724938,
+      "grad_norm": 0.1601051241159439,
+      "learning_rate": 4.027854386993565e-06,
+      "loss": 0.864,
+      "step": 134300
+    },
+    {
+      "epoch": 0.9722252383330799,
+      "grad_norm": 0.14622844755649567,
+      "learning_rate": 4.0277820003329785e-06,
+      "loss": 0.8713,
+      "step": 134310
+    },
+    {
+      "epoch": 0.9722976249936661,
+      "grad_norm": 0.14876750111579895,
+      "learning_rate": 4.027709613672392e-06,
+      "loss": 0.8719,
+      "step": 134320
+    },
+    {
+      "epoch": 0.9723700116542524,
+      "grad_norm": 0.17520910501480103,
+      "learning_rate": 4.027637227011807e-06,
+      "loss": 0.8661,
+      "step": 134330
+    },
+    {
+      "epoch": 0.9724423983148386,
+      "grad_norm": 0.1505105048418045,
+      "learning_rate": 4.02756484035122e-06,
+      "loss": 0.8659,
+      "step": 134340
+    },
+    {
+      "epoch": 0.9725147849754248,
+      "grad_norm": 0.18113350868225098,
+      "learning_rate": 4.027492453690634e-06,
+      "loss": 0.8705,
+      "step": 134350
+    },
+    {
+      "epoch": 0.9725871716360109,
+      "grad_norm": 0.14961007237434387,
+      "learning_rate": 4.0274200670300475e-06,
+      "loss": 0.8694,
+      "step": 134360
+    },
+    {
+      "epoch": 0.9726595582965971,
+      "grad_norm": 0.14745749533176422,
+      "learning_rate": 4.027347680369462e-06,
+      "loss": 0.8727,
+      "step": 134370
+    },
+    {
+      "epoch": 0.9727319449571833,
+      "grad_norm": 0.14691106975078583,
+      "learning_rate": 4.0272752937088755e-06,
+      "loss": 0.8673,
+      "step": 134380
+    },
+    {
+      "epoch": 0.9728043316177695,
+      "grad_norm": 0.145847350358963,
+      "learning_rate": 4.027202907048289e-06,
+      "loss": 0.8603,
+      "step": 134390
+    },
+    {
+      "epoch": 0.9728767182783556,
+      "grad_norm": 0.16000697016716003,
+      "learning_rate": 4.027130520387703e-06,
+      "loss": 0.8592,
+      "step": 134400
+    },
+    {
+      "epoch": 0.9729491049389418,
+      "grad_norm": 0.15306724607944489,
+      "learning_rate": 4.027058133727117e-06,
+      "loss": 0.8608,
+      "step": 134410
+    },
+    {
+      "epoch": 0.973021491599528,
+      "grad_norm": 0.1496717482805252,
+      "learning_rate": 4.026985747066531e-06,
+      "loss": 0.8751,
+      "step": 134420
+    },
+    {
+      "epoch": 0.9730938782601142,
+      "grad_norm": 0.1527402400970459,
+      "learning_rate": 4.0269133604059445e-06,
+      "loss": 0.8678,
+      "step": 134430
+    },
+    {
+      "epoch": 0.9731662649207005,
+      "grad_norm": 0.1510053426027298,
+      "learning_rate": 4.026840973745358e-06,
+      "loss": 0.8683,
+      "step": 134440
+    },
+    {
+      "epoch": 0.9732386515812866,
+      "grad_norm": 0.157288059592247,
+      "learning_rate": 4.0267685870847725e-06,
+      "loss": 0.8571,
+      "step": 134450
+    },
+    {
+      "epoch": 0.9733110382418728,
+      "grad_norm": 0.1452171355485916,
+      "learning_rate": 4.026696200424186e-06,
+      "loss": 0.8649,
+      "step": 134460
+    },
+    {
+      "epoch": 0.973383424902459,
+      "grad_norm": 0.15778832137584686,
+      "learning_rate": 4.0266238137636e-06,
+      "loss": 0.8636,
+      "step": 134470
+    },
+    {
+      "epoch": 0.9734558115630452,
+      "grad_norm": 0.14281177520751953,
+      "learning_rate": 4.026551427103013e-06,
+      "loss": 0.8588,
+      "step": 134480
+    },
+    {
+      "epoch": 0.9735281982236313,
+      "grad_norm": 0.14823725819587708,
+      "learning_rate": 4.026479040442428e-06,
+      "loss": 0.8644,
+      "step": 134490
+    },
+    {
+      "epoch": 0.9736005848842175,
+      "grad_norm": 0.14900419116020203,
+      "learning_rate": 4.0264066537818415e-06,
+      "loss": 0.8796,
+      "step": 134500
+    },
+    {
+      "epoch": 0.9736729715448037,
+      "grad_norm": 0.1535101979970932,
+      "learning_rate": 4.026334267121255e-06,
+      "loss": 0.8621,
+      "step": 134510
+    },
+    {
+      "epoch": 0.9737453582053899,
+      "grad_norm": 0.1498989462852478,
+      "learning_rate": 4.026261880460669e-06,
+      "loss": 0.8546,
+      "step": 134520
+    },
+    {
+      "epoch": 0.973817744865976,
+      "grad_norm": 0.16558587551116943,
+      "learning_rate": 4.026189493800083e-06,
+      "loss": 0.8622,
+      "step": 134530
+    },
+    {
+      "epoch": 0.9738901315265622,
+      "grad_norm": 0.15267030894756317,
+      "learning_rate": 4.026117107139497e-06,
+      "loss": 0.8666,
+      "step": 134540
+    },
+    {
+      "epoch": 0.9739625181871485,
+      "grad_norm": 0.15370555222034454,
+      "learning_rate": 4.02604472047891e-06,
+      "loss": 0.8474,
+      "step": 134550
+    },
+    {
+      "epoch": 0.9740349048477347,
+      "grad_norm": 0.1497911959886551,
+      "learning_rate": 4.025972333818324e-06,
+      "loss": 0.8697,
+      "step": 134560
+    },
+    {
+      "epoch": 0.9741072915083209,
+      "grad_norm": 0.15965795516967773,
+      "learning_rate": 4.0258999471577385e-06,
+      "loss": 0.869,
+      "step": 134570
+    },
+    {
+      "epoch": 0.974179678168907,
+      "grad_norm": 0.14840437471866608,
+      "learning_rate": 4.025827560497152e-06,
+      "loss": 0.8601,
+      "step": 134580
+    },
+    {
+      "epoch": 0.9742520648294932,
+      "grad_norm": 0.15610599517822266,
+      "learning_rate": 4.025755173836566e-06,
+      "loss": 0.8606,
+      "step": 134590
+    },
+    {
+      "epoch": 0.9743244514900794,
+      "grad_norm": 0.14686964452266693,
+      "learning_rate": 4.025682787175979e-06,
+      "loss": 0.8566,
+      "step": 134600
+    },
+    {
+      "epoch": 0.9743968381506656,
+      "grad_norm": 0.16029107570648193,
+      "learning_rate": 4.025610400515394e-06,
+      "loss": 0.8528,
+      "step": 134610
+    },
+    {
+      "epoch": 0.9744692248112518,
+      "grad_norm": 0.15175573527812958,
+      "learning_rate": 4.025538013854807e-06,
+      "loss": 0.8688,
+      "step": 134620
+    },
+    {
+      "epoch": 0.9745416114718379,
+      "grad_norm": 0.16867345571517944,
+      "learning_rate": 4.025465627194221e-06,
+      "loss": 0.8564,
+      "step": 134630
+    },
+    {
+      "epoch": 0.9746139981324241,
+      "grad_norm": 0.1488535851240158,
+      "learning_rate": 4.025393240533635e-06,
+      "loss": 0.8589,
+      "step": 134640
+    },
+    {
+      "epoch": 0.9746863847930104,
+      "grad_norm": 0.147738978266716,
+      "learning_rate": 4.025320853873049e-06,
+      "loss": 0.8754,
+      "step": 134650
+    },
+    {
+      "epoch": 0.9747587714535966,
+      "grad_norm": 0.14682996273040771,
+      "learning_rate": 4.025248467212463e-06,
+      "loss": 0.8585,
+      "step": 134660
+    },
+    {
+      "epoch": 0.9748311581141827,
+      "grad_norm": 0.17877760529518127,
+      "learning_rate": 4.025176080551876e-06,
+      "loss": 0.8617,
+      "step": 134670
+    },
+    {
+      "epoch": 0.9749035447747689,
+      "grad_norm": 0.3457520008087158,
+      "learning_rate": 4.02510369389129e-06,
+      "loss": 0.8692,
+      "step": 134680
+    },
+    {
+      "epoch": 0.9749759314353551,
+      "grad_norm": 0.1764483004808426,
+      "learning_rate": 4.025031307230704e-06,
+      "loss": 0.8658,
+      "step": 134690
+    },
+    {
+      "epoch": 0.9750483180959413,
+      "grad_norm": 0.17922911047935486,
+      "learning_rate": 4.024958920570118e-06,
+      "loss": 0.8704,
+      "step": 134700
+    },
+    {
+      "epoch": 0.9751207047565275,
+      "grad_norm": 0.1529940515756607,
+      "learning_rate": 4.024886533909532e-06,
+      "loss": 0.8758,
+      "step": 134710
+    },
+    {
+      "epoch": 0.9751930914171136,
+      "grad_norm": 0.1562705934047699,
+      "learning_rate": 4.024814147248945e-06,
+      "loss": 0.8639,
+      "step": 134720
+    },
+    {
+      "epoch": 0.9752654780776998,
+      "grad_norm": 0.15895576775074005,
+      "learning_rate": 4.024741760588359e-06,
+      "loss": 0.8625,
+      "step": 134730
+    },
+    {
+      "epoch": 0.975337864738286,
+      "grad_norm": 0.14802898466587067,
+      "learning_rate": 4.024669373927773e-06,
+      "loss": 0.8735,
+      "step": 134740
+    },
+    {
+      "epoch": 0.9754102513988722,
+      "grad_norm": 0.17037954926490784,
+      "learning_rate": 4.024596987267187e-06,
+      "loss": 0.873,
+      "step": 134750
+    },
+    {
+      "epoch": 0.9754826380594585,
+      "grad_norm": 0.16572614014148712,
+      "learning_rate": 4.024524600606601e-06,
+      "loss": 0.859,
+      "step": 134760
+    },
+    {
+      "epoch": 0.9755550247200446,
+      "grad_norm": 0.15422946214675903,
+      "learning_rate": 4.024452213946014e-06,
+      "loss": 0.8608,
+      "step": 134770
+    },
+    {
+      "epoch": 0.9756274113806308,
+      "grad_norm": 0.14978817105293274,
+      "learning_rate": 4.024379827285429e-06,
+      "loss": 0.8539,
+      "step": 134780
+    },
+    {
+      "epoch": 0.975699798041217,
+      "grad_norm": 0.15020768344402313,
+      "learning_rate": 4.024307440624842e-06,
+      "loss": 0.8647,
+      "step": 134790
+    },
+    {
+      "epoch": 0.9757721847018032,
+      "grad_norm": 0.1630668342113495,
+      "learning_rate": 4.024235053964256e-06,
+      "loss": 0.8748,
+      "step": 134800
+    },
+    {
+      "epoch": 0.9758445713623893,
+      "grad_norm": 0.1647442728281021,
+      "learning_rate": 4.0241626673036695e-06,
+      "loss": 0.8916,
+      "step": 134810
+    },
+    {
+      "epoch": 0.9759169580229755,
+      "grad_norm": 0.16220815479755402,
+      "learning_rate": 4.024090280643083e-06,
+      "loss": 0.8647,
+      "step": 134820
+    },
+    {
+      "epoch": 0.9759893446835617,
+      "grad_norm": 0.15508753061294556,
+      "learning_rate": 4.024017893982497e-06,
+      "loss": 0.8776,
+      "step": 134830
+    },
+    {
+      "epoch": 0.9760617313441479,
+      "grad_norm": 0.14630411565303802,
+      "learning_rate": 4.023945507321911e-06,
+      "loss": 0.862,
+      "step": 134840
+    },
+    {
+      "epoch": 0.976134118004734,
+      "grad_norm": 0.16463924944400787,
+      "learning_rate": 4.023873120661325e-06,
+      "loss": 0.8738,
+      "step": 134850
+    },
+    {
+      "epoch": 0.9762065046653203,
+      "grad_norm": 0.14505267143249512,
+      "learning_rate": 4.0238007340007384e-06,
+      "loss": 0.8578,
+      "step": 134860
+    },
+    {
+      "epoch": 0.9762788913259065,
+      "grad_norm": 0.15662142634391785,
+      "learning_rate": 4.023728347340152e-06,
+      "loss": 0.8646,
+      "step": 134870
+    },
+    {
+      "epoch": 0.9763512779864927,
+      "grad_norm": 0.3217943012714386,
+      "learning_rate": 4.0236559606795665e-06,
+      "loss": 0.87,
+      "step": 134880
+    },
+    {
+      "epoch": 0.9764236646470789,
+      "grad_norm": 0.1467258632183075,
+      "learning_rate": 4.02358357401898e-06,
+      "loss": 0.8735,
+      "step": 134890
+    },
+    {
+      "epoch": 0.976496051307665,
+      "grad_norm": 0.1595139056444168,
+      "learning_rate": 4.023511187358394e-06,
+      "loss": 0.8677,
+      "step": 134900
+    },
+    {
+      "epoch": 0.9765684379682512,
+      "grad_norm": 0.16887742280960083,
+      "learning_rate": 4.023438800697807e-06,
+      "loss": 0.8625,
+      "step": 134910
+    },
+    {
+      "epoch": 0.9766408246288374,
+      "grad_norm": 0.2070261687040329,
+      "learning_rate": 4.023366414037221e-06,
+      "loss": 0.8599,
+      "step": 134920
+    },
+    {
+      "epoch": 0.9767132112894236,
+      "grad_norm": 0.1530458927154541,
+      "learning_rate": 4.0232940273766354e-06,
+      "loss": 0.859,
+      "step": 134930
+    },
+    {
+      "epoch": 0.9767855979500097,
+      "grad_norm": 0.17413754761219025,
+      "learning_rate": 4.023221640716049e-06,
+      "loss": 0.8744,
+      "step": 134940
+    },
+    {
+      "epoch": 0.9768579846105959,
+      "grad_norm": 0.14189285039901733,
+      "learning_rate": 4.023149254055463e-06,
+      "loss": 0.8678,
+      "step": 134950
+    },
+    {
+      "epoch": 0.9769303712711821,
+      "grad_norm": 0.16996034979820251,
+      "learning_rate": 4.023076867394876e-06,
+      "loss": 0.8654,
+      "step": 134960
+    },
+    {
+      "epoch": 0.9770027579317684,
+      "grad_norm": 0.15220043063163757,
+      "learning_rate": 4.023004480734291e-06,
+      "loss": 0.8705,
+      "step": 134970
+    },
+    {
+      "epoch": 0.9770751445923546,
+      "grad_norm": 0.1522054523229599,
+      "learning_rate": 4.022932094073704e-06,
+      "loss": 0.8677,
+      "step": 134980
+    },
+    {
+      "epoch": 0.9771475312529407,
+      "grad_norm": 0.14856263995170593,
+      "learning_rate": 4.022859707413118e-06,
+      "loss": 0.8604,
+      "step": 134990
+    },
+    {
+      "epoch": 0.9772199179135269,
+      "grad_norm": 0.17470620572566986,
+      "learning_rate": 4.022787320752532e-06,
+      "loss": 0.8735,
+      "step": 135000
+    },
+    {
+      "epoch": 0.9772923045741131,
+      "grad_norm": 0.16149313747882843,
+      "learning_rate": 4.022714934091946e-06,
+      "loss": 0.8694,
+      "step": 135010
+    },
+    {
+      "epoch": 0.9773646912346993,
+      "grad_norm": 0.1480475217103958,
+      "learning_rate": 4.02264254743136e-06,
+      "loss": 0.8597,
+      "step": 135020
+    },
+    {
+      "epoch": 0.9774370778952854,
+      "grad_norm": 0.16102159023284912,
+      "learning_rate": 4.022570160770773e-06,
+      "loss": 0.8735,
+      "step": 135030
+    },
+    {
+      "epoch": 0.9775094645558716,
+      "grad_norm": 0.14369907975196838,
+      "learning_rate": 4.022497774110187e-06,
+      "loss": 0.8643,
+      "step": 135040
+    },
+    {
+      "epoch": 0.9775818512164578,
+      "grad_norm": 0.1809437870979309,
+      "learning_rate": 4.022425387449601e-06,
+      "loss": 0.8584,
+      "step": 135050
+    },
+    {
+      "epoch": 0.977654237877044,
+      "grad_norm": 0.15481232106685638,
+      "learning_rate": 4.022353000789015e-06,
+      "loss": 0.8688,
+      "step": 135060
+    },
+    {
+      "epoch": 0.9777266245376302,
+      "grad_norm": 0.1535213589668274,
+      "learning_rate": 4.022280614128429e-06,
+      "loss": 0.8592,
+      "step": 135070
+    },
+    {
+      "epoch": 0.9777990111982164,
+      "grad_norm": 0.14975209534168243,
+      "learning_rate": 4.022208227467842e-06,
+      "loss": 0.8657,
+      "step": 135080
+    },
+    {
+      "epoch": 0.9778713978588026,
+      "grad_norm": 0.15290719270706177,
+      "learning_rate": 4.022135840807257e-06,
+      "loss": 0.8633,
+      "step": 135090
+    },
+    {
+      "epoch": 0.9779437845193888,
+      "grad_norm": 0.15675273537635803,
+      "learning_rate": 4.02206345414667e-06,
+      "loss": 0.8642,
+      "step": 135100
+    },
+    {
+      "epoch": 0.978016171179975,
+      "grad_norm": 0.17015263438224792,
+      "learning_rate": 4.021991067486084e-06,
+      "loss": 0.8693,
+      "step": 135110
+    },
+    {
+      "epoch": 0.9780885578405611,
+      "grad_norm": 0.15289832651615143,
+      "learning_rate": 4.0219186808254975e-06,
+      "loss": 0.8696,
+      "step": 135120
+    },
+    {
+      "epoch": 0.9781609445011473,
+      "grad_norm": 0.15227100253105164,
+      "learning_rate": 4.021846294164912e-06,
+      "loss": 0.8625,
+      "step": 135130
+    },
+    {
+      "epoch": 0.9782333311617335,
+      "grad_norm": 0.19440871477127075,
+      "learning_rate": 4.021773907504326e-06,
+      "loss": 0.8697,
+      "step": 135140
+    },
+    {
+      "epoch": 0.9783057178223197,
+      "grad_norm": 0.15097714960575104,
+      "learning_rate": 4.021701520843739e-06,
+      "loss": 0.854,
+      "step": 135150
+    },
+    {
+      "epoch": 0.9783781044829059,
+      "grad_norm": 0.16234318912029266,
+      "learning_rate": 4.021629134183153e-06,
+      "loss": 0.8774,
+      "step": 135160
+    },
+    {
+      "epoch": 0.978450491143492,
+      "grad_norm": 0.19647642970085144,
+      "learning_rate": 4.021556747522567e-06,
+      "loss": 0.8644,
+      "step": 135170
+    },
+    {
+      "epoch": 0.9785228778040783,
+      "grad_norm": 0.14601005613803864,
+      "learning_rate": 4.021484360861981e-06,
+      "loss": 0.8577,
+      "step": 135180
+    },
+    {
+      "epoch": 0.9785952644646645,
+      "grad_norm": 0.15095117688179016,
+      "learning_rate": 4.0214119742013945e-06,
+      "loss": 0.8604,
+      "step": 135190
+    },
+    {
+      "epoch": 0.9786676511252507,
+      "grad_norm": 0.14409467577934265,
+      "learning_rate": 4.021339587540808e-06,
+      "loss": 0.8597,
+      "step": 135200
+    },
+    {
+      "epoch": 0.9787400377858368,
+      "grad_norm": 0.14906567335128784,
+      "learning_rate": 4.021267200880223e-06,
+      "loss": 0.8643,
+      "step": 135210
+    },
+    {
+      "epoch": 0.978812424446423,
+      "grad_norm": 0.15724803507328033,
+      "learning_rate": 4.021194814219636e-06,
+      "loss": 0.8458,
+      "step": 135220
+    },
+    {
+      "epoch": 0.9788848111070092,
+      "grad_norm": 0.14980719983577728,
+      "learning_rate": 4.02112242755905e-06,
+      "loss": 0.8623,
+      "step": 135230
+    },
+    {
+      "epoch": 0.9789571977675954,
+      "grad_norm": 0.14800138771533966,
+      "learning_rate": 4.0210500408984635e-06,
+      "loss": 0.8602,
+      "step": 135240
+    },
+    {
+      "epoch": 0.9790295844281816,
+      "grad_norm": 0.24829064309597015,
+      "learning_rate": 4.020977654237878e-06,
+      "loss": 0.8596,
+      "step": 135250
+    },
+    {
+      "epoch": 0.9791019710887677,
+      "grad_norm": 0.159278005361557,
+      "learning_rate": 4.0209052675772916e-06,
+      "loss": 0.8695,
+      "step": 135260
+    },
+    {
+      "epoch": 0.9791743577493539,
+      "grad_norm": 0.16504423320293427,
+      "learning_rate": 4.020832880916705e-06,
+      "loss": 0.8636,
+      "step": 135270
+    },
+    {
+      "epoch": 0.9792467444099401,
+      "grad_norm": 0.1613035798072815,
+      "learning_rate": 4.020760494256119e-06,
+      "loss": 0.8675,
+      "step": 135280
+    },
+    {
+      "epoch": 0.9793191310705264,
+      "grad_norm": 0.16693058609962463,
+      "learning_rate": 4.020688107595533e-06,
+      "loss": 0.8688,
+      "step": 135290
+    },
+    {
+      "epoch": 0.9793915177311125,
+      "grad_norm": 0.14621016383171082,
+      "learning_rate": 4.020615720934947e-06,
+      "loss": 0.8563,
+      "step": 135300
+    },
+    {
+      "epoch": 0.9794639043916987,
+      "grad_norm": 0.15493078529834747,
+      "learning_rate": 4.0205433342743605e-06,
+      "loss": 0.865,
+      "step": 135310
+    },
+    {
+      "epoch": 0.9795362910522849,
+      "grad_norm": 0.1841004192829132,
+      "learning_rate": 4.020470947613774e-06,
+      "loss": 0.8632,
+      "step": 135320
+    },
+    {
+      "epoch": 0.9796086777128711,
+      "grad_norm": 0.1623116284608841,
+      "learning_rate": 4.0203985609531886e-06,
+      "loss": 0.864,
+      "step": 135330
+    },
+    {
+      "epoch": 0.9796810643734573,
+      "grad_norm": 0.14910165965557098,
+      "learning_rate": 4.020326174292601e-06,
+      "loss": 0.8692,
+      "step": 135340
+    },
+    {
+      "epoch": 0.9797534510340434,
+      "grad_norm": 0.17828305065631866,
+      "learning_rate": 4.020253787632015e-06,
+      "loss": 0.8739,
+      "step": 135350
+    },
+    {
+      "epoch": 0.9798258376946296,
+      "grad_norm": 0.1424138993024826,
+      "learning_rate": 4.020181400971429e-06,
+      "loss": 0.8643,
+      "step": 135360
+    },
+    {
+      "epoch": 0.9798982243552158,
+      "grad_norm": 0.16113391518592834,
+      "learning_rate": 4.020109014310843e-06,
+      "loss": 0.8732,
+      "step": 135370
+    },
+    {
+      "epoch": 0.979970611015802,
+      "grad_norm": 0.17208503186702728,
+      "learning_rate": 4.020036627650257e-06,
+      "loss": 0.85,
+      "step": 135380
+    },
+    {
+      "epoch": 0.9800429976763883,
+      "grad_norm": 0.17494194209575653,
+      "learning_rate": 4.01996424098967e-06,
+      "loss": 0.8593,
+      "step": 135390
+    },
+    {
+      "epoch": 0.9801153843369744,
+      "grad_norm": 0.15494690835475922,
+      "learning_rate": 4.019891854329085e-06,
+      "loss": 0.8766,
+      "step": 135400
+    },
+    {
+      "epoch": 0.9801877709975606,
+      "grad_norm": 0.15408842265605927,
+      "learning_rate": 4.019819467668498e-06,
+      "loss": 0.8711,
+      "step": 135410
+    },
+    {
+      "epoch": 0.9802601576581468,
+      "grad_norm": 0.15154585242271423,
+      "learning_rate": 4.019747081007912e-06,
+      "loss": 0.8571,
+      "step": 135420
+    },
+    {
+      "epoch": 0.980332544318733,
+      "grad_norm": 0.16975507140159607,
+      "learning_rate": 4.0196746943473256e-06,
+      "loss": 0.8818,
+      "step": 135430
+    },
+    {
+      "epoch": 0.9804049309793191,
+      "grad_norm": 0.14920401573181152,
+      "learning_rate": 4.01960230768674e-06,
+      "loss": 0.869,
+      "step": 135440
+    },
+    {
+      "epoch": 0.9804773176399053,
+      "grad_norm": 0.15514759719371796,
+      "learning_rate": 4.019529921026154e-06,
+      "loss": 0.8593,
+      "step": 135450
+    },
+    {
+      "epoch": 0.9805497043004915,
+      "grad_norm": 0.16312751173973083,
+      "learning_rate": 4.019457534365567e-06,
+      "loss": 0.8585,
+      "step": 135460
+    },
+    {
+      "epoch": 0.9806220909610777,
+      "grad_norm": 0.14903192222118378,
+      "learning_rate": 4.019385147704981e-06,
+      "loss": 0.8557,
+      "step": 135470
+    },
+    {
+      "epoch": 0.9806944776216638,
+      "grad_norm": 0.15658678114414215,
+      "learning_rate": 4.019312761044395e-06,
+      "loss": 0.8683,
+      "step": 135480
+    },
+    {
+      "epoch": 0.98076686428225,
+      "grad_norm": 0.15279506146907806,
+      "learning_rate": 4.019240374383809e-06,
+      "loss": 0.8585,
+      "step": 135490
+    },
+    {
+      "epoch": 0.9808392509428363,
+      "grad_norm": 0.14755424857139587,
+      "learning_rate": 4.019167987723223e-06,
+      "loss": 0.8533,
+      "step": 135500
+    },
+    {
+      "epoch": 0.9809116376034225,
+      "grad_norm": 0.14603859186172485,
+      "learning_rate": 4.019095601062636e-06,
+      "loss": 0.8648,
+      "step": 135510
+    },
+    {
+      "epoch": 0.9809840242640087,
+      "grad_norm": 0.1589314043521881,
+      "learning_rate": 4.01902321440205e-06,
+      "loss": 0.8613,
+      "step": 135520
+    },
+    {
+      "epoch": 0.9810564109245948,
+      "grad_norm": 0.1501326709985733,
+      "learning_rate": 4.018950827741464e-06,
+      "loss": 0.8706,
+      "step": 135530
+    },
+    {
+      "epoch": 0.981128797585181,
+      "grad_norm": 0.15923535823822021,
+      "learning_rate": 4.018878441080878e-06,
+      "loss": 0.8739,
+      "step": 135540
+    },
+    {
+      "epoch": 0.9812011842457672,
+      "grad_norm": 0.15568740665912628,
+      "learning_rate": 4.0188060544202915e-06,
+      "loss": 0.8517,
+      "step": 135550
+    },
+    {
+      "epoch": 0.9812735709063534,
+      "grad_norm": 0.1511840522289276,
+      "learning_rate": 4.018733667759705e-06,
+      "loss": 0.8474,
+      "step": 135560
+    },
+    {
+      "epoch": 0.9813459575669395,
+      "grad_norm": 0.155585378408432,
+      "learning_rate": 4.01866128109912e-06,
+      "loss": 0.8598,
+      "step": 135570
+    },
+    {
+      "epoch": 0.9814183442275257,
+      "grad_norm": 0.15840795636177063,
+      "learning_rate": 4.018588894438533e-06,
+      "loss": 0.8683,
+      "step": 135580
+    },
+    {
+      "epoch": 0.9814907308881119,
+      "grad_norm": 0.1394014209508896,
+      "learning_rate": 4.018516507777947e-06,
+      "loss": 0.8745,
+      "step": 135590
+    },
+    {
+      "epoch": 0.9815631175486981,
+      "grad_norm": 0.15405994653701782,
+      "learning_rate": 4.0184441211173604e-06,
+      "loss": 0.876,
+      "step": 135600
+    },
+    {
+      "epoch": 0.9816355042092844,
+      "grad_norm": 0.15901432931423187,
+      "learning_rate": 4.018371734456775e-06,
+      "loss": 0.8676,
+      "step": 135610
+    },
+    {
+      "epoch": 0.9817078908698705,
+      "grad_norm": 0.150760680437088,
+      "learning_rate": 4.0182993477961885e-06,
+      "loss": 0.8553,
+      "step": 135620
+    },
+    {
+      "epoch": 0.9817802775304567,
+      "grad_norm": 0.22623537480831146,
+      "learning_rate": 4.018226961135602e-06,
+      "loss": 0.861,
+      "step": 135630
+    },
+    {
+      "epoch": 0.9818526641910429,
+      "grad_norm": 0.15437884628772736,
+      "learning_rate": 4.018154574475016e-06,
+      "loss": 0.8691,
+      "step": 135640
+    },
+    {
+      "epoch": 0.9819250508516291,
+      "grad_norm": 0.15233850479125977,
+      "learning_rate": 4.01808218781443e-06,
+      "loss": 0.8692,
+      "step": 135650
+    },
+    {
+      "epoch": 0.9819974375122152,
+      "grad_norm": 0.1725022792816162,
+      "learning_rate": 4.018009801153844e-06,
+      "loss": 0.8626,
+      "step": 135660
+    },
+    {
+      "epoch": 0.9820698241728014,
+      "grad_norm": 0.5819795727729797,
+      "learning_rate": 4.0179374144932574e-06,
+      "loss": 0.8739,
+      "step": 135670
+    },
+    {
+      "epoch": 0.9821422108333876,
+      "grad_norm": 0.20204482972621918,
+      "learning_rate": 4.017865027832671e-06,
+      "loss": 0.8681,
+      "step": 135680
+    },
+    {
+      "epoch": 0.9822145974939738,
+      "grad_norm": 0.15785406529903412,
+      "learning_rate": 4.0177926411720855e-06,
+      "loss": 0.8545,
+      "step": 135690
+    },
+    {
+      "epoch": 0.98228698415456,
+      "grad_norm": 0.17018094658851624,
+      "learning_rate": 4.017720254511499e-06,
+      "loss": 0.8569,
+      "step": 135700
+    },
+    {
+      "epoch": 0.9823593708151462,
+      "grad_norm": 0.1424306482076645,
+      "learning_rate": 4.017647867850913e-06,
+      "loss": 0.8755,
+      "step": 135710
+    },
+    {
+      "epoch": 0.9824317574757324,
+      "grad_norm": 0.146108016371727,
+      "learning_rate": 4.017575481190326e-06,
+      "loss": 0.8748,
+      "step": 135720
+    },
+    {
+      "epoch": 0.9825041441363186,
+      "grad_norm": 0.16462048888206482,
+      "learning_rate": 4.017503094529741e-06,
+      "loss": 0.8623,
+      "step": 135730
+    },
+    {
+      "epoch": 0.9825765307969048,
+      "grad_norm": 0.14882566034793854,
+      "learning_rate": 4.0174307078691544e-06,
+      "loss": 0.8583,
+      "step": 135740
+    },
+    {
+      "epoch": 0.982648917457491,
+      "grad_norm": 0.15076982975006104,
+      "learning_rate": 4.017358321208568e-06,
+      "loss": 0.8668,
+      "step": 135750
+    },
+    {
+      "epoch": 0.9827213041180771,
+      "grad_norm": 0.14368361234664917,
+      "learning_rate": 4.017285934547982e-06,
+      "loss": 0.8635,
+      "step": 135760
+    },
+    {
+      "epoch": 0.9827936907786633,
+      "grad_norm": 0.16168615221977234,
+      "learning_rate": 4.017213547887396e-06,
+      "loss": 0.8649,
+      "step": 135770
+    },
+    {
+      "epoch": 0.9828660774392495,
+      "grad_norm": 0.1403905153274536,
+      "learning_rate": 4.01714116122681e-06,
+      "loss": 0.8717,
+      "step": 135780
+    },
+    {
+      "epoch": 0.9829384640998357,
+      "grad_norm": 0.14840847253799438,
+      "learning_rate": 4.017068774566223e-06,
+      "loss": 0.8564,
+      "step": 135790
+    },
+    {
+      "epoch": 0.9830108507604218,
+      "grad_norm": 0.174706369638443,
+      "learning_rate": 4.016996387905637e-06,
+      "loss": 0.8632,
+      "step": 135800
+    },
+    {
+      "epoch": 0.983083237421008,
+      "grad_norm": 0.16106009483337402,
+      "learning_rate": 4.0169240012450515e-06,
+      "loss": 0.8746,
+      "step": 135810
+    },
+    {
+      "epoch": 0.9831556240815943,
+      "grad_norm": 0.15594612061977386,
+      "learning_rate": 4.016851614584465e-06,
+      "loss": 0.8661,
+      "step": 135820
+    },
+    {
+      "epoch": 0.9832280107421805,
+      "grad_norm": 0.14947769045829773,
+      "learning_rate": 4.016779227923879e-06,
+      "loss": 0.8688,
+      "step": 135830
+    },
+    {
+      "epoch": 0.9833003974027666,
+      "grad_norm": 0.15984922647476196,
+      "learning_rate": 4.016706841263292e-06,
+      "loss": 0.8712,
+      "step": 135840
+    },
+    {
+      "epoch": 0.9833727840633528,
+      "grad_norm": 0.1563669741153717,
+      "learning_rate": 4.016634454602707e-06,
+      "loss": 0.8738,
+      "step": 135850
+    },
+    {
+      "epoch": 0.983445170723939,
+      "grad_norm": 0.15901511907577515,
+      "learning_rate": 4.01656206794212e-06,
+      "loss": 0.8672,
+      "step": 135860
+    },
+    {
+      "epoch": 0.9835175573845252,
+      "grad_norm": 0.15088501572608948,
+      "learning_rate": 4.016489681281533e-06,
+      "loss": 0.8752,
+      "step": 135870
+    },
+    {
+      "epoch": 0.9835899440451114,
+      "grad_norm": 0.4412364065647125,
+      "learning_rate": 4.016417294620948e-06,
+      "loss": 0.8594,
+      "step": 135880
+    },
+    {
+      "epoch": 0.9836623307056975,
+      "grad_norm": 0.14504173398017883,
+      "learning_rate": 4.016344907960361e-06,
+      "loss": 0.8658,
+      "step": 135890
+    },
+    {
+      "epoch": 0.9837347173662837,
+      "grad_norm": 0.15376900136470795,
+      "learning_rate": 4.016272521299775e-06,
+      "loss": 0.8752,
+      "step": 135900
+    },
+    {
+      "epoch": 0.9838071040268699,
+      "grad_norm": 0.42392078042030334,
+      "learning_rate": 4.0162001346391885e-06,
+      "loss": 0.8586,
+      "step": 135910
+    },
+    {
+      "epoch": 0.9838794906874562,
+      "grad_norm": 0.17398519814014435,
+      "learning_rate": 4.016127747978603e-06,
+      "loss": 0.8659,
+      "step": 135920
+    },
+    {
+      "epoch": 0.9839518773480423,
+      "grad_norm": 0.15040136873722076,
+      "learning_rate": 4.0160553613180165e-06,
+      "loss": 0.863,
+      "step": 135930
+    },
+    {
+      "epoch": 0.9840242640086285,
+      "grad_norm": 0.15446005761623383,
+      "learning_rate": 4.01598297465743e-06,
+      "loss": 0.8706,
+      "step": 135940
+    },
+    {
+      "epoch": 0.9840966506692147,
+      "grad_norm": 0.14991377294063568,
+      "learning_rate": 4.015910587996844e-06,
+      "loss": 0.8666,
+      "step": 135950
+    },
+    {
+      "epoch": 0.9841690373298009,
+      "grad_norm": 0.15501756966114044,
+      "learning_rate": 4.015838201336258e-06,
+      "loss": 0.867,
+      "step": 135960
+    },
+    {
+      "epoch": 0.984241423990387,
+      "grad_norm": 0.1403217762708664,
+      "learning_rate": 4.015765814675672e-06,
+      "loss": 0.8682,
+      "step": 135970
+    },
+    {
+      "epoch": 0.9843138106509732,
+      "grad_norm": 0.1542348712682724,
+      "learning_rate": 4.0156934280150855e-06,
+      "loss": 0.8656,
+      "step": 135980
+    },
+    {
+      "epoch": 0.9843861973115594,
+      "grad_norm": 0.16103000938892365,
+      "learning_rate": 4.015621041354499e-06,
+      "loss": 0.8642,
+      "step": 135990
+    },
+    {
+      "epoch": 0.9844585839721456,
+      "grad_norm": 0.14834414422512054,
+      "learning_rate": 4.0155486546939136e-06,
+      "loss": 0.8572,
+      "step": 136000
+    },
+    {
+      "epoch": 0.9845309706327318,
+      "grad_norm": 0.16515208780765533,
+      "learning_rate": 4.015476268033327e-06,
+      "loss": 0.8589,
+      "step": 136010
+    },
+    {
+      "epoch": 0.9846033572933179,
+      "grad_norm": 0.16281743347644806,
+      "learning_rate": 4.015403881372741e-06,
+      "loss": 0.867,
+      "step": 136020
+    },
+    {
+      "epoch": 0.9846757439539042,
+      "grad_norm": 0.14419202506542206,
+      "learning_rate": 4.015331494712154e-06,
+      "loss": 0.8527,
+      "step": 136030
+    },
+    {
+      "epoch": 0.9847481306144904,
+      "grad_norm": 0.16212014853954315,
+      "learning_rate": 4.015259108051569e-06,
+      "loss": 0.8629,
+      "step": 136040
+    },
+    {
+      "epoch": 0.9848205172750766,
+      "grad_norm": 0.16474084556102753,
+      "learning_rate": 4.0151867213909825e-06,
+      "loss": 0.8652,
+      "step": 136050
+    },
+    {
+      "epoch": 0.9848929039356628,
+      "grad_norm": 0.19735056161880493,
+      "learning_rate": 4.015114334730396e-06,
+      "loss": 0.8733,
+      "step": 136060
+    },
+    {
+      "epoch": 0.9849652905962489,
+      "grad_norm": 0.164767324924469,
+      "learning_rate": 4.01504194806981e-06,
+      "loss": 0.8641,
+      "step": 136070
+    },
+    {
+      "epoch": 0.9850376772568351,
+      "grad_norm": 0.14573654532432556,
+      "learning_rate": 4.014969561409224e-06,
+      "loss": 0.8648,
+      "step": 136080
+    },
+    {
+      "epoch": 0.9851100639174213,
+      "grad_norm": 0.15756194293498993,
+      "learning_rate": 4.014897174748638e-06,
+      "loss": 0.8658,
+      "step": 136090
+    },
+    {
+      "epoch": 0.9851824505780075,
+      "grad_norm": 0.16832998394966125,
+      "learning_rate": 4.014824788088051e-06,
+      "loss": 0.8561,
+      "step": 136100
+    },
+    {
+      "epoch": 0.9852548372385936,
+      "grad_norm": 0.1867925524711609,
+      "learning_rate": 4.014752401427465e-06,
+      "loss": 0.8723,
+      "step": 136110
+    },
+    {
+      "epoch": 0.9853272238991798,
+      "grad_norm": 0.1441367268562317,
+      "learning_rate": 4.0146800147668795e-06,
+      "loss": 0.8627,
+      "step": 136120
+    },
+    {
+      "epoch": 0.985399610559766,
+      "grad_norm": 0.1538180559873581,
+      "learning_rate": 4.014607628106293e-06,
+      "loss": 0.8647,
+      "step": 136130
+    },
+    {
+      "epoch": 0.9854719972203523,
+      "grad_norm": 0.15871921181678772,
+      "learning_rate": 4.014535241445707e-06,
+      "loss": 0.8555,
+      "step": 136140
+    },
+    {
+      "epoch": 0.9855443838809385,
+      "grad_norm": 0.15353409945964813,
+      "learning_rate": 4.01446285478512e-06,
+      "loss": 0.8575,
+      "step": 136150
+    },
+    {
+      "epoch": 0.9856167705415246,
+      "grad_norm": 0.15181076526641846,
+      "learning_rate": 4.014390468124534e-06,
+      "loss": 0.8714,
+      "step": 136160
+    },
+    {
+      "epoch": 0.9856891572021108,
+      "grad_norm": 0.15089160203933716,
+      "learning_rate": 4.014318081463948e-06,
+      "loss": 0.8763,
+      "step": 136170
+    },
+    {
+      "epoch": 0.985761543862697,
+      "grad_norm": 0.15391786396503448,
+      "learning_rate": 4.014245694803362e-06,
+      "loss": 0.8725,
+      "step": 136180
+    },
+    {
+      "epoch": 0.9858339305232832,
+      "grad_norm": 0.14188186824321747,
+      "learning_rate": 4.014173308142776e-06,
+      "loss": 0.8615,
+      "step": 136190
+    },
+    {
+      "epoch": 0.9859063171838693,
+      "grad_norm": 0.16253648698329926,
+      "learning_rate": 4.014100921482189e-06,
+      "loss": 0.8683,
+      "step": 136200
+    },
+    {
+      "epoch": 0.9859787038444555,
+      "grad_norm": 0.154387429356575,
+      "learning_rate": 4.014028534821604e-06,
+      "loss": 0.859,
+      "step": 136210
+    },
+    {
+      "epoch": 0.9860510905050417,
+      "grad_norm": 0.15761414170265198,
+      "learning_rate": 4.013956148161017e-06,
+      "loss": 0.8724,
+      "step": 136220
+    },
+    {
+      "epoch": 0.9861234771656279,
+      "grad_norm": 0.14925530552864075,
+      "learning_rate": 4.013883761500431e-06,
+      "loss": 0.8658,
+      "step": 136230
+    },
+    {
+      "epoch": 0.9861958638262142,
+      "grad_norm": 0.15609656274318695,
+      "learning_rate": 4.013811374839845e-06,
+      "loss": 0.8684,
+      "step": 136240
+    },
+    {
+      "epoch": 0.9862682504868003,
+      "grad_norm": 0.1742173135280609,
+      "learning_rate": 4.013738988179259e-06,
+      "loss": 0.8502,
+      "step": 136250
+    },
+    {
+      "epoch": 0.9863406371473865,
+      "grad_norm": 0.15762127935886383,
+      "learning_rate": 4.013666601518673e-06,
+      "loss": 0.8603,
+      "step": 136260
+    },
+    {
+      "epoch": 0.9864130238079727,
+      "grad_norm": 0.178837388753891,
+      "learning_rate": 4.013594214858086e-06,
+      "loss": 0.8645,
+      "step": 136270
+    },
+    {
+      "epoch": 0.9864854104685589,
+      "grad_norm": 0.16362306475639343,
+      "learning_rate": 4.0135218281975e-06,
+      "loss": 0.8626,
+      "step": 136280
+    },
+    {
+      "epoch": 0.986557797129145,
+      "grad_norm": 0.14631548523902893,
+      "learning_rate": 4.013449441536914e-06,
+      "loss": 0.8687,
+      "step": 136290
+    },
+    {
+      "epoch": 0.9866301837897312,
+      "grad_norm": 0.14602495729923248,
+      "learning_rate": 4.013377054876328e-06,
+      "loss": 0.8654,
+      "step": 136300
+    },
+    {
+      "epoch": 0.9867025704503174,
+      "grad_norm": 0.15310421586036682,
+      "learning_rate": 4.013304668215742e-06,
+      "loss": 0.8742,
+      "step": 136310
+    },
+    {
+      "epoch": 0.9867749571109036,
+      "grad_norm": 0.14649897813796997,
+      "learning_rate": 4.013232281555155e-06,
+      "loss": 0.8571,
+      "step": 136320
+    },
+    {
+      "epoch": 0.9868473437714897,
+      "grad_norm": 0.15480268001556396,
+      "learning_rate": 4.01315989489457e-06,
+      "loss": 0.8737,
+      "step": 136330
+    },
+    {
+      "epoch": 0.9869197304320759,
+      "grad_norm": 0.161812886595726,
+      "learning_rate": 4.013087508233983e-06,
+      "loss": 0.8601,
+      "step": 136340
+    },
+    {
+      "epoch": 0.9869921170926622,
+      "grad_norm": 0.17308774590492249,
+      "learning_rate": 4.013015121573397e-06,
+      "loss": 0.8818,
+      "step": 136350
+    },
+    {
+      "epoch": 0.9870645037532484,
+      "grad_norm": 0.14882414042949677,
+      "learning_rate": 4.0129427349128105e-06,
+      "loss": 0.8744,
+      "step": 136360
+    },
+    {
+      "epoch": 0.9871368904138346,
+      "grad_norm": 0.14824624359607697,
+      "learning_rate": 4.012870348252225e-06,
+      "loss": 0.8686,
+      "step": 136370
+    },
+    {
+      "epoch": 0.9872092770744207,
+      "grad_norm": 0.16246159374713898,
+      "learning_rate": 4.012797961591639e-06,
+      "loss": 0.8634,
+      "step": 136380
+    },
+    {
+      "epoch": 0.9872816637350069,
+      "grad_norm": 0.17609475553035736,
+      "learning_rate": 4.012725574931052e-06,
+      "loss": 0.8657,
+      "step": 136390
+    },
+    {
+      "epoch": 0.9873540503955931,
+      "grad_norm": 0.14540891349315643,
+      "learning_rate": 4.012653188270466e-06,
+      "loss": 0.8443,
+      "step": 136400
+    },
+    {
+      "epoch": 0.9874264370561793,
+      "grad_norm": 0.15471509099006653,
+      "learning_rate": 4.0125808016098794e-06,
+      "loss": 0.8676,
+      "step": 136410
+    },
+    {
+      "epoch": 0.9874988237167655,
+      "grad_norm": 0.15584422647953033,
+      "learning_rate": 4.012508414949293e-06,
+      "loss": 0.8639,
+      "step": 136420
+    },
+    {
+      "epoch": 0.9875712103773516,
+      "grad_norm": 0.1576376110315323,
+      "learning_rate": 4.012436028288707e-06,
+      "loss": 0.8676,
+      "step": 136430
+    },
+    {
+      "epoch": 0.9876435970379378,
+      "grad_norm": 0.19695936143398285,
+      "learning_rate": 4.012363641628121e-06,
+      "loss": 0.8697,
+      "step": 136440
+    },
+    {
+      "epoch": 0.9877159836985241,
+      "grad_norm": 0.1736014187335968,
+      "learning_rate": 4.012291254967535e-06,
+      "loss": 0.8727,
+      "step": 136450
+    },
+    {
+      "epoch": 0.9877883703591103,
+      "grad_norm": 0.14788353443145752,
+      "learning_rate": 4.012218868306948e-06,
+      "loss": 0.8549,
+      "step": 136460
+    },
+    {
+      "epoch": 0.9878607570196964,
+      "grad_norm": 0.16188114881515503,
+      "learning_rate": 4.012146481646362e-06,
+      "loss": 0.8411,
+      "step": 136470
+    },
+    {
+      "epoch": 0.9879331436802826,
+      "grad_norm": 0.14878244698047638,
+      "learning_rate": 4.0120740949857764e-06,
+      "loss": 0.8626,
+      "step": 136480
+    },
+    {
+      "epoch": 0.9880055303408688,
+      "grad_norm": 0.16054673492908478,
+      "learning_rate": 4.01200170832519e-06,
+      "loss": 0.8722,
+      "step": 136490
+    },
+    {
+      "epoch": 0.988077917001455,
+      "grad_norm": 0.18446345627307892,
+      "learning_rate": 4.011929321664604e-06,
+      "loss": 0.8663,
+      "step": 136500
+    },
+    {
+      "epoch": 0.9881503036620412,
+      "grad_norm": 0.15131662786006927,
+      "learning_rate": 4.011856935004017e-06,
+      "loss": 0.8707,
+      "step": 136510
+    },
+    {
+      "epoch": 0.9882226903226273,
+      "grad_norm": 0.15739837288856506,
+      "learning_rate": 4.011784548343432e-06,
+      "loss": 0.8606,
+      "step": 136520
+    },
+    {
+      "epoch": 0.9882950769832135,
+      "grad_norm": 0.17093366384506226,
+      "learning_rate": 4.011712161682845e-06,
+      "loss": 0.8555,
+      "step": 136530
+    },
+    {
+      "epoch": 0.9883674636437997,
+      "grad_norm": 0.1519378423690796,
+      "learning_rate": 4.011639775022259e-06,
+      "loss": 0.8563,
+      "step": 136540
+    },
+    {
+      "epoch": 0.9884398503043859,
+      "grad_norm": 0.13930349051952362,
+      "learning_rate": 4.011567388361673e-06,
+      "loss": 0.8667,
+      "step": 136550
+    },
+    {
+      "epoch": 0.9885122369649721,
+      "grad_norm": 0.1486721783876419,
+      "learning_rate": 4.011495001701087e-06,
+      "loss": 0.8662,
+      "step": 136560
+    },
+    {
+      "epoch": 0.9885846236255583,
+      "grad_norm": 0.14712847769260406,
+      "learning_rate": 4.011422615040501e-06,
+      "loss": 0.8593,
+      "step": 136570
+    },
+    {
+      "epoch": 0.9886570102861445,
+      "grad_norm": 0.15622474253177643,
+      "learning_rate": 4.011350228379914e-06,
+      "loss": 0.8692,
+      "step": 136580
+    },
+    {
+      "epoch": 0.9887293969467307,
+      "grad_norm": 0.15203312039375305,
+      "learning_rate": 4.011277841719328e-06,
+      "loss": 0.8552,
+      "step": 136590
+    },
+    {
+      "epoch": 0.9888017836073169,
+      "grad_norm": 0.15190285444259644,
+      "learning_rate": 4.011205455058742e-06,
+      "loss": 0.8699,
+      "step": 136600
+    },
+    {
+      "epoch": 0.988874170267903,
+      "grad_norm": 0.14632107317447662,
+      "learning_rate": 4.011133068398156e-06,
+      "loss": 0.8657,
+      "step": 136610
+    },
+    {
+      "epoch": 0.9889465569284892,
+      "grad_norm": 0.16472133994102478,
+      "learning_rate": 4.01106068173757e-06,
+      "loss": 0.8669,
+      "step": 136620
+    },
+    {
+      "epoch": 0.9890189435890754,
+      "grad_norm": 0.17338450253009796,
+      "learning_rate": 4.010988295076983e-06,
+      "loss": 0.8675,
+      "step": 136630
+    },
+    {
+      "epoch": 0.9890913302496616,
+      "grad_norm": 0.1531408429145813,
+      "learning_rate": 4.010915908416398e-06,
+      "loss": 0.8706,
+      "step": 136640
+    },
+    {
+      "epoch": 0.9891637169102477,
+      "grad_norm": 0.2573656141757965,
+      "learning_rate": 4.010843521755811e-06,
+      "loss": 0.8592,
+      "step": 136650
+    },
+    {
+      "epoch": 0.9892361035708339,
+      "grad_norm": 0.17421439290046692,
+      "learning_rate": 4.010771135095225e-06,
+      "loss": 0.8501,
+      "step": 136660
+    },
+    {
+      "epoch": 0.9893084902314202,
+      "grad_norm": 0.1428343653678894,
+      "learning_rate": 4.0106987484346385e-06,
+      "loss": 0.8451,
+      "step": 136670
+    },
+    {
+      "epoch": 0.9893808768920064,
+      "grad_norm": 0.1828918159008026,
+      "learning_rate": 4.010626361774053e-06,
+      "loss": 0.8663,
+      "step": 136680
+    },
+    {
+      "epoch": 0.9894532635525926,
+      "grad_norm": 0.14895032346248627,
+      "learning_rate": 4.010553975113467e-06,
+      "loss": 0.8611,
+      "step": 136690
+    },
+    {
+      "epoch": 0.9895256502131787,
+      "grad_norm": 0.17357683181762695,
+      "learning_rate": 4.01048158845288e-06,
+      "loss": 0.8655,
+      "step": 136700
+    },
+    {
+      "epoch": 0.9895980368737649,
+      "grad_norm": 0.1530769318342209,
+      "learning_rate": 4.010409201792294e-06,
+      "loss": 0.852,
+      "step": 136710
+    },
+    {
+      "epoch": 0.9896704235343511,
+      "grad_norm": 0.14855614304542542,
+      "learning_rate": 4.010336815131708e-06,
+      "loss": 0.8639,
+      "step": 136720
+    },
+    {
+      "epoch": 0.9897428101949373,
+      "grad_norm": 0.15136641263961792,
+      "learning_rate": 4.010264428471122e-06,
+      "loss": 0.8603,
+      "step": 136730
+    },
+    {
+      "epoch": 0.9898151968555234,
+      "grad_norm": 0.15471212565898895,
+      "learning_rate": 4.0101920418105356e-06,
+      "loss": 0.8762,
+      "step": 136740
+    },
+    {
+      "epoch": 0.9898875835161096,
+      "grad_norm": 0.15268218517303467,
+      "learning_rate": 4.010119655149949e-06,
+      "loss": 0.8656,
+      "step": 136750
+    },
+    {
+      "epoch": 0.9899599701766958,
+      "grad_norm": 0.15985944867134094,
+      "learning_rate": 4.010047268489364e-06,
+      "loss": 0.8624,
+      "step": 136760
+    },
+    {
+      "epoch": 0.9900323568372821,
+      "grad_norm": 0.1638825535774231,
+      "learning_rate": 4.009974881828777e-06,
+      "loss": 0.8669,
+      "step": 136770
+    },
+    {
+      "epoch": 0.9901047434978683,
+      "grad_norm": 0.15115493535995483,
+      "learning_rate": 4.009902495168191e-06,
+      "loss": 0.8826,
+      "step": 136780
+    },
+    {
+      "epoch": 0.9901771301584544,
+      "grad_norm": 0.14795368909835815,
+      "learning_rate": 4.0098301085076045e-06,
+      "loss": 0.8596,
+      "step": 136790
+    },
+    {
+      "epoch": 0.9902495168190406,
+      "grad_norm": 0.17082315683364868,
+      "learning_rate": 4.009757721847018e-06,
+      "loss": 0.8785,
+      "step": 136800
+    },
+    {
+      "epoch": 0.9903219034796268,
+      "grad_norm": 0.1484622359275818,
+      "learning_rate": 4.0096853351864326e-06,
+      "loss": 0.8675,
+      "step": 136810
+    },
+    {
+      "epoch": 0.990394290140213,
+      "grad_norm": 0.16692779958248138,
+      "learning_rate": 4.009612948525846e-06,
+      "loss": 0.8716,
+      "step": 136820
+    },
+    {
+      "epoch": 0.9904666768007991,
+      "grad_norm": 0.16581696271896362,
+      "learning_rate": 4.00954056186526e-06,
+      "loss": 0.8664,
+      "step": 136830
+    },
+    {
+      "epoch": 0.9905390634613853,
+      "grad_norm": 0.15264694392681122,
+      "learning_rate": 4.009468175204673e-06,
+      "loss": 0.8654,
+      "step": 136840
+    },
+    {
+      "epoch": 0.9906114501219715,
+      "grad_norm": 0.17103955149650574,
+      "learning_rate": 4.009395788544088e-06,
+      "loss": 0.8827,
+      "step": 136850
+    },
+    {
+      "epoch": 0.9906838367825577,
+      "grad_norm": 0.15730604529380798,
+      "learning_rate": 4.0093234018835015e-06,
+      "loss": 0.8646,
+      "step": 136860
+    },
+    {
+      "epoch": 0.9907562234431438,
+      "grad_norm": 0.1455976516008377,
+      "learning_rate": 4.009251015222915e-06,
+      "loss": 0.8596,
+      "step": 136870
+    },
+    {
+      "epoch": 0.9908286101037301,
+      "grad_norm": 0.1492074877023697,
+      "learning_rate": 4.009178628562329e-06,
+      "loss": 0.8698,
+      "step": 136880
+    },
