{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 30324,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00065955447095487,
      "grad_norm": 2.413745403289795,
      "learning_rate": 1.4836795252225522e-07,
      "loss": 2.2717,
      "step": 10
    },
    {
      "epoch": 0.00131910894190974,
      "grad_norm": 2.212498426437378,
      "learning_rate": 3.1322123310253876e-07,
      "loss": 2.2465,
      "step": 20
    },
    {
      "epoch": 0.00197866341286461,
      "grad_norm": 2.1338729858398438,
      "learning_rate": 4.780745136828223e-07,
      "loss": 2.2954,
      "step": 30
    },
    {
      "epoch": 0.00263821788381948,
      "grad_norm": 1.911232352256775,
      "learning_rate": 6.429277942631058e-07,
      "loss": 2.2998,
      "step": 40
    },
    {
      "epoch": 0.00329777235477435,
      "grad_norm": 2.3705928325653076,
      "learning_rate": 8.077810748433894e-07,
      "loss": 2.2484,
      "step": 50
    },
    {
      "epoch": 0.00395732682572922,
      "grad_norm": 2.46266508102417,
      "learning_rate": 9.72634355423673e-07,
      "loss": 2.2459,
      "step": 60
    },
    {
      "epoch": 0.00461688129668409,
      "grad_norm": 2.2639856338500977,
      "learning_rate": 1.1374876360039564e-06,
      "loss": 2.1686,
      "step": 70
    },
    {
      "epoch": 0.00527643576763896,
      "grad_norm": 2.0852646827697754,
      "learning_rate": 1.3023409165842401e-06,
      "loss": 2.1975,
      "step": 80
    },
    {
      "epoch": 0.00593599023859383,
      "grad_norm": 3.11877179145813,
      "learning_rate": 1.4671941971645236e-06,
      "loss": 2.2112,
      "step": 90
    },
    {
      "epoch": 0.0065955447095487,
      "grad_norm": 1.8753658533096313,
      "learning_rate": 1.6320474777448073e-06,
      "loss": 2.2048,
      "step": 100
    },
    {
      "epoch": 0.00725509918050357,
      "grad_norm": 2.552967071533203,
      "learning_rate": 1.796900758325091e-06,
      "loss": 2.1385,
      "step": 110
    },
    {
      "epoch": 0.00791465365145844,
      "grad_norm": 2.5597939491271973,
      "learning_rate": 1.9617540389053744e-06,
      "loss": 2.106,
      "step": 120
    },
    {
      "epoch": 0.00857420812241331,
      "grad_norm": 2.459486722946167,
      "learning_rate": 2.126607319485658e-06,
      "loss": 2.1547,
      "step": 130
    },
    {
      "epoch": 0.00923376259336818,
      "grad_norm": 2.0308799743652344,
      "learning_rate": 2.2914606000659413e-06,
      "loss": 2.0145,
      "step": 140
    },
    {
      "epoch": 0.009893317064323049,
      "grad_norm": 2.0622429847717285,
      "learning_rate": 2.456313880646225e-06,
      "loss": 1.9606,
      "step": 150
    },
    {
      "epoch": 0.01055287153527792,
      "grad_norm": 1.494454026222229,
      "learning_rate": 2.6211671612265087e-06,
      "loss": 1.7742,
      "step": 160
    },
    {
      "epoch": 0.01121242600623279,
      "grad_norm": 1.3403377532958984,
      "learning_rate": 2.786020441806792e-06,
      "loss": 1.7,
      "step": 170
    },
    {
      "epoch": 0.01187198047718766,
      "grad_norm": 1.1371599435806274,
      "learning_rate": 2.9508737223870756e-06,
      "loss": 1.693,
      "step": 180
    },
    {
      "epoch": 0.01253153494814253,
      "grad_norm": 1.0440349578857422,
      "learning_rate": 3.115727002967359e-06,
      "loss": 1.6249,
      "step": 190
    },
    {
      "epoch": 0.0131910894190974,
      "grad_norm": 0.8132500052452087,
      "learning_rate": 3.280580283547643e-06,
      "loss": 1.5336,
      "step": 200
    },
    {
      "epoch": 0.013850643890052269,
      "grad_norm": 0.8266546130180359,
      "learning_rate": 3.4454335641279264e-06,
      "loss": 1.4346,
      "step": 210
    },
    {
      "epoch": 0.01451019836100714,
      "grad_norm": 0.6412438154220581,
      "learning_rate": 3.61028684470821e-06,
      "loss": 1.4265,
      "step": 220
    },
    {
      "epoch": 0.01516975283196201,
      "grad_norm": 0.5342578887939453,
      "learning_rate": 3.7751401252884933e-06,
      "loss": 1.2813,
      "step": 230
    },
    {
      "epoch": 0.01582930730291688,
      "grad_norm": 0.3916451036930084,
      "learning_rate": 3.939993405868777e-06,
      "loss": 1.3416,
      "step": 240
    },
    {
      "epoch": 0.01648886177387175,
      "grad_norm": 0.4304097294807434,
      "learning_rate": 4.104846686449061e-06,
      "loss": 1.3471,
      "step": 250
    },
    {
      "epoch": 0.01714841624482662,
      "grad_norm": 0.46680748462677,
      "learning_rate": 4.269699967029344e-06,
      "loss": 1.2997,
      "step": 260
    },
    {
      "epoch": 0.01780797071578149,
      "grad_norm": 0.37838730216026306,
      "learning_rate": 4.434553247609628e-06,
      "loss": 1.3106,
      "step": 270
    },
    {
      "epoch": 0.01846752518673636,
      "grad_norm": 0.3575062155723572,
      "learning_rate": 4.599406528189911e-06,
      "loss": 1.219,
      "step": 280
    },
    {
      "epoch": 0.01912707965769123,
      "grad_norm": 0.43343842029571533,
      "learning_rate": 4.7642598087701945e-06,
      "loss": 1.2032,
      "step": 290
    },
    {
      "epoch": 0.019786634128646098,
      "grad_norm": 0.4978269636631012,
      "learning_rate": 4.929113089350478e-06,
      "loss": 1.2446,
      "step": 300
    },
    {
      "epoch": 0.02044618859960097,
      "grad_norm": 0.3755980432033539,
      "learning_rate": 5.0939663699307614e-06,
      "loss": 1.3004,
      "step": 310
    },
    {
      "epoch": 0.02110574307055584,
      "grad_norm": 0.5298649668693542,
      "learning_rate": 5.258819650511045e-06,
      "loss": 1.2482,
      "step": 320
    },
    {
      "epoch": 0.02176529754151071,
      "grad_norm": 0.38744017481803894,
      "learning_rate": 5.423672931091329e-06,
      "loss": 1.1939,
      "step": 330
    },
    {
      "epoch": 0.02242485201246558,
      "grad_norm": 0.46404343843460083,
      "learning_rate": 5.588526211671613e-06,
      "loss": 1.1862,
      "step": 340
    },
    {
      "epoch": 0.023084406483420448,
      "grad_norm": 0.46498826146125793,
      "learning_rate": 5.753379492251896e-06,
      "loss": 1.1794,
      "step": 350
    },
    {
      "epoch": 0.02374396095437532,
      "grad_norm": 0.5952163934707642,
      "learning_rate": 5.91823277283218e-06,
      "loss": 1.129,
      "step": 360
    },
    {
      "epoch": 0.02440351542533019,
      "grad_norm": 0.5538243055343628,
      "learning_rate": 6.083086053412463e-06,
      "loss": 1.1399,
      "step": 370
    },
    {
      "epoch": 0.02506306989628506,
      "grad_norm": 0.6970240473747253,
      "learning_rate": 6.2479393339927465e-06,
      "loss": 1.1051,
      "step": 380
    },
    {
      "epoch": 0.02572262436723993,
      "grad_norm": 0.7881635427474976,
      "learning_rate": 6.412792614573031e-06,
      "loss": 1.1201,
      "step": 390
    },
    {
      "epoch": 0.0263821788381948,
      "grad_norm": 0.39730632305145264,
      "learning_rate": 6.577645895153314e-06,
      "loss": 1.1826,
      "step": 400
    },
    {
      "epoch": 0.02704173330914967,
      "grad_norm": 0.5324128866195679,
      "learning_rate": 6.742499175733598e-06,
      "loss": 1.1355,
      "step": 410
    },
    {
      "epoch": 0.027701287780104538,
      "grad_norm": 0.446857213973999,
      "learning_rate": 6.907352456313881e-06,
      "loss": 1.1181,
      "step": 420
    },
    {
      "epoch": 0.02836084225105941,
      "grad_norm": 0.7503374218940735,
      "learning_rate": 7.072205736894165e-06,
      "loss": 1.1523,
      "step": 430
    },
    {
      "epoch": 0.02902039672201428,
      "grad_norm": 0.6200227737426758,
      "learning_rate": 7.237059017474448e-06,
      "loss": 1.0842,
      "step": 440
    },
    {
      "epoch": 0.02967995119296915,
      "grad_norm": 0.5611582398414612,
      "learning_rate": 7.401912298054732e-06,
      "loss": 1.1075,
      "step": 450
    },
    {
      "epoch": 0.03033950566392402,
      "grad_norm": 0.41381916403770447,
      "learning_rate": 7.566765578635016e-06,
      "loss": 1.075,
      "step": 460
    },
    {
      "epoch": 0.030999060134878888,
      "grad_norm": 0.657398521900177,
      "learning_rate": 7.731618859215299e-06,
      "loss": 1.1104,
      "step": 470
    },
    {
      "epoch": 0.03165861460583376,
      "grad_norm": 0.9275014400482178,
      "learning_rate": 7.896472139795583e-06,
      "loss": 1.1215,
      "step": 480
    },
    {
      "epoch": 0.03231816907678863,
      "grad_norm": 0.6613135933876038,
      "learning_rate": 8.061325420375865e-06,
      "loss": 1.1092,
      "step": 490
    },
    {
      "epoch": 0.0329777235477435,
      "grad_norm": 0.7839298248291016,
      "learning_rate": 8.22617870095615e-06,
      "loss": 1.0522,
      "step": 500
    },
    {
      "epoch": 0.03363727801869837,
      "grad_norm": 0.6185851693153381,
      "learning_rate": 8.391031981536434e-06,
      "loss": 1.074,
      "step": 510
    },
    {
      "epoch": 0.03429683248965324,
      "grad_norm": 0.6220463514328003,
      "learning_rate": 8.555885262116717e-06,
      "loss": 1.0636,
      "step": 520
    },
    {
      "epoch": 0.034956386960608106,
      "grad_norm": 0.6293200254440308,
      "learning_rate": 8.720738542697001e-06,
      "loss": 1.0748,
      "step": 530
    },
    {
      "epoch": 0.03561594143156298,
      "grad_norm": 0.6412210464477539,
      "learning_rate": 8.885591823277284e-06,
      "loss": 1.0866,
      "step": 540
    },
    {
      "epoch": 0.03627549590251785,
      "grad_norm": 0.8587160706520081,
      "learning_rate": 9.050445103857568e-06,
      "loss": 1.0973,
      "step": 550
    },
    {
      "epoch": 0.03693505037347272,
      "grad_norm": 0.6351908445358276,
      "learning_rate": 9.21529838443785e-06,
      "loss": 1.0165,
      "step": 560
    },
    {
      "epoch": 0.03759460484442759,
      "grad_norm": 0.8718788623809814,
      "learning_rate": 9.380151665018135e-06,
      "loss": 1.0535,
      "step": 570
    },
    {
      "epoch": 0.03825415931538246,
      "grad_norm": 0.6819501519203186,
      "learning_rate": 9.545004945598417e-06,
      "loss": 1.0329,
      "step": 580
    },
    {
      "epoch": 0.03891371378633733,
      "grad_norm": 0.6043257117271423,
      "learning_rate": 9.709858226178702e-06,
      "loss": 0.9932,
      "step": 590
    },
    {
      "epoch": 0.039573268257292196,
      "grad_norm": 0.8498331308364868,
      "learning_rate": 9.874711506758984e-06,
      "loss": 1.0686,
      "step": 600
    },
    {
      "epoch": 0.04023282272824707,
      "grad_norm": 0.6236681938171387,
      "learning_rate": 1.0039564787339269e-05,
      "loss": 1.0402,
      "step": 610
    },
    {
      "epoch": 0.04089237719920194,
      "grad_norm": 0.8242946267127991,
      "learning_rate": 1.0204418067919553e-05,
      "loss": 1.0139,
      "step": 620
    },
    {
      "epoch": 0.04155193167015681,
      "grad_norm": 0.7272655367851257,
      "learning_rate": 1.0369271348499836e-05,
      "loss": 1.0754,
      "step": 630
    },
    {
      "epoch": 0.04221148614111168,
      "grad_norm": 0.6365894079208374,
      "learning_rate": 1.053412462908012e-05,
      "loss": 1.0287,
      "step": 640
    },
    {
      "epoch": 0.04287104061206655,
      "grad_norm": 0.5730026960372925,
      "learning_rate": 1.0698977909660403e-05,
      "loss": 1.0192,
      "step": 650
    },
    {
      "epoch": 0.04353059508302142,
      "grad_norm": 0.8422584533691406,
      "learning_rate": 1.0863831190240687e-05,
      "loss": 1.0297,
      "step": 660
    },
    {
      "epoch": 0.044190149553976286,
      "grad_norm": 0.5889145135879517,
      "learning_rate": 1.102868447082097e-05,
      "loss": 1.0257,
      "step": 670
    },
    {
      "epoch": 0.04484970402493116,
      "grad_norm": 0.7314364910125732,
      "learning_rate": 1.1193537751401254e-05,
      "loss": 1.0222,
      "step": 680
    },
    {
      "epoch": 0.04550925849588603,
      "grad_norm": 0.755777895450592,
      "learning_rate": 1.1358391031981536e-05,
      "loss": 1.0133,
      "step": 690
    },
    {
      "epoch": 0.046168812966840896,
      "grad_norm": 0.712040364742279,
      "learning_rate": 1.152324431256182e-05,
      "loss": 1.0136,
      "step": 700
    },
    {
      "epoch": 0.04682836743779577,
      "grad_norm": 0.843165397644043,
      "learning_rate": 1.1688097593142103e-05,
      "loss": 1.0341,
      "step": 710
    },
    {
      "epoch": 0.04748792190875064,
      "grad_norm": 0.75517338514328,
      "learning_rate": 1.1852950873722388e-05,
      "loss": 0.9576,
      "step": 720
    },
    {
      "epoch": 0.04814747637970551,
      "grad_norm": 0.619825541973114,
      "learning_rate": 1.2017804154302672e-05,
      "loss": 1.0275,
      "step": 730
    },
    {
      "epoch": 0.04880703085066038,
      "grad_norm": 0.5250429511070251,
      "learning_rate": 1.2182657434882955e-05,
      "loss": 0.9979,
      "step": 740
    },
    {
      "epoch": 0.04946658532161525,
      "grad_norm": 0.7191141843795776,
      "learning_rate": 1.2347510715463239e-05,
      "loss": 0.9829,
      "step": 750
    },
    {
      "epoch": 0.05012613979257012,
      "grad_norm": 0.6733546257019043,
      "learning_rate": 1.2512363996043521e-05,
      "loss": 1.0365,
      "step": 760
    },
    {
      "epoch": 0.050785694263524986,
      "grad_norm": 0.5763381719589233,
      "learning_rate": 1.2677217276623808e-05,
      "loss": 1.0175,
      "step": 770
    },
    {
      "epoch": 0.05144524873447986,
      "grad_norm": 0.6568360924720764,
      "learning_rate": 1.2842070557204088e-05,
      "loss": 0.9574,
      "step": 780
    },
    {
      "epoch": 0.05210480320543473,
      "grad_norm": 0.5359265208244324,
      "learning_rate": 1.3006923837784374e-05,
      "loss": 1.0253,
      "step": 790
    },
    {
      "epoch": 0.0527643576763896,
      "grad_norm": 0.6500004529953003,
      "learning_rate": 1.3171777118364655e-05,
      "loss": 1.0054,
      "step": 800
    },
    {
      "epoch": 0.05342391214734447,
      "grad_norm": 0.5643205642700195,
      "learning_rate": 1.3336630398944941e-05,
      "loss": 1.0252,
      "step": 810
    },
    {
      "epoch": 0.05408346661829934,
      "grad_norm": 0.675159752368927,
      "learning_rate": 1.3501483679525222e-05,
      "loss": 0.9847,
      "step": 820
    },
    {
      "epoch": 0.05474302108925421,
      "grad_norm": 0.6375489234924316,
      "learning_rate": 1.3666336960105508e-05,
      "loss": 0.9607,
      "step": 830
    },
    {
      "epoch": 0.055402575560209076,
      "grad_norm": 0.5561162233352661,
      "learning_rate": 1.3831190240685791e-05,
      "loss": 1.0017,
      "step": 840
    },
    {
      "epoch": 0.05606213003116395,
      "grad_norm": 0.5009911060333252,
      "learning_rate": 1.3996043521266075e-05,
      "loss": 0.9559,
      "step": 850
    },
    {
      "epoch": 0.05672168450211882,
      "grad_norm": 0.6449034214019775,
      "learning_rate": 1.4160896801846358e-05,
      "loss": 1.0037,
      "step": 860
    },
    {
      "epoch": 0.05738123897307369,
      "grad_norm": 0.5964744687080383,
      "learning_rate": 1.4325750082426642e-05,
      "loss": 1.0194,
      "step": 870
    },
    {
      "epoch": 0.05804079344402856,
      "grad_norm": 0.8459124565124512,
      "learning_rate": 1.4490603363006925e-05,
      "loss": 0.9388,
      "step": 880
    },
    {
      "epoch": 0.05870034791498343,
      "grad_norm": 0.630128026008606,
      "learning_rate": 1.4655456643587209e-05,
      "loss": 0.9996,
      "step": 890
    },
    {
      "epoch": 0.0593599023859383,
      "grad_norm": 0.5865899920463562,
      "learning_rate": 1.4820309924167492e-05,
      "loss": 0.9518,
      "step": 900
    },
    {
      "epoch": 0.060019456856893166,
      "grad_norm": 0.6959146857261658,
      "learning_rate": 1.4985163204747776e-05,
      "loss": 0.9677,
      "step": 910
    },
    {
      "epoch": 0.06067901132784804,
      "grad_norm": 0.5942516922950745,
      "learning_rate": 1.5150016485328059e-05,
      "loss": 0.9827,
      "step": 920
    },
    {
      "epoch": 0.06133856579880291,
      "grad_norm": 0.545923113822937,
      "learning_rate": 1.5314869765908343e-05,
      "loss": 0.952,
      "step": 930
    },
    {
      "epoch": 0.061998120269757775,
      "grad_norm": 0.6769543886184692,
      "learning_rate": 1.5479723046488627e-05,
      "loss": 0.9755,
      "step": 940
    },
    {
      "epoch": 0.06265767474071265,
      "grad_norm": 0.6097630262374878,
      "learning_rate": 1.564457632706891e-05,
      "loss": 1.004,
      "step": 950
    },
    {
      "epoch": 0.06331722921166752,
      "grad_norm": 0.6021948456764221,
      "learning_rate": 1.5809429607649192e-05,
      "loss": 0.9628,
      "step": 960
    },
    {
      "epoch": 0.06397678368262238,
      "grad_norm": 0.6364136934280396,
      "learning_rate": 1.5974282888229477e-05,
      "loss": 0.9575,
      "step": 970
    },
    {
      "epoch": 0.06463633815357726,
      "grad_norm": 0.6477786302566528,
      "learning_rate": 1.613913616880976e-05,
      "loss": 1.0281,
      "step": 980
    },
    {
      "epoch": 0.06529589262453213,
      "grad_norm": 0.6194107532501221,
      "learning_rate": 1.6303989449390045e-05,
      "loss": 1.0091,
      "step": 990
    },
    {
      "epoch": 0.065955447095487,
      "grad_norm": 0.6899977922439575,
      "learning_rate": 1.6468842729970326e-05,
      "loss": 0.9559,
      "step": 1000
    },
    {
      "epoch": 0.06661500156644187,
      "grad_norm": 0.6325615048408508,
      "learning_rate": 1.663369601055061e-05,
      "loss": 0.9716,
      "step": 1010
    },
    {
      "epoch": 0.06727455603739674,
      "grad_norm": 0.5693467259407043,
      "learning_rate": 1.6798549291130895e-05,
      "loss": 0.981,
      "step": 1020
    },
    {
      "epoch": 0.0679341105083516,
      "grad_norm": 0.6441985368728638,
      "learning_rate": 1.696340257171118e-05,
      "loss": 0.9821,
      "step": 1030
    },
    {
      "epoch": 0.06859366497930648,
      "grad_norm": 0.6741989850997925,
      "learning_rate": 1.712825585229146e-05,
      "loss": 0.9725,
      "step": 1040
    },
    {
      "epoch": 0.06925321945026135,
      "grad_norm": 0.6150327920913696,
      "learning_rate": 1.7293109132871744e-05,
      "loss": 0.9871,
      "step": 1050
    },
    {
      "epoch": 0.06991277392121621,
      "grad_norm": 0.5626732110977173,
      "learning_rate": 1.745796241345203e-05,
      "loss": 0.9707,
      "step": 1060
    },
    {
      "epoch": 0.07057232839217109,
      "grad_norm": 0.5015641450881958,
      "learning_rate": 1.7622815694032313e-05,
      "loss": 0.974,
      "step": 1070
    },
    {
      "epoch": 0.07123188286312596,
      "grad_norm": 0.5498797297477722,
      "learning_rate": 1.7787668974612594e-05,
      "loss": 0.9254,
      "step": 1080
    },
    {
      "epoch": 0.07189143733408083,
      "grad_norm": 0.6015375852584839,
      "learning_rate": 1.795252225519288e-05,
      "loss": 0.9425,
      "step": 1090
    },
    {
      "epoch": 0.0725509918050357,
      "grad_norm": 0.6349003911018372,
      "learning_rate": 1.8117375535773163e-05,
      "loss": 0.9542,
      "step": 1100
    },
    {
      "epoch": 0.07321054627599057,
      "grad_norm": 0.5576942563056946,
      "learning_rate": 1.8282228816353447e-05,
      "loss": 0.9377,
      "step": 1110
    },
    {
      "epoch": 0.07387010074694544,
      "grad_norm": 0.5731057524681091,
      "learning_rate": 1.8447082096933728e-05,
      "loss": 0.9705,
      "step": 1120
    },
    {
      "epoch": 0.07452965521790031,
      "grad_norm": 0.5626952648162842,
      "learning_rate": 1.8611935377514016e-05,
      "loss": 0.9557,
      "step": 1130
    },
    {
      "epoch": 0.07518920968885517,
      "grad_norm": 0.5902084708213806,
      "learning_rate": 1.8776788658094296e-05,
      "loss": 0.9629,
      "step": 1140
    },
    {
      "epoch": 0.07584876415981005,
      "grad_norm": 0.640737771987915,
      "learning_rate": 1.894164193867458e-05,
      "loss": 1.0292,
      "step": 1150
    },
    {
      "epoch": 0.07650831863076492,
      "grad_norm": 0.5191999673843384,
      "learning_rate": 1.9106495219254865e-05,
      "loss": 1.0054,
      "step": 1160
    },
    {
      "epoch": 0.07716787310171978,
      "grad_norm": 0.5990976095199585,
      "learning_rate": 1.927134849983515e-05,
      "loss": 0.9793,
      "step": 1170
    },
    {
      "epoch": 0.07782742757267466,
      "grad_norm": 0.5550162196159363,
      "learning_rate": 1.943620178041543e-05,
      "loss": 0.9916,
      "step": 1180
    },
    {
      "epoch": 0.07848698204362953,
      "grad_norm": 0.6130297183990479,
      "learning_rate": 1.9601055060995715e-05,
      "loss": 0.9418,
      "step": 1190
    },
    {
      "epoch": 0.07914653651458439,
      "grad_norm": 0.6504642963409424,
      "learning_rate": 1.9765908341576e-05,
      "loss": 0.9064,
      "step": 1200
    },
    {
      "epoch": 0.07980609098553927,
      "grad_norm": 0.6972726583480835,
      "learning_rate": 1.9930761622156283e-05,
      "loss": 0.9085,
      "step": 1210
    },
    {
      "epoch": 0.08046564545649414,
      "grad_norm": 0.5826172828674316,
      "learning_rate": 2.0095614902736564e-05,
      "loss": 0.9312,
      "step": 1220
    },
    {
      "epoch": 0.081125199927449,
      "grad_norm": 0.6361492872238159,
      "learning_rate": 2.026046818331685e-05,
      "loss": 0.9591,
      "step": 1230
    },
    {
      "epoch": 0.08178475439840388,
      "grad_norm": 0.6554346084594727,
      "learning_rate": 2.0425321463897133e-05,
      "loss": 0.9251,
      "step": 1240
    },
    {
      "epoch": 0.08244430886935875,
      "grad_norm": 0.5757082104682922,
      "learning_rate": 2.0590174744477417e-05,
      "loss": 0.9673,
      "step": 1250
    },
    {
      "epoch": 0.08310386334031362,
      "grad_norm": 0.7459129095077515,
      "learning_rate": 2.0755028025057698e-05,
      "loss": 0.9309,
      "step": 1260
    },
    {
      "epoch": 0.08376341781126849,
      "grad_norm": 0.670549750328064,
      "learning_rate": 2.0919881305637982e-05,
      "loss": 0.9372,
      "step": 1270
    },
    {
      "epoch": 0.08442297228222335,
      "grad_norm": 0.6436546444892883,
      "learning_rate": 2.1084734586218267e-05,
      "loss": 0.9501,
      "step": 1280
    },
    {
      "epoch": 0.08508252675317823,
      "grad_norm": 0.6000171303749084,
      "learning_rate": 2.124958786679855e-05,
      "loss": 0.9261,
      "step": 1290
    },
    {
      "epoch": 0.0857420812241331,
      "grad_norm": 0.6313101053237915,
      "learning_rate": 2.1414441147378832e-05,
      "loss": 0.9222,
      "step": 1300
    },
    {
      "epoch": 0.08640163569508796,
      "grad_norm": 0.833135724067688,
      "learning_rate": 2.157929442795912e-05,
      "loss": 0.9619,
      "step": 1310
    },
    {
      "epoch": 0.08706119016604284,
      "grad_norm": 0.5731329321861267,
      "learning_rate": 2.17441477085394e-05,
      "loss": 0.967,
      "step": 1320
    },
    {
      "epoch": 0.08772074463699771,
      "grad_norm": 0.6162570714950562,
      "learning_rate": 2.1909000989119685e-05,
      "loss": 0.9502,
      "step": 1330
    },
    {
      "epoch": 0.08838029910795257,
      "grad_norm": 0.6271604895591736,
      "learning_rate": 2.2073854269699966e-05,
      "loss": 0.9719,
      "step": 1340
    },
    {
      "epoch": 0.08903985357890745,
      "grad_norm": 0.7032260894775391,
      "learning_rate": 2.2238707550280253e-05,
      "loss": 0.9074,
      "step": 1350
    },
    {
      "epoch": 0.08969940804986232,
      "grad_norm": 0.62782222032547,
      "learning_rate": 2.2403560830860534e-05,
      "loss": 0.9138,
      "step": 1360
    },
    {
      "epoch": 0.09035896252081718,
      "grad_norm": 0.5636465549468994,
      "learning_rate": 2.256841411144082e-05,
      "loss": 0.9307,
      "step": 1370
    },
    {
      "epoch": 0.09101851699177206,
      "grad_norm": 0.6232348084449768,
      "learning_rate": 2.2733267392021103e-05,
      "loss": 0.9217,
      "step": 1380
    },
    {
      "epoch": 0.09167807146272693,
      "grad_norm": 0.5922995209693909,
      "learning_rate": 2.2898120672601387e-05,
      "loss": 0.9524,
      "step": 1390
    },
    {
      "epoch": 0.09233762593368179,
      "grad_norm": 0.5596097111701965,
      "learning_rate": 2.3062973953181668e-05,
      "loss": 0.9447,
      "step": 1400
    },
    {
      "epoch": 0.09299718040463667,
      "grad_norm": 0.5623578429222107,
      "learning_rate": 2.3227827233761952e-05,
      "loss": 0.9515,
      "step": 1410
    },
    {
      "epoch": 0.09365673487559154,
      "grad_norm": 0.6712508797645569,
      "learning_rate": 2.3392680514342237e-05,
      "loss": 0.9612,
      "step": 1420
    },
    {
      "epoch": 0.09431628934654641,
      "grad_norm": 0.781801164150238,
      "learning_rate": 2.355753379492252e-05,
      "loss": 0.9395,
      "step": 1430
    },
    {
      "epoch": 0.09497584381750128,
      "grad_norm": 0.6273669004440308,
      "learning_rate": 2.3722387075502802e-05,
      "loss": 0.9254,
      "step": 1440
    },
    {
      "epoch": 0.09563539828845614,
      "grad_norm": 0.5787047743797302,
      "learning_rate": 2.3887240356083086e-05,
      "loss": 0.8994,
      "step": 1450
    },
    {
      "epoch": 0.09629495275941102,
      "grad_norm": 0.5493133664131165,
      "learning_rate": 2.405209363666337e-05,
      "loss": 0.8898,
      "step": 1460
    },
    {
      "epoch": 0.09695450723036589,
      "grad_norm": 0.5380603671073914,
      "learning_rate": 2.4216946917243655e-05,
      "loss": 0.8835,
      "step": 1470
    },
    {
      "epoch": 0.09761406170132075,
      "grad_norm": 0.6819421052932739,
      "learning_rate": 2.4381800197823936e-05,
      "loss": 0.9235,
      "step": 1480
    },
    {
      "epoch": 0.09827361617227563,
      "grad_norm": 0.5991073250770569,
      "learning_rate": 2.454665347840422e-05,
      "loss": 0.8996,
      "step": 1490
    },
    {
      "epoch": 0.0989331706432305,
      "grad_norm": 0.6017669439315796,
      "learning_rate": 2.4711506758984505e-05,
      "loss": 0.946,
      "step": 1500
    },
    {
      "epoch": 0.09959272511418536,
      "grad_norm": 0.6362821459770203,
      "learning_rate": 2.487636003956479e-05,
      "loss": 0.9609,
      "step": 1510
    },
    {
      "epoch": 0.10025227958514024,
      "grad_norm": 0.6949606537818909,
      "learning_rate": 2.5041213320145073e-05,
      "loss": 0.9661,
      "step": 1520
    },
    {
      "epoch": 0.1009118340560951,
      "grad_norm": 0.6807888746261597,
      "learning_rate": 2.5206066600725354e-05,
      "loss": 0.9111,
      "step": 1530
    },
    {
      "epoch": 0.10157138852704997,
      "grad_norm": 0.6306922435760498,
      "learning_rate": 2.537091988130564e-05,
      "loss": 0.9029,
      "step": 1540
    },
    {
      "epoch": 0.10223094299800485,
      "grad_norm": 0.578364372253418,
      "learning_rate": 2.5535773161885923e-05,
      "loss": 0.9389,
      "step": 1550
    },
    {
      "epoch": 0.10289049746895972,
      "grad_norm": 0.5773734450340271,
      "learning_rate": 2.5700626442466207e-05,
      "loss": 0.944,
      "step": 1560
    },
    {
      "epoch": 0.10355005193991458,
      "grad_norm": 0.7100627422332764,
      "learning_rate": 2.5865479723046488e-05,
      "loss": 0.9705,
      "step": 1570
    },
    {
      "epoch": 0.10420960641086946,
      "grad_norm": 0.6515047550201416,
      "learning_rate": 2.6030333003626772e-05,
      "loss": 0.9532,
      "step": 1580
    },
    {
      "epoch": 0.10486916088182432,
      "grad_norm": 0.6922847032546997,
      "learning_rate": 2.6195186284207057e-05,
      "loss": 0.9124,
      "step": 1590
    },
    {
      "epoch": 0.1055287153527792,
      "grad_norm": 0.7992035150527954,
      "learning_rate": 2.636003956478734e-05,
      "loss": 0.9313,
      "step": 1600
    },
    {
      "epoch": 0.10618826982373407,
      "grad_norm": 0.6953135132789612,
      "learning_rate": 2.6524892845367622e-05,
      "loss": 0.9398,
      "step": 1610
    },
    {
      "epoch": 0.10684782429468893,
      "grad_norm": 0.7350666522979736,
      "learning_rate": 2.6689746125947906e-05,
      "loss": 0.8814,
      "step": 1620
    },
    {
      "epoch": 0.10750737876564381,
      "grad_norm": 0.7054291367530823,
      "learning_rate": 2.685459940652819e-05,
      "loss": 0.9421,
      "step": 1630
    },
    {
      "epoch": 0.10816693323659868,
      "grad_norm": 0.5834742784500122,
      "learning_rate": 2.7019452687108475e-05,
      "loss": 0.9328,
      "step": 1640
    },
    {
      "epoch": 0.10882648770755354,
      "grad_norm": 0.6302939653396606,
      "learning_rate": 2.7184305967688756e-05,
      "loss": 1.0036,
      "step": 1650
    },
    {
      "epoch": 0.10948604217850842,
      "grad_norm": 0.6931148767471313,
      "learning_rate": 2.734915924826904e-05,
      "loss": 0.9166,
      "step": 1660
    },
    {
      "epoch": 0.11014559664946329,
      "grad_norm": 0.5533245801925659,
      "learning_rate": 2.7514012528849324e-05,
      "loss": 0.8807,
      "step": 1670
    },
    {
      "epoch": 0.11080515112041815,
      "grad_norm": 0.6529316306114197,
      "learning_rate": 2.7678865809429612e-05,
      "loss": 0.8954,
      "step": 1680
    },
    {
      "epoch": 0.11146470559137303,
      "grad_norm": 0.567509114742279,
      "learning_rate": 2.784371909000989e-05,
      "loss": 0.878,
      "step": 1690
    },
    {
      "epoch": 0.1121242600623279,
      "grad_norm": 0.6449151039123535,
      "learning_rate": 2.8008572370590174e-05,
      "loss": 0.9475,
      "step": 1700
    },
    {
      "epoch": 0.11278381453328276,
      "grad_norm": 0.6593036651611328,
      "learning_rate": 2.8173425651170458e-05,
      "loss": 0.9124,
      "step": 1710
    },
    {
      "epoch": 0.11344336900423764,
      "grad_norm": 0.5768759846687317,
      "learning_rate": 2.8338278931750746e-05,
      "loss": 0.9553,
      "step": 1720
    },
    {
      "epoch": 0.1141029234751925,
      "grad_norm": 0.7477861046791077,
      "learning_rate": 2.8503132212331023e-05,
      "loss": 0.9001,
      "step": 1730
    },
    {
      "epoch": 0.11476247794614738,
      "grad_norm": 0.6525564193725586,
      "learning_rate": 2.8667985492911308e-05,
      "loss": 0.9107,
      "step": 1740
    },
    {
      "epoch": 0.11542203241710225,
      "grad_norm": 0.6128164529800415,
      "learning_rate": 2.8832838773491595e-05,
      "loss": 0.95,
      "step": 1750
    },
    {
      "epoch": 0.11608158688805711,
      "grad_norm": 0.5953107476234436,
      "learning_rate": 2.899769205407188e-05,
      "loss": 0.9213,
      "step": 1760
    },
    {
      "epoch": 0.116741141359012,
      "grad_norm": 0.679503858089447,
      "learning_rate": 2.9162545334652157e-05,
      "loss": 0.9104,
      "step": 1770
    },
    {
      "epoch": 0.11740069582996686,
      "grad_norm": 0.6580171585083008,
      "learning_rate": 2.932739861523244e-05,
      "loss": 0.9161,
      "step": 1780
    },
    {
      "epoch": 0.11806025030092172,
      "grad_norm": 0.6093540787696838,
      "learning_rate": 2.949225189581273e-05,
      "loss": 0.9435,
      "step": 1790
    },
    {
      "epoch": 0.1187198047718766,
      "grad_norm": 0.5813942551612854,
      "learning_rate": 2.9657105176393013e-05,
      "loss": 0.9012,
      "step": 1800
    },
    {
      "epoch": 0.11937935924283147,
      "grad_norm": 0.6638506650924683,
      "learning_rate": 2.9821958456973298e-05,
      "loss": 0.9067,
      "step": 1810
    },
    {
      "epoch": 0.12003891371378633,
      "grad_norm": 0.5925025343894958,
      "learning_rate": 2.998681173755358e-05,
      "loss": 0.9353,
      "step": 1820
    },
    {
      "epoch": 0.12069846818474121,
      "grad_norm": 0.688334584236145,
      "learning_rate": 3.0151665018133863e-05,
      "loss": 0.935,
      "step": 1830
    },
    {
      "epoch": 0.12135802265569608,
      "grad_norm": 0.6738244891166687,
      "learning_rate": 3.0316518298714147e-05,
      "loss": 0.9315,
      "step": 1840
    },
    {
      "epoch": 0.12201757712665094,
      "grad_norm": 0.6192309856414795,
      "learning_rate": 3.048137157929443e-05,
      "loss": 0.923,
      "step": 1850
    },
    {
      "epoch": 0.12267713159760582,
      "grad_norm": 0.6709905862808228,
      "learning_rate": 3.064622485987471e-05,
      "loss": 0.8941,
      "step": 1860
    },
    {
      "epoch": 0.12333668606856069,
      "grad_norm": 0.6975002288818359,
      "learning_rate": 3.0811078140455e-05,
      "loss": 0.95,
      "step": 1870
    },
    {
      "epoch": 0.12399624053951555,
      "grad_norm": 0.6484049558639526,
      "learning_rate": 3.0975931421035285e-05,
      "loss": 0.8955,
      "step": 1880
    },
    {
      "epoch": 0.12465579501047043,
      "grad_norm": 0.5713815093040466,
      "learning_rate": 3.1140784701615565e-05,
      "loss": 0.8926,
      "step": 1890
    },
    {
      "epoch": 0.1253153494814253,
      "grad_norm": 0.7232243418693542,
      "learning_rate": 3.1305637982195846e-05,
      "loss": 0.9017,
      "step": 1900
    },
    {
      "epoch": 0.12597490395238017,
      "grad_norm": 0.604157030582428,
      "learning_rate": 3.147049126277613e-05,
      "loss": 0.9463,
      "step": 1910
    },
    {
      "epoch": 0.12663445842333504,
      "grad_norm": 0.6578916311264038,
      "learning_rate": 3.1635344543356415e-05,
      "loss": 0.9106,
      "step": 1920
    },
    {
      "epoch": 0.1272940128942899,
      "grad_norm": 0.6498791575431824,
      "learning_rate": 3.18001978239367e-05,
      "loss": 0.8915,
      "step": 1930
    },
    {
      "epoch": 0.12795356736524477,
      "grad_norm": 0.5826857686042786,
      "learning_rate": 3.196505110451698e-05,
      "loss": 0.8851,
      "step": 1940
    },
    {
      "epoch": 0.12861312183619963,
      "grad_norm": 0.622488796710968,
      "learning_rate": 3.2129904385097265e-05,
      "loss": 0.8704,
      "step": 1950
    },
    {
      "epoch": 0.12927267630715453,
      "grad_norm": 0.5181499719619751,
      "learning_rate": 3.229475766567755e-05,
      "loss": 0.925,
      "step": 1960
    },
    {
      "epoch": 0.1299322307781094,
      "grad_norm": 0.6072003841400146,
      "learning_rate": 3.245961094625783e-05,
      "loss": 0.8958,
      "step": 1970
    },
    {
      "epoch": 0.13059178524906426,
      "grad_norm": 0.5885782241821289,
      "learning_rate": 3.2624464226838114e-05,
      "loss": 0.9435,
      "step": 1980
    },
    {
      "epoch": 0.13125133972001912,
      "grad_norm": 0.6380855441093445,
      "learning_rate": 3.27893175074184e-05,
      "loss": 0.9126,
      "step": 1990
    },
    {
      "epoch": 0.131910894190974,
      "grad_norm": 0.6280835866928101,
      "learning_rate": 3.295417078799868e-05,
      "loss": 0.9274,
      "step": 2000
    },
    {
      "epoch": 0.13257044866192888,
      "grad_norm": 0.6791607141494751,
      "learning_rate": 3.311902406857897e-05,
      "loss": 0.9361,
      "step": 2010
    },
    {
      "epoch": 0.13323000313288375,
      "grad_norm": 0.6630708575248718,
      "learning_rate": 3.328387734915925e-05,
      "loss": 0.9806,
      "step": 2020
    },
    {
      "epoch": 0.1338895576038386,
      "grad_norm": 0.6305481195449829,
      "learning_rate": 3.344873062973953e-05,
      "loss": 0.9137,
      "step": 2030
    },
    {
      "epoch": 0.13454911207479348,
      "grad_norm": 0.7444617748260498,
      "learning_rate": 3.361358391031982e-05,
      "loss": 0.92,
      "step": 2040
    },
    {
      "epoch": 0.13520866654574834,
      "grad_norm": 0.6238999962806702,
      "learning_rate": 3.37784371909001e-05,
      "loss": 0.9301,
      "step": 2050
    },
    {
      "epoch": 0.1358682210167032,
      "grad_norm": 0.6913322806358337,
      "learning_rate": 3.394329047148038e-05,
      "loss": 0.909,
      "step": 2060
    },
    {
      "epoch": 0.1365277754876581,
      "grad_norm": 0.6888453960418701,
      "learning_rate": 3.410814375206067e-05,
      "loss": 0.8859,
      "step": 2070
    },
    {
      "epoch": 0.13718732995861296,
      "grad_norm": 0.644161581993103,
      "learning_rate": 3.427299703264095e-05,
      "loss": 0.8954,
      "step": 2080
    },
    {
      "epoch": 0.13784688442956783,
      "grad_norm": 0.6524642705917358,
      "learning_rate": 3.443785031322124e-05,
      "loss": 0.8958,
      "step": 2090
    },
    {
      "epoch": 0.1385064389005227,
      "grad_norm": 0.6420117020606995,
      "learning_rate": 3.460270359380152e-05,
      "loss": 0.9183,
      "step": 2100
    },
    {
      "epoch": 0.13916599337147756,
      "grad_norm": 0.595023512840271,
      "learning_rate": 3.47675568743818e-05,
      "loss": 0.9226,
      "step": 2110
    },
    {
      "epoch": 0.13982554784243242,
      "grad_norm": 0.6285088062286377,
      "learning_rate": 3.493241015496209e-05,
      "loss": 0.8814,
      "step": 2120
    },
    {
      "epoch": 0.14048510231338732,
      "grad_norm": 0.6196804642677307,
      "learning_rate": 3.509726343554237e-05,
      "loss": 0.8598,
      "step": 2130
    },
    {
      "epoch": 0.14114465678434218,
      "grad_norm": 0.599234938621521,
      "learning_rate": 3.526211671612265e-05,
      "loss": 0.8996,
      "step": 2140
    },
    {
      "epoch": 0.14180421125529705,
      "grad_norm": 0.5277974009513855,
      "learning_rate": 3.542696999670294e-05,
      "loss": 0.9318,
      "step": 2150
    },
    {
      "epoch": 0.1424637657262519,
      "grad_norm": 0.6323888897895813,
      "learning_rate": 3.559182327728322e-05,
      "loss": 0.8764,
      "step": 2160
    },
    {
      "epoch": 0.14312332019720678,
      "grad_norm": 0.6974547505378723,
      "learning_rate": 3.5756676557863506e-05,
      "loss": 0.9063,
      "step": 2170
    },
    {
      "epoch": 0.14378287466816167,
      "grad_norm": 0.6774551272392273,
      "learning_rate": 3.592152983844379e-05,
      "loss": 0.8704,
      "step": 2180
    },
    {
      "epoch": 0.14444242913911653,
      "grad_norm": 0.673445999622345,
      "learning_rate": 3.608638311902407e-05,
      "loss": 0.9432,
      "step": 2190
    },
    {
      "epoch": 0.1451019836100714,
      "grad_norm": 0.6598107218742371,
      "learning_rate": 3.6251236399604355e-05,
      "loss": 0.9339,
      "step": 2200
    },
    {
      "epoch": 0.14576153808102627,
      "grad_norm": 0.6883885264396667,
      "learning_rate": 3.6416089680184636e-05,
      "loss": 0.9098,
      "step": 2210
    },
    {
      "epoch": 0.14642109255198113,
      "grad_norm": 0.6021624207496643,
      "learning_rate": 3.658094296076492e-05,
      "loss": 0.8606,
      "step": 2220
    },
    {
      "epoch": 0.147080647022936,
      "grad_norm": 0.5796533226966858,
      "learning_rate": 3.6745796241345205e-05,
      "loss": 0.899,
      "step": 2230
    },
    {
      "epoch": 0.1477402014938909,
      "grad_norm": 0.6242685317993164,
      "learning_rate": 3.6910649521925486e-05,
      "loss": 0.9218,
      "step": 2240
    },
    {
      "epoch": 0.14839975596484575,
      "grad_norm": 0.7199188470840454,
      "learning_rate": 3.7075502802505774e-05,
      "loss": 0.8937,
      "step": 2250
    },
    {
      "epoch": 0.14905931043580062,
      "grad_norm": 0.694506049156189,
      "learning_rate": 3.7240356083086054e-05,
      "loss": 0.8794,
      "step": 2260
    },
    {
      "epoch": 0.14971886490675548,
      "grad_norm": 0.6348100304603577,
      "learning_rate": 3.7405209363666335e-05,
      "loss": 0.9325,
      "step": 2270
    },
    {
      "epoch": 0.15037841937771035,
      "grad_norm": 0.730395495891571,
      "learning_rate": 3.757006264424662e-05,
      "loss": 0.9095,
      "step": 2280
    },
    {
      "epoch": 0.1510379738486652,
      "grad_norm": 0.639094352722168,
      "learning_rate": 3.773491592482691e-05,
      "loss": 0.8881,
      "step": 2290
    },
    {
      "epoch": 0.1516975283196201,
      "grad_norm": 0.7560307383537292,
      "learning_rate": 3.7899769205407185e-05,
      "loss": 0.8777,
      "step": 2300
    },
    {
      "epoch": 0.15235708279057497,
      "grad_norm": 0.6120066046714783,
      "learning_rate": 3.806462248598747e-05,
      "loss": 0.9273,
      "step": 2310
    },
    {
      "epoch": 0.15301663726152984,
      "grad_norm": 0.6584673523902893,
      "learning_rate": 3.822947576656776e-05,
      "loss": 0.9507,
      "step": 2320
    },
    {
      "epoch": 0.1536761917324847,
      "grad_norm": 0.7297015190124512,
      "learning_rate": 3.839432904714804e-05,
      "loss": 0.9074,
      "step": 2330
    },
    {
      "epoch": 0.15433574620343957,
      "grad_norm": 0.6097325682640076,
      "learning_rate": 3.855918232772832e-05,
      "loss": 0.9008,
      "step": 2340
    },
    {
      "epoch": 0.15499530067439446,
      "grad_norm": 0.6594650149345398,
      "learning_rate": 3.87240356083086e-05,
      "loss": 0.9434,
      "step": 2350
    },
    {
      "epoch": 0.15565485514534932,
      "grad_norm": 0.7125281691551208,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.8952,
      "step": 2360
    },
    {
      "epoch": 0.1563144096163042,
      "grad_norm": 0.6079130172729492,
      "learning_rate": 3.905374216946918e-05,
      "loss": 0.8867,
      "step": 2370
    },
    {
      "epoch": 0.15697396408725905,
      "grad_norm": 0.6090205907821655,
      "learning_rate": 3.921859545004945e-05,
      "loss": 0.8838,
      "step": 2380
    },
    {
      "epoch": 0.15763351855821392,
      "grad_norm": 0.7213959097862244,
      "learning_rate": 3.938344873062974e-05,
      "loss": 0.8975,
      "step": 2390
    },
    {
      "epoch": 0.15829307302916878,
      "grad_norm": 0.7012157440185547,
      "learning_rate": 3.954830201121003e-05,
      "loss": 0.9334,
      "step": 2400
    },
    {
      "epoch": 0.15895262750012368,
      "grad_norm": 0.8252381682395935,
      "learning_rate": 3.971315529179031e-05,
      "loss": 0.9535,
      "step": 2410
    },
    {
      "epoch": 0.15961218197107854,
      "grad_norm": 0.7257195711135864,
      "learning_rate": 3.987800857237059e-05,
      "loss": 0.92,
      "step": 2420
    },
    {
      "epoch": 0.1602717364420334,
      "grad_norm": 0.6549400091171265,
      "learning_rate": 4.004286185295088e-05,
      "loss": 0.9118,
      "step": 2430
    },
    {
      "epoch": 0.16093129091298827,
      "grad_norm": 0.6405781507492065,
      "learning_rate": 4.020771513353116e-05,
      "loss": 0.8934,
      "step": 2440
    },
    {
      "epoch": 0.16159084538394314,
      "grad_norm": 0.6783310770988464,
      "learning_rate": 4.0372568414111446e-05,
      "loss": 0.9367,
      "step": 2450
    },
    {
      "epoch": 0.162250399854898,
      "grad_norm": 0.6306205987930298,
      "learning_rate": 4.053742169469173e-05,
      "loss": 0.8781,
      "step": 2460
    },
    {
      "epoch": 0.1629099543258529,
      "grad_norm": 0.7541260123252869,
      "learning_rate": 4.070227497527201e-05,
      "loss": 0.9119,
      "step": 2470
    },
    {
      "epoch": 0.16356950879680776,
      "grad_norm": 0.6016479134559631,
      "learning_rate": 4.0867128255852296e-05,
      "loss": 0.9254,
      "step": 2480
    },
    {
      "epoch": 0.16422906326776263,
      "grad_norm": 0.6229802966117859,
      "learning_rate": 4.103198153643258e-05,
      "loss": 0.8907,
      "step": 2490
    },
    {
      "epoch": 0.1648886177387175,
      "grad_norm": 0.5935705900192261,
      "learning_rate": 4.119683481701286e-05,
      "loss": 0.901,
      "step": 2500
    },
    {
      "epoch": 0.16554817220967236,
      "grad_norm": 0.7060070037841797,
      "learning_rate": 4.1361688097593145e-05,
      "loss": 0.8912,
      "step": 2510
    },
    {
      "epoch": 0.16620772668062725,
      "grad_norm": 0.704928457736969,
      "learning_rate": 4.1526541378173426e-05,
      "loss": 0.9041,
      "step": 2520
    },
    {
      "epoch": 0.16686728115158211,
      "grad_norm": 0.7448880672454834,
      "learning_rate": 4.1691394658753714e-05,
      "loss": 0.8736,
      "step": 2530
    },
    {
      "epoch": 0.16752683562253698,
      "grad_norm": 0.6432943940162659,
      "learning_rate": 4.1856247939333995e-05,
      "loss": 0.8916,
      "step": 2540
    },
    {
      "epoch": 0.16818639009349184,
      "grad_norm": 0.692283034324646,
      "learning_rate": 4.2021101219914276e-05,
      "loss": 0.8787,
      "step": 2550
    },
    {
      "epoch": 0.1688459445644467,
      "grad_norm": 0.6590387225151062,
      "learning_rate": 4.2185954500494563e-05,
      "loss": 0.8528,
      "step": 2560
    },
    {
      "epoch": 0.16950549903540157,
      "grad_norm": 0.6531996726989746,
      "learning_rate": 4.2350807781074844e-05,
      "loss": 0.8857,
      "step": 2570
    },
    {
      "epoch": 0.17016505350635647,
      "grad_norm": 0.6231783628463745,
      "learning_rate": 4.2515661061655125e-05,
      "loss": 0.9153,
      "step": 2580
    },
    {
      "epoch": 0.17082460797731133,
      "grad_norm": 0.6619783639907837,
      "learning_rate": 4.268051434223541e-05,
      "loss": 0.929,
      "step": 2590
    },
    {
      "epoch": 0.1714841624482662,
      "grad_norm": 0.6574437618255615,
      "learning_rate": 4.2845367622815694e-05,
      "loss": 0.9076,
      "step": 2600
    },
    {
      "epoch": 0.17214371691922106,
      "grad_norm": 0.6736916303634644,
      "learning_rate": 4.301022090339598e-05,
      "loss": 0.9293,
      "step": 2610
    },
    {
      "epoch": 0.17280327139017593,
      "grad_norm": 0.7516410946846008,
      "learning_rate": 4.317507418397626e-05,
      "loss": 0.9069,
      "step": 2620
    },
    {
      "epoch": 0.1734628258611308,
      "grad_norm": 0.5540705919265747,
      "learning_rate": 4.3339927464556543e-05,
      "loss": 0.8764,
      "step": 2630
    },
    {
      "epoch": 0.17412238033208569,
      "grad_norm": 0.6677078008651733,
      "learning_rate": 4.350478074513683e-05,
      "loss": 0.9231,
      "step": 2640
    },
    {
      "epoch": 0.17478193480304055,
      "grad_norm": 0.5827996730804443,
      "learning_rate": 4.366963402571712e-05,
      "loss": 0.8681,
      "step": 2650
    },
    {
      "epoch": 0.17544148927399542,
      "grad_norm": 0.6666778922080994,
      "learning_rate": 4.383448730629739e-05,
      "loss": 0.9257,
      "step": 2660
    },
    {
      "epoch": 0.17610104374495028,
      "grad_norm": 0.5299721956253052,
      "learning_rate": 4.399934058687768e-05,
      "loss": 0.9241,
      "step": 2670
    },
    {
      "epoch": 0.17676059821590515,
      "grad_norm": 0.5545915365219116,
      "learning_rate": 4.416419386745796e-05,
      "loss": 0.862,
      "step": 2680
    },
    {
      "epoch": 0.17742015268686004,
      "grad_norm": 0.6056398749351501,
      "learning_rate": 4.432904714803825e-05,
      "loss": 0.9276,
      "step": 2690
    },
    {
      "epoch": 0.1780797071578149,
      "grad_norm": 0.740776538848877,
      "learning_rate": 4.449390042861853e-05,
      "loss": 0.9382,
      "step": 2700
    },
    {
      "epoch": 0.17873926162876977,
      "grad_norm": 0.6620045900344849,
      "learning_rate": 4.465875370919881e-05,
      "loss": 0.9182,
      "step": 2710
    },
    {
      "epoch": 0.17939881609972463,
      "grad_norm": 0.6999577879905701,
      "learning_rate": 4.48236069897791e-05,
      "loss": 0.9137,
      "step": 2720
    },
    {
      "epoch": 0.1800583705706795,
      "grad_norm": 0.6420654058456421,
      "learning_rate": 4.4988460270359386e-05,
      "loss": 0.9166,
      "step": 2730
    },
    {
      "epoch": 0.18071792504163436,
      "grad_norm": 0.7118934392929077,
      "learning_rate": 4.515331355093966e-05,
      "loss": 0.9266,
      "step": 2740
    },
    {
      "epoch": 0.18137747951258926,
      "grad_norm": 0.5614007115364075,
      "learning_rate": 4.531816683151995e-05,
      "loss": 0.9392,
      "step": 2750
    },
    {
      "epoch": 0.18203703398354412,
      "grad_norm": 0.6258883476257324,
      "learning_rate": 4.5483020112100236e-05,
      "loss": 0.8736,
      "step": 2760
    },
    {
      "epoch": 0.182696588454499,
      "grad_norm": 0.5963823795318604,
      "learning_rate": 4.564787339268052e-05,
      "loss": 0.8908,
      "step": 2770
    },
    {
      "epoch": 0.18335614292545385,
      "grad_norm": 0.703870415687561,
      "learning_rate": 4.58127266732608e-05,
      "loss": 0.9045,
      "step": 2780
    },
    {
      "epoch": 0.18401569739640872,
      "grad_norm": 0.5508717894554138,
      "learning_rate": 4.5977579953841086e-05,
      "loss": 0.8899,
      "step": 2790
    },
    {
      "epoch": 0.18467525186736358,
      "grad_norm": 0.6593130230903625,
      "learning_rate": 4.6142433234421366e-05,
      "loss": 0.9196,
      "step": 2800
    },
    {
      "epoch": 0.18533480633831848,
      "grad_norm": 0.6316845417022705,
      "learning_rate": 4.6307286515001654e-05,
      "loss": 0.8912,
      "step": 2810
    },
    {
      "epoch": 0.18599436080927334,
      "grad_norm": 0.6781242489814758,
      "learning_rate": 4.647213979558193e-05,
      "loss": 0.8971,
      "step": 2820
    },
    {
      "epoch": 0.1866539152802282,
      "grad_norm": 0.6441785097122192,
      "learning_rate": 4.6636993076162216e-05,
      "loss": 0.917,
      "step": 2830
    },
    {
      "epoch": 0.18731346975118307,
      "grad_norm": 0.636024534702301,
      "learning_rate": 4.6801846356742504e-05,
      "loss": 0.8869,
      "step": 2840
    },
    {
      "epoch": 0.18797302422213794,
      "grad_norm": 0.5574973821640015,
      "learning_rate": 4.6966699637322785e-05,
      "loss": 0.8681,
      "step": 2850
    },
    {
      "epoch": 0.18863257869309283,
      "grad_norm": 0.6535305976867676,
      "learning_rate": 4.713155291790307e-05,
      "loss": 0.8982,
      "step": 2860
    },
    {
      "epoch": 0.1892921331640477,
      "grad_norm": 0.5861814618110657,
      "learning_rate": 4.729640619848335e-05,
      "loss": 0.9378,
      "step": 2870
    },
    {
      "epoch": 0.18995168763500256,
      "grad_norm": 0.6891550421714783,
      "learning_rate": 4.7461259479063634e-05,
      "loss": 0.9189,
      "step": 2880
    },
    {
      "epoch": 0.19061124210595742,
      "grad_norm": 0.5748773813247681,
      "learning_rate": 4.762611275964392e-05,
      "loss": 0.8761,
      "step": 2890
    },
    {
      "epoch": 0.1912707965769123,
      "grad_norm": 0.673592209815979,
      "learning_rate": 4.77909660402242e-05,
      "loss": 0.8495,
      "step": 2900
    },
    {
      "epoch": 0.19193035104786715,
      "grad_norm": 0.539668083190918,
      "learning_rate": 4.7955819320804484e-05,
      "loss": 0.8722,
      "step": 2910
    },
    {
      "epoch": 0.19258990551882205,
      "grad_norm": 0.6330294013023376,
      "learning_rate": 4.812067260138477e-05,
      "loss": 0.9154,
      "step": 2920
    },
    {
      "epoch": 0.1932494599897769,
      "grad_norm": 0.7211683392524719,
      "learning_rate": 4.828552588196505e-05,
      "loss": 0.8343,
      "step": 2930
    },
    {
      "epoch": 0.19390901446073178,
      "grad_norm": 0.6515793800354004,
      "learning_rate": 4.845037916254534e-05,
      "loss": 0.9021,
      "step": 2940
    },
    {
      "epoch": 0.19456856893168664,
      "grad_norm": 0.6246060132980347,
      "learning_rate": 4.861523244312562e-05,
      "loss": 0.8516,
      "step": 2950
    },
    {
      "epoch": 0.1952281234026415,
      "grad_norm": 0.6002302765846252,
      "learning_rate": 4.87800857237059e-05,
      "loss": 0.862,
      "step": 2960
    },
    {
      "epoch": 0.19588767787359637,
      "grad_norm": 0.6544924378395081,
      "learning_rate": 4.894493900428619e-05,
      "loss": 0.8543,
      "step": 2970
    },
    {
      "epoch": 0.19654723234455126,
      "grad_norm": 0.6579413414001465,
      "learning_rate": 4.910979228486647e-05,
      "loss": 0.9218,
      "step": 2980
    },
    {
      "epoch": 0.19720678681550613,
      "grad_norm": 0.6065923571586609,
      "learning_rate": 4.927464556544675e-05,
      "loss": 0.8727,
      "step": 2990
    },
    {
      "epoch": 0.197866341286461,
      "grad_norm": 0.8123323917388916,
      "learning_rate": 4.943949884602704e-05,
      "loss": 0.911,
      "step": 3000
    },
    {
      "epoch": 0.19852589575741586,
      "grad_norm": 0.6149587631225586,
      "learning_rate": 4.960435212660732e-05,
      "loss": 0.8881,
      "step": 3010
    },
    {
      "epoch": 0.19918545022837073,
      "grad_norm": 0.6561806201934814,
      "learning_rate": 4.976920540718761e-05,
      "loss": 0.8765,
      "step": 3020
    },
    {
      "epoch": 0.19984500469932562,
      "grad_norm": 0.5923125743865967,
      "learning_rate": 4.993405868776789e-05,
      "loss": 0.9018,
      "step": 3030
    },
    {
      "epoch": 0.20050455917028048,
      "grad_norm": 0.616742730140686,
      "learning_rate": 4.999999403688283e-05,
      "loss": 0.8803,
      "step": 3040
    },
    {
      "epoch": 0.20116411364123535,
      "grad_norm": 0.5280983448028564,
      "learning_rate": 4.9999957595621525e-05,
      "loss": 0.9116,
      "step": 3050
    },
    {
      "epoch": 0.2018236681121902,
      "grad_norm": 0.5969537496566772,
      "learning_rate": 4.999988802599003e-05,
      "loss": 0.8889,
      "step": 3060
    },
    {
      "epoch": 0.20248322258314508,
      "grad_norm": 0.6396189332008362,
      "learning_rate": 4.999978532808053e-05,
      "loss": 0.8778,
      "step": 3070
    },
    {
      "epoch": 0.20314277705409994,
      "grad_norm": 0.6153563857078552,
      "learning_rate": 4.99996495020291e-05,
      "loss": 0.8914,
      "step": 3080
    },
    {
      "epoch": 0.20380233152505484,
      "grad_norm": 0.6899839043617249,
      "learning_rate": 4.999948054801574e-05,
      "loss": 0.8639,
      "step": 3090
    },
    {
      "epoch": 0.2044618859960097,
      "grad_norm": 0.6921502351760864,
      "learning_rate": 4.999927846626434e-05,
      "loss": 0.8647,
      "step": 3100
    },
    {
      "epoch": 0.20512144046696457,
      "grad_norm": 0.6061365008354187,
      "learning_rate": 4.999904325704268e-05,
      "loss": 0.8942,
      "step": 3110
    },
    {
      "epoch": 0.20578099493791943,
      "grad_norm": 0.6922749280929565,
      "learning_rate": 4.999877492066245e-05,
      "loss": 0.8897,
      "step": 3120
    },
    {
      "epoch": 0.2064405494088743,
      "grad_norm": 0.6063687205314636,
      "learning_rate": 4.9998473457479234e-05,
      "loss": 0.8576,
      "step": 3130
    },
    {
      "epoch": 0.20710010387982916,
      "grad_norm": 0.6878769993782043,
      "learning_rate": 4.99981388678925e-05,
      "loss": 0.8683,
      "step": 3140
    },
    {
      "epoch": 0.20775965835078405,
      "grad_norm": 0.7053576707839966,
      "learning_rate": 4.999777115234563e-05,
      "loss": 0.8899,
      "step": 3150
    },
    {
      "epoch": 0.20841921282173892,
      "grad_norm": 0.6048821806907654,
      "learning_rate": 4.9997370311325904e-05,
      "loss": 0.8854,
      "step": 3160
    },
    {
      "epoch": 0.20907876729269378,
      "grad_norm": 0.5910998582839966,