+    {
+      "epoch": 0.9909009967643163,
+      "grad_norm": 0.14965695142745972,
+      "learning_rate": 4.009106241901743e-06,
+      "loss": 0.867,
+      "step": 136890
+    },
+    {
+      "epoch": 0.9909733834249025,
+      "grad_norm": 0.19759173691272736,
+      "learning_rate": 4.009033855241157e-06,
+      "loss": 0.8611,
+      "step": 136900
+    },
+    {
+      "epoch": 0.9910457700854887,
+      "grad_norm": 0.17178118228912354,
+      "learning_rate": 4.00896146858057e-06,
+      "loss": 0.8834,
+      "step": 136910
+    },
+    {
+      "epoch": 0.9911181567460748,
+      "grad_norm": 0.1561564952135086,
+      "learning_rate": 4.008889081919984e-06,
+      "loss": 0.8574,
+      "step": 136920
+    },
+    {
+      "epoch": 0.991190543406661,
+      "grad_norm": 0.16888576745986938,
+      "learning_rate": 4.008816695259398e-06,
+      "loss": 0.8651,
+      "step": 136930
+    },
+    {
+      "epoch": 0.9912629300672472,
+      "grad_norm": 0.1438201516866684,
+      "learning_rate": 4.008744308598811e-06,
+      "loss": 0.8731,
+      "step": 136940
+    },
+    {
+      "epoch": 0.9913353167278334,
+      "grad_norm": 0.20436105132102966,
+      "learning_rate": 4.008671921938225e-06,
+      "loss": 0.8619,
+      "step": 136950
+    },
+    {
+      "epoch": 0.9914077033884195,
+      "grad_norm": 0.15767519176006317,
+      "learning_rate": 4.008599535277639e-06,
+      "loss": 0.8681,
+      "step": 136960
+    },
+    {
+      "epoch": 0.9914800900490057,
+      "grad_norm": 0.14873795211315155,
+      "learning_rate": 4.008527148617053e-06,
+      "loss": 0.8852,
+      "step": 136970
+    },
+    {
+      "epoch": 0.9915524767095919,
+      "grad_norm": 0.1454266458749771,
+      "learning_rate": 4.0084547619564666e-06,
+      "loss": 0.8646,
+      "step": 136980
+    },
+    {
+      "epoch": 0.9916248633701782,
+      "grad_norm": 0.14362385869026184,
+      "learning_rate": 4.00838237529588e-06,
+      "loss": 0.8638,
+      "step": 136990
+    },
+    {
+      "epoch": 0.9916972500307644,
+      "grad_norm": 0.2043345868587494,
+      "learning_rate": 4.008309988635295e-06,
+      "loss": 0.8635,
+      "step": 137000
+    },
+    {
+      "epoch": 0.9917696366913505,
+      "grad_norm": 0.15229232609272003,
+      "learning_rate": 4.008237601974708e-06,
+      "loss": 0.8684,
+      "step": 137010
+    },
+    {
+      "epoch": 0.9918420233519367,
+      "grad_norm": 0.17375898361206055,
+      "learning_rate": 4.008165215314122e-06,
+      "loss": 0.8571,
+      "step": 137020
+    },
+    {
+      "epoch": 0.9919144100125229,
+      "grad_norm": 0.16473175585269928,
+      "learning_rate": 4.0080928286535355e-06,
+      "loss": 0.8716,
+      "step": 137030
+    },
+    {
+      "epoch": 0.9919867966731091,
+      "grad_norm": 0.1511010229587555,
+      "learning_rate": 4.00802044199295e-06,
+      "loss": 0.8704,
+      "step": 137040
+    },
+    {
+      "epoch": 0.9920591833336952,
+      "grad_norm": 0.1537337750196457,
+      "learning_rate": 4.007948055332364e-06,
+      "loss": 0.8637,
+      "step": 137050
+    },
+    {
+      "epoch": 0.9921315699942814,
+      "grad_norm": 0.46637365221977234,
+      "learning_rate": 4.007875668671777e-06,
+      "loss": 0.8569,
+      "step": 137060
+    },
+    {
+      "epoch": 0.9922039566548676,
+      "grad_norm": 0.1638781875371933,
+      "learning_rate": 4.007803282011191e-06,
+      "loss": 0.8521,
+      "step": 137070
+    },
+    {
+      "epoch": 0.9922763433154538,
+      "grad_norm": 0.1593899428844452,
+      "learning_rate": 4.007730895350605e-06,
+      "loss": 0.8755,
+      "step": 137080
+    },
+    {
+      "epoch": 0.9923487299760401,
+      "grad_norm": 0.15153998136520386,
+      "learning_rate": 4.007658508690019e-06,
+      "loss": 0.864,
+      "step": 137090
+    },
+    {
+      "epoch": 0.9924211166366262,
+      "grad_norm": 0.15272100269794464,
+      "learning_rate": 4.0075861220294325e-06,
+      "loss": 0.8598,
+      "step": 137100
+    },
+    {
+      "epoch": 0.9924935032972124,
+      "grad_norm": 0.1559947431087494,
+      "learning_rate": 4.007513735368846e-06,
+      "loss": 0.8733,
+      "step": 137110
+    },
+    {
+      "epoch": 0.9925658899577986,
+      "grad_norm": 0.15066994726657867,
+      "learning_rate": 4.007441348708261e-06,
+      "loss": 0.8694,
+      "step": 137120
+    },
+    {
+      "epoch": 0.9926382766183848,
+      "grad_norm": 0.14346741139888763,
+      "learning_rate": 4.007368962047674e-06,
+      "loss": 0.8621,
+      "step": 137130
+    },
+    {
+      "epoch": 0.992710663278971,
+      "grad_norm": 0.14388960599899292,
+      "learning_rate": 4.007296575387088e-06,
+      "loss": 0.863,
+      "step": 137140
+    },
+    {
+      "epoch": 0.9927830499395571,
+      "grad_norm": 0.14946885406970978,
+      "learning_rate": 4.0072241887265014e-06,
+      "loss": 0.8577,
+      "step": 137150
+    },
+    {
+      "epoch": 0.9928554366001433,
+      "grad_norm": 0.15995468199253082,
+      "learning_rate": 4.007151802065916e-06,
+      "loss": 0.8613,
+      "step": 137160
+    },
+    {
+      "epoch": 0.9929278232607295,
+      "grad_norm": 0.15818224847316742,
+      "learning_rate": 4.0070794154053295e-06,
+      "loss": 0.871,
+      "step": 137170
+    },
+    {
+      "epoch": 0.9930002099213157,
+      "grad_norm": 0.15521134436130524,
+      "learning_rate": 4.007007028744743e-06,
+      "loss": 0.8568,
+      "step": 137180
+    },
+    {
+      "epoch": 0.9930725965819018,
+      "grad_norm": 0.15648885071277618,
+      "learning_rate": 4.006934642084157e-06,
+      "loss": 0.8688,
+      "step": 137190
+    },
+    {
+      "epoch": 0.9931449832424881,
+      "grad_norm": 0.15690802037715912,
+      "learning_rate": 4.006862255423571e-06,
+      "loss": 0.8696,
+      "step": 137200
+    },
+    {
+      "epoch": 0.9932173699030743,
+      "grad_norm": 0.17651623487472534,
+      "learning_rate": 4.006789868762985e-06,
+      "loss": 0.858,
+      "step": 137210
+    },
+    {
+      "epoch": 0.9932897565636605,
+      "grad_norm": 0.1453973650932312,
+      "learning_rate": 4.0067174821023984e-06,
+      "loss": 0.8623,
+      "step": 137220
+    },
+    {
+      "epoch": 0.9933621432242467,
+      "grad_norm": 0.16270014643669128,
+      "learning_rate": 4.006645095441812e-06,
+      "loss": 0.8702,
+      "step": 137230
+    },
+    {
+      "epoch": 0.9934345298848328,
+      "grad_norm": 0.2834886610507965,
+      "learning_rate": 4.0065727087812265e-06,
+      "loss": 0.8558,
+      "step": 137240
+    },
+    {
+      "epoch": 0.993506916545419,
+      "grad_norm": 0.1739642173051834,
+      "learning_rate": 4.00650032212064e-06,
+      "loss": 0.8631,
+      "step": 137250
+    },
+    {
+      "epoch": 0.9935793032060052,
+      "grad_norm": 0.16178597509860992,
+      "learning_rate": 4.006427935460054e-06,
+      "loss": 0.8616,
+      "step": 137260
+    },
+    {
+      "epoch": 0.9936516898665914,
+      "grad_norm": 0.1678241342306137,
+      "learning_rate": 4.006355548799467e-06,
+      "loss": 0.861,
+      "step": 137270
+    },
+    {
+      "epoch": 0.9937240765271775,
+      "grad_norm": 0.15866810083389282,
+      "learning_rate": 4.006283162138882e-06,
+      "loss": 0.8549,
+      "step": 137280
+    },
+    {
+      "epoch": 0.9937964631877637,
+      "grad_norm": 0.14100779592990875,
+      "learning_rate": 4.0062107754782955e-06,
+      "loss": 0.8714,
+      "step": 137290
+    },
+    {
+      "epoch": 0.99386884984835,
+      "grad_norm": 0.1390538066625595,
+      "learning_rate": 4.006138388817709e-06,
+      "loss": 0.8686,
+      "step": 137300
+    },
+    {
+      "epoch": 0.9939412365089362,
+      "grad_norm": 0.1549810916185379,
+      "learning_rate": 4.006066002157123e-06,
+      "loss": 0.8599,
+      "step": 137310
+    },
+    {
+      "epoch": 0.9940136231695224,
+      "grad_norm": 0.14956602454185486,
+      "learning_rate": 4.005993615496537e-06,
+      "loss": 0.8582,
+      "step": 137320
+    },
+    {
+      "epoch": 0.9940860098301085,
+      "grad_norm": 0.15790285170078278,
+      "learning_rate": 4.005921228835951e-06,
+      "loss": 0.8617,
+      "step": 137330
+    },
+    {
+      "epoch": 0.9941583964906947,
+      "grad_norm": 0.153904527425766,
+      "learning_rate": 4.005848842175364e-06,
+      "loss": 0.873,
+      "step": 137340
+    },
+    {
+      "epoch": 0.9942307831512809,
+      "grad_norm": 0.15151570737361908,
+      "learning_rate": 4.005776455514778e-06,
+      "loss": 0.8688,
+      "step": 137350
+    },
+    {
+      "epoch": 0.9943031698118671,
+      "grad_norm": 0.15569916367530823,
+      "learning_rate": 4.0057040688541925e-06,
+      "loss": 0.8518,
+      "step": 137360
+    },
+    {
+      "epoch": 0.9943755564724532,
+      "grad_norm": 0.14543747901916504,
+      "learning_rate": 4.005631682193606e-06,
+      "loss": 0.8739,
+      "step": 137370
+    },
+    {
+      "epoch": 0.9944479431330394,
+      "grad_norm": 0.16408121585845947,
+      "learning_rate": 4.00555929553302e-06,
+      "loss": 0.868,
+      "step": 137380
+    },
+    {
+      "epoch": 0.9945203297936256,
+      "grad_norm": 0.1823720484972,
+      "learning_rate": 4.005486908872433e-06,
+      "loss": 0.8547,
+      "step": 137390
+    },
+    {
+      "epoch": 0.9945927164542118,
+      "grad_norm": 0.15718266367912292,
+      "learning_rate": 4.005414522211847e-06,
+      "loss": 0.8677,
+      "step": 137400
+    },
+    {
+      "epoch": 0.994665103114798,
+      "grad_norm": 0.14774106442928314,
+      "learning_rate": 4.005342135551261e-06,
+      "loss": 0.8584,
+      "step": 137410
+    },
+    {
+      "epoch": 0.9947374897753842,
+      "grad_norm": 0.14286896586418152,
+      "learning_rate": 4.005269748890675e-06,
+      "loss": 0.8526,
+      "step": 137420
+    },
+    {
+      "epoch": 0.9948098764359704,
+      "grad_norm": 0.17574062943458557,
+      "learning_rate": 4.005197362230089e-06,
+      "loss": 0.8604,
+      "step": 137430
+    },
+    {
+      "epoch": 0.9948822630965566,
+      "grad_norm": 0.15135468542575836,
+      "learning_rate": 4.005124975569502e-06,
+      "loss": 0.8667,
+      "step": 137440
+    },
+    {
+      "epoch": 0.9949546497571428,
+      "grad_norm": 0.16362632811069489,
+      "learning_rate": 4.005052588908917e-06,
+      "loss": 0.8574,
+      "step": 137450
+    },
+    {
+      "epoch": 0.9950270364177289,
+      "grad_norm": 0.19329750537872314,
+      "learning_rate": 4.0049802022483295e-06,
+      "loss": 0.8783,
+      "step": 137460
+    },
+    {
+      "epoch": 0.9950994230783151,
+      "grad_norm": 0.1522989422082901,
+      "learning_rate": 4.004907815587744e-06,
+      "loss": 0.8557,
+      "step": 137470
+    },
+    {
+      "epoch": 0.9951718097389013,
+      "grad_norm": 0.17292699217796326,
+      "learning_rate": 4.0048354289271576e-06,
+      "loss": 0.8522,
+      "step": 137480
+    },
+    {
+      "epoch": 0.9952441963994875,
+      "grad_norm": 0.1753862351179123,
+      "learning_rate": 4.004763042266571e-06,
+      "loss": 0.8537,
+      "step": 137490
+    },
+    {
+      "epoch": 0.9953165830600736,
+      "grad_norm": 0.1441674530506134,
+      "learning_rate": 4.004690655605985e-06,
+      "loss": 0.8697,
+      "step": 137500
+    },
+    {
+      "epoch": 0.9953889697206598,
+      "grad_norm": 0.17849737405776978,
+      "learning_rate": 4.004618268945399e-06,
+      "loss": 0.8556,
+      "step": 137510
+    },
+    {
+      "epoch": 0.9954613563812461,
+      "grad_norm": 0.1520448625087738,
+      "learning_rate": 4.004545882284813e-06,
+      "loss": 0.8633,
+      "step": 137520
+    },
+    {
+      "epoch": 0.9955337430418323,
+      "grad_norm": 0.1547725945711136,
+      "learning_rate": 4.0044734956242265e-06,
+      "loss": 0.86,
+      "step": 137530
+    },
+    {
+      "epoch": 0.9956061297024185,
+      "grad_norm": 0.17297573387622833,
+      "learning_rate": 4.00440110896364e-06,
+      "loss": 0.8737,
+      "step": 137540
+    },
+    {
+      "epoch": 0.9956785163630046,
+      "grad_norm": 0.15290264785289764,
+      "learning_rate": 4.0043287223030546e-06,
+      "loss": 0.8615,
+      "step": 137550
+    },
+    {
+      "epoch": 0.9957509030235908,
+      "grad_norm": 0.15568116307258606,
+      "learning_rate": 4.004256335642468e-06,
+      "loss": 0.8603,
+      "step": 137560
+    },
+    {
+      "epoch": 0.995823289684177,
+      "grad_norm": 0.1544651985168457,
+      "learning_rate": 4.004183948981882e-06,
+      "loss": 0.8653,
+      "step": 137570
+    },
+    {
+      "epoch": 0.9958956763447632,
+      "grad_norm": 0.15817542374134064,
+      "learning_rate": 4.004111562321295e-06,
+      "loss": 0.8681,
+      "step": 137580
+    },
+    {
+      "epoch": 0.9959680630053493,
+      "grad_norm": 0.15345756709575653,
+      "learning_rate": 4.004039175660709e-06,
+      "loss": 0.8569,
+      "step": 137590
+    },
+    {
+      "epoch": 0.9960404496659355,
+      "grad_norm": 0.18895161151885986,
+      "learning_rate": 4.0039667890001235e-06,
+      "loss": 0.8733,
+      "step": 137600
+    },
+    {
+      "epoch": 0.9961128363265217,
+      "grad_norm": 0.1601308286190033,
+      "learning_rate": 4.003894402339537e-06,
+      "loss": 0.8627,
+      "step": 137610
+    },
+    {
+      "epoch": 0.996185222987108,
+      "grad_norm": 0.15279871225357056,
+      "learning_rate": 4.003822015678951e-06,
+      "loss": 0.8684,
+      "step": 137620
+    },
+    {
+      "epoch": 0.9962576096476942,
+      "grad_norm": 0.15466183423995972,
+      "learning_rate": 4.003749629018364e-06,
+      "loss": 0.8547,
+      "step": 137630
+    },
+    {
+      "epoch": 0.9963299963082803,
+      "grad_norm": 0.14806103706359863,
+      "learning_rate": 4.003677242357779e-06,
+      "loss": 0.8609,
+      "step": 137640
+    },
+    {
+      "epoch": 0.9964023829688665,
+      "grad_norm": 0.14731131494045258,
+      "learning_rate": 4.003604855697192e-06,
+      "loss": 0.8463,
+      "step": 137650
+    },
+    {
+      "epoch": 0.9964747696294527,
+      "grad_norm": 0.16127149760723114,
+      "learning_rate": 4.003532469036606e-06,
+      "loss": 0.8709,
+      "step": 137660
+    },
+    {
+      "epoch": 0.9965471562900389,
+      "grad_norm": 0.1602654606103897,
+      "learning_rate": 4.00346008237602e-06,
+      "loss": 0.8626,
+      "step": 137670
+    },
+    {
+      "epoch": 0.996619542950625,
+      "grad_norm": 0.18218478560447693,
+      "learning_rate": 4.003387695715434e-06,
+      "loss": 0.8643,
+      "step": 137680
+    },
+    {
+      "epoch": 0.9966919296112112,
+      "grad_norm": 0.19056251645088196,
+      "learning_rate": 4.003315309054848e-06,
+      "loss": 0.8618,
+      "step": 137690
+    },
+    {
+      "epoch": 0.9967643162717974,
+      "grad_norm": 0.15740570425987244,
+      "learning_rate": 4.003242922394261e-06,
+      "loss": 0.8707,
+      "step": 137700
+    },
+    {
+      "epoch": 0.9968367029323836,
+      "grad_norm": 0.15710806846618652,
+      "learning_rate": 4.003170535733675e-06,
+      "loss": 0.8691,
+      "step": 137710
+    },
+    {
+      "epoch": 0.9969090895929698,
+      "grad_norm": 0.16603271663188934,
+      "learning_rate": 4.003098149073089e-06,
+      "loss": 0.8569,
+      "step": 137720
+    },
+    {
+      "epoch": 0.996981476253556,
+      "grad_norm": 0.14341507852077484,
+      "learning_rate": 4.003025762412503e-06,
+      "loss": 0.8617,
+      "step": 137730
+    },
+    {
+      "epoch": 0.9970538629141422,
+      "grad_norm": 0.1590244024991989,
+      "learning_rate": 4.002953375751917e-06,
+      "loss": 0.8796,
+      "step": 137740
+    },
+    {
+      "epoch": 0.9971262495747284,
+      "grad_norm": 0.5351874232292175,
+      "learning_rate": 4.00288098909133e-06,
+      "loss": 0.8664,
+      "step": 137750
+    },
+    {
+      "epoch": 0.9971986362353146,
+      "grad_norm": 0.15063074231147766,
+      "learning_rate": 4.002808602430745e-06,
+      "loss": 0.862,
+      "step": 137760
+    },
+    {
+      "epoch": 0.9972710228959007,
+      "grad_norm": 0.18847909569740295,
+      "learning_rate": 4.002736215770158e-06,
+      "loss": 0.8713,
+      "step": 137770
+    },
+    {
+      "epoch": 0.9973434095564869,
+      "grad_norm": 0.1513339728116989,
+      "learning_rate": 4.002663829109572e-06,
+      "loss": 0.8714,
+      "step": 137780
+    },
+    {
+      "epoch": 0.9974157962170731,
+      "grad_norm": 0.14384783804416656,
+      "learning_rate": 4.002591442448986e-06,
+      "loss": 0.8723,
+      "step": 137790
+    },
+    {
+      "epoch": 0.9974881828776593,
+      "grad_norm": 0.1539144068956375,
+      "learning_rate": 4.0025190557884e-06,
+      "loss": 0.8618,
+      "step": 137800
+    },
+    {
+      "epoch": 0.9975605695382455,
+      "grad_norm": 0.1452489197254181,
+      "learning_rate": 4.002446669127814e-06,
+      "loss": 0.8714,
+      "step": 137810
+    },
+    {
+      "epoch": 0.9976329561988316,
+      "grad_norm": 0.15493297576904297,
+      "learning_rate": 4.002374282467227e-06,
+      "loss": 0.8562,
+      "step": 137820
+    },
+    {
+      "epoch": 0.9977053428594179,
+      "grad_norm": 0.154521182179451,
+      "learning_rate": 4.002301895806641e-06,
+      "loss": 0.8663,
+      "step": 137830
+    },
+    {
+      "epoch": 0.9977777295200041,
+      "grad_norm": 0.14628452062606812,
+      "learning_rate": 4.002229509146055e-06,
+      "loss": 0.8623,
+      "step": 137840
+    },
+    {
+      "epoch": 0.9978501161805903,
+      "grad_norm": 0.14919553697109222,
+      "learning_rate": 4.002157122485469e-06,
+      "loss": 0.8685,
+      "step": 137850
+    },
+    {
+      "epoch": 0.9979225028411765,
+      "grad_norm": 0.1762702614068985,
+      "learning_rate": 4.002084735824883e-06,
+      "loss": 0.8657,
+      "step": 137860
+    },
+    {
+      "epoch": 0.9979948895017626,
+      "grad_norm": 0.151996448636055,
+      "learning_rate": 4.002012349164296e-06,
+      "loss": 0.8568,
+      "step": 137870
+    },
+    {
+      "epoch": 0.9980672761623488,
+      "grad_norm": 0.15440037846565247,
+      "learning_rate": 4.001939962503711e-06,
+      "loss": 0.852,
+      "step": 137880
+    },
+    {
+      "epoch": 0.998139662822935,
+      "grad_norm": 0.16397789120674133,
+      "learning_rate": 4.001867575843124e-06,
+      "loss": 0.854,
+      "step": 137890
+    },
+    {
+      "epoch": 0.9982120494835212,
+      "grad_norm": 0.19279788434505463,
+      "learning_rate": 4.001795189182538e-06,
+      "loss": 0.8581,
+      "step": 137900
+    },
+    {
+      "epoch": 0.9982844361441073,
+      "grad_norm": 0.15897057950496674,
+      "learning_rate": 4.0017228025219515e-06,
+      "loss": 0.8711,
+      "step": 137910
+    },
+    {
+      "epoch": 0.9983568228046935,
+      "grad_norm": 0.1570604145526886,
+      "learning_rate": 4.001650415861366e-06,
+      "loss": 0.8645,
+      "step": 137920
+    },
+    {
+      "epoch": 0.9984292094652797,
+      "grad_norm": 0.16585461795330048,
+      "learning_rate": 4.00157802920078e-06,
+      "loss": 0.8796,
+      "step": 137930
+    },
+    {
+      "epoch": 0.998501596125866,
+      "grad_norm": 0.15717382729053497,
+      "learning_rate": 4.001505642540193e-06,
+      "loss": 0.8606,
+      "step": 137940
+    },
+    {
+      "epoch": 0.9985739827864522,
+      "grad_norm": 0.16818828880786896,
+      "learning_rate": 4.001433255879607e-06,
+      "loss": 0.8777,
+      "step": 137950
+    },
+    {
+      "epoch": 0.9986463694470383,
+      "grad_norm": 0.18249420821666718,
+      "learning_rate": 4.001360869219021e-06,
+      "loss": 0.866,
+      "step": 137960
+    },
+    {
+      "epoch": 0.9987187561076245,
+      "grad_norm": 0.16219259798526764,
+      "learning_rate": 4.001288482558435e-06,
+      "loss": 0.8638,
+      "step": 137970
+    },
+    {
+      "epoch": 0.9987911427682107,
+      "grad_norm": 0.15621773898601532,
+      "learning_rate": 4.0012160958978485e-06,
+      "loss": 0.8691,
+      "step": 137980
+    },
+    {
+      "epoch": 0.9988635294287969,
+      "grad_norm": 0.1662006676197052,
+      "learning_rate": 4.001143709237262e-06,
+      "loss": 0.8813,
+      "step": 137990
+    },
+    {
+      "epoch": 0.998935916089383,
+      "grad_norm": 0.13916003704071045,
+      "learning_rate": 4.001071322576676e-06,
+      "loss": 0.8568,
+      "step": 138000
+    },
+    {
+      "epoch": 0.9990083027499692,
+      "grad_norm": 0.15064559876918793,
+      "learning_rate": 4.000998935916089e-06,
+      "loss": 0.8619,
+      "step": 138010
+    },
+    {
+      "epoch": 0.9990806894105554,
+      "grad_norm": 0.16008462011814117,
+      "learning_rate": 4.000926549255503e-06,
+      "loss": 0.8738,
+      "step": 138020
+    },
+    {
+      "epoch": 0.9991530760711416,
+      "grad_norm": 0.1454089879989624,
+      "learning_rate": 4.0008541625949175e-06,
+      "loss": 0.8681,
+      "step": 138030
+    },
+    {
+      "epoch": 0.9992254627317277,
+      "grad_norm": 0.3634811341762543,
+      "learning_rate": 4.000781775934331e-06,
+      "loss": 0.8603,
+      "step": 138040
+    },
+    {
+      "epoch": 0.999297849392314,
+      "grad_norm": 0.16307184100151062,
+      "learning_rate": 4.000709389273745e-06,
+      "loss": 0.8636,
+      "step": 138050
+    },
+    {
+      "epoch": 0.9993702360529002,
+      "grad_norm": 0.15558481216430664,
+      "learning_rate": 4.000637002613158e-06,
+      "loss": 0.8703,
+      "step": 138060
+    },
+    {
+      "epoch": 0.9994426227134864,
+      "grad_norm": 0.1434854418039322,
+      "learning_rate": 4.000564615952573e-06,
+      "loss": 0.8678,
+      "step": 138070
+    },
+    {
+      "epoch": 0.9995150093740726,
+      "grad_norm": 0.13864096999168396,
+      "learning_rate": 4.000492229291986e-06,
+      "loss": 0.8635,
+      "step": 138080
+    },
+    {
+      "epoch": 0.9995873960346587,
+      "grad_norm": 0.1576741486787796,
+      "learning_rate": 4.0004198426314e-06,
+      "loss": 0.8626,
+      "step": 138090
+    },
+    {
+      "epoch": 0.9996597826952449,
+      "grad_norm": 0.14672447741031647,
+      "learning_rate": 4.000347455970814e-06,
+      "loss": 0.8483,
+      "step": 138100
+    },
+    {
+      "epoch": 0.9997321693558311,
+      "grad_norm": 0.16229544579982758,
+      "learning_rate": 4.000275069310228e-06,
+      "loss": 0.8708,
+      "step": 138110
+    },
+    {
+      "epoch": 0.9998045560164173,
+      "grad_norm": 0.14865851402282715,
+      "learning_rate": 4.000202682649642e-06,
+      "loss": 0.8589,
+      "step": 138120
+    },
+    {
+      "epoch": 0.9998769426770034,
+      "grad_norm": 0.14516963064670563,
+      "learning_rate": 4.000130295989055e-06,
+      "loss": 0.8713,
+      "step": 138130
+    },
+    {
+      "epoch": 0.9999493293375896,
+      "grad_norm": 0.14701320230960846,
+      "learning_rate": 4.000057909328469e-06,
+      "loss": 0.8514,
+      "step": 138140
+    },
+    {
+      "epoch": 1.000021715998176,
+      "grad_norm": 0.15110774338245392,
+      "learning_rate": 3.999985522667883e-06,
+      "loss": 0.8665,
+      "step": 138150
+    },
+    {
+      "epoch": 1.000094102658762,
+      "grad_norm": 0.14296278357505798,
+      "learning_rate": 3.999913136007297e-06,
+      "loss": 0.8619,
+      "step": 138160
+    },
+    {
+      "epoch": 1.0001664893193483,
+      "grad_norm": 0.14643356204032898,
+      "learning_rate": 3.999840749346711e-06,
+      "loss": 0.8675,
+      "step": 138170
+    },
+    {
+      "epoch": 1.0002388759799343,
+      "grad_norm": 0.16440226137638092,
+      "learning_rate": 3.999768362686124e-06,
+      "loss": 0.8456,
+      "step": 138180
+    },
+    {
+      "epoch": 1.0003112626405206,
+      "grad_norm": 0.14877723157405853,
+      "learning_rate": 3.999695976025538e-06,
+      "loss": 0.8591,
+      "step": 138190
+    },
+    {
+      "epoch": 1.0003836493011067,
+      "grad_norm": 0.150392085313797,
+      "learning_rate": 3.999623589364952e-06,
+      "loss": 0.8588,
+      "step": 138200
+    },
+    {
+      "epoch": 1.000456035961693,
+      "grad_norm": 0.1557486355304718,
+      "learning_rate": 3.999551202704366e-06,
+      "loss": 0.8715,
+      "step": 138210
+    },
+    {
+      "epoch": 1.0005284226222793,
+      "grad_norm": 0.16424834728240967,
+      "learning_rate": 3.9994788160437795e-06,
+      "loss": 0.8752,
+      "step": 138220
+    },
+    {
+      "epoch": 1.0006008092828653,
+      "grad_norm": 0.1615014523267746,
+      "learning_rate": 3.999406429383193e-06,
+      "loss": 0.8734,
+      "step": 138230
+    },
+    {
+      "epoch": 1.0006731959434516,
+      "grad_norm": 0.1606891006231308,
+      "learning_rate": 3.999334042722608e-06,
+      "loss": 0.8675,
+      "step": 138240
+    },
+    {
+      "epoch": 1.0007455826040377,
+      "grad_norm": 0.16052010655403137,
+      "learning_rate": 3.999261656062021e-06,
+      "loss": 0.8599,
+      "step": 138250
+    },
+    {
+      "epoch": 1.000817969264624,
+      "grad_norm": 0.154453843832016,
+      "learning_rate": 3.999189269401435e-06,
+      "loss": 0.8503,
+      "step": 138260
+    },
+    {
+      "epoch": 1.00089035592521,
+      "grad_norm": 0.1499967873096466,
+      "learning_rate": 3.9991168827408485e-06,
+      "loss": 0.8631,
+      "step": 138270
+    },
+    {
+      "epoch": 1.0009627425857963,
+      "grad_norm": 0.1510349065065384,
+      "learning_rate": 3.999044496080263e-06,
+      "loss": 0.8663,
+      "step": 138280
+    },
+    {
+      "epoch": 1.0010351292463824,
+      "grad_norm": 0.18189969658851624,
+      "learning_rate": 3.9989721094196766e-06,
+      "loss": 0.8725,
+      "step": 138290
+    },
+    {
+      "epoch": 1.0011075159069687,
+      "grad_norm": 0.1505780965089798,
+      "learning_rate": 3.99889972275909e-06,
+      "loss": 0.8589,
+      "step": 138300
+    },
+    {
+      "epoch": 1.001179902567555,
+      "grad_norm": 0.15750384330749512,
+      "learning_rate": 3.998827336098504e-06,
+      "loss": 0.8675,
+      "step": 138310
+    },
+    {
+      "epoch": 1.001252289228141,
+      "grad_norm": 0.14412176609039307,
+      "learning_rate": 3.998754949437918e-06,
+      "loss": 0.8613,
+      "step": 138320
+    },
+    {
+      "epoch": 1.0013246758887273,
+      "grad_norm": 0.1520223170518875,
+      "learning_rate": 3.998682562777332e-06,
+      "loss": 0.8639,
+      "step": 138330
+    },
+    {
+      "epoch": 1.0013970625493134,
+      "grad_norm": 0.15672039985656738,
+      "learning_rate": 3.9986101761167455e-06,
+      "loss": 0.8651,
+      "step": 138340
+    },
+    {
+      "epoch": 1.0014694492098997,
+      "grad_norm": 0.19693726301193237,
+      "learning_rate": 3.998537789456159e-06,
+      "loss": 0.8727,
+      "step": 138350
+    },
+    {
+      "epoch": 1.0015418358704857,
+      "grad_norm": 0.18891595304012299,
+      "learning_rate": 3.9984654027955736e-06,
+      "loss": 0.8756,
+      "step": 138360
+    },
+    {
+      "epoch": 1.001614222531072,
+      "grad_norm": 0.14918550848960876,
+      "learning_rate": 3.998393016134987e-06,
+      "loss": 0.8603,
+      "step": 138370
+    },
+    {
+      "epoch": 1.001686609191658,
+      "grad_norm": 0.16212646663188934,
+      "learning_rate": 3.998320629474401e-06,
+      "loss": 0.8653,
+      "step": 138380
+    },
+    {
+      "epoch": 1.0017589958522444,
+      "grad_norm": 0.14969107508659363,
+      "learning_rate": 3.998248242813814e-06,
+      "loss": 0.8777,
+      "step": 138390
+    },
+    {
+      "epoch": 1.0018313825128304,
+      "grad_norm": 0.14508505165576935,
+      "learning_rate": 3.998175856153229e-06,
+      "loss": 0.8573,
+      "step": 138400
+    },
+    {
+      "epoch": 1.0019037691734167,
+      "grad_norm": 0.1480925977230072,
+      "learning_rate": 3.9981034694926425e-06,
+      "loss": 0.872,
+      "step": 138410
+    },
+    {
+      "epoch": 1.001976155834003,
+      "grad_norm": 0.16480182111263275,
+      "learning_rate": 3.998031082832056e-06,
+      "loss": 0.881,
+      "step": 138420
+    },
+    {
+      "epoch": 1.002048542494589,
+      "grad_norm": 0.14416438341140747,
+      "learning_rate": 3.99795869617147e-06,
+      "loss": 0.867,
+      "step": 138430
+    },
+    {
+      "epoch": 1.0021209291551754,
+      "grad_norm": 0.14154788851737976,
+      "learning_rate": 3.997886309510884e-06,
+      "loss": 0.8568,
+      "step": 138440
+    },
+    {
+      "epoch": 1.0021933158157614,
+      "grad_norm": 0.1673395335674286,
+      "learning_rate": 3.997813922850298e-06,
+      "loss": 0.8553,
+      "step": 138450
+    },
+    {
+      "epoch": 1.0022657024763477,
+      "grad_norm": 0.16951638460159302,
+      "learning_rate": 3.997741536189711e-06,
+      "loss": 0.8569,
+      "step": 138460
+    },
+    {
+      "epoch": 1.0023380891369338,
+      "grad_norm": 0.16748495399951935,
+      "learning_rate": 3.997669149529125e-06,
+      "loss": 0.8657,
+      "step": 138470
+    },
+    {
+      "epoch": 1.00241047579752,
+      "grad_norm": 0.1609235256910324,
+      "learning_rate": 3.9975967628685395e-06,
+      "loss": 0.8493,
+      "step": 138480
+    },
+    {
+      "epoch": 1.0024828624581061,
+      "grad_norm": 0.15038661658763885,
+      "learning_rate": 3.997524376207953e-06,
+      "loss": 0.8577,
+      "step": 138490
+    },
+    {
+      "epoch": 1.0025552491186924,
+      "grad_norm": 0.15699024498462677,
+      "learning_rate": 3.997451989547367e-06,
+      "loss": 0.8584,
+      "step": 138500
+    },
+    {
+      "epoch": 1.0026276357792785,
+      "grad_norm": 0.153981551527977,
+      "learning_rate": 3.99737960288678e-06,
+      "loss": 0.8621,
+      "step": 138510
+    },
+    {
+      "epoch": 1.0027000224398648,
+      "grad_norm": 0.14797601103782654,
+      "learning_rate": 3.997307216226194e-06,
+      "loss": 0.8603,
+      "step": 138520
+    },
+    {
+      "epoch": 1.002772409100451,
+      "grad_norm": 0.18665072321891785,
+      "learning_rate": 3.997234829565608e-06,
+      "loss": 0.8615,
+      "step": 138530
+    },
+    {
+      "epoch": 1.0028447957610371,
+      "grad_norm": 0.18154151737689972,
+      "learning_rate": 3.997162442905021e-06,
+      "loss": 0.868,
+      "step": 138540
+    },
+    {
+      "epoch": 1.0029171824216234,
+      "grad_norm": 0.14707796275615692,
+      "learning_rate": 3.997090056244436e-06,
+      "loss": 0.8544,
+      "step": 138550
+    },
+    {
+      "epoch": 1.0029895690822095,
+      "grad_norm": 0.1541696935892105,
+      "learning_rate": 3.997017669583849e-06,
+      "loss": 0.8633,
+      "step": 138560
+    },
+    {
+      "epoch": 1.0030619557427958,
+      "grad_norm": 0.15165592730045319,
+      "learning_rate": 3.996945282923263e-06,
+      "loss": 0.8684,
+      "step": 138570
+    },
+    {
+      "epoch": 1.0031343424033818,
+      "grad_norm": 0.1640002727508545,
+      "learning_rate": 3.9968728962626765e-06,
+      "loss": 0.8738,
+      "step": 138580
+    },
+    {
+      "epoch": 1.0032067290639681,
+      "grad_norm": 0.14847826957702637,
+      "learning_rate": 3.996800509602091e-06,
+      "loss": 0.8608,
+      "step": 138590
+    },
+    {
+      "epoch": 1.0032791157245542,
+      "grad_norm": 0.16807755827903748,
+      "learning_rate": 3.996728122941505e-06,
+      "loss": 0.8656,
+      "step": 138600
+    },
+    {
+      "epoch": 1.0033515023851405,
+      "grad_norm": 0.175052210688591,
+      "learning_rate": 3.996655736280918e-06,
+      "loss": 0.8663,
+      "step": 138610
+    },
+    {
+      "epoch": 1.0034238890457265,
+      "grad_norm": 0.18418249487876892,
+      "learning_rate": 3.996583349620332e-06,
+      "loss": 0.8619,
+      "step": 138620
+    },
+    {
+      "epoch": 1.0034962757063128,
+      "grad_norm": 0.14789465069770813,
+      "learning_rate": 3.996510962959746e-06,
+      "loss": 0.8597,
+      "step": 138630
+    },
+    {
+      "epoch": 1.0035686623668991,
+      "grad_norm": 0.14474569261074066,
+      "learning_rate": 3.99643857629916e-06,
+      "loss": 0.8721,
+      "step": 138640
+    },
+    {
+      "epoch": 1.0036410490274852,
+      "grad_norm": 0.14751043915748596,
+      "learning_rate": 3.9963661896385735e-06,
+      "loss": 0.8922,
+      "step": 138650
+    },
+    {
+      "epoch": 1.0037134356880715,
+      "grad_norm": 0.14138489961624146,
+      "learning_rate": 3.996293802977987e-06,
+      "loss": 0.8643,
+      "step": 138660
+    },
+    {
+      "epoch": 1.0037858223486575,
+      "grad_norm": 0.14229081571102142,
+      "learning_rate": 3.996221416317402e-06,
+      "loss": 0.8595,
+      "step": 138670
+    },
+    {
+      "epoch": 1.0038582090092438,
+      "grad_norm": 0.24338965117931366,
+      "learning_rate": 3.996149029656815e-06,
+      "loss": 0.861,
+      "step": 138680
+    },
+    {
+      "epoch": 1.00393059566983,
+      "grad_norm": 0.2033909410238266,
+      "learning_rate": 3.996076642996229e-06,
+      "loss": 0.864,
+      "step": 138690
+    },
+    {
+      "epoch": 1.0040029823304162,
+      "grad_norm": 0.15376165509223938,
+      "learning_rate": 3.9960042563356424e-06,
+      "loss": 0.8662,
+      "step": 138700
+    },
+    {
+      "epoch": 1.0040753689910022,
+      "grad_norm": 0.14948670566082,
+      "learning_rate": 3.995931869675057e-06,
+      "loss": 0.8744,
+      "step": 138710
+    },
+    {
+      "epoch": 1.0041477556515885,
+      "grad_norm": 0.15459775924682617,
+      "learning_rate": 3.9958594830144705e-06,
+      "loss": 0.8763,
+      "step": 138720
+    },
+    {
+      "epoch": 1.0042201423121746,
+      "grad_norm": 0.16300074756145477,
+      "learning_rate": 3.995787096353884e-06,
+      "loss": 0.8532,
+      "step": 138730
+    },
+    {
+      "epoch": 1.004292528972761,
+      "grad_norm": 0.14748498797416687,
+      "learning_rate": 3.995714709693298e-06,
+      "loss": 0.8659,
+      "step": 138740
+    },
+    {
+      "epoch": 1.0043649156333472,
+      "grad_norm": 0.14953255653381348,
+      "learning_rate": 3.995642323032712e-06,
+      "loss": 0.8737,
+      "step": 138750
+    },
+    {
+      "epoch": 1.0044373022939332,
+      "grad_norm": 0.17030136287212372,
+      "learning_rate": 3.995569936372126e-06,
+      "loss": 0.8659,
+      "step": 138760
+    },
+    {
+      "epoch": 1.0045096889545195,
+      "grad_norm": 0.15466326475143433,
+      "learning_rate": 3.9954975497115395e-06,
+      "loss": 0.8597,
+      "step": 138770
+    },
+    {
+      "epoch": 1.0045820756151056,
+      "grad_norm": 0.2016264945268631,
+      "learning_rate": 3.995425163050953e-06,
+      "loss": 0.8675,
+      "step": 138780
+    },
+    {
+      "epoch": 1.0046544622756919,
+      "grad_norm": 0.2277584671974182,
+      "learning_rate": 3.9953527763903675e-06,
+      "loss": 0.8689,
+      "step": 138790
+    },
+    {
+      "epoch": 1.004726848936278,
+      "grad_norm": 0.14678841829299927,
+      "learning_rate": 3.995280389729781e-06,
+      "loss": 0.8658,
+      "step": 138800
+    },
+    {
+      "epoch": 1.0047992355968642,
+      "grad_norm": 0.18819843232631683,
+      "learning_rate": 3.995208003069195e-06,
+      "loss": 0.8627,
+      "step": 138810
+    },
+    {
+      "epoch": 1.0048716222574503,
+      "grad_norm": 0.14515884220600128,
+      "learning_rate": 3.995135616408608e-06,
+      "loss": 0.8478,
+      "step": 138820
+    },
+    {
+      "epoch": 1.0049440089180366,
+      "grad_norm": 0.15614381432533264,
+      "learning_rate": 3.995063229748022e-06,
+      "loss": 0.8636,
+      "step": 138830
+    },
+    {
+      "epoch": 1.0050163955786229,
+      "grad_norm": 0.14769770205020905,
+      "learning_rate": 3.9949908430874365e-06,
+      "loss": 0.8577,
+      "step": 138840
+    },
+    {
+      "epoch": 1.005088782239209,
+      "grad_norm": 0.15718652307987213,
+      "learning_rate": 3.99491845642685e-06,
+      "loss": 0.8645,
+      "step": 138850
+    },
+    {
+      "epoch": 1.0051611688997952,
+      "grad_norm": 0.14552608132362366,
+      "learning_rate": 3.994846069766264e-06,
+      "loss": 0.8609,
+      "step": 138860
+    },
+    {
+      "epoch": 1.0052335555603813,
+      "grad_norm": 0.1629728376865387,
+      "learning_rate": 3.994773683105677e-06,
+      "loss": 0.8725,
+      "step": 138870
+    },
+    {
+      "epoch": 1.0053059422209676,
+      "grad_norm": 0.1548013985157013,
+      "learning_rate": 3.994701296445092e-06,
+      "loss": 0.8592,
+      "step": 138880
+    },
+    {
+      "epoch": 1.0053783288815537,
+      "grad_norm": 0.1612953543663025,
+      "learning_rate": 3.994628909784505e-06,
+      "loss": 0.8649,
+      "step": 138890
+    },
+    {
+      "epoch": 1.00545071554214,
+      "grad_norm": 0.15613959729671478,
+      "learning_rate": 3.994556523123919e-06,
+      "loss": 0.852,
+      "step": 138900
+    },
+    {
+      "epoch": 1.005523102202726,
+      "grad_norm": 0.14688079059123993,
+      "learning_rate": 3.994484136463333e-06,
+      "loss": 0.8625,
+      "step": 138910
+    },
+    {
+      "epoch": 1.0055954888633123,
+      "grad_norm": 0.16856560111045837,
+      "learning_rate": 3.994411749802747e-06,
+      "loss": 0.866,
+      "step": 138920
+    },
+    {
+      "epoch": 1.0056678755238984,
+      "grad_norm": 0.14123956859111786,
+      "learning_rate": 3.994339363142161e-06,
+      "loss": 0.8595,
+      "step": 138930
+    },
+    {
+      "epoch": 1.0057402621844846,
+      "grad_norm": 0.15855082869529724,
+      "learning_rate": 3.994266976481574e-06,
+      "loss": 0.863,
+      "step": 138940
+    },
+    {
+      "epoch": 1.005812648845071,
+      "grad_norm": 0.16692784428596497,
+      "learning_rate": 3.994194589820988e-06,
+      "loss": 0.8534,
+      "step": 138950
+    },
+    {
+      "epoch": 1.005885035505657,
+      "grad_norm": 0.1530875861644745,
+      "learning_rate": 3.994122203160402e-06,
+      "loss": 0.855,
+      "step": 138960
+    },
+    {
+      "epoch": 1.0059574221662433,
+      "grad_norm": 0.16144618391990662,
+      "learning_rate": 3.994049816499816e-06,
+      "loss": 0.8496,
+      "step": 138970
+    },
+    {
+      "epoch": 1.0060298088268294,
+      "grad_norm": 0.15884657204151154,
+      "learning_rate": 3.99397742983923e-06,
+      "loss": 0.8641,
+      "step": 138980
+    },
+    {
+      "epoch": 1.0061021954874156,
+      "grad_norm": 0.1514037847518921,
+      "learning_rate": 3.993905043178643e-06,
+      "loss": 0.8562,
+      "step": 138990
+    },
+    {
+      "epoch": 1.0061745821480017,
+      "grad_norm": 0.1448083519935608,
+      "learning_rate": 3.993832656518058e-06,
+      "loss": 0.8545,
+      "step": 139000
+    },
+    {
+      "epoch": 1.006246968808588,
+      "grad_norm": 0.153413325548172,
+      "learning_rate": 3.993760269857471e-06,
+      "loss": 0.8668,
+      "step": 139010
+    },
+    {
+      "epoch": 1.006319355469174,
+      "grad_norm": 0.16846509277820587,
+      "learning_rate": 3.993687883196885e-06,
+      "loss": 0.8534,
+      "step": 139020
+    },
+    {
+      "epoch": 1.0063917421297603,
+      "grad_norm": 0.14701557159423828,
+      "learning_rate": 3.9936154965362986e-06,
+      "loss": 0.8597,
+      "step": 139030
+    },
+    {
+      "epoch": 1.0064641287903464,
+      "grad_norm": 0.15452063083648682,
+      "learning_rate": 3.993543109875713e-06,
+      "loss": 0.8734,
+      "step": 139040
+    },
+    {
+      "epoch": 1.0065365154509327,
+      "grad_norm": 0.14439797401428223,
+      "learning_rate": 3.993470723215126e-06,
+      "loss": 0.8762,
+      "step": 139050
+    },
+    {
+      "epoch": 1.006608902111519,
+      "grad_norm": 0.15302760899066925,
+      "learning_rate": 3.993398336554539e-06,
+      "loss": 0.8597,
+      "step": 139060
+    },
+    {
+      "epoch": 1.006681288772105,
+      "grad_norm": 0.1926603466272354,
+      "learning_rate": 3.993325949893954e-06,
+      "loss": 0.8524,
+      "step": 139070
+    },
+    {
+      "epoch": 1.0067536754326913,
+      "grad_norm": 0.15659856796264648,
+      "learning_rate": 3.9932535632333675e-06,
+      "loss": 0.8558,
+      "step": 139080
+    },
+    {
+      "epoch": 1.0068260620932774,
+      "grad_norm": 0.18837852776050568,
+      "learning_rate": 3.993181176572781e-06,
+      "loss": 0.8756,
+      "step": 139090
+    },
+    {
+      "epoch": 1.0068984487538637,
+      "grad_norm": 0.14618010818958282,
+      "learning_rate": 3.993108789912195e-06,
+      "loss": 0.8688,
+      "step": 139100
+    },
+    {
+      "epoch": 1.0069708354144498,
+      "grad_norm": 0.1539430171251297,
+      "learning_rate": 3.993036403251609e-06,
+      "loss": 0.8525,
+      "step": 139110
+    },
+    {
+      "epoch": 1.007043222075036,
+      "grad_norm": 0.1436515599489212,
+      "learning_rate": 3.992964016591023e-06,
+      "loss": 0.8657,
+      "step": 139120
+    },
+    {
+      "epoch": 1.0071156087356221,
+      "grad_norm": 0.15015138685703278,
+      "learning_rate": 3.992891629930436e-06,
+      "loss": 0.8606,
+      "step": 139130
+    },
+    {
+      "epoch": 1.0071879953962084,
+      "grad_norm": 0.15750154852867126,
+      "learning_rate": 3.99281924326985e-06,
+      "loss": 0.849,
+      "step": 139140
+    },
+    {
+      "epoch": 1.0072603820567945,
+      "grad_norm": 0.15625479817390442,
+      "learning_rate": 3.9927468566092645e-06,
+      "loss": 0.8614,
+      "step": 139150
+    },
+    {
+      "epoch": 1.0073327687173808,
+      "grad_norm": 0.1461600959300995,
+      "learning_rate": 3.992674469948678e-06,
+      "loss": 0.8485,
+      "step": 139160
+    },
+    {
+      "epoch": 1.007405155377967,
+      "grad_norm": 0.16277381777763367,
+      "learning_rate": 3.992602083288092e-06,
+      "loss": 0.8705,
+      "step": 139170
+    },
+    {
+      "epoch": 1.007477542038553,
+      "grad_norm": 0.3381327986717224,
+      "learning_rate": 3.992529696627505e-06,
+      "loss": 0.8686,
+      "step": 139180
+    },
+    {
+      "epoch": 1.0075499286991394,
+      "grad_norm": 0.14702801406383514,
+      "learning_rate": 3.99245730996692e-06,
+      "loss": 0.8526,
+      "step": 139190
+    },
+    {
+      "epoch": 1.0076223153597255,
+      "grad_norm": 0.15568077564239502,
+      "learning_rate": 3.992384923306333e-06,
+      "loss": 0.8497,
+      "step": 139200
+    },
+    {
+      "epoch": 1.0076947020203117,
+      "grad_norm": 0.16276511549949646,
+      "learning_rate": 3.992312536645747e-06,
+      "loss": 0.8634,
+      "step": 139210
+    },
+    {
+      "epoch": 1.0077670886808978,
+      "grad_norm": 0.1559554636478424,
+      "learning_rate": 3.992240149985161e-06,
+      "loss": 0.8703,
+      "step": 139220
+    },
+    {
+      "epoch": 1.007839475341484,
+      "grad_norm": 0.156097874045372,
+      "learning_rate": 3.992167763324575e-06,
+      "loss": 0.8721,
+      "step": 139230
+    },
+    {
+      "epoch": 1.0079118620020702,
+      "grad_norm": 0.1414407193660736,
+      "learning_rate": 3.992095376663989e-06,
+      "loss": 0.8661,
+      "step": 139240
+    },
+    {
+      "epoch": 1.0079842486626565,
+      "grad_norm": 0.15653790533542633,
+      "learning_rate": 3.992022990003402e-06,
+      "loss": 0.8547,
+      "step": 139250
+    },
+    {
+      "epoch": 1.0080566353232427,
+      "grad_norm": 0.15031273663043976,
+      "learning_rate": 3.991950603342816e-06,
+      "loss": 0.858,
+      "step": 139260
+    },
+    {
+      "epoch": 1.0081290219838288,
+      "grad_norm": 0.17239518463611603,
+      "learning_rate": 3.9918782166822304e-06,
+      "loss": 0.8548,
+      "step": 139270
+    },
+    {
+      "epoch": 1.008201408644415,
+      "grad_norm": 0.1503749042749405,
+      "learning_rate": 3.991805830021644e-06,
+      "loss": 0.8711,
+      "step": 139280
+    },
+    {
+      "epoch": 1.0082737953050012,
+      "grad_norm": 0.15691126883029938,
+      "learning_rate": 3.991733443361058e-06,
+      "loss": 0.8525,
+      "step": 139290
+    },
+    {
+      "epoch": 1.0083461819655875,
+      "grad_norm": 0.15124544501304626,
+      "learning_rate": 3.991661056700471e-06,
+      "loss": 0.8684,
+      "step": 139300
+    },
+    {
+      "epoch": 1.0084185686261735,
+      "grad_norm": 0.16640368103981018,
+      "learning_rate": 3.991588670039886e-06,
+      "loss": 0.8621,
+      "step": 139310
+    },
+    {
+      "epoch": 1.0084909552867598,
+      "grad_norm": 0.14733107388019562,
+      "learning_rate": 3.991516283379299e-06,
+      "loss": 0.8595,
+      "step": 139320
+    },
+    {
+      "epoch": 1.0085633419473459,
+      "grad_norm": 0.14905431866645813,
+      "learning_rate": 3.991443896718713e-06,
+      "loss": 0.8636,
+      "step": 139330
+    },
+    {
+      "epoch": 1.0086357286079322,
+      "grad_norm": 0.16022346913814545,
+      "learning_rate": 3.991371510058127e-06,
+      "loss": 0.8564,
+      "step": 139340
+    },
+    {
+      "epoch": 1.0087081152685182,
+      "grad_norm": 0.14993324875831604,
+      "learning_rate": 3.991299123397541e-06,
+      "loss": 0.8704,
+      "step": 139350
+    },
+    {
+      "epoch": 1.0087805019291045,
+      "grad_norm": 0.14608387649059296,
+      "learning_rate": 3.991226736736955e-06,
+      "loss": 0.848,
+      "step": 139360
+    },
+    {
+      "epoch": 1.0088528885896908,
+      "grad_norm": 0.14675495028495789,
+      "learning_rate": 3.991154350076368e-06,
+      "loss": 0.8736,
+      "step": 139370
+    },
+    {
+      "epoch": 1.0089252752502769,
+      "grad_norm": 0.13962878286838531,
+      "learning_rate": 3.991081963415782e-06,
+      "loss": 0.8634,
+      "step": 139380
+    },
+    {
+      "epoch": 1.0089976619108632,
+      "grad_norm": 0.1584140658378601,
+      "learning_rate": 3.991009576755196e-06,
+      "loss": 0.8708,
+      "step": 139390
+    },
+    {
+      "epoch": 1.0090700485714492,
+      "grad_norm": 0.1543683111667633,
+      "learning_rate": 3.99093719009461e-06,
+      "loss": 0.8651,
+      "step": 139400
+    },
+    {
+      "epoch": 1.0091424352320355,
+      "grad_norm": 0.1464136689901352,
+      "learning_rate": 3.990864803434024e-06,
+      "loss": 0.8693,
+      "step": 139410
+    },
+    {
+      "epoch": 1.0092148218926216,
+      "grad_norm": 0.1557295322418213,
+      "learning_rate": 3.990792416773437e-06,
+      "loss": 0.8609,
+      "step": 139420
+    },
+    {
+      "epoch": 1.0092872085532079,
+      "grad_norm": 0.14724186062812805,
+      "learning_rate": 3.990720030112851e-06,
+      "loss": 0.8753,
+      "step": 139430
+    },
+    {
+      "epoch": 1.009359595213794,
+      "grad_norm": 0.1482950747013092,
+      "learning_rate": 3.990647643452265e-06,
+      "loss": 0.8607,
+      "step": 139440
+    },
+    {
+      "epoch": 1.0094319818743802,
+      "grad_norm": 0.15687072277069092,
+      "learning_rate": 3.990575256791679e-06,
+      "loss": 0.857,
+      "step": 139450
+    },
+    {
+      "epoch": 1.0095043685349663,
+      "grad_norm": 0.15880662202835083,
+      "learning_rate": 3.9905028701310925e-06,
+      "loss": 0.8595,
+      "step": 139460
+    },
+    {
+      "epoch": 1.0095767551955526,
+      "grad_norm": 0.14912816882133484,
+      "learning_rate": 3.990430483470506e-06,
+      "loss": 0.8613,
+      "step": 139470
+    },