      "learning_rate": 4.9996936345364485e-05,
      "loss": 0.8622,
      "step": 3170
    },
    {
      "epoch": 0.20973832176364865,
      "grad_norm": 0.5993175506591797,
      "learning_rate": 4.999646925503644e-05,
      "loss": 0.9227,
      "step": 3180
    },
    {
      "epoch": 0.21039787623460351,
      "grad_norm": 0.5830239653587341,
      "learning_rate": 4.999596904096073e-05,
      "loss": 0.883,
      "step": 3190
    },
    {
      "epoch": 0.2110574307055584,
      "grad_norm": 0.5703751444816589,
      "learning_rate": 4.99954357038002e-05,
      "loss": 0.9075,
      "step": 3200
    },
    {
      "epoch": 0.21171698517651327,
      "grad_norm": 0.6441596150398254,
      "learning_rate": 4.99948692442616e-05,
      "loss": 0.9032,
      "step": 3210
    },
    {
      "epoch": 0.21237653964746814,
      "grad_norm": 0.6591752171516418,
      "learning_rate": 4.999426966309557e-05,
      "loss": 0.8922,
      "step": 3220
    },
    {
      "epoch": 0.213036094118423,
      "grad_norm": 0.6065967679023743,
      "learning_rate": 4.999363696109663e-05,
      "loss": 0.8931,
      "step": 3230
    },
    {
      "epoch": 0.21369564858937787,
      "grad_norm": 0.5592631101608276,
      "learning_rate": 4.9992971139103194e-05,
      "loss": 0.8682,
      "step": 3240
    },
    {
      "epoch": 0.21435520306033273,
      "grad_norm": 0.5951939821243286,
      "learning_rate": 4.999227219799758e-05,
      "loss": 0.8172,
      "step": 3250
    },
    {
      "epoch": 0.21501475753128763,
      "grad_norm": 0.6042636036872864,
      "learning_rate": 4.9991540138705975e-05,
      "loss": 0.9106,
      "step": 3260
    },
    {
      "epoch": 0.2156743120022425,
      "grad_norm": 0.5808688998222351,
      "learning_rate": 4.999077496219845e-05,
      "loss": 0.926,
      "step": 3270
    },
    {
      "epoch": 0.21633386647319736,
      "grad_norm": 0.7375866174697876,
      "learning_rate": 4.998997666948898e-05,
      "loss": 0.9025,
      "step": 3280
    },
    {
      "epoch": 0.21699342094415222,
      "grad_norm": 0.5514616370201111,
      "learning_rate": 4.9989145261635415e-05,
      "loss": 0.9091,
      "step": 3290
    },
    {
      "epoch": 0.21765297541510709,
      "grad_norm": 0.5729548931121826,
      "learning_rate": 4.998828073973947e-05,
      "loss": 0.933,
      "step": 3300
    },
    {
      "epoch": 0.21831252988606198,
      "grad_norm": 0.644983172416687,
      "learning_rate": 4.998738310494676e-05,
      "loss": 0.8818,
      "step": 3310
    },
    {
      "epoch": 0.21897208435701684,
      "grad_norm": 0.5531386137008667,
      "learning_rate": 4.998645235844678e-05,
      "loss": 0.9143,
      "step": 3320
    },
    {
      "epoch": 0.2196316388279717,
      "grad_norm": 0.5550761222839355,
      "learning_rate": 4.9985488501472885e-05,
      "loss": 0.975,
      "step": 3330
    },
    {
      "epoch": 0.22029119329892657,
      "grad_norm": 0.5695024132728577,
      "learning_rate": 4.998449153530234e-05,
      "loss": 0.9387,
      "step": 3340
    },
    {
      "epoch": 0.22095074776988144,
      "grad_norm": 0.621046781539917,
      "learning_rate": 4.998346146125623e-05,
      "loss": 0.9138,
      "step": 3350
    },
    {
      "epoch": 0.2216103022408363,
      "grad_norm": 0.6257467269897461,
      "learning_rate": 4.9982398280699575e-05,
      "loss": 0.9066,
      "step": 3360
    },
    {
      "epoch": 0.2222698567117912,
      "grad_norm": 0.5778996348381042,
      "learning_rate": 4.998130199504122e-05,
      "loss": 0.906,
      "step": 3370
    },
    {
      "epoch": 0.22292941118274606,
      "grad_norm": 0.5707346200942993,
      "learning_rate": 4.998017260573389e-05,
      "loss": 0.9184,
      "step": 3380
    },
    {
      "epoch": 0.22358896565370093,
      "grad_norm": 0.6047236919403076,
      "learning_rate": 4.9979010114274195e-05,
      "loss": 0.8492,
      "step": 3390
    },
    {
      "epoch": 0.2242485201246558,
      "grad_norm": 0.6120710372924805,
      "learning_rate": 4.9977814522202574e-05,
      "loss": 0.8874,
      "step": 3400
    },
    {
      "epoch": 0.22490807459561066,
      "grad_norm": 0.5748836398124695,
      "learning_rate": 4.997658583110337e-05,
      "loss": 0.8931,
      "step": 3410
    },
    {
      "epoch": 0.22556762906656552,
      "grad_norm": 0.6227717399597168,
      "learning_rate": 4.997532404260476e-05,
      "loss": 0.8616,
      "step": 3420
    },
    {
      "epoch": 0.22622718353752042,
      "grad_norm": 0.5952726006507874,
      "learning_rate": 4.997402915837879e-05,
      "loss": 0.8802,
      "step": 3430
    },
    {
      "epoch": 0.22688673800847528,
      "grad_norm": 0.5556307435035706,
      "learning_rate": 4.997270118014135e-05,
      "loss": 0.9117,
      "step": 3440
    },
    {
      "epoch": 0.22754629247943015,
      "grad_norm": 0.6270772814750671,
      "learning_rate": 4.99713401096522e-05,
      "loss": 0.8638,
      "step": 3450
    },
    {
      "epoch": 0.228205846950385,
      "grad_norm": 0.5514934062957764,
      "learning_rate": 4.996994594871495e-05,
      "loss": 0.8688,
      "step": 3460
    },
    {
      "epoch": 0.22886540142133988,
      "grad_norm": 0.6048349738121033,
      "learning_rate": 4.9968518699177035e-05,
      "loss": 0.8659,
      "step": 3470
    },
    {
      "epoch": 0.22952495589229477,
      "grad_norm": 0.6672205924987793,
      "learning_rate": 4.996705836292979e-05,
      "loss": 0.884,
      "step": 3480
    },
    {
      "epoch": 0.23018451036324963,
      "grad_norm": 0.5777316093444824,
      "learning_rate": 4.996556494190832e-05,
      "loss": 0.9188,
      "step": 3490
    },
    {
      "epoch": 0.2308440648342045,
      "grad_norm": 0.5783727765083313,
      "learning_rate": 4.996403843809164e-05,
      "loss": 0.884,
      "step": 3500
    },
    {
      "epoch": 0.23150361930515936,
      "grad_norm": 0.5978668928146362,
      "learning_rate": 4.996247885350257e-05,
      "loss": 0.8894,
      "step": 3510
    },
    {
      "epoch": 0.23216317377611423,
      "grad_norm": 0.5237613320350647,
      "learning_rate": 4.9960886190207774e-05,
      "loss": 0.9107,
      "step": 3520
    },
    {
      "epoch": 0.2328227282470691,
      "grad_norm": 0.6135637760162354,
      "learning_rate": 4.9959260450317746e-05,
      "loss": 0.8781,
      "step": 3530
    },
    {
      "epoch": 0.233482282718024,
      "grad_norm": 0.6045156717300415,
      "learning_rate": 4.9957601635986815e-05,
      "loss": 0.9053,
      "step": 3540
    },
    {
      "epoch": 0.23414183718897885,
      "grad_norm": 0.6048551797866821,
      "learning_rate": 4.995590974941314e-05,
      "loss": 0.9054,
      "step": 3550
    },
    {
      "epoch": 0.23480139165993372,
      "grad_norm": 0.6331484317779541,
      "learning_rate": 4.99541847928387e-05,
      "loss": 0.8917,
      "step": 3560
    },
    {
      "epoch": 0.23546094613088858,
      "grad_norm": 0.5210322737693787,
      "learning_rate": 4.99524267685493e-05,
      "loss": 0.899,
      "step": 3570
    },
    {
      "epoch": 0.23612050060184345,
      "grad_norm": 0.6640326976776123,
      "learning_rate": 4.9950635678874564e-05,
      "loss": 0.8845,
      "step": 3580
    },
    {
      "epoch": 0.2367800550727983,
      "grad_norm": 0.7048312425613403,
      "learning_rate": 4.9948811526187935e-05,
      "loss": 0.8704,
      "step": 3590
    },
    {
      "epoch": 0.2374396095437532,
      "grad_norm": 0.5205860137939453,
      "learning_rate": 4.9946954312906645e-05,
      "loss": 0.907,
      "step": 3600
    },
    {
      "epoch": 0.23809916401470807,
      "grad_norm": 0.6026431918144226,
      "learning_rate": 4.9945064041491786e-05,
      "loss": 0.8609,
      "step": 3610
    },
    {
      "epoch": 0.23875871848566294,
      "grad_norm": 0.7078092694282532,
      "learning_rate": 4.994314071444821e-05,
      "loss": 0.8966,
      "step": 3620
    },
    {
      "epoch": 0.2394182729566178,
      "grad_norm": 0.6005507707595825,
      "learning_rate": 4.994118433432459e-05,
      "loss": 0.8864,
      "step": 3630
    },
    {
      "epoch": 0.24007782742757267,
      "grad_norm": 0.5615901350975037,
      "learning_rate": 4.99391949037134e-05,
      "loss": 0.8206,
      "step": 3640
    },
    {
      "epoch": 0.24073738189852756,
      "grad_norm": 0.5781015157699585,
      "learning_rate": 4.99371724252509e-05,
      "loss": 0.891,
      "step": 3650
    },
    {
      "epoch": 0.24139693636948242,
      "grad_norm": 0.516188383102417,
      "learning_rate": 4.993511690161717e-05,
      "loss": 0.8413,
      "step": 3660
    },
    {
      "epoch": 0.2420564908404373,
      "grad_norm": 0.5980455875396729,
      "learning_rate": 4.993302833553605e-05,
      "loss": 0.8802,
      "step": 3670
    },
    {
      "epoch": 0.24271604531139215,
      "grad_norm": 0.6498844623565674,
      "learning_rate": 4.9930906729775164e-05,
      "loss": 0.8944,
      "step": 3680
    },
    {
      "epoch": 0.24337559978234702,
      "grad_norm": 0.6646761298179626,
      "learning_rate": 4.9928752087145945e-05,
      "loss": 0.8729,
      "step": 3690
    },
    {
      "epoch": 0.24403515425330188,
      "grad_norm": 0.5892811417579651,
      "learning_rate": 4.9926564410503593e-05,
      "loss": 0.9241,
      "step": 3700
    },
    {
      "epoch": 0.24469470872425678,
      "grad_norm": 0.588645339012146,
      "learning_rate": 4.992434370274707e-05,
      "loss": 0.8868,
      "step": 3710
    },
    {
      "epoch": 0.24535426319521164,
      "grad_norm": 0.6737022399902344,
      "learning_rate": 4.992208996681912e-05,
      "loss": 0.8572,
      "step": 3720
    },
    {
      "epoch": 0.2460138176661665,
      "grad_norm": 0.5201873779296875,
      "learning_rate": 4.991980320570625e-05,
      "loss": 0.8591,
      "step": 3730
    },
    {
      "epoch": 0.24667337213712137,
      "grad_norm": 0.6893309354782104,
      "learning_rate": 4.991748342243874e-05,
      "loss": 0.9197,
      "step": 3740
    },
    {
      "epoch": 0.24733292660807624,
      "grad_norm": 0.6144750118255615,
      "learning_rate": 4.9915130620090616e-05,
      "loss": 0.8604,
      "step": 3750
    },
    {
      "epoch": 0.2479924810790311,
      "grad_norm": 0.6070724129676819,
      "learning_rate": 4.9912744801779664e-05,
      "loss": 0.8649,
      "step": 3760
    },
    {
      "epoch": 0.248652035549986,
      "grad_norm": 0.5637763142585754,
      "learning_rate": 4.991032597066742e-05,
      "loss": 0.8973,
      "step": 3770
    },
    {
      "epoch": 0.24931159002094086,
      "grad_norm": 0.6145963668823242,
      "learning_rate": 4.990787412995916e-05,
      "loss": 0.8506,
      "step": 3780
    },
    {
      "epoch": 0.24997114449189572,
      "grad_norm": 0.6317866444587708,
      "learning_rate": 4.990538928290392e-05,
      "loss": 0.8798,
      "step": 3790
    },
    {
      "epoch": 0.2506306989628506,
      "grad_norm": 0.6320456862449646,
      "learning_rate": 4.990287143279445e-05,
      "loss": 0.9093,
      "step": 3800
    },
    {
      "epoch": 0.25129025343380545,
      "grad_norm": 0.651431143283844,
      "learning_rate": 4.9900320582967263e-05,
      "loss": 0.8826,
      "step": 3810
    },
    {
      "epoch": 0.25194980790476035,
      "grad_norm": 0.6451202034950256,
      "learning_rate": 4.989773673680258e-05,
      "loss": 0.9028,
      "step": 3820
    },
    {
      "epoch": 0.2526093623757152,
      "grad_norm": 0.6628925800323486,
      "learning_rate": 4.989511989772434e-05,
      "loss": 0.8489,
      "step": 3830
    },
    {
      "epoch": 0.2532689168466701,
      "grad_norm": 0.8010949492454529,
      "learning_rate": 4.989247006920023e-05,
      "loss": 0.8915,
      "step": 3840
    },
    {
      "epoch": 0.25392847131762497,
      "grad_norm": 0.5844048261642456,
      "learning_rate": 4.988978725474162e-05,
      "loss": 0.9375,
      "step": 3850
    },
    {
      "epoch": 0.2545880257885798,
      "grad_norm": 0.6060191988945007,
      "learning_rate": 4.988707145790361e-05,
      "loss": 0.86,
      "step": 3860
    },
    {
      "epoch": 0.2552475802595347,
      "grad_norm": 0.5984094738960266,
      "learning_rate": 4.988432268228501e-05,
      "loss": 0.8814,
      "step": 3870
    },
    {
      "epoch": 0.25590713473048954,
      "grad_norm": 0.6448334455490112,
      "learning_rate": 4.988154093152833e-05,
      "loss": 0.8941,
      "step": 3880
    },
    {
      "epoch": 0.25656668920144443,
      "grad_norm": 0.5510440468788147,
      "learning_rate": 4.987872620931975e-05,
      "loss": 0.8881,
      "step": 3890
    },
    {
      "epoch": 0.25722624367239927,
      "grad_norm": 0.5189316868782043,
      "learning_rate": 4.987587851938918e-05,
      "loss": 0.8564,
      "step": 3900
    },
    {
      "epoch": 0.25788579814335416,
      "grad_norm": 0.5691280961036682,
      "learning_rate": 4.98729978655102e-05,
      "loss": 0.8402,
      "step": 3910
    },
    {
      "epoch": 0.25854535261430905,
      "grad_norm": 0.6573358178138733,
      "learning_rate": 4.987008425150005e-05,
      "loss": 0.8846,
      "step": 3920
    },
    {
      "epoch": 0.2592049070852639,
      "grad_norm": 0.5777154564857483,
      "learning_rate": 4.98671376812197e-05,
      "loss": 0.8662,
      "step": 3930
    },
    {
      "epoch": 0.2598644615562188,
      "grad_norm": 0.5385237336158752,
      "learning_rate": 4.986415815857374e-05,
      "loss": 0.8887,
      "step": 3940
    },
    {
      "epoch": 0.2605240160271736,
      "grad_norm": 0.6187642216682434,
      "learning_rate": 4.9861145687510445e-05,
      "loss": 0.84,
      "step": 3950
    },
    {
      "epoch": 0.2611835704981285,
      "grad_norm": 0.5783065557479858,
      "learning_rate": 4.9858100272021754e-05,
      "loss": 0.86,
      "step": 3960
    },
    {
      "epoch": 0.2618431249690834,
      "grad_norm": 0.5744683742523193,
      "learning_rate": 4.985502191614326e-05,
      "loss": 0.9069,
      "step": 3970
    },
    {
      "epoch": 0.26250267944003824,
      "grad_norm": 0.5620474815368652,
      "learning_rate": 4.985191062395422e-05,
      "loss": 0.8804,
      "step": 3980
    },
    {
      "epoch": 0.26316223391099314,
      "grad_norm": 0.6070266366004944,
      "learning_rate": 4.984876639957751e-05,
      "loss": 0.8759,
      "step": 3990
    },
    {
      "epoch": 0.263821788381948,
      "grad_norm": 0.689795196056366,
      "learning_rate": 4.984558924717965e-05,
      "loss": 0.8584,
      "step": 4000
    },
    {
      "epoch": 0.26448134285290287,
      "grad_norm": 0.5451216101646423,
      "learning_rate": 4.984237917097081e-05,
      "loss": 0.8829,
      "step": 4010
    },
    {
      "epoch": 0.26514089732385776,
      "grad_norm": 0.5520846843719482,
      "learning_rate": 4.9839136175204795e-05,
      "loss": 0.9094,
      "step": 4020
    },
    {
      "epoch": 0.2658004517948126,
      "grad_norm": 0.6379942297935486,
      "learning_rate": 4.983586026417899e-05,
      "loss": 0.9022,
      "step": 4030
    },
    {
      "epoch": 0.2664600062657675,
      "grad_norm": 0.5754224061965942,
      "learning_rate": 4.983255144223445e-05,
      "loss": 0.8302,
      "step": 4040
    },
    {
      "epoch": 0.26711956073672233,
      "grad_norm": 0.5679827332496643,
      "learning_rate": 4.9829209713755815e-05,
      "loss": 0.84,
      "step": 4050
    },
    {
      "epoch": 0.2677791152076772,
      "grad_norm": 0.5987281203269958,
      "learning_rate": 4.9825835083171326e-05,
      "loss": 0.8516,
      "step": 4060
    },
    {
      "epoch": 0.26843866967863206,
      "grad_norm": 0.5965194702148438,
      "learning_rate": 4.9822427554952826e-05,
      "loss": 0.8453,
      "step": 4070
    },
    {
      "epoch": 0.26909822414958695,
      "grad_norm": 0.6309374570846558,
      "learning_rate": 4.981898713361577e-05,
      "loss": 0.8861,
      "step": 4080
    },
    {
      "epoch": 0.26975777862054184,
      "grad_norm": 0.5262725949287415,
      "learning_rate": 4.9815513823719185e-05,
      "loss": 0.8558,
      "step": 4090
    },
    {
      "epoch": 0.2704173330914967,
      "grad_norm": 0.564869225025177,
      "learning_rate": 4.9812007629865676e-05,
      "loss": 0.8498,
      "step": 4100
    },
    {
      "epoch": 0.2710768875624516,
      "grad_norm": 0.6223157644271851,
      "learning_rate": 4.980846855670143e-05,
      "loss": 0.9029,
      "step": 4110
    },
    {
      "epoch": 0.2717364420334064,
      "grad_norm": 0.5574937462806702,
      "learning_rate": 4.9804896608916215e-05,
      "loss": 0.8667,
      "step": 4120
    },
    {
      "epoch": 0.2723959965043613,
      "grad_norm": 0.5297979712486267,
      "learning_rate": 4.980129179124334e-05,
      "loss": 0.8944,
      "step": 4130
    },
    {
      "epoch": 0.2730555509753162,
      "grad_norm": 0.5349012017250061,
      "learning_rate": 4.979765410845969e-05,
      "loss": 0.9075,
      "step": 4140
    },
    {
      "epoch": 0.27371510544627103,
      "grad_norm": 0.5487902164459229,
      "learning_rate": 4.979398356538568e-05,
      "loss": 0.8637,
      "step": 4150
    },
    {
      "epoch": 0.2743746599172259,
      "grad_norm": 0.6216309666633606,
      "learning_rate": 4.9790280166885306e-05,
      "loss": 0.888,
      "step": 4160
    },
    {
      "epoch": 0.27503421438818076,
      "grad_norm": 0.6944143772125244,
      "learning_rate": 4.9786543917866057e-05,
      "loss": 0.9287,
      "step": 4170
    },
    {
      "epoch": 0.27569376885913566,
      "grad_norm": 0.5620063543319702,
      "learning_rate": 4.978277482327899e-05,
      "loss": 0.9018,
      "step": 4180
    },
    {
      "epoch": 0.27635332333009055,
      "grad_norm": 0.6146485209465027,
      "learning_rate": 4.9778972888118655e-05,
      "loss": 0.8805,
      "step": 4190
    },
    {
      "epoch": 0.2770128778010454,
      "grad_norm": 0.4971786439418793,
      "learning_rate": 4.977513811742316e-05,
      "loss": 0.8887,
      "step": 4200
    },
    {
      "epoch": 0.2776724322720003,
      "grad_norm": 0.6936700940132141,
      "learning_rate": 4.977127051627408e-05,
      "loss": 0.9038,
      "step": 4210
    },
    {
      "epoch": 0.2783319867429551,
      "grad_norm": 0.5615195035934448,
      "learning_rate": 4.976737008979653e-05,
      "loss": 0.8549,
      "step": 4220
    },
    {
      "epoch": 0.27899154121391,
      "grad_norm": 0.5746992230415344,
      "learning_rate": 4.9763436843159103e-05,
      "loss": 0.8949,
      "step": 4230
    },
    {
      "epoch": 0.27965109568486485,
      "grad_norm": 0.5601503849029541,
      "learning_rate": 4.97594707815739e-05,
      "loss": 0.8856,
      "step": 4240
    },
    {
      "epoch": 0.28031065015581974,
      "grad_norm": 0.5905441045761108,
      "learning_rate": 4.975547191029648e-05,
      "loss": 0.8566,
      "step": 4250
    },
    {
      "epoch": 0.28097020462677463,
      "grad_norm": 0.5898568034172058,
      "learning_rate": 4.97514402346259e-05,
      "loss": 0.8271,
      "step": 4260
    },
    {
      "epoch": 0.28162975909772947,
      "grad_norm": 0.5321268439292908,
      "learning_rate": 4.9747375759904694e-05,
      "loss": 0.9163,
      "step": 4270
    },
    {
      "epoch": 0.28228931356868436,
      "grad_norm": 0.5248478055000305,
      "learning_rate": 4.974327849151884e-05,
      "loss": 0.8777,
      "step": 4280
    },
    {
      "epoch": 0.2829488680396392,
      "grad_norm": 0.6414687037467957,
      "learning_rate": 4.973914843489779e-05,
      "loss": 0.8378,
      "step": 4290
    },
    {
      "epoch": 0.2836084225105941,
      "grad_norm": 0.66009122133255,
      "learning_rate": 4.9734985595514404e-05,
      "loss": 0.8621,
      "step": 4300
    },
    {
      "epoch": 0.284267976981549,
      "grad_norm": 0.5482301115989685,
      "learning_rate": 4.973078997888505e-05,
      "loss": 0.8923,
      "step": 4310
    },
    {
      "epoch": 0.2849275314525038,
      "grad_norm": 0.5216924548149109,
      "learning_rate": 4.972656159056949e-05,
      "loss": 0.8672,
      "step": 4320
    },
    {
      "epoch": 0.2855870859234587,
      "grad_norm": 0.6303116679191589,
      "learning_rate": 4.97223004361709e-05,
      "loss": 0.9094,
      "step": 4330
    },
    {
      "epoch": 0.28624664039441355,
      "grad_norm": 0.5526524782180786,
      "learning_rate": 4.97180065213359e-05,
      "loss": 0.8468,
      "step": 4340
    },
    {
      "epoch": 0.28690619486536845,
      "grad_norm": 0.5693706274032593,
      "learning_rate": 4.9713679851754526e-05,
      "loss": 0.8807,
      "step": 4350
    },
    {
      "epoch": 0.28756574933632334,
      "grad_norm": 0.55186927318573,
      "learning_rate": 4.97093204331602e-05,
      "loss": 0.8718,
      "step": 4360
    },
    {
      "epoch": 0.2882253038072782,
      "grad_norm": 0.6338434219360352,
      "learning_rate": 4.970492827132975e-05,
      "loss": 0.8295,
      "step": 4370
    },
    {
      "epoch": 0.28888485827823307,
      "grad_norm": 0.6424077749252319,
      "learning_rate": 4.970050337208339e-05,
      "loss": 0.8789,
      "step": 4380
    },
    {
      "epoch": 0.2895444127491879,
      "grad_norm": 0.6192151308059692,
      "learning_rate": 4.969604574128472e-05,
      "loss": 0.8799,
      "step": 4390
    },
    {
      "epoch": 0.2902039672201428,
      "grad_norm": 0.5858079195022583,
      "learning_rate": 4.969155538484071e-05,
      "loss": 0.8652,
      "step": 4400
    },
    {
      "epoch": 0.29086352169109764,
      "grad_norm": 0.6232174038887024,
      "learning_rate": 4.9687032308701715e-05,
      "loss": 0.8596,
      "step": 4410
    },
    {
      "epoch": 0.29152307616205253,
      "grad_norm": 0.6015047430992126,
      "learning_rate": 4.968247651886141e-05,
      "loss": 0.9113,
      "step": 4420
    },
    {
      "epoch": 0.2921826306330074,
      "grad_norm": 0.5569378137588501,
      "learning_rate": 4.9677888021356846e-05,
      "loss": 0.8941,
      "step": 4430
    },
    {
      "epoch": 0.29284218510396226,
      "grad_norm": 0.6491048336029053,
      "learning_rate": 4.9673266822268405e-05,
      "loss": 0.8547,
      "step": 4440
    },
    {
      "epoch": 0.29350173957491715,
      "grad_norm": 0.6606380939483643,
      "learning_rate": 4.966861292771982e-05,
      "loss": 0.8491,
      "step": 4450
    },
    {
      "epoch": 0.294161294045872,
      "grad_norm": 0.5726791620254517,
      "learning_rate": 4.966392634387814e-05,
      "loss": 0.8852,
      "step": 4460
    },
    {
      "epoch": 0.2948208485168269,
      "grad_norm": 0.6234275102615356,
      "learning_rate": 4.965920707695372e-05,
      "loss": 0.862,
      "step": 4470
    },
    {
      "epoch": 0.2954804029877818,
      "grad_norm": 0.5283318161964417,
      "learning_rate": 4.965445513320025e-05,
      "loss": 0.8395,
      "step": 4480
    },
    {
      "epoch": 0.2961399574587366,
      "grad_norm": 0.5711292028427124,
      "learning_rate": 4.9649670518914696e-05,
      "loss": 0.8736,
      "step": 4490
    },
    {
      "epoch": 0.2967995119296915,
      "grad_norm": 0.5051360130310059,
      "learning_rate": 4.9644853240437323e-05,
      "loss": 0.8488,
      "step": 4500
    },
    {
      "epoch": 0.29745906640064634,
      "grad_norm": 0.5224673748016357,
      "learning_rate": 4.96400033041517e-05,
      "loss": 0.8795,
      "step": 4510
    },
    {
      "epoch": 0.29811862087160124,
      "grad_norm": 0.5676709413528442,
      "learning_rate": 4.963512071648464e-05,
      "loss": 0.8647,
      "step": 4520
    },
    {
      "epoch": 0.29877817534255613,
      "grad_norm": 0.5240904092788696,
      "learning_rate": 4.963020548390626e-05,
      "loss": 0.8622,
      "step": 4530
    },
    {
      "epoch": 0.29943772981351097,
      "grad_norm": 0.6189314723014832,
      "learning_rate": 4.96252576129299e-05,
      "loss": 0.8645,
      "step": 4540
    },
    {
      "epoch": 0.30009728428446586,
      "grad_norm": 0.6082637906074524,
      "learning_rate": 4.962027711011218e-05,
      "loss": 0.8999,
      "step": 4550
    },
    {
      "epoch": 0.3007568387554207,
      "grad_norm": 0.4707930088043213,
      "learning_rate": 4.9615263982052945e-05,
      "loss": 0.8538,
      "step": 4560
    },
    {
      "epoch": 0.3014163932263756,
      "grad_norm": 0.5452629327774048,
      "learning_rate": 4.961021823539527e-05,
      "loss": 0.8774,
      "step": 4570
    },
    {
      "epoch": 0.3020759476973304,
      "grad_norm": 0.6488273739814758,
      "learning_rate": 4.960513987682547e-05,
      "loss": 0.8377,
      "step": 4580
    },
    {
      "epoch": 0.3027355021682853,
      "grad_norm": 0.6546377539634705,
      "learning_rate": 4.960002891307306e-05,
      "loss": 0.8474,
      "step": 4590
    },
    {
      "epoch": 0.3033950566392402,
      "grad_norm": 0.5141822099685669,
      "learning_rate": 4.9594885350910775e-05,
      "loss": 0.9065,
      "step": 4600
    },
    {
      "epoch": 0.30405461111019505,
      "grad_norm": 0.5785159468650818,
      "learning_rate": 4.958970919715453e-05,
      "loss": 0.8542,
      "step": 4610
    },
    {
      "epoch": 0.30471416558114994,
      "grad_norm": 0.5312493443489075,
      "learning_rate": 4.958450045866344e-05,
      "loss": 0.875,
      "step": 4620
    },
    {
      "epoch": 0.3053737200521048,
      "grad_norm": 0.5114297866821289,
      "learning_rate": 4.9579259142339803e-05,
      "loss": 0.865,
      "step": 4630
    },
    {
      "epoch": 0.3060332745230597,
      "grad_norm": 0.5772585868835449,
      "learning_rate": 4.957398525512908e-05,
      "loss": 0.8766,
      "step": 4640
    },
    {
      "epoch": 0.30669282899401457,
      "grad_norm": 0.6451815366744995,
      "learning_rate": 4.956867880401989e-05,
      "loss": 0.8471,
      "step": 4650
    },
    {
      "epoch": 0.3073523834649694,
      "grad_norm": 0.5841494798660278,
      "learning_rate": 4.9563339796044004e-05,
      "loss": 0.8858,
      "step": 4660
    },
    {
      "epoch": 0.3080119379359243,
      "grad_norm": 0.6031407117843628,
      "learning_rate": 4.955796823827636e-05,
      "loss": 0.8522,
      "step": 4670
    },
    {
      "epoch": 0.30867149240687913,
      "grad_norm": 0.6057419776916504,
      "learning_rate": 4.955256413783499e-05,
      "loss": 0.8584,
      "step": 4680
    },
    {
      "epoch": 0.309331046877834,
      "grad_norm": 0.6070857048034668,
      "learning_rate": 4.954712750188106e-05,
      "loss": 0.8726,
      "step": 4690
    },
    {
      "epoch": 0.3099906013487889,
      "grad_norm": 0.5966079235076904,
      "learning_rate": 4.954165833761889e-05,
      "loss": 0.8269,
      "step": 4700
    },
    {
      "epoch": 0.31065015581974376,
      "grad_norm": 0.6299596428871155,
      "learning_rate": 4.953615665229584e-05,
      "loss": 0.853,
      "step": 4710
    },
    {
      "epoch": 0.31130971029069865,
      "grad_norm": 0.531351625919342,
      "learning_rate": 4.9530622453202414e-05,
      "loss": 0.8558,
      "step": 4720
    },
    {
      "epoch": 0.3119692647616535,
      "grad_norm": 0.5519111752510071,
      "learning_rate": 4.952505574767217e-05,
      "loss": 0.8969,
      "step": 4730
    },
    {
      "epoch": 0.3126288192326084,
      "grad_norm": 0.542658805847168,
      "learning_rate": 4.951945654308178e-05,
      "loss": 0.8505,
      "step": 4740
    },
    {
      "epoch": 0.3132883737035632,
      "grad_norm": 0.6531395316123962,
      "learning_rate": 4.951382484685093e-05,
      "loss": 0.8559,
      "step": 4750
    },
    {
      "epoch": 0.3139479281745181,
      "grad_norm": 0.5937902331352234,
      "learning_rate": 4.9508160666442404e-05,
      "loss": 0.8593,
      "step": 4760
    },
    {
      "epoch": 0.314607482645473,
      "grad_norm": 0.5703521370887756,
      "learning_rate": 4.9502464009362015e-05,
      "loss": 0.8447,
      "step": 4770
    },
    {
      "epoch": 0.31526703711642784,
      "grad_norm": 0.5206552147865295,
      "learning_rate": 4.949673488315862e-05,
      "loss": 0.8426,
      "step": 4780
    },
    {
      "epoch": 0.31592659158738273,
      "grad_norm": 0.6926104426383972,
      "learning_rate": 4.949097329542409e-05,
      "loss": 0.8905,
      "step": 4790
    },
    {
      "epoch": 0.31658614605833757,
      "grad_norm": 0.5506643652915955,
      "learning_rate": 4.948517925379332e-05,
      "loss": 0.8486,
      "step": 4800
    },
    {
      "epoch": 0.31724570052929246,
      "grad_norm": 0.48270678520202637,
      "learning_rate": 4.947935276594421e-05,
      "loss": 0.8676,
      "step": 4810
    },
    {
      "epoch": 0.31790525500024736,
      "grad_norm": 0.6450955867767334,
      "learning_rate": 4.947349383959765e-05,
      "loss": 0.8027,
      "step": 4820
    },
    {
      "epoch": 0.3185648094712022,
      "grad_norm": 0.6653099656105042,
      "learning_rate": 4.946760248251752e-05,
      "loss": 0.8735,
      "step": 4830
    },
    {
      "epoch": 0.3192243639421571,
      "grad_norm": 0.6305930614471436,
      "learning_rate": 4.946167870251069e-05,
      "loss": 0.8307,
      "step": 4840
    },
    {
      "epoch": 0.3198839184131119,
      "grad_norm": 0.5568416118621826,
      "learning_rate": 4.945572250742696e-05,
      "loss": 0.8584,
      "step": 4850
    },
    {
      "epoch": 0.3205434728840668,
      "grad_norm": 0.5373754501342773,
      "learning_rate": 4.9449733905159125e-05,
      "loss": 0.9037,
      "step": 4860
    },
    {
      "epoch": 0.3212030273550217,
      "grad_norm": 0.6957933306694031,
      "learning_rate": 4.944371290364289e-05,
      "loss": 0.8741,
      "step": 4870
    },
    {
      "epoch": 0.32186258182597655,
      "grad_norm": 0.5444389581680298,
      "learning_rate": 4.943765951085691e-05,
      "loss": 0.888,
      "step": 4880
    },
    {
      "epoch": 0.32252213629693144,
      "grad_norm": 0.5078149437904358,
      "learning_rate": 4.943157373482276e-05,
      "loss": 0.8617,
      "step": 4890
    },
    {
      "epoch": 0.3231816907678863,
      "grad_norm": 0.5339057445526123,
      "learning_rate": 4.942545558360493e-05,
      "loss": 0.8304,
      "step": 4900
    },
    {
      "epoch": 0.32384124523884117,
      "grad_norm": 0.5843328237533569,
      "learning_rate": 4.941930506531081e-05,
      "loss": 0.8492,
      "step": 4910
    },
    {
      "epoch": 0.324500799709796,
      "grad_norm": 0.5101600885391235,
      "learning_rate": 4.9413122188090676e-05,
      "loss": 0.8542,
      "step": 4920
    },
    {
      "epoch": 0.3251603541807509,
      "grad_norm": 0.4721750020980835,
      "learning_rate": 4.940690696013769e-05,
      "loss": 0.8755,
      "step": 4930
    },
    {
      "epoch": 0.3258199086517058,
      "grad_norm": 0.5229089260101318,
      "learning_rate": 4.940065938968787e-05,
      "loss": 0.8836,
      "step": 4940
    },
    {
      "epoch": 0.32647946312266063,
      "grad_norm": 0.4514390528202057,
      "learning_rate": 4.9394379485020124e-05,
      "loss": 0.8757,
      "step": 4950
    },
    {
      "epoch": 0.3271390175936155,
      "grad_norm": 0.5412085652351379,
      "learning_rate": 4.938806725445617e-05,
      "loss": 0.8467,
      "step": 4960
    },
    {
      "epoch": 0.32779857206457036,
      "grad_norm": 0.5420681238174438,
      "learning_rate": 4.938172270636059e-05,
      "loss": 0.8671,
      "step": 4970
    },
    {
      "epoch": 0.32845812653552525,
      "grad_norm": 0.5458070635795593,
      "learning_rate": 4.937534584914076e-05,
      "loss": 0.8665,
      "step": 4980
    },
    {
      "epoch": 0.32911768100648015,
      "grad_norm": 0.5852974057197571,
      "learning_rate": 4.936893669124691e-05,
      "loss": 0.8574,
      "step": 4990
    },
    {
      "epoch": 0.329777235477435,
      "grad_norm": 0.525571346282959,
      "learning_rate": 4.936249524117206e-05,
      "loss": 0.852,
      "step": 5000
    },
    {
      "epoch": 0.3304367899483899,
      "grad_norm": 0.5699165463447571,
      "learning_rate": 4.935602150745198e-05,
      "loss": 0.8888,
      "step": 5010
    },
    {
      "epoch": 0.3310963444193447,
      "grad_norm": 0.5384656190872192,
      "learning_rate": 4.934951549866529e-05,
      "loss": 0.8972,
      "step": 5020
    },
    {
      "epoch": 0.3317558988902996,
      "grad_norm": 0.6388717889785767,
      "learning_rate": 4.934297722343332e-05,
      "loss": 0.8702,
      "step": 5030
    },
    {
      "epoch": 0.3324154533612545,
      "grad_norm": 0.5993049740791321,
      "learning_rate": 4.933640669042019e-05,
      "loss": 0.9064,
      "step": 5040
    },
    {
      "epoch": 0.33307500783220934,
      "grad_norm": 0.49405619502067566,
      "learning_rate": 4.932980390833275e-05,
      "loss": 0.8532,
      "step": 5050
    },
    {
      "epoch": 0.33373456230316423,
      "grad_norm": 0.5631572604179382,
      "learning_rate": 4.9323168885920604e-05,
      "loss": 0.8464,
      "step": 5060
    },
    {
      "epoch": 0.33439411677411907,
      "grad_norm": 0.5703868269920349,
      "learning_rate": 4.9316501631976056e-05,
      "loss": 0.8859,
      "step": 5070
    },
    {
      "epoch": 0.33505367124507396,
      "grad_norm": 0.5599934458732605,
      "learning_rate": 4.930980215533413e-05,
      "loss": 0.8796,
      "step": 5080
    },
    {
      "epoch": 0.3357132257160288,
      "grad_norm": 0.5828800797462463,
      "learning_rate": 4.9303070464872556e-05,
      "loss": 0.8847,
      "step": 5090
    },
    {
      "epoch": 0.3363727801869837,
      "grad_norm": 0.5994510650634766,
      "learning_rate": 4.929630656951175e-05,
      "loss": 0.914,
      "step": 5100
    },
    {
      "epoch": 0.3370323346579386,
      "grad_norm": 0.5373006463050842,
      "learning_rate": 4.9289510478214795e-05,
      "loss": 0.9014,
      "step": 5110
    },
    {
      "epoch": 0.3376918891288934,
      "grad_norm": 0.5972238779067993,
      "learning_rate": 4.9282682199987446e-05,
      "loss": 0.8267,
      "step": 5120
    },
    {
      "epoch": 0.3383514435998483,
      "grad_norm": 0.5829002261161804,
      "learning_rate": 4.92758217438781e-05,
      "loss": 0.8404,
      "step": 5130
    },
    {
      "epoch": 0.33901099807080315,
      "grad_norm": 0.5077376961708069,
      "learning_rate": 4.926892911897782e-05,
      "loss": 0.8797,
      "step": 5140
    },
    {
      "epoch": 0.33967055254175804,
      "grad_norm": 0.5396333932876587,
      "learning_rate": 4.926200433442025e-05,
      "loss": 0.8627,
      "step": 5150
    },
    {
      "epoch": 0.34033010701271293,
      "grad_norm": 0.5384257435798645,
      "learning_rate": 4.92550473993817e-05,
      "loss": 0.901,
      "step": 5160
    },
    {
      "epoch": 0.34098966148366777,
      "grad_norm": 0.5948606729507446,
      "learning_rate": 4.924805832308106e-05,
      "loss": 0.8521,
      "step": 5170
    },
    {
      "epoch": 0.34164921595462266,
      "grad_norm": 0.5079233646392822,
      "learning_rate": 4.92410371147798e-05,
      "loss": 0.8591,
      "step": 5180
    },
    {
      "epoch": 0.3423087704255775,
      "grad_norm": 0.49939635396003723,
      "learning_rate": 4.9233983783782005e-05,
      "loss": 0.8761,
      "step": 5190
    },
    {
      "epoch": 0.3429683248965324,
      "grad_norm": 0.5150532126426697,
      "learning_rate": 4.9226898339434294e-05,
      "loss": 0.8774,
      "step": 5200
    },
    {
      "epoch": 0.3436278793674873,
      "grad_norm": 0.5863020420074463,
      "learning_rate": 4.921978079112585e-05,
      "loss": 0.8877,
      "step": 5210
    },
    {
      "epoch": 0.3442874338384421,
      "grad_norm": 0.6055648326873779,
      "learning_rate": 4.92126311482884e-05,
      "loss": 0.8641,
      "step": 5220
    },
    {
      "epoch": 0.344946988309397,
      "grad_norm": 0.5826128125190735,
      "learning_rate": 4.92054494203962e-05,
      "loss": 0.8957,
      "step": 5230
    },
    {
      "epoch": 0.34560654278035186,
      "grad_norm": 0.47612279653549194,
      "learning_rate": 4.9198235616966034e-05,
      "loss": 0.8525,
      "step": 5240
    },
    {
      "epoch": 0.34626609725130675,
      "grad_norm": 0.5482989549636841,
      "learning_rate": 4.919098974755717e-05,
      "loss": 0.9164,
      "step": 5250
    },
    {
      "epoch": 0.3469256517222616,
      "grad_norm": 0.5377515554428101,
      "learning_rate": 4.918371182177138e-05,
      "loss": 0.867,
      "step": 5260
    },
    {
      "epoch": 0.3475852061932165,
      "grad_norm": 0.6391288638114929,
      "learning_rate": 4.917640184925292e-05,
      "loss": 0.8724,
      "step": 5270
    },
    {
      "epoch": 0.34824476066417137,
      "grad_norm": 0.550122082233429,
      "learning_rate": 4.916905983968849e-05,
      "loss": 0.8763,
      "step": 5280
    },
    {
      "epoch": 0.3489043151351262,
      "grad_norm": 0.5827004313468933,
      "learning_rate": 4.9161685802807276e-05,
      "loss": 0.8563,
      "step": 5290
    },
    {
      "epoch": 0.3495638696060811,
      "grad_norm": 0.5390400886535645,
      "learning_rate": 4.9154279748380884e-05,
      "loss": 0.8564,
      "step": 5300
    },
    {
      "epoch": 0.35022342407703594,
      "grad_norm": 0.4911317229270935,
      "learning_rate": 4.9146841686223336e-05,
      "loss": 0.8702,
      "step": 5310
    },
    {
      "epoch": 0.35088297854799083,
      "grad_norm": 0.5611599087715149,
      "learning_rate": 4.91393716261911e-05,
      "loss": 0.8464,
      "step": 5320
    },
    {
      "epoch": 0.3515425330189457,
      "grad_norm": 0.6718890070915222,
      "learning_rate": 4.913186957818303e-05,
      "loss": 0.8805,
      "step": 5330
    },
    {
      "epoch": 0.35220208748990056,
      "grad_norm": 0.6611254215240479,
      "learning_rate": 4.9124335552140353e-05,
      "loss": 0.8526,
      "step": 5340
    },
    {
      "epoch": 0.35286164196085545,
      "grad_norm": 0.5236311554908752,
      "learning_rate": 4.91167695580467e-05,
      "loss": 0.8691,
      "step": 5350
    },
    {
      "epoch": 0.3535211964318103,
      "grad_norm": 0.6203802824020386,
      "learning_rate": 4.910917160592804e-05,
      "loss": 0.859,
      "step": 5360
    },
    {
      "epoch": 0.3541807509027652,
      "grad_norm": 0.4989217519760132,
      "learning_rate": 4.910154170585272e-05,
      "loss": 0.9082,
      "step": 5370
    },
    {
      "epoch": 0.3548403053737201,
      "grad_norm": 0.5358820557594299,
      "learning_rate": 4.909387986793137e-05,
      "loss": 0.8199,
      "step": 5380
    },
    {
      "epoch": 0.3554998598446749,
      "grad_norm": 0.5449334979057312,
      "learning_rate": 4.908618610231701e-05,
      "loss": 0.866,
      "step": 5390
    },
    {
      "epoch": 0.3561594143156298,
      "grad_norm": 0.5716169476509094,
      "learning_rate": 4.9078460419204905e-05,
      "loss": 0.8263,
      "step": 5400
    },
    {
      "epoch": 0.35681896878658464,
      "grad_norm": 0.5559737682342529,
      "learning_rate": 4.907070282883267e-05,
      "loss": 0.8876,
      "step": 5410
    },
    {
      "epoch": 0.35747852325753954,
      "grad_norm": 0.5706018805503845,
      "learning_rate": 4.9062913341480163e-05,
      "loss": 0.8184,
      "step": 5420
    },
    {
      "epoch": 0.3581380777284944,
      "grad_norm": 0.4763959050178528,
      "learning_rate": 4.905509196746952e-05,
      "loss": 0.8376,
      "step": 5430
    },
    {
      "epoch": 0.35879763219944927,
      "grad_norm": 0.6196017265319824,
      "learning_rate": 4.9047238717165135e-05,
      "loss": 0.8774,
      "step": 5440
    },
    {
      "epoch": 0.35945718667040416,
      "grad_norm": 0.5639172196388245,
      "learning_rate": 4.9039353600973645e-05,
      "loss": 0.8914,
      "step": 5450
    },
    {
      "epoch": 0.360116741141359,
      "grad_norm": 0.539543867111206,
      "learning_rate": 4.90314366293439e-05,
      "loss": 0.8769,
      "step": 5460
    },
    {
      "epoch": 0.3607762956123139,
      "grad_norm": 0.5149198770523071,
      "learning_rate": 4.902348781276699e-05,
      "loss": 0.8533,
      "step": 5470
    },
    {
      "epoch": 0.36143585008326873,
      "grad_norm": 0.5052158832550049,
      "learning_rate": 4.901550716177616e-05,
      "loss": 0.8773,
      "step": 5480
    },
    {
      "epoch": 0.3620954045542236,
      "grad_norm": 0.48542529344558716,
      "learning_rate": 4.9007494686946896e-05,
      "loss": 0.8475,
      "step": 5490
    },
    {
      "epoch": 0.3627549590251785,
      "grad_norm": 0.5748151540756226,
      "learning_rate": 4.8999450398896804e-05,
      "loss": 0.8471,
      "step": 5500
    },
    {
      "epoch": 0.36341451349613335,
      "grad_norm": 0.5332309007644653,
      "learning_rate": 4.899137430828568e-05,
      "loss": 0.8578,
      "step": 5510
    },
    {
      "epoch": 0.36407406796708824,
      "grad_norm": 0.5747234225273132,
      "learning_rate": 4.8983266425815446e-05,
      "loss": 0.8573,
      "step": 5520
    },
    {
      "epoch": 0.3647336224380431,
      "grad_norm": 0.5445331335067749,
      "learning_rate": 4.897512676223016e-05,
      "loss": 0.8938,
      "step": 5530
    },
    {
      "epoch": 0.365393176908998,
      "grad_norm": 0.5994349122047424,
      "learning_rate": 4.8966955328315985e-05,
      "loss": 0.8726,
      "step": 5540
    },
    {
      "epoch": 0.36605273137995287,
      "grad_norm": 0.656285285949707,
      "learning_rate": 4.8958752134901196e-05,
      "loss": 0.8654,
      "step": 5550
    },
    {
      "epoch": 0.3667122858509077,
      "grad_norm": 0.4907941222190857,
      "learning_rate": 4.8950517192856146e-05,
      "loss": 0.8306,
      "step": 5560
    },
    {
      "epoch": 0.3673718403218626,
      "grad_norm": 0.544032633304596,
      "learning_rate": 4.894225051309327e-05,
      "loss": 0.9078,
      "step": 5570
    },
    {
      "epoch": 0.36803139479281743,
      "grad_norm": 0.5852176547050476,
      "learning_rate": 4.893395210656704e-05,
      "loss": 0.8421,
      "step": 5580
    },
    {
      "epoch": 0.3686909492637723,
      "grad_norm": 0.49195295572280884,
      "learning_rate": 4.8925621984274e-05,
      "loss": 0.8807,
      "step": 5590
    },
    {
      "epoch": 0.36935050373472716,
      "grad_norm": 0.6170905828475952,
      "learning_rate": 4.8917260157252674e-05,
      "loss": 0.8737,
      "step": 5600
    },
    {
      "epoch": 0.37001005820568206,
      "grad_norm": 0.5538198351860046,
      "learning_rate": 4.890886663658366e-05,
      "loss": 0.896,
      "step": 5610
    },
    {
      "epoch": 0.37066961267663695,
      "grad_norm": 0.5857877135276794,
      "learning_rate": 4.89004414333895e-05,
      "loss": 0.895,
      "step": 5620
    },
    {
      "epoch": 0.3713291671475918,
      "grad_norm": 0.46444687247276306,
      "learning_rate": 4.889198455883476e-05,
      "loss": 0.8998,
      "step": 5630
    },
    {
      "epoch": 0.3719887216185467,
      "grad_norm": 0.6267735362052917,
      "learning_rate": 4.888349602412595e-05,
      "loss": 0.8512,
      "step": 5640
    },
    {
      "epoch": 0.3726482760895015,
      "grad_norm": 0.5671398043632507,
      "learning_rate": 4.887497584051154e-05,
      "loss": 0.8621,
      "step": 5650
    },
    {
      "epoch": 0.3733078305604564,
      "grad_norm": 0.596243679523468,
      "learning_rate": 4.8866424019281935e-05,
      "loss": 0.8371,
      "step": 5660
    },
    {
      "epoch": 0.3739673850314113,
      "grad_norm": 0.5437570810317993,
      "learning_rate": 4.885784057176949e-05,
      "loss": 0.8937,
      "step": 5670
    },
    {
      "epoch": 0.37462693950236614,
      "grad_norm": 0.5501529574394226,
      "learning_rate": 4.884922550934843e-05,
      "loss": 0.8591,
      "step": 5680
    },
    {
      "epoch": 0.37528649397332103,
      "grad_norm": 0.5578798651695251,
      "learning_rate": 4.884057884343491e-05,
      "loss": 0.8518,
      "step": 5690
    },
    {
      "epoch": 0.37594604844427587,
      "grad_norm": 0.56142258644104,
      "learning_rate": 4.8831900585486936e-05,
      "loss": 0.9065,
      "step": 5700
    },
    {
      "epoch": 0.37660560291523076,
      "grad_norm": 0.5245373845100403,
      "learning_rate": 4.882319074700439e-05,
      "loss": 0.8545,
      "step": 5710
    },
    {
      "epoch": 0.37726515738618566,
      "grad_norm": 0.5014517903327942,
      "learning_rate": 4.8814449339529e-05,
      "loss": 0.8646,
      "step": 5720
    },
    {
      "epoch": 0.3779247118571405,
      "grad_norm": 0.5961427688598633,
      "learning_rate": 4.880567637464434e-05,
      "loss": 0.855,
      "step": 5730
    },
    {
      "epoch": 0.3785842663280954,
      "grad_norm": 0.5327067971229553,
      "learning_rate": 4.8796871863975776e-05,
      "loss": 0.8777,
      "step": 5740
    },
    {
      "epoch": 0.3792438207990502,
      "grad_norm": 0.4703054130077362,
      "learning_rate": 4.8788035819190504e-05,
      "loss": 0.8834,
      "step": 5750
    },
    {
      "epoch": 0.3799033752700051,
      "grad_norm": 0.6308974027633667,
      "learning_rate": 4.877916825199749e-05,
      "loss": 0.8532,
      "step": 5760
    },
    {
      "epoch": 0.38056292974095995,
      "grad_norm": 0.45056989789009094,
      "learning_rate": 4.8770269174147465e-05,
      "loss": 0.8441,
      "step": 5770
    },
    {
      "epoch": 0.38122248421191485,
      "grad_norm": 0.5260857939720154,
      "learning_rate": 4.8761338597432945e-05,
      "loss": 0.8776,
      "step": 5780
    },
    {
      "epoch": 0.38188203868286974,
      "grad_norm": 0.5394748449325562,
      "learning_rate": 4.875237653368816e-05,
      "loss": 0.8358,
      "step": 5790
    },
    {
      "epoch": 0.3825415931538246,
      "grad_norm": 0.5730496048927307,
      "learning_rate": 4.874338299478907e-05,
      "loss": 0.9312,
      "step": 5800
    },
    {
      "epoch": 0.38320114762477947,
      "grad_norm": 0.5718327760696411,
      "learning_rate": 4.8734357992653355e-05,
      "loss": 0.8601,
      "step": 5810
    },
    {
      "epoch": 0.3838607020957343,
      "grad_norm": 0.5419636368751526,
      "learning_rate": 4.872530153924037e-05,
      "loss": 0.857,
      "step": 5820
    },
    {
      "epoch": 0.3845202565666892,
      "grad_norm": 0.63425213098526,
      "learning_rate": 4.871621364655116e-05,
      "loss": 0.888,
      "step": 5830
    },
    {
      "epoch": 0.3851798110376441,
      "grad_norm": 0.567084789276123,
      "learning_rate": 4.870709432662843e-05,
      "loss": 0.8543,
      "step": 5840
    },
    {
      "epoch": 0.38583936550859893,
      "grad_norm": 0.7061660289764404,
      "learning_rate": 4.869794359155653e-05,
      "loss": 0.8304,
      "step": 5850
    },
    {
      "epoch": 0.3864989199795538,
      "grad_norm": 0.5780564546585083,
      "learning_rate": 4.868876145346144e-05,
      "loss": 0.9157,
      "step": 5860
    },
    {
      "epoch": 0.38715847445050866,
      "grad_norm": 0.6022658348083496,
      "learning_rate": 4.8679547924510735e-05,
      "loss": 0.8334,
      "step": 5870
    },
    {
      "epoch": 0.38781802892146355,
      "grad_norm": 0.5134875774383545,
      "learning_rate": 4.867030301691363e-05,
      "loss": 0.8873,
      "step": 5880
    },
    {
      "epoch": 0.38847758339241845,
      "grad_norm": 0.5205914378166199,
      "learning_rate": 4.866102674292088e-05,
      "loss": 0.8931,
      "step": 5890
    },
    {
      "epoch": 0.3891371378633733,
      "grad_norm": 0.5714764595031738,
      "learning_rate": 4.865171911482481e-05,
      "loss": 0.86,
      "step": 5900
    },
    {
      "epoch": 0.3897966923343282,
      "grad_norm": 0.5874856114387512,
      "learning_rate": 4.864238014495933e-05,
      "loss": 0.862,
      "step": 5910
    },
    {
      "epoch": 0.390456246805283,
      "grad_norm": 0.5221552848815918,
      "learning_rate": 4.8633009845699824e-05,
      "loss": 0.8148,
      "step": 5920
    },
    {
      "epoch": 0.3911158012762379,
      "grad_norm": 0.5328670144081116,
      "learning_rate": 4.862360822946324e-05,
      "loss": 0.8696,
      "step": 5930
    },
    {
      "epoch": 0.39177535574719274,
      "grad_norm": 0.5030723214149475,
      "learning_rate": 4.861417530870801e-05,
      "loss": 0.8793,
      "step": 5940
    },
    {
      "epoch": 0.39243491021814764,
      "grad_norm": 0.5878767967224121,
      "learning_rate": 4.860471109593404e-05,
      "loss": 0.8511,
      "step": 5950
    },
    {
      "epoch": 0.39309446468910253,
      "grad_norm": 0.441631019115448,
      "learning_rate": 4.859521560368271e-05,
      "loss": 0.9012,
      "step": 5960
    },
    {
      "epoch": 0.39375401916005737,
      "grad_norm": 0.5347902774810791,
      "learning_rate": 4.858568884453685e-05,
      "loss": 0.9162,
      "step": 5970
    },
    {
      "epoch": 0.39441357363101226,
      "grad_norm": 0.49979016184806824,
      "learning_rate": 4.857613083112072e-05,
      "loss": 0.8954,
      "step": 5980
    },
    {
      "epoch": 0.3950731281019671,
      "grad_norm": 0.5070289969444275,
      "learning_rate": 4.85665415761e-05,
      "loss": 0.9112,
      "step": 5990
    },
    {
      "epoch": 0.395732682572922,
      "grad_norm": 0.49998021125793457,
      "learning_rate": 4.855692109218177e-05,
      "loss": 0.849,
      "step": 6000
    },
    {
      "epoch": 0.3963922370438769,
      "grad_norm": 0.5425897836685181,
      "learning_rate": 4.854726939211448e-05,
      "loss": 0.8485,
      "step": 6010
    },
    {
      "epoch": 0.3970517915148317,
      "grad_norm": 0.5229442119598389,
      "learning_rate": 4.8537586488687964e-05,
      "loss": 0.8821,
      "step": 6020
    },
    {
      "epoch": 0.3977113459857866,
      "grad_norm": 0.5876744985580444,
      "learning_rate": 4.852787239473339e-05,
      "loss": 0.8534,
      "step": 6030
    },
    {
      "epoch": 0.39837090045674145,
      "grad_norm": 0.5264390707015991,
      "learning_rate": 4.851812712312327e-05,
      "loss": 0.8467,
      "step": 6040
    },
    {
      "epoch": 0.39903045492769634,
      "grad_norm": 0.48363423347473145,
      "learning_rate": 4.850835068677143e-05,
      "loss": 0.8573,
      "step": 6050
    },
    {
      "epoch": 0.39969000939865124,
      "grad_norm": 0.5530417561531067,
      "learning_rate": 4.849854309863297e-05,
      "loss": 0.9285,
      "step": 6060
    },
    {
      "epoch": 0.4003495638696061,
      "grad_norm": 0.5027384161949158,
      "learning_rate": 4.84887043717043e-05,
      "loss": 0.8676,
      "step": 6070
    },
    {
      "epoch": 0.40100911834056097,
      "grad_norm": 0.546110212802887,
      "learning_rate": 4.8478834519023084e-05,
      "loss": 0.8827,
      "step": 6080
    },
    {
      "epoch": 0.4016686728115158,
      "grad_norm": 0.614757776260376,
      "learning_rate": 4.8468933553668225e-05,
      "loss": 0.8417,
      "step": 6090
    },
    {
      "epoch": 0.4023282272824707,
      "grad_norm": 0.6290607452392578,
      "learning_rate": 4.845900148875986e-05,
      "loss": 0.8582,
      "step": 6100
    },
    {
      "epoch": 0.40298778175342553,
      "grad_norm": 0.5472835898399353,
      "learning_rate": 4.844903833745933e-05,
      "loss": 0.8602,
      "step": 6110
    },
    {
      "epoch": 0.4036473362243804,
      "grad_norm": 0.5838117599487305,
      "learning_rate": 4.8439044112969186e-05,
      "loss": 0.8411,
      "step": 6120
    },
    {
      "epoch": 0.4043068906953353,
      "grad_norm": 0.4958944022655487,
      "learning_rate": 4.842901882853314e-05,
      "loss": 0.8941,
      "step": 6130
    },
    {
      "epoch": 0.40496644516629016,
      "grad_norm": 0.5282686352729797,
      "learning_rate": 4.841896249743607e-05,
      "loss": 0.8996,
      "step": 6140
    },
    {
      "epoch": 0.40562599963724505,
      "grad_norm": 0.4729621410369873,
      "learning_rate": 4.8408875133003995e-05,
      "loss": 0.8869,
      "step": 6150
    },
    {
      "epoch": 0.4062855541081999,
      "grad_norm": 0.4891512095928192,
      "learning_rate": 4.839875674860405e-05,
      "loss": 0.9026,
      "step": 6160
    },
    {
      "epoch": 0.4069451085791548,
      "grad_norm": 0.5252025127410889,
      "learning_rate": 4.8388607357644475e-05,
      "loss": 0.8563,
      "step": 6170
    },
    {
      "epoch": 0.4076046630501097,
      "grad_norm": 0.5011799931526184,
      "learning_rate": 4.837842697357462e-05,
      "loss": 0.8434,
      "step": 6180
    },
    {
      "epoch": 0.4082642175210645,
      "grad_norm": 0.542057454586029,
      "learning_rate": 4.836821560988489e-05,
      "loss": 0.8237,
      "step": 6190
    },
    {
      "epoch": 0.4089237719920194,
      "grad_norm": 0.5338612794876099,
      "learning_rate": 4.835797328010672e-05,
      "loss": 0.8907,
      "step": 6200
    },
    {
      "epoch": 0.40958332646297424,
      "grad_norm": 0.5403425693511963,
      "learning_rate": 4.834769999781262e-05,
      "loss": 0.8345,
      "step": 6210
    },
    {
      "epoch": 0.41024288093392913,
      "grad_norm": 0.5578640699386597,
      "learning_rate": 4.833739577661609e-05,
      "loss": 0.8822,
      "step": 6220
    },
    {
      "epoch": 0.410902435404884,
      "grad_norm": 0.556004524230957,
      "learning_rate": 4.832706063017164e-05,
      "loss": 0.8276,
      "step": 6230
    },
    {
      "epoch": 0.41156198987583886,
      "grad_norm": 0.5602332949638367,
      "learning_rate": 4.831669457217475e-05,
      "loss": 0.8635,
      "step": 6240
    },
    {
      "epoch": 0.41222154434679376,
      "grad_norm": 0.5787188410758972,
      "learning_rate": 4.830629761636186e-05,
      "loss": 0.8235,
      "step": 6250
    },
    {
      "epoch": 0.4128810988177486,
      "grad_norm": 0.6306541562080383,
      "learning_rate": 4.829586977651038e-05,
      "loss": 0.876,
      "step": 6260
    },
    {
      "epoch": 0.4135406532887035,
      "grad_norm": 0.5497257113456726,
      "learning_rate": 4.828541106643862e-05,
      "loss": 0.7759,
      "step": 6270
    },
    {
      "epoch": 0.4142002077596583,
      "grad_norm": 0.48434269428253174,
      "learning_rate": 4.82749215000058e-05,
      "loss": 0.9067,
      "step": 6280
    },
    {
      "epoch": 0.4148597622306132,
      "grad_norm": 0.5243717432022095,
      "learning_rate": 4.8264401091112035e-05,
      "loss": 0.8673,
      "step": 6290
    },
    {
      "epoch": 0.4155193167015681,
      "grad_norm": 0.5414081811904907,
      "learning_rate": 4.8253849853698307e-05,
      "loss": 0.8885,
      "step": 6300
    },
    {
      "epoch": 0.41617887117252295,
      "grad_norm": 0.5486317873001099,
      "learning_rate": 4.824326780174645e-05,
      "loss": 0.8785,
      "step": 6310
    },
    {
      "epoch": 0.41683842564347784,
      "grad_norm": 0.5972316861152649,
      "learning_rate": 4.8232654949279145e-05,
      "loss": 0.8628,
      "step": 6320
    },
    {
      "epoch": 0.4174979801144327,
      "grad_norm": 0.4799783229827881,
      "learning_rate": 4.822201131035986e-05,
      "loss": 0.8222,
      "step": 6330
    },
    {
      "epoch": 0.41815753458538757,
      "grad_norm": 0.4656096398830414,
      "learning_rate": 4.821133689909288e-05,