+    {
+      "epoch": 1.0096491418561389,
+      "grad_norm": 0.15200045704841614,
+      "learning_rate": 3.990358096809921e-06,
+      "loss": 0.8597,
+      "step": 139480
+    },
+    {
+      "epoch": 1.009721528516725,
+      "grad_norm": 0.1544422060251236,
+      "learning_rate": 3.990285710149334e-06,
+      "loss": 0.858,
+      "step": 139490
+    },
+    {
+      "epoch": 1.0097939151773112,
+      "grad_norm": 0.16103899478912354,
+      "learning_rate": 3.990213323488748e-06,
+      "loss": 0.8596,
+      "step": 139500
+    },
+    {
+      "epoch": 1.0098663018378973,
+      "grad_norm": 0.1505272537469864,
+      "learning_rate": 3.9901409368281615e-06,
+      "loss": 0.859,
+      "step": 139510
+    },
+    {
+      "epoch": 1.0099386884984836,
+      "grad_norm": 0.16239230334758759,
+      "learning_rate": 3.990068550167576e-06,
+      "loss": 0.8547,
+      "step": 139520
+    },
+    {
+      "epoch": 1.0100110751590696,
+      "grad_norm": 0.15705235302448273,
+      "learning_rate": 3.9899961635069895e-06,
+      "loss": 0.8604,
+      "step": 139530
+    },
+    {
+      "epoch": 1.010083461819656,
+      "grad_norm": 0.15264788269996643,
+      "learning_rate": 3.989923776846403e-06,
+      "loss": 0.8729,
+      "step": 139540
+    },
+    {
+      "epoch": 1.010155848480242,
+      "grad_norm": 0.15049117803573608,
+      "learning_rate": 3.989851390185817e-06,
+      "loss": 0.8613,
+      "step": 139550
+    },
+    {
+      "epoch": 1.0102282351408283,
+      "grad_norm": 0.17415769398212433,
+      "learning_rate": 3.989779003525231e-06,
+      "loss": 0.8613,
+      "step": 139560
+    },
+    {
+      "epoch": 1.0103006218014143,
+      "grad_norm": 0.151180699467659,
+      "learning_rate": 3.989706616864645e-06,
+      "loss": 0.8631,
+      "step": 139570
+    },
+    {
+      "epoch": 1.0103730084620006,
+      "grad_norm": 0.14194343984127045,
+      "learning_rate": 3.9896342302040585e-06,
+      "loss": 0.8597,
+      "step": 139580
+    },
+    {
+      "epoch": 1.010445395122587,
+      "grad_norm": 0.15992465615272522,
+      "learning_rate": 3.989561843543472e-06,
+      "loss": 0.8636,
+      "step": 139590
+    },
+    {
+      "epoch": 1.010517781783173,
+      "grad_norm": 0.15397891402244568,
+      "learning_rate": 3.989489456882886e-06,
+      "loss": 0.8599,
+      "step": 139600
+    },
+    {
+      "epoch": 1.0105901684437593,
+      "grad_norm": 0.16052521765232086,
+      "learning_rate": 3.989417070222299e-06,
+      "loss": 0.8576,
+      "step": 139610
+    },
+    {
+      "epoch": 1.0106625551043453,
+      "grad_norm": 0.15102127194404602,
+      "learning_rate": 3.989344683561713e-06,
+      "loss": 0.872,
+      "step": 139620
+    },
+    {
+      "epoch": 1.0107349417649316,
+      "grad_norm": 0.1572069674730301,
+      "learning_rate": 3.989272296901127e-06,
+      "loss": 0.8688,
+      "step": 139630
+    },
+    {
+      "epoch": 1.0108073284255177,
+      "grad_norm": 0.14729195833206177,
+      "learning_rate": 3.989199910240541e-06,
+      "loss": 0.8668,
+      "step": 139640
+    },
+    {
+      "epoch": 1.010879715086104,
+      "grad_norm": 0.16329136490821838,
+      "learning_rate": 3.989127523579955e-06,
+      "loss": 0.8643,
+      "step": 139650
+    },
+    {
+      "epoch": 1.01095210174669,
+      "grad_norm": 0.15864083170890808,
+      "learning_rate": 3.989055136919368e-06,
+      "loss": 0.8686,
+      "step": 139660
+    },
+    {
+      "epoch": 1.0110244884072763,
+      "grad_norm": 0.1520940214395523,
+      "learning_rate": 3.988982750258783e-06,
+      "loss": 0.8573,
+      "step": 139670
+    },
+    {
+      "epoch": 1.0110968750678624,
+      "grad_norm": 0.2237161546945572,
+      "learning_rate": 3.988910363598196e-06,
+      "loss": 0.8685,
+      "step": 139680
+    },
+    {
+      "epoch": 1.0111692617284487,
+      "grad_norm": 0.15291093289852142,
+      "learning_rate": 3.98883797693761e-06,
+      "loss": 0.8554,
+      "step": 139690
+    },
+    {
+      "epoch": 1.011241648389035,
+      "grad_norm": 0.1433836966753006,
+      "learning_rate": 3.9887655902770235e-06,
+      "loss": 0.8553,
+      "step": 139700
+    },
+    {
+      "epoch": 1.011314035049621,
+      "grad_norm": 0.15727286040782928,
+      "learning_rate": 3.988693203616438e-06,
+      "loss": 0.8692,
+      "step": 139710
+    },
+    {
+      "epoch": 1.0113864217102073,
+      "grad_norm": 0.14590954780578613,
+      "learning_rate": 3.988620816955852e-06,
+      "loss": 0.8654,
+      "step": 139720
+    },
+    {
+      "epoch": 1.0114588083707934,
+      "grad_norm": 0.15376925468444824,
+      "learning_rate": 3.988548430295265e-06,
+      "loss": 0.8614,
+      "step": 139730
+    },
+    {
+      "epoch": 1.0115311950313797,
+      "grad_norm": 0.14909863471984863,
+      "learning_rate": 3.988476043634679e-06,
+      "loss": 0.8432,
+      "step": 139740
+    },
+    {
+      "epoch": 1.0116035816919657,
+      "grad_norm": 0.15601171553134918,
+      "learning_rate": 3.988403656974093e-06,
+      "loss": 0.8543,
+      "step": 139750
+    },
+    {
+      "epoch": 1.011675968352552,
+      "grad_norm": 0.1624634563922882,
+      "learning_rate": 3.988331270313507e-06,
+      "loss": 0.8587,
+      "step": 139760
+    },
+    {
+      "epoch": 1.011748355013138,
+      "grad_norm": 0.15104293823242188,
+      "learning_rate": 3.9882588836529206e-06,
+      "loss": 0.8621,
+      "step": 139770
+    },
+    {
+      "epoch": 1.0118207416737244,
+      "grad_norm": 0.15875457227230072,
+      "learning_rate": 3.988186496992334e-06,
+      "loss": 0.8702,
+      "step": 139780
+    },
+    {
+      "epoch": 1.0118931283343104,
+      "grad_norm": 0.14369283616542816,
+      "learning_rate": 3.988114110331749e-06,
+      "loss": 0.8654,
+      "step": 139790
+    },
+    {
+      "epoch": 1.0119655149948967,
+      "grad_norm": 0.15194371342658997,
+      "learning_rate": 3.988041723671162e-06,
+      "loss": 0.8584,
+      "step": 139800
+    },
+    {
+      "epoch": 1.012037901655483,
+      "grad_norm": 0.18892666697502136,
+      "learning_rate": 3.987969337010576e-06,
+      "loss": 0.8606,
+      "step": 139810
+    },
+    {
+      "epoch": 1.012110288316069,
+      "grad_norm": 0.16798314452171326,
+      "learning_rate": 3.9878969503499895e-06,
+      "loss": 0.8621,
+      "step": 139820
+    },
+    {
+      "epoch": 1.0121826749766554,
+      "grad_norm": 0.15889109671115875,
+      "learning_rate": 3.987824563689404e-06,
+      "loss": 0.8675,
+      "step": 139830
+    },
+    {
+      "epoch": 1.0122550616372414,
+      "grad_norm": 0.14668938517570496,
+      "learning_rate": 3.9877521770288176e-06,
+      "loss": 0.8611,
+      "step": 139840
+    },
+    {
+      "epoch": 1.0123274482978277,
+      "grad_norm": 0.15897607803344727,
+      "learning_rate": 3.987679790368231e-06,
+      "loss": 0.8696,
+      "step": 139850
+    },
+    {
+      "epoch": 1.0123998349584138,
+      "grad_norm": 0.14037030935287476,
+      "learning_rate": 3.987607403707645e-06,
+      "loss": 0.8669,
+      "step": 139860
+    },
+    {
+      "epoch": 1.012472221619,
+      "grad_norm": 0.161557137966156,
+      "learning_rate": 3.987535017047059e-06,
+      "loss": 0.8397,
+      "step": 139870
+    },
+    {
+      "epoch": 1.0125446082795861,
+      "grad_norm": 0.14961223304271698,
+      "learning_rate": 3.987462630386473e-06,
+      "loss": 0.8637,
+      "step": 139880
+    },
+    {
+      "epoch": 1.0126169949401724,
+      "grad_norm": 0.16417111456394196,
+      "learning_rate": 3.9873902437258865e-06,
+      "loss": 0.8535,
+      "step": 139890
+    },
+    {
+      "epoch": 1.0126893816007587,
+      "grad_norm": 0.1531134694814682,
+      "learning_rate": 3.9873178570653e-06,
+      "loss": 0.8662,
+      "step": 139900
+    },
+    {
+      "epoch": 1.0127617682613448,
+      "grad_norm": 0.16314208507537842,
+      "learning_rate": 3.9872454704047146e-06,
+      "loss": 0.874,
+      "step": 139910
+    },
+    {
+      "epoch": 1.012834154921931,
+      "grad_norm": 0.1492464244365692,
+      "learning_rate": 3.987173083744128e-06,
+      "loss": 0.8594,
+      "step": 139920
+    },
+    {
+      "epoch": 1.0129065415825171,
+      "grad_norm": 0.1546127200126648,
+      "learning_rate": 3.987100697083542e-06,
+      "loss": 0.8683,
+      "step": 139930
+    },
+    {
+      "epoch": 1.0129789282431034,
+      "grad_norm": 0.23250380158424377,
+      "learning_rate": 3.987028310422955e-06,
+      "loss": 0.8602,
+      "step": 139940
+    },
+    {
+      "epoch": 1.0130513149036895,
+      "grad_norm": 0.15371747314929962,
+      "learning_rate": 3.98695592376237e-06,
+      "loss": 0.8577,
+      "step": 139950
+    },
+    {
+      "epoch": 1.0131237015642758,
+      "grad_norm": 0.15039215981960297,
+      "learning_rate": 3.9868835371017835e-06,
+      "loss": 0.8775,
+      "step": 139960
+    },
+    {
+      "epoch": 1.0131960882248618,
+      "grad_norm": 0.14372408390045166,
+      "learning_rate": 3.986811150441197e-06,
+      "loss": 0.8603,
+      "step": 139970
+    },
+    {
+      "epoch": 1.0132684748854481,
+      "grad_norm": 0.18086668848991394,
+      "learning_rate": 3.986738763780611e-06,
+      "loss": 0.8697,
+      "step": 139980
+    },
+    {
+      "epoch": 1.0133408615460342,
+      "grad_norm": 0.16421173512935638,
+      "learning_rate": 3.986666377120025e-06,
+      "loss": 0.8727,
+      "step": 139990
+    },
+    {
+      "epoch": 1.0134132482066205,
+      "grad_norm": 0.149836003780365,
+      "learning_rate": 3.986593990459439e-06,
+      "loss": 0.8633,
+      "step": 140000
+    },
+    {
+      "epoch": 1.0134856348672068,
+      "grad_norm": 0.15972940623760223,
+      "learning_rate": 3.9865216037988524e-06,
+      "loss": 0.8598,
+      "step": 140010
+    },
+    {
+      "epoch": 1.0135580215277928,
+      "grad_norm": 0.15127001702785492,
+      "learning_rate": 3.986449217138266e-06,
+      "loss": 0.8639,
+      "step": 140020
+    },
+    {
+      "epoch": 1.0136304081883791,
+      "grad_norm": 0.1499137580394745,
+      "learning_rate": 3.9863768304776805e-06,
+      "loss": 0.8712,
+      "step": 140030
+    },
+    {
+      "epoch": 1.0137027948489652,
+      "grad_norm": 0.1619751751422882,
+      "learning_rate": 3.986304443817094e-06,
+      "loss": 0.8751,
+      "step": 140040
+    },
+    {
+      "epoch": 1.0137751815095515,
+      "grad_norm": 0.15871897339820862,
+      "learning_rate": 3.986232057156508e-06,
+      "loss": 0.8627,
+      "step": 140050
+    },
+    {
+      "epoch": 1.0138475681701375,
+      "grad_norm": 0.15638139843940735,
+      "learning_rate": 3.986159670495921e-06,
+      "loss": 0.8587,
+      "step": 140060
+    },
+    {
+      "epoch": 1.0139199548307238,
+      "grad_norm": 0.1525679975748062,
+      "learning_rate": 3.986087283835335e-06,
+      "loss": 0.8644,
+      "step": 140070
+    },
+    {
+      "epoch": 1.01399234149131,
+      "grad_norm": 0.14840306341648102,
+      "learning_rate": 3.9860148971747494e-06,
+      "loss": 0.8653,
+      "step": 140080
+    },
+    {
+      "epoch": 1.0140647281518962,
+      "grad_norm": 0.15386714041233063,
+      "learning_rate": 3.985942510514163e-06,
+      "loss": 0.8709,
+      "step": 140090
+    },
+    {
+      "epoch": 1.0141371148124823,
+      "grad_norm": 0.17089219391345978,
+      "learning_rate": 3.985870123853577e-06,
+      "loss": 0.8653,
+      "step": 140100
+    },
+    {
+      "epoch": 1.0142095014730685,
+      "grad_norm": 0.16779905557632446,
+      "learning_rate": 3.98579773719299e-06,
+      "loss": 0.8503,
+      "step": 140110
+    },
+    {
+      "epoch": 1.0142818881336548,
+      "grad_norm": 0.1514463871717453,
+      "learning_rate": 3.985725350532404e-06,
+      "loss": 0.8604,
+      "step": 140120
+    },
+    {
+      "epoch": 1.014354274794241,
+      "grad_norm": 0.15929542481899261,
+      "learning_rate": 3.9856529638718175e-06,
+      "loss": 0.87,
+      "step": 140130
+    },
+    {
+      "epoch": 1.0144266614548272,
+      "grad_norm": 0.1467376947402954,
+      "learning_rate": 3.985580577211232e-06,
+      "loss": 0.8597,
+      "step": 140140
+    },
+    {
+      "epoch": 1.0144990481154132,
+      "grad_norm": 0.14556573331356049,
+      "learning_rate": 3.985508190550646e-06,
+      "loss": 0.8666,
+      "step": 140150
+    },
+    {
+      "epoch": 1.0145714347759995,
+      "grad_norm": 0.15759846568107605,
+      "learning_rate": 3.985435803890059e-06,
+      "loss": 0.8605,
+      "step": 140160
+    },
+    {
+      "epoch": 1.0146438214365856,
+      "grad_norm": 0.16892042756080627,
+      "learning_rate": 3.985363417229473e-06,
+      "loss": 0.8745,
+      "step": 140170
+    },
+    {
+      "epoch": 1.014716208097172,
+      "grad_norm": 0.16491712629795074,
+      "learning_rate": 3.985291030568887e-06,
+      "loss": 0.863,
+      "step": 140180
+    },
+    {
+      "epoch": 1.014788594757758,
+      "grad_norm": 0.14999713003635406,
+      "learning_rate": 3.985218643908301e-06,
+      "loss": 0.8698,
+      "step": 140190
+    },
+    {
+      "epoch": 1.0148609814183442,
+      "grad_norm": 0.15313737094402313,
+      "learning_rate": 3.9851462572477145e-06,
+      "loss": 0.8486,
+      "step": 140200
+    },
+    {
+      "epoch": 1.0149333680789303,
+      "grad_norm": 0.15523867309093475,
+      "learning_rate": 3.985073870587128e-06,
+      "loss": 0.8521,
+      "step": 140210
+    },
+    {
+      "epoch": 1.0150057547395166,
+      "grad_norm": 1.109740972518921,
+      "learning_rate": 3.985001483926543e-06,
+      "loss": 0.8812,
+      "step": 140220
+    },
+    {
+      "epoch": 1.0150781414001029,
+      "grad_norm": 0.15085531771183014,
+      "learning_rate": 3.984929097265956e-06,
+      "loss": 0.8549,
+      "step": 140230
+    },
+    {
+      "epoch": 1.015150528060689,
+      "grad_norm": 0.17226199805736542,
+      "learning_rate": 3.98485671060537e-06,
+      "loss": 0.8643,
+      "step": 140240
+    },
+    {
+      "epoch": 1.0152229147212752,
+      "grad_norm": 0.15006664395332336,
+      "learning_rate": 3.9847843239447835e-06,
+      "loss": 0.8721,
+      "step": 140250
+    },
+    {
+      "epoch": 1.0152953013818613,
+      "grad_norm": 0.1507822424173355,
+      "learning_rate": 3.984711937284197e-06,
+      "loss": 0.8708,
+      "step": 140260
+    },
+    {
+      "epoch": 1.0153676880424476,
+      "grad_norm": 0.1650809794664383,
+      "learning_rate": 3.9846395506236115e-06,
+      "loss": 0.8617,
+      "step": 140270
+    },
+    {
+      "epoch": 1.0154400747030337,
+      "grad_norm": 0.14646883308887482,
+      "learning_rate": 3.984567163963025e-06,
+      "loss": 0.8711,
+      "step": 140280
+    },
+    {
+      "epoch": 1.01551246136362,
+      "grad_norm": 0.16238850355148315,
+      "learning_rate": 3.984494777302439e-06,
+      "loss": 0.8666,
+      "step": 140290
+    },
+    {
+      "epoch": 1.015584848024206,
+      "grad_norm": 0.15635709464550018,
+      "learning_rate": 3.984422390641852e-06,
+      "loss": 0.8715,
+      "step": 140300
+    },
+    {
+      "epoch": 1.0156572346847923,
+      "grad_norm": 0.15769830346107483,
+      "learning_rate": 3.984350003981267e-06,
+      "loss": 0.8673,
+      "step": 140310
+    },
+    {
+      "epoch": 1.0157296213453786,
+      "grad_norm": 0.1550094187259674,
+      "learning_rate": 3.9842776173206805e-06,
+      "loss": 0.8631,
+      "step": 140320
+    },
+    {
+      "epoch": 1.0158020080059647,
+      "grad_norm": 0.16687199473381042,
+      "learning_rate": 3.984205230660094e-06,
+      "loss": 0.8557,
+      "step": 140330
+    },
+    {
+      "epoch": 1.015874394666551,
+      "grad_norm": 0.19312235713005066,
+      "learning_rate": 3.984132843999508e-06,
+      "loss": 0.868,
+      "step": 140340
+    },
+    {
+      "epoch": 1.015946781327137,
+      "grad_norm": 0.15332721173763275,
+      "learning_rate": 3.984060457338922e-06,
+      "loss": 0.8644,
+      "step": 140350
+    },
+    {
+      "epoch": 1.0160191679877233,
+      "grad_norm": 0.15439341962337494,
+      "learning_rate": 3.983988070678336e-06,
+      "loss": 0.8557,
+      "step": 140360
+    },
+    {
+      "epoch": 1.0160915546483094,
+      "grad_norm": 0.140192449092865,
+      "learning_rate": 3.983915684017749e-06,
+      "loss": 0.864,
+      "step": 140370
+    },
+    {
+      "epoch": 1.0161639413088956,
+      "grad_norm": 0.15483497083187103,
+      "learning_rate": 3.983843297357163e-06,
+      "loss": 0.8618,
+      "step": 140380
+    },
+    {
+      "epoch": 1.0162363279694817,
+      "grad_norm": 0.15083526074886322,
+      "learning_rate": 3.9837709106965775e-06,
+      "loss": 0.8493,
+      "step": 140390
+    },
+    {
+      "epoch": 1.016308714630068,
+      "grad_norm": 0.15921801328659058,
+      "learning_rate": 3.983698524035991e-06,
+      "loss": 0.8575,
+      "step": 140400
+    },
+    {
+      "epoch": 1.016381101290654,
+      "grad_norm": 0.15354293584823608,
+      "learning_rate": 3.983626137375405e-06,
+      "loss": 0.8575,
+      "step": 140410
+    },
+    {
+      "epoch": 1.0164534879512404,
+      "grad_norm": 0.159376859664917,
+      "learning_rate": 3.983553750714818e-06,
+      "loss": 0.8627,
+      "step": 140420
+    },
+    {
+      "epoch": 1.0165258746118266,
+      "grad_norm": 0.16422039270401,
+      "learning_rate": 3.983481364054233e-06,
+      "loss": 0.8659,
+      "step": 140430
+    },
+    {
+      "epoch": 1.0165982612724127,
+      "grad_norm": 0.16659992933273315,
+      "learning_rate": 3.983408977393646e-06,
+      "loss": 0.8698,
+      "step": 140440
+    },
+    {
+      "epoch": 1.016670647932999,
+      "grad_norm": 0.155796617269516,
+      "learning_rate": 3.98333659073306e-06,
+      "loss": 0.8626,
+      "step": 140450
+    },
+    {
+      "epoch": 1.016743034593585,
+      "grad_norm": 0.15610454976558685,
+      "learning_rate": 3.983264204072474e-06,
+      "loss": 0.8471,
+      "step": 140460
+    },
+    {
+      "epoch": 1.0168154212541713,
+      "grad_norm": 0.17221376299858093,
+      "learning_rate": 3.983191817411888e-06,
+      "loss": 0.8756,
+      "step": 140470
+    },
+    {
+      "epoch": 1.0168878079147574,
+      "grad_norm": 0.1467757523059845,
+      "learning_rate": 3.983119430751302e-06,
+      "loss": 0.8566,
+      "step": 140480
+    },
+    {
+      "epoch": 1.0169601945753437,
+      "grad_norm": 0.14400814473628998,
+      "learning_rate": 3.983047044090715e-06,
+      "loss": 0.8487,
+      "step": 140490
+    },
+    {
+      "epoch": 1.0170325812359298,
+      "grad_norm": 0.15027961134910583,
+      "learning_rate": 3.982974657430129e-06,
+      "loss": 0.8633,
+      "step": 140500
+    },
+    {
+      "epoch": 1.017104967896516,
+      "grad_norm": 0.14819438755512238,
+      "learning_rate": 3.982902270769543e-06,
+      "loss": 0.854,
+      "step": 140510
+    },
+    {
+      "epoch": 1.0171773545571021,
+      "grad_norm": 0.1465308964252472,
+      "learning_rate": 3.982829884108957e-06,
+      "loss": 0.8568,
+      "step": 140520
+    },
+    {
+      "epoch": 1.0172497412176884,
+      "grad_norm": 0.32729414105415344,
+      "learning_rate": 3.982757497448371e-06,
+      "loss": 0.8546,
+      "step": 140530
+    },
+    {
+      "epoch": 1.0173221278782747,
+      "grad_norm": 0.15239901840686798,
+      "learning_rate": 3.982685110787784e-06,
+      "loss": 0.8547,
+      "step": 140540
+    },
+    {
+      "epoch": 1.0173945145388608,
+      "grad_norm": 0.15983672440052032,
+      "learning_rate": 3.982612724127199e-06,
+      "loss": 0.8539,
+      "step": 140550
+    },
+    {
+      "epoch": 1.017466901199447,
+      "grad_norm": 0.15606176853179932,
+      "learning_rate": 3.982540337466612e-06,
+      "loss": 0.8574,
+      "step": 140560
+    },
+    {
+      "epoch": 1.0175392878600331,
+      "grad_norm": 0.1544933319091797,
+      "learning_rate": 3.982467950806026e-06,
+      "loss": 0.8641,
+      "step": 140570
+    },
+    {
+      "epoch": 1.0176116745206194,
+      "grad_norm": 0.14850009977817535,
+      "learning_rate": 3.9823955641454396e-06,
+      "loss": 0.8581,
+      "step": 140580
+    },
+    {
+      "epoch": 1.0176840611812055,
+      "grad_norm": 0.16443096101284027,
+      "learning_rate": 3.982323177484854e-06,
+      "loss": 0.853,
+      "step": 140590
+    },
+    {
+      "epoch": 1.0177564478417918,
+      "grad_norm": 0.15893018245697021,
+      "learning_rate": 3.982250790824268e-06,
+      "loss": 0.8707,
+      "step": 140600
+    },
+    {
+      "epoch": 1.0178288345023778,
+      "grad_norm": 0.16136851906776428,
+      "learning_rate": 3.982178404163681e-06,
+      "loss": 0.8528,
+      "step": 140610
+    },
+    {
+      "epoch": 1.017901221162964,
+      "grad_norm": 0.14998020231723785,
+      "learning_rate": 3.982106017503095e-06,
+      "loss": 0.8641,
+      "step": 140620
+    },
+    {
+      "epoch": 1.0179736078235502,
+      "grad_norm": 0.16709375381469727,
+      "learning_rate": 3.982033630842509e-06,
+      "loss": 0.8502,
+      "step": 140630
+    },
+    {
+      "epoch": 1.0180459944841365,
+      "grad_norm": 0.15164078772068024,
+      "learning_rate": 3.981961244181923e-06,
+      "loss": 0.8566,
+      "step": 140640
+    },
+    {
+      "epoch": 1.0181183811447228,
+      "grad_norm": 0.16210389137268066,
+      "learning_rate": 3.981888857521336e-06,
+      "loss": 0.8707,
+      "step": 140650
+    },
+    {
+      "epoch": 1.0181907678053088,
+      "grad_norm": 0.15444253385066986,
+      "learning_rate": 3.98181647086075e-06,
+      "loss": 0.8628,
+      "step": 140660
+    },
+    {
+      "epoch": 1.018263154465895,
+      "grad_norm": 0.15437939763069153,
+      "learning_rate": 3.981744084200164e-06,
+      "loss": 0.8504,
+      "step": 140670
+    },
+    {
+      "epoch": 1.0183355411264812,
+      "grad_norm": 0.15111126005649567,
+      "learning_rate": 3.981671697539577e-06,
+      "loss": 0.8609,
+      "step": 140680
+    },
+    {
+      "epoch": 1.0184079277870675,
+      "grad_norm": 0.1448373794555664,
+      "learning_rate": 3.981599310878991e-06,
+      "loss": 0.8588,
+      "step": 140690
+    },
+    {
+      "epoch": 1.0184803144476535,
+      "grad_norm": 0.14929819107055664,
+      "learning_rate": 3.9815269242184055e-06,
+      "loss": 0.868,
+      "step": 140700
+    },
+    {
+      "epoch": 1.0185527011082398,
+      "grad_norm": 0.1808697134256363,
+      "learning_rate": 3.981454537557819e-06,
+      "loss": 0.8635,
+      "step": 140710
+    },
+    {
+      "epoch": 1.0186250877688259,
+      "grad_norm": 0.1492968201637268,
+      "learning_rate": 3.981382150897233e-06,
+      "loss": 0.8852,
+      "step": 140720
+    },
+    {
+      "epoch": 1.0186974744294122,
+      "grad_norm": 0.15429550409317017,
+      "learning_rate": 3.981309764236646e-06,
+      "loss": 0.87,
+      "step": 140730
+    },
+    {
+      "epoch": 1.0187698610899982,
+      "grad_norm": 0.1998952180147171,
+      "learning_rate": 3.981237377576061e-06,
+      "loss": 0.8681,
+      "step": 140740
+    },
+    {
+      "epoch": 1.0188422477505845,
+      "grad_norm": 0.14868871867656708,
+      "learning_rate": 3.9811649909154744e-06,
+      "loss": 0.8709,
+      "step": 140750
+    },
+    {
+      "epoch": 1.0189146344111708,
+      "grad_norm": 0.16644510626792908,
+      "learning_rate": 3.981092604254888e-06,
+      "loss": 0.8687,
+      "step": 140760
+    },
+    {
+      "epoch": 1.0189870210717569,
+      "grad_norm": 0.14398843050003052,
+      "learning_rate": 3.981020217594302e-06,
+      "loss": 0.8643,
+      "step": 140770
+    },
+    {
+      "epoch": 1.0190594077323432,
+      "grad_norm": 0.14528527855873108,
+      "learning_rate": 3.980947830933716e-06,
+      "loss": 0.8783,
+      "step": 140780
+    },
+    {
+      "epoch": 1.0191317943929292,
+      "grad_norm": 0.15322700142860413,
+      "learning_rate": 3.98087544427313e-06,
+      "loss": 0.8527,
+      "step": 140790
+    },
+    {
+      "epoch": 1.0192041810535155,
+      "grad_norm": 0.1468924880027771,
+      "learning_rate": 3.980803057612543e-06,
+      "loss": 0.8566,
+      "step": 140800
+    },
+    {
+      "epoch": 1.0192765677141016,
+      "grad_norm": 0.1553269475698471,
+      "learning_rate": 3.980730670951957e-06,
+      "loss": 0.8583,
+      "step": 140810
+    },
+    {
+      "epoch": 1.0193489543746879,
+      "grad_norm": 0.14764392375946045,
+      "learning_rate": 3.9806582842913714e-06,
+      "loss": 0.8567,
+      "step": 140820
+    },
+    {
+      "epoch": 1.019421341035274,
+      "grad_norm": 0.16374526917934418,
+      "learning_rate": 3.980585897630785e-06,
+      "loss": 0.8495,
+      "step": 140830
+    },
+    {
+      "epoch": 1.0194937276958602,
+      "grad_norm": 0.1522345393896103,
+      "learning_rate": 3.980513510970199e-06,
+      "loss": 0.8623,
+      "step": 140840
+    },
+    {
+      "epoch": 1.0195661143564463,
+      "grad_norm": 0.16136972606182098,
+      "learning_rate": 3.980441124309612e-06,
+      "loss": 0.8658,
+      "step": 140850
+    },
+    {
+      "epoch": 1.0196385010170326,
+      "grad_norm": 0.15984009206295013,
+      "learning_rate": 3.980368737649026e-06,
+      "loss": 0.8635,
+      "step": 140860
+    },
+    {
+      "epoch": 1.0197108876776189,
+      "grad_norm": 0.14360426366329193,
+      "learning_rate": 3.98029635098844e-06,
+      "loss": 0.8678,
+      "step": 140870
+    },
+    {
+      "epoch": 1.019783274338205,
+      "grad_norm": 0.17367321252822876,
+      "learning_rate": 3.980223964327854e-06,
+      "loss": 0.869,
+      "step": 140880
+    },
+    {
+      "epoch": 1.0198556609987912,
+      "grad_norm": 0.14711663126945496,
+      "learning_rate": 3.980151577667268e-06,
+      "loss": 0.8577,
+      "step": 140890
+    },
+    {
+      "epoch": 1.0199280476593773,
+      "grad_norm": 0.15861208736896515,
+      "learning_rate": 3.980079191006681e-06,
+      "loss": 0.8562,
+      "step": 140900
+    },
+    {
+      "epoch": 1.0200004343199636,
+      "grad_norm": 0.15030620992183685,
+      "learning_rate": 3.980006804346096e-06,
+      "loss": 0.863,
+      "step": 140910
+    },
+    {
+      "epoch": 1.0200728209805496,
+      "grad_norm": 0.14951910078525543,
+      "learning_rate": 3.979934417685509e-06,
+      "loss": 0.8614,
+      "step": 140920
+    },
+    {
+      "epoch": 1.020145207641136,
+      "grad_norm": 0.1687590330839157,
+      "learning_rate": 3.979862031024923e-06,
+      "loss": 0.878,
+      "step": 140930
+    },
+    {
+      "epoch": 1.020217594301722,
+      "grad_norm": 0.17207682132720947,
+      "learning_rate": 3.9797896443643365e-06,
+      "loss": 0.8576,
+      "step": 140940
+    },
+    {
+      "epoch": 1.0202899809623083,
+      "grad_norm": 0.14084044098854065,
+      "learning_rate": 3.979717257703751e-06,
+      "loss": 0.8559,
+      "step": 140950
+    },
+    {
+      "epoch": 1.0203623676228946,
+      "grad_norm": 0.19523248076438904,
+      "learning_rate": 3.979644871043165e-06,
+      "loss": 0.8645,
+      "step": 140960
+    },
+    {
+      "epoch": 1.0204347542834806,
+      "grad_norm": 0.14832082390785217,
+      "learning_rate": 3.979572484382578e-06,
+      "loss": 0.8652,
+      "step": 140970
+    },
+    {
+      "epoch": 1.020507140944067,
+      "grad_norm": 0.21129783987998962,
+      "learning_rate": 3.979500097721992e-06,
+      "loss": 0.8643,
+      "step": 140980
+    },
+    {
+      "epoch": 1.020579527604653,
+      "grad_norm": 0.16113729774951935,
+      "learning_rate": 3.979427711061406e-06,
+      "loss": 0.8615,
+      "step": 140990
+    },
+    {
+      "epoch": 1.0206519142652393,
+      "grad_norm": 0.15041108429431915,
+      "learning_rate": 3.97935532440082e-06,
+      "loss": 0.8556,
+      "step": 141000
+    },
+    {
+      "epoch": 1.0207243009258253,
+      "grad_norm": 0.1588127315044403,
+      "learning_rate": 3.9792829377402335e-06,
+      "loss": 0.8617,
+      "step": 141010
+    },
+    {
+      "epoch": 1.0207966875864116,
+      "grad_norm": 0.1841687262058258,
+      "learning_rate": 3.979210551079647e-06,
+      "loss": 0.8581,
+      "step": 141020
+    },
+    {
+      "epoch": 1.0208690742469977,
+      "grad_norm": 0.1596136838197708,
+      "learning_rate": 3.979138164419062e-06,
+      "loss": 0.87,
+      "step": 141030
+    },
+    {
+      "epoch": 1.020941460907584,
+      "grad_norm": 0.15191321074962616,
+      "learning_rate": 3.979065777758475e-06,
+      "loss": 0.865,
+      "step": 141040
+    },
+    {
+      "epoch": 1.02101384756817,
+      "grad_norm": 0.14728322625160217,
+      "learning_rate": 3.978993391097889e-06,
+      "loss": 0.8643,
+      "step": 141050
+    },
+    {
+      "epoch": 1.0210862342287563,
+      "grad_norm": 0.15328173339366913,
+      "learning_rate": 3.9789210044373025e-06,
+      "loss": 0.8609,
+      "step": 141060
+    },
+    {
+      "epoch": 1.0211586208893426,
+      "grad_norm": 0.14939600229263306,
+      "learning_rate": 3.978848617776717e-06,
+      "loss": 0.8711,
+      "step": 141070
+    },
+    {
+      "epoch": 1.0212310075499287,
+      "grad_norm": 0.16299229860305786,
+      "learning_rate": 3.9787762311161305e-06,
+      "loss": 0.8558,
+      "step": 141080
+    },
+    {
+      "epoch": 1.021303394210515,
+      "grad_norm": 0.14965036511421204,
+      "learning_rate": 3.978703844455544e-06,
+      "loss": 0.8579,
+      "step": 141090
+    },
+    {
+      "epoch": 1.021375780871101,
+      "grad_norm": 0.1521276980638504,
+      "learning_rate": 3.978631457794958e-06,
+      "loss": 0.8616,
+      "step": 141100
+    },
+    {
+      "epoch": 1.0214481675316873,
+      "grad_norm": 0.17580242455005646,
+      "learning_rate": 3.978559071134372e-06,
+      "loss": 0.8586,
+      "step": 141110
+    },
+    {
+      "epoch": 1.0215205541922734,
+      "grad_norm": 0.15669845044612885,
+      "learning_rate": 3.978486684473786e-06,
+      "loss": 0.8737,
+      "step": 141120
+    },
+    {
+      "epoch": 1.0215929408528597,
+      "grad_norm": 0.1495697796344757,
+      "learning_rate": 3.9784142978131995e-06,
+      "loss": 0.8624,
+      "step": 141130
+    },
+    {
+      "epoch": 1.0216653275134457,
+      "grad_norm": 0.16285309195518494,
+      "learning_rate": 3.978341911152613e-06,
+      "loss": 0.8776,
+      "step": 141140
+    },
+    {
+      "epoch": 1.021737714174032,
+      "grad_norm": 0.14582541584968567,
+      "learning_rate": 3.9782695244920275e-06,
+      "loss": 0.8666,
+      "step": 141150
+    },
+    {
+      "epoch": 1.021810100834618,
+      "grad_norm": 0.15949222445487976,
+      "learning_rate": 3.978197137831441e-06,
+      "loss": 0.8504,
+      "step": 141160
+    },
+    {
+      "epoch": 1.0218824874952044,
+      "grad_norm": 0.1575903445482254,
+      "learning_rate": 3.978124751170855e-06,
+      "loss": 0.8695,
+      "step": 141170
+    },
+    {
+      "epoch": 1.0219548741557907,
+      "grad_norm": 0.17722275853157043,
+      "learning_rate": 3.978052364510268e-06,
+      "loss": 0.8657,
+      "step": 141180
+    },
+    {
+      "epoch": 1.0220272608163767,
+      "grad_norm": 0.14309902489185333,
+      "learning_rate": 3.977979977849682e-06,
+      "loss": 0.8619,
+      "step": 141190
+    },
+    {
+      "epoch": 1.022099647476963,
+      "grad_norm": 0.14356312155723572,
+      "learning_rate": 3.977907591189096e-06,
+      "loss": 0.8674,
+      "step": 141200
+    },
+    {
+      "epoch": 1.022172034137549,
+      "grad_norm": 0.15702299773693085,
+      "learning_rate": 3.977835204528509e-06,
+      "loss": 0.8666,
+      "step": 141210
+    },
+    {
+      "epoch": 1.0222444207981354,
+      "grad_norm": 0.16147883236408234,
+      "learning_rate": 3.977762817867924e-06,
+      "loss": 0.8675,
+      "step": 141220
+    },
+    {
+      "epoch": 1.0223168074587214,
+      "grad_norm": 0.1556180715560913,
+      "learning_rate": 3.977690431207337e-06,
+      "loss": 0.8506,
+      "step": 141230
+    },
+    {
+      "epoch": 1.0223891941193077,
+      "grad_norm": 0.17178305983543396,
+      "learning_rate": 3.977618044546751e-06,
+      "loss": 0.8649,
+      "step": 141240
+    },
+    {
+      "epoch": 1.0224615807798938,
+      "grad_norm": 0.16344556212425232,
+      "learning_rate": 3.9775456578861646e-06,
+      "loss": 0.8561,
+      "step": 141250
+    },
+    {
+      "epoch": 1.02253396744048,
+      "grad_norm": 0.20177888870239258,
+      "learning_rate": 3.977473271225579e-06,
+      "loss": 0.8556,
+      "step": 141260
+    },
+    {
+      "epoch": 1.0226063541010662,
+      "grad_norm": 0.16756558418273926,
+      "learning_rate": 3.977400884564993e-06,
+      "loss": 0.8527,
+      "step": 141270
+    },
+    {
+      "epoch": 1.0226787407616524,
+      "grad_norm": 0.14939232170581818,
+      "learning_rate": 3.977328497904406e-06,
+      "loss": 0.8596,
+      "step": 141280
+    },
+    {
+      "epoch": 1.0227511274222387,
+      "grad_norm": 0.14247731864452362,
+      "learning_rate": 3.97725611124382e-06,
+      "loss": 0.8656,
+      "step": 141290
+    },
+    {
+      "epoch": 1.0228235140828248,
+      "grad_norm": 0.18383385241031647,
+      "learning_rate": 3.977183724583234e-06,
+      "loss": 0.86,
+      "step": 141300
+    },
+    {
+      "epoch": 1.022895900743411,
+      "grad_norm": 0.16358555853366852,
+      "learning_rate": 3.977111337922648e-06,
+      "loss": 0.873,
+      "step": 141310
+    },
+    {
+      "epoch": 1.0229682874039971,
+      "grad_norm": 0.1494084596633911,
+      "learning_rate": 3.9770389512620616e-06,
+      "loss": 0.8498,
+      "step": 141320
+    },
+    {
+      "epoch": 1.0230406740645834,
+      "grad_norm": 0.14796346426010132,
+      "learning_rate": 3.976966564601475e-06,
+      "loss": 0.8665,
+      "step": 141330
+    },
+    {
+      "epoch": 1.0231130607251695,
+      "grad_norm": 0.2109985500574112,
+      "learning_rate": 3.97689417794089e-06,
+      "loss": 0.8533,
+      "step": 141340
+    },
+    {
+      "epoch": 1.0231854473857558,
+      "grad_norm": 0.14924803376197815,
+      "learning_rate": 3.976821791280303e-06,
+      "loss": 0.867,
+      "step": 141350
+    },
+    {
+      "epoch": 1.0232578340463419,
+      "grad_norm": 0.16223694384098053,
+      "learning_rate": 3.976749404619717e-06,
+      "loss": 0.87,
+      "step": 141360
+    },
+    {
+      "epoch": 1.0233302207069281,
+      "grad_norm": 0.14216773211956024,
+      "learning_rate": 3.9766770179591305e-06,
+      "loss": 0.8584,
+      "step": 141370
+    },
+    {
+      "epoch": 1.0234026073675142,
+      "grad_norm": 0.15058837831020355,
+      "learning_rate": 3.976604631298545e-06,
+      "loss": 0.8655,
+      "step": 141380
+    },
+    {
+      "epoch": 1.0234749940281005,
+      "grad_norm": 0.14986705780029297,
+      "learning_rate": 3.9765322446379586e-06,
+      "loss": 0.8506,
+      "step": 141390
+    },
+    {
+      "epoch": 1.0235473806886868,
+      "grad_norm": 0.1716388761997223,
+      "learning_rate": 3.976459857977372e-06,
+      "loss": 0.8623,
+      "step": 141400
+    },
+    {
+      "epoch": 1.0236197673492728,
+      "grad_norm": 0.1996438354253769,
+      "learning_rate": 3.976387471316786e-06,
+      "loss": 0.8684,
+      "step": 141410
+    },
+    {
+      "epoch": 1.0236921540098591,
+      "grad_norm": 0.14589093625545502,
+      "learning_rate": 3.9763150846562e-06,
+      "loss": 0.8659,
+      "step": 141420
+    },
+    {
+      "epoch": 1.0237645406704452,
+      "grad_norm": 0.1426560878753662,
+      "learning_rate": 3.976242697995614e-06,
+      "loss": 0.8633,
+      "step": 141430
+    },
+    {
+      "epoch": 1.0238369273310315,
+      "grad_norm": 0.21811816096305847,
+      "learning_rate": 3.9761703113350275e-06,
+      "loss": 0.857,
+      "step": 141440
+    },
+    {
+      "epoch": 1.0239093139916176,
+      "grad_norm": 0.13757973909378052,
+      "learning_rate": 3.976097924674441e-06,
+      "loss": 0.8659,
+      "step": 141450
+    },
+    {
+      "epoch": 1.0239817006522038,
+      "grad_norm": 0.15585637092590332,
+      "learning_rate": 3.9760255380138556e-06,
+      "loss": 0.8663,
+      "step": 141460
+    },
+    {
+      "epoch": 1.02405408731279,
+      "grad_norm": 0.1470966935157776,
+      "learning_rate": 3.975953151353269e-06,
+      "loss": 0.8585,
+      "step": 141470
+    },
+    {
+      "epoch": 1.0241264739733762,
+      "grad_norm": 0.15229181945323944,
+      "learning_rate": 3.975880764692683e-06,
+      "loss": 0.8514,
+      "step": 141480
+    },
+    {
+      "epoch": 1.0241988606339625,
+      "grad_norm": 0.1502394825220108,
+      "learning_rate": 3.9758083780320964e-06,
+      "loss": 0.8598,
+      "step": 141490
+    },
+    {
+      "epoch": 1.0242712472945485,
+      "grad_norm": 0.1439654380083084,
+      "learning_rate": 3.97573599137151e-06,
+      "loss": 0.8768,
+      "step": 141500
+    },
+    {
+      "epoch": 1.0243436339551348,
+      "grad_norm": 0.14857643842697144,
+      "learning_rate": 3.9756636047109245e-06,
+      "loss": 0.871,
+      "step": 141510
+    },
+    {
+      "epoch": 1.024416020615721,
+      "grad_norm": 0.1647205948829651,
+      "learning_rate": 3.975591218050338e-06,
+      "loss": 0.8624,
+      "step": 141520
+    },
+    {
+      "epoch": 1.0244884072763072,
+      "grad_norm": 0.16993491351604462,
+      "learning_rate": 3.975518831389752e-06,
+      "loss": 0.8779,
+      "step": 141530
+    },
+    {
+      "epoch": 1.0245607939368933,
+      "grad_norm": 0.16677454113960266,
+      "learning_rate": 3.975446444729165e-06,
+      "loss": 0.861,
+      "step": 141540
+    },
+    {
+      "epoch": 1.0246331805974795,
+      "grad_norm": 0.15071935951709747,
+      "learning_rate": 3.97537405806858e-06,
+      "loss": 0.87,
+      "step": 141550
+    },
+    {
+      "epoch": 1.0247055672580656,
+      "grad_norm": 0.15021325647830963,
+      "learning_rate": 3.9753016714079934e-06,
+      "loss": 0.8523,
+      "step": 141560
+    },
+    {
+      "epoch": 1.024777953918652,
+      "grad_norm": 0.14703767001628876,
+      "learning_rate": 3.975229284747407e-06,
+      "loss": 0.8572,
+      "step": 141570
+    },
+    {
+      "epoch": 1.024850340579238,
+      "grad_norm": 0.17009884119033813,
+      "learning_rate": 3.975156898086821e-06,
+      "loss": 0.8612,
+      "step": 141580
+    },
+    {
+      "epoch": 1.0249227272398242,
+      "grad_norm": 0.21612617373466492,
+      "learning_rate": 3.975084511426235e-06,
+      "loss": 0.8604,
+      "step": 141590
+    },
+    {
+      "epoch": 1.0249951139004105,
+      "grad_norm": 0.1644405573606491,
+      "learning_rate": 3.975012124765649e-06,
+      "loss": 0.858,
+      "step": 141600
+    },
+    {
+      "epoch": 1.0250675005609966,
+      "grad_norm": 0.1592147797346115,
+      "learning_rate": 3.974939738105062e-06,
+      "loss": 0.8701,
+      "step": 141610
+    },
+    {
+      "epoch": 1.025139887221583,
+      "grad_norm": 0.1543736755847931,
+      "learning_rate": 3.974867351444476e-06,
+      "loss": 0.851,
+      "step": 141620
+    },
+    {
+      "epoch": 1.025212273882169,
+      "grad_norm": 0.15851731598377228,
+      "learning_rate": 3.9747949647838904e-06,
+      "loss": 0.8625,
+      "step": 141630
+    },
+    {
+      "epoch": 1.0252846605427552,
+      "grad_norm": 0.1598898321390152,
+      "learning_rate": 3.974722578123304e-06,
+      "loss": 0.8513,
+      "step": 141640
+    },
+    {
+      "epoch": 1.0253570472033413,
+      "grad_norm": 0.1485353261232376,
+      "learning_rate": 3.974650191462718e-06,
+      "loss": 0.8666,
+      "step": 141650
+    },
+    {
+      "epoch": 1.0254294338639276,
+      "grad_norm": 0.15137670934200287,
+      "learning_rate": 3.974577804802131e-06,
+      "loss": 0.8551,
+      "step": 141660
+    },
+    {
+      "epoch": 1.0255018205245137,
+      "grad_norm": 0.14559021592140198,
+      "learning_rate": 3.974505418141546e-06,
+      "loss": 0.8602,
+      "step": 141670
+    },
+    {
+      "epoch": 1.0255742071851,
+      "grad_norm": 0.15725764632225037,
+      "learning_rate": 3.974433031480959e-06,
+      "loss": 0.8529,
+      "step": 141680
+    },
+    {
+      "epoch": 1.025646593845686,
+      "grad_norm": 0.15133067965507507,
+      "learning_rate": 3.974360644820373e-06,
+      "loss": 0.8677,
+      "step": 141690
+    },
+    {
+      "epoch": 1.0257189805062723,
+      "grad_norm": 0.16057194769382477,
+      "learning_rate": 3.974288258159787e-06,
+      "loss": 0.8644,
+      "step": 141700
+    },
+    {
+      "epoch": 1.0257913671668586,
+      "grad_norm": 0.14923806488513947,
+      "learning_rate": 3.9742158714992e-06,
+      "loss": 0.8539,
+      "step": 141710
+    },
+    {
+      "epoch": 1.0258637538274447,
+      "grad_norm": 0.1376243233680725,
+      "learning_rate": 3.974143484838614e-06,
+      "loss": 0.8581,
+      "step": 141720
+    },
+    {
+      "epoch": 1.025936140488031,
+      "grad_norm": 0.15476512908935547,
+      "learning_rate": 3.9740710981780274e-06,
+      "loss": 0.8533,
+      "step": 141730
+    },
+    {
+      "epoch": 1.026008527148617,
+      "grad_norm": 0.16103792190551758,
+      "learning_rate": 3.973998711517442e-06,
+      "loss": 0.8599,
+      "step": 141740
+    },
+    {
+      "epoch": 1.0260809138092033,
+      "grad_norm": 0.1493249237537384,
+      "learning_rate": 3.9739263248568555e-06,
+      "loss": 0.8548,
+      "step": 141750
+    },
+    {
+      "epoch": 1.0261533004697894,
+      "grad_norm": 0.1508176475763321,
+      "learning_rate": 3.973853938196269e-06,
+      "loss": 0.8681,
+      "step": 141760
+    },
+    {
+      "epoch": 1.0262256871303757,
+      "grad_norm": 0.16366364061832428,
+      "learning_rate": 3.973781551535683e-06,
+      "loss": 0.8723,
+      "step": 141770
+    },
+    {
+      "epoch": 1.0262980737909617,
+      "grad_norm": 0.1778716742992401,
+      "learning_rate": 3.973709164875097e-06,
+      "loss": 0.87,
+      "step": 141780
+    },
+    {
+      "epoch": 1.026370460451548,
+      "grad_norm": 0.1537426859140396,
+      "learning_rate": 3.973636778214511e-06,
+      "loss": 0.8588,
+      "step": 141790
+    },
+    {
+      "epoch": 1.026442847112134,
+      "grad_norm": 0.15936316549777985,
+      "learning_rate": 3.9735643915539245e-06,
+      "loss": 0.8524,
+      "step": 141800
+    },
+    {
+      "epoch": 1.0265152337727204,
+      "grad_norm": 0.15303193032741547,
+      "learning_rate": 3.973492004893338e-06,
+      "loss": 0.8655,
+      "step": 141810
+    },
+    {
+      "epoch": 1.0265876204333066,
+      "grad_norm": 0.1577739119529724,
+      "learning_rate": 3.9734196182327525e-06,
+      "loss": 0.8629,
+      "step": 141820
+    },
+    {
+      "epoch": 1.0266600070938927,
+      "grad_norm": 0.15587298572063446,
+      "learning_rate": 3.973347231572166e-06,
+      "loss": 0.8514,
+      "step": 141830
+    },
+    {
+      "epoch": 1.026732393754479,
+      "grad_norm": 0.15615636110305786,
+      "learning_rate": 3.97327484491158e-06,
+      "loss": 0.8699,
+      "step": 141840
+    },
+    {
+      "epoch": 1.026804780415065,
+      "grad_norm": 0.15278807282447815,
+      "learning_rate": 3.973202458250993e-06,
+      "loss": 0.8551,
+      "step": 141850
+    },
+    {
+      "epoch": 1.0268771670756514,
+      "grad_norm": 0.1699153482913971,
+      "learning_rate": 3.973130071590408e-06,
+      "loss": 0.8701,
+      "step": 141860
+    },
+    {
+      "epoch": 1.0269495537362374,
+      "grad_norm": 0.1508774608373642,
+      "learning_rate": 3.9730576849298215e-06,
+      "loss": 0.8502,
+      "step": 141870
+    },
+    {
+      "epoch": 1.0270219403968237,
+      "grad_norm": 0.1427360475063324,
+      "learning_rate": 3.972985298269235e-06,
+      "loss": 0.8529,
+      "step": 141880
+    },
+    {
+      "epoch": 1.0270943270574098,
+      "grad_norm": 0.15359556674957275,
+      "learning_rate": 3.972912911608649e-06,
+      "loss": 0.8513,
+      "step": 141890
+    },
+    {
+      "epoch": 1.027166713717996,
+      "grad_norm": 0.15233898162841797,
+      "learning_rate": 3.972840524948063e-06,
+      "loss": 0.8758,
+      "step": 141900
+    },
+    {
+      "epoch": 1.0272391003785821,
+      "grad_norm": 0.1512978971004486,
+      "learning_rate": 3.972768138287477e-06,
+      "loss": 0.8616,
+      "step": 141910
+    },
+    {
+      "epoch": 1.0273114870391684,
+      "grad_norm": 0.15592820942401886,
+      "learning_rate": 3.97269575162689e-06,
+      "loss": 0.8621,
+      "step": 141920
+    },
+    {
+      "epoch": 1.0273838736997547,
+      "grad_norm": 0.15350952744483948,
+      "learning_rate": 3.972623364966304e-06,
+      "loss": 0.8509,
+      "step": 141930
+    },
+    {
+      "epoch": 1.0274562603603408,
+      "grad_norm": 0.16110540926456451,
+      "learning_rate": 3.9725509783057185e-06,
+      "loss": 0.8634,
+      "step": 141940
+    },
+    {
+      "epoch": 1.027528647020927,
+      "grad_norm": 0.15037639439105988,
+      "learning_rate": 3.972478591645132e-06,
+      "loss": 0.8537,
+      "step": 141950
+    },
+    {
+      "epoch": 1.0276010336815131,
+      "grad_norm": 0.16510790586471558,
+      "learning_rate": 3.972406204984546e-06,
+      "loss": 0.8515,
+      "step": 141960
+    },
+    {
+      "epoch": 1.0276734203420994,
+      "grad_norm": 0.16185833513736725,
+      "learning_rate": 3.972333818323959e-06,
+      "loss": 0.8547,
+      "step": 141970
+    },
+    {
+      "epoch": 1.0277458070026855,
+      "grad_norm": 0.14903537929058075,
+      "learning_rate": 3.972261431663374e-06,
+      "loss": 0.8765,
+      "step": 141980
+    },
+    {
+      "epoch": 1.0278181936632718,
+      "grad_norm": 0.14744271337985992,
+      "learning_rate": 3.972189045002787e-06,
+      "loss": 0.8588,
+      "step": 141990
+    },
+    {
+      "epoch": 1.0278905803238578,
+      "grad_norm": 0.16065943241119385,
+      "learning_rate": 3.972116658342201e-06,
+      "loss": 0.8532,
+      "step": 142000
+    },
+    {
+      "epoch": 1.0279629669844441,
+      "grad_norm": 0.15768252313137054,
+      "learning_rate": 3.972044271681615e-06,
+      "loss": 0.8627,
+      "step": 142010
+    },
+    {
+      "epoch": 1.0280353536450304,
+      "grad_norm": 0.16931724548339844,
+      "learning_rate": 3.971971885021029e-06,
+      "loss": 0.8667,
+      "step": 142020
+    },
+    {
+      "epoch": 1.0281077403056165,
+      "grad_norm": 0.14971376955509186,
+      "learning_rate": 3.971899498360443e-06,
+      "loss": 0.8592,
+      "step": 142030
+    },
+    {
+      "epoch": 1.0281801269662028,
+      "grad_norm": 0.1479533463716507,
+      "learning_rate": 3.971827111699856e-06,
+      "loss": 0.8577,
+      "step": 142040
+    },
+    {
+      "epoch": 1.0282525136267888,
+      "grad_norm": 0.15381501615047455,
+      "learning_rate": 3.97175472503927e-06,
+      "loss": 0.8658,
+      "step": 142050
+    },
+    {
+      "epoch": 1.028324900287375,
+      "grad_norm": 0.1562289297580719,
+      "learning_rate": 3.971682338378684e-06,
+      "loss": 0.8571,
+      "step": 142060
+    },
+    {