      "loss": 0.8367,
      "step": 6340
    },
    {
      "epoch": 0.41881708905634246,
      "grad_norm": 0.4220138490200043,
      "learning_rate": 4.820063172962328e-05,
      "loss": 0.8608,
      "step": 6350
    },
    {
      "epoch": 0.4194766435272973,
      "grad_norm": 0.5239322185516357,
      "learning_rate": 4.8189895816136855e-05,
      "loss": 0.8596,
      "step": 6360
    },
    {
      "epoch": 0.4201361979982522,
      "grad_norm": 0.562239944934845,
      "learning_rate": 4.8179129172860173e-05,
      "loss": 0.892,
      "step": 6370
    },
    {
      "epoch": 0.42079575246920703,
      "grad_norm": 0.5668872594833374,
      "learning_rate": 4.8168331814060505e-05,
      "loss": 0.8036,
      "step": 6380
    },
    {
      "epoch": 0.4214553069401619,
      "grad_norm": 0.5321053266525269,
      "learning_rate": 4.8157503754045835e-05,
      "loss": 0.8519,
      "step": 6390
    },
    {
      "epoch": 0.4221148614111168,
      "grad_norm": 0.5495429635047913,
      "learning_rate": 4.814664500716483e-05,
      "loss": 0.8603,
      "step": 6400
    },
    {
      "epoch": 0.42277441588207165,
      "grad_norm": 0.5138208270072937,
      "learning_rate": 4.8135755587806806e-05,
      "loss": 0.8398,
      "step": 6410
    },
    {
      "epoch": 0.42343397035302655,
      "grad_norm": 0.549922525882721,
      "learning_rate": 4.812483551040176e-05,
      "loss": 0.8564,
      "step": 6420
    },
    {
      "epoch": 0.4240935248239814,
      "grad_norm": 0.49002325534820557,
      "learning_rate": 4.811388478942026e-05,
      "loss": 0.8315,
      "step": 6430
    },
    {
      "epoch": 0.4247530792949363,
      "grad_norm": 0.43466535210609436,
      "learning_rate": 4.810290343937353e-05,
      "loss": 0.8757,
      "step": 6440
    },
    {
      "epoch": 0.4254126337658911,
      "grad_norm": 0.5885916948318481,
      "learning_rate": 4.8091891474813366e-05,
      "loss": 0.8468,
      "step": 6450
    },
    {
      "epoch": 0.426072188236846,
      "grad_norm": 0.6256899237632751,
      "learning_rate": 4.808084891033212e-05,
      "loss": 0.8636,
      "step": 6460
    },
    {
      "epoch": 0.4267317427078009,
      "grad_norm": 0.49635109305381775,
      "learning_rate": 4.806977576056272e-05,
      "loss": 0.8742,
      "step": 6470
    },
    {
      "epoch": 0.42739129717875574,
      "grad_norm": 0.5088977217674255,
      "learning_rate": 4.8058672040178586e-05,
      "loss": 0.8276,
      "step": 6480
    },
    {
      "epoch": 0.42805085164971063,
      "grad_norm": 0.4978698194026947,
      "learning_rate": 4.804753776389369e-05,
      "loss": 0.8841,
      "step": 6490
    },
    {
      "epoch": 0.42871040612066547,
      "grad_norm": 0.5396628379821777,
      "learning_rate": 4.803637294646245e-05,
      "loss": 0.8505,
      "step": 6500
    },
    {
      "epoch": 0.42936996059162036,
      "grad_norm": 0.5257627367973328,
      "learning_rate": 4.8025177602679805e-05,
      "loss": 0.8388,
      "step": 6510
    },
    {
      "epoch": 0.43002951506257525,
      "grad_norm": 0.4984934628009796,
      "learning_rate": 4.8013951747381104e-05,
      "loss": 0.8656,
      "step": 6520
    },
    {
      "epoch": 0.4306890695335301,
      "grad_norm": 0.5394895076751709,
      "learning_rate": 4.800269539544214e-05,
      "loss": 0.9021,
      "step": 6530
    },
    {
      "epoch": 0.431348624004485,
      "grad_norm": 0.5981066226959229,
      "learning_rate": 4.799140856177914e-05,
      "loss": 0.847,
      "step": 6540
    },
    {
      "epoch": 0.4320081784754398,
      "grad_norm": 0.5400550365447998,
      "learning_rate": 4.798009126134869e-05,
      "loss": 0.8804,
      "step": 6550
    },
    {
      "epoch": 0.4326677329463947,
      "grad_norm": 0.5571463108062744,
      "learning_rate": 4.7968743509147776e-05,
      "loss": 0.8671,
      "step": 6560
    },
    {
      "epoch": 0.4333272874173496,
      "grad_norm": 0.5644498467445374,
      "learning_rate": 4.795736532021372e-05,
      "loss": 0.8762,
      "step": 6570
    },
    {
      "epoch": 0.43398684188830444,
      "grad_norm": 0.4889192581176758,
      "learning_rate": 4.794595670962417e-05,
      "loss": 0.8543,
      "step": 6580
    },
    {
      "epoch": 0.43464639635925933,
      "grad_norm": 0.6175575256347656,
      "learning_rate": 4.793451769249712e-05,
      "loss": 0.8644,
      "step": 6590
    },
    {
      "epoch": 0.43530595083021417,
      "grad_norm": 0.548516035079956,
      "learning_rate": 4.7923048283990814e-05,
      "loss": 0.9286,
      "step": 6600
    },
    {
      "epoch": 0.43596550530116907,
      "grad_norm": 0.5567188262939453,
      "learning_rate": 4.7911548499303806e-05,
      "loss": 0.8522,
      "step": 6610
    },
    {
      "epoch": 0.43662505977212396,
      "grad_norm": 0.602759599685669,
      "learning_rate": 4.790001835367488e-05,
      "loss": 0.8517,
      "step": 6620
    },
    {
      "epoch": 0.4372846142430788,
      "grad_norm": 0.5450781583786011,
      "learning_rate": 4.7888457862383064e-05,
      "loss": 0.8317,
      "step": 6630
    },
    {
      "epoch": 0.4379441687140337,
      "grad_norm": 0.5020095705986023,
      "learning_rate": 4.787686704074759e-05,
      "loss": 0.8516,
      "step": 6640
    },
    {
      "epoch": 0.4386037231849885,
      "grad_norm": 0.527065098285675,
      "learning_rate": 4.7865245904127874e-05,
      "loss": 0.839,
      "step": 6650
    },
    {
      "epoch": 0.4392632776559434,
      "grad_norm": 0.5484679937362671,
      "learning_rate": 4.7853594467923533e-05,
      "loss": 0.8759,
      "step": 6660
    },
    {
      "epoch": 0.43992283212689826,
      "grad_norm": 0.5647040605545044,
      "learning_rate": 4.784191274757431e-05,
      "loss": 0.8639,
      "step": 6670
    },
    {
      "epoch": 0.44058238659785315,
      "grad_norm": 0.5041146874427795,
      "learning_rate": 4.783020075856009e-05,
      "loss": 0.8089,
      "step": 6680
    },
    {
      "epoch": 0.44124194106880804,
      "grad_norm": 0.5434051156044006,
      "learning_rate": 4.781845851640086e-05,
      "loss": 0.8887,
      "step": 6690
    },
    {
      "epoch": 0.4419014955397629,
      "grad_norm": 0.5963016152381897,
      "learning_rate": 4.780668603665669e-05,
      "loss": 0.8474,
      "step": 6700
    },
    {
      "epoch": 0.44256105001071777,
      "grad_norm": 0.6325017213821411,
      "learning_rate": 4.7794883334927756e-05,
      "loss": 0.867,
      "step": 6710
    },
    {
      "epoch": 0.4432206044816726,
      "grad_norm": 0.514064610004425,
      "learning_rate": 4.778305042685422e-05,
      "loss": 0.8779,
      "step": 6720
    },
    {
      "epoch": 0.4438801589526275,
      "grad_norm": 0.5603863596916199,
      "learning_rate": 4.777118732811633e-05,
      "loss": 0.8183,
      "step": 6730
    },
    {
      "epoch": 0.4445397134235824,
      "grad_norm": 0.5008962750434875,
      "learning_rate": 4.7759294054434314e-05,
      "loss": 0.8398,
      "step": 6740
    },
    {
      "epoch": 0.44519926789453723,
      "grad_norm": 0.4778035581111908,
      "learning_rate": 4.774737062156839e-05,
      "loss": 0.8348,
      "step": 6750
    },
    {
      "epoch": 0.4458588223654921,
      "grad_norm": 0.4997784495353699,
      "learning_rate": 4.773541704531873e-05,
      "loss": 0.8549,
      "step": 6760
    },
    {
      "epoch": 0.44651837683644696,
      "grad_norm": 0.5219247341156006,
      "learning_rate": 4.772343334152548e-05,
      "loss": 0.8739,
      "step": 6770
    },
    {
      "epoch": 0.44717793130740185,
      "grad_norm": 0.5919073224067688,
      "learning_rate": 4.771141952606868e-05,
      "loss": 0.8334,
      "step": 6780
    },
    {
      "epoch": 0.44783748577835675,
      "grad_norm": 0.5725424885749817,
      "learning_rate": 4.7699375614868275e-05,
      "loss": 0.8633,
      "step": 6790
    },
    {
      "epoch": 0.4484970402493116,
      "grad_norm": 0.592842161655426,
      "learning_rate": 4.7687301623884105e-05,
      "loss": 0.8395,
      "step": 6800
    },
    {
      "epoch": 0.4491565947202665,
      "grad_norm": 0.4690439999103546,
      "learning_rate": 4.767519756911587e-05,
      "loss": 0.8987,
      "step": 6810
    },
    {
      "epoch": 0.4498161491912213,
      "grad_norm": 0.48259902000427246,
      "learning_rate": 4.766306346660309e-05,
      "loss": 0.8764,
      "step": 6820
    },
    {
      "epoch": 0.4504757036621762,
      "grad_norm": 0.565538763999939,
      "learning_rate": 4.765089933242511e-05,
      "loss": 0.8634,
      "step": 6830
    },
    {
      "epoch": 0.45113525813313105,
      "grad_norm": 0.5153759717941284,
      "learning_rate": 4.763870518270109e-05,
      "loss": 0.8678,
      "step": 6840
    },
    {
      "epoch": 0.45179481260408594,
      "grad_norm": 0.4666423797607422,
      "learning_rate": 4.762648103358995e-05,
      "loss": 0.8452,
      "step": 6850
    },
    {
      "epoch": 0.45245436707504083,
      "grad_norm": 0.543533980846405,
      "learning_rate": 4.761422690129035e-05,
      "loss": 0.8482,
      "step": 6860
    },
    {
      "epoch": 0.45311392154599567,
      "grad_norm": 0.5554654598236084,
      "learning_rate": 4.7601942802040696e-05,
      "loss": 0.8684,
      "step": 6870
    },
    {
      "epoch": 0.45377347601695056,
      "grad_norm": 0.5532326102256775,
      "learning_rate": 4.758962875211911e-05,
      "loss": 0.8608,
      "step": 6880
    },
    {
      "epoch": 0.4544330304879054,
      "grad_norm": 0.47074058651924133,
      "learning_rate": 4.757728476784339e-05,
      "loss": 0.8137,
      "step": 6890
    },
    {
      "epoch": 0.4550925849588603,
      "grad_norm": 0.4835397005081177,
      "learning_rate": 4.756491086557102e-05,
      "loss": 0.8586,
      "step": 6900
    },
    {
      "epoch": 0.4557521394298152,
      "grad_norm": 0.49764856696128845,
      "learning_rate": 4.7552507061699094e-05,
      "loss": 0.8595,
      "step": 6910
    },
    {
      "epoch": 0.45641169390077,
      "grad_norm": 0.5029451847076416,
      "learning_rate": 4.754007337266437e-05,
      "loss": 0.8581,
      "step": 6920
    },
    {
      "epoch": 0.4570712483717249,
      "grad_norm": 0.5425214767456055,
      "learning_rate": 4.752760981494319e-05,
      "loss": 0.8458,
      "step": 6930
    },
    {
      "epoch": 0.45773080284267975,
      "grad_norm": 0.48952221870422363,
      "learning_rate": 4.751511640505147e-05,
      "loss": 0.8655,
      "step": 6940
    },
    {
      "epoch": 0.45839035731363464,
      "grad_norm": 0.5727320909500122,
      "learning_rate": 4.750259315954469e-05,
      "loss": 0.8527,
      "step": 6950
    },
    {
      "epoch": 0.45904991178458954,
      "grad_norm": 0.4817902147769928,
      "learning_rate": 4.749004009501787e-05,
      "loss": 0.8289,
      "step": 6960
    },
    {
      "epoch": 0.4597094662555444,
      "grad_norm": 0.5458888411521912,
      "learning_rate": 4.7477457228105535e-05,
      "loss": 0.8764,
      "step": 6970
    },
    {
      "epoch": 0.46036902072649927,
      "grad_norm": 0.5658895373344421,
      "learning_rate": 4.746484457548171e-05,
      "loss": 0.8694,
      "step": 6980
    },
    {
      "epoch": 0.4610285751974541,
      "grad_norm": 0.5030274987220764,
      "learning_rate": 4.74522021538599e-05,
      "loss": 0.8161,
      "step": 6990
    },
    {
      "epoch": 0.461688129668409,
      "grad_norm": 0.6383715271949768,
      "learning_rate": 4.743952997999303e-05,
      "loss": 0.8224,
      "step": 7000
    },
    {
      "epoch": 0.46234768413936383,
      "grad_norm": 0.5009424090385437,
      "learning_rate": 4.7426828070673476e-05,
      "loss": 0.87,
      "step": 7010
    },
    {
      "epoch": 0.4630072386103187,
      "grad_norm": 0.5131643414497375,
      "learning_rate": 4.7414096442733015e-05,
      "loss": 0.8434,
      "step": 7020
    },
    {
      "epoch": 0.4636667930812736,
      "grad_norm": 0.516572117805481,
      "learning_rate": 4.740133511304278e-05,
      "loss": 0.8653,
      "step": 7030
    },
    {
      "epoch": 0.46432634755222846,
      "grad_norm": 0.5132176876068115,
      "learning_rate": 4.7388544098513295e-05,
      "loss": 0.8502,
      "step": 7040
    },
    {
      "epoch": 0.46498590202318335,
      "grad_norm": 0.5855733156204224,
      "learning_rate": 4.737572341609441e-05,
      "loss": 0.8573,
      "step": 7050
    },
    {
      "epoch": 0.4656454564941382,
      "grad_norm": 0.585620641708374,
      "learning_rate": 4.736287308277527e-05,
      "loss": 0.8575,
      "step": 7060
    },
    {
      "epoch": 0.4663050109650931,
      "grad_norm": 0.506710410118103,
      "learning_rate": 4.7349993115584354e-05,
      "loss": 0.8729,
      "step": 7070
    },
    {
      "epoch": 0.466964565436048,
      "grad_norm": 0.5265117883682251,
      "learning_rate": 4.733708353158936e-05,
      "loss": 0.8097,
      "step": 7080
    },
    {
      "epoch": 0.4676241199070028,
      "grad_norm": 0.5592107772827148,
      "learning_rate": 4.732414434789727e-05,
      "loss": 0.8626,
      "step": 7090
    },
    {
      "epoch": 0.4682836743779577,
      "grad_norm": 0.5142652988433838,
      "learning_rate": 4.7311175581654266e-05,
      "loss": 0.8759,
      "step": 7100
    },
    {
      "epoch": 0.46894322884891254,
      "grad_norm": 0.4814768433570862,
      "learning_rate": 4.729817725004575e-05,
      "loss": 0.8939,
      "step": 7110
    },
    {
      "epoch": 0.46960278331986743,
      "grad_norm": 0.4561408460140228,
      "learning_rate": 4.728514937029629e-05,
      "loss": 0.8657,
      "step": 7120
    },
    {
      "epoch": 0.4702623377908223,
      "grad_norm": 0.573836088180542,
      "learning_rate": 4.7272091959669606e-05,
      "loss": 0.8114,
      "step": 7130
    },
    {
      "epoch": 0.47092189226177716,
      "grad_norm": 0.5914710164070129,
      "learning_rate": 4.725900503546856e-05,
      "loss": 0.8626,
      "step": 7140
    },
    {
      "epoch": 0.47158144673273206,
      "grad_norm": 0.4675734341144562,
      "learning_rate": 4.724588861503513e-05,
      "loss": 0.8399,
      "step": 7150
    },
    {
      "epoch": 0.4722410012036869,
      "grad_norm": 0.5302778482437134,
      "learning_rate": 4.7232742715750356e-05,
      "loss": 0.8578,
      "step": 7160
    },
    {
      "epoch": 0.4729005556746418,
      "grad_norm": 0.48141634464263916,
      "learning_rate": 4.721956735503437e-05,
      "loss": 0.8934,
      "step": 7170
    },
    {
      "epoch": 0.4735601101455966,
      "grad_norm": 0.46267980337142944,
      "learning_rate": 4.7206362550346314e-05,
      "loss": 0.8764,
      "step": 7180
    },
    {
      "epoch": 0.4742196646165515,
      "grad_norm": 0.5466578006744385,
      "learning_rate": 4.719312831918437e-05,
      "loss": 0.8604,
      "step": 7190
    },
    {
      "epoch": 0.4748792190875064,
      "grad_norm": 0.5205459594726562,
      "learning_rate": 4.717986467908573e-05,
      "loss": 0.8327,
      "step": 7200
    },
    {
      "epoch": 0.47553877355846125,
      "grad_norm": 0.6165599822998047,
      "learning_rate": 4.71665716476265e-05,
      "loss": 0.8117,
      "step": 7210
    },
    {
      "epoch": 0.47619832802941614,
      "grad_norm": 0.5680373311042786,
      "learning_rate": 4.7153249242421794e-05,
      "loss": 0.8502,
      "step": 7220
    },
    {
      "epoch": 0.476857882500371,
      "grad_norm": 0.46300211548805237,
      "learning_rate": 4.713989748112561e-05,
      "loss": 0.8657,
      "step": 7230
    },
    {
      "epoch": 0.47751743697132587,
      "grad_norm": 0.5191219449043274,
      "learning_rate": 4.7126516381430877e-05,
      "loss": 0.8613,
      "step": 7240
    },
    {
      "epoch": 0.47817699144228076,
      "grad_norm": 0.5691511631011963,
      "learning_rate": 4.711310596106937e-05,
      "loss": 0.8751,
      "step": 7250
    },
    {
      "epoch": 0.4788365459132356,
      "grad_norm": 0.5520512461662292,
      "learning_rate": 4.7099666237811736e-05,
      "loss": 0.8751,
      "step": 7260
    },
    {
      "epoch": 0.4794961003841905,
      "grad_norm": 0.5244108438491821,
      "learning_rate": 4.708619722946746e-05,
      "loss": 0.8205,
      "step": 7270
    },
    {
      "epoch": 0.48015565485514533,
      "grad_norm": 0.4873800277709961,
      "learning_rate": 4.707269895388482e-05,
      "loss": 0.8659,
      "step": 7280
    },
    {
      "epoch": 0.4808152093261002,
      "grad_norm": 0.5746051669120789,
      "learning_rate": 4.705917142895087e-05,
      "loss": 0.8204,
      "step": 7290
    },
    {
      "epoch": 0.4814747637970551,
      "grad_norm": 0.506810188293457,
      "learning_rate": 4.7045614672591466e-05,
      "loss": 0.8652,
      "step": 7300
    },
    {
      "epoch": 0.48213431826800995,
      "grad_norm": 0.46993669867515564,
      "learning_rate": 4.703202870277113e-05,
      "loss": 0.861,
      "step": 7310
    },
    {
      "epoch": 0.48279387273896485,
      "grad_norm": 0.5876075625419617,
      "learning_rate": 4.701841353749316e-05,
      "loss": 0.8545,
      "step": 7320
    },
    {
      "epoch": 0.4834534272099197,
      "grad_norm": 0.5827206969261169,
      "learning_rate": 4.7004769194799495e-05,
      "loss": 0.8505,
      "step": 7330
    },
    {
      "epoch": 0.4841129816808746,
      "grad_norm": 0.5576984882354736,
      "learning_rate": 4.699109569277078e-05,
      "loss": 0.8291,
      "step": 7340
    },
    {
      "epoch": 0.4847725361518294,
      "grad_norm": 0.5203624367713928,
      "learning_rate": 4.697739304952627e-05,
      "loss": 0.8648,
      "step": 7350
    },
    {
      "epoch": 0.4854320906227843,
      "grad_norm": 0.5325854420661926,
      "learning_rate": 4.696366128322385e-05,
      "loss": 0.8756,
      "step": 7360
    },
    {
      "epoch": 0.4860916450937392,
      "grad_norm": 0.5148476958274841,
      "learning_rate": 4.6949900412059976e-05,
      "loss": 0.8653,
      "step": 7370
    },
    {
      "epoch": 0.48675119956469404,
      "grad_norm": 0.5403642654418945,
      "learning_rate": 4.6936110454269706e-05,
      "loss": 0.8666,
      "step": 7380
    },
    {
      "epoch": 0.48741075403564893,
      "grad_norm": 0.6091113090515137,
      "learning_rate": 4.692229142812662e-05,
      "loss": 0.8594,
      "step": 7390
    },
    {
      "epoch": 0.48807030850660377,
      "grad_norm": 0.5950354337692261,
      "learning_rate": 4.6908443351942824e-05,
      "loss": 0.8594,
      "step": 7400
    },
    {
      "epoch": 0.48872986297755866,
      "grad_norm": 0.5460037589073181,
      "learning_rate": 4.68945662440689e-05,
      "loss": 0.8146,
      "step": 7410
    },
    {
      "epoch": 0.48938941744851355,
      "grad_norm": 0.4589973986148834,
      "learning_rate": 4.688066012289394e-05,
      "loss": 0.8469,
      "step": 7420
    },
    {
      "epoch": 0.4900489719194684,
      "grad_norm": 0.5626265406608582,
      "learning_rate": 4.686672500684545e-05,
      "loss": 0.8655,
      "step": 7430
    },
    {
      "epoch": 0.4907085263904233,
      "grad_norm": 0.5438959002494812,
      "learning_rate": 4.685276091438937e-05,
      "loss": 0.8631,
      "step": 7440
    },
    {
      "epoch": 0.4913680808613781,
      "grad_norm": 0.5762490034103394,
      "learning_rate": 4.683876786403003e-05,
      "loss": 0.8649,
      "step": 7450
    },
    {
      "epoch": 0.492027635332333,
      "grad_norm": 0.5192339420318604,
      "learning_rate": 4.6824745874310155e-05,
      "loss": 0.8716,
      "step": 7460
    },
    {
      "epoch": 0.4926871898032879,
      "grad_norm": 0.6523929238319397,
      "learning_rate": 4.6810694963810784e-05,
      "loss": 0.8636,
      "step": 7470
    },
    {
      "epoch": 0.49334674427424274,
      "grad_norm": 0.4878346920013428,
      "learning_rate": 4.679661515115132e-05,
      "loss": 0.8706,
      "step": 7480
    },
    {
      "epoch": 0.49400629874519764,
      "grad_norm": 0.5933941602706909,
      "learning_rate": 4.678250645498943e-05,
      "loss": 0.842,
      "step": 7490
    },
    {
      "epoch": 0.4946658532161525,
      "grad_norm": 0.5156781673431396,
      "learning_rate": 4.6768368894021074e-05,
      "loss": 0.8545,
      "step": 7500
    },
    {
      "epoch": 0.49532540768710737,
      "grad_norm": 0.5238683223724365,
      "learning_rate": 4.675420248698045e-05,
      "loss": 0.8971,
      "step": 7510
    },
    {
      "epoch": 0.4959849621580622,
      "grad_norm": 0.5473213791847229,
      "learning_rate": 4.6740007252640006e-05,
      "loss": 0.8299,
      "step": 7520
    },
    {
      "epoch": 0.4966445166290171,
      "grad_norm": 0.5568445920944214,
      "learning_rate": 4.672578320981036e-05,
      "loss": 0.9001,
      "step": 7530
    },
    {
      "epoch": 0.497304071099972,
      "grad_norm": 0.5153902769088745,
      "learning_rate": 4.671153037734032e-05,
      "loss": 0.8358,
      "step": 7540
    },
    {
      "epoch": 0.4979636255709268,
      "grad_norm": 0.6011136174201965,
      "learning_rate": 4.669724877411685e-05,
      "loss": 0.8451,
      "step": 7550
    },
    {
      "epoch": 0.4986231800418817,
      "grad_norm": 0.5671474933624268,
      "learning_rate": 4.668293841906502e-05,
      "loss": 0.8882,
      "step": 7560
    },
    {
      "epoch": 0.49928273451283656,
      "grad_norm": 0.5087859034538269,
      "learning_rate": 4.6668599331148025e-05,
      "loss": 0.817,
      "step": 7570
    },
    {
      "epoch": 0.49994228898379145,
      "grad_norm": 0.5660974383354187,
      "learning_rate": 4.66542315293671e-05,
      "loss": 0.8382,
      "step": 7580
    },
    {
      "epoch": 0.5006018434547463,
      "grad_norm": 0.5032976269721985,
      "learning_rate": 4.663983503276158e-05,
      "loss": 0.8495,
      "step": 7590
    },
    {
      "epoch": 0.5012613979257012,
      "grad_norm": 0.6020365357398987,
      "learning_rate": 4.662540986040878e-05,
      "loss": 0.8475,
      "step": 7600
    },
    {
      "epoch": 0.501920952396656,
      "grad_norm": 0.5403103232383728,
      "learning_rate": 4.6610956031424026e-05,
      "loss": 0.8442,
      "step": 7610
    },
    {
      "epoch": 0.5025805068676109,
      "grad_norm": 0.5416961908340454,
      "learning_rate": 4.659647356496063e-05,
      "loss": 0.8514,
      "step": 7620
    },
    {
      "epoch": 0.5032400613385658,
      "grad_norm": 0.5349296927452087,
      "learning_rate": 4.658196248020984e-05,
      "loss": 0.8336,
      "step": 7630
    },
    {
      "epoch": 0.5038996158095207,
      "grad_norm": 0.5633635520935059,
      "learning_rate": 4.6567422796400846e-05,
      "loss": 0.8008,
      "step": 7640
    },
    {
      "epoch": 0.5045591702804756,
      "grad_norm": 0.5842314958572388,
      "learning_rate": 4.6552854532800705e-05,
      "loss": 0.8616,
      "step": 7650
    },
    {
      "epoch": 0.5052187247514304,
      "grad_norm": 0.5005068182945251,
      "learning_rate": 4.6538257708714375e-05,
      "loss": 0.8432,
      "step": 7660
    },
    {
      "epoch": 0.5058782792223853,
      "grad_norm": 0.5463336110115051,
      "learning_rate": 4.652363234348464e-05,
      "loss": 0.8573,
      "step": 7670
    },
    {
      "epoch": 0.5065378336933402,
      "grad_norm": 0.5448721647262573,
      "learning_rate": 4.6508978456492115e-05,
      "loss": 0.9031,
      "step": 7680
    },
    {
      "epoch": 0.507197388164295,
      "grad_norm": 0.5476658940315247,
      "learning_rate": 4.649429606715521e-05,
      "loss": 0.8108,
      "step": 7690
    },
    {
      "epoch": 0.5078569426352499,
      "grad_norm": 0.5261156558990479,
      "learning_rate": 4.6479585194930105e-05,
      "loss": 0.8532,
      "step": 7700
    },
    {
      "epoch": 0.5085164971062047,
      "grad_norm": 0.7172783613204956,
      "learning_rate": 4.646484585931072e-05,
      "loss": 0.8915,
      "step": 7710
    },
    {
      "epoch": 0.5091760515771596,
      "grad_norm": 0.4560132324695587,
      "learning_rate": 4.64500780798287e-05,
      "loss": 0.8142,
      "step": 7720
    },
    {
      "epoch": 0.5098356060481145,
      "grad_norm": 0.5214531421661377,
      "learning_rate": 4.643528187605336e-05,
      "loss": 0.8289,
      "step": 7730
    },
    {
      "epoch": 0.5104951605190694,
      "grad_norm": 0.5078032612800598,
      "learning_rate": 4.6420457267591724e-05,
      "loss": 0.8697,
      "step": 7740
    },
    {
      "epoch": 0.5111547149900242,
      "grad_norm": 0.5201875567436218,
      "learning_rate": 4.64056042740884e-05,
      "loss": 0.8752,
      "step": 7750
    },
    {
      "epoch": 0.5118142694609791,
      "grad_norm": 0.4340056777000427,
      "learning_rate": 4.639072291522567e-05,
      "loss": 0.8809,
      "step": 7760
    },
    {
      "epoch": 0.512473823931934,
      "grad_norm": 0.5306442379951477,
      "learning_rate": 4.637581321072335e-05,
      "loss": 0.832,
      "step": 7770
    },
    {
      "epoch": 0.5131333784028889,
      "grad_norm": 0.48031529784202576,
      "learning_rate": 4.636087518033886e-05,
      "loss": 0.8208,
      "step": 7780
    },
    {
      "epoch": 0.5137929328738438,
      "grad_norm": 0.4846910834312439,
      "learning_rate": 4.6345908843867126e-05,
      "loss": 0.9174,
      "step": 7790
    },
    {
      "epoch": 0.5144524873447985,
      "grad_norm": 0.5625187158584595,
      "learning_rate": 4.6330914221140596e-05,
      "loss": 0.8346,
      "step": 7800
    },
    {
      "epoch": 0.5151120418157534,
      "grad_norm": 0.6281886100769043,
      "learning_rate": 4.631589133202921e-05,
      "loss": 0.8391,
      "step": 7810
    },
    {
      "epoch": 0.5157715962867083,
      "grad_norm": 0.516608715057373,
      "learning_rate": 4.630084019644034e-05,
      "loss": 0.8417,
      "step": 7820
    },
    {
      "epoch": 0.5164311507576632,
      "grad_norm": 0.5382763147354126,
      "learning_rate": 4.6285760834318814e-05,
      "loss": 0.8164,
      "step": 7830
    },
    {
      "epoch": 0.5170907052286181,
      "grad_norm": 0.4136751592159271,
      "learning_rate": 4.6270653265646855e-05,
      "loss": 0.8465,
      "step": 7840
    },
    {
      "epoch": 0.5177502596995729,
      "grad_norm": 0.4957684874534607,
      "learning_rate": 4.6255517510444054e-05,
      "loss": 0.8992,
      "step": 7850
    },
    {
      "epoch": 0.5184098141705278,
      "grad_norm": 0.4853816628456116,
      "learning_rate": 4.624035358876736e-05,
      "loss": 0.8319,
      "step": 7860
    },
    {
      "epoch": 0.5190693686414827,
      "grad_norm": 0.5266784429550171,
      "learning_rate": 4.622516152071106e-05,
      "loss": 0.893,
      "step": 7870
    },
    {
      "epoch": 0.5197289231124376,
      "grad_norm": 0.5038085579872131,
      "learning_rate": 4.620994132640671e-05,
      "loss": 0.8838,
      "step": 7880
    },
    {
      "epoch": 0.5203884775833925,
      "grad_norm": 0.511411190032959,
      "learning_rate": 4.619469302602317e-05,
      "loss": 0.8379,
      "step": 7890
    },
    {
      "epoch": 0.5210480320543472,
      "grad_norm": 0.4921332001686096,
      "learning_rate": 4.6179416639766524e-05,
      "loss": 0.863,
      "step": 7900
    },
    {
      "epoch": 0.5217075865253021,
      "grad_norm": 0.5133247971534729,
      "learning_rate": 4.616411218788007e-05,
      "loss": 0.8444,
      "step": 7910
    },
    {
      "epoch": 0.522367140996257,
      "grad_norm": 0.4950665235519409,
      "learning_rate": 4.614877969064431e-05,
      "loss": 0.8385,
      "step": 7920
    },
    {
      "epoch": 0.5230266954672119,