+      "epoch": 1.0283972869479612,
+      "grad_norm": 0.15410132706165314,
+      "learning_rate": 3.971609951718098e-06,
+      "loss": 0.8519,
+      "step": 142070
+    },
+    {
+      "epoch": 1.0284696736085475,
+      "grad_norm": 0.19577772915363312,
+      "learning_rate": 3.971537565057512e-06,
+      "loss": 0.8558,
+      "step": 142080
+    },
+    {
+      "epoch": 1.0285420602691335,
+      "grad_norm": 0.15742763876914978,
+      "learning_rate": 3.971465178396925e-06,
+      "loss": 0.8577,
+      "step": 142090
+    },
+    {
+      "epoch": 1.0286144469297198,
+      "grad_norm": 0.16709017753601074,
+      "learning_rate": 3.971392791736339e-06,
+      "loss": 0.865,
+      "step": 142100
+    },
+    {
+      "epoch": 1.0286868335903059,
+      "grad_norm": 0.15220534801483154,
+      "learning_rate": 3.971320405075753e-06,
+      "loss": 0.8565,
+      "step": 142110
+    },
+    {
+      "epoch": 1.0287592202508922,
+      "grad_norm": 0.18524552881717682,
+      "learning_rate": 3.971248018415167e-06,
+      "loss": 0.8583,
+      "step": 142120
+    },
+    {
+      "epoch": 1.0288316069114785,
+      "grad_norm": 0.1528216153383255,
+      "learning_rate": 3.9711756317545806e-06,
+      "loss": 0.8781,
+      "step": 142130
+    },
+    {
+      "epoch": 1.0289039935720645,
+      "grad_norm": 0.15787044167518616,
+      "learning_rate": 3.971103245093994e-06,
+      "loss": 0.861,
+      "step": 142140
+    },
+    {
+      "epoch": 1.0289763802326508,
+      "grad_norm": 0.15298911929130554,
+      "learning_rate": 3.971030858433409e-06,
+      "loss": 0.8647,
+      "step": 142150
+    },
+    {
+      "epoch": 1.0290487668932369,
+      "grad_norm": 0.15209560096263885,
+      "learning_rate": 3.970958471772822e-06,
+      "loss": 0.8553,
+      "step": 142160
+    },
+    {
+      "epoch": 1.0291211535538232,
+      "grad_norm": 0.14856834709644318,
+      "learning_rate": 3.970886085112236e-06,
+      "loss": 0.8556,
+      "step": 142170
+    },
+    {
+      "epoch": 1.0291935402144092,
+      "grad_norm": 0.14934612810611725,
+      "learning_rate": 3.9708136984516495e-06,
+      "loss": 0.867,
+      "step": 142180
+    },
+    {
+      "epoch": 1.0292659268749955,
+      "grad_norm": 0.1754796802997589,
+      "learning_rate": 3.970741311791064e-06,
+      "loss": 0.8633,
+      "step": 142190
+    },
+    {
+      "epoch": 1.0293383135355816,
+      "grad_norm": 0.15860222280025482,
+      "learning_rate": 3.9706689251304776e-06,
+      "loss": 0.8635,
+      "step": 142200
+    },
+    {
+      "epoch": 1.0294107001961679,
+      "grad_norm": 0.16497546434402466,
+      "learning_rate": 3.970596538469891e-06,
+      "loss": 0.8575,
+      "step": 142210
+    },
+    {
+      "epoch": 1.029483086856754,
+      "grad_norm": 0.20866011083126068,
+      "learning_rate": 3.970524151809305e-06,
+      "loss": 0.8699,
+      "step": 142220
+    },
+    {
+      "epoch": 1.0295554735173402,
+      "grad_norm": 0.17648346722126007,
+      "learning_rate": 3.970451765148719e-06,
+      "loss": 0.8578,
+      "step": 142230
+    },
+    {
+      "epoch": 1.0296278601779265,
+      "grad_norm": 0.1537626087665558,
+      "learning_rate": 3.970379378488132e-06,
+      "loss": 0.8485,
+      "step": 142240
+    },
+    {
+      "epoch": 1.0297002468385126,
+      "grad_norm": 0.14427883923053741,
+      "learning_rate": 3.9703069918275465e-06,
+      "loss": 0.856,
+      "step": 142250
+    },
+    {
+      "epoch": 1.0297726334990989,
+      "grad_norm": 0.1432482898235321,
+      "learning_rate": 3.97023460516696e-06,
+      "loss": 0.8569,
+      "step": 142260
+    },
+    {
+      "epoch": 1.029845020159685,
+      "grad_norm": 0.14573949575424194,
+      "learning_rate": 3.970162218506374e-06,
+      "loss": 0.8622,
+      "step": 142270
+    },
+    {
+      "epoch": 1.0299174068202712,
+      "grad_norm": 0.1648380309343338,
+      "learning_rate": 3.970089831845787e-06,
+      "loss": 0.8654,
+      "step": 142280
+    },
+    {
+      "epoch": 1.0299897934808573,
+      "grad_norm": 0.16308258473873138,
+      "learning_rate": 3.970017445185201e-06,
+      "loss": 0.8474,
+      "step": 142290
+    },
+    {
+      "epoch": 1.0300621801414436,
+      "grad_norm": 0.1557963788509369,
+      "learning_rate": 3.9699450585246154e-06,
+      "loss": 0.8688,
+      "step": 142300
+    },
+    {
+      "epoch": 1.0301345668020296,
+      "grad_norm": 0.16031388938426971,
+      "learning_rate": 3.969872671864029e-06,
+      "loss": 0.8593,
+      "step": 142310
+    },
+    {
+      "epoch": 1.030206953462616,
+      "grad_norm": 0.17652627825737,
+      "learning_rate": 3.969800285203443e-06,
+      "loss": 0.8639,
+      "step": 142320
+    },
+    {
+      "epoch": 1.030279340123202,
+      "grad_norm": 0.1523296982049942,
+      "learning_rate": 3.969727898542856e-06,
+      "loss": 0.8444,
+      "step": 142330
+    },
+    {
+      "epoch": 1.0303517267837883,
+      "grad_norm": 0.16042089462280273,
+      "learning_rate": 3.969655511882271e-06,
+      "loss": 0.8583,
+      "step": 142340
+    },
+    {
+      "epoch": 1.0304241134443746,
+      "grad_norm": 0.15083900094032288,
+      "learning_rate": 3.969583125221684e-06,
+      "loss": 0.8611,
+      "step": 142350
+    },
+    {
+      "epoch": 1.0304965001049606,
+      "grad_norm": 0.16533105075359344,
+      "learning_rate": 3.969510738561098e-06,
+      "loss": 0.8657,
+      "step": 142360
+    },
+    {
+      "epoch": 1.030568886765547,
+      "grad_norm": 0.16816429793834686,
+      "learning_rate": 3.969438351900512e-06,
+      "loss": 0.8586,
+      "step": 142370
+    },
+    {
+      "epoch": 1.030641273426133,
+      "grad_norm": 0.15156440436840057,
+      "learning_rate": 3.969365965239926e-06,
+      "loss": 0.8609,
+      "step": 142380
+    },
+    {
+      "epoch": 1.0307136600867193,
+      "grad_norm": 0.15166370570659637,
+      "learning_rate": 3.96929357857934e-06,
+      "loss": 0.8651,
+      "step": 142390
+    },
+    {
+      "epoch": 1.0307860467473053,
+      "grad_norm": 0.15877971053123474,
+      "learning_rate": 3.969221191918753e-06,
+      "loss": 0.8612,
+      "step": 142400
+    },
+    {
+      "epoch": 1.0308584334078916,
+      "grad_norm": 0.15745185315608978,
+      "learning_rate": 3.969148805258167e-06,
+      "loss": 0.8582,
+      "step": 142410
+    },
+    {
+      "epoch": 1.0309308200684777,
+      "grad_norm": 0.16884928941726685,
+      "learning_rate": 3.969076418597581e-06,
+      "loss": 0.8673,
+      "step": 142420
+    },
+    {
+      "epoch": 1.031003206729064,
+      "grad_norm": 0.14777499437332153,
+      "learning_rate": 3.969004031936995e-06,
+      "loss": 0.8598,
+      "step": 142430
+    },
+    {
+      "epoch": 1.03107559338965,
+      "grad_norm": 0.17125903069972992,
+      "learning_rate": 3.968931645276409e-06,
+      "loss": 0.8579,
+      "step": 142440
+    },
+    {
+      "epoch": 1.0311479800502363,
+      "grad_norm": 0.18040338158607483,
+      "learning_rate": 3.968859258615822e-06,
+      "loss": 0.8556,
+      "step": 142450
+    },
+    {
+      "epoch": 1.0312203667108226,
+      "grad_norm": 0.14645922183990479,
+      "learning_rate": 3.968786871955237e-06,
+      "loss": 0.8515,
+      "step": 142460
+    },
+    {
+      "epoch": 1.0312927533714087,
+      "grad_norm": 0.17631933093070984,
+      "learning_rate": 3.96871448529465e-06,
+      "loss": 0.8559,
+      "step": 142470
+    },
+    {
+      "epoch": 1.031365140031995,
+      "grad_norm": 0.19517827033996582,
+      "learning_rate": 3.968642098634064e-06,
+      "loss": 0.8675,
+      "step": 142480
+    },
+    {
+      "epoch": 1.031437526692581,
+      "grad_norm": 0.5200027227401733,
+      "learning_rate": 3.9685697119734775e-06,
+      "loss": 0.8619,
+      "step": 142490
+    },
+    {
+      "epoch": 1.0315099133531673,
+      "grad_norm": 0.15519946813583374,
+      "learning_rate": 3.968497325312892e-06,
+      "loss": 0.8543,
+      "step": 142500
+    },
+    {
+      "epoch": 1.0315823000137534,
+      "grad_norm": 0.147262305021286,
+      "learning_rate": 3.968424938652306e-06,
+      "loss": 0.8527,
+      "step": 142510
+    },
+    {
+      "epoch": 1.0316546866743397,
+      "grad_norm": 0.15960846841335297,
+      "learning_rate": 3.968352551991719e-06,
+      "loss": 0.8642,
+      "step": 142520
+    },
+    {
+      "epoch": 1.0317270733349257,
+      "grad_norm": 0.16142988204956055,
+      "learning_rate": 3.968280165331133e-06,
+      "loss": 0.853,
+      "step": 142530
+    },
+    {
+      "epoch": 1.031799459995512,
+      "grad_norm": 0.14241138100624084,
+      "learning_rate": 3.968207778670547e-06,
+      "loss": 0.8441,
+      "step": 142540
+    },
+    {
+      "epoch": 1.0318718466560983,
+      "grad_norm": 0.15053856372833252,
+      "learning_rate": 3.968135392009961e-06,
+      "loss": 0.8477,
+      "step": 142550
+    },
+    {
+      "epoch": 1.0319442333166844,
+      "grad_norm": 0.14978879690170288,
+      "learning_rate": 3.9680630053493745e-06,
+      "loss": 0.852,
+      "step": 142560
+    },
+    {
+      "epoch": 1.0320166199772707,
+      "grad_norm": 0.15513886511325836,
+      "learning_rate": 3.967990618688788e-06,
+      "loss": 0.8575,
+      "step": 142570
+    },
+    {
+      "epoch": 1.0320890066378567,
+      "grad_norm": 0.14662916958332062,
+      "learning_rate": 3.967918232028203e-06,
+      "loss": 0.8626,
+      "step": 142580
+    },
+    {
+      "epoch": 1.032161393298443,
+      "grad_norm": 0.15396258234977722,
+      "learning_rate": 3.967845845367616e-06,
+      "loss": 0.8612,
+      "step": 142590
+    },
+    {
+      "epoch": 1.032233779959029,
+      "grad_norm": 0.1600462794303894,
+      "learning_rate": 3.96777345870703e-06,
+      "loss": 0.873,
+      "step": 142600
+    },
+    {
+      "epoch": 1.0323061666196154,
+      "grad_norm": 0.15220437943935394,
+      "learning_rate": 3.9677010720464435e-06,
+      "loss": 0.8591,
+      "step": 142610
+    },
+    {
+      "epoch": 1.0323785532802014,
+      "grad_norm": 0.14761093258857727,
+      "learning_rate": 3.967628685385858e-06,
+      "loss": 0.8484,
+      "step": 142620
+    },
+    {
+      "epoch": 1.0324509399407877,
+      "grad_norm": 0.15190240740776062,
+      "learning_rate": 3.9675562987252715e-06,
+      "loss": 0.8556,
+      "step": 142630
+    },
+    {
+      "epoch": 1.0325233266013738,
+      "grad_norm": 0.15413150191307068,
+      "learning_rate": 3.967483912064685e-06,
+      "loss": 0.8524,
+      "step": 142640
+    },
+    {
+      "epoch": 1.03259571326196,
+      "grad_norm": 0.16066095232963562,
+      "learning_rate": 3.967411525404099e-06,
+      "loss": 0.8588,
+      "step": 142650
+    },
+    {
+      "epoch": 1.0326680999225464,
+      "grad_norm": 0.1533208042383194,
+      "learning_rate": 3.967339138743513e-06,
+      "loss": 0.8539,
+      "step": 142660
+    },
+    {
+      "epoch": 1.0327404865831324,
+      "grad_norm": 0.1481548249721527,
+      "learning_rate": 3.967266752082927e-06,
+      "loss": 0.8609,
+      "step": 142670
+    },
+    {
+      "epoch": 1.0328128732437187,
+      "grad_norm": 0.16002528369426727,
+      "learning_rate": 3.9671943654223405e-06,
+      "loss": 0.8602,
+      "step": 142680
+    },
+    {
+      "epoch": 1.0328852599043048,
+      "grad_norm": 0.14188747107982635,
+      "learning_rate": 3.967121978761754e-06,
+      "loss": 0.8648,
+      "step": 142690
+    },
+    {
+      "epoch": 1.032957646564891,
+      "grad_norm": 0.1650628298521042,
+      "learning_rate": 3.9670495921011685e-06,
+      "loss": 0.8579,
+      "step": 142700
+    },
+    {
+      "epoch": 1.0330300332254772,
+      "grad_norm": 0.15831433236598969,
+      "learning_rate": 3.966977205440582e-06,
+      "loss": 0.8578,
+      "step": 142710
+    },
+    {
+      "epoch": 1.0331024198860634,
+      "grad_norm": 0.14619334042072296,
+      "learning_rate": 3.966904818779996e-06,
+      "loss": 0.8735,
+      "step": 142720
+    },
+    {
+      "epoch": 1.0331748065466495,
+      "grad_norm": 0.1563662737607956,
+      "learning_rate": 3.966832432119409e-06,
+      "loss": 0.8541,
+      "step": 142730
+    },
+    {
+      "epoch": 1.0332471932072358,
+      "grad_norm": 0.155498206615448,
+      "learning_rate": 3.966760045458823e-06,
+      "loss": 0.8615,
+      "step": 142740
+    },
+    {
+      "epoch": 1.0333195798678219,
+      "grad_norm": 0.1640671342611313,
+      "learning_rate": 3.9666876587982375e-06,
+      "loss": 0.8612,
+      "step": 142750
+    },
+    {
+      "epoch": 1.0333919665284081,
+      "grad_norm": 0.15346024930477142,
+      "learning_rate": 3.966615272137651e-06,
+      "loss": 0.8689,
+      "step": 142760
+    },
+    {
+      "epoch": 1.0334643531889944,
+      "grad_norm": 0.14893843233585358,
+      "learning_rate": 3.966542885477065e-06,
+      "loss": 0.8617,
+      "step": 142770
+    },
+    {
+      "epoch": 1.0335367398495805,
+      "grad_norm": 0.1711651086807251,
+      "learning_rate": 3.966470498816478e-06,
+      "loss": 0.8623,
+      "step": 142780
+    },
+    {
+      "epoch": 1.0336091265101668,
+      "grad_norm": 0.1499413400888443,
+      "learning_rate": 3.966398112155892e-06,
+      "loss": 0.8654,
+      "step": 142790
+    },
+    {
+      "epoch": 1.0336815131707529,
+      "grad_norm": 0.16550101339817047,
+      "learning_rate": 3.9663257254953056e-06,
+      "loss": 0.8597,
+      "step": 142800
+    },
+    {
+      "epoch": 1.0337538998313391,
+      "grad_norm": 0.15514644980430603,
+      "learning_rate": 3.96625333883472e-06,
+      "loss": 0.8546,
+      "step": 142810
+    },
+    {
+      "epoch": 1.0338262864919252,
+      "grad_norm": 0.15630175173282623,
+      "learning_rate": 3.966180952174134e-06,
+      "loss": 0.8613,
+      "step": 142820
+    },
+    {
+      "epoch": 1.0338986731525115,
+      "grad_norm": 0.15911681950092316,
+      "learning_rate": 3.966108565513547e-06,
+      "loss": 0.8679,
+      "step": 142830
+    },
+    {
+      "epoch": 1.0339710598130976,
+      "grad_norm": 0.14707563817501068,
+      "learning_rate": 3.966036178852961e-06,
+      "loss": 0.8566,
+      "step": 142840
+    },
+    {
+      "epoch": 1.0340434464736838,
+      "grad_norm": 0.15566620230674744,
+      "learning_rate": 3.965963792192375e-06,
+      "loss": 0.8664,
+      "step": 142850
+    },
+    {
+      "epoch": 1.03411583313427,
+      "grad_norm": 0.16189688444137573,
+      "learning_rate": 3.965891405531789e-06,
+      "loss": 0.8587,
+      "step": 142860
+    },
+    {
+      "epoch": 1.0341882197948562,
+      "grad_norm": 0.1611325591802597,
+      "learning_rate": 3.9658190188712026e-06,
+      "loss": 0.8551,
+      "step": 142870
+    },
+    {
+      "epoch": 1.0342606064554425,
+      "grad_norm": 0.15380004048347473,
+      "learning_rate": 3.965746632210616e-06,
+      "loss": 0.8503,
+      "step": 142880
+    },
+    {
+      "epoch": 1.0343329931160286,
+      "grad_norm": 0.15085004270076752,
+      "learning_rate": 3.96567424555003e-06,
+      "loss": 0.8579,
+      "step": 142890
+    },
+    {
+      "epoch": 1.0344053797766148,
+      "grad_norm": 0.15508107841014862,
+      "learning_rate": 3.965601858889444e-06,
+      "loss": 0.8644,
+      "step": 142900
+    },
+    {
+      "epoch": 1.034477766437201,
+      "grad_norm": 0.16797524690628052,
+      "learning_rate": 3.965529472228858e-06,
+      "loss": 0.8579,
+      "step": 142910
+    },
+    {
+      "epoch": 1.0345501530977872,
+      "grad_norm": 0.16412454843521118,
+      "learning_rate": 3.9654570855682715e-06,
+      "loss": 0.8653,
+      "step": 142920
+    },
+    {
+      "epoch": 1.0346225397583733,
+      "grad_norm": 0.1560967117547989,
+      "learning_rate": 3.965384698907685e-06,
+      "loss": 0.8617,
+      "step": 142930
+    },
+    {
+      "epoch": 1.0346949264189595,
+      "grad_norm": 0.1702379286289215,
+      "learning_rate": 3.9653123122470996e-06,
+      "loss": 0.8819,
+      "step": 142940
+    },
+    {
+      "epoch": 1.0347673130795456,
+      "grad_norm": 0.16101233661174774,
+      "learning_rate": 3.965239925586513e-06,
+      "loss": 0.8702,
+      "step": 142950
+    },
+    {
+      "epoch": 1.034839699740132,
+      "grad_norm": 0.1507912427186966,
+      "learning_rate": 3.965167538925927e-06,
+      "loss": 0.8644,
+      "step": 142960
+    },
+    {
+      "epoch": 1.034912086400718,
+      "grad_norm": 0.14601871371269226,
+      "learning_rate": 3.96509515226534e-06,
+      "loss": 0.8614,
+      "step": 142970
+    },
+    {
+      "epoch": 1.0349844730613043,
+      "grad_norm": 0.1494961827993393,
+      "learning_rate": 3.965022765604755e-06,
+      "loss": 0.8673,
+      "step": 142980
+    },
+    {
+      "epoch": 1.0350568597218905,
+      "grad_norm": 0.1469757854938507,
+      "learning_rate": 3.9649503789441685e-06,
+      "loss": 0.8586,
+      "step": 142990
+    },
+    {
+      "epoch": 1.0351292463824766,
+      "grad_norm": 0.1592247635126114,
+      "learning_rate": 3.964877992283582e-06,
+      "loss": 0.8657,
+      "step": 143000
+    },
+    {
+      "epoch": 1.035201633043063,
+      "grad_norm": 0.1555202305316925,
+      "learning_rate": 3.964805605622996e-06,
+      "loss": 0.8588,
+      "step": 143010
+    },
+    {
+      "epoch": 1.035274019703649,
+      "grad_norm": 0.15788942575454712,
+      "learning_rate": 3.96473321896241e-06,
+      "loss": 0.8612,
+      "step": 143020
+    },
+    {
+      "epoch": 1.0353464063642352,
+      "grad_norm": 0.15312904119491577,
+      "learning_rate": 3.964660832301824e-06,
+      "loss": 0.8609,
+      "step": 143030
+    },
+    {
+      "epoch": 1.0354187930248213,
+      "grad_norm": 0.1483820229768753,
+      "learning_rate": 3.9645884456412374e-06,
+      "loss": 0.8604,
+      "step": 143040
+    },
+    {
+      "epoch": 1.0354911796854076,
+      "grad_norm": 0.15752215683460236,
+      "learning_rate": 3.964516058980651e-06,
+      "loss": 0.8596,
+      "step": 143050
+    },
+    {
+      "epoch": 1.0355635663459937,
+      "grad_norm": 0.15568195283412933,
+      "learning_rate": 3.9644436723200655e-06,
+      "loss": 0.8665,
+      "step": 143060
+    },
+    {
+      "epoch": 1.03563595300658,
+      "grad_norm": 0.14673691987991333,
+      "learning_rate": 3.964371285659479e-06,
+      "loss": 0.8612,
+      "step": 143070
+    },
+    {
+      "epoch": 1.035708339667166,
+      "grad_norm": 0.15292707085609436,
+      "learning_rate": 3.964298898998893e-06,
+      "loss": 0.8511,
+      "step": 143080
+    },
+    {
+      "epoch": 1.0357807263277523,
+      "grad_norm": 0.1723155379295349,
+      "learning_rate": 3.964226512338306e-06,
+      "loss": 0.8634,
+      "step": 143090
+    },
+    {
+      "epoch": 1.0358531129883386,
+      "grad_norm": 0.15540607273578644,
+      "learning_rate": 3.964154125677721e-06,
+      "loss": 0.8704,
+      "step": 143100
+    },
+    {
+      "epoch": 1.0359254996489247,
+      "grad_norm": 0.14979644119739532,
+      "learning_rate": 3.9640817390171344e-06,
+      "loss": 0.8563,
+      "step": 143110
+    },
+    {
+      "epoch": 1.035997886309511,
+      "grad_norm": 0.1446203589439392,
+      "learning_rate": 3.964009352356548e-06,
+      "loss": 0.8557,
+      "step": 143120
+    },
+    {
+      "epoch": 1.036070272970097,
+      "grad_norm": 0.15356165170669556,
+      "learning_rate": 3.963936965695962e-06,
+      "loss": 0.8653,
+      "step": 143130
+    },
+    {
+      "epoch": 1.0361426596306833,
+      "grad_norm": 0.1651306450366974,
+      "learning_rate": 3.963864579035376e-06,
+      "loss": 0.857,
+      "step": 143140
+    },
+    {
+      "epoch": 1.0362150462912694,
+      "grad_norm": 0.16041286289691925,
+      "learning_rate": 3.96379219237479e-06,
+      "loss": 0.8544,
+      "step": 143150
+    },
+    {
+      "epoch": 1.0362874329518557,
+      "grad_norm": 0.14968006312847137,
+      "learning_rate": 3.963719805714203e-06,
+      "loss": 0.8596,
+      "step": 143160
+    },
+    {
+      "epoch": 1.0363598196124417,
+      "grad_norm": 0.1907356232404709,
+      "learning_rate": 3.963647419053617e-06,
+      "loss": 0.8688,
+      "step": 143170
+    },
+    {
+      "epoch": 1.036432206273028,
+      "grad_norm": 0.15427495539188385,
+      "learning_rate": 3.9635750323930314e-06,
+      "loss": 0.8584,
+      "step": 143180
+    },
+    {
+      "epoch": 1.0365045929336143,
+      "grad_norm": 0.15541379153728485,
+      "learning_rate": 3.963502645732445e-06,
+      "loss": 0.8563,
+      "step": 143190
+    },
+    {
+      "epoch": 1.0365769795942004,
+      "grad_norm": 0.14748956263065338,
+      "learning_rate": 3.963430259071859e-06,
+      "loss": 0.8564,
+      "step": 143200
+    },
+    {
+      "epoch": 1.0366493662547867,
+      "grad_norm": 0.15645940601825714,
+      "learning_rate": 3.963357872411272e-06,
+      "loss": 0.8467,
+      "step": 143210
+    },
+    {
+      "epoch": 1.0367217529153727,
+      "grad_norm": 0.21893535554409027,
+      "learning_rate": 3.963285485750687e-06,
+      "loss": 0.8673,
+      "step": 143220
+    },
+    {
+      "epoch": 1.036794139575959,
+      "grad_norm": 0.15868252515792847,
+      "learning_rate": 3.9632130990901e-06,
+      "loss": 0.8582,
+      "step": 143230
+    },
+    {
+      "epoch": 1.036866526236545,
+      "grad_norm": 0.1767452508211136,
+      "learning_rate": 3.963140712429514e-06,
+      "loss": 0.8549,
+      "step": 143240
+    },
+    {
+      "epoch": 1.0369389128971314,
+      "grad_norm": 0.14982333779335022,
+      "learning_rate": 3.963068325768928e-06,
+      "loss": 0.8538,
+      "step": 143250
+    },
+    {
+      "epoch": 1.0370112995577174,
+      "grad_norm": 0.1804773211479187,
+      "learning_rate": 3.962995939108342e-06,
+      "loss": 0.8644,
+      "step": 143260
+    },
+    {
+      "epoch": 1.0370836862183037,
+      "grad_norm": 0.15341947972774506,
+      "learning_rate": 3.962923552447756e-06,
+      "loss": 0.8605,
+      "step": 143270
+    },
+    {
+      "epoch": 1.0371560728788898,
+      "grad_norm": 0.15054786205291748,
+      "learning_rate": 3.962851165787169e-06,
+      "loss": 0.8607,
+      "step": 143280
+    },
+    {
+      "epoch": 1.037228459539476,
+      "grad_norm": 0.15896694362163544,
+      "learning_rate": 3.962778779126583e-06,
+      "loss": 0.845,
+      "step": 143290
+    },
+    {
+      "epoch": 1.0373008462000624,
+      "grad_norm": 0.1585065871477127,
+      "learning_rate": 3.9627063924659965e-06,
+      "loss": 0.8481,
+      "step": 143300
+    },
+    {
+      "epoch": 1.0373732328606484,
+      "grad_norm": 0.1433423012495041,
+      "learning_rate": 3.96263400580541e-06,
+      "loss": 0.8576,
+      "step": 143310
+    },
+    {
+      "epoch": 1.0374456195212347,
+      "grad_norm": 0.14972993731498718,
+      "learning_rate": 3.962561619144824e-06,
+      "loss": 0.8564,
+      "step": 143320
+    },
+    {
+      "epoch": 1.0375180061818208,
+      "grad_norm": 0.1547120362520218,
+      "learning_rate": 3.962489232484238e-06,
+      "loss": 0.8754,
+      "step": 143330
+    },
+    {
+      "epoch": 1.037590392842407,
+      "grad_norm": 0.15354642271995544,
+      "learning_rate": 3.962416845823652e-06,
+      "loss": 0.8516,
+      "step": 143340
+    },
+    {
+      "epoch": 1.0376627795029931,
+      "grad_norm": 0.15381920337677002,
+      "learning_rate": 3.9623444591630655e-06,
+      "loss": 0.8588,
+      "step": 143350
+    },
+    {
+      "epoch": 1.0377351661635794,
+      "grad_norm": 0.1736004650592804,
+      "learning_rate": 3.962272072502479e-06,
+      "loss": 0.863,
+      "step": 143360
+    },
+    {
+      "epoch": 1.0378075528241655,
+      "grad_norm": 0.1495305895805359,
+      "learning_rate": 3.9621996858418935e-06,
+      "loss": 0.861,
+      "step": 143370
+    },
+    {
+      "epoch": 1.0378799394847518,
+      "grad_norm": 0.16278128325939178,
+      "learning_rate": 3.962127299181307e-06,
+      "loss": 0.8673,
+      "step": 143380
+    },
+    {
+      "epoch": 1.0379523261453378,
+      "grad_norm": 0.16452762484550476,
+      "learning_rate": 3.962054912520721e-06,
+      "loss": 0.8692,
+      "step": 143390
+    },
+    {
+      "epoch": 1.0380247128059241,
+      "grad_norm": 0.15840129554271698,
+      "learning_rate": 3.961982525860134e-06,
+      "loss": 0.8589,
+      "step": 143400
+    },
+    {
+      "epoch": 1.0380970994665104,
+      "grad_norm": 0.14838318526744843,
+      "learning_rate": 3.961910139199549e-06,
+      "loss": 0.8547,
+      "step": 143410
+    },
+    {
+      "epoch": 1.0381694861270965,
+      "grad_norm": 0.16569004952907562,
+      "learning_rate": 3.9618377525389625e-06,
+      "loss": 0.847,
+      "step": 143420
+    },
+    {
+      "epoch": 1.0382418727876828,
+      "grad_norm": 0.1536937952041626,
+      "learning_rate": 3.961765365878376e-06,
+      "loss": 0.8547,
+      "step": 143430
+    },
+    {
+      "epoch": 1.0383142594482688,
+      "grad_norm": 0.16339300572872162,
+      "learning_rate": 3.96169297921779e-06,
+      "loss": 0.8637,
+      "step": 143440
+    },
+    {
+      "epoch": 1.0383866461088551,
+      "grad_norm": 0.1539309173822403,
+      "learning_rate": 3.961620592557204e-06,
+      "loss": 0.8667,
+      "step": 143450
+    },
+    {
+      "epoch": 1.0384590327694412,
+      "grad_norm": 0.16559149324893951,
+      "learning_rate": 3.961548205896618e-06,
+      "loss": 0.8595,
+      "step": 143460
+    },
+    {
+      "epoch": 1.0385314194300275,
+      "grad_norm": 0.1649179905653,
+      "learning_rate": 3.961475819236031e-06,
+      "loss": 0.8685,
+      "step": 143470
+    },
+    {
+      "epoch": 1.0386038060906135,
+      "grad_norm": 0.15256711840629578,
+      "learning_rate": 3.961403432575445e-06,
+      "loss": 0.856,
+      "step": 143480
+    },
+    {
+      "epoch": 1.0386761927511998,
+      "grad_norm": 0.1560206562280655,
+      "learning_rate": 3.9613310459148595e-06,
+      "loss": 0.8654,
+      "step": 143490
+    },
+    {
+      "epoch": 1.038748579411786,
+      "grad_norm": 0.16528300940990448,
+      "learning_rate": 3.961258659254273e-06,
+      "loss": 0.8678,
+      "step": 143500
+    },
+    {
+      "epoch": 1.0388209660723722,
+      "grad_norm": 0.1495116949081421,
+      "learning_rate": 3.961186272593687e-06,
+      "loss": 0.8661,
+      "step": 143510
+    },
+    {
+      "epoch": 1.0388933527329585,
+      "grad_norm": 0.16050077974796295,
+      "learning_rate": 3.9611138859331e-06,
+      "loss": 0.8632,
+      "step": 143520
+    },
+    {
+      "epoch": 1.0389657393935445,
+      "grad_norm": 0.14846397936344147,
+      "learning_rate": 3.961041499272514e-06,
+      "loss": 0.8513,
+      "step": 143530
+    },
+    {
+      "epoch": 1.0390381260541308,
+      "grad_norm": 0.15063372254371643,
+      "learning_rate": 3.960969112611928e-06,
+      "loss": 0.8696,
+      "step": 143540
+    },
+    {
+      "epoch": 1.0391105127147169,
+      "grad_norm": 0.15840241312980652,
+      "learning_rate": 3.960896725951342e-06,
+      "loss": 0.8648,
+      "step": 143550
+    },
+    {
+      "epoch": 1.0391828993753032,
+      "grad_norm": 0.22342956066131592,
+      "learning_rate": 3.960824339290756e-06,
+      "loss": 0.8598,
+      "step": 143560
+    },
+    {
+      "epoch": 1.0392552860358892,
+      "grad_norm": 0.15167798101902008,
+      "learning_rate": 3.960751952630169e-06,
+      "loss": 0.8531,
+      "step": 143570
+    },
+    {
+      "epoch": 1.0393276726964755,
+      "grad_norm": 0.158100888133049,
+      "learning_rate": 3.960679565969584e-06,
+      "loss": 0.8568,
+      "step": 143580
+    },
+    {
+      "epoch": 1.0394000593570616,
+      "grad_norm": 0.1609964668750763,
+      "learning_rate": 3.960607179308997e-06,
+      "loss": 0.8516,
+      "step": 143590
+    },
+    {
+      "epoch": 1.0394724460176479,
+      "grad_norm": 0.21195641160011292,
+      "learning_rate": 3.960534792648411e-06,
+      "loss": 0.8686,
+      "step": 143600
+    },
+    {
+      "epoch": 1.0395448326782342,
+      "grad_norm": 0.1418725848197937,
+      "learning_rate": 3.9604624059878246e-06,
+      "loss": 0.8644,
+      "step": 143610
+    },
+    {
+      "epoch": 1.0396172193388202,
+      "grad_norm": 0.15254880487918854,
+      "learning_rate": 3.960390019327239e-06,
+      "loss": 0.8557,
+      "step": 143620
+    },
+    {
+      "epoch": 1.0396896059994065,
+      "grad_norm": 0.1374930441379547,
+      "learning_rate": 3.960317632666653e-06,
+      "loss": 0.8588,
+      "step": 143630
+    },
+    {
+      "epoch": 1.0397619926599926,
+      "grad_norm": 0.1622588187456131,
+      "learning_rate": 3.960245246006066e-06,
+      "loss": 0.8537,
+      "step": 143640
+    },
+    {
+      "epoch": 1.0398343793205789,
+      "grad_norm": 0.1530616283416748,
+      "learning_rate": 3.96017285934548e-06,
+      "loss": 0.863,
+      "step": 143650
+    },
+    {
+      "epoch": 1.039906765981165,
+      "grad_norm": 0.16928766667842865,
+      "learning_rate": 3.960100472684894e-06,
+      "loss": 0.8646,
+      "step": 143660
+    },
+    {
+      "epoch": 1.0399791526417512,
+      "grad_norm": 0.15035808086395264,
+      "learning_rate": 3.960028086024308e-06,
+      "loss": 0.8539,
+      "step": 143670
+    },
+    {
+      "epoch": 1.0400515393023373,
+      "grad_norm": 0.1483323574066162,
+      "learning_rate": 3.9599556993637216e-06,
+      "loss": 0.8753,
+      "step": 143680
+    },
+    {
+      "epoch": 1.0401239259629236,
+      "grad_norm": 0.15935401618480682,
+      "learning_rate": 3.959883312703135e-06,
+      "loss": 0.8563,
+      "step": 143690
+    },
+    {
+      "epoch": 1.0401963126235096,
+      "grad_norm": 0.15291538834571838,
+      "learning_rate": 3.95981092604255e-06,
+      "loss": 0.8726,
+      "step": 143700
+    },
+    {
+      "epoch": 1.040268699284096,
+      "grad_norm": 0.14946019649505615,
+      "learning_rate": 3.959738539381963e-06,
+      "loss": 0.8623,
+      "step": 143710
+    },
+    {
+      "epoch": 1.0403410859446822,
+      "grad_norm": 0.13929761946201324,
+      "learning_rate": 3.959666152721377e-06,
+      "loss": 0.8612,
+      "step": 143720
+    },
+    {
+      "epoch": 1.0404134726052683,
+      "grad_norm": 0.1517852246761322,
+      "learning_rate": 3.9595937660607905e-06,
+      "loss": 0.8605,
+      "step": 143730
+    },
+    {
+      "epoch": 1.0404858592658546,
+      "grad_norm": 0.14445209503173828,
+      "learning_rate": 3.959521379400205e-06,
+      "loss": 0.8802,
+      "step": 143740
+    },
+    {
+      "epoch": 1.0405582459264406,
+      "grad_norm": 0.1581372618675232,
+      "learning_rate": 3.959448992739619e-06,
+      "loss": 0.8513,
+      "step": 143750
+    },
+    {
+      "epoch": 1.040630632587027,
+      "grad_norm": 0.16304130852222443,
+      "learning_rate": 3.959376606079032e-06,
+      "loss": 0.869,
+      "step": 143760
+    },
+    {
+      "epoch": 1.040703019247613,
+      "grad_norm": 0.17016209661960602,
+      "learning_rate": 3.959304219418446e-06,
+      "loss": 0.8516,
+      "step": 143770
+    },
+    {
+      "epoch": 1.0407754059081993,
+      "grad_norm": 0.1455002725124359,
+      "learning_rate": 3.95923183275786e-06,
+      "loss": 0.8541,
+      "step": 143780
+    },
+    {
+      "epoch": 1.0408477925687853,
+      "grad_norm": 0.3031005263328552,
+      "learning_rate": 3.959159446097274e-06,
+      "loss": 0.8615,
+      "step": 143790
+    },
+    {
+      "epoch": 1.0409201792293716,
+      "grad_norm": 0.15537060797214508,
+      "learning_rate": 3.9590870594366875e-06,
+      "loss": 0.8534,
+      "step": 143800
+    },
+    {
+      "epoch": 1.0409925658899577,
+      "grad_norm": 0.18739604949951172,
+      "learning_rate": 3.959014672776101e-06,
+      "loss": 0.8595,
+      "step": 143810
+    },
+    {
+      "epoch": 1.041064952550544,
+      "grad_norm": 0.15703223645687103,
+      "learning_rate": 3.958942286115516e-06,
+      "loss": 0.8672,
+      "step": 143820
+    },
+    {
+      "epoch": 1.0411373392111303,
+      "grad_norm": 0.15714220702648163,
+      "learning_rate": 3.958869899454928e-06,
+      "loss": 0.8801,
+      "step": 143830
+    },
+    {
+      "epoch": 1.0412097258717163,
+      "grad_norm": 0.14824450016021729,
+      "learning_rate": 3.958797512794342e-06,
+      "loss": 0.8673,
+      "step": 143840
+    },
+    {
+      "epoch": 1.0412821125323026,
+      "grad_norm": 0.14315825700759888,
+      "learning_rate": 3.9587251261337564e-06,
+      "loss": 0.8553,
+      "step": 143850
+    },
+    {
+      "epoch": 1.0413544991928887,
+      "grad_norm": 0.1702081710100174,
+      "learning_rate": 3.95865273947317e-06,
+      "loss": 0.8644,
+      "step": 143860
+    },
+    {
+      "epoch": 1.041426885853475,
+      "grad_norm": 0.18592719733715057,
+      "learning_rate": 3.958580352812584e-06,
+      "loss": 0.865,
+      "step": 143870
+    },
+    {
+      "epoch": 1.041499272514061,
+      "grad_norm": 0.15795159339904785,
+      "learning_rate": 3.958507966151997e-06,
+      "loss": 0.8566,
+      "step": 143880
+    },
+    {
+      "epoch": 1.0415716591746473,
+      "grad_norm": 0.16113890707492828,
+      "learning_rate": 3.958435579491412e-06,
+      "loss": 0.8577,
+      "step": 143890
+    },
+    {
+      "epoch": 1.0416440458352334,
+      "grad_norm": 0.29650846123695374,
+      "learning_rate": 3.958363192830825e-06,
+      "loss": 0.8594,
+      "step": 143900
+    },
+    {
+      "epoch": 1.0417164324958197,
+      "grad_norm": 0.14694488048553467,
+      "learning_rate": 3.958290806170239e-06,
+      "loss": 0.8693,
+      "step": 143910
+    },
+    {
+      "epoch": 1.0417888191564058,
+      "grad_norm": 0.15277814865112305,
+      "learning_rate": 3.958218419509653e-06,
+      "loss": 0.8697,
+      "step": 143920
+    },
+    {
+      "epoch": 1.041861205816992,
+      "grad_norm": 0.16288557648658752,
+      "learning_rate": 3.958146032849067e-06,
+      "loss": 0.8669,
+      "step": 143930
+    },
+    {
+      "epoch": 1.0419335924775783,
+      "grad_norm": 0.14526166021823883,
+      "learning_rate": 3.958073646188481e-06,
+      "loss": 0.8649,
+      "step": 143940
+    },
+    {
+      "epoch": 1.0420059791381644,
+      "grad_norm": 0.15091979503631592,
+      "learning_rate": 3.958001259527894e-06,
+      "loss": 0.859,
+      "step": 143950
+    },
+    {
+      "epoch": 1.0420783657987507,
+      "grad_norm": 0.14309091866016388,
+      "learning_rate": 3.957928872867308e-06,
+      "loss": 0.8614,
+      "step": 143960
+    },
+    {
+      "epoch": 1.0421507524593367,
+      "grad_norm": 0.15367577970027924,
+      "learning_rate": 3.957856486206722e-06,
+      "loss": 0.8606,
+      "step": 143970
+    },
+    {
+      "epoch": 1.042223139119923,
+      "grad_norm": 0.15176285803318024,
+      "learning_rate": 3.957784099546136e-06,
+      "loss": 0.8542,
+      "step": 143980
+    },
+    {
+      "epoch": 1.042295525780509,
+      "grad_norm": 0.1759510636329651,
+      "learning_rate": 3.95771171288555e-06,
+      "loss": 0.858,
+      "step": 143990
+    },
+    {
+      "epoch": 1.0423679124410954,
+      "grad_norm": 0.17628808319568634,
+      "learning_rate": 3.957639326224963e-06,
+      "loss": 0.8684,
+      "step": 144000
+    },
+    {
+      "epoch": 1.0424402991016815,
+      "grad_norm": 0.14174871146678925,
+      "learning_rate": 3.957566939564378e-06,
+      "loss": 0.8565,
+      "step": 144010
+    },
+    {
+      "epoch": 1.0425126857622677,
+      "grad_norm": 0.14700306951999664,
+      "learning_rate": 3.957494552903791e-06,
+      "loss": 0.8608,
+      "step": 144020
+    },
+    {
+      "epoch": 1.0425850724228538,
+      "grad_norm": 0.14982403814792633,
+      "learning_rate": 3.957422166243205e-06,
+      "loss": 0.8552,
+      "step": 144030
+    },
+    {
+      "epoch": 1.04265745908344,
+      "grad_norm": 0.1603231132030487,
+      "learning_rate": 3.9573497795826185e-06,
+      "loss": 0.8628,
+      "step": 144040
+    },
+    {
+      "epoch": 1.0427298457440264,
+      "grad_norm": 0.173030823469162,
+      "learning_rate": 3.957277392922033e-06,
+      "loss": 0.8695,
+      "step": 144050
+    },
+    {
+      "epoch": 1.0428022324046125,
+      "grad_norm": 0.18161053955554962,
+      "learning_rate": 3.957205006261447e-06,
+      "loss": 0.8633,
+      "step": 144060
+    },
+    {
+      "epoch": 1.0428746190651987,
+      "grad_norm": 0.15263399481773376,
+      "learning_rate": 3.95713261960086e-06,
+      "loss": 0.8574,
+      "step": 144070
+    },
+    {
+      "epoch": 1.0429470057257848,
+      "grad_norm": 0.15920861065387726,
+      "learning_rate": 3.957060232940274e-06,
+      "loss": 0.8629,
+      "step": 144080
+    },
+    {
+      "epoch": 1.043019392386371,
+      "grad_norm": 0.15555734932422638,
+      "learning_rate": 3.956987846279688e-06,
+      "loss": 0.8768,
+      "step": 144090
+    },
+    {
+      "epoch": 1.0430917790469572,
+      "grad_norm": 0.1550634801387787,
+      "learning_rate": 3.956915459619102e-06,
+      "loss": 0.8533,
+      "step": 144100
+    },
+    {
+      "epoch": 1.0431641657075434,
+      "grad_norm": 0.14551453292369843,
+      "learning_rate": 3.9568430729585155e-06,
+      "loss": 0.8579,
+      "step": 144110
+    },
+    {
+      "epoch": 1.0432365523681295,
+      "grad_norm": 0.14766032993793488,
+      "learning_rate": 3.956770686297929e-06,
+      "loss": 0.8678,
+      "step": 144120
+    },
+    {
+      "epoch": 1.0433089390287158,
+      "grad_norm": 0.15198858082294464,
+      "learning_rate": 3.956698299637344e-06,
+      "loss": 0.8638,
+      "step": 144130
+    },
+    {
+      "epoch": 1.0433813256893019,
+      "grad_norm": 0.16540324687957764,
+      "learning_rate": 3.956625912976757e-06,
+      "loss": 0.8662,
+      "step": 144140
+    },
+    {
+      "epoch": 1.0434537123498882,
+      "grad_norm": 0.14517223834991455,
+      "learning_rate": 3.956553526316171e-06,
+      "loss": 0.8454,
+      "step": 144150
+    },
+    {
+      "epoch": 1.0435260990104744,
+      "grad_norm": 0.15670832991600037,
+      "learning_rate": 3.9564811396555845e-06,
+      "loss": 0.8574,
+      "step": 144160
+    },
+    {
+      "epoch": 1.0435984856710605,
+      "grad_norm": 0.16372807323932648,
+      "learning_rate": 3.956408752994998e-06,
+      "loss": 0.8549,
+      "step": 144170
+    },
+    {
+      "epoch": 1.0436708723316468,
+      "grad_norm": 0.157616525888443,
+      "learning_rate": 3.9563363663344125e-06,
+      "loss": 0.844,
+      "step": 144180
+    },
+    {
+      "epoch": 1.0437432589922329,
+      "grad_norm": 0.14867444336414337,
+      "learning_rate": 3.956263979673826e-06,
+      "loss": 0.8581,
+      "step": 144190
+    },
+    {
+      "epoch": 1.0438156456528191,
+      "grad_norm": 0.15793468058109283,
+      "learning_rate": 3.95619159301324e-06,
+      "loss": 0.8503,
+      "step": 144200
+    },
+    {
+      "epoch": 1.0438880323134052,
+      "grad_norm": 0.14294303953647614,
+      "learning_rate": 3.956119206352653e-06,
+      "loss": 0.8643,
+      "step": 144210
+    },
+    {
+      "epoch": 1.0439604189739915,
+      "grad_norm": 0.15803156793117523,
+      "learning_rate": 3.956046819692068e-06,
+      "loss": 0.8626,
+      "step": 144220
+    },
+    {
+      "epoch": 1.0440328056345776,
+      "grad_norm": 0.16307860612869263,
+      "learning_rate": 3.9559744330314815e-06,
+      "loss": 0.8663,
+      "step": 144230
+    },
+    {
+      "epoch": 1.0441051922951639,
+      "grad_norm": 0.14012962579727173,
+      "learning_rate": 3.955902046370895e-06,
+      "loss": 0.8566,
+      "step": 144240
+    },
+    {
+      "epoch": 1.0441775789557501,
+      "grad_norm": 0.19442588090896606,
+      "learning_rate": 3.955829659710309e-06,
+      "loss": 0.8765,
+      "step": 144250
+    },
+    {
+      "epoch": 1.0442499656163362,
+      "grad_norm": 0.15497778356075287,
+      "learning_rate": 3.955757273049723e-06,
+      "loss": 0.8596,
+      "step": 144260
+    },
+    {
+      "epoch": 1.0443223522769225,
+      "grad_norm": 0.15223382413387299,
+      "learning_rate": 3.955684886389137e-06,
+      "loss": 0.8815,
+      "step": 144270
+    },
+    {
+      "epoch": 1.0443947389375086,
+      "grad_norm": 0.15136075019836426,
+      "learning_rate": 3.95561249972855e-06,
+      "loss": 0.8648,
+      "step": 144280
+    },
+    {
+      "epoch": 1.0444671255980948,
+      "grad_norm": 0.15024292469024658,
+      "learning_rate": 3.955540113067964e-06,
+      "loss": 0.86,
+      "step": 144290
+    },
+    {
+      "epoch": 1.044539512258681,
+      "grad_norm": 0.1588437706232071,
+      "learning_rate": 3.9554677264073785e-06,
+      "loss": 0.8607,
+      "step": 144300
+    },
+    {
+      "epoch": 1.0446118989192672,
+      "grad_norm": 0.15004202723503113,
+      "learning_rate": 3.955395339746792e-06,
+      "loss": 0.8609,
+      "step": 144310
+    },
+    {
+      "epoch": 1.0446842855798533,
+      "grad_norm": 0.14561429619789124,
+      "learning_rate": 3.955322953086206e-06,
+      "loss": 0.8673,
+      "step": 144320
+    },
+    {
+      "epoch": 1.0447566722404396,
+      "grad_norm": 0.1565464287996292,
+      "learning_rate": 3.955250566425619e-06,
+      "loss": 0.8591,
+      "step": 144330
+    },
+    {
+      "epoch": 1.0448290589010256,
+      "grad_norm": 0.1545923948287964,
+      "learning_rate": 3.955178179765034e-06,
+      "loss": 0.8445,
+      "step": 144340
+    },
+    {
+      "epoch": 1.044901445561612,
+      "grad_norm": 0.15572689473628998,
+      "learning_rate": 3.955105793104447e-06,
+      "loss": 0.8552,
+      "step": 144350
+    },
+    {
+      "epoch": 1.0449738322221982,
+      "grad_norm": 0.1447010338306427,
+      "learning_rate": 3.95503340644386e-06,
+      "loss": 0.8587,
+      "step": 144360
+    },
+    {
+      "epoch": 1.0450462188827843,
+      "grad_norm": 0.15390467643737793,
+      "learning_rate": 3.954961019783275e-06,
+      "loss": 0.8602,
+      "step": 144370
+    },
+    {
+      "epoch": 1.0451186055433705,
+      "grad_norm": 0.16179361939430237,
+      "learning_rate": 3.954888633122688e-06,
+      "loss": 0.8546,
+      "step": 144380
+    },
+    {
+      "epoch": 1.0451909922039566,
+      "grad_norm": 0.1796809881925583,
+      "learning_rate": 3.954816246462102e-06,
+      "loss": 0.8601,
+      "step": 144390
+    },
+    {
+      "epoch": 1.045263378864543,
+      "grad_norm": 0.16800828278064728,
+      "learning_rate": 3.9547438598015155e-06,
+      "loss": 0.8624,
+      "step": 144400
+    },
+    {
+      "epoch": 1.045335765525129,
+      "grad_norm": 0.14844734966754913,
+      "learning_rate": 3.95467147314093e-06,
+      "loss": 0.8666,
+      "step": 144410
+    },
+    {
+      "epoch": 1.0454081521857153,
+      "grad_norm": 0.1545286625623703,
+      "learning_rate": 3.9545990864803436e-06,
+      "loss": 0.8598,
+      "step": 144420
+    },
+    {
+      "epoch": 1.0454805388463013,
+      "grad_norm": 0.1570822298526764,
+      "learning_rate": 3.954526699819757e-06,
+      "loss": 0.8608,
+      "step": 144430
+    },
+    {
+      "epoch": 1.0455529255068876,
+      "grad_norm": 0.14228971302509308,
+      "learning_rate": 3.954454313159171e-06,
+      "loss": 0.8523,
+      "step": 144440
+    },
+    {
+      "epoch": 1.0456253121674737,
+      "grad_norm": 0.14609818160533905,
+      "learning_rate": 3.954381926498585e-06,
+      "loss": 0.8584,
+      "step": 144450
+    },
+    {
+      "epoch": 1.04569769882806,
+      "grad_norm": 0.1522809863090515,
+      "learning_rate": 3.954309539837999e-06,
+      "loss": 0.8569,
+      "step": 144460
+    },
+    {
+      "epoch": 1.0457700854886463,
+      "grad_norm": 0.2182849943637848,
+      "learning_rate": 3.9542371531774125e-06,
+      "loss": 0.8412,
+      "step": 144470
+    },
+    {
+      "epoch": 1.0458424721492323,
+      "grad_norm": 0.148259699344635,
+      "learning_rate": 3.954164766516826e-06,
+      "loss": 0.8521,
+      "step": 144480
+    },
+    {
+      "epoch": 1.0459148588098186,
+      "grad_norm": 0.14706644415855408,
+      "learning_rate": 3.954092379856241e-06,
+      "loss": 0.8622,
+      "step": 144490
+    },
+    {
+      "epoch": 1.0459872454704047,
+      "grad_norm": 0.15702740848064423,
+      "learning_rate": 3.954019993195654e-06,
+      "loss": 0.8563,
+      "step": 144500
+    },
+    {
+      "epoch": 1.046059632130991,
+      "grad_norm": 0.1426214873790741,
+      "learning_rate": 3.953947606535068e-06,
+      "loss": 0.8628,
+      "step": 144510
+    },
+    {
+      "epoch": 1.046132018791577,
+      "grad_norm": 0.14611200988292694,
+      "learning_rate": 3.9538752198744814e-06,
+      "loss": 0.8544,
+      "step": 144520
+    },
+    {
+      "epoch": 1.0462044054521633,
+      "grad_norm": 0.15756775438785553,
+      "learning_rate": 3.953802833213896e-06,
+      "loss": 0.8568,
+      "step": 144530
+    },
+    {
+      "epoch": 1.0462767921127494,
+      "grad_norm": 0.15759259462356567,
+      "learning_rate": 3.9537304465533095e-06,
+      "loss": 0.8634,
+      "step": 144540
+    },
+    {
+      "epoch": 1.0463491787733357,
+      "grad_norm": 0.15958790481090546,
+      "learning_rate": 3.953658059892723e-06,
+      "loss": 0.8763,
+      "step": 144550
+    },
+    {
+      "epoch": 1.0464215654339217,
+      "grad_norm": 0.15510474145412445,
+      "learning_rate": 3.953585673232137e-06,
+      "loss": 0.873,
+      "step": 144560
+    },
+    {
+      "epoch": 1.046493952094508,
+      "grad_norm": 0.1597568690776825,
+      "learning_rate": 3.953513286571551e-06,
+      "loss": 0.8488,
+      "step": 144570
+    },
+    {
+      "epoch": 1.0465663387550943,
+      "grad_norm": 0.14834840595722198,
+      "learning_rate": 3.953440899910965e-06,
+      "loss": 0.8545,
+      "step": 144580
+    },
+    {
+      "epoch": 1.0466387254156804,
+      "grad_norm": 0.13909775018692017,
+      "learning_rate": 3.9533685132503784e-06,
+      "loss": 0.8478,
+      "step": 144590
+    },
+    {
+      "epoch": 1.0467111120762667,
+      "grad_norm": 0.14474329352378845,
+      "learning_rate": 3.953296126589792e-06,
+      "loss": 0.8699,
+      "step": 144600
+    },
+    {
+      "epoch": 1.0467834987368527,
+      "grad_norm": 0.16412761807441711,
+      "learning_rate": 3.9532237399292065e-06,
+      "loss": 0.8616,
+      "step": 144610
+    },
+    {
+      "epoch": 1.046855885397439,
+      "grad_norm": 0.1485072374343872,
+      "learning_rate": 3.95315135326862e-06,
+      "loss": 0.8594,
+      "step": 144620
+    },
+    {
+      "epoch": 1.046928272058025,
+      "grad_norm": 0.1748332530260086,
+      "learning_rate": 3.953078966608034e-06,
+      "loss": 0.8552,
+      "step": 144630
+    },
+    {
+      "epoch": 1.0470006587186114,
+      "grad_norm": 0.15388593077659607,
+      "learning_rate": 3.953006579947447e-06,
+      "loss": 0.8633,
+      "step": 144640
+    },
+    {
+      "epoch": 1.0470730453791974,
+      "grad_norm": 0.1663622111082077,
+      "learning_rate": 3.952934193286862e-06,
+      "loss": 0.8675,
+      "step": 144650
+    },
+    {