      "grad_norm": 0.6108483076095581,
      "learning_rate": 4.613341916837689e-05,
      "loss": 0.8486,
      "step": 7930
    },
    {
      "epoch": 0.5236862499381668,
      "grad_norm": 0.5163877010345459,
      "learning_rate": 4.611803064143263e-05,
      "loss": 0.8779,
      "step": 7940
    },
    {
      "epoch": 0.5243458044091216,
      "grad_norm": 0.4994930624961853,
      "learning_rate": 4.6102614130203424e-05,
      "loss": 0.8292,
      "step": 7950
    },
    {
      "epoch": 0.5250053588800765,
      "grad_norm": 0.45950081944465637,
      "learning_rate": 4.608716965511827e-05,
      "loss": 0.8807,
      "step": 7960
    },
    {
      "epoch": 0.5256649133510314,
      "grad_norm": 0.5794354677200317,
      "learning_rate": 4.60716972366432e-05,
      "loss": 0.8288,
      "step": 7970
    },
    {
      "epoch": 0.5263244678219863,
      "grad_norm": 0.593536376953125,
      "learning_rate": 4.60561968952813e-05,
      "loss": 0.8667,
      "step": 7980
    },
    {
      "epoch": 0.5269840222929412,
      "grad_norm": 0.5221846103668213,
      "learning_rate": 4.604066865157265e-05,
      "loss": 0.8593,
      "step": 7990
    },
    {
      "epoch": 0.527643576763896,
      "grad_norm": 0.5120049118995667,
      "learning_rate": 4.602511252609429e-05,
      "loss": 0.8116,
      "step": 8000
    },
    {
      "epoch": 0.5283031312348508,
      "grad_norm": 0.594819962978363,
      "learning_rate": 4.600952853946023e-05,
      "loss": 0.8882,
      "step": 8010
    },
    {
      "epoch": 0.5289626857058057,
      "grad_norm": 0.4388793110847473,
      "learning_rate": 4.599391671232138e-05,
      "loss": 0.854,
      "step": 8020
    },
    {
      "epoch": 0.5296222401767606,
      "grad_norm": 0.49736639857292175,
      "learning_rate": 4.597827706536556e-05,
      "loss": 0.8745,
      "step": 8030
    },
    {
      "epoch": 0.5302817946477155,
      "grad_norm": 0.6666948795318604,
      "learning_rate": 4.5962609619317434e-05,
      "loss": 0.8671,
      "step": 8040
    },
    {
      "epoch": 0.5309413491186703,
      "grad_norm": 0.4953151047229767,
      "learning_rate": 4.5946914394938514e-05,
      "loss": 0.874,
      "step": 8050
    },
    {
      "epoch": 0.5316009035896252,
      "grad_norm": 0.6658579707145691,
      "learning_rate": 4.593119141302714e-05,
      "loss": 0.8242,
      "step": 8060
    },
    {
      "epoch": 0.5322604580605801,
      "grad_norm": 0.484624445438385,
      "learning_rate": 4.59154406944184e-05,
      "loss": 0.8286,
      "step": 8070
    },
    {
      "epoch": 0.532920012531535,
      "grad_norm": 0.4317564368247986,
      "learning_rate": 4.5899662259984166e-05,
      "loss": 0.8598,
      "step": 8080
    },
    {
      "epoch": 0.5335795670024898,
      "grad_norm": 0.6070392727851868,
      "learning_rate": 4.588385613063302e-05,
      "loss": 0.8192,
      "step": 8090
    },
    {
      "epoch": 0.5342391214734447,
      "grad_norm": 0.5171172022819519,
      "learning_rate": 4.5868022327310264e-05,
      "loss": 0.84,
      "step": 8100
    },
    {
      "epoch": 0.5348986759443995,
      "grad_norm": 0.482692688703537,
      "learning_rate": 4.585216087099783e-05,
      "loss": 0.8015,
      "step": 8110
    },
    {
      "epoch": 0.5355582304153544,
      "grad_norm": 0.5581316351890564,
      "learning_rate": 4.583627178271435e-05,
      "loss": 0.8459,
      "step": 8120
    },
    {
      "epoch": 0.5362177848863093,
      "grad_norm": 0.6078279614448547,
      "learning_rate": 4.582035508351502e-05,
      "loss": 0.8771,
      "step": 8130
    },
    {
      "epoch": 0.5368773393572641,
      "grad_norm": 0.5564948916435242,
      "learning_rate": 4.580441079449167e-05,
      "loss": 0.8413,
      "step": 8140
    },
    {
      "epoch": 0.537536893828219,
      "grad_norm": 0.5623395442962646,
      "learning_rate": 4.5788438936772647e-05,
      "loss": 0.8595,
      "step": 8150
    },
    {
      "epoch": 0.5381964482991739,
      "grad_norm": 0.4650050699710846,
      "learning_rate": 4.5772439531522874e-05,
      "loss": 0.8825,
      "step": 8160
    },
    {
      "epoch": 0.5388560027701288,
      "grad_norm": 0.4734998643398285,
      "learning_rate": 4.5756412599943734e-05,
      "loss": 0.8794,
      "step": 8170
    },
    {
      "epoch": 0.5395155572410837,
      "grad_norm": 0.48804497718811035,
      "learning_rate": 4.5740358163273114e-05,
      "loss": 0.8392,
      "step": 8180
    },
    {
      "epoch": 0.5401751117120385,
      "grad_norm": 0.518511176109314,
      "learning_rate": 4.572427624278536e-05,
      "loss": 0.8408,
      "step": 8190
    },
    {
      "epoch": 0.5408346661829934,
      "grad_norm": 0.4985922574996948,
      "learning_rate": 4.57081668597912e-05,
      "loss": 0.8528,
      "step": 8200
    },
    {
      "epoch": 0.5414942206539483,
      "grad_norm": 0.5364166498184204,
      "learning_rate": 4.569203003563778e-05,
      "loss": 0.8069,
      "step": 8210
    },
    {
      "epoch": 0.5421537751249031,
      "grad_norm": 0.5090025067329407,
      "learning_rate": 4.56758657917086e-05,
      "loss": 0.854,
      "step": 8220
    },
    {
      "epoch": 0.542813329595858,
      "grad_norm": 0.5399594306945801,
      "learning_rate": 4.565967414942351e-05,
      "loss": 0.8182,
      "step": 8230
    },
    {
      "epoch": 0.5434728840668128,
      "grad_norm": 0.48023220896720886,
      "learning_rate": 4.564345513023865e-05,
      "loss": 0.8335,
      "step": 8240
    },
    {
      "epoch": 0.5441324385377677,
      "grad_norm": 0.47295162081718445,
      "learning_rate": 4.562720875564644e-05,
      "loss": 0.8727,
      "step": 8250
    },
    {
      "epoch": 0.5447919930087226,
      "grad_norm": 0.46377071738243103,
      "learning_rate": 4.561093504717555e-05,
      "loss": 0.82,
      "step": 8260
    },
    {
      "epoch": 0.5454515474796775,
      "grad_norm": 0.5018013119697571,
      "learning_rate": 4.5594634026390896e-05,
      "loss": 0.8799,
      "step": 8270
    },
    {
      "epoch": 0.5461111019506324,
      "grad_norm": 0.5019428730010986,
      "learning_rate": 4.5578305714893535e-05,
      "loss": 0.8223,
      "step": 8280
    },
    {
      "epoch": 0.5467706564215872,
      "grad_norm": 0.5506460666656494,
      "learning_rate": 4.556195013432073e-05,
      "loss": 0.9144,
      "step": 8290
    },
    {
      "epoch": 0.5474302108925421,
      "grad_norm": 0.5080007910728455,
      "learning_rate": 4.5545567306345874e-05,
      "loss": 0.9037,
      "step": 8300
    },
    {
      "epoch": 0.548089765363497,
      "grad_norm": 0.47640281915664673,
      "learning_rate": 4.5529157252678446e-05,
      "loss": 0.8722,
      "step": 8310
    },
    {
      "epoch": 0.5487493198344519,
      "grad_norm": 0.5623273849487305,
      "learning_rate": 4.551271999506403e-05,
      "loss": 0.8497,
      "step": 8320
    },
    {
      "epoch": 0.5494088743054067,
      "grad_norm": 0.4991755187511444,
      "learning_rate": 4.5496255555284236e-05,
      "loss": 0.8393,
      "step": 8330
    },
    {
      "epoch": 0.5500684287763615,
      "grad_norm": 0.48409467935562134,
      "learning_rate": 4.5479763955156706e-05,
      "loss": 0.836,
      "step": 8340
    },
    {
      "epoch": 0.5507279832473164,
      "grad_norm": 0.5671091675758362,
      "learning_rate": 4.546324521653507e-05,
      "loss": 0.9025,
      "step": 8350
    },
    {
      "epoch": 0.5513875377182713,
      "grad_norm": 0.506266713142395,
      "learning_rate": 4.5446699361308914e-05,
      "loss": 0.8011,
      "step": 8360
    },
    {
      "epoch": 0.5520470921892262,
      "grad_norm": 0.5483802556991577,
      "learning_rate": 4.5430126411403776e-05,
      "loss": 0.851,
      "step": 8370
    },
    {
      "epoch": 0.5527066466601811,
      "grad_norm": 0.46331244707107544,
      "learning_rate": 4.541352638878108e-05,
      "loss": 0.8544,
      "step": 8380
    },
    {
      "epoch": 0.5533662011311359,
      "grad_norm": 0.5024694800376892,
      "learning_rate": 4.5396899315438135e-05,
      "loss": 0.8617,
      "step": 8390
    },
    {
      "epoch": 0.5540257556020908,
      "grad_norm": 0.4586072564125061,
      "learning_rate": 4.538024521340809e-05,
      "loss": 0.8427,
      "step": 8400
    },
    {
      "epoch": 0.5546853100730457,
      "grad_norm": 0.5485617518424988,
      "learning_rate": 4.5363564104759905e-05,
      "loss": 0.8684,
      "step": 8410
    },
    {
      "epoch": 0.5553448645440006,
      "grad_norm": 0.4568479359149933,
      "learning_rate": 4.5346856011598345e-05,
      "loss": 0.8274,
      "step": 8420
    },
    {
      "epoch": 0.5560044190149553,
      "grad_norm": 0.574080765247345,
      "learning_rate": 4.533012095606391e-05,
      "loss": 0.8643,
      "step": 8430
    },
    {
      "epoch": 0.5566639734859102,
      "grad_norm": 0.5243128538131714,
      "learning_rate": 4.5313358960332864e-05,
      "loss": 0.8536,
      "step": 8440
    },
    {
      "epoch": 0.5573235279568651,
      "grad_norm": 0.4808977544307709,
      "learning_rate": 4.5296570046617136e-05,
      "loss": 0.8712,
      "step": 8450
    },
    {
      "epoch": 0.55798308242782,
      "grad_norm": 0.5325448513031006,
      "learning_rate": 4.527975423716434e-05,
      "loss": 0.8249,
      "step": 8460
    },
    {
      "epoch": 0.5586426368987749,
      "grad_norm": 0.53151535987854,
      "learning_rate": 4.5262911554257723e-05,
      "loss": 0.8731,
      "step": 8470
    },
    {
      "epoch": 0.5593021913697297,
      "grad_norm": 0.48107099533081055,
      "learning_rate": 4.524604202021615e-05,
      "loss": 0.8429,
      "step": 8480
    },
    {
      "epoch": 0.5599617458406846,
      "grad_norm": 0.49375805258750916,
      "learning_rate": 4.5229145657394075e-05,
      "loss": 0.8207,
      "step": 8490
    },
    {
      "epoch": 0.5606213003116395,
      "grad_norm": 0.5281050205230713,
      "learning_rate": 4.521222248818148e-05,
      "loss": 0.8357,
      "step": 8500
    },
    {
      "epoch": 0.5612808547825944,
      "grad_norm": 0.5373192429542542,
      "learning_rate": 4.51952725350039e-05,
      "loss": 0.8558,
      "step": 8510
    },
    {
      "epoch": 0.5619404092535493,
      "grad_norm": 0.42790329456329346,
      "learning_rate": 4.5178295820322326e-05,
      "loss": 0.8332,
      "step": 8520
    },
    {
      "epoch": 0.562599963724504,
      "grad_norm": 0.5260501503944397,
      "learning_rate": 4.5161292366633246e-05,
      "loss": 0.853,
      "step": 8530
    },
    {
      "epoch": 0.5632595181954589,
      "grad_norm": 0.4377802312374115,
      "learning_rate": 4.514426219646857e-05,
      "loss": 0.8405,
      "step": 8540
    },
    {
      "epoch": 0.5639190726664138,
      "grad_norm": 0.5629652738571167,
      "learning_rate": 4.512720533239558e-05,
      "loss": 0.8894,
      "step": 8550
    },
    {
      "epoch": 0.5645786271373687,
      "grad_norm": 0.5291570425033569,
      "learning_rate": 4.5110121797017005e-05,
      "loss": 0.8372,
      "step": 8560
    },
    {
      "epoch": 0.5652381816083236,
      "grad_norm": 0.5068237781524658,
      "learning_rate": 4.509301161297083e-05,
      "loss": 0.8577,
      "step": 8570
    },
    {
      "epoch": 0.5658977360792784,
      "grad_norm": 0.4621034264564514,
      "learning_rate": 4.507587480293041e-05,
      "loss": 0.8374,
      "step": 8580
    },
    {
      "epoch": 0.5665572905502333,
      "grad_norm": 0.572524905204773,
      "learning_rate": 4.505871138960437e-05,
      "loss": 0.8846,
      "step": 8590
    },
    {
      "epoch": 0.5672168450211882,
      "grad_norm": 0.49350565671920776,
      "learning_rate": 4.504152139573659e-05,
      "loss": 0.8411,
      "step": 8600
    },
    {
      "epoch": 0.5678763994921431,
      "grad_norm": 0.4351065754890442,
      "learning_rate": 4.502430484410614e-05,
      "loss": 0.8809,
      "step": 8610
    },
    {
      "epoch": 0.568535953963098,
      "grad_norm": 0.5555775761604309,
      "learning_rate": 4.500706175752733e-05,
      "loss": 0.8555,
      "step": 8620
    },
    {
      "epoch": 0.5691955084340528,
      "grad_norm": 0.48449385166168213,
      "learning_rate": 4.498979215884963e-05,
      "loss": 0.8341,
      "step": 8630
    },
    {
      "epoch": 0.5698550629050076,
      "grad_norm": 0.5540316104888916,
      "learning_rate": 4.4972496070957596e-05,
      "loss": 0.8344,
      "step": 8640
    },
    {
      "epoch": 0.5705146173759625,
      "grad_norm": 0.6112542152404785,
      "learning_rate": 4.495517351677093e-05,
      "loss": 0.8555,
      "step": 8650
    },
    {
      "epoch": 0.5711741718469174,
      "grad_norm": 0.4980996549129486,
      "learning_rate": 4.4937824519244384e-05,
      "loss": 0.8395,
      "step": 8660
    },
    {
      "epoch": 0.5718337263178723,
      "grad_norm": 0.5588312149047852,
      "learning_rate": 4.4920449101367765e-05,
      "loss": 0.8709,
      "step": 8670
    },
    {
      "epoch": 0.5724932807888271,
      "grad_norm": 0.5733287930488586,
      "learning_rate": 4.4903047286165875e-05,
      "loss": 0.8522,
      "step": 8680
    },
    {
      "epoch": 0.573152835259782,
      "grad_norm": 0.5462709665298462,
      "learning_rate": 4.488561909669851e-05,
      "loss": 0.8962,
      "step": 8690
    },
    {
      "epoch": 0.5738123897307369,
      "grad_norm": 0.506364107131958,
      "learning_rate": 4.48681645560604e-05,
      "loss": 0.8295,
      "step": 8700
    },
    {
      "epoch": 0.5744719442016918,
      "grad_norm": 0.5398162603378296,
      "learning_rate": 4.485068368738122e-05,
      "loss": 0.8408,
      "step": 8710
    },
    {
      "epoch": 0.5751314986726467,
      "grad_norm": 0.5593112111091614,
      "learning_rate": 4.4833176513825504e-05,
      "loss": 0.8576,
      "step": 8720
    },
    {
      "epoch": 0.5757910531436015,
      "grad_norm": 0.5460782051086426,
      "learning_rate": 4.481564305859266e-05,
      "loss": 0.8577,
      "step": 8730
    },
    {
      "epoch": 0.5764506076145564,
      "grad_norm": 0.4949951767921448,
      "learning_rate": 4.479808334491692e-05,
      "loss": 0.8214,
      "step": 8740
    },
    {
      "epoch": 0.5771101620855112,
      "grad_norm": 0.5325927138328552,
      "learning_rate": 4.478049739606731e-05,
      "loss": 0.8378,
      "step": 8750
    },
    {
      "epoch": 0.5777697165564661,
      "grad_norm": 0.4853087067604065,
      "learning_rate": 4.476288523534762e-05,
      "loss": 0.8298,
      "step": 8760
    },
    {
      "epoch": 0.578429271027421,
      "grad_norm": 0.49878838658332825,
      "learning_rate": 4.4745246886096385e-05,
      "loss": 0.8152,
      "step": 8770
    },
    {
      "epoch": 0.5790888254983758,
      "grad_norm": 0.46912261843681335,
      "learning_rate": 4.4727582371686825e-05,
      "loss": 0.8127,
      "step": 8780
    },
    {
      "epoch": 0.5797483799693307,
      "grad_norm": 0.5477004647254944,
      "learning_rate": 4.470989171552685e-05,
      "loss": 0.8446,
      "step": 8790
    },
    {
      "epoch": 0.5804079344402856,
      "grad_norm": 0.5392802953720093,
      "learning_rate": 4.4692174941059006e-05,
      "loss": 0.8225,
      "step": 8800
    },
    {
      "epoch": 0.5810674889112405,
      "grad_norm": 0.5593754053115845,
      "learning_rate": 4.467443207176044e-05,
      "loss": 0.8241,
      "step": 8810
    },
    {
      "epoch": 0.5817270433821953,
      "grad_norm": 0.5052483677864075,
      "learning_rate": 4.4656663131142894e-05,
      "loss": 0.8303,
      "step": 8820
    },
    {
      "epoch": 0.5823865978531502,
      "grad_norm": 0.5355677604675293,
      "learning_rate": 4.463886814275264e-05,
      "loss": 0.8278,
      "step": 8830
    },
    {
      "epoch": 0.5830461523241051,
      "grad_norm": 0.4736327826976776,
      "learning_rate": 4.462104713017048e-05,
      "loss": 0.8249,
      "step": 8840
    },
    {
      "epoch": 0.58370570679506,
      "grad_norm": 0.5423553586006165,
      "learning_rate": 4.460320011701169e-05,
      "loss": 0.8718,
      "step": 8850
    },
    {
      "epoch": 0.5843652612660148,
      "grad_norm": 0.5245490074157715,
      "learning_rate": 4.458532712692603e-05,
      "loss": 0.8582,
      "step": 8860
    },
    {
      "epoch": 0.5850248157369696,
      "grad_norm": 0.5676358342170715,
      "learning_rate": 4.456742818359765e-05,
      "loss": 0.9181,
      "step": 8870
    },
    {
      "epoch": 0.5856843702079245,
      "grad_norm": 0.5473547577857971,
      "learning_rate": 4.45495033107451e-05,
      "loss": 0.8865,
      "step": 8880
    },
    {
      "epoch": 0.5863439246788794,
      "grad_norm": 0.6121447086334229,
      "learning_rate": 4.453155253212129e-05,
      "loss": 0.8642,
      "step": 8890
    },
    {
      "epoch": 0.5870034791498343,
      "grad_norm": 0.4944417178630829,
      "learning_rate": 4.451357587151347e-05,
      "loss": 0.8767,
      "step": 8900
    },
    {
      "epoch": 0.5876630336207892,
      "grad_norm": 0.5933345556259155,
      "learning_rate": 4.449557335274318e-05,
      "loss": 0.875,
      "step": 8910
    },
    {
      "epoch": 0.588322588091744,
      "grad_norm": 0.5733058452606201,
      "learning_rate": 4.4477544999666214e-05,
      "loss": 0.8545,
      "step": 8920
    },
    {
      "epoch": 0.5889821425626989,
      "grad_norm": 0.5385257005691528,
      "learning_rate": 4.4459490836172626e-05,
      "loss": 0.8053,
      "step": 8930
    },
    {
      "epoch": 0.5896416970336538,
      "grad_norm": 0.5365327596664429,
      "learning_rate": 4.4441410886186644e-05,
      "loss": 0.8491,
      "step": 8940
    },
    {
      "epoch": 0.5903012515046087,
      "grad_norm": 0.5605241656303406,
      "learning_rate": 4.442330517366668e-05,
      "loss": 0.8735,
      "step": 8950
    },
    {
      "epoch": 0.5909608059755636,
      "grad_norm": 0.4860929548740387,
      "learning_rate": 4.44051737226053e-05,
      "loss": 0.8801,
      "step": 8960
    },
    {
      "epoch": 0.5916203604465183,
      "grad_norm": 0.5219595432281494,
      "learning_rate": 4.438701655702914e-05,
      "loss": 0.8252,
      "step": 8970
    },
    {
      "epoch": 0.5922799149174732,
      "grad_norm": 0.49364590644836426,
      "learning_rate": 4.436883370099895e-05,
      "loss": 0.8633,
      "step": 8980
    },
    {
      "epoch": 0.5929394693884281,
      "grad_norm": 0.5454347729682922,
      "learning_rate": 4.43506251786095e-05,
      "loss": 0.9105,
      "step": 8990
    },
    {
      "epoch": 0.593599023859383,
      "grad_norm": 0.5115081667900085,
      "learning_rate": 4.4332391013989574e-05,
      "loss": 0.808,
      "step": 9000
    },
    {
      "epoch": 0.5942585783303379,
      "grad_norm": 0.5812116861343384,
      "learning_rate": 4.4314131231301944e-05,
      "loss": 0.8058,
      "step": 9010
    },
    {
      "epoch": 0.5949181328012927,
      "grad_norm": 0.5520011186599731,
      "learning_rate": 4.4295845854743326e-05,
      "loss": 0.854,
      "step": 9020
    },
    {
      "epoch": 0.5955776872722476,
      "grad_norm": 0.5113941431045532,
      "learning_rate": 4.4277534908544335e-05,
      "loss": 0.8387,
      "step": 9030
    },
    {
      "epoch": 0.5962372417432025,
      "grad_norm": 0.5096448659896851,
      "learning_rate": 4.425919841696952e-05,
      "loss": 0.8877,
      "step": 9040
    },
    {
      "epoch": 0.5968967962141574,
      "grad_norm": 0.5333675146102905,
      "learning_rate": 4.42408364043172e-05,
      "loss": 0.8456,
      "step": 9050
    },
    {
      "epoch": 0.5975563506851123,
      "grad_norm": 0.5552167296409607,
      "learning_rate": 4.42224488949196e-05,
      "loss": 0.8681,
      "step": 9060
    },
    {
      "epoch": 0.598215905156067,
      "grad_norm": 0.45667341351509094,
      "learning_rate": 4.420403591314266e-05,
      "loss": 0.8792,
      "step": 9070
    },
    {
      "epoch": 0.5988754596270219,
      "grad_norm": 0.6120560169219971,
      "learning_rate": 4.418559748338611e-05,
      "loss": 0.8355,
      "step": 9080
    },
    {
      "epoch": 0.5995350140979768,
      "grad_norm": 0.4900742471218109,
      "learning_rate": 4.416713363008341e-05,
      "loss": 0.8652,
      "step": 9090
    },
    {
      "epoch": 0.6001945685689317,
      "grad_norm": 0.5414189696311951,
      "learning_rate": 4.414864437770169e-05,
      "loss": 0.8779,
      "step": 9100
    },
    {
      "epoch": 0.6008541230398866,
      "grad_norm": 0.5061354637145996,
      "learning_rate": 4.413012975074174e-05,
      "loss": 0.8185,
      "step": 9110
    },
    {
      "epoch": 0.6015136775108414,
      "grad_norm": 0.5018121600151062,
      "learning_rate": 4.4111589773737976e-05,
      "loss": 0.8634,
      "step": 9120
    },
    {
      "epoch": 0.6021732319817963,
      "grad_norm": 0.5141509175300598,
      "learning_rate": 4.4093024471258414e-05,
      "loss": 0.8258,
      "step": 9130
    },
    {
      "epoch": 0.6028327864527512,
      "grad_norm": 0.5262400507926941,
      "learning_rate": 4.407443386790462e-05,
      "loss": 0.8597,
      "step": 9140
    },
    {
      "epoch": 0.6034923409237061,
      "grad_norm": 0.5369293093681335,
      "learning_rate": 4.40558179883117e-05,
      "loss": 0.9019,
      "step": 9150
    },
    {
      "epoch": 0.6041518953946609,
      "grad_norm": 0.5289623141288757,
      "learning_rate": 4.4037176857148235e-05,
      "loss": 0.8262,
      "step": 9160
    },
    {
      "epoch": 0.6048114498656157,
      "grad_norm": 0.5164844393730164,
      "learning_rate": 4.401851049911629e-05,
      "loss": 0.8415,
      "step": 9170
    },
    {
      "epoch": 0.6054710043365706,
      "grad_norm": 0.47341203689575195,
      "learning_rate": 4.399981893895133e-05,
      "loss": 0.8433,
      "step": 9180
    },
    {
      "epoch": 0.6061305588075255,
      "grad_norm": 0.4960174560546875,
      "learning_rate": 4.398110220142225e-05,
      "loss": 0.8375,
      "step": 9190
    },
    {
      "epoch": 0.6067901132784804,
      "grad_norm": 0.5548086762428284,
      "learning_rate": 4.39623603113313e-05,
      "loss": 0.8448,
      "step": 9200
    },
    {
      "epoch": 0.6074496677494352,
      "grad_norm": 0.5246376991271973,
      "learning_rate": 4.3943593293514024e-05,
      "loss": 0.865,
      "step": 9210
    },
    {
      "epoch": 0.6081092222203901,
      "grad_norm": 0.5469093918800354,
      "learning_rate": 4.3924801172839315e-05,
      "loss": 0.8506,
      "step": 9220
    },
    {
      "epoch": 0.608768776691345,
      "grad_norm": 0.4608471393585205,
      "learning_rate": 4.390598397420931e-05,
      "loss": 0.864,
      "step": 9230
    },
    {
      "epoch": 0.6094283311622999,
      "grad_norm": 0.5221256017684937,
      "learning_rate": 4.3887141722559356e-05,
      "loss": 0.8849,
      "step": 9240
    },
    {
      "epoch": 0.6100878856332548,
      "grad_norm": 0.49641260504722595,
      "learning_rate": 4.3868274442858034e-05,
      "loss": 0.8584,
      "step": 9250
    },
    {
      "epoch": 0.6107474401042096,
      "grad_norm": 0.5827483534812927,
      "learning_rate": 4.384938216010708e-05,
      "loss": 0.8585,
      "step": 9260
    },
    {
      "epoch": 0.6114069945751645,
      "grad_norm": 0.48872390389442444,
      "learning_rate": 4.383046489934135e-05,
      "loss": 0.8287,
      "step": 9270
    },
    {
      "epoch": 0.6120665490461193,
      "grad_norm": 0.6012468934059143,
      "learning_rate": 4.3811522685628806e-05,
      "loss": 0.8265,
      "step": 9280
    },
    {
      "epoch": 0.6127261035170742,
      "grad_norm": 0.48777174949645996,
      "learning_rate": 4.379255554407048e-05,
      "loss": 0.8712,
      "step": 9290
    },
    {
      "epoch": 0.6133856579880291,
      "grad_norm": 0.5294257998466492,
      "learning_rate": 4.377356349980044e-05,
      "loss": 0.8713,
      "step": 9300
    },
    {
      "epoch": 0.6140452124589839,
      "grad_norm": 0.534724771976471,
      "learning_rate": 4.3754546577985734e-05,
      "loss": 0.8705,
      "step": 9310
    },
    {
      "epoch": 0.6147047669299388,
      "grad_norm": 0.49359211325645447,
      "learning_rate": 4.37355048038264e-05,
      "loss": 0.8156,
      "step": 9320
    },
    {
      "epoch": 0.6153643214008937,
      "grad_norm": 0.48252472281455994,
      "learning_rate": 4.37164382025554e-05,
      "loss": 0.8668,
      "step": 9330
    },
    {
      "epoch": 0.6160238758718486,
      "grad_norm": 0.5378819704055786,
      "learning_rate": 4.3697346799438584e-05,
      "loss": 0.8605,
      "step": 9340
    },
    {
      "epoch": 0.6166834303428035,
      "grad_norm": 0.46722471714019775,
      "learning_rate": 4.3678230619774685e-05,
      "loss": 0.8482,
      "step": 9350
    },
    {
      "epoch": 0.6173429848137583,
      "grad_norm": 0.5129494667053223,
      "learning_rate": 4.3659089688895264e-05,
      "loss": 0.8591,
      "step": 9360
    },
    {
      "epoch": 0.6180025392847132,
      "grad_norm": 0.5393134951591492,
      "learning_rate": 4.363992403216467e-05,
      "loss": 0.8435,
      "step": 9370
    },
    {
      "epoch": 0.618662093755668,
      "grad_norm": 0.5477026104927063,
      "learning_rate": 4.362073367498002e-05,
      "loss": 0.8633,
      "step": 9380
    },
    {
      "epoch": 0.619321648226623,
      "grad_norm": 0.49567267298698425,
      "learning_rate": 4.360151864277119e-05,
      "loss": 0.8314,
      "step": 9390
    },
    {
      "epoch": 0.6199812026975778,
      "grad_norm": 0.5262508392333984,
      "learning_rate": 4.358227896100071e-05,
      "loss": 0.8515,
      "step": 9400
    },
    {
      "epoch": 0.6206407571685326,
      "grad_norm": 0.4469965100288391,
      "learning_rate": 4.3563014655163804e-05,
      "loss": 0.8674,
      "step": 9410
    },
    {
      "epoch": 0.6213003116394875,
      "grad_norm": 0.5764192938804626,
      "learning_rate": 4.354372575078831e-05,
      "loss": 0.875,
      "step": 9420
    },
    {
      "epoch": 0.6219598661104424,
      "grad_norm": 0.5805740356445312,
      "learning_rate": 4.3524412273434676e-05,
      "loss": 0.8142,
      "step": 9430
    },
    {
      "epoch": 0.6226194205813973,
      "grad_norm": 0.5671592950820923,
      "learning_rate": 4.350507424869591e-05,
      "loss": 0.8581,
      "step": 9440
    },
    {
      "epoch": 0.6232789750523522,
      "grad_norm": 0.4964553713798523,
      "learning_rate": 4.348571170219754e-05,
      "loss": 0.8241,
      "step": 9450
    },
    {
      "epoch": 0.623938529523307,
      "grad_norm": 0.5607429146766663,
      "learning_rate": 4.34663246595976e-05,
      "loss": 0.8813,
      "step": 9460
    },
    {
      "epoch": 0.6245980839942619,
      "grad_norm": 0.4957119822502136,
      "learning_rate": 4.344691314658657e-05,
      "loss": 0.8554,
      "step": 9470
    },
    {
      "epoch": 0.6252576384652168,
      "grad_norm": 0.5064859986305237,
      "learning_rate": 4.3427477188887366e-05,
      "loss": 0.8196,
      "step": 9480
    },
    {
      "epoch": 0.6259171929361717,
      "grad_norm": 0.5270329713821411,
      "learning_rate": 4.3408016812255305e-05,
      "loss": 0.8072,
      "step": 9490
    },
    {
      "epoch": 0.6265767474071264,
      "grad_norm": 0.6006602048873901,
      "learning_rate": 4.3388532042478046e-05,
      "loss": 0.8736,
      "step": 9500
    },
    {
      "epoch": 0.6272363018780813,
      "grad_norm": 0.5389840006828308,
      "learning_rate": 4.336902290537558e-05,
      "loss": 0.879,
      "step": 9510
    },
    {
      "epoch": 0.6278958563490362,