+      "epoch": 1.0471454320397837,
+      "grad_norm": 0.1697302758693695,
+      "learning_rate": 3.9528618066262754e-06,
+      "loss": 0.8561,
+      "step": 144660
+    },
+    {
+      "epoch": 1.04721781870037,
+      "grad_norm": 0.15296570956707,
+      "learning_rate": 3.952789419965689e-06,
+      "loss": 0.8699,
+      "step": 144670
+    },
+    {
+      "epoch": 1.047290205360956,
+      "grad_norm": 0.1404290646314621,
+      "learning_rate": 3.952717033305103e-06,
+      "loss": 0.8566,
+      "step": 144680
+    },
+    {
+      "epoch": 1.0473625920215424,
+      "grad_norm": 0.16811597347259521,
+      "learning_rate": 3.952644646644517e-06,
+      "loss": 0.8771,
+      "step": 144690
+    },
+    {
+      "epoch": 1.0474349786821284,
+      "grad_norm": 0.15282821655273438,
+      "learning_rate": 3.952572259983931e-06,
+      "loss": 0.8612,
+      "step": 144700
+    },
+    {
+      "epoch": 1.0475073653427147,
+      "grad_norm": 0.16427479684352875,
+      "learning_rate": 3.952499873323344e-06,
+      "loss": 0.8633,
+      "step": 144710
+    },
+    {
+      "epoch": 1.0475797520033008,
+      "grad_norm": 0.16405102610588074,
+      "learning_rate": 3.952427486662758e-06,
+      "loss": 0.8652,
+      "step": 144720
+    },
+    {
+      "epoch": 1.047652138663887,
+      "grad_norm": 0.14191152155399323,
+      "learning_rate": 3.9523551000021724e-06,
+      "loss": 0.8628,
+      "step": 144730
+    },
+    {
+      "epoch": 1.0477245253244731,
+      "grad_norm": 0.15622809529304504,
+      "learning_rate": 3.952282713341586e-06,
+      "loss": 0.8697,
+      "step": 144740
+    },
+    {
+      "epoch": 1.0477969119850594,
+      "grad_norm": 0.15117841958999634,
+      "learning_rate": 3.952210326681e-06,
+      "loss": 0.8628,
+      "step": 144750
+    },
+    {
+      "epoch": 1.0478692986456455,
+      "grad_norm": 0.14459557831287384,
+      "learning_rate": 3.952137940020413e-06,
+      "loss": 0.861,
+      "step": 144760
+    },
+    {
+      "epoch": 1.0479416853062318,
+      "grad_norm": 0.14188896119594574,
+      "learning_rate": 3.952065553359827e-06,
+      "loss": 0.8772,
+      "step": 144770
+    },
+    {
+      "epoch": 1.048014071966818,
+      "grad_norm": 0.16736344993114471,
+      "learning_rate": 3.951993166699241e-06,
+      "loss": 0.8569,
+      "step": 144780
+    },
+    {
+      "epoch": 1.0480864586274041,
+      "grad_norm": 0.16879191994667053,
+      "learning_rate": 3.951920780038655e-06,
+      "loss": 0.8496,
+      "step": 144790
+    },
+    {
+      "epoch": 1.0481588452879904,
+      "grad_norm": 0.16716881096363068,
+      "learning_rate": 3.951848393378069e-06,
+      "loss": 0.8536,
+      "step": 144800
+    },
+    {
+      "epoch": 1.0482312319485765,
+      "grad_norm": 0.14509662985801697,
+      "learning_rate": 3.951776006717482e-06,
+      "loss": 0.8582,
+      "step": 144810
+    },
+    {
+      "epoch": 1.0483036186091628,
+      "grad_norm": 0.16360674798488617,
+      "learning_rate": 3.951703620056897e-06,
+      "loss": 0.8664,
+      "step": 144820
+    },
+    {
+      "epoch": 1.0483760052697488,
+      "grad_norm": 0.15307435393333435,
+      "learning_rate": 3.95163123339631e-06,
+      "loss": 0.8481,
+      "step": 144830
+    },
+    {
+      "epoch": 1.0484483919303351,
+      "grad_norm": 0.1502341479063034,
+      "learning_rate": 3.951558846735724e-06,
+      "loss": 0.8533,
+      "step": 144840
+    },
+    {
+      "epoch": 1.0485207785909212,
+      "grad_norm": 0.15502779185771942,
+      "learning_rate": 3.9514864600751375e-06,
+      "loss": 0.8588,
+      "step": 144850
+    },
+    {
+      "epoch": 1.0485931652515075,
+      "grad_norm": 0.15955467522144318,
+      "learning_rate": 3.951414073414552e-06,
+      "loss": 0.858,
+      "step": 144860
+    },
+    {
+      "epoch": 1.0486655519120935,
+      "grad_norm": 0.14335303008556366,
+      "learning_rate": 3.951341686753966e-06,
+      "loss": 0.8591,
+      "step": 144870
+    },
+    {
+      "epoch": 1.0487379385726798,
+      "grad_norm": 0.15716949105262756,
+      "learning_rate": 3.951269300093379e-06,
+      "loss": 0.8433,
+      "step": 144880
+    },
+    {
+      "epoch": 1.0488103252332661,
+      "grad_norm": 0.16508370637893677,
+      "learning_rate": 3.951196913432793e-06,
+      "loss": 0.8454,
+      "step": 144890
+    },
+    {
+      "epoch": 1.0488827118938522,
+      "grad_norm": 0.14553935825824738,
+      "learning_rate": 3.9511245267722065e-06,
+      "loss": 0.8657,
+      "step": 144900
+    },
+    {
+      "epoch": 1.0489550985544385,
+      "grad_norm": 0.16298197209835052,
+      "learning_rate": 3.95105214011162e-06,
+      "loss": 0.8496,
+      "step": 144910
+    },
+    {
+      "epoch": 1.0490274852150245,
+      "grad_norm": 0.15694205462932587,
+      "learning_rate": 3.9509797534510345e-06,
+      "loss": 0.8463,
+      "step": 144920
+    },
+    {
+      "epoch": 1.0490998718756108,
+      "grad_norm": 0.14872339367866516,
+      "learning_rate": 3.950907366790448e-06,
+      "loss": 0.8512,
+      "step": 144930
+    },
+    {
+      "epoch": 1.0491722585361969,
+      "grad_norm": 0.14613871276378632,
+      "learning_rate": 3.950834980129862e-06,
+      "loss": 0.8468,
+      "step": 144940
+    },
+    {
+      "epoch": 1.0492446451967832,
+      "grad_norm": 0.175027996301651,
+      "learning_rate": 3.950762593469275e-06,
+      "loss": 0.8601,
+      "step": 144950
+    },
+    {
+      "epoch": 1.0493170318573692,
+      "grad_norm": 0.15894457697868347,
+      "learning_rate": 3.950690206808689e-06,
+      "loss": 0.8597,
+      "step": 144960
+    },
+    {
+      "epoch": 1.0493894185179555,
+      "grad_norm": 0.14936257898807526,
+      "learning_rate": 3.9506178201481035e-06,
+      "loss": 0.8526,
+      "step": 144970
+    },
+    {
+      "epoch": 1.0494618051785416,
+      "grad_norm": 0.15866105258464813,
+      "learning_rate": 3.950545433487517e-06,
+      "loss": 0.8633,
+      "step": 144980
+    },
+    {
+      "epoch": 1.0495341918391279,
+      "grad_norm": 0.1673537641763687,
+      "learning_rate": 3.950473046826931e-06,
+      "loss": 0.8502,
+      "step": 144990
+    },
+    {
+      "epoch": 1.0496065784997142,
+      "grad_norm": 0.14420992136001587,
+      "learning_rate": 3.950400660166344e-06,
+      "loss": 0.8576,
+      "step": 145000
+    },
+    {
+      "epoch": 1.0496789651603002,
+      "grad_norm": 0.16589012742042542,
+      "learning_rate": 3.950328273505759e-06,
+      "loss": 0.8598,
+      "step": 145010
+    },
+    {
+      "epoch": 1.0497513518208865,
+      "grad_norm": 0.15447083115577698,
+      "learning_rate": 3.950255886845172e-06,
+      "loss": 0.8461,
+      "step": 145020
+    },
+    {
+      "epoch": 1.0498237384814726,
+      "grad_norm": 0.15255771577358246,
+      "learning_rate": 3.950183500184586e-06,
+      "loss": 0.8643,
+      "step": 145030
+    },
+    {
+      "epoch": 1.0498961251420589,
+      "grad_norm": 0.17621764540672302,
+      "learning_rate": 3.950111113524e-06,
+      "loss": 0.8766,
+      "step": 145040
+    },
+    {
+      "epoch": 1.049968511802645,
+      "grad_norm": 0.15463955700397491,
+      "learning_rate": 3.950038726863414e-06,
+      "loss": 0.8675,
+      "step": 145050
+    },
+    {
+      "epoch": 1.0500408984632312,
+      "grad_norm": 0.14184853434562683,
+      "learning_rate": 3.949966340202828e-06,
+      "loss": 0.8533,
+      "step": 145060
+    },
+    {
+      "epoch": 1.0501132851238173,
+      "grad_norm": 0.1818603277206421,
+      "learning_rate": 3.949893953542241e-06,
+      "loss": 0.8644,
+      "step": 145070
+    },
+    {
+      "epoch": 1.0501856717844036,
+      "grad_norm": 0.1516370177268982,
+      "learning_rate": 3.949821566881655e-06,
+      "loss": 0.8632,
+      "step": 145080
+    },
+    {
+      "epoch": 1.0502580584449897,
+      "grad_norm": 0.15722733736038208,
+      "learning_rate": 3.949749180221069e-06,
+      "loss": 0.8562,
+      "step": 145090
+    },
+    {
+      "epoch": 1.050330445105576,
+      "grad_norm": 0.1518261581659317,
+      "learning_rate": 3.949676793560483e-06,
+      "loss": 0.8694,
+      "step": 145100
+    },
+    {
+      "epoch": 1.0504028317661622,
+      "grad_norm": 0.15784966945648193,
+      "learning_rate": 3.949604406899897e-06,
+      "loss": 0.8699,
+      "step": 145110
+    },
+    {
+      "epoch": 1.0504752184267483,
+      "grad_norm": 0.15036171674728394,
+      "learning_rate": 3.94953202023931e-06,
+      "loss": 0.8759,
+      "step": 145120
+    },
+    {
+      "epoch": 1.0505476050873346,
+      "grad_norm": 0.1495043933391571,
+      "learning_rate": 3.949459633578725e-06,
+      "loss": 0.8592,
+      "step": 145130
+    },
+    {
+      "epoch": 1.0506199917479206,
+      "grad_norm": 0.16956430673599243,
+      "learning_rate": 3.949387246918138e-06,
+      "loss": 0.8543,
+      "step": 145140
+    },
+    {
+      "epoch": 1.050692378408507,
+      "grad_norm": 0.1504315882921219,
+      "learning_rate": 3.949314860257552e-06,
+      "loss": 0.8614,
+      "step": 145150
+    },
+    {
+      "epoch": 1.050764765069093,
+      "grad_norm": 0.16208255290985107,
+      "learning_rate": 3.9492424735969656e-06,
+      "loss": 0.8619,
+      "step": 145160
+    },
+    {
+      "epoch": 1.0508371517296793,
+      "grad_norm": 0.15502387285232544,
+      "learning_rate": 3.94917008693638e-06,
+      "loss": 0.8673,
+      "step": 145170
+    },
+    {
+      "epoch": 1.0509095383902654,
+      "grad_norm": 0.1422945111989975,
+      "learning_rate": 3.949097700275794e-06,
+      "loss": 0.8584,
+      "step": 145180
+    },
+    {
+      "epoch": 1.0509819250508516,
+      "grad_norm": 0.15001416206359863,
+      "learning_rate": 3.949025313615207e-06,
+      "loss": 0.8614,
+      "step": 145190
+    },
+    {
+      "epoch": 1.0510543117114377,
+      "grad_norm": 0.1533329039812088,
+      "learning_rate": 3.948952926954621e-06,
+      "loss": 0.8755,
+      "step": 145200
+    },
+    {
+      "epoch": 1.051126698372024,
+      "grad_norm": 0.16097886860370636,
+      "learning_rate": 3.948880540294035e-06,
+      "loss": 0.8498,
+      "step": 145210
+    },
+    {
+      "epoch": 1.0511990850326103,
+      "grad_norm": 0.16414912045001984,
+      "learning_rate": 3.948808153633449e-06,
+      "loss": 0.8579,
+      "step": 145220
+    },
+    {
+      "epoch": 1.0512714716931963,
+      "grad_norm": 0.1741316169500351,
+      "learning_rate": 3.948735766972863e-06,
+      "loss": 0.8486,
+      "step": 145230
+    },
+    {
+      "epoch": 1.0513438583537826,
+      "grad_norm": 0.15596316754817963,
+      "learning_rate": 3.948663380312276e-06,
+      "loss": 0.8546,
+      "step": 145240
+    },
+    {
+      "epoch": 1.0514162450143687,
+      "grad_norm": 0.15392540395259857,
+      "learning_rate": 3.948590993651691e-06,
+      "loss": 0.8443,
+      "step": 145250
+    },
+    {
+      "epoch": 1.051488631674955,
+      "grad_norm": 0.15028880536556244,
+      "learning_rate": 3.948518606991104e-06,
+      "loss": 0.8501,
+      "step": 145260
+    },
+    {
+      "epoch": 1.051561018335541,
+      "grad_norm": 0.14961551129817963,
+      "learning_rate": 3.948446220330518e-06,
+      "loss": 0.8549,
+      "step": 145270
+    },
+    {
+      "epoch": 1.0516334049961273,
+      "grad_norm": 0.16350728273391724,
+      "learning_rate": 3.9483738336699315e-06,
+      "loss": 0.8723,
+      "step": 145280
+    },
+    {
+      "epoch": 1.0517057916567134,
+      "grad_norm": 0.1536688506603241,
+      "learning_rate": 3.948301447009346e-06,
+      "loss": 0.8512,
+      "step": 145290
+    },
+    {
+      "epoch": 1.0517781783172997,
+      "grad_norm": 0.1536385864019394,
+      "learning_rate": 3.94822906034876e-06,
+      "loss": 0.8598,
+      "step": 145300
+    },
+    {
+      "epoch": 1.051850564977886,
+      "grad_norm": 0.1381438672542572,
+      "learning_rate": 3.948156673688173e-06,
+      "loss": 0.8638,
+      "step": 145310
+    },
+    {
+      "epoch": 1.051922951638472,
+      "grad_norm": 0.17076510190963745,
+      "learning_rate": 3.948084287027587e-06,
+      "loss": 0.8629,
+      "step": 145320
+    },
+    {
+      "epoch": 1.0519953382990583,
+      "grad_norm": 0.15944698452949524,
+      "learning_rate": 3.948011900367001e-06,
+      "loss": 0.8617,
+      "step": 145330
+    },
+    {
+      "epoch": 1.0520677249596444,
+      "grad_norm": 0.15273165702819824,
+      "learning_rate": 3.947939513706415e-06,
+      "loss": 0.8553,
+      "step": 145340
+    },
+    {
+      "epoch": 1.0521401116202307,
+      "grad_norm": 0.14267678558826447,
+      "learning_rate": 3.9478671270458285e-06,
+      "loss": 0.8529,
+      "step": 145350
+    },
+    {
+      "epoch": 1.0522124982808168,
+      "grad_norm": 0.1617887020111084,
+      "learning_rate": 3.947794740385242e-06,
+      "loss": 0.8506,
+      "step": 145360
+    },
+    {
+      "epoch": 1.052284884941403,
+      "grad_norm": 0.1612195074558258,
+      "learning_rate": 3.947722353724657e-06,
+      "loss": 0.8655,
+      "step": 145370
+    },
+    {
+      "epoch": 1.052357271601989,
+      "grad_norm": 0.16076426208019257,
+      "learning_rate": 3.94764996706407e-06,
+      "loss": 0.853,
+      "step": 145380
+    },
+    {
+      "epoch": 1.0524296582625754,
+      "grad_norm": 0.14427517354488373,
+      "learning_rate": 3.947577580403484e-06,
+      "loss": 0.8647,
+      "step": 145390
+    },
+    {
+      "epoch": 1.0525020449231615,
+      "grad_norm": 0.1446637511253357,
+      "learning_rate": 3.9475051937428974e-06,
+      "loss": 0.8605,
+      "step": 145400
+    },
+    {
+      "epoch": 1.0525744315837477,
+      "grad_norm": 0.15358710289001465,
+      "learning_rate": 3.947432807082311e-06,
+      "loss": 0.8512,
+      "step": 145410
+    },
+    {
+      "epoch": 1.052646818244334,
+      "grad_norm": 0.1505446434020996,
+      "learning_rate": 3.947360420421725e-06,
+      "loss": 0.8567,
+      "step": 145420
+    },
+    {
+      "epoch": 1.05271920490492,
+      "grad_norm": 0.1601179987192154,
+      "learning_rate": 3.947288033761138e-06,
+      "loss": 0.8596,
+      "step": 145430
+    },
+    {
+      "epoch": 1.0527915915655064,
+      "grad_norm": 0.1553214192390442,
+      "learning_rate": 3.947215647100553e-06,
+      "loss": 0.8566,
+      "step": 145440
+    },
+    {
+      "epoch": 1.0528639782260925,
+      "grad_norm": 0.1488734483718872,
+      "learning_rate": 3.947143260439966e-06,
+      "loss": 0.8725,
+      "step": 145450
+    },
+    {
+      "epoch": 1.0529363648866787,
+      "grad_norm": 0.15073513984680176,
+      "learning_rate": 3.94707087377938e-06,
+      "loss": 0.8653,
+      "step": 145460
+    },
+    {
+      "epoch": 1.0530087515472648,
+      "grad_norm": 0.14692334830760956,
+      "learning_rate": 3.946998487118794e-06,
+      "loss": 0.8713,
+      "step": 145470
+    },
+    {
+      "epoch": 1.053081138207851,
+      "grad_norm": 0.14102409780025482,
+      "learning_rate": 3.946926100458208e-06,
+      "loss": 0.866,
+      "step": 145480
+    },
+    {
+      "epoch": 1.0531535248684372,
+      "grad_norm": 0.15512901544570923,
+      "learning_rate": 3.946853713797622e-06,
+      "loss": 0.8619,
+      "step": 145490
+    },
+    {
+      "epoch": 1.0532259115290235,
+      "grad_norm": 0.15089280903339386,
+      "learning_rate": 3.946781327137035e-06,
+      "loss": 0.8563,
+      "step": 145500
+    },
+    {
+      "epoch": 1.0532982981896095,
+      "grad_norm": 0.14949414134025574,
+      "learning_rate": 3.946708940476449e-06,
+      "loss": 0.8751,
+      "step": 145510
+    },
+    {
+      "epoch": 1.0533706848501958,
+      "grad_norm": 0.18709976971149445,
+      "learning_rate": 3.946636553815863e-06,
+      "loss": 0.8556,
+      "step": 145520
+    },
+    {
+      "epoch": 1.053443071510782,
+      "grad_norm": 0.15413489937782288,
+      "learning_rate": 3.946564167155277e-06,
+      "loss": 0.8572,
+      "step": 145530
+    },
+    {
+      "epoch": 1.0535154581713682,
+      "grad_norm": 0.16273820400238037,
+      "learning_rate": 3.946491780494691e-06,
+      "loss": 0.8619,
+      "step": 145540
+    },
+    {
+      "epoch": 1.0535878448319544,
+      "grad_norm": 0.1684591919183731,
+      "learning_rate": 3.946419393834104e-06,
+      "loss": 0.8659,
+      "step": 145550
+    },
+    {
+      "epoch": 1.0536602314925405,
+      "grad_norm": 0.14503608644008636,
+      "learning_rate": 3.946347007173518e-06,
+      "loss": 0.8567,
+      "step": 145560
+    },
+    {
+      "epoch": 1.0537326181531268,
+      "grad_norm": 0.1562882363796234,
+      "learning_rate": 3.946274620512932e-06,
+      "loss": 0.8636,
+      "step": 145570
+    },
+    {
+      "epoch": 1.0538050048137129,
+      "grad_norm": 0.1774781495332718,
+      "learning_rate": 3.946202233852346e-06,
+      "loss": 0.8586,
+      "step": 145580
+    },
+    {
+      "epoch": 1.0538773914742992,
+      "grad_norm": 0.1477038562297821,
+      "learning_rate": 3.9461298471917595e-06,
+      "loss": 0.8533,
+      "step": 145590
+    },
+    {
+      "epoch": 1.0539497781348852,
+      "grad_norm": 0.15526357293128967,
+      "learning_rate": 3.946057460531173e-06,
+      "loss": 0.8632,
+      "step": 145600
+    },
+    {
+      "epoch": 1.0540221647954715,
+      "grad_norm": 0.15866219997406006,
+      "learning_rate": 3.945985073870588e-06,
+      "loss": 0.8617,
+      "step": 145610
+    },
+    {
+      "epoch": 1.0540945514560576,
+      "grad_norm": 0.1477918028831482,
+      "learning_rate": 3.945912687210001e-06,
+      "loss": 0.8589,
+      "step": 145620
+    },
+    {
+      "epoch": 1.0541669381166439,
+      "grad_norm": 0.15647591650485992,
+      "learning_rate": 3.945840300549415e-06,
+      "loss": 0.858,
+      "step": 145630
+    },
+    {
+      "epoch": 1.0542393247772301,
+      "grad_norm": 0.15782573819160461,
+      "learning_rate": 3.9457679138888285e-06,
+      "loss": 0.8497,
+      "step": 145640
+    },
+    {
+      "epoch": 1.0543117114378162,
+      "grad_norm": 0.15249016880989075,
+      "learning_rate": 3.945695527228243e-06,
+      "loss": 0.8575,
+      "step": 145650
+    },
+    {
+      "epoch": 1.0543840980984025,
+      "grad_norm": 0.15300941467285156,
+      "learning_rate": 3.9456231405676565e-06,
+      "loss": 0.8714,
+      "step": 145660
+    },
+    {
+      "epoch": 1.0544564847589886,
+      "grad_norm": 0.15666401386260986,
+      "learning_rate": 3.94555075390707e-06,
+      "loss": 0.8728,
+      "step": 145670
+    },
+    {
+      "epoch": 1.0545288714195749,
+      "grad_norm": 0.13935892283916473,
+      "learning_rate": 3.945478367246484e-06,
+      "loss": 0.8641,
+      "step": 145680
+    },
+    {
+      "epoch": 1.054601258080161,
+      "grad_norm": 0.1481926590204239,
+      "learning_rate": 3.945405980585898e-06,
+      "loss": 0.8426,
+      "step": 145690
+    },
+    {
+      "epoch": 1.0546736447407472,
+      "grad_norm": 0.14881548285484314,
+      "learning_rate": 3.945333593925312e-06,
+      "loss": 0.8627,
+      "step": 145700
+    },
+    {
+      "epoch": 1.0547460314013333,
+      "grad_norm": 0.15182042121887207,
+      "learning_rate": 3.9452612072647255e-06,
+      "loss": 0.8521,
+      "step": 145710
+    },
+    {
+      "epoch": 1.0548184180619196,
+      "grad_norm": 0.14741186797618866,
+      "learning_rate": 3.945188820604139e-06,
+      "loss": 0.8544,
+      "step": 145720
+    },
+    {
+      "epoch": 1.0548908047225058,
+      "grad_norm": 0.1624380648136139,
+      "learning_rate": 3.9451164339435536e-06,
+      "loss": 0.8632,
+      "step": 145730
+    },
+    {
+      "epoch": 1.054963191383092,
+      "grad_norm": 0.14663566648960114,
+      "learning_rate": 3.945044047282967e-06,
+      "loss": 0.8584,
+      "step": 145740
+    },
+    {
+      "epoch": 1.0550355780436782,
+      "grad_norm": 0.1530272662639618,
+      "learning_rate": 3.944971660622381e-06,
+      "loss": 0.8637,
+      "step": 145750
+    },
+    {
+      "epoch": 1.0551079647042643,
+      "grad_norm": 0.14073669910430908,
+      "learning_rate": 3.944899273961794e-06,
+      "loss": 0.8538,
+      "step": 145760
+    },
+    {
+      "epoch": 1.0551803513648506,
+      "grad_norm": 0.1432807892560959,
+      "learning_rate": 3.944826887301209e-06,
+      "loss": 0.8553,
+      "step": 145770
+    },
+    {
+      "epoch": 1.0552527380254366,
+      "grad_norm": 0.15058694779872894,
+      "learning_rate": 3.9447545006406225e-06,
+      "loss": 0.8688,
+      "step": 145780
+    },
+    {
+      "epoch": 1.055325124686023,
+      "grad_norm": 0.14936070144176483,
+      "learning_rate": 3.944682113980036e-06,
+      "loss": 0.8571,
+      "step": 145790
+    },
+    {
+      "epoch": 1.055397511346609,
+      "grad_norm": 0.1473982185125351,
+      "learning_rate": 3.94460972731945e-06,
+      "loss": 0.8464,
+      "step": 145800
+    },
+    {
+      "epoch": 1.0554698980071953,
+      "grad_norm": 0.14957574009895325,
+      "learning_rate": 3.944537340658864e-06,
+      "loss": 0.863,
+      "step": 145810
+    },
+    {
+      "epoch": 1.0555422846677813,
+      "grad_norm": 0.16294653713703156,
+      "learning_rate": 3.944464953998278e-06,
+      "loss": 0.8702,
+      "step": 145820
+    },
+    {
+      "epoch": 1.0556146713283676,
+      "grad_norm": 0.15721982717514038,
+      "learning_rate": 3.944392567337691e-06,
+      "loss": 0.8644,
+      "step": 145830
+    },
+    {
+      "epoch": 1.055687057988954,
+      "grad_norm": 0.15910634398460388,
+      "learning_rate": 3.944320180677105e-06,
+      "loss": 0.8531,
+      "step": 145840
+    },
+    {
+      "epoch": 1.05575944464954,
+      "grad_norm": 0.14400066435337067,
+      "learning_rate": 3.9442477940165195e-06,
+      "loss": 0.8654,
+      "step": 145850
+    },
+    {
+      "epoch": 1.0558318313101263,
+      "grad_norm": 0.15780583024024963,
+      "learning_rate": 3.944175407355933e-06,
+      "loss": 0.838,
+      "step": 145860
+    },
+    {
+      "epoch": 1.0559042179707123,
+      "grad_norm": 0.14663490653038025,
+      "learning_rate": 3.944103020695347e-06,
+      "loss": 0.8495,
+      "step": 145870
+    },
+    {
+      "epoch": 1.0559766046312986,
+      "grad_norm": 0.14983581006526947,
+      "learning_rate": 3.94403063403476e-06,
+      "loss": 0.8636,
+      "step": 145880
+    },
+    {
+      "epoch": 1.0560489912918847,
+      "grad_norm": 0.15457117557525635,
+      "learning_rate": 3.943958247374175e-06,
+      "loss": 0.8544,
+      "step": 145890
+    },
+    {
+      "epoch": 1.056121377952471,
+      "grad_norm": 0.15235184133052826,
+      "learning_rate": 3.943885860713588e-06,
+      "loss": 0.8751,
+      "step": 145900
+    },
+    {
+      "epoch": 1.056193764613057,
+      "grad_norm": 0.14950567483901978,
+      "learning_rate": 3.943813474053002e-06,
+      "loss": 0.8557,
+      "step": 145910
+    },
+    {
+      "epoch": 1.0562661512736433,
+      "grad_norm": 0.15957853198051453,
+      "learning_rate": 3.943741087392416e-06,
+      "loss": 0.8671,
+      "step": 145920
+    },
+    {
+      "epoch": 1.0563385379342294,
+      "grad_norm": 0.15900222957134247,
+      "learning_rate": 3.94366870073183e-06,
+      "loss": 0.8576,
+      "step": 145930
+    },
+    {
+      "epoch": 1.0564109245948157,
+      "grad_norm": 0.16637851297855377,
+      "learning_rate": 3.943596314071244e-06,
+      "loss": 0.862,
+      "step": 145940
+    },
+    {
+      "epoch": 1.056483311255402,
+      "grad_norm": 0.14471204578876495,
+      "learning_rate": 3.9435239274106565e-06,
+      "loss": 0.8473,
+      "step": 145950
+    },
+    {
+      "epoch": 1.056555697915988,
+      "grad_norm": 0.16126057505607605,
+      "learning_rate": 3.943451540750071e-06,
+      "loss": 0.8646,
+      "step": 145960
+    },
+    {
+      "epoch": 1.0566280845765743,
+      "grad_norm": 0.15724702179431915,
+      "learning_rate": 3.943379154089485e-06,
+      "loss": 0.8513,
+      "step": 145970
+    },
+    {
+      "epoch": 1.0567004712371604,
+      "grad_norm": 0.14262165129184723,
+      "learning_rate": 3.943306767428898e-06,
+      "loss": 0.8657,
+      "step": 145980
+    },
+    {
+      "epoch": 1.0567728578977467,
+      "grad_norm": 0.15832731127738953,
+      "learning_rate": 3.943234380768312e-06,
+      "loss": 0.8564,
+      "step": 145990
+    },
+    {
+      "epoch": 1.0568452445583327,
+      "grad_norm": 0.16884136199951172,
+      "learning_rate": 3.943161994107726e-06,
+      "loss": 0.8436,
+      "step": 146000
+    },
+    {
+      "epoch": 1.056917631218919,
+      "grad_norm": 0.16266973316669464,
+      "learning_rate": 3.94308960744714e-06,
+      "loss": 0.866,
+      "step": 146010
+    },
+    {
+      "epoch": 1.056990017879505,
+      "grad_norm": 0.15265819430351257,
+      "learning_rate": 3.9430172207865535e-06,
+      "loss": 0.8672,
+      "step": 146020
+    },
+    {
+      "epoch": 1.0570624045400914,
+      "grad_norm": 0.14945794641971588,
+      "learning_rate": 3.942944834125967e-06,
+      "loss": 0.852,
+      "step": 146030
+    },
+    {
+      "epoch": 1.0571347912006774,
+      "grad_norm": 0.15644454956054688,
+      "learning_rate": 3.942872447465382e-06,
+      "loss": 0.8661,
+      "step": 146040
+    },
+    {
+      "epoch": 1.0572071778612637,
+      "grad_norm": 0.21663767099380493,
+      "learning_rate": 3.942800060804795e-06,
+      "loss": 0.8755,
+      "step": 146050
+    },
+    {
+      "epoch": 1.05727956452185,
+      "grad_norm": 0.16206009685993195,
+      "learning_rate": 3.942727674144209e-06,
+      "loss": 0.8679,
+      "step": 146060
+    },
+    {
+      "epoch": 1.057351951182436,
+      "grad_norm": 0.15549540519714355,
+      "learning_rate": 3.9426552874836224e-06,
+      "loss": 0.8578,
+      "step": 146070
+    },
+    {
+      "epoch": 1.0574243378430224,
+      "grad_norm": 0.16803592443466187,
+      "learning_rate": 3.942582900823037e-06,
+      "loss": 0.8662,
+      "step": 146080
+    },
+    {
+      "epoch": 1.0574967245036084,
+      "grad_norm": 0.1614563912153244,
+      "learning_rate": 3.9425105141624505e-06,
+      "loss": 0.8658,
+      "step": 146090
+    },
+    {
+      "epoch": 1.0575691111641947,
+      "grad_norm": 0.14888714253902435,
+      "learning_rate": 3.942438127501864e-06,
+      "loss": 0.8591,
+      "step": 146100
+    },
+    {
+      "epoch": 1.0576414978247808,
+      "grad_norm": 0.16620272397994995,
+      "learning_rate": 3.942365740841278e-06,
+      "loss": 0.8625,
+      "step": 146110
+    },
+    {
+      "epoch": 1.057713884485367,
+      "grad_norm": 0.28313207626342773,
+      "learning_rate": 3.942293354180692e-06,
+      "loss": 0.8607,
+      "step": 146120
+    },
+    {
+      "epoch": 1.0577862711459531,
+      "grad_norm": 0.17238175868988037,
+      "learning_rate": 3.942220967520106e-06,
+      "loss": 0.8728,
+      "step": 146130
+    },
+    {
+      "epoch": 1.0578586578065394,
+      "grad_norm": 0.14445391297340393,
+      "learning_rate": 3.9421485808595194e-06,
+      "loss": 0.8665,
+      "step": 146140
+    },
+    {
+      "epoch": 1.0579310444671255,
+      "grad_norm": 0.15155573189258575,
+      "learning_rate": 3.942076194198933e-06,
+      "loss": 0.8549,
+      "step": 146150
+    },
+    {
+      "epoch": 1.0580034311277118,
+      "grad_norm": 0.16635553538799286,
+      "learning_rate": 3.9420038075383475e-06,
+      "loss": 0.8578,
+      "step": 146160
+    },
+    {
+      "epoch": 1.058075817788298,
+      "grad_norm": 0.19425910711288452,
+      "learning_rate": 3.941931420877761e-06,
+      "loss": 0.8583,
+      "step": 146170
+    },
+    {
+      "epoch": 1.0581482044488841,
+      "grad_norm": 0.14456264674663544,
+      "learning_rate": 3.941859034217175e-06,
+      "loss": 0.8481,
+      "step": 146180
+    },
+    {
+      "epoch": 1.0582205911094704,
+      "grad_norm": 0.15371955931186676,
+      "learning_rate": 3.941786647556588e-06,
+      "loss": 0.8503,
+      "step": 146190
+    },
+    {
+      "epoch": 1.0582929777700565,
+      "grad_norm": 0.14967955648899078,
+      "learning_rate": 3.941714260896002e-06,
+      "loss": 0.8658,
+      "step": 146200
+    },
+    {
+      "epoch": 1.0583653644306428,
+      "grad_norm": 0.14629121124744415,
+      "learning_rate": 3.9416418742354164e-06,
+      "loss": 0.8577,
+      "step": 146210
+    },
+    {
+      "epoch": 1.0584377510912288,
+      "grad_norm": 0.13868474960327148,
+      "learning_rate": 3.94156948757483e-06,
+      "loss": 0.8489,
+      "step": 146220
+    },
+    {
+      "epoch": 1.0585101377518151,
+      "grad_norm": 0.16160985827445984,
+      "learning_rate": 3.941497100914244e-06,
+      "loss": 0.8684,
+      "step": 146230
+    },
+    {
+      "epoch": 1.0585825244124012,
+      "grad_norm": 0.23618648946285248,
+      "learning_rate": 3.941424714253657e-06,
+      "loss": 0.8524,
+      "step": 146240
+    },
+    {
+      "epoch": 1.0586549110729875,
+      "grad_norm": 0.15129047632217407,
+      "learning_rate": 3.941352327593072e-06,
+      "loss": 0.8782,
+      "step": 146250
+    },
+    {
+      "epoch": 1.0587272977335735,
+      "grad_norm": 0.144764244556427,
+      "learning_rate": 3.941279940932485e-06,
+      "loss": 0.8546,
+      "step": 146260
+    },
+    {
+      "epoch": 1.0587996843941598,
+      "grad_norm": 0.14293217658996582,
+      "learning_rate": 3.941207554271899e-06,
+      "loss": 0.8508,
+      "step": 146270
+    },
+    {
+      "epoch": 1.0588720710547461,
+      "grad_norm": 0.1651974767446518,
+      "learning_rate": 3.941135167611313e-06,
+      "loss": 0.8733,
+      "step": 146280
+    },
+    {
+      "epoch": 1.0589444577153322,
+      "grad_norm": 0.15078642964363098,
+      "learning_rate": 3.941062780950727e-06,
+      "loss": 0.8563,
+      "step": 146290
+    },
+    {
+      "epoch": 1.0590168443759185,
+      "grad_norm": 0.1709497570991516,
+      "learning_rate": 3.940990394290141e-06,
+      "loss": 0.8564,
+      "step": 146300
+    },
+    {
+      "epoch": 1.0590892310365045,
+      "grad_norm": 0.15568095445632935,
+      "learning_rate": 3.940918007629554e-06,
+      "loss": 0.8716,
+      "step": 146310
+    },
+    {
+      "epoch": 1.0591616176970908,
+      "grad_norm": 0.15851420164108276,
+      "learning_rate": 3.940845620968968e-06,
+      "loss": 0.861,
+      "step": 146320
+    },
+    {
+      "epoch": 1.059234004357677,
+      "grad_norm": 0.15701240301132202,
+      "learning_rate": 3.940773234308382e-06,
+      "loss": 0.861,
+      "step": 146330
+    },
+    {
+      "epoch": 1.0593063910182632,
+      "grad_norm": 0.160984069108963,
+      "learning_rate": 3.940700847647796e-06,
+      "loss": 0.8659,
+      "step": 146340
+    },
+    {
+      "epoch": 1.0593787776788492,
+      "grad_norm": 0.15175053477287292,
+      "learning_rate": 3.94062846098721e-06,
+      "loss": 0.8628,
+      "step": 146350
+    },
+    {
+      "epoch": 1.0594511643394355,
+      "grad_norm": 0.15705060958862305,
+      "learning_rate": 3.940556074326623e-06,
+      "loss": 0.8726,
+      "step": 146360
+    },
+    {
+      "epoch": 1.0595235510000216,
+      "grad_norm": 0.1432766169309616,
+      "learning_rate": 3.940483687666038e-06,
+      "loss": 0.8457,
+      "step": 146370
+    },
+    {
+      "epoch": 1.059595937660608,
+      "grad_norm": 0.15701285004615784,
+      "learning_rate": 3.940411301005451e-06,
+      "loss": 0.8669,
+      "step": 146380
+    },
+    {
+      "epoch": 1.0596683243211942,
+      "grad_norm": 0.15151244401931763,
+      "learning_rate": 3.940338914344865e-06,
+      "loss": 0.8623,
+      "step": 146390
+    },
+    {
+      "epoch": 1.0597407109817802,
+      "grad_norm": 0.19881872832775116,
+      "learning_rate": 3.9402665276842785e-06,
+      "loss": 0.8554,
+      "step": 146400
+    },
+    {
+      "epoch": 1.0598130976423665,
+      "grad_norm": 0.14868730306625366,
+      "learning_rate": 3.940194141023693e-06,
+      "loss": 0.8686,
+      "step": 146410
+    },
+    {
+      "epoch": 1.0598854843029526,
+      "grad_norm": 0.14795683324337006,
+      "learning_rate": 3.940121754363107e-06,
+      "loss": 0.858,
+      "step": 146420
+    },
+    {
+      "epoch": 1.0599578709635389,
+      "grad_norm": 0.144784614443779,
+      "learning_rate": 3.94004936770252e-06,
+      "loss": 0.8613,
+      "step": 146430
+    },
+    {
+      "epoch": 1.060030257624125,
+      "grad_norm": 0.15263043344020844,
+      "learning_rate": 3.939976981041934e-06,
+      "loss": 0.8662,
+      "step": 146440
+    },
+    {
+      "epoch": 1.0601026442847112,
+      "grad_norm": 0.1521047055721283,
+      "learning_rate": 3.939904594381348e-06,
+      "loss": 0.8434,
+      "step": 146450
+    },
+    {
+      "epoch": 1.0601750309452973,
+      "grad_norm": 0.15608127415180206,
+      "learning_rate": 3.939832207720762e-06,
+      "loss": 0.855,
+      "step": 146460
+    },
+    {
+      "epoch": 1.0602474176058836,
+      "grad_norm": 0.15321072936058044,
+      "learning_rate": 3.9397598210601756e-06,
+      "loss": 0.8676,
+      "step": 146470
+    },
+    {
+      "epoch": 1.0603198042664699,
+      "grad_norm": 0.175469309091568,
+      "learning_rate": 3.939687434399589e-06,
+      "loss": 0.857,
+      "step": 146480
+    },
+    {
+      "epoch": 1.060392190927056,
+      "grad_norm": 0.15530693531036377,
+      "learning_rate": 3.939615047739003e-06,
+      "loss": 0.8568,
+      "step": 146490
+    },
+    {
+      "epoch": 1.0604645775876422,
+      "grad_norm": 0.14557047188282013,
+      "learning_rate": 3.939542661078416e-06,
+      "loss": 0.8633,
+      "step": 146500
+    },
+    {
+      "epoch": 1.0605369642482283,
+      "grad_norm": 0.1573035567998886,
+      "learning_rate": 3.93947027441783e-06,
+      "loss": 0.8631,
+      "step": 146510
+    },
+    {
+      "epoch": 1.0606093509088146,
+      "grad_norm": 0.15351547300815582,
+      "learning_rate": 3.9393978877572445e-06,
+      "loss": 0.8401,
+      "step": 146520
+    },
+    {
+      "epoch": 1.0606817375694007,
+      "grad_norm": 0.21780115365982056,
+      "learning_rate": 3.939325501096658e-06,
+      "loss": 0.8566,
+      "step": 146530
+    },
+    {
+      "epoch": 1.060754124229987,
+      "grad_norm": 0.19611738622188568,
+      "learning_rate": 3.939253114436072e-06,
+      "loss": 0.8488,
+      "step": 146540
+    },
+    {
+      "epoch": 1.060826510890573,
+      "grad_norm": 0.1587083786725998,
+      "learning_rate": 3.939180727775485e-06,
+      "loss": 0.8542,
+      "step": 146550
+    },
+    {
+      "epoch": 1.0608988975511593,
+      "grad_norm": 0.17451640963554382,
+      "learning_rate": 3.9391083411149e-06,
+      "loss": 0.8686,
+      "step": 146560
+    },
+    {
+      "epoch": 1.0609712842117454,
+      "grad_norm": 0.1856866180896759,
+      "learning_rate": 3.939035954454313e-06,
+      "loss": 0.8569,
+      "step": 146570
+    },
+    {
+      "epoch": 1.0610436708723316,
+      "grad_norm": 0.14287951588630676,
+      "learning_rate": 3.938963567793727e-06,
+      "loss": 0.8559,
+      "step": 146580
+    },
+    {
+      "epoch": 1.061116057532918,
+      "grad_norm": 0.17988088726997375,
+      "learning_rate": 3.938891181133141e-06,
+      "loss": 0.8711,
+      "step": 146590
+    },
+    {
+      "epoch": 1.061188444193504,
+      "grad_norm": 0.1705288589000702,
+      "learning_rate": 3.938818794472555e-06,
+      "loss": 0.8664,
+      "step": 146600
+    },
+    {
+      "epoch": 1.0612608308540903,
+      "grad_norm": 0.16595390439033508,
+      "learning_rate": 3.938746407811969e-06,
+      "loss": 0.8768,
+      "step": 146610
+    },
+    {
+      "epoch": 1.0613332175146764,
+      "grad_norm": 0.15512683987617493,
+      "learning_rate": 3.938674021151382e-06,
+      "loss": 0.86,
+      "step": 146620
+    },
+    {
+      "epoch": 1.0614056041752626,
+      "grad_norm": 0.1589556485414505,
+      "learning_rate": 3.938601634490796e-06,
+      "loss": 0.8608,
+      "step": 146630
+    },
+    {
+      "epoch": 1.0614779908358487,
+      "grad_norm": 0.1618974655866623,
+      "learning_rate": 3.93852924783021e-06,
+      "loss": 0.8557,
+      "step": 146640
+    },
+    {
+      "epoch": 1.061550377496435,
+      "grad_norm": 0.1717182695865631,
+      "learning_rate": 3.938456861169624e-06,
+      "loss": 0.8621,
+      "step": 146650
+    },
+    {
+      "epoch": 1.061622764157021,
+      "grad_norm": 0.1516030728816986,
+      "learning_rate": 3.938384474509038e-06,
+      "loss": 0.8814,
+      "step": 146660
+    },
+    {
+      "epoch": 1.0616951508176073,
+      "grad_norm": 0.14224623143672943,
+      "learning_rate": 3.938312087848451e-06,
+      "loss": 0.8649,
+      "step": 146670
+    },
+    {
+      "epoch": 1.0617675374781934,
+      "grad_norm": 0.1500004678964615,
+      "learning_rate": 3.938239701187866e-06,
+      "loss": 0.8508,
+      "step": 146680
+    },
+    {
+      "epoch": 1.0618399241387797,
+      "grad_norm": 0.14933015406131744,
+      "learning_rate": 3.938167314527279e-06,
+      "loss": 0.8643,
+      "step": 146690
+    },
+    {
+      "epoch": 1.061912310799366,
+      "grad_norm": 0.15696154534816742,
+      "learning_rate": 3.938094927866693e-06,
+      "loss": 0.8618,
+      "step": 146700
+    },
+    {
+      "epoch": 1.061984697459952,
+      "grad_norm": 0.1518896073102951,
+      "learning_rate": 3.9380225412061066e-06,
+      "loss": 0.8556,
+      "step": 146710
+    },
+    {
+      "epoch": 1.0620570841205383,
+      "grad_norm": 0.1495170295238495,
+      "learning_rate": 3.937950154545521e-06,
+      "loss": 0.855,
+      "step": 146720
+    },
+    {
+      "epoch": 1.0621294707811244,
+      "grad_norm": 0.15522539615631104,
+      "learning_rate": 3.937877767884935e-06,
+      "loss": 0.8592,
+      "step": 146730
+    },
+    {
+      "epoch": 1.0622018574417107,
+      "grad_norm": 0.14690908789634705,
+      "learning_rate": 3.937805381224348e-06,
+      "loss": 0.8489,
+      "step": 146740
+    },
+    {
+      "epoch": 1.0622742441022968,
+      "grad_norm": 0.17294001579284668,
+      "learning_rate": 3.937732994563762e-06,
+      "loss": 0.8548,
+      "step": 146750
+    },
+    {
+      "epoch": 1.062346630762883,
+      "grad_norm": 0.16013190150260925,
+      "learning_rate": 3.937660607903176e-06,
+      "loss": 0.8541,
+      "step": 146760
+    },
+    {
+      "epoch": 1.0624190174234691,
+      "grad_norm": 0.15690450370311737,
+      "learning_rate": 3.93758822124259e-06,
+      "loss": 0.8535,
+      "step": 146770
+    },
+    {
+      "epoch": 1.0624914040840554,
+      "grad_norm": 0.1489821821451187,
+      "learning_rate": 3.937515834582004e-06,
+      "loss": 0.8614,
+      "step": 146780
+    },
+    {
+      "epoch": 1.0625637907446417,
+      "grad_norm": 0.15481127798557281,
+      "learning_rate": 3.937443447921417e-06,
+      "loss": 0.8483,
+      "step": 146790
+    },
+    {
+      "epoch": 1.0626361774052278,
+      "grad_norm": 0.16512636840343475,
+      "learning_rate": 3.937371061260832e-06,
+      "loss": 0.8513,
+      "step": 146800
+    },
+    {
+      "epoch": 1.062708564065814,
+      "grad_norm": 0.1462734192609787,
+      "learning_rate": 3.937298674600245e-06,
+      "loss": 0.8618,
+      "step": 146810
+    },
+    {
+      "epoch": 1.0627809507264,
+      "grad_norm": 0.1540970355272293,
+      "learning_rate": 3.937226287939659e-06,
+      "loss": 0.8699,
+      "step": 146820
+    },
+    {
+      "epoch": 1.0628533373869864,
+      "grad_norm": 0.14309470355510712,
+      "learning_rate": 3.9371539012790725e-06,
+      "loss": 0.8544,
+      "step": 146830
+    },
+    {
+      "epoch": 1.0629257240475725,
+      "grad_norm": 0.14380641281604767,
+      "learning_rate": 3.937081514618486e-06,
+      "loss": 0.8579,
+      "step": 146840
+    },
+    {
+      "epoch": 1.0629981107081587,
+      "grad_norm": 0.15650497376918793,
+      "learning_rate": 3.937009127957901e-06,
+      "loss": 0.8614,
+      "step": 146850
+    },
+    {
+      "epoch": 1.0630704973687448,
+      "grad_norm": 0.15885330736637115,
+      "learning_rate": 3.936936741297314e-06,
+      "loss": 0.8604,
+      "step": 146860
+    },
+    {
+      "epoch": 1.063142884029331,
+      "grad_norm": 0.15795765817165375,
+      "learning_rate": 3.936864354636728e-06,
+      "loss": 0.8521,
+      "step": 146870
+    },
+    {
+      "epoch": 1.0632152706899172,
+      "grad_norm": 0.15613383054733276,
+      "learning_rate": 3.9367919679761414e-06,
+      "loss": 0.8605,
+      "step": 146880
+    },
+    {
+      "epoch": 1.0632876573505035,
+      "grad_norm": 0.14996646344661713,
+      "learning_rate": 3.936719581315556e-06,
+      "loss": 0.863,
+      "step": 146890
+    },
+    {
+      "epoch": 1.0633600440110897,
+      "grad_norm": 0.1522240936756134,
+      "learning_rate": 3.9366471946549695e-06,
+      "loss": 0.8565,
+      "step": 146900
+    },
+    {
+      "epoch": 1.0634324306716758,
+      "grad_norm": 0.1428409069776535,
+      "learning_rate": 3.936574807994383e-06,
+      "loss": 0.8535,
+      "step": 146910
+    },
+    {
+      "epoch": 1.063504817332262,
+      "grad_norm": 0.14927205443382263,
+      "learning_rate": 3.936502421333797e-06,
+      "loss": 0.8608,
+      "step": 146920
+    },
+    {
+      "epoch": 1.0635772039928482,
+      "grad_norm": 0.15207083523273468,
+      "learning_rate": 3.936430034673211e-06,
+      "loss": 0.8657,
+      "step": 146930
+    },
+    {
+      "epoch": 1.0636495906534345,
+      "grad_norm": 0.15743815898895264,
+      "learning_rate": 3.936357648012625e-06,
+      "loss": 0.8576,
+      "step": 146940
+    },
+    {
+      "epoch": 1.0637219773140205,
+      "grad_norm": 0.14594559371471405,
+      "learning_rate": 3.9362852613520384e-06,
+      "loss": 0.8709,
+      "step": 146950
+    },
+    {
+      "epoch": 1.0637943639746068,
+      "grad_norm": 0.1405821442604065,
+      "learning_rate": 3.936212874691452e-06,
+      "loss": 0.8572,
+      "step": 146960
+    },
+    {
+      "epoch": 1.0638667506351929,
+      "grad_norm": 0.15676797926425934,
+      "learning_rate": 3.9361404880308665e-06,
+      "loss": 0.8538,
+      "step": 146970
+    },
+    {
+      "epoch": 1.0639391372957792,
+      "grad_norm": 0.14403656125068665,
+      "learning_rate": 3.93606810137028e-06,
+      "loss": 0.8515,
+      "step": 146980
+    },
+    {
+      "epoch": 1.0640115239563652,
+      "grad_norm": 0.14301113784313202,
+      "learning_rate": 3.935995714709694e-06,
+      "loss": 0.8572,
+      "step": 146990
+    },
+    {
+      "epoch": 1.0640839106169515,
+      "grad_norm": 0.15309615433216095,
+      "learning_rate": 3.935923328049107e-06,
+      "loss": 0.861,
+      "step": 147000
+    },
+    {
+      "epoch": 1.0641562972775378,
+      "grad_norm": 0.3443004786968231,
+      "learning_rate": 3.935850941388521e-06,
+      "loss": 0.8656,
+      "step": 147010
+    },
+    {
+      "epoch": 1.0642286839381239,
+      "grad_norm": 0.13839933276176453,
+      "learning_rate": 3.935778554727935e-06,
+      "loss": 0.8555,
+      "step": 147020
+    },
+    {
+      "epoch": 1.0643010705987102,
+      "grad_norm": 0.14977069199085236,
+      "learning_rate": 3.935706168067348e-06,
+      "loss": 0.8623,
+      "step": 147030
+    },
+    {
+      "epoch": 1.0643734572592962,
+      "grad_norm": 0.166509747505188,
+      "learning_rate": 3.935633781406763e-06,
+      "loss": 0.8703,
+      "step": 147040
+    },
+    {
+      "epoch": 1.0644458439198825,
+      "grad_norm": 0.15191148221492767,
+      "learning_rate": 3.935561394746176e-06,
+      "loss": 0.8508,
+      "step": 147050
+    },
+    {
+      "epoch": 1.0645182305804686,
+      "grad_norm": 0.15983009338378906,
+      "learning_rate": 3.93548900808559e-06,
+      "loss": 0.8578,
+      "step": 147060
+    },
+    {
+      "epoch": 1.0645906172410549,
+      "grad_norm": 0.1482071876525879,
+      "learning_rate": 3.9354166214250035e-06,
+      "loss": 0.8419,
+      "step": 147070
+    },
+    {
+      "epoch": 1.064663003901641,
+      "grad_norm": 0.1566699892282486,
+      "learning_rate": 3.935344234764418e-06,
+      "loss": 0.8658,
+      "step": 147080
+    },
+    {
+      "epoch": 1.0647353905622272,
+      "grad_norm": 0.14756731688976288,
+      "learning_rate": 3.935271848103832e-06,
+      "loss": 0.8493,
+      "step": 147090
+    },
+    {
+      "epoch": 1.0648077772228133,
+      "grad_norm": 0.15499506890773773,
+      "learning_rate": 3.935199461443245e-06,
+      "loss": 0.8484,
+      "step": 147100
+    },
+    {
+      "epoch": 1.0648801638833996,
+      "grad_norm": 0.15901382267475128,
+      "learning_rate": 3.935127074782659e-06,
+      "loss": 0.8613,
+      "step": 147110
+    },
+    {
+      "epoch": 1.0649525505439859,
+      "grad_norm": 0.1550762802362442,
+      "learning_rate": 3.935054688122073e-06,
+      "loss": 0.8441,
+      "step": 147120
+    },
+    {
+      "epoch": 1.065024937204572,
+      "grad_norm": 0.17926537990570068,
+      "learning_rate": 3.934982301461487e-06,
+      "loss": 0.8558,
+      "step": 147130
+    },
+    {
+      "epoch": 1.0650973238651582,
+      "grad_norm": 0.31348535418510437,
+      "learning_rate": 3.9349099148009005e-06,
+      "loss": 0.8624,
+      "step": 147140
+    },
+    {
+      "epoch": 1.0651697105257443,
+      "grad_norm": 0.15891113877296448,
+      "learning_rate": 3.934837528140314e-06,
+      "loss": 0.8495,
+      "step": 147150
+    },
+    {
+      "epoch": 1.0652420971863306,
+      "grad_norm": 0.15291360020637512,
+      "learning_rate": 3.934765141479729e-06,
+      "loss": 0.8455,
+      "step": 147160
+    },
+    {
+      "epoch": 1.0653144838469166,
+      "grad_norm": 0.25229060649871826,
+      "learning_rate": 3.934692754819142e-06,
+      "loss": 0.8589,
+      "step": 147170
+    },
+    {
+      "epoch": 1.065386870507503,
+      "grad_norm": 0.17564287781715393,
+      "learning_rate": 3.934620368158556e-06,
+      "loss": 0.8675,
+      "step": 147180
+    },
+    {
+      "epoch": 1.065459257168089,
+      "grad_norm": 0.15342505276203156,
+      "learning_rate": 3.9345479814979695e-06,
+      "loss": 0.8534,
+      "step": 147190
+    },
+    {
+      "epoch": 1.0655316438286753,
+      "grad_norm": 0.15959271788597107,
+      "learning_rate": 3.934475594837384e-06,
+      "loss": 0.8614,
+      "step": 147200
+    },
+    {
+      "epoch": 1.0656040304892613,
+      "grad_norm": 0.15670059621334076,
+      "learning_rate": 3.9344032081767976e-06,
+      "loss": 0.8668,
+      "step": 147210
+    },
+    {
+      "epoch": 1.0656764171498476,
+      "grad_norm": 0.16998760402202606,
+      "learning_rate": 3.934330821516211e-06,
+      "loss": 0.8682,
+      "step": 147220
+    },
+    {
+      "epoch": 1.065748803810434,
+      "grad_norm": 0.1576838493347168,
+      "learning_rate": 3.934258434855625e-06,
+      "loss": 0.848,
+      "step": 147230
+    },
+    {
+      "epoch": 1.06582119047102,
+      "grad_norm": 0.13807885348796844,
+      "learning_rate": 3.934186048195039e-06,
+      "loss": 0.8532,
+      "step": 147240
+    },
+    {
+      "epoch": 1.0658935771316063,
+      "grad_norm": 0.15785497426986694,
+      "learning_rate": 3.934113661534453e-06,
+      "loss": 0.8594,
+      "step": 147250
+    },
+    {
+      "epoch": 1.0659659637921923,
+      "grad_norm": 0.16951316595077515,
+      "learning_rate": 3.9340412748738665e-06,
+      "loss": 0.8581,
+      "step": 147260
+    },
+    {
+      "epoch": 1.0660383504527786,
+      "grad_norm": 0.2050899863243103,
+      "learning_rate": 3.93396888821328e-06,
+      "loss": 0.8607,
+      "step": 147270
+    },
+    {
+      "epoch": 1.0661107371133647,
+      "grad_norm": 0.17105598747730255,
+      "learning_rate": 3.9338965015526946e-06,
+      "loss": 0.8594,
+      "step": 147280
+    },
+    {
+      "epoch": 1.066183123773951,
+      "grad_norm": 0.15434543788433075,
+      "learning_rate": 3.933824114892108e-06,
+      "loss": 0.8475,
+      "step": 147290
+    },
+    {
+      "epoch": 1.066255510434537,
+      "grad_norm": 0.1630607694387436,
+      "learning_rate": 3.933751728231522e-06,
+      "loss": 0.8562,
+      "step": 147300
+    },
+    {
+      "epoch": 1.0663278970951233,
+      "grad_norm": 0.15573401749134064,
+      "learning_rate": 3.933679341570935e-06,
+      "loss": 0.8537,
+      "step": 147310
+    },
+    {
+      "epoch": 1.0664002837557094,
+      "grad_norm": 0.15144288539886475,
+      "learning_rate": 3.93360695491035e-06,
+      "loss": 0.8598,
+      "step": 147320
+    },
+    {
+      "epoch": 1.0664726704162957,
+      "grad_norm": 0.14449411630630493,
+      "learning_rate": 3.9335345682497635e-06,
+      "loss": 0.8529,
+      "step": 147330
+    },
+    {
+      "epoch": 1.066545057076882,
+      "grad_norm": 0.15416362881660461,
+      "learning_rate": 3.933462181589177e-06,
+      "loss": 0.8507,
+      "step": 147340
+    },
+    {
+      "epoch": 1.066617443737468,
+      "grad_norm": 0.1476205438375473,
+      "learning_rate": 3.933389794928591e-06,
+      "loss": 0.8652,
+      "step": 147350
+    },
+    {
+      "epoch": 1.0666898303980543,
+      "grad_norm": 0.1484995037317276,
+      "learning_rate": 3.933317408268005e-06,
+      "loss": 0.8602,
+      "step": 147360
+    },
+    {
+      "epoch": 1.0667622170586404,
+      "grad_norm": 0.15167966485023499,
+      "learning_rate": 3.933245021607419e-06,
+      "loss": 0.841,
+      "step": 147370
+    },
+    {
+      "epoch": 1.0668346037192267,
+      "grad_norm": 0.15195885300636292,
+      "learning_rate": 3.933172634946832e-06,
+      "loss": 0.8577,
+      "step": 147380
+    },
+    {
+      "epoch": 1.0669069903798127,
+      "grad_norm": 0.15130667388439178,
+      "learning_rate": 3.933100248286246e-06,
+      "loss": 0.8491,
+      "step": 147390
+    },
+    {
+      "epoch": 1.066979377040399,
+      "grad_norm": 0.15865103900432587,
+      "learning_rate": 3.9330278616256605e-06,
+      "loss": 0.8707,
+      "step": 147400
+    },
+    {
+      "epoch": 1.067051763700985,
+      "grad_norm": 0.15017808973789215,
+      "learning_rate": 3.932955474965074e-06,
+      "loss": 0.8559,
+      "step": 147410
+    },
+    {
+      "epoch": 1.0671241503615714,
+      "grad_norm": 0.15235696732997894,
+      "learning_rate": 3.932883088304488e-06,
+      "loss": 0.8593,
+      "step": 147420
+    },
+    {
+      "epoch": 1.0671965370221574,
+      "grad_norm": 0.19113440811634064,
+      "learning_rate": 3.932810701643901e-06,
+      "loss": 0.8677,
+      "step": 147430
+    },
+    {
+      "epoch": 1.0672689236827437,
+      "grad_norm": 0.19408351182937622,
+      "learning_rate": 3.932738314983315e-06,
+      "loss": 0.854,
+      "step": 147440
+    },
+    {
+      "epoch": 1.06734131034333,
+      "grad_norm": 0.15390466153621674,
+      "learning_rate": 3.932665928322729e-06,
+      "loss": 0.8565,
+      "step": 147450
+    },
+    {
+      "epoch": 1.067413697003916,
+      "grad_norm": 0.1523483693599701,
+      "learning_rate": 3.932593541662143e-06,
+      "loss": 0.8531,
+      "step": 147460
+    },
+    {
+      "epoch": 1.0674860836645024,
+      "grad_norm": 0.17553357779979706,
+      "learning_rate": 3.932521155001557e-06,
+      "loss": 0.8569,
+      "step": 147470
+    },
+    {
+      "epoch": 1.0675584703250884,
+      "grad_norm": 0.19476906955242157,
+      "learning_rate": 3.93244876834097e-06,
+      "loss": 0.8691,
+      "step": 147480
+    },
+    {
+      "epoch": 1.0676308569856747,
+      "grad_norm": 0.2501210570335388,
+      "learning_rate": 3.932376381680385e-06,
+      "loss": 0.8737,
+      "step": 147490
+    },
+    {
+      "epoch": 1.0677032436462608,
+      "grad_norm": 0.17125388979911804,
+      "learning_rate": 3.932303995019798e-06,
+      "loss": 0.8577,
+      "step": 147500
+    },
+    {
+      "epoch": 1.067775630306847,
+      "grad_norm": 0.15461327135562897,
+      "learning_rate": 3.932231608359212e-06,
+      "loss": 0.8669,
+      "step": 147510
+    },
+    {
+      "epoch": 1.0678480169674331,
+      "grad_norm": 0.1496717482805252,
+      "learning_rate": 3.932159221698626e-06,
+      "loss": 0.864,
+      "step": 147520
+    },
+    {
+      "epoch": 1.0679204036280194,
+      "grad_norm": 0.14370813965797424,
+      "learning_rate": 3.93208683503804e-06,
+      "loss": 0.8496,
+      "step": 147530
+    },
+    {
+      "epoch": 1.0679927902886055,
+      "grad_norm": 0.16169150173664093,
+      "learning_rate": 3.932014448377453e-06,
+      "loss": 0.8587,
+      "step": 147540
+    },
+    {
+      "epoch": 1.0680651769491918,
+      "grad_norm": 0.15165531635284424,
+      "learning_rate": 3.931942061716867e-06,
+      "loss": 0.8602,
+      "step": 147550
+    },
+    {
+      "epoch": 1.068137563609778,
+      "grad_norm": 0.14875246584415436,
+      "learning_rate": 3.931869675056281e-06,
+      "loss": 0.8488,
+      "step": 147560
+    },
+    {
+      "epoch": 1.0682099502703641,
+      "grad_norm": 0.15369826555252075,
+      "learning_rate": 3.9317972883956945e-06,
+      "loss": 0.8615,
+      "step": 147570
+    },
+    {
+      "epoch": 1.0682823369309504,
+      "grad_norm": 0.15338794887065887,
+      "learning_rate": 3.931724901735108e-06,
+      "loss": 0.8682,
+      "step": 147580
+    },
+    {
+      "epoch": 1.0683547235915365,
+      "grad_norm": 0.17521625757217407,
+      "learning_rate": 3.931652515074523e-06,
+      "loss": 0.8585,
+      "step": 147590
+    },
+    {
+      "epoch": 1.0684271102521228,
+      "grad_norm": 0.15786945819854736,
+      "learning_rate": 3.931580128413936e-06,
+      "loss": 0.8668,
+      "step": 147600
+    },
+    {
+      "epoch": 1.0684994969127088,
+      "grad_norm": 0.16188044846057892,
+      "learning_rate": 3.93150774175335e-06,
+      "loss": 0.8529,
+      "step": 147610
+    },
+    {
+      "epoch": 1.0685718835732951,
+      "grad_norm": 0.15496951341629028,
+      "learning_rate": 3.9314353550927634e-06,
+      "loss": 0.8538,
+      "step": 147620
+    },
+    {
+      "epoch": 1.0686442702338812,
+      "grad_norm": 0.1538427770137787,
+      "learning_rate": 3.931362968432177e-06,
+      "loss": 0.871,
+      "step": 147630
+    },
+    {
+      "epoch": 1.0687166568944675,
+      "grad_norm": 0.14457300305366516,
+      "learning_rate": 3.9312905817715915e-06,
+      "loss": 0.8551,
+      "step": 147640
+    },
+    {
+      "epoch": 1.0687890435550538,
+      "grad_norm": 0.14885105192661285,
+      "learning_rate": 3.931218195111005e-06,
+      "loss": 0.8498,
+      "step": 147650
+    },
+    {
+      "epoch": 1.0688614302156398,
+      "grad_norm": 0.16504499316215515,
+      "learning_rate": 3.931145808450419e-06,
+      "loss": 0.853,
+      "step": 147660
+    },
+    {
+      "epoch": 1.0689338168762261,
+      "grad_norm": 0.15119002759456635,
+      "learning_rate": 3.931073421789832e-06,
+      "loss": 0.8574,
+      "step": 147670
+    },
+    {
+      "epoch": 1.0690062035368122,
+      "grad_norm": 0.15814948081970215,
+      "learning_rate": 3.931001035129247e-06,
+      "loss": 0.8593,
+      "step": 147680
+    },
+    {
+      "epoch": 1.0690785901973985,
+      "grad_norm": 0.14475545287132263,
+      "learning_rate": 3.9309286484686604e-06,
+      "loss": 0.8555,
+      "step": 147690
+    },
+    {
+      "epoch": 1.0691509768579845,
+      "grad_norm": 0.1500001847743988,
+      "learning_rate": 3.930856261808074e-06,
+      "loss": 0.8552,
+      "step": 147700
+    },
+    {
+      "epoch": 1.0692233635185708,
+      "grad_norm": 0.15069542825222015,
+      "learning_rate": 3.930783875147488e-06,
+      "loss": 0.8675,
+      "step": 147710
+    },
+    {
+      "epoch": 1.069295750179157,
+      "grad_norm": 0.14854957163333893,
+      "learning_rate": 3.930711488486902e-06,
+      "loss": 0.8658,
+      "step": 147720
+    },
+    {
+      "epoch": 1.0693681368397432,
+      "grad_norm": 0.17693032324314117,
+      "learning_rate": 3.930639101826316e-06,
+      "loss": 0.8453,
+      "step": 147730
+    },
+    {
+      "epoch": 1.0694405235003295,
+      "grad_norm": 0.15209737420082092,
+      "learning_rate": 3.930566715165729e-06,
+      "loss": 0.8566,
+      "step": 147740
+    },
+    {
+      "epoch": 1.0695129101609155,
+      "grad_norm": 0.1621839553117752,
+      "learning_rate": 3.930494328505143e-06,
+      "loss": 0.8516,
+      "step": 147750
+    },
+    {
+      "epoch": 1.0695852968215018,
+      "grad_norm": 0.14790567755699158,
+      "learning_rate": 3.9304219418445575e-06,
+      "loss": 0.8514,
+      "step": 147760
+    },
+    {
+      "epoch": 1.069657683482088,
+      "grad_norm": 0.14647607505321503,
+      "learning_rate": 3.930349555183971e-06,
+      "loss": 0.8419,
+      "step": 147770
+    },
+    {
+      "epoch": 1.0697300701426742,
+      "grad_norm": 0.16838887333869934,
+      "learning_rate": 3.930277168523385e-06,
+      "loss": 0.8465,
+      "step": 147780
+    },
+    {
+      "epoch": 1.0698024568032602,
+      "grad_norm": 0.1428205966949463,
+      "learning_rate": 3.930204781862798e-06,
+      "loss": 0.8642,
+      "step": 147790
+    },
+    {
+      "epoch": 1.0698748434638465,
+      "grad_norm": 0.14637254178524017,
+      "learning_rate": 3.930132395202213e-06,
+      "loss": 0.862,
+      "step": 147800
+    },
+    {
+      "epoch": 1.0699472301244326,
+      "grad_norm": 0.14838822185993195,
+      "learning_rate": 3.930060008541626e-06,
+      "loss": 0.8473,
+      "step": 147810
+    },
+    {
+      "epoch": 1.070019616785019,
+      "grad_norm": 0.15595309436321259,
+      "learning_rate": 3.92998762188104e-06,
+      "loss": 0.8575,
+      "step": 147820
+    },
+    {
+      "epoch": 1.070092003445605,
+      "grad_norm": 0.15181201696395874,
+      "learning_rate": 3.929915235220454e-06,
+      "loss": 0.8763,
+      "step": 147830
+    },
+    {
+      "epoch": 1.0701643901061912,
+      "grad_norm": 0.14543366432189941,
+      "learning_rate": 3.929842848559868e-06,
+      "loss": 0.8537,
+      "step": 147840
+    },
+    {
+      "epoch": 1.0702367767667775,
+      "grad_norm": 0.16064667701721191,
+      "learning_rate": 3.929770461899282e-06,
+      "loss": 0.8557,
+      "step": 147850
+    },
+    {
+      "epoch": 1.0703091634273636,
+      "grad_norm": 0.1469428986310959,
+      "learning_rate": 3.929698075238695e-06,
+      "loss": 0.8442,
+      "step": 147860
+    },
+    {
+      "epoch": 1.0703815500879499,
+      "grad_norm": 0.16561655700206757,
+      "learning_rate": 3.929625688578109e-06,
+      "loss": 0.849,
+      "step": 147870
+    },
+    {
+      "epoch": 1.070453936748536,
+      "grad_norm": 0.14810919761657715,
+      "learning_rate": 3.929553301917523e-06,
+      "loss": 0.8645,
+      "step": 147880
+    },
+    {
+      "epoch": 1.0705263234091222,
+      "grad_norm": 0.15601563453674316,
+      "learning_rate": 3.929480915256937e-06,
+      "loss": 0.8638,
+      "step": 147890
+    },
+    {
+      "epoch": 1.0705987100697083,
+      "grad_norm": 0.15862402319908142,
+      "learning_rate": 3.929408528596351e-06,
+      "loss": 0.8574,
+      "step": 147900
+    },
+    {
+      "epoch": 1.0706710967302946,
+      "grad_norm": 0.1481819897890091,
+      "learning_rate": 3.929336141935764e-06,
+      "loss": 0.8584,
+      "step": 147910
+    },
+    {
+      "epoch": 1.0707434833908807,
+      "grad_norm": 0.17046231031417847,
+      "learning_rate": 3.929263755275179e-06,
+      "loss": 0.8599,
+      "step": 147920
+    },
+    {
+      "epoch": 1.070815870051467,
+      "grad_norm": 0.155575230717659,
+      "learning_rate": 3.929191368614592e-06,
+      "loss": 0.8527,
+      "step": 147930
+    },
+    {
+      "epoch": 1.070888256712053,
+      "grad_norm": 0.1649925410747528,
+      "learning_rate": 3.929118981954006e-06,
+      "loss": 0.8621,
+      "step": 147940
+    },
+    {
+      "epoch": 1.0709606433726393,
+      "grad_norm": 0.14515741169452667,
+      "learning_rate": 3.9290465952934195e-06,
+      "loss": 0.8572,
+      "step": 147950
+    },
+    {
+      "epoch": 1.0710330300332256,
+      "grad_norm": 0.1484888792037964,
+      "learning_rate": 3.928974208632834e-06,
+      "loss": 0.8535,
+      "step": 147960
+    },
+    {
+      "epoch": 1.0711054166938117,
+      "grad_norm": 0.14969849586486816,
+      "learning_rate": 3.928901821972248e-06,
+      "loss": 0.8612,
+      "step": 147970
+    },
+    {
+      "epoch": 1.071177803354398,
+      "grad_norm": 0.1582787185907364,
+      "learning_rate": 3.928829435311661e-06,
+      "loss": 0.862,
+      "step": 147980
+    },
+    {
+      "epoch": 1.071250190014984,
+      "grad_norm": 0.14706555008888245,
+      "learning_rate": 3.928757048651075e-06,
+      "loss": 0.8579,
+      "step": 147990
+    },
+    {
+      "epoch": 1.0713225766755703,
+      "grad_norm": 0.1515503078699112,
+      "learning_rate": 3.928684661990489e-06,
+      "loss": 0.8742,
+      "step": 148000
+    },
+    {
+      "epoch": 1.0713949633361564,
+      "grad_norm": 0.16300860047340393,
+      "learning_rate": 3.928612275329903e-06,
+      "loss": 0.8495,
+      "step": 148010
+    },
+    {
+      "epoch": 1.0714673499967426,
+      "grad_norm": 0.15342256426811218,
+      "learning_rate": 3.9285398886693166e-06,
+      "loss": 0.864,
+      "step": 148020
+    },
+    {
+      "epoch": 1.0715397366573287,
+      "grad_norm": 0.1569945365190506,
+      "learning_rate": 3.92846750200873e-06,
+      "loss": 0.8606,
+      "step": 148030
+    },
+    {
+      "epoch": 1.071612123317915,
+      "grad_norm": 0.15621396899223328,
+      "learning_rate": 3.928395115348145e-06,
+      "loss": 0.8419,
+      "step": 148040
+    },
+    {
+      "epoch": 1.071684509978501,
+      "grad_norm": 0.14876516163349152,
+      "learning_rate": 3.928322728687558e-06,
+      "loss": 0.8576,
+      "step": 148050
+    },
+    {
+      "epoch": 1.0717568966390874,
+      "grad_norm": 0.15889860689640045,
+      "learning_rate": 3.928250342026972e-06,
+      "loss": 0.8586,
+      "step": 148060
+    },
+    {
+      "epoch": 1.0718292832996736,
+      "grad_norm": 0.15577679872512817,
+      "learning_rate": 3.9281779553663855e-06,
+      "loss": 0.8577,
+      "step": 148070
+    },
+    {
+      "epoch": 1.0719016699602597,
+      "grad_norm": 0.1523328423500061,
+      "learning_rate": 3.928105568705799e-06,
+      "loss": 0.8556,
+      "step": 148080
+    },
+    {
+      "epoch": 1.071974056620846,
+      "grad_norm": 0.15466256439685822,
+      "learning_rate": 3.928033182045213e-06,
+      "loss": 0.8548,
+      "step": 148090
+    },
+    {
+      "epoch": 1.072046443281432,
+      "grad_norm": 0.14170676469802856,
+      "learning_rate": 3.927960795384626e-06,
+      "loss": 0.8438,
+      "step": 148100
+    },
+    {
+      "epoch": 1.0721188299420183,
+      "grad_norm": 0.1515391618013382,
+      "learning_rate": 3.927888408724041e-06,
+      "loss": 0.8612,
+      "step": 148110
+    },
+    {
+      "epoch": 1.0721912166026044,
+      "grad_norm": 0.17052294313907623,
+      "learning_rate": 3.927816022063454e-06,
+      "loss": 0.846,
+      "step": 148120
+    },
+    {
+      "epoch": 1.0722636032631907,
+      "grad_norm": 0.16018043458461761,
+      "learning_rate": 3.927743635402868e-06,
+      "loss": 0.8622,
+      "step": 148130
+    },
+    {
+      "epoch": 1.0723359899237768,
+      "grad_norm": 0.15600261092185974,
+      "learning_rate": 3.927671248742282e-06,
+      "loss": 0.8633,
+      "step": 148140
+    },
+    {
+      "epoch": 1.072408376584363,
+      "grad_norm": 0.15749092400074005,
+      "learning_rate": 3.927598862081696e-06,
+      "loss": 0.843,
+      "step": 148150
+    },
+    {
+      "epoch": 1.0724807632449491,
+      "grad_norm": 0.15799428522586823,
+      "learning_rate": 3.92752647542111e-06,
+      "loss": 0.8613,
+      "step": 148160
+    },
+    {
+      "epoch": 1.0725531499055354,
+      "grad_norm": 0.16896884143352509,
+      "learning_rate": 3.927454088760523e-06,
+      "loss": 0.8452,
+      "step": 148170
+    },
+    {
+      "epoch": 1.0726255365661217,
+      "grad_norm": 0.1738024652004242,
+      "learning_rate": 3.927381702099937e-06,
+      "loss": 0.8582,
+      "step": 148180
+    },
+    {
+      "epoch": 1.0726979232267078,
+      "grad_norm": 0.1570395976305008,
+      "learning_rate": 3.927309315439351e-06,
+      "loss": 0.8687,
+      "step": 148190
+    },
+    {
+      "epoch": 1.072770309887294,
+      "grad_norm": 0.14952725172042847,
+      "learning_rate": 3.927236928778765e-06,
+      "loss": 0.8601,
+      "step": 148200
+    },
+    {
+      "epoch": 1.0728426965478801,
+      "grad_norm": 0.14488628506660461,
+      "learning_rate": 3.927164542118179e-06,
+      "loss": 0.847,
+      "step": 148210
+    },
+    {
+      "epoch": 1.0729150832084664,
+      "grad_norm": 0.16457195580005646,
+      "learning_rate": 3.927092155457592e-06,
+      "loss": 0.862,
+      "step": 148220
+    },
+    {
+      "epoch": 1.0729874698690525,
+      "grad_norm": 0.15635018050670624,
+      "learning_rate": 3.927019768797006e-06,
+      "loss": 0.85,
+      "step": 148230
+    },
+    {
+      "epoch": 1.0730598565296388,
+      "grad_norm": 0.15710517764091492,
+      "learning_rate": 3.92694738213642e-06,
+      "loss": 0.85,
+      "step": 148240
+    },
+    {
+      "epoch": 1.0731322431902248,
+      "grad_norm": 0.14973808825016022,
+      "learning_rate": 3.926874995475834e-06,
+      "loss": 0.8561,
+      "step": 148250
+    },
+    {
+      "epoch": 1.073204629850811,
+      "grad_norm": 0.18000011146068573,
+      "learning_rate": 3.926802608815248e-06,
+      "loss": 0.8432,
+      "step": 148260
+    },
+    {
+      "epoch": 1.0732770165113972,
+      "grad_norm": 0.14735475182533264,
+      "learning_rate": 3.926730222154661e-06,
+      "loss": 0.8719,
+      "step": 148270
+    },
+    {
+      "epoch": 1.0733494031719835,
+      "grad_norm": 0.1539383828639984,
+      "learning_rate": 3.926657835494076e-06,
+      "loss": 0.8637,
+      "step": 148280
+    },
+    {
+      "epoch": 1.0734217898325698,
+      "grad_norm": 0.26289451122283936,
+      "learning_rate": 3.926585448833489e-06,
+      "loss": 0.8712,
+      "step": 148290
+    },
+    {
+      "epoch": 1.0734941764931558,
+      "grad_norm": 0.16455796360969543,
+      "learning_rate": 3.926513062172903e-06,
+      "loss": 0.8612,
+      "step": 148300
+    },
+    {
+      "epoch": 1.073566563153742,
+      "grad_norm": 0.15031065046787262,
+      "learning_rate": 3.9264406755123165e-06,
+      "loss": 0.8556,
+      "step": 148310
+    },
+    {
+      "epoch": 1.0736389498143282,
+      "grad_norm": 0.14603126049041748,
+      "learning_rate": 3.926368288851731e-06,
+      "loss": 0.8603,
+      "step": 148320
+    },
+    {
+      "epoch": 1.0737113364749145,
+      "grad_norm": 0.1502143293619156,
+      "learning_rate": 3.926295902191145e-06,
+      "loss": 0.8621,
+      "step": 148330
+    },
+    {
+      "epoch": 1.0737837231355005,
+      "grad_norm": 0.15308043360710144,
+      "learning_rate": 3.926223515530558e-06,
+      "loss": 0.8633,
+      "step": 148340
+    },
+    {
+      "epoch": 1.0738561097960868,
+      "grad_norm": 0.15625415742397308,
+      "learning_rate": 3.926151128869972e-06,
+      "loss": 0.8654,
+      "step": 148350
+    },
+    {
+      "epoch": 1.0739284964566729,
+      "grad_norm": 0.16124477982521057,
+      "learning_rate": 3.926078742209386e-06,
+      "loss": 0.8583,
+      "step": 148360
+    },
+    {
+      "epoch": 1.0740008831172592,
+      "grad_norm": 0.14821426570415497,
+      "learning_rate": 3.9260063555488e-06,
+      "loss": 0.8497,
+      "step": 148370
+    },
+    {
+      "epoch": 1.0740732697778452,
+      "grad_norm": 0.14356663823127747,
+      "learning_rate": 3.9259339688882135e-06,
+      "loss": 0.8511,
+      "step": 148380
+    },
+    {
+      "epoch": 1.0741456564384315,
+      "grad_norm": 0.14811255037784576,
+      "learning_rate": 3.925861582227627e-06,
+      "loss": 0.8453,
+      "step": 148390
+    },
+    {
+      "epoch": 1.0742180430990178,
+      "grad_norm": 0.18157772719860077,
+      "learning_rate": 3.925789195567042e-06,
+      "loss": 0.8601,
+      "step": 148400
+    },
+    {
+      "epoch": 1.0742904297596039,
+      "grad_norm": 0.16508889198303223,
+      "learning_rate": 3.925716808906455e-06,
+      "loss": 0.8523,
+      "step": 148410
+    },
+    {
+      "epoch": 1.0743628164201902,
+      "grad_norm": 0.15497782826423645,
+      "learning_rate": 3.925644422245869e-06,
+      "loss": 0.8588,
+      "step": 148420
+    },
+    {
+      "epoch": 1.0744352030807762,
+      "grad_norm": 0.1482570469379425,
+      "learning_rate": 3.9255720355852824e-06,
+      "loss": 0.8657,
+      "step": 148430
+    },
+    {
+      "epoch": 1.0745075897413625,
+      "grad_norm": 0.15788806974887848,
+      "learning_rate": 3.925499648924697e-06,
+      "loss": 0.8588,
+      "step": 148440
+    },
+    {
+      "epoch": 1.0745799764019486,
+      "grad_norm": 0.1528482884168625,
+      "learning_rate": 3.9254272622641105e-06,
+      "loss": 0.8502,
+      "step": 148450
+    },
+    {
+      "epoch": 1.0746523630625349,
+      "grad_norm": 0.14317600429058075,
+      "learning_rate": 3.925354875603524e-06,
+      "loss": 0.8432,
+      "step": 148460
+    },
+    {
+      "epoch": 1.074724749723121,
+      "grad_norm": 0.15992312133312225,
+      "learning_rate": 3.925282488942938e-06,
+      "loss": 0.8555,
+      "step": 148470
+    },
+    {
+      "epoch": 1.0747971363837072,
+      "grad_norm": 0.16133040189743042,
+      "learning_rate": 3.925210102282352e-06,
+      "loss": 0.8568,
+      "step": 148480
+    },
+    {
+      "epoch": 1.0748695230442933,
+      "grad_norm": 0.1480935513973236,
+      "learning_rate": 3.925137715621766e-06,
+      "loss": 0.8637,
+      "step": 148490
+    },
+    {
+      "epoch": 1.0749419097048796,
+      "grad_norm": 0.16463685035705566,
+      "learning_rate": 3.9250653289611795e-06,
+      "loss": 0.8598,
+      "step": 148500
+    },
+    {
+      "epoch": 1.0750142963654659,
+      "grad_norm": 0.1424337923526764,
+      "learning_rate": 3.924992942300593e-06,
+      "loss": 0.855,
+      "step": 148510
+    },
+    {
+      "epoch": 1.075086683026052,
+      "grad_norm": 0.25730541348457336,
+      "learning_rate": 3.9249205556400075e-06,
+      "loss": 0.8715,
+      "step": 148520
+    },
+    {
+      "epoch": 1.0751590696866382,
+      "grad_norm": 0.15047651529312134,
+      "learning_rate": 3.924848168979421e-06,
+      "loss": 0.8611,
+      "step": 148530
+    },
+    {
+      "epoch": 1.0752314563472243,
+      "grad_norm": 0.15243446826934814,
+      "learning_rate": 3.924775782318835e-06,
+      "loss": 0.8515,
+      "step": 148540
+    },
+    {
+      "epoch": 1.0753038430078106,
+      "grad_norm": 0.14074011147022247,
+      "learning_rate": 3.924703395658248e-06,
+      "loss": 0.8418,
+      "step": 148550
+    },
+    {
+      "epoch": 1.0753762296683966,
+      "grad_norm": 0.14536842703819275,
+      "learning_rate": 3.924631008997663e-06,
+      "loss": 0.8604,
+      "step": 148560
+    },
+    {
+      "epoch": 1.075448616328983,
+      "grad_norm": 0.1618451476097107,
+      "learning_rate": 3.9245586223370765e-06,
+      "loss": 0.8389,
+      "step": 148570
+    },
+    {
+      "epoch": 1.075521002989569,
+      "grad_norm": 0.1468305140733719,
+      "learning_rate": 3.92448623567649e-06,
+      "loss": 0.8525,
+      "step": 148580
+    },
+    {
+      "epoch": 1.0755933896501553,
+      "grad_norm": 0.22342850267887115,
+      "learning_rate": 3.924413849015904e-06,
+      "loss": 0.8623,
+      "step": 148590
+    },
+    {
+      "epoch": 1.0756657763107413,
+      "grad_norm": 0.1797487586736679,
+      "learning_rate": 3.924341462355317e-06,
+      "loss": 0.8519,
+      "step": 148600
+    },
+    {
+      "epoch": 1.0757381629713276,
+      "grad_norm": 0.16016285121440887,
+      "learning_rate": 3.924269075694731e-06,
+      "loss": 0.8712,
+      "step": 148610
+    },
+    {
+      "epoch": 1.075810549631914,
+      "grad_norm": 0.16447670757770538,
+      "learning_rate": 3.9241966890341445e-06,
+      "loss": 0.8615,
+      "step": 148620
+    },
+    {
+      "epoch": 1.0758829362925,
+      "grad_norm": 0.18734236061573029,
+      "learning_rate": 3.924124302373559e-06,
+      "loss": 0.8522,
+      "step": 148630
+    },
+    {
+      "epoch": 1.0759553229530863,
+      "grad_norm": 0.14720183610916138,
+      "learning_rate": 3.924051915712973e-06,
+      "loss": 0.8525,
+      "step": 148640
+    },
+    {
+      "epoch": 1.0760277096136723,
+      "grad_norm": 0.1844429075717926,
+      "learning_rate": 3.923979529052386e-06,
+      "loss": 0.8465,
+      "step": 148650
+    },
+    {
+      "epoch": 1.0761000962742586,
+      "grad_norm": 0.15892651677131653,
+      "learning_rate": 3.9239071423918e-06,
+      "loss": 0.8523,
+      "step": 148660
+    },
+    {
+      "epoch": 1.0761724829348447,
+      "grad_norm": 0.15322613716125488,
+      "learning_rate": 3.923834755731214e-06,
+      "loss": 0.8584,
+      "step": 148670
+    },
+    {
+      "epoch": 1.076244869595431,
+      "grad_norm": 0.1480284184217453,
+      "learning_rate": 3.923762369070628e-06,
+      "loss": 0.8334,
+      "step": 148680
+    },
+    {
+      "epoch": 1.076317256256017,
+      "grad_norm": 0.16445012390613556,
+      "learning_rate": 3.9236899824100415e-06,
+      "loss": 0.8602,
+      "step": 148690
+    },
+    {
+      "epoch": 1.0763896429166033,
+      "grad_norm": 0.15210187435150146,
+      "learning_rate": 3.923617595749455e-06,
+      "loss": 0.8627,
+      "step": 148700
+    },
+    {
+      "epoch": 1.0764620295771896,
+      "grad_norm": 0.1529064029455185,
+      "learning_rate": 3.92354520908887e-06,
+      "loss": 0.8646,
+      "step": 148710
+    },
+    {
+      "epoch": 1.0765344162377757,
+      "grad_norm": 0.15188395977020264,
+      "learning_rate": 3.923472822428283e-06,
+      "loss": 0.8636,
+      "step": 148720
+    },
+    {
+      "epoch": 1.076606802898362,
+      "grad_norm": 0.14736254513263702,
+      "learning_rate": 3.923400435767697e-06,
+      "loss": 0.8703,
+      "step": 148730
+    },
+    {
+      "epoch": 1.076679189558948,
+      "grad_norm": 0.14823542535305023,
+      "learning_rate": 3.9233280491071105e-06,
+      "loss": 0.863,
+      "step": 148740
+    },
+    {
+      "epoch": 1.0767515762195343,
+      "grad_norm": 0.15504339337348938,
+      "learning_rate": 3.923255662446525e-06,
+      "loss": 0.845,
+      "step": 148750
+    },
+    {
+      "epoch": 1.0768239628801204,
+      "grad_norm": 0.1641693115234375,
+      "learning_rate": 3.9231832757859386e-06,
+      "loss": 0.8574,
+      "step": 148760
+    },
+    {
+      "epoch": 1.0768963495407067,
+      "grad_norm": 0.1393650621175766,
+      "learning_rate": 3.923110889125352e-06,
+      "loss": 0.8528,
+      "step": 148770
+    },
+    {
+      "epoch": 1.0769687362012927,
+      "grad_norm": 0.7887364625930786,
+      "learning_rate": 3.923038502464766e-06,
+      "loss": 0.8631,
+      "step": 148780
+    },
+    {
+      "epoch": 1.077041122861879,
+      "grad_norm": 0.14820659160614014,
+      "learning_rate": 3.92296611580418e-06,
+      "loss": 0.8757,
+      "step": 148790
+    },
+    {
+      "epoch": 1.0771135095224653,
+      "grad_norm": 0.1467932164669037,
+      "learning_rate": 3.922893729143594e-06,
+      "loss": 0.8715,
+      "step": 148800
+    },
+    {
+      "epoch": 1.0771858961830514,
+      "grad_norm": 0.1651478409767151,
+      "learning_rate": 3.9228213424830075e-06,
+      "loss": 0.8704,
+      "step": 148810
+    },
+    {
+      "epoch": 1.0772582828436377,
+      "grad_norm": 0.1475069671869278,
+      "learning_rate": 3.922748955822421e-06,
+      "loss": 0.8651,
+      "step": 148820
+    },
+    {
+      "epoch": 1.0773306695042237,
+      "grad_norm": 0.1452004313468933,
+      "learning_rate": 3.9226765691618356e-06,
+      "loss": 0.8628,
+      "step": 148830
+    },
+    {
+      "epoch": 1.07740305616481,
+      "grad_norm": 0.15325690805912018,
+      "learning_rate": 3.922604182501249e-06,
+      "loss": 0.8556,
+      "step": 148840
+    },
+    {
+      "epoch": 1.077475442825396,
+      "grad_norm": 0.16279549896717072,
+      "learning_rate": 3.922531795840663e-06,
+      "loss": 0.8468,
+      "step": 148850
+    },
+    {
+      "epoch": 1.0775478294859824,
+      "grad_norm": 0.1613786667585373,
+      "learning_rate": 3.922459409180076e-06,
+      "loss": 0.8632,
+      "step": 148860
+    },
+    {
+      "epoch": 1.0776202161465684,
+      "grad_norm": 0.15279968082904816,
+      "learning_rate": 3.92238702251949e-06,
+      "loss": 0.8663,
+      "step": 148870
+    },
+    {
+      "epoch": 1.0776926028071547,
+      "grad_norm": 0.16398665308952332,
+      "learning_rate": 3.9223146358589045e-06,
+      "loss": 0.8583,
+      "step": 148880
+    },
+    {
+      "epoch": 1.0777649894677408,
+      "grad_norm": 0.15407854318618774,
+      "learning_rate": 3.922242249198318e-06,
+      "loss": 0.8703,
+      "step": 148890
+    },
+    {
+      "epoch": 1.077837376128327,
+      "grad_norm": 0.16437429189682007,
+      "learning_rate": 3.922169862537732e-06,
+      "loss": 0.8619,
+      "step": 148900
+    },
+    {
+      "epoch": 1.0779097627889134,
+      "grad_norm": 0.14691048860549927,
+      "learning_rate": 3.922097475877145e-06,
+      "loss": 0.8478,
+      "step": 148910
+    },
+    {
+      "epoch": 1.0779821494494994,
+      "grad_norm": 0.14637692272663116,
+      "learning_rate": 3.92202508921656e-06,
+      "loss": 0.8654,
+      "step": 148920
+    },
+    {
+      "epoch": 1.0780545361100857,
+      "grad_norm": 0.1593993455171585,
+      "learning_rate": 3.921952702555973e-06,
+      "loss": 0.8506,
+      "step": 148930
+    },
+    {
+      "epoch": 1.0781269227706718,
+      "grad_norm": 0.14419616758823395,
+      "learning_rate": 3.921880315895387e-06,
+      "loss": 0.846,
+      "step": 148940
+    },
+    {
+      "epoch": 1.078199309431258,
+      "grad_norm": 0.15830476582050323,
+      "learning_rate": 3.921807929234801e-06,
+      "loss": 0.8575,
+      "step": 148950
+    },
+    {
+      "epoch": 1.0782716960918441,
+      "grad_norm": 0.16840828955173492,
+      "learning_rate": 3.921735542574215e-06,
+      "loss": 0.8483,
+      "step": 148960
+    },
+    {
+      "epoch": 1.0783440827524304,
+      "grad_norm": 0.15455256402492523,
+      "learning_rate": 3.921663155913629e-06,
+      "loss": 0.8433,
+      "step": 148970
+    },
+    {
+      "epoch": 1.0784164694130165,
+      "grad_norm": 0.1700802743434906,
+      "learning_rate": 3.921590769253042e-06,
+      "loss": 0.8655,
+      "step": 148980
+    },
+    {
+      "epoch": 1.0784888560736028,
+      "grad_norm": 0.1804065704345703,
+      "learning_rate": 3.921518382592456e-06,
+      "loss": 0.8648,
+      "step": 148990
+    },
+    {
+      "epoch": 1.0785612427341889,
+      "grad_norm": 0.16987882554531097,
+      "learning_rate": 3.9214459959318704e-06,
+      "loss": 0.8674,
+      "step": 149000
+    },
+    {
+      "epoch": 1.0786336293947751,
+      "grad_norm": 0.14747925102710724,
+      "learning_rate": 3.921373609271284e-06,
+      "loss": 0.858,
+      "step": 149010
+    },
+    {
+      "epoch": 1.0787060160553614,
+      "grad_norm": 0.1391853392124176,
+      "learning_rate": 3.921301222610698e-06,
+      "loss": 0.8458,
+      "step": 149020
+    },
+    {
+      "epoch": 1.0787784027159475,
+      "grad_norm": 0.15131983160972595,
+      "learning_rate": 3.921228835950111e-06,
+      "loss": 0.8572,
+      "step": 149030
+    },
+    {
+      "epoch": 1.0788507893765338,
+      "grad_norm": 0.14283746480941772,
+      "learning_rate": 3.921156449289526e-06,
+      "loss": 0.8434,
+      "step": 149040
+    },
+    {
+      "epoch": 1.0789231760371198,
+      "grad_norm": 0.1681165099143982,
+      "learning_rate": 3.921084062628939e-06,
+      "loss": 0.8539,
+      "step": 149050
+    },
+    {
+      "epoch": 1.0789955626977061,
+      "grad_norm": 0.15962868928909302,
+      "learning_rate": 3.921011675968353e-06,
+      "loss": 0.8488,
+      "step": 149060
+    },
+    {
+      "epoch": 1.0790679493582922,
+      "grad_norm": 0.14654859900474548,
+      "learning_rate": 3.920939289307767e-06,
+      "loss": 0.848,
+      "step": 149070
+    },
+    {
+      "epoch": 1.0791403360188785,
+      "grad_norm": 0.1605050265789032,
+      "learning_rate": 3.920866902647181e-06,
+      "loss": 0.8612,
+      "step": 149080
+    },
+    {
+      "epoch": 1.0792127226794646,
+      "grad_norm": 0.15435899794101715,
+      "learning_rate": 3.920794515986595e-06,
+      "loss": 0.8665,
+      "step": 149090
+    },
+    {
+      "epoch": 1.0792851093400508,
+      "grad_norm": 0.15466824173927307,
+      "learning_rate": 3.920722129326008e-06,
+      "loss": 0.8633,
+      "step": 149100
+    },
+    {
+      "epoch": 1.079357496000637,
+      "grad_norm": 0.1811055988073349,
+      "learning_rate": 3.920649742665422e-06,
+      "loss": 0.8626,
+      "step": 149110
+    },
+    {
+      "epoch": 1.0794298826612232,
+      "grad_norm": 0.15361252427101135,
+      "learning_rate": 3.920577356004836e-06,
+      "loss": 0.8588,
+      "step": 149120
+    },
+    {
+      "epoch": 1.0795022693218095,
+      "grad_norm": 0.15100868046283722,
+      "learning_rate": 3.920504969344249e-06,
+      "loss": 0.8744,
+      "step": 149130
+    },
+    {
+      "epoch": 1.0795746559823955,
+      "grad_norm": 0.16140015423297882,
+      "learning_rate": 3.920432582683663e-06,
+      "loss": 0.8592,
+      "step": 149140
+    },
+    {
+      "epoch": 1.0796470426429818,
+      "grad_norm": 0.1692638099193573,
+      "learning_rate": 3.920360196023077e-06,
+      "loss": 0.8445,
+      "step": 149150
+    },
+    {
+      "epoch": 1.079719429303568,
+      "grad_norm": 0.1461089849472046,
+      "learning_rate": 3.920287809362491e-06,
+      "loss": 0.8389,
+      "step": 149160
+    },
+    {
+      "epoch": 1.0797918159641542,
+      "grad_norm": 0.15460637211799622,
+      "learning_rate": 3.9202154227019044e-06,
+      "loss": 0.8514,
+      "step": 149170
+    },
+    {
+      "epoch": 1.0798642026247403,
+      "grad_norm": 0.15934206545352936,
+      "learning_rate": 3.920143036041318e-06,
+      "loss": 0.8502,
+      "step": 149180
+    },
+    {
+      "epoch": 1.0799365892853265,
+      "grad_norm": 0.17823490500450134,
+      "learning_rate": 3.9200706493807325e-06,
+      "loss": 0.8531,
+      "step": 149190
+    },
+    {
+      "epoch": 1.0800089759459126,
+      "grad_norm": 0.15246717631816864,
+      "learning_rate": 3.919998262720146e-06,
+      "loss": 0.8649,
+      "step": 149200
+    },
+    {
+      "epoch": 1.080081362606499,
+      "grad_norm": 0.1905028522014618,
+      "learning_rate": 3.91992587605956e-06,
+      "loss": 0.8662,
+      "step": 149210
+    },
+    {
+      "epoch": 1.080153749267085,
+      "grad_norm": 0.16161444783210754,
+      "learning_rate": 3.919853489398973e-06,
+      "loss": 0.8635,
+      "step": 149220
+    },
+    {
+      "epoch": 1.0802261359276712,
+      "grad_norm": 0.1495893895626068,
+      "learning_rate": 3.919781102738388e-06,
+      "loss": 0.8637,
+      "step": 149230
+    },
+    {
+      "epoch": 1.0802985225882575,
+      "grad_norm": 0.15014596283435822,
+      "learning_rate": 3.9197087160778015e-06,
+      "loss": 0.8642,
+      "step": 149240
+    },
+    {
+      "epoch": 1.0803709092488436,
+      "grad_norm": 0.13995687663555145,
+      "learning_rate": 3.919636329417215e-06,
+      "loss": 0.8641,
+      "step": 149250
+    },
+    {
+      "epoch": 1.08044329590943,
+      "grad_norm": 0.15207460522651672,
+      "learning_rate": 3.919563942756629e-06,
+      "loss": 0.852,
+      "step": 149260
+    },
+    {
+      "epoch": 1.080515682570016,
+      "grad_norm": 0.15077516436576843,
+      "learning_rate": 3.919491556096043e-06,
+      "loss": 0.8663,
+      "step": 149270
+    },
+    {
+      "epoch": 1.0805880692306022,
+      "grad_norm": 0.15563450753688812,
+      "learning_rate": 3.919419169435457e-06,
+      "loss": 0.8547,
+      "step": 149280
+    },
+    {
+      "epoch": 1.0806604558911883,
+      "grad_norm": 0.1566159725189209,
+      "learning_rate": 3.91934678277487e-06,
+      "loss": 0.8581,
+      "step": 149290
+    },
+    {
+      "epoch": 1.0807328425517746,
+      "grad_norm": 0.1530294120311737,
+      "learning_rate": 3.919274396114284e-06,
+      "loss": 0.8684,
+      "step": 149300
+    },
+    {
+      "epoch": 1.0808052292123607,
+      "grad_norm": 0.14563488960266113,
+      "learning_rate": 3.9192020094536985e-06,
+      "loss": 0.8668,
+      "step": 149310
+    },
+    {
+      "epoch": 1.080877615872947,
+      "grad_norm": 0.36675575375556946,
+      "learning_rate": 3.919129622793112e-06,
+      "loss": 0.8571,
+      "step": 149320
+    },
+    {
+      "epoch": 1.080950002533533,
+      "grad_norm": 0.17297199368476868,
+      "learning_rate": 3.919057236132526e-06,
+      "loss": 0.867,
+      "step": 149330
+    },
+    {
+      "epoch": 1.0810223891941193,
+      "grad_norm": 0.1599206179380417,
+      "learning_rate": 3.918984849471939e-06,
+      "loss": 0.8633,
+      "step": 149340
+    },
+    {
+      "epoch": 1.0810947758547056,
+      "grad_norm": 0.16135400533676147,
+      "learning_rate": 3.918912462811354e-06,
+      "loss": 0.8661,
+      "step": 149350
+    },
+    {
+      "epoch": 1.0811671625152917,
+      "grad_norm": 0.16339783370494843,
+      "learning_rate": 3.918840076150767e-06,
+      "loss": 0.8668,
+      "step": 149360
+    },
+    {
+      "epoch": 1.081239549175878,
+      "grad_norm": 0.14185838401317596,
+      "learning_rate": 3.918767689490181e-06,
+      "loss": 0.851,
+      "step": 149370
+    },
+    {
+      "epoch": 1.081311935836464,
+      "grad_norm": 0.14981046319007874,
+      "learning_rate": 3.918695302829595e-06,
+      "loss": 0.862,
+      "step": 149380
+    },
+    {
+      "epoch": 1.0813843224970503,
+      "grad_norm": 0.14530541002750397,
+      "learning_rate": 3.918622916169009e-06,
+      "loss": 0.8553,
+      "step": 149390
+    },
+    {
+      "epoch": 1.0814567091576364,
+      "grad_norm": 0.14957144856452942,
+      "learning_rate": 3.918550529508423e-06,
+      "loss": 0.8599,
+      "step": 149400
+    },
+    {
+      "epoch": 1.0815290958182227,
+      "grad_norm": 0.14457090198993683,
+      "learning_rate": 3.918478142847836e-06,
+      "loss": 0.8483,
+      "step": 149410
+    },
+    {
+      "epoch": 1.0816014824788087,
+      "grad_norm": 0.1516246795654297,
+      "learning_rate": 3.91840575618725e-06,
+      "loss": 0.8654,
+      "step": 149420
+    },
+    {
+      "epoch": 1.081673869139395,
+      "grad_norm": 0.15999293327331543,
+      "learning_rate": 3.918333369526664e-06,
+      "loss": 0.8624,
+      "step": 149430
+    },
+    {
+      "epoch": 1.081746255799981,
+      "grad_norm": 0.16079509258270264,
+      "learning_rate": 3.918260982866078e-06,
+      "loss": 0.8486,
+      "step": 149440
+    },
+    {
+      "epoch": 1.0818186424605674,
+      "grad_norm": 0.16731907427310944,
+      "learning_rate": 3.918188596205492e-06,
+      "loss": 0.8651,
+      "step": 149450
+    },
+    {
+      "epoch": 1.0818910291211536,
+      "grad_norm": 0.16714203357696533,
+      "learning_rate": 3.918116209544905e-06,
+      "loss": 0.8646,
+      "step": 149460
+    },
+    {
+      "epoch": 1.0819634157817397,
+      "grad_norm": 0.14884908497333527,
+      "learning_rate": 3.918043822884319e-06,
+      "loss": 0.8628,
+      "step": 149470
+    },
+    {
+      "epoch": 1.082035802442326,
+      "grad_norm": 0.1482784003019333,
+      "learning_rate": 3.917971436223733e-06,
+      "loss": 0.86,
+      "step": 149480
+    },
+    {
+      "epoch": 1.082108189102912,
+      "grad_norm": 0.17224667966365814,
+      "learning_rate": 3.917899049563147e-06,
+      "loss": 0.863,
+      "step": 149490
+    },
+    {
+      "epoch": 1.0821805757634984,
+      "grad_norm": 0.15392819046974182,
+      "learning_rate": 3.9178266629025606e-06,
+      "loss": 0.8601,
+      "step": 149500
+    },
+    {
+      "epoch": 1.0822529624240844,
+      "grad_norm": 0.15251944959163666,
+      "learning_rate": 3.917754276241974e-06,
+      "loss": 0.8585,
+      "step": 149510
+    },
+    {
+      "epoch": 1.0823253490846707,
+      "grad_norm": 0.14979909360408783,
+      "learning_rate": 3.917681889581389e-06,
+      "loss": 0.8553,
+      "step": 149520
+    },
+    {
+      "epoch": 1.0823977357452568,
+      "grad_norm": 0.1591353863477707,
+      "learning_rate": 3.917609502920802e-06,
+      "loss": 0.8604,
+      "step": 149530
+    },
+    {
+      "epoch": 1.082470122405843,
+      "grad_norm": 0.1540420949459076,
+      "learning_rate": 3.917537116260216e-06,
+      "loss": 0.8674,
+      "step": 149540
+    },
+    {
+      "epoch": 1.0825425090664291,
+      "grad_norm": 0.43315520882606506,
+      "learning_rate": 3.9174647295996295e-06,
+      "loss": 0.8565,
+      "step": 149550
+    },
+    {
+      "epoch": 1.0826148957270154,
+      "grad_norm": 0.14986060559749603,
+      "learning_rate": 3.917392342939044e-06,
+      "loss": 0.8639,
+      "step": 149560
+    },
+    {
+      "epoch": 1.0826872823876017,
+      "grad_norm": 0.1607617884874344,
+      "learning_rate": 3.9173199562784576e-06,
+      "loss": 0.8578,
+      "step": 149570
+    },
+    {
+      "epoch": 1.0827596690481878,
+      "grad_norm": 0.1520787626504898,
+      "learning_rate": 3.917247569617871e-06,
+      "loss": 0.8506,
+      "step": 149580
+    },
+    {
+      "epoch": 1.082832055708774,
+      "grad_norm": 0.15222980082035065,
+      "learning_rate": 3.917175182957285e-06,
+      "loss": 0.8564,
+      "step": 149590
+    },
+    {
+      "epoch": 1.0829044423693601,
+      "grad_norm": 0.14734259247779846,
+      "learning_rate": 3.917102796296699e-06,
+      "loss": 0.8571,
+      "step": 149600
+    },
+    {
+      "epoch": 1.0829768290299464,
+      "grad_norm": 0.15078797936439514,
+      "learning_rate": 3.917030409636113e-06,
+      "loss": 0.8483,
+      "step": 149610
+    },
+    {
+      "epoch": 1.0830492156905325,
+      "grad_norm": 0.15337948501110077,
+      "learning_rate": 3.9169580229755265e-06,
+      "loss": 0.8463,
+      "step": 149620
+    },
+    {
+      "epoch": 1.0831216023511188,
+      "grad_norm": 0.24916933476924896,
+      "learning_rate": 3.91688563631494e-06,
+      "loss": 0.8624,
+      "step": 149630
+    },
+    {
+      "epoch": 1.0831939890117048,
+      "grad_norm": 0.15374572575092316,
+      "learning_rate": 3.9168132496543546e-06,
+      "loss": 0.8475,
+      "step": 149640
+    },
+    {
+      "epoch": 1.0832663756722911,
+      "grad_norm": 0.17309990525245667,
+      "learning_rate": 3.916740862993768e-06,
+      "loss": 0.8458,
+      "step": 149650
+    },
+    {
+      "epoch": 1.0833387623328772,
+      "grad_norm": 0.16714857518672943,
+      "learning_rate": 3.916668476333181e-06,
+      "loss": 0.8569,
+      "step": 149660
+    },
+    {
+      "epoch": 1.0834111489934635,
+      "grad_norm": 0.15130509436130524,
+      "learning_rate": 3.916596089672595e-06,
+      "loss": 0.8534,
+      "step": 149670
+    },
+    {
+      "epoch": 1.0834835356540498,
+      "grad_norm": 0.1503904163837433,
+      "learning_rate": 3.916523703012009e-06,
+      "loss": 0.8673,
+      "step": 149680
+    },
+    {
+      "epoch": 1.0835559223146358,
+      "grad_norm": 0.15241362154483795,
+      "learning_rate": 3.916451316351423e-06,
+      "loss": 0.8575,
+      "step": 149690
+    },
+    {
+      "epoch": 1.083628308975222,
+      "grad_norm": 0.15771149098873138,
+      "learning_rate": 3.916378929690836e-06,
+      "loss": 0.8571,
+      "step": 149700
+    },
+    {
+      "epoch": 1.0837006956358082,
+      "grad_norm": 0.17620976269245148,
+      "learning_rate": 3.916306543030251e-06,
+      "loss": 0.8588,
+      "step": 149710
+    },
+    {
+      "epoch": 1.0837730822963945,
+      "grad_norm": 0.15407635271549225,
+      "learning_rate": 3.916234156369664e-06,
+      "loss": 0.8585,
+      "step": 149720
+    },
+    {
+      "epoch": 1.0838454689569805,
+      "grad_norm": 0.14716872572898865,
+      "learning_rate": 3.916161769709078e-06,
+      "loss": 0.8473,
+      "step": 149730
+    },
+    {
+      "epoch": 1.0839178556175668,
+      "grad_norm": 0.14605097472667694,
+      "learning_rate": 3.916089383048492e-06,
+      "loss": 0.8604,
+      "step": 149740
+    },
+    {
+      "epoch": 1.0839902422781529,
+      "grad_norm": 0.15472210943698883,
+      "learning_rate": 3.916016996387906e-06,
+      "loss": 0.8726,
+      "step": 149750
+    },
+    {
+      "epoch": 1.0840626289387392,
+      "grad_norm": 0.15628187358379364,
+      "learning_rate": 3.91594460972732e-06,
+      "loss": 0.8521,
+      "step": 149760
+    },
+    {
+      "epoch": 1.0841350155993255,
+      "grad_norm": 0.1554306596517563,
+      "learning_rate": 3.915872223066733e-06,
+      "loss": 0.8629,
+      "step": 149770
+    },
+    {
+      "epoch": 1.0842074022599115,
+      "grad_norm": 0.14699378609657288,
+      "learning_rate": 3.915799836406147e-06,
+      "loss": 0.8479,
+      "step": 149780
+    },
+    {
+      "epoch": 1.0842797889204978,
+      "grad_norm": 0.1461055874824524,
+      "learning_rate": 3.915727449745561e-06,
+      "loss": 0.8519,
+      "step": 149790
+    },
+    {
+      "epoch": 1.0843521755810839,
+      "grad_norm": 0.1457691639661789,
+      "learning_rate": 3.915655063084975e-06,
+      "loss": 0.8723,
+      "step": 149800
+    },
+    {
+      "epoch": 1.0844245622416702,
+      "grad_norm": 0.1460854858160019,
+      "learning_rate": 3.915582676424389e-06,
+      "loss": 0.8567,
+      "step": 149810
+    },
+    {
+      "epoch": 1.0844969489022562,
+      "grad_norm": 0.14653198421001434,
+      "learning_rate": 3.915510289763802e-06,
+      "loss": 0.8545,
+      "step": 149820
+    },
+    {
+      "epoch": 1.0845693355628425,
+      "grad_norm": 0.1542053073644638,
+      "learning_rate": 3.915437903103217e-06,
+      "loss": 0.8549,
+      "step": 149830
+    },