      "grad_norm": 0.551805853843689,
      "learning_rate": 4.334948942680018e-05,
      "loss": 0.8526,
      "step": 9520
    },
    {
      "epoch": 0.6285554108199911,
      "grad_norm": 0.512858510017395,
      "learning_rate": 4.33299316326364e-05,
      "loss": 0.8474,
      "step": 9530
    },
    {
      "epoch": 0.629214965290946,
      "grad_norm": 0.4484001100063324,
      "learning_rate": 4.331034954880098e-05,
      "loss": 0.8638,
      "step": 9540
    },
    {
      "epoch": 0.6298745197619008,
      "grad_norm": 0.47472772002220154,
      "learning_rate": 4.329074320124288e-05,
      "loss": 0.8579,
      "step": 9550
    },
    {
      "epoch": 0.6305340742328557,
      "grad_norm": 0.4683813750743866,
      "learning_rate": 4.327111261594318e-05,
      "loss": 0.845,
      "step": 9560
    },
    {
      "epoch": 0.6311936287038106,
      "grad_norm": 0.5788334012031555,
      "learning_rate": 4.325145781891511e-05,
      "loss": 0.8298,
      "step": 9570
    },
    {
      "epoch": 0.6318531831747655,
      "grad_norm": 0.5225703716278076,
      "learning_rate": 4.323177883620397e-05,
      "loss": 0.8652,
      "step": 9580
    },
    {
      "epoch": 0.6325127376457204,
      "grad_norm": 0.5140812397003174,
      "learning_rate": 4.321207569388711e-05,
      "loss": 0.8526,
      "step": 9590
    },
    {
      "epoch": 0.6331722921166751,
      "grad_norm": 0.5085943937301636,
      "learning_rate": 4.319234841807388e-05,
      "loss": 0.8725,
      "step": 9600
    },
    {
      "epoch": 0.63383184658763,
      "grad_norm": 0.5138686299324036,
      "learning_rate": 4.317259703490564e-05,
      "loss": 0.8721,
      "step": 9610
    },
    {
      "epoch": 0.6344914010585849,
      "grad_norm": 0.5103156566619873,
      "learning_rate": 4.315282157055567e-05,
      "loss": 0.8488,
      "step": 9620
    },
    {
      "epoch": 0.6351509555295398,
      "grad_norm": 0.4980023205280304,
      "learning_rate": 4.313302205122918e-05,
      "loss": 0.8676,
      "step": 9630
    },
    {
      "epoch": 0.6358105100004947,
      "grad_norm": 0.5121146440505981,
      "learning_rate": 4.311319850316323e-05,
      "loss": 0.8798,
      "step": 9640
    },
    {
      "epoch": 0.6364700644714495,
      "grad_norm": 0.4848631024360657,
      "learning_rate": 4.309335095262676e-05,
      "loss": 0.8552,
      "step": 9650
    },
    {
      "epoch": 0.6371296189424044,
      "grad_norm": 0.5741939544677734,
      "learning_rate": 4.307347942592047e-05,
      "loss": 0.8477,
      "step": 9660
    },
    {
      "epoch": 0.6377891734133593,
      "grad_norm": 0.62933349609375,
      "learning_rate": 4.3053583949376886e-05,
      "loss": 0.8501,
      "step": 9670
    },
    {
      "epoch": 0.6384487278843142,
      "grad_norm": 0.5315343737602234,
      "learning_rate": 4.3033664549360215e-05,
      "loss": 0.8368,
      "step": 9680
    },
    {
      "epoch": 0.6391082823552691,
      "grad_norm": 0.5381109714508057,
      "learning_rate": 4.3013721252266404e-05,
      "loss": 0.8963,
      "step": 9690
    },
    {
      "epoch": 0.6397678368262238,
      "grad_norm": 0.45453372597694397,
      "learning_rate": 4.2993754084523055e-05,
      "loss": 0.8361,
      "step": 9700
    },
    {
      "epoch": 0.6404273912971787,
      "grad_norm": 0.5069344639778137,
      "learning_rate": 4.29737630725894e-05,
      "loss": 0.8874,
      "step": 9710
    },
    {
      "epoch": 0.6410869457681336,
      "grad_norm": 0.45548906922340393,
      "learning_rate": 4.295374824295627e-05,
      "loss": 0.8352,
      "step": 9720
    },
    {
      "epoch": 0.6417465002390885,
      "grad_norm": 0.5694235563278198,
      "learning_rate": 4.293370962214605e-05,
      "loss": 0.7936,
      "step": 9730
    },
    {
      "epoch": 0.6424060547100434,
      "grad_norm": 0.5499682426452637,
      "learning_rate": 4.291364723671267e-05,
      "loss": 0.8706,
      "step": 9740
    },
    {
      "epoch": 0.6430656091809982,
      "grad_norm": 0.5891594290733337,
      "learning_rate": 4.289356111324154e-05,
      "loss": 0.8444,
      "step": 9750
    },
    {
      "epoch": 0.6437251636519531,
      "grad_norm": 0.5436473488807678,
      "learning_rate": 4.287345127834952e-05,
      "loss": 0.8477,
      "step": 9760
    },
    {
      "epoch": 0.644384718122908,
      "grad_norm": 0.5410086512565613,
      "learning_rate": 4.285331775868491e-05,
      "loss": 0.8566,
      "step": 9770
    },
    {
      "epoch": 0.6450442725938629,
      "grad_norm": 0.5957720875740051,
      "learning_rate": 4.2833160580927354e-05,
      "loss": 0.8366,
      "step": 9780
    },
    {
      "epoch": 0.6457038270648178,
      "grad_norm": 0.5900678634643555,
      "learning_rate": 4.281297977178791e-05,
      "loss": 0.8485,
      "step": 9790
    },
    {
      "epoch": 0.6463633815357726,
      "grad_norm": 0.46779167652130127,
      "learning_rate": 4.27927753580089e-05,
      "loss": 0.8603,
      "step": 9800
    },
    {
      "epoch": 0.6470229360067274,
      "grad_norm": 0.5066987872123718,
      "learning_rate": 4.2772547366363935e-05,
      "loss": 0.811,
      "step": 9810
    },
    {
      "epoch": 0.6476824904776823,
      "grad_norm": 0.4854544699192047,
      "learning_rate": 4.275229582365789e-05,
      "loss": 0.8084,
      "step": 9820
    },
    {
      "epoch": 0.6483420449486372,
      "grad_norm": 0.5501772165298462,
      "learning_rate": 4.273202075672682e-05,
      "loss": 0.8513,
      "step": 9830
    },
    {
      "epoch": 0.649001599419592,
      "grad_norm": 0.5020511150360107,
      "learning_rate": 4.271172219243798e-05,
      "loss": 0.8533,
      "step": 9840
    },
    {
      "epoch": 0.6496611538905469,
      "grad_norm": 0.5452893972396851,
      "learning_rate": 4.269140015768974e-05,
      "loss": 0.872,
      "step": 9850
    },
    {
      "epoch": 0.6503207083615018,
      "grad_norm": 0.5562975406646729,
      "learning_rate": 4.2671054679411605e-05,
      "loss": 0.8741,
      "step": 9860
    },
    {
      "epoch": 0.6509802628324567,
      "grad_norm": 0.5018196105957031,
      "learning_rate": 4.2650685784564085e-05,
      "loss": 0.8489,
      "step": 9870
    },
    {
      "epoch": 0.6516398173034116,
      "grad_norm": 0.5853243470191956,
      "learning_rate": 4.263029350013878e-05,
      "loss": 0.8669,
      "step": 9880
    },
    {
      "epoch": 0.6522993717743664,
      "grad_norm": 0.5556926131248474,
      "learning_rate": 4.260987785315826e-05,
      "loss": 0.8974,
      "step": 9890
    },
    {
      "epoch": 0.6529589262453213,
      "grad_norm": 0.4784821569919586,
      "learning_rate": 4.258943887067605e-05,
      "loss": 0.8753,
      "step": 9900
    },
    {
      "epoch": 0.6536184807162762,
      "grad_norm": 0.4800739884376526,
      "learning_rate": 4.256897657977661e-05,
      "loss": 0.8215,
      "step": 9910
    },
    {
      "epoch": 0.654278035187231,
      "grad_norm": 0.5254513621330261,
      "learning_rate": 4.254849100757527e-05,
      "loss": 0.8362,
      "step": 9920
    },
    {
      "epoch": 0.6549375896581859,
      "grad_norm": 0.5375306010246277,
      "learning_rate": 4.252798218121823e-05,
      "loss": 0.8502,
      "step": 9930
    },
    {
      "epoch": 0.6555971441291407,
      "grad_norm": 0.5457652807235718,
      "learning_rate": 4.2507450127882496e-05,
      "loss": 0.8469,
      "step": 9940
    },
    {
      "epoch": 0.6562566986000956,
      "grad_norm": 0.49668949842453003,
      "learning_rate": 4.248689487477584e-05,
      "loss": 0.8365,
      "step": 9950
    },
    {
      "epoch": 0.6569162530710505,
      "grad_norm": 0.4837133586406708,
      "learning_rate": 4.246631644913681e-05,
      "loss": 0.876,
      "step": 9960
    },
    {
      "epoch": 0.6575758075420054,
      "grad_norm": 0.5117802023887634,
      "learning_rate": 4.244571487823462e-05,
      "loss": 0.8463,
      "step": 9970
    },
    {
      "epoch": 0.6582353620129603,
      "grad_norm": 0.4953489601612091,
      "learning_rate": 4.242509018936919e-05,
      "loss": 0.8449,
      "step": 9980
    },
    {
      "epoch": 0.6588949164839151,
      "grad_norm": 0.48785701394081116,
      "learning_rate": 4.2404442409871056e-05,
      "loss": 0.8167,
      "step": 9990
    },
    {
      "epoch": 0.65955447095487,
      "grad_norm": 0.48401758074760437,
      "learning_rate": 4.238377156710135e-05,
      "loss": 0.8466,
      "step": 10000
    },
    {
      "epoch": 0.6602140254258249,
      "grad_norm": 0.5855180621147156,
      "learning_rate": 4.236307768845177e-05,
      "loss": 0.8657,
      "step": 10010
    },
    {
      "epoch": 0.6608735798967798,
      "grad_norm": 0.510765790939331,
      "learning_rate": 4.234236080134455e-05,
      "loss": 0.8918,
      "step": 10020
    },
    {
      "epoch": 0.6615331343677346,
      "grad_norm": 0.569969654083252,
      "learning_rate": 4.232162093323241e-05,
      "loss": 0.8425,
      "step": 10030
    },
    {
      "epoch": 0.6621926888386894,
      "grad_norm": 0.4895627498626709,
      "learning_rate": 4.23008581115985e-05,
      "loss": 0.8155,
      "step": 10040
    },
    {
      "epoch": 0.6628522433096443,
      "grad_norm": 0.5220820903778076,
      "learning_rate": 4.2280072363956415e-05,
      "loss": 0.8213,
      "step": 10050
    },
    {
      "epoch": 0.6635117977805992,
      "grad_norm": 0.5542130470275879,
      "learning_rate": 4.225926371785012e-05,
      "loss": 0.828,
      "step": 10060
    },
    {
      "epoch": 0.6641713522515541,
      "grad_norm": 0.5008354187011719,
      "learning_rate": 4.2238432200853914e-05,
      "loss": 0.8678,
      "step": 10070
    },
    {
      "epoch": 0.664830906722509,
      "grad_norm": 0.504158079624176,
      "learning_rate": 4.221757784057243e-05,
      "loss": 0.8491,
      "step": 10080
    },
    {
      "epoch": 0.6654904611934638,
      "grad_norm": 0.5875375270843506,
      "learning_rate": 4.219670066464053e-05,
      "loss": 0.8821,
      "step": 10090
    },
    {
      "epoch": 0.6661500156644187,
      "grad_norm": 0.5509904026985168,
      "learning_rate": 4.217580070072336e-05,
      "loss": 0.8331,
      "step": 10100
    },
    {
      "epoch": 0.6668095701353736,
      "grad_norm": 0.4190371632575989,
      "learning_rate": 4.215487797651621e-05,
      "loss": 0.8379,
      "step": 10110
    },
    {
      "epoch": 0.6674691246063285,
      "grad_norm": 0.49308207631111145,
      "learning_rate": 4.213393251974458e-05,
      "loss": 0.8703,
      "step": 10120
    },
    {
      "epoch": 0.6681286790772833,
      "grad_norm": 0.47358161211013794,
      "learning_rate": 4.2112964358164056e-05,
      "loss": 0.8307,
      "step": 10130
    },
    {
      "epoch": 0.6687882335482381,
      "grad_norm": 0.4836333692073822,
      "learning_rate": 4.209197351956032e-05,
      "loss": 0.8192,
      "step": 10140
    },
    {
      "epoch": 0.669447788019193,
      "grad_norm": 0.46903476119041443,
      "learning_rate": 4.207096003174914e-05,
      "loss": 0.871,
      "step": 10150
    },
    {
      "epoch": 0.6701073424901479,
      "grad_norm": 0.45738083124160767,
      "learning_rate": 4.204992392257624e-05,
      "loss": 0.8276,
      "step": 10160
    },
    {
      "epoch": 0.6707668969611028,
      "grad_norm": 0.5130215287208557,
      "learning_rate": 4.202886521991735e-05,
      "loss": 0.8392,
      "step": 10170
    },
    {
      "epoch": 0.6714264514320576,
      "grad_norm": 0.4978039860725403,
      "learning_rate": 4.200778395167815e-05,
      "loss": 0.8435,
      "step": 10180
    },
    {
      "epoch": 0.6720860059030125,
      "grad_norm": 0.4951028823852539,
      "learning_rate": 4.19866801457942e-05,
      "loss": 0.8354,
      "step": 10190
    },
    {
      "epoch": 0.6727455603739674,
      "grad_norm": 0.5361219644546509,
      "learning_rate": 4.196555383023094e-05,
      "loss": 0.8466,
      "step": 10200
    },
    {
      "epoch": 0.6734051148449223,
      "grad_norm": 0.5519912838935852,
      "learning_rate": 4.194440503298363e-05,
      "loss": 0.8152,
      "step": 10210
    },
    {
      "epoch": 0.6740646693158772,
      "grad_norm": 0.5510390996932983,
      "learning_rate": 4.1923233782077334e-05,
      "loss": 0.816,
      "step": 10220
    },
    {
      "epoch": 0.674724223786832,
      "grad_norm": 0.5109158754348755,
      "learning_rate": 4.1902040105566844e-05,
      "loss": 0.873,
      "step": 10230
    },
    {
      "epoch": 0.6753837782577868,
      "grad_norm": 0.47579264640808105,
      "learning_rate": 4.188082403153669e-05,
      "loss": 0.8319,
      "step": 10240
    },
    {
      "epoch": 0.6760433327287417,
      "grad_norm": 0.5012093186378479,
      "learning_rate": 4.1859585588101095e-05,
      "loss": 0.8384,
      "step": 10250
    },
    {
      "epoch": 0.6767028871996966,
      "grad_norm": 0.5368977189064026,
      "learning_rate": 4.1838324803403894e-05,
      "loss": 0.8212,
      "step": 10260
    },
    {
      "epoch": 0.6773624416706515,
      "grad_norm": 0.5367230772972107,
      "learning_rate": 4.181704170561854e-05,
      "loss": 0.8183,
      "step": 10270
    },
    {
      "epoch": 0.6780219961416063,
      "grad_norm": 0.606758713722229,
      "learning_rate": 4.179573632294806e-05,
      "loss": 0.8444,
      "step": 10280
    },
    {
      "epoch": 0.6786815506125612,
      "grad_norm": 0.5097817182540894,
      "learning_rate": 4.1774408683625e-05,
      "loss": 0.8486,
      "step": 10290
    },
    {
      "epoch": 0.6793411050835161,
      "grad_norm": 0.4653185307979584,
      "learning_rate": 4.175305881591141e-05,
      "loss": 0.8338,
      "step": 10300
    },
    {
      "epoch": 0.680000659554471,
      "grad_norm": 0.46998143196105957,
      "learning_rate": 4.1731686748098795e-05,
      "loss": 0.8318,
      "step": 10310
    },
    {
      "epoch": 0.6806602140254259,
      "grad_norm": 0.5375470519065857,
      "learning_rate": 4.1710292508508055e-05,
      "loss": 0.8311,
      "step": 10320
    },
    {
      "epoch": 0.6813197684963807,
      "grad_norm": 0.4817380905151367,
      "learning_rate": 4.168887612548951e-05,
      "loss": 0.867,
      "step": 10330
    },
    {
      "epoch": 0.6819793229673355,
      "grad_norm": 0.4903552234172821,
      "learning_rate": 4.16674376274228e-05,
      "loss": 0.8758,
      "step": 10340
    },
    {
      "epoch": 0.6826388774382904,
      "grad_norm": 0.5465636253356934,
      "learning_rate": 4.164597704271686e-05,
      "loss": 0.8036,
      "step": 10350
    },
    {
      "epoch": 0.6832984319092453,
      "grad_norm": 0.5072474479675293,
      "learning_rate": 4.1624494399809923e-05,
      "loss": 0.7913,
      "step": 10360
    },
    {
      "epoch": 0.6839579863802002,
      "grad_norm": 0.5057288408279419,
      "learning_rate": 4.1602989727169414e-05,
      "loss": 0.8626,
      "step": 10370
    },
    {
      "epoch": 0.684617540851155,
      "grad_norm": 0.4934908151626587,
      "learning_rate": 4.1581463053292e-05,
      "loss": 0.8859,
      "step": 10380
    },
    {
      "epoch": 0.6852770953221099,
      "grad_norm": 0.45849695801734924,
      "learning_rate": 4.1559914406703446e-05,
      "loss": 0.7938,
      "step": 10390
    },
    {
      "epoch": 0.6859366497930648,
      "grad_norm": 0.5369190573692322,
      "learning_rate": 4.153834381595869e-05,
      "loss": 0.8507,
      "step": 10400
    },
    {
      "epoch": 0.6865962042640197,
      "grad_norm": 0.48543307185173035,
      "learning_rate": 4.151675130964169e-05,
      "loss": 0.8423,
      "step": 10410
    },
    {
      "epoch": 0.6872557587349746,
      "grad_norm": 0.5510615706443787,
      "learning_rate": 4.149513691636551e-05,
      "loss": 0.8856,
      "step": 10420
    },
    {
      "epoch": 0.6879153132059294,
      "grad_norm": 0.495917946100235,
      "learning_rate": 4.147350066477215e-05,
      "loss": 0.8118,
      "step": 10430
    },
    {
      "epoch": 0.6885748676768843,
      "grad_norm": 0.5462771058082581,
      "learning_rate": 4.145184258353263e-05,
      "loss": 0.8556,
      "step": 10440
    },
    {
      "epoch": 0.6892344221478391,
      "grad_norm": 0.5103912949562073,
      "learning_rate": 4.143016270134686e-05,
      "loss": 0.8913,
      "step": 10450
    },
    {
      "epoch": 0.689893976618794,
      "grad_norm": 0.5578963756561279,
      "learning_rate": 4.140846104694368e-05,
      "loss": 0.8662,
      "step": 10460
    },
    {
      "epoch": 0.6905535310897489,
      "grad_norm": 0.51603102684021,
      "learning_rate": 4.138673764908074e-05,
      "loss": 0.8427,
      "step": 10470
    },
    {
      "epoch": 0.6912130855607037,
      "grad_norm": 0.5023142695426941,
      "learning_rate": 4.136499253654452e-05,
      "loss": 0.8672,
      "step": 10480
    },
    {
      "epoch": 0.6918726400316586,
      "grad_norm": 0.49633529782295227,
      "learning_rate": 4.134322573815028e-05,
      "loss": 0.8462,
      "step": 10490
    },
    {
      "epoch": 0.6925321945026135,
      "grad_norm": 0.48940548300743103,
      "learning_rate": 4.1321437282742e-05,
      "loss": 0.835,
      "step": 10500
    },
    {
      "epoch": 0.6931917489735684,
      "grad_norm": 0.481265664100647,
      "learning_rate": 4.129962719919238e-05,
      "loss": 0.8219,
      "step": 10510
    },
    {
      "epoch": 0.6938513034445232,
      "grad_norm": 0.5408857464790344,
      "learning_rate": 4.127779551640276e-05,
      "loss": 0.7663,
      "step": 10520
    },
    {
      "epoch": 0.6945108579154781,
      "grad_norm": 0.5185447931289673,
      "learning_rate": 4.1255942263303127e-05,
      "loss": 0.8399,
      "step": 10530
    },
    {
      "epoch": 0.695170412386433,
      "grad_norm": 0.46000975370407104,
      "learning_rate": 4.1234067468852023e-05,
      "loss": 0.8613,
      "step": 10540
    },
    {
      "epoch": 0.6958299668573878,
      "grad_norm": 0.47393959760665894,
      "learning_rate": 4.121217116203656e-05,
      "loss": 0.8552,
      "step": 10550
    },
    {
      "epoch": 0.6964895213283427,
      "grad_norm": 0.5146589875221252,
      "learning_rate": 4.1190253371872335e-05,
      "loss": 0.8211,
      "step": 10560
    },
    {
      "epoch": 0.6971490757992975,
      "grad_norm": 0.530214250087738,
      "learning_rate": 4.116831412740342e-05,
      "loss": 0.8309,
      "step": 10570
    },
    {
      "epoch": 0.6978086302702524,
      "grad_norm": 0.4867837131023407,
      "learning_rate": 4.114635345770234e-05,
      "loss": 0.8603,
      "step": 10580
    },
    {
      "epoch": 0.6984681847412073,
      "grad_norm": 0.4901922345161438,
      "learning_rate": 4.112437139186998e-05,
      "loss": 0.839,
      "step": 10590
    },
    {
      "epoch": 0.6991277392121622,
      "grad_norm": 0.5245567560195923,
      "learning_rate": 4.110236795903559e-05,
      "loss": 0.809,
      "step": 10600
    },
    {
      "epoch": 0.6997872936831171,
      "grad_norm": 0.5026851892471313,
      "learning_rate": 4.1080343188356734e-05,
      "loss": 0.8655,
      "step": 10610
    },
    {
      "epoch": 0.7004468481540719,
      "grad_norm": 0.4868167042732239,
      "learning_rate": 4.105829710901926e-05,
      "loss": 0.8839,
      "step": 10620
    },
    {
      "epoch": 0.7011064026250268,
      "grad_norm": 0.5156816840171814,
      "learning_rate": 4.1036229750237225e-05,
      "loss": 0.8233,
      "step": 10630
    },
    {
      "epoch": 0.7017659570959817,
      "grad_norm": 0.5261308550834656,
      "learning_rate": 4.1014141141252925e-05,
      "loss": 0.8245,
      "step": 10640
    },
    {
      "epoch": 0.7024255115669366,
      "grad_norm": 0.484281450510025,
      "learning_rate": 4.099203131133678e-05,
      "loss": 0.8896,
      "step": 10650
    },
    {
      "epoch": 0.7030850660378914,
      "grad_norm": 0.5240840315818787,
      "learning_rate": 4.0969900289787355e-05,
      "loss": 0.8351,
      "step": 10660
    },
    {
      "epoch": 0.7037446205088462,
      "grad_norm": 0.5067978501319885,
      "learning_rate": 4.094774810593128e-05,
      "loss": 0.8063,
      "step": 10670
    },
    {
      "epoch": 0.7044041749798011,
      "grad_norm": 0.4869655668735504,
      "learning_rate": 4.0925574789123245e-05,
      "loss": 0.8333,
      "step": 10680
    },
    {
      "epoch": 0.705063729450756,
      "grad_norm": 0.4879327118396759,
      "learning_rate": 4.090338036874592e-05,
      "loss": 0.8978,
      "step": 10690
    },
    {
      "epoch": 0.7057232839217109,
      "grad_norm": 0.503984808921814,
      "learning_rate": 4.088116487420995e-05,
      "loss": 0.8225,
      "step": 10700
    },
    {
      "epoch": 0.7063828383926658,
      "grad_norm": 0.5124876499176025,
      "learning_rate": 4.085892833495394e-05,
      "loss": 0.8247,
      "step": 10710
    },
    {
      "epoch": 0.7070423928636206,
      "grad_norm": 0.471996933221817,
      "learning_rate": 4.083667078044432e-05,
      "loss": 0.8366,
      "step": 10720
    },
    {
      "epoch": 0.7077019473345755,
      "grad_norm": 0.495221346616745,
      "learning_rate": 4.081439224017541e-05,
      "loss": 0.836,
      "step": 10730
    },
    {
      "epoch": 0.7083615018055304,
      "grad_norm": 0.501978874206543,
      "learning_rate": 4.0792092743669344e-05,
      "loss": 0.8057,
      "step": 10740
    },
    {
      "epoch": 0.7090210562764853,
      "grad_norm": 0.517574667930603,
      "learning_rate": 4.076977232047599e-05,
      "loss": 0.8133,
      "step": 10750
    },
    {
      "epoch": 0.7096806107474402,
      "grad_norm": 0.5121254920959473,
      "learning_rate": 4.074743100017298e-05,
      "loss": 0.8108,
      "step": 10760
    },
    {
      "epoch": 0.7103401652183949,
      "grad_norm": 0.5190317630767822,
      "learning_rate": 4.072506881236561e-05,
      "loss": 0.854,
      "step": 10770
    },
    {
      "epoch": 0.7109997196893498,
      "grad_norm": 0.5075172185897827,
      "learning_rate": 4.070268578668687e-05,
      "loss": 0.8511,
      "step": 10780
    },
    {
      "epoch": 0.7116592741603047,
      "grad_norm": 0.5328771471977234,
      "learning_rate": 4.068028195279731e-05,
      "loss": 0.761,
      "step": 10790
    },
    {
      "epoch": 0.7123188286312596,
      "grad_norm": 0.5213194489479065,
      "learning_rate": 4.06578573403851e-05,
      "loss": 0.822,
      "step": 10800
    },
    {
      "epoch": 0.7129783831022145,
      "grad_norm": 0.476565957069397,
      "learning_rate": 4.063541197916592e-05,
      "loss": 0.8382,
      "step": 10810
    },
    {
      "epoch": 0.7136379375731693,
      "grad_norm": 0.5263732075691223,
      "learning_rate": 4.0612945898882946e-05,
      "loss": 0.8203,
      "step": 10820
    },
    {
      "epoch": 0.7142974920441242,
      "grad_norm": 0.46448376774787903,
      "learning_rate": 4.0590459129306815e-05,
      "loss": 0.8388,
      "step": 10830
    },
    {
      "epoch": 0.7149570465150791,
      "grad_norm": 0.5530455708503723,
      "learning_rate": 4.056795170023557e-05,
      "loss": 0.8416,
      "step": 10840
    },
    {
      "epoch": 0.715616600986034,
      "grad_norm": 0.5095324516296387,
      "learning_rate": 4.054542364149466e-05,
      "loss": 0.8055,
      "step": 10850
    },
    {
      "epoch": 0.7162761554569888,
      "grad_norm": 0.48809099197387695,
      "learning_rate": 4.052287498293683e-05,
      "loss": 0.8078,
      "step": 10860
    },
    {
      "epoch": 0.7169357099279436,
      "grad_norm": 0.44991356134414673,
      "learning_rate": 4.050030575444216e-05,
      "loss": 0.8114,
      "step": 10870
    },
    {
      "epoch": 0.7175952643988985,
      "grad_norm": 0.5083290338516235,
      "learning_rate": 4.047771598591795e-05,
      "loss": 0.8888,
      "step": 10880
    },
    {
      "epoch": 0.7182548188698534,
      "grad_norm": 0.5286580920219421,
      "learning_rate": 4.045510570729877e-05,
      "loss": 0.8571,
      "step": 10890
    },
    {
      "epoch": 0.7189143733408083,
      "grad_norm": 0.4988810420036316,
      "learning_rate": 4.04324749485463e-05,
      "loss": 0.8191,
      "step": 10900
    },
    {
      "epoch": 0.7195739278117631,
      "grad_norm": 0.49191129207611084,
      "learning_rate": 4.040982373964943e-05,
      "loss": 0.802,
      "step": 10910
    },
    {
      "epoch": 0.720233482282718,
      "grad_norm": 0.5243684649467468,
      "learning_rate": 4.0387152110624096e-05,
      "loss": 0.8477,
      "step": 10920
    },
    {
      "epoch": 0.7208930367536729,
      "grad_norm": 0.44547760486602783,
      "learning_rate": 4.0364460091513325e-05,
      "loss": 0.8592,
      "step": 10930
    },
    {
      "epoch": 0.7215525912246278,
      "grad_norm": 0.5532107353210449,
      "learning_rate": 4.034174771238715e-05,
      "loss": 0.8735,
      "step": 10940
    },
    {
      "epoch": 0.7222121456955827,
      "grad_norm": 0.5193709135055542,
      "learning_rate": 4.0319015003342574e-05,
      "loss": 0.841,
      "step": 10950
    },
    {
      "epoch": 0.7228717001665375,
      "grad_norm": 0.44667038321495056,
      "learning_rate": 4.029626199450357e-05,
      "loss": 0.8137,
      "step": 10960
    },
    {
      "epoch": 0.7235312546374923,
      "grad_norm": 0.482125461101532,
      "learning_rate": 4.0273488716020986e-05,
      "loss": 0.8391,
      "step": 10970
    },
    {
      "epoch": 0.7241908091084472,
      "grad_norm": 0.5427354574203491,
      "learning_rate": 4.025069519807253e-05,
      "loss": 0.7965,
      "step": 10980
    },
    {
      "epoch": 0.7248503635794021,
      "grad_norm": 0.5725659728050232,
      "learning_rate": 4.022788147086275e-05,
      "loss": 0.8399,
      "step": 10990
    },
    {
      "epoch": 0.725509918050357,
      "grad_norm": 0.5463594198226929,
      "learning_rate": 4.020504756462297e-05,
      "loss": 0.8382,
      "step": 11000
    },
    {
      "epoch": 0.7261694725213118,
      "grad_norm": 0.5389062166213989,
      "learning_rate": 4.018219350961121e-05,
      "loss": 0.8262,
      "step": 11010
    },
    {
      "epoch": 0.7268290269922667,
      "grad_norm": 0.5420926213264465,
      "learning_rate": 4.015931933611225e-05,
      "loss": 0.837,
      "step": 11020
    },
    {
      "epoch": 0.7274885814632216,
      "grad_norm": 0.5854398012161255,
      "learning_rate": 4.01364250744375e-05,
      "loss": 0.8117,
      "step": 11030
    },
    {
      "epoch": 0.7281481359341765,
      "grad_norm": 0.512503981590271,
      "learning_rate": 4.0113510754925e-05,
      "loss": 0.8319,
      "step": 11040
    },
    {
      "epoch": 0.7288076904051314,
      "grad_norm": 0.4752545654773712,
      "learning_rate": 4.0090576407939354e-05,
      "loss": 0.859,
      "step": 11050
    },
    {
      "epoch": 0.7294672448760862,
      "grad_norm": 0.5591562986373901,
      "learning_rate": 4.006762206387172e-05,
      "loss": 0.83,
      "step": 11060
    },
    {
      "epoch": 0.7301267993470411,
      "grad_norm": 0.5371208190917969,
      "learning_rate": 4.004464775313976e-05,
      "loss": 0.8883,
      "step": 11070
    },
    {
      "epoch": 0.730786353817996,
      "grad_norm": 0.471000075340271,
      "learning_rate": 4.002165350618756e-05,
      "loss": 0.8262,
      "step": 11080
    },
    {
      "epoch": 0.7314459082889508,
      "grad_norm": 0.5140637159347534,
      "learning_rate": 3.999863935348568e-05,
      "loss": 0.8794,
      "step": 11090
    },
    {
      "epoch": 0.7321054627599057,
      "grad_norm": 0.5318006277084351,
      "learning_rate": 3.9975605325530995e-05,