+    {
+      "epoch": 1.0846417222234286,
+      "grad_norm": 0.1533014178276062,
+      "learning_rate": 3.91536551644263e-06,
+      "loss": 0.8684,
+      "step": 149840
+    },
+    {
+      "epoch": 1.0847141088840149,
+      "grad_norm": 0.1554461568593979,
+      "learning_rate": 3.915293129782044e-06,
+      "loss": 0.8568,
+      "step": 149850
+    },
+    {
+      "epoch": 1.084786495544601,
+      "grad_norm": 0.15294988453388214,
+      "learning_rate": 3.9152207431214575e-06,
+      "loss": 0.8692,
+      "step": 149860
+    },
+    {
+      "epoch": 1.0848588822051872,
+      "grad_norm": 0.1588299423456192,
+      "learning_rate": 3.915148356460872e-06,
+      "loss": 0.8545,
+      "step": 149870
+    },
+    {
+      "epoch": 1.0849312688657735,
+      "grad_norm": 0.14999845623970032,
+      "learning_rate": 3.915075969800286e-06,
+      "loss": 0.8722,
+      "step": 149880
+    },
+    {
+      "epoch": 1.0850036555263596,
+      "grad_norm": 0.1531199961900711,
+      "learning_rate": 3.915003583139699e-06,
+      "loss": 0.8564,
+      "step": 149890
+    },
+    {
+      "epoch": 1.0850760421869459,
+      "grad_norm": 0.1617061197757721,
+      "learning_rate": 3.914931196479113e-06,
+      "loss": 0.8435,
+      "step": 149900
+    },
+    {
+      "epoch": 1.085148428847532,
+      "grad_norm": 0.1557346135377884,
+      "learning_rate": 3.914858809818527e-06,
+      "loss": 0.8582,
+      "step": 149910
+    },
+    {
+      "epoch": 1.0852208155081182,
+      "grad_norm": 0.14734655618667603,
+      "learning_rate": 3.914786423157941e-06,
+      "loss": 0.8694,
+      "step": 149920
+    },
+    {
+      "epoch": 1.0852932021687043,
+      "grad_norm": 0.1435697227716446,
+      "learning_rate": 3.9147140364973545e-06,
+      "loss": 0.8429,
+      "step": 149930
+    },
+    {
+      "epoch": 1.0853655888292906,
+      "grad_norm": 0.15137584507465363,
+      "learning_rate": 3.914641649836768e-06,
+      "loss": 0.866,
+      "step": 149940
+    },
+    {
+      "epoch": 1.0854379754898766,
+      "grad_norm": 0.1859600692987442,
+      "learning_rate": 3.914569263176183e-06,
+      "loss": 0.8518,
+      "step": 149950
+    },
+    {
+      "epoch": 1.085510362150463,
+      "grad_norm": 0.15186361968517303,
+      "learning_rate": 3.914496876515596e-06,
+      "loss": 0.8694,
+      "step": 149960
+    },
+    {
+      "epoch": 1.0855827488110492,
+      "grad_norm": 0.1560952514410019,
+      "learning_rate": 3.91442448985501e-06,
+      "loss": 0.8628,
+      "step": 149970
+    },
+    {
+      "epoch": 1.0856551354716353,
+      "grad_norm": 0.14067547023296356,
+      "learning_rate": 3.9143521031944234e-06,
+      "loss": 0.8585,
+      "step": 149980
+    },
+    {
+      "epoch": 1.0857275221322216,
+      "grad_norm": 0.15680783987045288,
+      "learning_rate": 3.914279716533838e-06,
+      "loss": 0.8498,
+      "step": 149990
+    },
+    {
+      "epoch": 1.0857999087928076,
+      "grad_norm": 0.15567612648010254,
+      "learning_rate": 3.9142073298732515e-06,
+      "loss": 0.8562,
+      "step": 150000
+    },
+    {
+      "epoch": 1.085872295453394,
+      "grad_norm": 0.1518632471561432,
+      "learning_rate": 3.914134943212665e-06,
+      "loss": 0.8667,
+      "step": 150010
+    },
+    {
+      "epoch": 1.08594468211398,
+      "grad_norm": 0.15622593462467194,
+      "learning_rate": 3.914062556552079e-06,
+      "loss": 0.8498,
+      "step": 150020
+    },
+    {
+      "epoch": 1.0860170687745663,
+      "grad_norm": 0.15988045930862427,
+      "learning_rate": 3.913990169891493e-06,
+      "loss": 0.8651,
+      "step": 150030
+    },
+    {
+      "epoch": 1.0860894554351523,
+      "grad_norm": 0.15659677982330322,
+      "learning_rate": 3.913917783230907e-06,
+      "loss": 0.8574,
+      "step": 150040
+    },
+    {
+      "epoch": 1.0861618420957386,
+      "grad_norm": 0.15888313949108124,
+      "learning_rate": 3.9138453965703205e-06,
+      "loss": 0.8573,
+      "step": 150050
+    },
+    {
+      "epoch": 1.0862342287563247,
+      "grad_norm": 0.16083279252052307,
+      "learning_rate": 3.913773009909734e-06,
+      "loss": 0.8572,
+      "step": 150060
+    },
+    {
+      "epoch": 1.086306615416911,
+      "grad_norm": 0.14704735577106476,
+      "learning_rate": 3.9137006232491485e-06,
+      "loss": 0.8524,
+      "step": 150070
+    },
+    {
+      "epoch": 1.0863790020774973,
+      "grad_norm": 0.15133285522460938,
+      "learning_rate": 3.913628236588562e-06,
+      "loss": 0.8759,
+      "step": 150080
+    },
+    {
+      "epoch": 1.0864513887380833,
+      "grad_norm": 0.14864712953567505,
+      "learning_rate": 3.913555849927976e-06,
+      "loss": 0.8561,
+      "step": 150090
+    },
+    {
+      "epoch": 1.0865237753986696,
+      "grad_norm": 0.14330703020095825,
+      "learning_rate": 3.913483463267389e-06,
+      "loss": 0.8613,
+      "step": 150100
+    },
+    {
+      "epoch": 1.0865961620592557,
+      "grad_norm": 0.1465221792459488,
+      "learning_rate": 3.913411076606803e-06,
+      "loss": 0.8581,
+      "step": 150110
+    },
+    {
+      "epoch": 1.086668548719842,
+      "grad_norm": 0.14701513946056366,
+      "learning_rate": 3.9133386899462175e-06,
+      "loss": 0.8542,
+      "step": 150120
+    },
+    {
+      "epoch": 1.086740935380428,
+      "grad_norm": 0.15257854759693146,
+      "learning_rate": 3.913266303285631e-06,
+      "loss": 0.868,
+      "step": 150130
+    },
+    {
+      "epoch": 1.0868133220410143,
+      "grad_norm": 0.15649712085723877,
+      "learning_rate": 3.913193916625045e-06,
+      "loss": 0.8602,
+      "step": 150140
+    },
+    {
+      "epoch": 1.0868857087016004,
+      "grad_norm": 0.14844316244125366,
+      "learning_rate": 3.913121529964458e-06,
+      "loss": 0.8488,
+      "step": 150150
+    },
+    {
+      "epoch": 1.0869580953621867,
+      "grad_norm": 0.14990560710430145,
+      "learning_rate": 3.913049143303873e-06,
+      "loss": 0.8656,
+      "step": 150160
+    },
+    {
+      "epoch": 1.0870304820227727,
+      "grad_norm": 0.17032568156719208,
+      "learning_rate": 3.912976756643286e-06,
+      "loss": 0.8592,
+      "step": 150170
+    },
+    {
+      "epoch": 1.087102868683359,
+      "grad_norm": 0.16417841613292694,
+      "learning_rate": 3.9129043699827e-06,
+      "loss": 0.8502,
+      "step": 150180
+    },
+    {
+      "epoch": 1.0871752553439453,
+      "grad_norm": 0.16040199995040894,
+      "learning_rate": 3.912831983322114e-06,
+      "loss": 0.8603,
+      "step": 150190
+    },
+    {
+      "epoch": 1.0872476420045314,
+      "grad_norm": 0.15688760578632355,
+      "learning_rate": 3.912759596661527e-06,
+      "loss": 0.8449,
+      "step": 150200
+    },
+    {
+      "epoch": 1.0873200286651177,
+      "grad_norm": 0.14727970957756042,
+      "learning_rate": 3.912687210000941e-06,
+      "loss": 0.8688,
+      "step": 150210
+    },
+    {
+      "epoch": 1.0873924153257037,
+      "grad_norm": 0.1582842767238617,
+      "learning_rate": 3.912614823340355e-06,
+      "loss": 0.8636,
+      "step": 150220
+    },
+    {
+      "epoch": 1.08746480198629,
+      "grad_norm": 0.15592309832572937,
+      "learning_rate": 3.912542436679769e-06,
+      "loss": 0.8656,
+      "step": 150230
+    },
+    {
+      "epoch": 1.087537188646876,
+      "grad_norm": 0.1545068770647049,
+      "learning_rate": 3.9124700500191826e-06,
+      "loss": 0.8603,
+      "step": 150240
+    },
+    {
+      "epoch": 1.0876095753074624,
+      "grad_norm": 0.15706755220890045,
+      "learning_rate": 3.912397663358596e-06,
+      "loss": 0.8549,
+      "step": 150250
+    },
+    {
+      "epoch": 1.0876819619680484,
+      "grad_norm": 0.15659025311470032,
+      "learning_rate": 3.912325276698011e-06,
+      "loss": 0.8617,
+      "step": 150260
+    },
+    {
+      "epoch": 1.0877543486286347,
+      "grad_norm": 0.18718166649341583,
+      "learning_rate": 3.912252890037424e-06,
+      "loss": 0.8653,
+      "step": 150270
+    },
+    {
+      "epoch": 1.0878267352892208,
+      "grad_norm": 0.15322886407375336,
+      "learning_rate": 3.912180503376838e-06,
+      "loss": 0.8603,
+      "step": 150280
+    },
+    {
+      "epoch": 1.087899121949807,
+      "grad_norm": 0.16585904359817505,
+      "learning_rate": 3.9121081167162515e-06,
+      "loss": 0.8705,
+      "step": 150290
+    },
+    {
+      "epoch": 1.0879715086103934,
+      "grad_norm": 0.13771013915538788,
+      "learning_rate": 3.912035730055665e-06,
+      "loss": 0.8482,
+      "step": 150300
+    },
+    {
+      "epoch": 1.0880438952709794,
+      "grad_norm": 0.16550597548484802,
+      "learning_rate": 3.9119633433950796e-06,
+      "loss": 0.8535,
+      "step": 150310
+    },
+    {
+      "epoch": 1.0881162819315657,
+      "grad_norm": 0.1450309306383133,
+      "learning_rate": 3.911890956734493e-06,
+      "loss": 0.8588,
+      "step": 150320
+    },
+    {
+      "epoch": 1.0881886685921518,
+      "grad_norm": 0.15524572134017944,
+      "learning_rate": 3.911818570073907e-06,
+      "loss": 0.8563,
+      "step": 150330
+    },
+    {
+      "epoch": 1.088261055252738,
+      "grad_norm": 0.1510109007358551,
+      "learning_rate": 3.91174618341332e-06,
+      "loss": 0.8731,
+      "step": 150340
+    },
+    {
+      "epoch": 1.0883334419133242,
+      "grad_norm": 0.15307964384555817,
+      "learning_rate": 3.911673796752735e-06,
+      "loss": 0.864,
+      "step": 150350
+    },
+    {
+      "epoch": 1.0884058285739104,
+      "grad_norm": 0.15317986905574799,
+      "learning_rate": 3.9116014100921485e-06,
+      "loss": 0.8565,
+      "step": 150360
+    },
+    {
+      "epoch": 1.0884782152344965,
+      "grad_norm": 0.15566273033618927,
+      "learning_rate": 3.911529023431562e-06,
+      "loss": 0.8512,
+      "step": 150370
+    },
+    {
+      "epoch": 1.0885506018950828,
+      "grad_norm": 0.14965330064296722,
+      "learning_rate": 3.911456636770976e-06,
+      "loss": 0.8651,
+      "step": 150380
+    },
+    {
+      "epoch": 1.0886229885556689,
+      "grad_norm": 0.1521342396736145,
+      "learning_rate": 3.91138425011039e-06,
+      "loss": 0.8568,
+      "step": 150390
+    },
+    {
+      "epoch": 1.0886953752162551,
+      "grad_norm": 0.1588059663772583,
+      "learning_rate": 3.911311863449804e-06,
+      "loss": 0.8595,
+      "step": 150400
+    },
+    {
+      "epoch": 1.0887677618768414,
+      "grad_norm": 0.145501509308815,
+      "learning_rate": 3.911239476789217e-06,
+      "loss": 0.8497,
+      "step": 150410
+    },
+    {
+      "epoch": 1.0888401485374275,
+      "grad_norm": 0.14711380004882812,
+      "learning_rate": 3.911167090128631e-06,
+      "loss": 0.849,
+      "step": 150420
+    },
+    {
+      "epoch": 1.0889125351980138,
+      "grad_norm": 0.1537601202726364,
+      "learning_rate": 3.9110947034680455e-06,
+      "loss": 0.8739,
+      "step": 150430
+    },
+    {
+      "epoch": 1.0889849218585999,
+      "grad_norm": 0.1484655886888504,
+      "learning_rate": 3.911022316807459e-06,
+      "loss": 0.8626,
+      "step": 150440
+    },
+    {
+      "epoch": 1.0890573085191861,
+      "grad_norm": 0.14920607209205627,
+      "learning_rate": 3.910949930146873e-06,
+      "loss": 0.8466,
+      "step": 150450
+    },
+    {
+      "epoch": 1.0891296951797722,
+      "grad_norm": 0.16342855989933014,
+      "learning_rate": 3.910877543486286e-06,
+      "loss": 0.858,
+      "step": 150460
+    },
+    {
+      "epoch": 1.0892020818403585,
+      "grad_norm": 0.15427295863628387,
+      "learning_rate": 3.910805156825701e-06,
+      "loss": 0.859,
+      "step": 150470
+    },
+    {
+      "epoch": 1.0892744685009446,
+      "grad_norm": 0.15797512233257294,
+      "learning_rate": 3.9107327701651144e-06,
+      "loss": 0.8466,
+      "step": 150480
+    },
+    {
+      "epoch": 1.0893468551615308,
+      "grad_norm": 0.15120099484920502,
+      "learning_rate": 3.910660383504528e-06,
+      "loss": 0.8429,
+      "step": 150490
+    },
+    {
+      "epoch": 1.089419241822117,
+      "grad_norm": 0.16131120920181274,
+      "learning_rate": 3.910587996843942e-06,
+      "loss": 0.8562,
+      "step": 150500
+    },
+    {
+      "epoch": 1.0894916284827032,
+      "grad_norm": 0.15284837782382965,
+      "learning_rate": 3.910515610183356e-06,
+      "loss": 0.8584,
+      "step": 150510
+    },
+    {
+      "epoch": 1.0895640151432895,
+      "grad_norm": 0.1720409095287323,
+      "learning_rate": 3.91044322352277e-06,
+      "loss": 0.8637,
+      "step": 150520
+    },
+    {
+      "epoch": 1.0896364018038756,
+      "grad_norm": 0.16043171286582947,
+      "learning_rate": 3.910370836862183e-06,
+      "loss": 0.8538,
+      "step": 150530
+    },
+    {
+      "epoch": 1.0897087884644618,
+      "grad_norm": 0.16862253844738007,
+      "learning_rate": 3.910298450201597e-06,
+      "loss": 0.8509,
+      "step": 150540
+    },
+    {
+      "epoch": 1.089781175125048,
+      "grad_norm": 0.1581002175807953,
+      "learning_rate": 3.9102260635410114e-06,
+      "loss": 0.851,
+      "step": 150550
+    },
+    {
+      "epoch": 1.0898535617856342,
+      "grad_norm": 0.15931819379329681,
+      "learning_rate": 3.910153676880425e-06,
+      "loss": 0.8528,
+      "step": 150560
+    },
+    {
+      "epoch": 1.0899259484462203,
+      "grad_norm": 0.1453382819890976,
+      "learning_rate": 3.910081290219839e-06,
+      "loss": 0.862,
+      "step": 150570
+    },
+    {
+      "epoch": 1.0899983351068065,
+      "grad_norm": 0.14687404036521912,
+      "learning_rate": 3.910008903559252e-06,
+      "loss": 0.8474,
+      "step": 150580
+    },
+    {
+      "epoch": 1.0900707217673926,
+      "grad_norm": 0.17053967714309692,
+      "learning_rate": 3.909936516898667e-06,
+      "loss": 0.8553,
+      "step": 150590
+    },
+    {
+      "epoch": 1.090143108427979,
+      "grad_norm": 0.1455235779285431,
+      "learning_rate": 3.90986413023808e-06,
+      "loss": 0.8503,
+      "step": 150600
+    },
+    {
+      "epoch": 1.090215495088565,
+      "grad_norm": 0.15172958374023438,
+      "learning_rate": 3.909791743577494e-06,
+      "loss": 0.8488,
+      "step": 150610
+    },
+    {
+      "epoch": 1.0902878817491513,
+      "grad_norm": 0.15800823271274567,
+      "learning_rate": 3.909719356916908e-06,
+      "loss": 0.8566,
+      "step": 150620
+    },
+    {
+      "epoch": 1.0903602684097375,
+      "grad_norm": 0.13901235163211823,
+      "learning_rate": 3.909646970256322e-06,
+      "loss": 0.8373,
+      "step": 150630
+    },
+    {
+      "epoch": 1.0904326550703236,
+      "grad_norm": 0.1460275948047638,
+      "learning_rate": 3.909574583595736e-06,
+      "loss": 0.8577,
+      "step": 150640
+    },
+    {
+      "epoch": 1.09050504173091,
+      "grad_norm": 0.1747177392244339,
+      "learning_rate": 3.909502196935149e-06,
+      "loss": 0.8456,
+      "step": 150650
+    },
+    {
+      "epoch": 1.090577428391496,
+      "grad_norm": 0.1587553173303604,
+      "learning_rate": 3.909429810274563e-06,
+      "loss": 0.8445,
+      "step": 150660
+    },
+    {
+      "epoch": 1.0906498150520822,
+      "grad_norm": 0.1453934758901596,
+      "learning_rate": 3.909357423613977e-06,
+      "loss": 0.8659,
+      "step": 150670
+    },
+    {
+      "epoch": 1.0907222017126683,
+      "grad_norm": 0.14467094838619232,
+      "learning_rate": 3.909285036953391e-06,
+      "loss": 0.8556,
+      "step": 150680
+    },
+    {
+      "epoch": 1.0907945883732546,
+      "grad_norm": 0.16606931388378143,
+      "learning_rate": 3.909212650292805e-06,
+      "loss": 0.8616,
+      "step": 150690
+    },
+    {
+      "epoch": 1.0908669750338407,
+      "grad_norm": 0.1615527719259262,
+      "learning_rate": 3.909140263632218e-06,
+      "loss": 0.866,
+      "step": 150700
+    },
+    {
+      "epoch": 1.090939361694427,
+      "grad_norm": 0.15768449008464813,
+      "learning_rate": 3.909067876971633e-06,
+      "loss": 0.8517,
+      "step": 150710
+    },
+    {
+      "epoch": 1.091011748355013,
+      "grad_norm": 0.15066152811050415,
+      "learning_rate": 3.9089954903110454e-06,
+      "loss": 0.8518,
+      "step": 150720
+    },
+    {
+      "epoch": 1.0910841350155993,
+      "grad_norm": 0.1462964415550232,
+      "learning_rate": 3.908923103650459e-06,
+      "loss": 0.8543,
+      "step": 150730
+    },
+    {
+      "epoch": 1.0911565216761856,
+      "grad_norm": 0.16555063426494598,
+      "learning_rate": 3.9088507169898735e-06,
+      "loss": 0.8591,
+      "step": 150740
+    },
+    {
+      "epoch": 1.0912289083367717,
+      "grad_norm": 0.14814066886901855,
+      "learning_rate": 3.908778330329287e-06,
+      "loss": 0.856,
+      "step": 150750
+    },
+    {
+      "epoch": 1.091301294997358,
+      "grad_norm": 0.16804130375385284,
+      "learning_rate": 3.908705943668701e-06,
+      "loss": 0.8561,
+      "step": 150760
+    },
+    {
+      "epoch": 1.091373681657944,
+      "grad_norm": 0.14669744670391083,
+      "learning_rate": 3.908633557008114e-06,
+      "loss": 0.8556,
+      "step": 150770
+    },
+    {
+      "epoch": 1.0914460683185303,
+      "grad_norm": 0.16342781484127045,
+      "learning_rate": 3.908561170347529e-06,
+      "loss": 0.8664,
+      "step": 150780
+    },
+    {
+      "epoch": 1.0915184549791164,
+      "grad_norm": 0.16906608641147614,
+      "learning_rate": 3.9084887836869425e-06,
+      "loss": 0.8687,
+      "step": 150790
+    },
+    {
+      "epoch": 1.0915908416397027,
+      "grad_norm": 0.15913209319114685,
+      "learning_rate": 3.908416397026356e-06,
+      "loss": 0.8517,
+      "step": 150800
+    },
+    {
+      "epoch": 1.0916632283002887,
+      "grad_norm": 0.15466493368148804,
+      "learning_rate": 3.90834401036577e-06,
+      "loss": 0.8546,
+      "step": 150810
+    },
+    {
+      "epoch": 1.091735614960875,
+      "grad_norm": 0.19748826324939728,
+      "learning_rate": 3.908271623705184e-06,
+      "loss": 0.866,
+      "step": 150820
+    },
+    {
+      "epoch": 1.0918080016214613,
+      "grad_norm": 0.15401212871074677,
+      "learning_rate": 3.908199237044598e-06,
+      "loss": 0.8585,
+      "step": 150830
+    },
+    {
+      "epoch": 1.0918803882820474,
+      "grad_norm": 0.1531715989112854,
+      "learning_rate": 3.908126850384011e-06,
+      "loss": 0.8631,
+      "step": 150840
+    },
+    {
+      "epoch": 1.0919527749426337,
+      "grad_norm": 0.14852984249591827,
+      "learning_rate": 3.908054463723425e-06,
+      "loss": 0.8497,
+      "step": 150850
+    },
+    {
+      "epoch": 1.0920251616032197,
+      "grad_norm": 0.1562282145023346,
+      "learning_rate": 3.9079820770628395e-06,
+      "loss": 0.8547,
+      "step": 150860
+    },
+    {
+      "epoch": 1.092097548263806,
+      "grad_norm": 0.14857418835163116,
+      "learning_rate": 3.907909690402253e-06,
+      "loss": 0.8633,
+      "step": 150870
+    },
+    {
+      "epoch": 1.092169934924392,
+      "grad_norm": 0.1435396671295166,
+      "learning_rate": 3.907837303741667e-06,
+      "loss": 0.8574,
+      "step": 150880
+    },
+    {
+      "epoch": 1.0922423215849784,
+      "grad_norm": 0.1506817787885666,
+      "learning_rate": 3.90776491708108e-06,
+      "loss": 0.8692,
+      "step": 150890
+    },
+    {
+      "epoch": 1.0923147082455644,
+      "grad_norm": 0.1591109335422516,
+      "learning_rate": 3.907692530420494e-06,
+      "loss": 0.8619,
+      "step": 150900
+    },
+    {
+      "epoch": 1.0923870949061507,
+      "grad_norm": 0.16759736835956573,
+      "learning_rate": 3.907620143759908e-06,
+      "loss": 0.8634,
+      "step": 150910
+    },
+    {
+      "epoch": 1.0924594815667368,
+      "grad_norm": 0.15088944137096405,
+      "learning_rate": 3.907547757099322e-06,
+      "loss": 0.8632,
+      "step": 150920
+    },
+    {
+      "epoch": 1.092531868227323,
+      "grad_norm": 0.16850528120994568,
+      "learning_rate": 3.907475370438736e-06,
+      "loss": 0.8562,
+      "step": 150930
+    },
+    {
+      "epoch": 1.0926042548879094,
+      "grad_norm": 0.15842682123184204,
+      "learning_rate": 3.907402983778149e-06,
+      "loss": 0.8653,
+      "step": 150940
+    },
+    {
+      "epoch": 1.0926766415484954,
+      "grad_norm": 0.30550140142440796,
+      "learning_rate": 3.907330597117564e-06,
+      "loss": 0.852,
+      "step": 150950
+    },
+    {
+      "epoch": 1.0927490282090817,
+      "grad_norm": 0.139174684882164,
+      "learning_rate": 3.907258210456977e-06,
+      "loss": 0.8543,
+      "step": 150960
+    },
+    {
+      "epoch": 1.0928214148696678,
+      "grad_norm": 0.15121084451675415,
+      "learning_rate": 3.907185823796391e-06,
+      "loss": 0.8517,
+      "step": 150970
+    },
+    {
+      "epoch": 1.092893801530254,
+      "grad_norm": 0.1478102058172226,
+      "learning_rate": 3.9071134371358046e-06,
+      "loss": 0.8651,
+      "step": 150980
+    },
+    {
+      "epoch": 1.0929661881908401,
+      "grad_norm": 0.1502637416124344,
+      "learning_rate": 3.907041050475219e-06,
+      "loss": 0.8538,
+      "step": 150990
+    },
+    {
+      "epoch": 1.0930385748514264,
+      "grad_norm": 0.14458067715168,
+      "learning_rate": 3.906968663814633e-06,
+      "loss": 0.8546,
+      "step": 151000
+    },
+    {
+      "epoch": 1.0931109615120125,
+      "grad_norm": 0.14992700517177582,
+      "learning_rate": 3.906896277154046e-06,
+      "loss": 0.8584,
+      "step": 151010
+    },
+    {
+      "epoch": 1.0931833481725988,
+      "grad_norm": 0.15242300927639008,
+      "learning_rate": 3.90682389049346e-06,
+      "loss": 0.8674,
+      "step": 151020
+    },
+    {
+      "epoch": 1.093255734833185,
+      "grad_norm": 0.14612126350402832,
+      "learning_rate": 3.906751503832874e-06,
+      "loss": 0.8515,
+      "step": 151030
+    },
+    {
+      "epoch": 1.0933281214937711,
+      "grad_norm": 0.1628275364637375,
+      "learning_rate": 3.906679117172288e-06,
+      "loss": 0.8642,
+      "step": 151040
+    },
+    {
+      "epoch": 1.0934005081543574,
+      "grad_norm": 0.1532110571861267,
+      "learning_rate": 3.9066067305117016e-06,
+      "loss": 0.8566,
+      "step": 151050
+    },
+    {
+      "epoch": 1.0934728948149435,
+      "grad_norm": 0.15354089438915253,
+      "learning_rate": 3.906534343851115e-06,
+      "loss": 0.8534,
+      "step": 151060
+    },
+    {
+      "epoch": 1.0935452814755298,
+      "grad_norm": 0.14480741322040558,
+      "learning_rate": 3.90646195719053e-06,
+      "loss": 0.8601,
+      "step": 151070
+    },
+    {
+      "epoch": 1.0936176681361158,
+      "grad_norm": 0.15542499721050262,
+      "learning_rate": 3.906389570529943e-06,
+      "loss": 0.8666,
+      "step": 151080
+    },
+    {
+      "epoch": 1.0936900547967021,
+      "grad_norm": 0.1690310835838318,
+      "learning_rate": 3.906317183869357e-06,
+      "loss": 0.8571,
+      "step": 151090
+    },
+    {
+      "epoch": 1.0937624414572882,
+      "grad_norm": 0.15053977072238922,
+      "learning_rate": 3.9062447972087705e-06,
+      "loss": 0.8588,
+      "step": 151100
+    },
+    {
+      "epoch": 1.0938348281178745,
+      "grad_norm": 0.15322209894657135,
+      "learning_rate": 3.906172410548185e-06,
+      "loss": 0.8564,
+      "step": 151110
+    },
+    {
+      "epoch": 1.0939072147784605,
+      "grad_norm": 0.15235215425491333,
+      "learning_rate": 3.9061000238875986e-06,
+      "loss": 0.8649,
+      "step": 151120
+    },
+    {
+      "epoch": 1.0939796014390468,
+      "grad_norm": 0.15845470130443573,
+      "learning_rate": 3.906027637227012e-06,
+      "loss": 0.8585,
+      "step": 151130
+    },
+    {
+      "epoch": 1.094051988099633,
+      "grad_norm": 0.1422303318977356,
+      "learning_rate": 3.905955250566426e-06,
+      "loss": 0.8602,
+      "step": 151140
+    },
+    {
+      "epoch": 1.0941243747602192,
+      "grad_norm": 0.16013941168785095,
+      "learning_rate": 3.90588286390584e-06,
+      "loss": 0.8649,
+      "step": 151150
+    },
+    {
+      "epoch": 1.0941967614208055,
+      "grad_norm": 0.16993825137615204,
+      "learning_rate": 3.905810477245254e-06,
+      "loss": 0.8522,
+      "step": 151160
+    },
+    {
+      "epoch": 1.0942691480813915,
+      "grad_norm": 0.16168084740638733,
+      "learning_rate": 3.9057380905846675e-06,
+      "loss": 0.8599,
+      "step": 151170
+    },
+    {
+      "epoch": 1.0943415347419778,
+      "grad_norm": 0.15797677636146545,
+      "learning_rate": 3.905665703924081e-06,
+      "loss": 0.8621,
+      "step": 151180
+    },
+    {
+      "epoch": 1.0944139214025639,
+      "grad_norm": 0.15897414088249207,
+      "learning_rate": 3.9055933172634956e-06,
+      "loss": 0.8508,
+      "step": 151190
+    },
+    {
+      "epoch": 1.0944863080631502,
+      "grad_norm": 0.16979162395000458,
+      "learning_rate": 3.905520930602909e-06,
+      "loss": 0.8518,
+      "step": 151200
+    },
+    {
+      "epoch": 1.0945586947237362,
+      "grad_norm": 0.1656196266412735,
+      "learning_rate": 3.905448543942323e-06,
+      "loss": 0.8492,
+      "step": 151210
+    },
+    {
+      "epoch": 1.0946310813843225,
+      "grad_norm": 0.15596655011177063,
+      "learning_rate": 3.905376157281736e-06,
+      "loss": 0.8483,
+      "step": 151220
+    },
+    {
+      "epoch": 1.0947034680449086,
+      "grad_norm": 0.15505443513393402,
+      "learning_rate": 3.905303770621151e-06,
+      "loss": 0.8654,
+      "step": 151230
+    },
+    {
+      "epoch": 1.0947758547054949,
+      "grad_norm": 0.1517280638217926,
+      "learning_rate": 3.9052313839605645e-06,
+      "loss": 0.8622,
+      "step": 151240
+    },
+    {
+      "epoch": 1.0948482413660812,
+      "grad_norm": 0.1682533472776413,
+      "learning_rate": 3.905158997299977e-06,
+      "loss": 0.8528,
+      "step": 151250
+    },
+    {
+      "epoch": 1.0949206280266672,
+      "grad_norm": 0.1618666797876358,
+      "learning_rate": 3.905086610639392e-06,
+      "loss": 0.8599,
+      "step": 151260
+    },
+    {
+      "epoch": 1.0949930146872535,
+      "grad_norm": 0.15311026573181152,
+      "learning_rate": 3.905014223978805e-06,
+      "loss": 0.8581,
+      "step": 151270
+    },
+    {
+      "epoch": 1.0950654013478396,
+      "grad_norm": 0.18201227486133575,
+      "learning_rate": 3.904941837318219e-06,
+      "loss": 0.8654,
+      "step": 151280
+    },
+    {
+      "epoch": 1.0951377880084259,
+      "grad_norm": 0.1799287348985672,
+      "learning_rate": 3.904869450657633e-06,
+      "loss": 0.8493,
+      "step": 151290
+    },
+    {
+      "epoch": 1.095210174669012,
+      "grad_norm": 0.13958740234375,
+      "learning_rate": 3.904797063997047e-06,
+      "loss": 0.8565,
+      "step": 151300
+    },
+    {
+      "epoch": 1.0952825613295982,
+      "grad_norm": 0.1535644382238388,
+      "learning_rate": 3.904724677336461e-06,
+      "loss": 0.8495,
+      "step": 151310
+    },
+    {
+      "epoch": 1.0953549479901843,
+      "grad_norm": 0.14460021257400513,
+      "learning_rate": 3.904652290675874e-06,
+      "loss": 0.8723,
+      "step": 151320
+    },
+    {
+      "epoch": 1.0954273346507706,
+      "grad_norm": 0.16038860380649567,
+      "learning_rate": 3.904579904015288e-06,
+      "loss": 0.8534,
+      "step": 151330
+    },
+    {
+      "epoch": 1.0954997213113566,
+      "grad_norm": 0.17048697173595428,
+      "learning_rate": 3.904507517354702e-06,
+      "loss": 0.8529,
+      "step": 151340
+    },
+    {
+      "epoch": 1.095572107971943,
+      "grad_norm": 0.15420754253864288,
+      "learning_rate": 3.904435130694116e-06,
+      "loss": 0.8602,
+      "step": 151350
+    },
+    {
+      "epoch": 1.0956444946325292,
+      "grad_norm": 0.14572323858737946,
+      "learning_rate": 3.90436274403353e-06,
+      "loss": 0.8527,
+      "step": 151360
+    },
+    {
+      "epoch": 1.0957168812931153,
+      "grad_norm": 0.16259993612766266,
+      "learning_rate": 3.904290357372943e-06,
+      "loss": 0.8697,
+      "step": 151370
+    },
+    {
+      "epoch": 1.0957892679537016,
+      "grad_norm": 0.15503832697868347,
+      "learning_rate": 3.904217970712358e-06,
+      "loss": 0.862,
+      "step": 151380
+    },
+    {
+      "epoch": 1.0958616546142876,
+      "grad_norm": 0.1643747240304947,
+      "learning_rate": 3.904145584051771e-06,
+      "loss": 0.8619,
+      "step": 151390
+    },
+    {
+      "epoch": 1.095934041274874,
+      "grad_norm": 0.1568668931722641,
+      "learning_rate": 3.904073197391185e-06,
+      "loss": 0.8584,
+      "step": 151400
+    },
+    {
+      "epoch": 1.09600642793546,
+      "grad_norm": 0.1458984911441803,
+      "learning_rate": 3.9040008107305985e-06,
+      "loss": 0.8597,
+      "step": 151410
+    },
+    {
+      "epoch": 1.0960788145960463,
+      "grad_norm": 0.15451470017433167,
+      "learning_rate": 3.903928424070013e-06,
+      "loss": 0.8573,
+      "step": 151420
+    },
+    {
+      "epoch": 1.0961512012566323,
+      "grad_norm": 0.15595468878746033,
+      "learning_rate": 3.903856037409427e-06,
+      "loss": 0.8785,
+      "step": 151430
+    },
+    {
+      "epoch": 1.0962235879172186,
+      "grad_norm": 0.15538018941879272,
+      "learning_rate": 3.90378365074884e-06,
+      "loss": 0.8642,
+      "step": 151440
+    },
+    {
+      "epoch": 1.0962959745778047,
+      "grad_norm": 0.15324239432811737,
+      "learning_rate": 3.903711264088254e-06,
+      "loss": 0.8584,
+      "step": 151450
+    },
+    {
+      "epoch": 1.096368361238391,
+      "grad_norm": 0.14837108552455902,
+      "learning_rate": 3.903638877427668e-06,
+      "loss": 0.8654,
+      "step": 151460
+    },
+    {
+      "epoch": 1.0964407478989773,
+      "grad_norm": 0.1704886555671692,
+      "learning_rate": 3.903566490767082e-06,
+      "loss": 0.8615,
+      "step": 151470
+    },
+    {
+      "epoch": 1.0965131345595633,
+      "grad_norm": 0.16142944991588593,
+      "learning_rate": 3.9034941041064955e-06,
+      "loss": 0.8701,
+      "step": 151480
+    },
+    {
+      "epoch": 1.0965855212201496,
+      "grad_norm": 0.16319426894187927,
+      "learning_rate": 3.903421717445909e-06,
+      "loss": 0.8603,
+      "step": 151490
+    },
+    {
+      "epoch": 1.0966579078807357,
+      "grad_norm": 0.18104782700538635,
+      "learning_rate": 3.903349330785324e-06,
+      "loss": 0.8598,
+      "step": 151500
+    },
+    {
+      "epoch": 1.096730294541322,
+      "grad_norm": 0.15671753883361816,
+      "learning_rate": 3.903276944124737e-06,
+      "loss": 0.8621,
+      "step": 151510
+    },
+    {
+      "epoch": 1.096802681201908,
+      "grad_norm": 0.1673654317855835,
+      "learning_rate": 3.903204557464151e-06,
+      "loss": 0.8553,
+      "step": 151520
+    },
+    {
+      "epoch": 1.0968750678624943,
+      "grad_norm": 0.14993630349636078,
+      "learning_rate": 3.9031321708035645e-06,
+      "loss": 0.846,
+      "step": 151530
+    },
+    {
+      "epoch": 1.0969474545230804,
+      "grad_norm": 0.16005301475524902,
+      "learning_rate": 3.903059784142978e-06,
+      "loss": 0.8595,
+      "step": 151540
+    },
+    {
+      "epoch": 1.0970198411836667,
+      "grad_norm": 0.1531304568052292,
+      "learning_rate": 3.9029873974823925e-06,
+      "loss": 0.8566,
+      "step": 151550
+    },
+    {
+      "epoch": 1.0970922278442528,
+      "grad_norm": 0.15316185355186462,
+      "learning_rate": 3.902915010821806e-06,
+      "loss": 0.848,
+      "step": 151560
+    },
+    {
+      "epoch": 1.097164614504839,
+      "grad_norm": 0.16039323806762695,
+      "learning_rate": 3.90284262416122e-06,
+      "loss": 0.8562,
+      "step": 151570
+    },
+    {
+      "epoch": 1.0972370011654253,
+      "grad_norm": 0.14785118401050568,
+      "learning_rate": 3.902770237500633e-06,
+      "loss": 0.8508,
+      "step": 151580
+    },
+    {
+      "epoch": 1.0973093878260114,
+      "grad_norm": 0.16044607758522034,
+      "learning_rate": 3.902697850840048e-06,
+      "loss": 0.8569,
+      "step": 151590
+    },
+    {
+      "epoch": 1.0973817744865977,
+      "grad_norm": 0.16360950469970703,
+      "learning_rate": 3.9026254641794615e-06,
+      "loss": 0.8555,
+      "step": 151600
+    },
+    {
+      "epoch": 1.0974541611471837,
+      "grad_norm": 0.15035872161388397,
+      "learning_rate": 3.902553077518875e-06,
+      "loss": 0.8524,
+      "step": 151610
+    },
+    {
+      "epoch": 1.09752654780777,
+      "grad_norm": 0.19104793667793274,
+      "learning_rate": 3.902480690858289e-06,
+      "loss": 0.8468,
+      "step": 151620
+    },
+    {
+      "epoch": 1.097598934468356,
+      "grad_norm": 0.1884557455778122,
+      "learning_rate": 3.902408304197703e-06,
+      "loss": 0.8692,
+      "step": 151630
+    },
+    {
+      "epoch": 1.0976713211289424,
+      "grad_norm": 0.15501312911510468,
+      "learning_rate": 3.902335917537117e-06,
+      "loss": 0.8693,
+      "step": 151640
+    },
+    {
+      "epoch": 1.0977437077895285,
+      "grad_norm": 0.1506832242012024,
+      "learning_rate": 3.90226353087653e-06,
+      "loss": 0.8568,
+      "step": 151650
+    },
+    {
+      "epoch": 1.0978160944501147,
+      "grad_norm": 0.1559952199459076,
+      "learning_rate": 3.902191144215944e-06,
+      "loss": 0.8556,
+      "step": 151660
+    },
+    {
+      "epoch": 1.0978884811107008,
+      "grad_norm": 0.17316295206546783,
+      "learning_rate": 3.9021187575553585e-06,
+      "loss": 0.8683,
+      "step": 151670
+    },
+    {
+      "epoch": 1.097960867771287,
+      "grad_norm": 0.1504284292459488,
+      "learning_rate": 3.902046370894772e-06,
+      "loss": 0.8607,
+      "step": 151680
+    },
+    {
+      "epoch": 1.0980332544318734,
+      "grad_norm": 0.15068010985851288,
+      "learning_rate": 3.901973984234186e-06,
+      "loss": 0.849,
+      "step": 151690
+    },
+    {
+      "epoch": 1.0981056410924595,
+      "grad_norm": 0.14526309072971344,
+      "learning_rate": 3.901901597573599e-06,
+      "loss": 0.8576,
+      "step": 151700
+    },
+    {
+      "epoch": 1.0981780277530457,
+      "grad_norm": 0.1651238054037094,
+      "learning_rate": 3.901829210913014e-06,
+      "loss": 0.8712,
+      "step": 151710
+    },
+    {
+      "epoch": 1.0982504144136318,
+      "grad_norm": 0.16406778991222382,
+      "learning_rate": 3.901756824252427e-06,
+      "loss": 0.8533,
+      "step": 151720
+    },
+    {
+      "epoch": 1.098322801074218,
+      "grad_norm": 0.15506918728351593,
+      "learning_rate": 3.901684437591841e-06,
+      "loss": 0.8644,
+      "step": 151730
+    },
+    {
+      "epoch": 1.0983951877348042,
+      "grad_norm": 0.1499270647764206,
+      "learning_rate": 3.901612050931255e-06,
+      "loss": 0.8435,
+      "step": 151740
+    },
+    {
+      "epoch": 1.0984675743953904,
+      "grad_norm": 0.15782611072063446,
+      "learning_rate": 3.901539664270669e-06,
+      "loss": 0.8602,
+      "step": 151750
+    },
+    {
+      "epoch": 1.0985399610559765,
+      "grad_norm": 0.14612479507923126,
+      "learning_rate": 3.901467277610083e-06,
+      "loss": 0.8569,
+      "step": 151760
+    },
+    {
+      "epoch": 1.0986123477165628,
+      "grad_norm": 0.15989379584789276,
+      "learning_rate": 3.901394890949496e-06,
+      "loss": 0.8473,
+      "step": 151770
+    },
+    {
+      "epoch": 1.0986847343771489,
+      "grad_norm": 0.14152558147907257,
+      "learning_rate": 3.90132250428891e-06,
+      "loss": 0.8588,
+      "step": 151780
+    },
+    {
+      "epoch": 1.0987571210377352,
+      "grad_norm": 0.14922699332237244,
+      "learning_rate": 3.9012501176283236e-06,
+      "loss": 0.8483,
+      "step": 151790
+    },
+    {
+      "epoch": 1.0988295076983214,
+      "grad_norm": 0.15178348124027252,
+      "learning_rate": 3.901177730967737e-06,
+      "loss": 0.8548,
+      "step": 151800
+    },
+    {
+      "epoch": 1.0989018943589075,
+      "grad_norm": 0.1589786410331726,
+      "learning_rate": 3.901105344307151e-06,
+      "loss": 0.8626,
+      "step": 151810
+    },
+    {
+      "epoch": 1.0989742810194938,
+      "grad_norm": 0.15005895495414734,
+      "learning_rate": 3.901032957646565e-06,
+      "loss": 0.8627,
+      "step": 151820
+    },
+    {
+      "epoch": 1.0990466676800799,
+      "grad_norm": 0.14310167729854584,
+      "learning_rate": 3.900960570985979e-06,
+      "loss": 0.8621,
+      "step": 151830
+    },
+    {
+      "epoch": 1.0991190543406661,
+      "grad_norm": 0.1532042920589447,
+      "learning_rate": 3.9008881843253925e-06,
+      "loss": 0.8571,
+      "step": 151840
+    },
+    {
+      "epoch": 1.0991914410012522,
+      "grad_norm": 0.15156692266464233,
+      "learning_rate": 3.900815797664806e-06,
+      "loss": 0.856,
+      "step": 151850
+    },
+    {
+      "epoch": 1.0992638276618385,
+      "grad_norm": 0.17657136917114258,
+      "learning_rate": 3.9007434110042206e-06,
+      "loss": 0.8521,
+      "step": 151860
+    },
+    {
+      "epoch": 1.0993362143224246,
+      "grad_norm": 0.17290574312210083,
+      "learning_rate": 3.900671024343634e-06,
+      "loss": 0.8681,
+      "step": 151870
+    },
+    {
+      "epoch": 1.0994086009830109,
+      "grad_norm": 0.14646093547344208,
+      "learning_rate": 3.900598637683048e-06,
+      "loss": 0.8464,
+      "step": 151880
+    },
+    {
+      "epoch": 1.0994809876435971,
+      "grad_norm": 0.14447811245918274,
+      "learning_rate": 3.900526251022461e-06,
+      "loss": 0.8395,
+      "step": 151890
+    },
+    {
+      "epoch": 1.0995533743041832,
+      "grad_norm": 0.16864459216594696,
+      "learning_rate": 3.900453864361876e-06,
+      "loss": 0.8542,
+      "step": 151900
+    },
+    {
+      "epoch": 1.0996257609647695,
+      "grad_norm": 0.1442280113697052,
+      "learning_rate": 3.9003814777012895e-06,
+      "loss": 0.853,
+      "step": 151910
+    },
+    {
+      "epoch": 1.0996981476253556,
+      "grad_norm": 0.15601858496665955,
+      "learning_rate": 3.900309091040703e-06,
+      "loss": 0.8549,
+      "step": 151920
+    },
+    {
+      "epoch": 1.0997705342859418,
+      "grad_norm": 0.14018070697784424,
+      "learning_rate": 3.900236704380117e-06,
+      "loss": 0.8535,
+      "step": 151930
+    },
+    {
+      "epoch": 1.099842920946528,
+      "grad_norm": 0.14601945877075195,
+      "learning_rate": 3.900164317719531e-06,
+      "loss": 0.8679,
+      "step": 151940
+    },
+    {
+      "epoch": 1.0999153076071142,
+      "grad_norm": 0.1592642217874527,
+      "learning_rate": 3.900091931058945e-06,
+      "loss": 0.8574,
+      "step": 151950
+    },
+    {
+      "epoch": 1.0999876942677003,
+      "grad_norm": 0.1683703064918518,
+      "learning_rate": 3.900019544398358e-06,
+      "loss": 0.8592,
+      "step": 151960
+    },
+    {
+      "epoch": 1.1000600809282866,
+      "grad_norm": 0.15897010266780853,
+      "learning_rate": 3.899947157737772e-06,
+      "loss": 0.8487,
+      "step": 151970
+    },
+    {
+      "epoch": 1.1001324675888726,
+      "grad_norm": 0.14617374539375305,
+      "learning_rate": 3.8998747710771865e-06,
+      "loss": 0.8556,
+      "step": 151980
+    },
+    {
+      "epoch": 1.100204854249459,
+      "grad_norm": 0.157510906457901,
+      "learning_rate": 3.8998023844166e-06,
+      "loss": 0.857,
+      "step": 151990
+    },
+    {
+      "epoch": 1.1002772409100452,
+      "grad_norm": 0.1601264476776123,
+      "learning_rate": 3.899729997756014e-06,
+      "loss": 0.8583,
+      "step": 152000
+    },
+    {
+      "epoch": 1.1003496275706313,
+      "grad_norm": 0.16230392456054688,
+      "learning_rate": 3.899657611095427e-06,
+      "loss": 0.8503,
+      "step": 152010
+    },
+    {
+      "epoch": 1.1004220142312175,
+      "grad_norm": 0.14697298407554626,
+      "learning_rate": 3.899585224434842e-06,
+      "loss": 0.8452,
+      "step": 152020
+    },
+    {
+      "epoch": 1.1004944008918036,
+      "grad_norm": 0.1654907613992691,
+      "learning_rate": 3.8995128377742554e-06,
+      "loss": 0.8531,
+      "step": 152030
+    },
+    {
+      "epoch": 1.10056678755239,
+      "grad_norm": 0.1564827561378479,
+      "learning_rate": 3.899440451113669e-06,
+      "loss": 0.8554,
+      "step": 152040
+    },
+    {
+      "epoch": 1.100639174212976,
+      "grad_norm": 0.14684629440307617,
+      "learning_rate": 3.899368064453083e-06,
+      "loss": 0.8562,
+      "step": 152050
+    },
+    {
+      "epoch": 1.1007115608735623,
+      "grad_norm": 0.15104876458644867,
+      "learning_rate": 3.899295677792497e-06,
+      "loss": 0.8607,
+      "step": 152060
+    },
+    {
+      "epoch": 1.1007839475341483,
+      "grad_norm": 0.15481539070606232,
+      "learning_rate": 3.899223291131911e-06,
+      "loss": 0.8587,
+      "step": 152070
+    },
+    {
+      "epoch": 1.1008563341947346,
+      "grad_norm": 0.15481428802013397,
+      "learning_rate": 3.899150904471324e-06,
+      "loss": 0.8526,
+      "step": 152080
+    },
+    {
+      "epoch": 1.100928720855321,
+      "grad_norm": 0.1640326827764511,
+      "learning_rate": 3.899078517810738e-06,
+      "loss": 0.8632,
+      "step": 152090
+    },
+    {
+      "epoch": 1.101001107515907,
+      "grad_norm": 0.1505155861377716,
+      "learning_rate": 3.8990061311501524e-06,
+      "loss": 0.8439,
+      "step": 152100
+    },
+    {
+      "epoch": 1.1010734941764933,
+      "grad_norm": 0.1723697930574417,
+      "learning_rate": 3.898933744489566e-06,
+      "loss": 0.8604,
+      "step": 152110
+    },
+    {
+      "epoch": 1.1011458808370793,
+      "grad_norm": 0.16168347001075745,
+      "learning_rate": 3.89886135782898e-06,
+      "loss": 0.8418,
+      "step": 152120
+    },
+    {
+      "epoch": 1.1012182674976656,
+      "grad_norm": 1.0787012577056885,
+      "learning_rate": 3.898788971168393e-06,
+      "loss": 0.8555,
+      "step": 152130
+    },
+    {
+      "epoch": 1.1012906541582517,
+      "grad_norm": 0.15610916912555695,
+      "learning_rate": 3.898716584507807e-06,
+      "loss": 0.8459,
+      "step": 152140
+    },
+    {
+      "epoch": 1.101363040818838,
+      "grad_norm": 0.1682717204093933,
+      "learning_rate": 3.898644197847221e-06,
+      "loss": 0.8496,
+      "step": 152150
+    },
+    {
+      "epoch": 1.101435427479424,
+      "grad_norm": 0.34548187255859375,
+      "learning_rate": 3.898571811186635e-06,
+      "loss": 0.8519,
+      "step": 152160
+    },
+    {
+      "epoch": 1.1015078141400103,
+      "grad_norm": 0.15088237822055817,
+      "learning_rate": 3.898499424526049e-06,
+      "loss": 0.8588,
+      "step": 152170
+    },
+    {
+      "epoch": 1.1015802008005964,
+      "grad_norm": 0.13609455525875092,
+      "learning_rate": 3.898427037865462e-06,
+      "loss": 0.847,
+      "step": 152180
+    },
+    {
+      "epoch": 1.1016525874611827,
+      "grad_norm": 0.14786964654922485,
+      "learning_rate": 3.898354651204877e-06,
+      "loss": 0.8687,
+      "step": 152190
+    },
+    {
+      "epoch": 1.101724974121769,
+      "grad_norm": 0.14596401154994965,
+      "learning_rate": 3.89828226454429e-06,
+      "loss": 0.8543,
+      "step": 152200
+    },
+    {
+      "epoch": 1.101797360782355,
+      "grad_norm": 0.1763083040714264,
+      "learning_rate": 3.898209877883704e-06,
+      "loss": 0.8499,
+      "step": 152210
+    },
+    {
+      "epoch": 1.1018697474429413,
+      "grad_norm": 0.19486874341964722,
+      "learning_rate": 3.8981374912231175e-06,
+      "loss": 0.857,
+      "step": 152220
+    },
+    {
+      "epoch": 1.1019421341035274,
+      "grad_norm": 0.15471166372299194,
+      "learning_rate": 3.898065104562532e-06,
+      "loss": 0.8454,
+      "step": 152230
+    },
+    {
+      "epoch": 1.1020145207641137,
+      "grad_norm": 0.14786270260810852,
+      "learning_rate": 3.897992717901946e-06,
+      "loss": 0.8673,
+      "step": 152240
+    },
+    {
+      "epoch": 1.1020869074246997,
+      "grad_norm": 0.14872819185256958,
+      "learning_rate": 3.897920331241359e-06,
+      "loss": 0.8543,
+      "step": 152250
+    },
+    {
+      "epoch": 1.102159294085286,
+      "grad_norm": 0.1608414351940155,
+      "learning_rate": 3.897847944580773e-06,
+      "loss": 0.8618,
+      "step": 152260
+    },
+    {
+      "epoch": 1.102231680745872,
+      "grad_norm": 0.1444377303123474,
+      "learning_rate": 3.897775557920187e-06,
+      "loss": 0.8486,
+      "step": 152270
+    },
+    {
+      "epoch": 1.1023040674064584,
+      "grad_norm": 0.16134855151176453,
+      "learning_rate": 3.897703171259601e-06,
+      "loss": 0.8642,
+      "step": 152280
+    },
+    {
+      "epoch": 1.1023764540670444,
+      "grad_norm": 0.1678602695465088,
+      "learning_rate": 3.8976307845990145e-06,
+      "loss": 0.8583,
+      "step": 152290
+    },
+    {
+      "epoch": 1.1024488407276307,
+      "grad_norm": 0.15565115213394165,
+      "learning_rate": 3.897558397938428e-06,
+      "loss": 0.8503,
+      "step": 152300
+    },
+    {
+      "epoch": 1.102521227388217,
+      "grad_norm": 0.16068263351917267,
+      "learning_rate": 3.897486011277842e-06,
+      "loss": 0.8734,
+      "step": 152310
+    },
+    {
+      "epoch": 1.102593614048803,
+      "grad_norm": 0.15741673111915588,
+      "learning_rate": 3.897413624617255e-06,
+      "loss": 0.8602,
+      "step": 152320
+    },
+    {
+      "epoch": 1.1026660007093894,
+      "grad_norm": 0.17430439591407776,
+      "learning_rate": 3.897341237956669e-06,
+      "loss": 0.8542,
+      "step": 152330
+    },
+    {
+      "epoch": 1.1027383873699754,
+      "grad_norm": 0.15103577077388763,
+      "learning_rate": 3.8972688512960835e-06,
+      "loss": 0.8624,
+      "step": 152340
+    },
+    {
+      "epoch": 1.1028107740305617,
+      "grad_norm": 0.15457139909267426,
+      "learning_rate": 3.897196464635497e-06,
+      "loss": 0.8546,
+      "step": 152350
+    },
+    {
+      "epoch": 1.1028831606911478,
+      "grad_norm": 0.1677873432636261,
+      "learning_rate": 3.897124077974911e-06,
+      "loss": 0.8774,
+      "step": 152360
+    },
+    {
+      "epoch": 1.102955547351734,
+      "grad_norm": 0.1446024477481842,
+      "learning_rate": 3.897051691314324e-06,
+      "loss": 0.8584,
+      "step": 152370
+    },
+    {
+      "epoch": 1.1030279340123201,
+      "grad_norm": 0.1553516536951065,
+      "learning_rate": 3.896979304653739e-06,
+      "loss": 0.855,
+      "step": 152380
+    },
+    {
+      "epoch": 1.1031003206729064,
+      "grad_norm": 0.15812979638576508,
+      "learning_rate": 3.896906917993152e-06,
+      "loss": 0.8538,
+      "step": 152390
+    },
+    {
+      "epoch": 1.1031727073334925,
+      "grad_norm": 0.16017301380634308,
+      "learning_rate": 3.896834531332566e-06,
+      "loss": 0.8485,
+      "step": 152400
+    },
+    {
+      "epoch": 1.1032450939940788,
+      "grad_norm": 0.15883536636829376,
+      "learning_rate": 3.89676214467198e-06,
+      "loss": 0.8547,
+      "step": 152410
+    },
+    {
+      "epoch": 1.103317480654665,
+      "grad_norm": 0.16042304039001465,
+      "learning_rate": 3.896689758011394e-06,
+      "loss": 0.8527,
+      "step": 152420
+    },
+    {
+      "epoch": 1.1033898673152511,
+      "grad_norm": 0.1522279679775238,
+      "learning_rate": 3.896617371350808e-06,
+      "loss": 0.8571,
+      "step": 152430
+    },
+    {
+      "epoch": 1.1034622539758374,
+      "grad_norm": 0.15726742148399353,
+      "learning_rate": 3.896544984690221e-06,
+      "loss": 0.8514,
+      "step": 152440
+    },
+    {
+      "epoch": 1.1035346406364235,
+      "grad_norm": 0.1631903052330017,
+      "learning_rate": 3.896472598029635e-06,
+      "loss": 0.8575,
+      "step": 152450
+    },
+    {
+      "epoch": 1.1036070272970098,
+      "grad_norm": 0.15091995894908905,
+      "learning_rate": 3.896400211369049e-06,
+      "loss": 0.8496,
+      "step": 152460
+    },
+    {
+      "epoch": 1.1036794139575958,
+      "grad_norm": 0.1674959510564804,
+      "learning_rate": 3.896327824708463e-06,
+      "loss": 0.863,
+      "step": 152470
+    },
+    {
+      "epoch": 1.1037518006181821,
+      "grad_norm": 0.15883536636829376,
+      "learning_rate": 3.896255438047877e-06,
+      "loss": 0.8545,
+      "step": 152480
+    },
+    {
+      "epoch": 1.1038241872787682,
+      "grad_norm": 0.17401011288166046,
+      "learning_rate": 3.89618305138729e-06,
+      "loss": 0.859,
+      "step": 152490
+    },
+    {
+      "epoch": 1.1038965739393545,
+      "grad_norm": 0.1531701236963272,
+      "learning_rate": 3.896110664726705e-06,
+      "loss": 0.8492,
+      "step": 152500
+    },
+    {
+      "epoch": 1.1039689605999405,
+      "grad_norm": 0.15053513646125793,
+      "learning_rate": 3.896038278066118e-06,
+      "loss": 0.8485,
+      "step": 152510
+    },
+    {
+      "epoch": 1.1040413472605268,
+      "grad_norm": 0.15643155574798584,
+      "learning_rate": 3.895965891405532e-06,
+      "loss": 0.8485,
+      "step": 152520
+    },
+    {
+      "epoch": 1.1041137339211131,
+      "grad_norm": 0.14518050849437714,
+      "learning_rate": 3.8958935047449456e-06,
+      "loss": 0.8522,
+      "step": 152530
+    },
+    {
+      "epoch": 1.1041861205816992,
+      "grad_norm": 0.16403423249721527,
+      "learning_rate": 3.89582111808436e-06,
+      "loss": 0.8595,
+      "step": 152540
+    },
+    {
+      "epoch": 1.1042585072422855,
+      "grad_norm": 0.18213549256324768,
+      "learning_rate": 3.895748731423774e-06,
+      "loss": 0.8591,
+      "step": 152550
+    },
+    {
+      "epoch": 1.1043308939028715,
+      "grad_norm": 0.15698887407779694,
+      "learning_rate": 3.895676344763187e-06,
+      "loss": 0.8547,
+      "step": 152560
+    },
+    {
+      "epoch": 1.1044032805634578,
+      "grad_norm": 0.1551319658756256,
+      "learning_rate": 3.895603958102601e-06,
+      "loss": 0.8563,
+      "step": 152570
+    },
+    {
+      "epoch": 1.1044756672240439,
+      "grad_norm": 0.1502751260995865,
+      "learning_rate": 3.895531571442015e-06,
+      "loss": 0.8594,
+      "step": 152580
+    },
+    {
+      "epoch": 1.1045480538846302,
+      "grad_norm": 0.15094530582427979,
+      "learning_rate": 3.895459184781429e-06,
+      "loss": 0.8501,
+      "step": 152590
+    },
+    {
+      "epoch": 1.1046204405452162,
+      "grad_norm": 0.14504896104335785,
+      "learning_rate": 3.8953867981208426e-06,
+      "loss": 0.8475,
+      "step": 152600
+    },
+    {
+      "epoch": 1.1046928272058025,
+      "grad_norm": 0.14708517491817474,
+      "learning_rate": 3.895314411460256e-06,
+      "loss": 0.8641,
+      "step": 152610
+    },
+    {
+      "epoch": 1.1047652138663886,
+      "grad_norm": 0.15217843651771545,
+      "learning_rate": 3.895242024799671e-06,
+      "loss": 0.8524,
+      "step": 152620
+    },
+    {
+      "epoch": 1.1048376005269749,
+      "grad_norm": 0.15076854825019836,
+      "learning_rate": 3.895169638139084e-06,
+      "loss": 0.8604,
+      "step": 152630
+    },
+    {
+      "epoch": 1.1049099871875612,
+      "grad_norm": 0.1549665480852127,
+      "learning_rate": 3.895097251478498e-06,
+      "loss": 0.8572,
+      "step": 152640
+    },
+    {
+      "epoch": 1.1049823738481472,
+      "grad_norm": 0.146447092294693,
+      "learning_rate": 3.8950248648179115e-06,
+      "loss": 0.8508,
+      "step": 152650
+    },
+    {
+      "epoch": 1.1050547605087335,
+      "grad_norm": 0.156124547123909,
+      "learning_rate": 3.894952478157326e-06,
+      "loss": 0.863,
+      "step": 152660
+    },
+    {
+      "epoch": 1.1051271471693196,
+      "grad_norm": 0.14300669729709625,
+      "learning_rate": 3.8948800914967396e-06,
+      "loss": 0.8673,
+      "step": 152670
+    },
+    {
+      "epoch": 1.1051995338299059,
+      "grad_norm": 0.1642896682024002,
+      "learning_rate": 3.894807704836153e-06,
+      "loss": 0.8495,
+      "step": 152680
+    },
+    {
+      "epoch": 1.105271920490492,
+      "grad_norm": 0.15257836878299713,
+      "learning_rate": 3.894735318175567e-06,
+      "loss": 0.8463,
+      "step": 152690
+    },
+    {
+      "epoch": 1.1053443071510782,
+      "grad_norm": 0.16251547634601593,
+      "learning_rate": 3.894662931514981e-06,
+      "loss": 0.8402,
+      "step": 152700
+    },
+    {
+      "epoch": 1.1054166938116643,
+      "grad_norm": 0.145891934633255,
+      "learning_rate": 3.894590544854395e-06,
+      "loss": 0.8492,
+      "step": 152710
+    },
+    {
+      "epoch": 1.1054890804722506,
+      "grad_norm": 0.16337795555591583,
+      "learning_rate": 3.8945181581938085e-06,
+      "loss": 0.8457,
+      "step": 152720
+    },
+    {
+      "epoch": 1.1055614671328367,
+      "grad_norm": 0.1507614105939865,
+      "learning_rate": 3.894445771533222e-06,
+      "loss": 0.8649,
+      "step": 152730
+    },
+    {
+      "epoch": 1.105633853793423,
+      "grad_norm": 0.16109922528266907,
+      "learning_rate": 3.894373384872637e-06,
+      "loss": 0.8611,
+      "step": 152740
+    },
+    {
+      "epoch": 1.1057062404540092,
+      "grad_norm": 0.14710114896297455,
+      "learning_rate": 3.89430099821205e-06,
+      "loss": 0.854,
+      "step": 152750
+    },
+    {
+      "epoch": 1.1057786271145953,
+      "grad_norm": 0.14651963114738464,
+      "learning_rate": 3.894228611551464e-06,
+      "loss": 0.8518,
+      "step": 152760
+    },
+    {
+      "epoch": 1.1058510137751816,
+      "grad_norm": 0.1510373204946518,
+      "learning_rate": 3.8941562248908774e-06,
+      "loss": 0.8726,
+      "step": 152770
+    },
+    {
+      "epoch": 1.1059234004357676,
+      "grad_norm": 0.14552970230579376,
+      "learning_rate": 3.894083838230291e-06,
+      "loss": 0.8645,
+      "step": 152780
+    },
+    {
+      "epoch": 1.105995787096354,
+      "grad_norm": 0.1477714329957962,
+      "learning_rate": 3.8940114515697055e-06,
+      "loss": 0.8663,
+      "step": 152790
+    },
+    {
+      "epoch": 1.10606817375694,
+      "grad_norm": 0.1511882096529007,
+      "learning_rate": 3.893939064909119e-06,
+      "loss": 0.8589,
+      "step": 152800
+    },
+    {
+      "epoch": 1.1061405604175263,
+      "grad_norm": 0.16555480659008026,
+      "learning_rate": 3.893866678248533e-06,
+      "loss": 0.8446,
+      "step": 152810
+    },
+    {
+      "epoch": 1.1062129470781124,
+      "grad_norm": 0.15422996878623962,
+      "learning_rate": 3.893794291587946e-06,
+      "loss": 0.8476,
+      "step": 152820
+    },
+    {
+      "epoch": 1.1062853337386986,
+      "grad_norm": 0.16590246558189392,
+      "learning_rate": 3.893721904927361e-06,
+      "loss": 0.8658,
+      "step": 152830
+    },
+    {
+      "epoch": 1.1063577203992847,
+      "grad_norm": 0.15505556762218475,
+      "learning_rate": 3.893649518266774e-06,
+      "loss": 0.8438,
+      "step": 152840
+    },
+    {
+      "epoch": 1.106430107059871,
+      "grad_norm": 0.15418249368667603,
+      "learning_rate": 3.893577131606188e-06,
+      "loss": 0.8507,
+      "step": 152850
+    },
+    {
+      "epoch": 1.1065024937204573,
+      "grad_norm": 0.15402624011039734,
+      "learning_rate": 3.893504744945602e-06,
+      "loss": 0.8656,
+      "step": 152860
+    },
+    {
+      "epoch": 1.1065748803810433,
+      "grad_norm": 0.15612943470478058,
+      "learning_rate": 3.893432358285015e-06,
+      "loss": 0.8594,
+      "step": 152870
+    },
+    {
+      "epoch": 1.1066472670416296,
+      "grad_norm": 0.1624891459941864,
+      "learning_rate": 3.893359971624429e-06,
+      "loss": 0.8577,
+      "step": 152880
+    },
+    {
+      "epoch": 1.1067196537022157,
+      "grad_norm": 0.18142132461071014,
+      "learning_rate": 3.893287584963843e-06,
+      "loss": 0.8541,
+      "step": 152890
+    },
+    {
+      "epoch": 1.106792040362802,
+      "grad_norm": 0.16291342675685883,
+      "learning_rate": 3.893215198303257e-06,
+      "loss": 0.8546,
+      "step": 152900
+    },
+    {
+      "epoch": 1.106864427023388,
+      "grad_norm": 0.15221039950847626,
+      "learning_rate": 3.893142811642671e-06,
+      "loss": 0.8548,
+      "step": 152910
+    },
+    {
+      "epoch": 1.1069368136839743,
+      "grad_norm": 0.15222559869289398,
+      "learning_rate": 3.893070424982084e-06,
+      "loss": 0.8612,
+      "step": 152920
+    },
+    {
+      "epoch": 1.1070092003445604,
+      "grad_norm": 0.15900181233882904,
+      "learning_rate": 3.892998038321498e-06,
+      "loss": 0.8725,
+      "step": 152930
+    },
+    {
+      "epoch": 1.1070815870051467,
+      "grad_norm": 0.1619795560836792,
+      "learning_rate": 3.892925651660912e-06,
+      "loss": 0.8511,
+      "step": 152940
+    },
+    {
+      "epoch": 1.107153973665733,
+      "grad_norm": 0.1571727842092514,
+      "learning_rate": 3.892853265000326e-06,
+      "loss": 0.8482,
+      "step": 152950
+    },
+    {
+      "epoch": 1.107226360326319,
+      "grad_norm": 0.13933265209197998,
+      "learning_rate": 3.8927808783397395e-06,
+      "loss": 0.8563,
+      "step": 152960
+    },
+    {
+      "epoch": 1.1072987469869053,
+      "grad_norm": 0.14725233614444733,
+      "learning_rate": 3.892708491679153e-06,
+      "loss": 0.8507,
+      "step": 152970
+    },
+    {
+      "epoch": 1.1073711336474914,
+      "grad_norm": 0.1523694097995758,
+      "learning_rate": 3.892636105018568e-06,
+      "loss": 0.8503,
+      "step": 152980
+    },
+    {
+      "epoch": 1.1074435203080777,
+      "grad_norm": 0.1705412119626999,
+      "learning_rate": 3.892563718357981e-06,
+      "loss": 0.8545,
+      "step": 152990
+    },
+    {
+      "epoch": 1.1075159069686638,
+      "grad_norm": 0.14305712282657623,
+      "learning_rate": 3.892491331697395e-06,
+      "loss": 0.8585,
+      "step": 153000
+    },
+    {
+      "epoch": 1.10758829362925,
+      "grad_norm": 0.1486334204673767,
+      "learning_rate": 3.8924189450368085e-06,
+      "loss": 0.847,
+      "step": 153010
+    },
+    {
+      "epoch": 1.107660680289836,
+      "grad_norm": 0.16062171757221222,
+      "learning_rate": 3.892346558376223e-06,
+      "loss": 0.8527,
+      "step": 153020
+    },
+    {
+      "epoch": 1.1077330669504224,
+      "grad_norm": 0.14781929552555084,
+      "learning_rate": 3.8922741717156365e-06,
+      "loss": 0.8677,
+      "step": 153030
+    },
+    {
+      "epoch": 1.1078054536110085,
+      "grad_norm": 0.14651741087436676,
+      "learning_rate": 3.89220178505505e-06,
+      "loss": 0.8535,
+      "step": 153040
+    },
+    {
+      "epoch": 1.1078778402715947,
+      "grad_norm": 0.14890103042125702,
+      "learning_rate": 3.892129398394464e-06,
+      "loss": 0.863,
+      "step": 153050
+    },
+    {
+      "epoch": 1.107950226932181,
+      "grad_norm": 0.14752887189388275,
+      "learning_rate": 3.892057011733878e-06,
+      "loss": 0.8494,
+      "step": 153060
+    },
+    {
+      "epoch": 1.108022613592767,
+      "grad_norm": 0.15425044298171997,
+      "learning_rate": 3.891984625073292e-06,
+      "loss": 0.8693,
+      "step": 153070
+    },
+    {
+      "epoch": 1.1080950002533534,
+      "grad_norm": 0.14572696387767792,
+      "learning_rate": 3.8919122384127055e-06,
+      "loss": 0.8467,
+      "step": 153080
+    },
+    {
+      "epoch": 1.1081673869139395,
+      "grad_norm": 0.16281406581401825,
+      "learning_rate": 3.891839851752119e-06,
+      "loss": 0.8794,
+      "step": 153090
+    },
+    {
+      "epoch": 1.1082397735745257,
+      "grad_norm": 0.152713343501091,
+      "learning_rate": 3.8917674650915335e-06,
+      "loss": 0.8605,
+      "step": 153100
+    },
+    {
+      "epoch": 1.1083121602351118,
+      "grad_norm": 0.14870685338974,
+      "learning_rate": 3.891695078430947e-06,
+      "loss": 0.8602,
+      "step": 153110
+    },
+    {
+      "epoch": 1.108384546895698,
+      "grad_norm": 0.15383371710777283,
+      "learning_rate": 3.891622691770361e-06,
+      "loss": 0.8538,
+      "step": 153120
+    },
+    {
+      "epoch": 1.1084569335562842,
+      "grad_norm": 0.1520891636610031,
+      "learning_rate": 3.891550305109774e-06,
+      "loss": 0.8539,
+      "step": 153130
+    },
+    {
+      "epoch": 1.1085293202168705,
+      "grad_norm": 0.15309153497219086,
+      "learning_rate": 3.891477918449189e-06,
+      "loss": 0.8459,
+      "step": 153140
+    },
+    {
+      "epoch": 1.1086017068774567,
+      "grad_norm": 0.16597382724285126,
+      "learning_rate": 3.8914055317886025e-06,
+      "loss": 0.8529,
+      "step": 153150
+    },
+    {
+      "epoch": 1.1086740935380428,
+      "grad_norm": 0.1510355919599533,
+      "learning_rate": 3.891333145128016e-06,
+      "loss": 0.8605,
+      "step": 153160
+    },
+    {
+      "epoch": 1.108746480198629,
+      "grad_norm": 0.14412082731723785,
+      "learning_rate": 3.89126075846743e-06,
+      "loss": 0.8532,
+      "step": 153170
+    },
+    {
+      "epoch": 1.1088188668592152,
+      "grad_norm": 0.1505080908536911,
+      "learning_rate": 3.891188371806844e-06,
+      "loss": 0.8613,
+      "step": 153180
+    },
+    {
+      "epoch": 1.1088912535198014,
+      "grad_norm": 0.14691261947155,
+      "learning_rate": 3.891115985146258e-06,
+      "loss": 0.8596,
+      "step": 153190
+    },
+    {
+      "epoch": 1.1089636401803875,
+      "grad_norm": 0.15509304404258728,
+      "learning_rate": 3.891043598485671e-06,
+      "loss": 0.8393,
+      "step": 153200
+    },
+    {
+      "epoch": 1.1090360268409738,
+      "grad_norm": 0.14733615517616272,
+      "learning_rate": 3.890971211825085e-06,
+      "loss": 0.8485,
+      "step": 153210
+    },
+    {
+      "epoch": 1.1091084135015599,
+      "grad_norm": 0.1337285041809082,
+      "learning_rate": 3.8908988251644995e-06,
+      "loss": 0.8556,
+      "step": 153220
+    },
+    {
+      "epoch": 1.1091808001621462,
+      "grad_norm": 0.14776895940303802,
+      "learning_rate": 3.890826438503913e-06,
+      "loss": 0.8796,
+      "step": 153230
+    },
+    {
+      "epoch": 1.1092531868227322,
+      "grad_norm": 0.1420595645904541,
+      "learning_rate": 3.890754051843327e-06,
+      "loss": 0.8565,
+      "step": 153240
+    },
+    {
+      "epoch": 1.1093255734833185,
+      "grad_norm": 0.16995348036289215,
+      "learning_rate": 3.89068166518274e-06,
+      "loss": 0.8469,
+      "step": 153250
+    },
+    {
+      "epoch": 1.1093979601439048,
+      "grad_norm": 0.14591839909553528,
+      "learning_rate": 3.890609278522155e-06,
+      "loss": 0.8588,
+      "step": 153260
+    },
+    {
+      "epoch": 1.1094703468044909,
+      "grad_norm": 0.15660324692726135,
+      "learning_rate": 3.890536891861568e-06,
+      "loss": 0.8571,
+      "step": 153270
+    },
+    {
+      "epoch": 1.1095427334650771,
+      "grad_norm": 0.16159191727638245,
+      "learning_rate": 3.890464505200982e-06,
+      "loss": 0.8689,
+      "step": 153280
+    },
+    {
+      "epoch": 1.1096151201256632,
+      "grad_norm": 0.15314868092536926,
+      "learning_rate": 3.890392118540396e-06,
+      "loss": 0.8534,
+      "step": 153290
+    },
+    {
+      "epoch": 1.1096875067862495,
+      "grad_norm": 0.1325388103723526,
+      "learning_rate": 3.89031973187981e-06,
+      "loss": 0.8674,
+      "step": 153300
+    },
+    {
+      "epoch": 1.1097598934468356,
+      "grad_norm": 0.15876524150371552,
+      "learning_rate": 3.890247345219224e-06,
+      "loss": 0.859,
+      "step": 153310
+    },
+    {
+      "epoch": 1.1098322801074219,
+      "grad_norm": 0.14967800676822662,
+      "learning_rate": 3.890174958558637e-06,
+      "loss": 0.8601,
+      "step": 153320
+    },
+    {
+      "epoch": 1.109904666768008,
+      "grad_norm": 0.14991550147533417,
+      "learning_rate": 3.890102571898051e-06,
+      "loss": 0.8454,
+      "step": 153330
+    },
+    {
+      "epoch": 1.1099770534285942,
+      "grad_norm": 0.15167036652565002,
+      "learning_rate": 3.890030185237465e-06,
+      "loss": 0.8558,
+      "step": 153340
+    },
+    {
+      "epoch": 1.1100494400891803,
+      "grad_norm": 0.15944884717464447,
+      "learning_rate": 3.889957798576879e-06,
+      "loss": 0.8634,
+      "step": 153350
+    },
+    {
+      "epoch": 1.1101218267497666,
+      "grad_norm": 0.14728944003582,
+      "learning_rate": 3.889885411916293e-06,
+      "loss": 0.8759,
+      "step": 153360
+    },
+    {
+      "epoch": 1.1101942134103528,
+      "grad_norm": 0.16900379955768585,
+      "learning_rate": 3.889813025255706e-06,
+      "loss": 0.8672,
+      "step": 153370
+    },
+    {
+      "epoch": 1.110266600070939,
+      "grad_norm": 0.14968164265155792,
+      "learning_rate": 3.88974063859512e-06,
+      "loss": 0.8545,
+      "step": 153380
+    },
+    {
+      "epoch": 1.1103389867315252,
+      "grad_norm": 0.16399233043193817,
+      "learning_rate": 3.8896682519345335e-06,
+      "loss": 0.864,
+      "step": 153390
+    },
+    {
+      "epoch": 1.1104113733921113,
+      "grad_norm": 0.1753818243741989,
+      "learning_rate": 3.889595865273947e-06,
+      "loss": 0.851,
+      "step": 153400
+    },
+    {
+      "epoch": 1.1104837600526976,
+      "grad_norm": 0.18223677575588226,
+      "learning_rate": 3.8895234786133616e-06,
+      "loss": 0.8468,
+      "step": 153410
+    },
+    {
+      "epoch": 1.1105561467132836,
+      "grad_norm": 0.1732460856437683,
+      "learning_rate": 3.889451091952775e-06,
+      "loss": 0.8565,
+      "step": 153420
+    },
+    {
+      "epoch": 1.11062853337387,
+      "grad_norm": 0.15468300879001617,
+      "learning_rate": 3.889378705292189e-06,
+      "loss": 0.8635,
+      "step": 153430
+    },
+    {
+      "epoch": 1.110700920034456,
+      "grad_norm": 0.15871159732341766,
+      "learning_rate": 3.889306318631602e-06,
+      "loss": 0.8611,
+      "step": 153440
+    },
+    {
+      "epoch": 1.1107733066950423,
+      "grad_norm": 0.14641880989074707,
+      "learning_rate": 3.889233931971017e-06,
+      "loss": 0.8447,
+      "step": 153450
+    },
+    {
+      "epoch": 1.1108456933556283,
+      "grad_norm": 0.1457509547472,
+      "learning_rate": 3.8891615453104305e-06,
+      "loss": 0.8662,
+      "step": 153460
+    },
+    {
+      "epoch": 1.1109180800162146,
+      "grad_norm": 0.1556580513715744,
+      "learning_rate": 3.889089158649844e-06,
+      "loss": 0.8524,
+      "step": 153470
+    },
+    {
+      "epoch": 1.110990466676801,
+      "grad_norm": 0.14962519705295563,
+      "learning_rate": 3.889016771989258e-06,
+      "loss": 0.8592,
+      "step": 153480
+    },
+    {
+      "epoch": 1.111062853337387,
+      "grad_norm": 0.1524748057126999,
+      "learning_rate": 3.888944385328672e-06,
+      "loss": 0.8583,
+      "step": 153490
+    },
+    {
+      "epoch": 1.1111352399979733,
+      "grad_norm": 0.17120623588562012,
+      "learning_rate": 3.888871998668086e-06,
+      "loss": 0.862,
+      "step": 153500
+    },
+    {
+      "epoch": 1.1112076266585593,
+      "grad_norm": 0.14351938664913177,
+      "learning_rate": 3.8887996120074994e-06,
+      "loss": 0.8434,
+      "step": 153510
+    },
+    {
+      "epoch": 1.1112800133191456,
+      "grad_norm": 0.14822779595851898,
+      "learning_rate": 3.888727225346913e-06,
+      "loss": 0.8371,
+      "step": 153520
+    },
+    {
+      "epoch": 1.1113523999797317,
+      "grad_norm": 0.20761913061141968,
+      "learning_rate": 3.8886548386863275e-06,
+      "loss": 0.8579,
+      "step": 153530
+    },
+    {
+      "epoch": 1.111424786640318,
+      "grad_norm": 0.15165117383003235,
+      "learning_rate": 3.888582452025741e-06,
+      "loss": 0.8596,
+      "step": 153540
+    },
+    {
+      "epoch": 1.111497173300904,
+      "grad_norm": 0.15417353808879852,
+      "learning_rate": 3.888510065365155e-06,
+      "loss": 0.8515,
+      "step": 153550
+    },
+    {
+      "epoch": 1.1115695599614903,
+      "grad_norm": 0.17663225531578064,
+      "learning_rate": 3.888437678704568e-06,
+      "loss": 0.8599,
+      "step": 153560
+    },
+    {
+      "epoch": 1.1116419466220764,
+      "grad_norm": 0.17057162523269653,
+      "learning_rate": 3.888365292043982e-06,
+      "loss": 0.8655,
+      "step": 153570
+    },
+    {
+      "epoch": 1.1117143332826627,
+      "grad_norm": 0.15689362585544586,
+      "learning_rate": 3.8882929053833964e-06,
+      "loss": 0.8619,
+      "step": 153580
+    },
+    {
+      "epoch": 1.111786719943249,
+      "grad_norm": 0.15272989869117737,
+      "learning_rate": 3.88822051872281e-06,
+      "loss": 0.8598,
+      "step": 153590
+    },
+    {
+      "epoch": 1.111859106603835,
+      "grad_norm": 0.15261130034923553,
+      "learning_rate": 3.888148132062224e-06,
+      "loss": 0.8659,
+      "step": 153600
+    },
+    {
+      "epoch": 1.1119314932644213,
+      "grad_norm": 0.15566353499889374,
+      "learning_rate": 3.888075745401637e-06,
+      "loss": 0.8488,
+      "step": 153610
+    },
+    {
+      "epoch": 1.1120038799250074,
+      "grad_norm": 0.155398890376091,
+      "learning_rate": 3.888003358741052e-06,
+      "loss": 0.8547,
+      "step": 153620
+    },
+    {
+      "epoch": 1.1120762665855937,
+      "grad_norm": 0.14617139101028442,
+      "learning_rate": 3.887930972080465e-06,
+      "loss": 0.8598,
+      "step": 153630
+    },
+    {
+      "epoch": 1.1121486532461797,
+      "grad_norm": 0.1497737020254135,
+      "learning_rate": 3.887858585419879e-06,
+      "loss": 0.8527,
+      "step": 153640
+    },
+    {
+      "epoch": 1.112221039906766,
+      "grad_norm": 0.143030047416687,
+      "learning_rate": 3.887786198759293e-06,
+      "loss": 0.8613,
+      "step": 153650
+    },
+    {
+      "epoch": 1.112293426567352,
+      "grad_norm": 0.15191875398159027,
+      "learning_rate": 3.887713812098707e-06,
+      "loss": 0.8623,
+      "step": 153660
+    },
+    {
+      "epoch": 1.1123658132279384,
+      "grad_norm": 0.14382189512252808,
+      "learning_rate": 3.887641425438121e-06,
+      "loss": 0.8587,
+      "step": 153670
+    },
+    {
+      "epoch": 1.1124381998885244,
+      "grad_norm": 0.15472808480262756,
+      "learning_rate": 3.887569038777534e-06,
+      "loss": 0.8605,
+      "step": 153680
+    },
+    {
+      "epoch": 1.1125105865491107,
+      "grad_norm": 0.15003816783428192,
+      "learning_rate": 3.887496652116948e-06,
+      "loss": 0.8393,
+      "step": 153690
+    },
+    {
+      "epoch": 1.112582973209697,
+      "grad_norm": 0.17156195640563965,
+      "learning_rate": 3.887424265456362e-06,
+      "loss": 0.8712,
+      "step": 153700
+    },
+    {
+      "epoch": 1.112655359870283,
+      "grad_norm": 0.13888807594776154,
+      "learning_rate": 3.887351878795776e-06,
+      "loss": 0.8557,
+      "step": 153710
+    },
+    {
+      "epoch": 1.1127277465308694,
+      "grad_norm": 0.13991951942443848,
+      "learning_rate": 3.88727949213519e-06,
+      "loss": 0.8459,
+      "step": 153720
+    },
+    {
+      "epoch": 1.1128001331914554,
+      "grad_norm": 0.16196198761463165,
+      "learning_rate": 3.887207105474603e-06,
+      "loss": 0.8508,
+      "step": 153730
+    },
+    {
+      "epoch": 1.1128725198520417,
+      "grad_norm": 0.14788612723350525,
+      "learning_rate": 3.887134718814018e-06,
+      "loss": 0.8425,
+      "step": 153740
+    },
+    {
+      "epoch": 1.1129449065126278,
+      "grad_norm": 0.1603209227323532,
+      "learning_rate": 3.887062332153431e-06,
+      "loss": 0.85,
+      "step": 153750
+    },
+    {
+      "epoch": 1.113017293173214,
+      "grad_norm": 0.16001610457897186,
+      "learning_rate": 3.886989945492845e-06,
+      "loss": 0.842,
+      "step": 153760
+    },
+    {
+      "epoch": 1.1130896798338001,
+      "grad_norm": 0.18758459389209747,
+      "learning_rate": 3.8869175588322585e-06,
+      "loss": 0.8503,
+      "step": 153770
+    },
+    {
+      "epoch": 1.1131620664943864,
+      "grad_norm": 0.1522257775068283,
+      "learning_rate": 3.886845172171673e-06,
+      "loss": 0.8627,
+      "step": 153780
+    },
+    {
+      "epoch": 1.1132344531549725,
+      "grad_norm": 0.15141525864601135,
+      "learning_rate": 3.886772785511087e-06,
+      "loss": 0.8585,
+      "step": 153790
+    },
+    {
+      "epoch": 1.1133068398155588,
+      "grad_norm": 0.22884730994701385,
+      "learning_rate": 3.8867003988505e-06,
+      "loss": 0.8604,
+      "step": 153800
+    },
+    {
+      "epoch": 1.113379226476145,
+      "grad_norm": 0.15862050652503967,
+      "learning_rate": 3.886628012189914e-06,
+      "loss": 0.8575,
+      "step": 153810
+    },
+    {
+      "epoch": 1.1134516131367311,
+      "grad_norm": 0.17081834375858307,
+      "learning_rate": 3.886555625529328e-06,
+      "loss": 0.8704,
+      "step": 153820
+    },
+    {
+      "epoch": 1.1135239997973174,
+      "grad_norm": 0.14992080628871918,
+      "learning_rate": 3.886483238868742e-06,
+      "loss": 0.864,
+      "step": 153830
+    },
+    {
+      "epoch": 1.1135963864579035,
+      "grad_norm": 0.16578203439712524,
+      "learning_rate": 3.8864108522081555e-06,
+      "loss": 0.8579,
+      "step": 153840
+    },
+    {
+      "epoch": 1.1136687731184898,
+      "grad_norm": 0.1445264369249344,
+      "learning_rate": 3.886338465547569e-06,
+      "loss": 0.8607,
+      "step": 153850
+    },
+    {
+      "epoch": 1.1137411597790758,
+      "grad_norm": 0.20468395948410034,
+      "learning_rate": 3.886266078886984e-06,
+      "loss": 0.8534,
+      "step": 153860
+    },
+    {
+      "epoch": 1.1138135464396621,
+      "grad_norm": 0.15679140388965607,
+      "learning_rate": 3.886193692226397e-06,
+      "loss": 0.8537,
+      "step": 153870
+    },
+    {
+      "epoch": 1.1138859331002482,
+      "grad_norm": 0.1447470486164093,
+      "learning_rate": 3.886121305565811e-06,
+      "loss": 0.8484,
+      "step": 153880
+    },
+    {
+      "epoch": 1.1139583197608345,
+      "grad_norm": 0.15149566531181335,
+      "learning_rate": 3.8860489189052245e-06,
+      "loss": 0.8667,
+      "step": 153890
+    },
+    {
+      "epoch": 1.1140307064214205,
+      "grad_norm": 0.15107187628746033,
+      "learning_rate": 3.885976532244638e-06,
+      "loss": 0.8702,
+      "step": 153900
+    },
+    {
+      "epoch": 1.1141030930820068,
+      "grad_norm": 0.15003450214862823,
+      "learning_rate": 3.885904145584052e-06,
+      "loss": 0.8469,
+      "step": 153910
+    },
+    {
+      "epoch": 1.1141754797425931,
+      "grad_norm": 0.1784355640411377,
+      "learning_rate": 3.885831758923465e-06,
+      "loss": 0.8632,
+      "step": 153920
+    },
+    {
+      "epoch": 1.1142478664031792,
+      "grad_norm": 0.16152457892894745,
+      "learning_rate": 3.88575937226288e-06,
+      "loss": 0.8586,
+      "step": 153930
+    },
+    {
+      "epoch": 1.1143202530637655,
+      "grad_norm": 0.15195733308792114,
+      "learning_rate": 3.885686985602293e-06,
+      "loss": 0.863,
+      "step": 153940
+    },
+    {
+      "epoch": 1.1143926397243515,
+      "grad_norm": 0.14799615740776062,
+      "learning_rate": 3.885614598941707e-06,
+      "loss": 0.8429,
+      "step": 153950
+    },
+    {
+      "epoch": 1.1144650263849378,
+      "grad_norm": 0.15094245970249176,
+      "learning_rate": 3.885542212281121e-06,
+      "loss": 0.8488,
+      "step": 153960
+    },
+    {
+      "epoch": 1.114537413045524,
+      "grad_norm": 0.15210039913654327,
+      "learning_rate": 3.885469825620535e-06,
+      "loss": 0.8584,
+      "step": 153970
+    },
+    {
+      "epoch": 1.1146097997061102,
+      "grad_norm": 0.16268359124660492,
+      "learning_rate": 3.885397438959949e-06,
+      "loss": 0.8587,
+      "step": 153980
+    },
+    {
+      "epoch": 1.1146821863666962,
+      "grad_norm": 0.15050064027309418,
+      "learning_rate": 3.885325052299362e-06,
+      "loss": 0.8557,
+      "step": 153990
+    },
+    {
+      "epoch": 1.1147545730272825,
+      "grad_norm": 0.15507516264915466,
+      "learning_rate": 3.885252665638776e-06,
+      "loss": 0.867,
+      "step": 154000
+    },
+    {
+      "epoch": 1.1148269596878686,
+      "grad_norm": 0.2021208256483078,
+      "learning_rate": 3.88518027897819e-06,
+      "loss": 0.8476,
+      "step": 154010
+    },
+    {
+      "epoch": 1.114899346348455,
+      "grad_norm": 0.15524426102638245,
+      "learning_rate": 3.885107892317604e-06,
+      "loss": 0.8567,
+      "step": 154020
+    },
+    {
+      "epoch": 1.1149717330090412,
+      "grad_norm": 0.14437155425548553,
+      "learning_rate": 3.885035505657018e-06,
+      "loss": 0.8575,
+      "step": 154030
+    },
+    {
+      "epoch": 1.1150441196696272,
+      "grad_norm": 0.15939441323280334,
+      "learning_rate": 3.884963118996431e-06,
+      "loss": 0.8591,
+      "step": 154040
+    },
+    {
+      "epoch": 1.1151165063302135,
+      "grad_norm": 0.15175363421440125,
+      "learning_rate": 3.884890732335846e-06,
+      "loss": 0.866,
+      "step": 154050
+    },
+    {
+      "epoch": 1.1151888929907996,
+      "grad_norm": 0.14421996474266052,
+      "learning_rate": 3.884818345675259e-06,
+      "loss": 0.8485,
+      "step": 154060
+    },
+    {
+      "epoch": 1.1152612796513859,
+      "grad_norm": 0.14754317700862885,
+      "learning_rate": 3.884745959014673e-06,
+      "loss": 0.8681,
+      "step": 154070
+    },
+    {
+      "epoch": 1.115333666311972,
+      "grad_norm": 0.16109803318977356,
+      "learning_rate": 3.8846735723540866e-06,
+      "loss": 0.8769,
+      "step": 154080
+    },
+    {
+      "epoch": 1.1154060529725582,
+      "grad_norm": 0.16321797668933868,
+      "learning_rate": 3.884601185693501e-06,
+      "loss": 0.8612,
+      "step": 154090
+    },
+    {
+      "epoch": 1.1154784396331443,
+      "grad_norm": 0.15551266074180603,
+      "learning_rate": 3.884528799032915e-06,
+      "loss": 0.865,
+      "step": 154100
+    },
+    {
+      "epoch": 1.1155508262937306,
+      "grad_norm": 0.15028153359889984,
+      "learning_rate": 3.884456412372328e-06,
+      "loss": 0.8584,
+      "step": 154110
+    },
+    {
+      "epoch": 1.1156232129543169,
+      "grad_norm": 0.14665335416793823,
+      "learning_rate": 3.884384025711742e-06,
+      "loss": 0.8349,
+      "step": 154120
+    },
+    {
+      "epoch": 1.115695599614903,
+      "grad_norm": 0.1615133285522461,
+      "learning_rate": 3.884311639051156e-06,
+      "loss": 0.8697,
+      "step": 154130
+    },
+    {
+      "epoch": 1.1157679862754892,
+      "grad_norm": 0.14545638859272003,
+      "learning_rate": 3.88423925239057e-06,
+      "loss": 0.8582,
+      "step": 154140
+    },
+    {
+      "epoch": 1.1158403729360753,
+      "grad_norm": 0.18917188048362732,
+      "learning_rate": 3.8841668657299836e-06,
+      "loss": 0.8593,
+      "step": 154150
+    },
+    {
+      "epoch": 1.1159127595966616,
+      "grad_norm": 0.1670539379119873,
+      "learning_rate": 3.884094479069397e-06,
+      "loss": 0.8695,
+      "step": 154160
+    },
+    {
+      "epoch": 1.1159851462572477,
+      "grad_norm": 0.1593600958585739,
+      "learning_rate": 3.884022092408812e-06,
+      "loss": 0.8738,
+      "step": 154170
+    },
+    {
+      "epoch": 1.116057532917834,
+      "grad_norm": 0.15494653582572937,
+      "learning_rate": 3.883949705748225e-06,
+      "loss": 0.8465,
+      "step": 154180
+    },
+    {
+      "epoch": 1.11612991957842,
+      "grad_norm": 0.15124863386154175,
+      "learning_rate": 3.883877319087639e-06,
+      "loss": 0.8538,
+      "step": 154190
+    },
+    {
+      "epoch": 1.1162023062390063,
+      "grad_norm": 0.1503598988056183,
+      "learning_rate": 3.8838049324270525e-06,
+      "loss": 0.8617,
+      "step": 154200
+    },
+    {
+      "epoch": 1.1162746928995926,
+      "grad_norm": 0.1482527107000351,
+      "learning_rate": 3.883732545766466e-06,
+      "loss": 0.8542,
+      "step": 154210
+    },
+    {
+      "epoch": 1.1163470795601786,
+      "grad_norm": 0.1634032428264618,
+      "learning_rate": 3.883660159105881e-06,
+      "loss": 0.8605,
+      "step": 154220
+    },
+    {
+      "epoch": 1.116419466220765,
+      "grad_norm": 0.14771081507205963,
+      "learning_rate": 3.883587772445294e-06,
+      "loss": 0.8564,
+      "step": 154230
+    },
+    {
+      "epoch": 1.116491852881351,
+      "grad_norm": 0.16932451725006104,
+      "learning_rate": 3.883515385784708e-06,
+      "loss": 0.8578,
+      "step": 154240
+    },
+    {
+      "epoch": 1.1165642395419373,
+      "grad_norm": 0.1720747947692871,
+      "learning_rate": 3.8834429991241214e-06,
+      "loss": 0.8435,
+      "step": 154250
+    },
+    {
+      "epoch": 1.1166366262025234,
+      "grad_norm": 0.15350791811943054,
+      "learning_rate": 3.883370612463536e-06,
+      "loss": 0.8628,
+      "step": 154260
+    },
+    {
+      "epoch": 1.1167090128631096,
+      "grad_norm": 0.162908136844635,
+      "learning_rate": 3.8832982258029495e-06,
+      "loss": 0.8472,
+      "step": 154270
+    },
+    {
+      "epoch": 1.1167813995236957,
+      "grad_norm": 0.1558998078107834,
+      "learning_rate": 3.883225839142363e-06,
+      "loss": 0.8482,
+      "step": 154280
+    },
+    {
+      "epoch": 1.116853786184282,
+      "grad_norm": 0.15881863236427307,
+      "learning_rate": 3.883153452481777e-06,
+      "loss": 0.8572,
+      "step": 154290
+    },
+    {
+      "epoch": 1.116926172844868,
+      "grad_norm": 0.15854284167289734,
+      "learning_rate": 3.883081065821191e-06,
+      "loss": 0.873,
+      "step": 154300
+    },
+    {
+      "epoch": 1.1169985595054543,
+      "grad_norm": 0.29879847168922424,
+      "learning_rate": 3.883008679160605e-06,
+      "loss": 0.8582,
+      "step": 154310
+    },
+    {
+      "epoch": 1.1170709461660406,
+      "grad_norm": 0.15802979469299316,
+      "learning_rate": 3.8829362925000184e-06,
+      "loss": 0.8642,
+      "step": 154320
+    },
+    {
+      "epoch": 1.1171433328266267,
+      "grad_norm": 0.1594230681657791,
+      "learning_rate": 3.882863905839432e-06,
+      "loss": 0.8585,
+      "step": 154330
+    },
+    {
+      "epoch": 1.117215719487213,
+      "grad_norm": 0.1390036940574646,
+      "learning_rate": 3.8827915191788465e-06,
+      "loss": 0.8615,
+      "step": 154340
+    },
+    {
+      "epoch": 1.117288106147799,
+      "grad_norm": 0.1467415988445282,
+      "learning_rate": 3.88271913251826e-06,
+      "loss": 0.8429,
+      "step": 154350
+    },
+    {
+      "epoch": 1.1173604928083853,
+      "grad_norm": 0.1545180082321167,
+      "learning_rate": 3.882646745857674e-06,
+      "loss": 0.857,
+      "step": 154360
+    },
+    {
+      "epoch": 1.1174328794689714,
+      "grad_norm": 0.1605067402124405,
+      "learning_rate": 3.882574359197087e-06,
+      "loss": 0.8628,
+      "step": 154370
+    },
+    {
+      "epoch": 1.1175052661295577,
+      "grad_norm": 0.1610914170742035,
+      "learning_rate": 3.882501972536502e-06,
+      "loss": 0.8567,
+      "step": 154380
+    },
+    {
+      "epoch": 1.1175776527901438,
+      "grad_norm": 0.19544139504432678,
+      "learning_rate": 3.8824295858759154e-06,
+      "loss": 0.851,
+      "step": 154390
+    },
+    {
+      "epoch": 1.11765003945073,
+      "grad_norm": 0.16312572360038757,
+      "learning_rate": 3.882357199215329e-06,
+      "loss": 0.8662,
+      "step": 154400
+    },
+    {
+      "epoch": 1.1177224261113161,
+      "grad_norm": 0.15379171073436737,
+      "learning_rate": 3.882284812554743e-06,
+      "loss": 0.8593,
+      "step": 154410
+    },
+    {
+      "epoch": 1.1177948127719024,
+      "grad_norm": 0.1567648947238922,
+      "learning_rate": 3.882212425894157e-06,
+      "loss": 0.8636,
+      "step": 154420
+    },
+    {
+      "epoch": 1.1178671994324887,
+      "grad_norm": 0.1432834267616272,
+      "learning_rate": 3.88214003923357e-06,
+      "loss": 0.8443,
+      "step": 154430
+    },
+    {
+      "epoch": 1.1179395860930748,
+      "grad_norm": 0.1487468183040619,
+      "learning_rate": 3.8820676525729835e-06,
+      "loss": 0.8509,
+      "step": 154440
+    },
+    {
+      "epoch": 1.118011972753661,
+      "grad_norm": 0.14194174110889435,
+      "learning_rate": 3.881995265912398e-06,
+      "loss": 0.8439,
+      "step": 154450
+    },
+    {
+      "epoch": 1.118084359414247,
+      "grad_norm": 0.15703342854976654,
+      "learning_rate": 3.881922879251812e-06,
+      "loss": 0.8675,
+      "step": 154460
+    },
+    {
+      "epoch": 1.1181567460748334,
+      "grad_norm": 0.15480118989944458,
+      "learning_rate": 3.881850492591225e-06,
+      "loss": 0.8605,
+      "step": 154470
+    },
+    {
+      "epoch": 1.1182291327354195,
+      "grad_norm": 0.14932669699192047,
+      "learning_rate": 3.881778105930639e-06,
+      "loss": 0.8424,
+      "step": 154480
+    },
+    {
+      "epoch": 1.1183015193960057,
+      "grad_norm": 0.16219106316566467,
+      "learning_rate": 3.881705719270053e-06,
+      "loss": 0.8539,
+      "step": 154490
+    },
+    {
+      "epoch": 1.1183739060565918,
+      "grad_norm": 0.1499941051006317,
+      "learning_rate": 3.881633332609467e-06,
+      "loss": 0.8615,
+      "step": 154500
+    },
+    {
+      "epoch": 1.118446292717178,
+      "grad_norm": 0.16468805074691772,
+      "learning_rate": 3.8815609459488805e-06,
+      "loss": 0.8536,
+      "step": 154510
+    },
+    {
+      "epoch": 1.1185186793777642,
+      "grad_norm": 0.1566213071346283,
+      "learning_rate": 3.881488559288294e-06,
+      "loss": 0.8603,
+      "step": 154520
+    },
+    {
+      "epoch": 1.1185910660383505,
+      "grad_norm": 0.14781454205513,
+      "learning_rate": 3.881416172627709e-06,
+      "loss": 0.8494,
+      "step": 154530
+    },
+    {
+      "epoch": 1.1186634526989367,
+      "grad_norm": 0.1491243541240692,
+      "learning_rate": 3.881343785967122e-06,
+      "loss": 0.8571,
+      "step": 154540
+    },
+    {
+      "epoch": 1.1187358393595228,
+      "grad_norm": 0.15416626632213593,
+      "learning_rate": 3.881271399306536e-06,
+      "loss": 0.8502,
+      "step": 154550
+    },
+    {
+      "epoch": 1.118808226020109,
+      "grad_norm": 0.1574944406747818,
+      "learning_rate": 3.8811990126459495e-06,
+      "loss": 0.8377,
+      "step": 154560
+    },
+    {
+      "epoch": 1.1188806126806952,
+      "grad_norm": 0.15383422374725342,
+      "learning_rate": 3.881126625985364e-06,
+      "loss": 0.8587,
+      "step": 154570
+    },
+    {
+      "epoch": 1.1189529993412815,
+      "grad_norm": 0.15116389095783234,
+      "learning_rate": 3.8810542393247775e-06,
+      "loss": 0.8461,
+      "step": 154580
+    },
+    {
+      "epoch": 1.1190253860018675,
+      "grad_norm": 0.15596124529838562,
+      "learning_rate": 3.880981852664191e-06,
+      "loss": 0.8434,
+      "step": 154590
+    },
+    {
+      "epoch": 1.1190977726624538,
+      "grad_norm": 0.15235568583011627,
+      "learning_rate": 3.880909466003605e-06,
+      "loss": 0.8597,
+      "step": 154600
+    },
+    {
+      "epoch": 1.1191701593230399,
+      "grad_norm": 0.1518736332654953,
+      "learning_rate": 3.880837079343019e-06,
+      "loss": 0.8583,
+      "step": 154610
+    },
+    {
+      "epoch": 1.1192425459836262,
+      "grad_norm": 0.14196760952472687,
+      "learning_rate": 3.880764692682433e-06,
+      "loss": 0.8663,
+      "step": 154620
+    },
+    {
+      "epoch": 1.1193149326442122,
+      "grad_norm": 0.15060287714004517,
+      "learning_rate": 3.8806923060218465e-06,
+      "loss": 0.8448,
+      "step": 154630
+    },
+    {
+      "epoch": 1.1193873193047985,
+      "grad_norm": 0.20425279438495636,
+      "learning_rate": 3.88061991936126e-06,
+      "loss": 0.8628,
+      "step": 154640
+    },
+    {
+      "epoch": 1.1194597059653848,
+      "grad_norm": 0.1478336602449417,
+      "learning_rate": 3.8805475327006745e-06,
+      "loss": 0.8626,
+      "step": 154650
+    },
+    {
+      "epoch": 1.1195320926259709,
+      "grad_norm": 0.14704637229442596,
+      "learning_rate": 3.880475146040088e-06,
+      "loss": 0.8552,
+      "step": 154660
+    },
+    {
+      "epoch": 1.1196044792865572,
+      "grad_norm": 0.2208547443151474,
+      "learning_rate": 3.880402759379502e-06,
+      "loss": 0.8501,
+      "step": 154670
+    },
+    {
+      "epoch": 1.1196768659471432,
+      "grad_norm": 0.15612949430942535,
+      "learning_rate": 3.880330372718915e-06,
+      "loss": 0.8516,
+      "step": 154680
+    },
+    {
+      "epoch": 1.1197492526077295,
+      "grad_norm": 0.15186789631843567,
+      "learning_rate": 3.88025798605833e-06,
+      "loss": 0.8542,
+      "step": 154690
+    },
+    {
+      "epoch": 1.1198216392683156,
+      "grad_norm": 0.16197703778743744,
+      "learning_rate": 3.8801855993977435e-06,
+      "loss": 0.8628,
+      "step": 154700
+    },
+    {
+      "epoch": 1.1198940259289019,
+      "grad_norm": 0.1460353136062622,
+      "learning_rate": 3.880113212737157e-06,
+      "loss": 0.871,
+      "step": 154710
+    },
+    {
+      "epoch": 1.119966412589488,
+      "grad_norm": 0.1506577581167221,
+      "learning_rate": 3.880040826076571e-06,
+      "loss": 0.8598,
+      "step": 154720
+    },
+    {
+      "epoch": 1.1200387992500742,
+      "grad_norm": 0.16504830121994019,
+      "learning_rate": 3.879968439415985e-06,
+      "loss": 0.8634,
+      "step": 154730
+    },
+    {
+      "epoch": 1.1201111859106603,
+      "grad_norm": 0.15591880679130554,
+      "learning_rate": 3.879896052755399e-06,
+      "loss": 0.8572,
+      "step": 154740
+    },
+    {
+      "epoch": 1.1201835725712466,
+      "grad_norm": 0.14666037261486053,
+      "learning_rate": 3.879823666094812e-06,
+      "loss": 0.8661,
+      "step": 154750
+    },
+    {
+      "epoch": 1.1202559592318329,
+      "grad_norm": 0.14609721302986145,
+      "learning_rate": 3.879751279434226e-06,
+      "loss": 0.8442,
+      "step": 154760
+    },
+    {
+      "epoch": 1.120328345892419,
+      "grad_norm": 0.15813161432743073,
+      "learning_rate": 3.8796788927736405e-06,
+      "loss": 0.8426,
+      "step": 154770
+    },
+    {
+      "epoch": 1.1204007325530052,
+      "grad_norm": 0.15691469609737396,
+      "learning_rate": 3.879606506113054e-06,
+      "loss": 0.8623,
+      "step": 154780
+    },
+    {
+      "epoch": 1.1204731192135913,
+      "grad_norm": 0.14903950691223145,
+      "learning_rate": 3.879534119452468e-06,
+      "loss": 0.8414,
+      "step": 154790
+    },
+    {
+      "epoch": 1.1205455058741776,
+      "grad_norm": 0.1798841953277588,
+      "learning_rate": 3.879461732791881e-06,
+      "loss": 0.8486,
+      "step": 154800
+    },
+    {
+      "epoch": 1.1206178925347636,
+      "grad_norm": 0.1467403918504715,
+      "learning_rate": 3.879389346131295e-06,
+      "loss": 0.8564,
+      "step": 154810
+    },
+    {
+      "epoch": 1.12069027919535,
+      "grad_norm": 0.14768868684768677,
+      "learning_rate": 3.879316959470709e-06,
+      "loss": 0.8532,
+      "step": 154820
+    },
+    {
+      "epoch": 1.120762665855936,
+      "grad_norm": 0.15052692592144012,
+      "learning_rate": 3.879244572810123e-06,
+      "loss": 0.8576,
+      "step": 154830
+    },
+    {
+      "epoch": 1.1208350525165223,
+      "grad_norm": 0.15779510140419006,
+      "learning_rate": 3.879172186149537e-06,
+      "loss": 0.8538,
+      "step": 154840
+    },
+    {
+      "epoch": 1.1209074391771083,
+      "grad_norm": 0.16950878500938416,
+      "learning_rate": 3.87909979948895e-06,
+      "loss": 0.852,
+      "step": 154850
+    },
+    {
+      "epoch": 1.1209798258376946,
+      "grad_norm": 0.16230851411819458,
+      "learning_rate": 3.879027412828365e-06,
+      "loss": 0.8514,
+      "step": 154860
+    },
+    {
+      "epoch": 1.121052212498281,
+      "grad_norm": 0.16090090572834015,
+      "learning_rate": 3.878955026167778e-06,
+      "loss": 0.8537,
+      "step": 154870
+    },
+    {
+      "epoch": 1.121124599158867,
+      "grad_norm": 0.15824168920516968,
+      "learning_rate": 3.878882639507192e-06,
+      "loss": 0.8446,
+      "step": 154880
+    },
+    {
+      "epoch": 1.1211969858194533,
+      "grad_norm": 0.145647332072258,
+      "learning_rate": 3.8788102528466056e-06,
+      "loss": 0.854,
+      "step": 154890
+    },
+    {
+      "epoch": 1.1212693724800393,
+      "grad_norm": 0.15013916790485382,
+      "learning_rate": 3.87873786618602e-06,
+      "loss": 0.8671,
+      "step": 154900
+    },
+    {
+      "epoch": 1.1213417591406256,
+      "grad_norm": 0.16897442936897278,
+      "learning_rate": 3.878665479525434e-06,
+      "loss": 0.8543,
+      "step": 154910
+    },
+    {
+      "epoch": 1.1214141458012117,
+      "grad_norm": 0.14996156096458435,
+      "learning_rate": 3.878593092864847e-06,
+      "loss": 0.8535,
+      "step": 154920
+    },
+    {
+      "epoch": 1.121486532461798,
+      "grad_norm": 0.16847410798072815,
+      "learning_rate": 3.878520706204261e-06,
+      "loss": 0.8508,
+      "step": 154930
+    },
+    {
+      "epoch": 1.121558919122384,
+      "grad_norm": 0.1546623855829239,
+      "learning_rate": 3.878448319543675e-06,
+      "loss": 0.8584,
+      "step": 154940
+    },
+    {
+      "epoch": 1.1216313057829703,
+      "grad_norm": 0.1652345508337021,
+      "learning_rate": 3.878375932883089e-06,
+      "loss": 0.8633,
+      "step": 154950
+    },
+    {
+      "epoch": 1.1217036924435564,
+      "grad_norm": 0.20880626142024994,
+      "learning_rate": 3.878303546222503e-06,
+      "loss": 0.8632,
+      "step": 154960
+    },
+    {
+      "epoch": 1.1217760791041427,
+      "grad_norm": 0.16059084236621857,
+      "learning_rate": 3.878231159561916e-06,
+      "loss": 0.8494,
+      "step": 154970
+    },
+    {
+      "epoch": 1.121848465764729,
+      "grad_norm": 0.15037423372268677,
+      "learning_rate": 3.87815877290133e-06,
+      "loss": 0.8516,
+      "step": 154980
+    },
+    {
+      "epoch": 1.121920852425315,
+      "grad_norm": 0.14622123539447784,
+      "learning_rate": 3.8780863862407434e-06,
+      "loss": 0.8621,
+      "step": 154990
+    },
+    {
+      "epoch": 1.1219932390859013,
+      "grad_norm": 0.1608671396970749,
+      "learning_rate": 3.878013999580157e-06,
+      "loss": 0.8603,
+      "step": 155000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 690735,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.131881219057621e+19,
+  "train_batch_size": 80,
+  "trial_name": null,
+  "trial_params": null
+}