      "loss": 0.8302,
      "step": 11100
    },
    {
      "epoch": 0.7327650172308605,
      "grad_norm": 0.5049440860748291,
      "learning_rate": 3.995255145284678e-05,
      "loss": 0.8566,
      "step": 11110
    },
    {
      "epoch": 0.7334245717018154,
      "grad_norm": 0.46503761410713196,
      "learning_rate": 3.992947776598254e-05,
      "loss": 0.8444,
      "step": 11120
    },
    {
      "epoch": 0.7340841261727703,
      "grad_norm": 0.5713391900062561,
      "learning_rate": 3.9906384295514106e-05,
      "loss": 0.8181,
      "step": 11130
    },
    {
      "epoch": 0.7347436806437252,
      "grad_norm": 0.499016672372818,
      "learning_rate": 3.988327107204347e-05,
      "loss": 0.8153,
      "step": 11140
    },
    {
      "epoch": 0.7354032351146801,
      "grad_norm": 0.4445125162601471,
      "learning_rate": 3.9860138126198834e-05,
      "loss": 0.8197,
      "step": 11150
    },
    {
      "epoch": 0.7360627895856349,
      "grad_norm": 0.5101156830787659,
      "learning_rate": 3.9836985488634514e-05,
      "loss": 0.7927,
      "step": 11160
    },
    {
      "epoch": 0.7367223440565898,
      "grad_norm": 0.47717103362083435,
      "learning_rate": 3.981381319003093e-05,
      "loss": 0.8049,
      "step": 11170
    },
    {
      "epoch": 0.7373818985275447,
      "grad_norm": 0.5079829692840576,
      "learning_rate": 3.9790621261094554e-05,
      "loss": 0.8554,
      "step": 11180
    },
    {
      "epoch": 0.7380414529984995,
      "grad_norm": 0.46226266026496887,
      "learning_rate": 3.976740973255787e-05,
      "loss": 0.8502,
      "step": 11190
    },
    {
      "epoch": 0.7387010074694543,
      "grad_norm": 0.4857373833656311,
      "learning_rate": 3.974417863517933e-05,
      "loss": 0.7853,
      "step": 11200
    },
    {
      "epoch": 0.7393605619404092,
      "grad_norm": 0.48326465487480164,
      "learning_rate": 3.972092799974333e-05,
      "loss": 0.8034,
      "step": 11210
    },
    {
      "epoch": 0.7400201164113641,
      "grad_norm": 0.5540308952331543,
      "learning_rate": 3.969765785706016e-05,
      "loss": 0.8369,
      "step": 11220
    },
    {
      "epoch": 0.740679670882319,
      "grad_norm": 0.5187276601791382,
      "learning_rate": 3.967436823796592e-05,
      "loss": 0.8689,
      "step": 11230
    },
    {
      "epoch": 0.7413392253532739,
      "grad_norm": 0.5328775644302368,
      "learning_rate": 3.965105917332256e-05,
      "loss": 0.8345,
      "step": 11240
    },
    {
      "epoch": 0.7419987798242287,
      "grad_norm": 0.4606923460960388,
      "learning_rate": 3.962773069401778e-05,
      "loss": 0.799,
      "step": 11250
    },
    {
      "epoch": 0.7426583342951836,
      "grad_norm": 0.5835855603218079,
      "learning_rate": 3.960438283096504e-05,
      "loss": 0.8093,
      "step": 11260
    },
    {
      "epoch": 0.7433178887661385,
      "grad_norm": 0.5286555886268616,
      "learning_rate": 3.958101561510342e-05,
      "loss": 0.8687,
      "step": 11270
    },
    {
      "epoch": 0.7439774432370934,
      "grad_norm": 0.5591707825660706,
      "learning_rate": 3.9557629077397705e-05,
      "loss": 0.8547,
      "step": 11280
    },
    {
      "epoch": 0.7446369977080483,
      "grad_norm": 0.48347362875938416,
      "learning_rate": 3.953422324883826e-05,
      "loss": 0.8076,
      "step": 11290
    },
    {
      "epoch": 0.745296552179003,
      "grad_norm": 0.4533845782279968,
      "learning_rate": 3.951079816044101e-05,
      "loss": 0.7979,
      "step": 11300
    },
    {
      "epoch": 0.7459561066499579,
      "grad_norm": 0.49205514788627625,
      "learning_rate": 3.948735384324742e-05,
      "loss": 0.858,
      "step": 11310
    },
    {
      "epoch": 0.7466156611209128,
      "grad_norm": 0.5320116281509399,
      "learning_rate": 3.94638903283244e-05,
      "loss": 0.8193,
      "step": 11320
    },
    {
      "epoch": 0.7472752155918677,
      "grad_norm": 0.5026687979698181,
      "learning_rate": 3.9440407646764355e-05,
      "loss": 0.8121,
      "step": 11330
    },
    {
      "epoch": 0.7479347700628226,
      "grad_norm": 0.5017873644828796,
      "learning_rate": 3.9416905829685034e-05,
      "loss": 0.8568,
      "step": 11340
    },
    {
      "epoch": 0.7485943245337774,
      "grad_norm": 0.5668044090270996,
      "learning_rate": 3.939338490822957e-05,
      "loss": 0.8492,
      "step": 11350
    },
    {
      "epoch": 0.7492538790047323,
      "grad_norm": 0.47770196199417114,
      "learning_rate": 3.9369844913566414e-05,
      "loss": 0.8135,
      "step": 11360
    },
    {
      "epoch": 0.7499134334756872,
      "grad_norm": 0.5444761514663696,
      "learning_rate": 3.934628587688928e-05,
      "loss": 0.8244,
      "step": 11370
    },
    {
      "epoch": 0.7505729879466421,
      "grad_norm": 0.4261077642440796,
      "learning_rate": 3.9322707829417116e-05,
      "loss": 0.8383,
      "step": 11380
    },
    {
      "epoch": 0.751232542417597,
      "grad_norm": 0.4794468581676483,
      "learning_rate": 3.929911080239407e-05,
      "loss": 0.8073,
      "step": 11390
    },
    {
      "epoch": 0.7518920968885517,
      "grad_norm": 0.4978042542934418,
      "learning_rate": 3.9275494827089445e-05,
      "loss": 0.8459,
      "step": 11400
    },
    {
      "epoch": 0.7525516513595066,
      "grad_norm": 0.5170986652374268,
      "learning_rate": 3.925185993479763e-05,
      "loss": 0.8554,
      "step": 11410
    },
    {
      "epoch": 0.7532112058304615,
      "grad_norm": 0.49450716376304626,
      "learning_rate": 3.922820615683811e-05,
      "loss": 0.8524,
      "step": 11420
    },
    {
      "epoch": 0.7538707603014164,
      "grad_norm": 0.530701756477356,
      "learning_rate": 3.920453352455538e-05,
      "loss": 0.8002,
      "step": 11430
    },
    {
      "epoch": 0.7545303147723713,
      "grad_norm": 0.4432068169116974,
      "learning_rate": 3.918084206931892e-05,
      "loss": 0.8412,
      "step": 11440
    },
    {
      "epoch": 0.7551898692433261,
      "grad_norm": 0.48112452030181885,
      "learning_rate": 3.915713182252315e-05,
      "loss": 0.853,
      "step": 11450
    },
    {
      "epoch": 0.755849423714281,
      "grad_norm": 0.5220328569412231,
      "learning_rate": 3.9133402815587404e-05,
      "loss": 0.8331,
      "step": 11460
    },
    {
      "epoch": 0.7565089781852359,
      "grad_norm": 0.4995824992656708,
      "learning_rate": 3.9109655079955864e-05,
      "loss": 0.8682,
      "step": 11470
    },
    {
      "epoch": 0.7571685326561908,
      "grad_norm": 0.4851597845554352,
      "learning_rate": 3.908588864709754e-05,
      "loss": 0.8431,
      "step": 11480
    },
    {
      "epoch": 0.7578280871271457,
      "grad_norm": 0.49197956919670105,
      "learning_rate": 3.9062103548506215e-05,
      "loss": 0.871,
      "step": 11490
    },
    {
      "epoch": 0.7584876415981004,
      "grad_norm": 0.4794784486293793,
      "learning_rate": 3.9038299815700385e-05,
      "loss": 0.8256,
      "step": 11500
    },
    {
      "epoch": 0.7591471960690553,
      "grad_norm": 0.5070955753326416,
      "learning_rate": 3.901447748022328e-05,
      "loss": 0.8067,
      "step": 11510
    },
    {
      "epoch": 0.7598067505400102,
      "grad_norm": 0.5093052983283997,
      "learning_rate": 3.899063657364275e-05,
      "loss": 0.804,
      "step": 11520
    },
    {
      "epoch": 0.7604663050109651,
      "grad_norm": 0.5523203611373901,
      "learning_rate": 3.896677712755127e-05,
      "loss": 0.8963,
      "step": 11530
    },
    {
      "epoch": 0.7611258594819199,
      "grad_norm": 0.5485988855361938,
      "learning_rate": 3.894289917356586e-05,
      "loss": 0.8572,
      "step": 11540
    },
    {
      "epoch": 0.7617854139528748,
      "grad_norm": 0.5128387212753296,
      "learning_rate": 3.8919002743328105e-05,
      "loss": 0.8724,
      "step": 11550
    },
    {
      "epoch": 0.7624449684238297,
      "grad_norm": 0.5283038020133972,
      "learning_rate": 3.889508786850403e-05,
      "loss": 0.8275,
      "step": 11560
    },
    {
      "epoch": 0.7631045228947846,
      "grad_norm": 0.5048648118972778,
      "learning_rate": 3.887115458078414e-05,
      "loss": 0.8876,
      "step": 11570
    },
    {
      "epoch": 0.7637640773657395,
      "grad_norm": 0.4777819812297821,
      "learning_rate": 3.884720291188331e-05,
      "loss": 0.8258,
      "step": 11580
    },
    {
      "epoch": 0.7644236318366943,
      "grad_norm": 0.4839961528778076,
      "learning_rate": 3.8823232893540774e-05,
      "loss": 0.8219,
      "step": 11590
    },
    {
      "epoch": 0.7650831863076492,
      "grad_norm": 0.5506725907325745,
      "learning_rate": 3.879924455752012e-05,
      "loss": 0.8657,
      "step": 11600
    },
    {
      "epoch": 0.765742740778604,
      "grad_norm": 0.505973756313324,
      "learning_rate": 3.877523793560915e-05,
      "loss": 0.8274,
      "step": 11610
    },
    {
      "epoch": 0.7664022952495589,
      "grad_norm": 0.5126368403434753,
      "learning_rate": 3.875121305961995e-05,
      "loss": 0.8468,
      "step": 11620
    },
    {
      "epoch": 0.7670618497205138,
      "grad_norm": 0.5028843879699707,
      "learning_rate": 3.872716996138876e-05,
      "loss": 0.8406,
      "step": 11630
    },
    {
      "epoch": 0.7677214041914686,
      "grad_norm": 0.5137712955474854,
      "learning_rate": 3.870310867277599e-05,
      "loss": 0.8498,
      "step": 11640
    },
    {
      "epoch": 0.7683809586624235,
      "grad_norm": 0.5019781589508057,
      "learning_rate": 3.8679029225666145e-05,
      "loss": 0.8329,
      "step": 11650
    },
    {
      "epoch": 0.7690405131333784,
      "grad_norm": 0.554481029510498,
      "learning_rate": 3.865493165196778e-05,
      "loss": 0.8429,
      "step": 11660
    },
    {
      "epoch": 0.7697000676043333,
      "grad_norm": 0.5420853495597839,
      "learning_rate": 3.86308159836135e-05,
      "loss": 0.7977,
      "step": 11670
    },
    {
      "epoch": 0.7703596220752882,
      "grad_norm": 0.5133484601974487,
      "learning_rate": 3.8606682252559856e-05,
      "loss": 0.8264,
      "step": 11680
    },
    {
      "epoch": 0.771019176546243,
      "grad_norm": 0.5209570527076721,
      "learning_rate": 3.8582530490787363e-05,
      "loss": 0.8342,
      "step": 11690
    },
    {
      "epoch": 0.7716787310171979,
      "grad_norm": 0.5342729091644287,
      "learning_rate": 3.85583607303004e-05,
      "loss": 0.8546,
      "step": 11700
    },
    {
      "epoch": 0.7723382854881528,
      "grad_norm": 0.512787938117981,
      "learning_rate": 3.853417300312724e-05,
      "loss": 0.852,
      "step": 11710
    },
    {
      "epoch": 0.7729978399591076,
      "grad_norm": 0.46172672510147095,
      "learning_rate": 3.8509967341319894e-05,
      "loss": 0.8465,
      "step": 11720
    },
    {
      "epoch": 0.7736573944300625,
      "grad_norm": 0.4403376877307892,
      "learning_rate": 3.848574377695421e-05,
      "loss": 0.8159,
      "step": 11730
    },
    {
      "epoch": 0.7743169489010173,
      "grad_norm": 0.46622994542121887,
      "learning_rate": 3.8461502342129736e-05,
      "loss": 0.8515,
      "step": 11740
    },
    {
      "epoch": 0.7749765033719722,
      "grad_norm": 0.4755195081233978,
      "learning_rate": 3.8437243068969674e-05,
      "loss": 0.8223,
      "step": 11750
    },
    {
      "epoch": 0.7756360578429271,
      "grad_norm": 0.48838281631469727,
      "learning_rate": 3.8412965989620905e-05,
      "loss": 0.8511,
      "step": 11760
    },
    {
      "epoch": 0.776295612313882,
      "grad_norm": 0.4364047944545746,
      "learning_rate": 3.838867113625387e-05,
      "loss": 0.8278,
      "step": 11770
    },
    {
      "epoch": 0.7769551667848369,
      "grad_norm": 0.5896735787391663,
      "learning_rate": 3.83643585410626e-05,
      "loss": 0.8102,
      "step": 11780
    },
    {
      "epoch": 0.7776147212557917,
      "grad_norm": 0.5205544829368591,
      "learning_rate": 3.834002823626459e-05,
      "loss": 0.8274,
      "step": 11790
    },
    {
      "epoch": 0.7782742757267466,
      "grad_norm": 0.5001713633537292,
      "learning_rate": 3.8315680254100846e-05,
      "loss": 0.8401,
      "step": 11800
    },
    {
      "epoch": 0.7789338301977015,
      "grad_norm": 0.49767130613327026,
      "learning_rate": 3.8291314626835784e-05,
      "loss": 0.827,
      "step": 11810
    },
    {
      "epoch": 0.7795933846686564,
      "grad_norm": 0.5469626188278198,
      "learning_rate": 3.826693138675719e-05,
      "loss": 0.8394,
      "step": 11820
    },
    {
      "epoch": 0.7802529391396112,
      "grad_norm": 0.4951440989971161,
      "learning_rate": 3.82425305661762e-05,
      "loss": 0.8815,
      "step": 11830
    },
    {
      "epoch": 0.780912493610566,
      "grad_norm": 0.5210491418838501,
      "learning_rate": 3.8218112197427244e-05,
      "loss": 0.8384,
      "step": 11840
    },
    {
      "epoch": 0.7815720480815209,
      "grad_norm": 0.5787615776062012,
      "learning_rate": 3.819367631286802e-05,
      "loss": 0.862,
      "step": 11850
    },
    {
      "epoch": 0.7822316025524758,
      "grad_norm": 0.5131245255470276,
      "learning_rate": 3.81692229448794e-05,
      "loss": 0.8308,
      "step": 11860
    },
    {
      "epoch": 0.7828911570234307,
      "grad_norm": 0.5724840760231018,
      "learning_rate": 3.814475212586548e-05,
      "loss": 0.8058,
      "step": 11870
    },
    {
      "epoch": 0.7835507114943855,
      "grad_norm": 0.5495114922523499,
      "learning_rate": 3.812026388825342e-05,
      "loss": 0.8645,
      "step": 11880
    },
    {
      "epoch": 0.7842102659653404,
      "grad_norm": 0.5270974636077881,
      "learning_rate": 3.8095758264493506e-05,
      "loss": 0.8307,
      "step": 11890
    },
    {
      "epoch": 0.7848698204362953,
      "grad_norm": 0.5352598428726196,
      "learning_rate": 3.807123528705905e-05,
      "loss": 0.8296,
      "step": 11900
    },
    {
      "epoch": 0.7855293749072502,
      "grad_norm": 0.517625093460083,
      "learning_rate": 3.804669498844635e-05,
      "loss": 0.8251,
      "step": 11910
    },
    {
      "epoch": 0.7861889293782051,
      "grad_norm": 0.5091532468795776,
      "learning_rate": 3.8022137401174676e-05,
      "loss": 0.8293,
      "step": 11920
    },
    {
      "epoch": 0.7868484838491598,
      "grad_norm": 0.5616522431373596,
      "learning_rate": 3.799756255778617e-05,
      "loss": 0.8256,
      "step": 11930
    },
    {
      "epoch": 0.7875080383201147,
      "grad_norm": 0.5082550644874573,
      "learning_rate": 3.79729704908459e-05,
      "loss": 0.8563,
      "step": 11940
    },
    {
      "epoch": 0.7881675927910696,
      "grad_norm": 0.4487828016281128,
      "learning_rate": 3.794836123294171e-05,
      "loss": 0.8187,
      "step": 11950
    },
    {
      "epoch": 0.7888271472620245,
      "grad_norm": 0.47633489966392517,
      "learning_rate": 3.792373481668424e-05,
      "loss": 0.8589,
      "step": 11960
    },
    {
      "epoch": 0.7894867017329794,
      "grad_norm": 0.5292962789535522,
      "learning_rate": 3.789909127470687e-05,
      "loss": 0.8168,
      "step": 11970
    },
    {
      "epoch": 0.7901462562039342,
      "grad_norm": 0.5220751762390137,
      "learning_rate": 3.787443063966567e-05,
      "loss": 0.8347,
      "step": 11980
    },
    {
      "epoch": 0.7908058106748891,
      "grad_norm": 0.46419110894203186,
      "learning_rate": 3.7849752944239356e-05,
      "loss": 0.8396,
      "step": 11990
    },
    {
      "epoch": 0.791465365145844,
      "grad_norm": 0.5395681262016296,
      "learning_rate": 3.7825058221129264e-05,
      "loss": 0.7882,
      "step": 12000
    },
    {
      "epoch": 0.7921249196167989,
      "grad_norm": 0.5546279549598694,
      "learning_rate": 3.780034650305928e-05,
      "loss": 0.8386,
      "step": 12010
    },
    {
      "epoch": 0.7927844740877538,
      "grad_norm": 0.526649534702301,
      "learning_rate": 3.7775617822775826e-05,
      "loss": 0.8477,
      "step": 12020
    },
    {
      "epoch": 0.7934440285587085,
      "grad_norm": 0.4924740195274353,
      "learning_rate": 3.775087221304779e-05,
      "loss": 0.7887,
      "step": 12030
    },
    {
      "epoch": 0.7941035830296634,
      "grad_norm": 0.5638056993484497,
      "learning_rate": 3.772610970666649e-05,
      "loss": 0.8524,
      "step": 12040
    },
    {
      "epoch": 0.7947631375006183,
      "grad_norm": 0.5174016952514648,
      "learning_rate": 3.770133033644565e-05,
      "loss": 0.8334,
      "step": 12050
    },
    {
      "epoch": 0.7954226919715732,
      "grad_norm": 0.5178600549697876,
      "learning_rate": 3.7676534135221306e-05,
      "loss": 0.8554,
      "step": 12060
    },
    {
      "epoch": 0.7960822464425281,
      "grad_norm": 0.5493482947349548,
      "learning_rate": 3.765172113585186e-05,
      "loss": 0.8356,
      "step": 12070
    },
    {
      "epoch": 0.7967418009134829,
      "grad_norm": 0.4699559211730957,
      "learning_rate": 3.762689137121792e-05,
      "loss": 0.8766,
      "step": 12080
    },
    {
      "epoch": 0.7974013553844378,
      "grad_norm": 0.5163393616676331,
      "learning_rate": 3.760204487422233e-05,
      "loss": 0.8348,
      "step": 12090
    },
    {
      "epoch": 0.7980609098553927,
      "grad_norm": 0.5244756937026978,
      "learning_rate": 3.75771816777901e-05,
      "loss": 0.8443,
      "step": 12100
    },
    {
      "epoch": 0.7987204643263476,
      "grad_norm": 0.5232709646224976,
      "learning_rate": 3.755230181486837e-05,
      "loss": 0.8496,
      "step": 12110
    },
    {
      "epoch": 0.7993800187973025,
      "grad_norm": 0.5520874261856079,
      "learning_rate": 3.7527405318426366e-05,
      "loss": 0.8428,
      "step": 12120
    },
    {
      "epoch": 0.8000395732682573,
      "grad_norm": 0.5144450664520264,
      "learning_rate": 3.7502492221455375e-05,
      "loss": 0.8245,
      "step": 12130
    },
    {
      "epoch": 0.8006991277392121,
      "grad_norm": 0.4852900207042694,
      "learning_rate": 3.747756255696866e-05,
      "loss": 0.8377,
      "step": 12140
    },
    {
      "epoch": 0.801358682210167,
      "grad_norm": 0.5000022649765015,
      "learning_rate": 3.745261635800142e-05,
      "loss": 0.8679,
      "step": 12150
    },
    {
      "epoch": 0.8020182366811219,
      "grad_norm": 0.45508676767349243,
      "learning_rate": 3.742765365761082e-05,
      "loss": 0.8336,
      "step": 12160
    },
    {
      "epoch": 0.8026777911520768,
      "grad_norm": 0.6492980718612671,
      "learning_rate": 3.7402674488875835e-05,
      "loss": 0.8476,
      "step": 12170
    },
    {
      "epoch": 0.8033373456230316,
      "grad_norm": 0.5550660490989685,
      "learning_rate": 3.7377678884897305e-05,
      "loss": 0.8518,
      "step": 12180
    },
    {
      "epoch": 0.8039969000939865,
      "grad_norm": 0.4505656957626343,
      "learning_rate": 3.735266687879782e-05,
      "loss": 0.812,
      "step": 12190
    },
    {
      "epoch": 0.8046564545649414,
      "grad_norm": 0.47150373458862305,
      "learning_rate": 3.732763850372172e-05,
      "loss": 0.8279,
      "step": 12200
    },
    {
      "epoch": 0.8053160090358963,
      "grad_norm": 0.587897777557373,
      "learning_rate": 3.730259379283503e-05,
      "loss": 0.8301,
      "step": 12210
    },
    {
      "epoch": 0.8059755635068511,
      "grad_norm": 0.501718282699585,
      "learning_rate": 3.727753277932543e-05,
      "loss": 0.8305,
      "step": 12220
    },
    {
      "epoch": 0.806635117977806,
      "grad_norm": 0.5089925527572632,
      "learning_rate": 3.725245549640218e-05,
      "loss": 0.8058,
      "step": 12230
    },
    {
      "epoch": 0.8072946724487609,
      "grad_norm": 0.4705756902694702,
      "learning_rate": 3.722736197729614e-05,
      "loss": 0.8453,
      "step": 12240
    },
    {
      "epoch": 0.8079542269197157,
      "grad_norm": 0.5152075290679932,
      "learning_rate": 3.720225225525966e-05,
      "loss": 0.8424,
      "step": 12250
    },
    {
      "epoch": 0.8086137813906706,
      "grad_norm": 0.5138453245162964,
      "learning_rate": 3.7177126363566536e-05,
      "loss": 0.8107,
      "step": 12260
    },
    {
      "epoch": 0.8092733358616254,
      "grad_norm": 0.5220133066177368,
      "learning_rate": 3.715198433551206e-05,
      "loss": 0.8531,
      "step": 12270
    },
    {
      "epoch": 0.8099328903325803,
      "grad_norm": 0.473230242729187,
      "learning_rate": 3.7126826204412824e-05,
      "loss": 0.8469,
      "step": 12280
    },
    {
      "epoch": 0.8105924448035352,
      "grad_norm": 0.49255087971687317,
      "learning_rate": 3.710165200360682e-05,
      "loss": 0.8169,
      "step": 12290
    },
    {
      "epoch": 0.8112519992744901,
      "grad_norm": 0.5157859325408936,
      "learning_rate": 3.707646176645332e-05,
      "loss": 0.8276,
      "step": 12300
    },
    {
      "epoch": 0.811911553745445,
      "grad_norm": 0.4958195686340332,
      "learning_rate": 3.7051255526332835e-05,
      "loss": 0.821,
      "step": 12310
    },
    {
      "epoch": 0.8125711082163998,
      "grad_norm": 0.447382390499115,
      "learning_rate": 3.702603331664708e-05,
      "loss": 0.8235,
      "step": 12320
    },
    {
      "epoch": 0.8132306626873547,
      "grad_norm": 0.5194646120071411,
      "learning_rate": 3.7000795170818945e-05,
      "loss": 0.8696,
      "step": 12330
    },
    {
      "epoch": 0.8138902171583096,
      "grad_norm": 0.4164382517337799,
      "learning_rate": 3.697554112229244e-05,
      "loss": 0.8149,
      "step": 12340
    },
    {
      "epoch": 0.8145497716292645,
      "grad_norm": 0.44350144267082214,
      "learning_rate": 3.6950271204532624e-05,
      "loss": 0.8193,
      "step": 12350
    },
    {
      "epoch": 0.8152093261002193,
      "grad_norm": 0.5147350430488586,
      "learning_rate": 3.6924985451025626e-05,
      "loss": 0.8306,
      "step": 12360
    },
    {
      "epoch": 0.8158688805711741,
      "grad_norm": 0.49076810479164124,
      "learning_rate": 3.689968389527851e-05,
      "loss": 0.8554,
      "step": 12370
    },
    {
      "epoch": 0.816528435042129,
      "grad_norm": 0.4826480746269226,
      "learning_rate": 3.6874366570819306e-05,
      "loss": 0.8187,
      "step": 12380
    },
    {
      "epoch": 0.8171879895130839,
      "grad_norm": 0.5235629081726074,
      "learning_rate": 3.684903351119696e-05,
      "loss": 0.86,
      "step": 12390
    },
    {
      "epoch": 0.8178475439840388,
      "grad_norm": 0.5923688411712646,
      "learning_rate": 3.6823684749981234e-05,
      "loss": 0.8607,
      "step": 12400
    },
    {
      "epoch": 0.8185070984549937,
      "grad_norm": 0.5231000781059265,
      "learning_rate": 3.6798320320762715e-05,
      "loss": 0.826,
      "step": 12410
    },
    {
      "epoch": 0.8191666529259485,
      "grad_norm": 0.6260348558425903,
      "learning_rate": 3.6772940257152736e-05,
      "loss": 0.825,
      "step": 12420
    },
    {
      "epoch": 0.8198262073969034,
      "grad_norm": 0.49430742859840393,
      "learning_rate": 3.674754459278338e-05,
      "loss": 0.8276,
      "step": 12430
    },
    {
      "epoch": 0.8204857618678583,
      "grad_norm": 0.5140009522438049,
      "learning_rate": 3.672213336130737e-05,
      "loss": 0.8656,
      "step": 12440
    },
    {
      "epoch": 0.8211453163388132,
      "grad_norm": 0.5198078751564026,
      "learning_rate": 3.669670659639807e-05,
      "loss": 0.8917,
      "step": 12450
    },
    {
      "epoch": 0.821804870809768,
      "grad_norm": 0.5191746354103088,
      "learning_rate": 3.667126433174943e-05,
      "loss": 0.841,
      "step": 12460
    },
    {
      "epoch": 0.8224644252807228,
      "grad_norm": 0.5955049395561218,
      "learning_rate": 3.6645806601075935e-05,
      "loss": 0.846,
      "step": 12470
    },
    {
      "epoch": 0.8231239797516777,
      "grad_norm": 0.4628657102584839,
      "learning_rate": 3.662033343811258e-05,
      "loss": 0.8451,
      "step": 12480
    },
    {
      "epoch": 0.8237835342226326,
      "grad_norm": 0.4703860580921173,
      "learning_rate": 3.6594844876614785e-05,
      "loss": 0.8507,
      "step": 12490
    },
    {
      "epoch": 0.8244430886935875,
      "grad_norm": 0.5772825479507446,
      "learning_rate": 3.6569340950358395e-05,
      "loss": 0.8423,
      "step": 12500
    },
    {
      "epoch": 0.8251026431645424,
      "grad_norm": 0.5526529550552368,
      "learning_rate": 3.6543821693139595e-05,
      "loss": 0.8013,
      "step": 12510
    },
    {
      "epoch": 0.8257621976354972,
      "grad_norm": 0.48744431138038635,
      "learning_rate": 3.651828713877491e-05,
      "loss": 0.8268,
      "step": 12520
    },
    {
      "epoch": 0.8264217521064521,
      "grad_norm": 0.5575173497200012,
      "learning_rate": 3.6492737321101115e-05,
      "loss": 0.8754,
      "step": 12530
    },
    {
      "epoch": 0.827081306577407,
      "grad_norm": 0.5068231821060181,
      "learning_rate": 3.646717227397524e-05,
      "loss": 0.8603,
      "step": 12540
    },
    {
      "epoch": 0.8277408610483619,
      "grad_norm": 0.5143702626228333,
      "learning_rate": 3.6441592031274454e-05,
      "loss": 0.7895,
      "step": 12550
    },
    {
      "epoch": 0.8284004155193166,
      "grad_norm": 0.45512455701828003,
      "learning_rate": 3.64159966268961e-05,
      "loss": 0.8014,
      "step": 12560
    },
    {
      "epoch": 0.8290599699902715,
      "grad_norm": 0.5064200162887573,
      "learning_rate": 3.639038609475758e-05,
      "loss": 0.8284,
      "step": 12570
    },
    {
      "epoch": 0.8297195244612264,
      "grad_norm": 0.5486271977424622,
      "learning_rate": 3.636476046879637e-05,
      "loss": 0.8553,
      "step": 12580
    },
    {
      "epoch": 0.8303790789321813,
      "grad_norm": 0.6321308016777039,
      "learning_rate": 3.633911978296994e-05,
      "loss": 0.8198,
      "step": 12590
    },
    {
      "epoch": 0.8310386334031362,
      "grad_norm": 0.47299566864967346,
      "learning_rate": 3.6313464071255703e-05,
      "loss": 0.8246,
      "step": 12600
    },
    {
      "epoch": 0.831698187874091,
      "grad_norm": 0.5452206134796143,
      "learning_rate": 3.6287793367651e-05,
      "loss": 0.8276,
      "step": 12610
    },
    {
      "epoch": 0.8323577423450459,
      "grad_norm": 0.5663066506385803,
      "learning_rate": 3.6262107706173034e-05,
      "loss": 0.8369,
      "step": 12620
    },
    {
      "epoch": 0.8330172968160008,
      "grad_norm": 0.5083215832710266,
      "learning_rate": 3.623640712085882e-05,
      "loss": 0.8315,
      "step": 12630
    },
    {
      "epoch": 0.8336768512869557,
      "grad_norm": 0.496698260307312,
      "learning_rate": 3.621069164576516e-05,
      "loss": 0.8508,
      "step": 12640
    },
    {
      "epoch": 0.8343364057579106,
      "grad_norm": 0.4482015073299408,
      "learning_rate": 3.618496131496858e-05,
      "loss": 0.8738,
      "step": 12650
    },
    {
      "epoch": 0.8349959602288654,
      "grad_norm": 0.5575942397117615,
      "learning_rate": 3.6159216162565296e-05,
      "loss": 0.8085,
      "step": 12660
    },
    {
      "epoch": 0.8356555146998202,
      "grad_norm": 0.5119989514350891,
      "learning_rate": 3.613345622267116e-05,
      "loss": 0.8429,
      "step": 12670
    },
    {
      "epoch": 0.8363150691707751,
      "grad_norm": 0.49572911858558655,
      "learning_rate": 3.6107681529421634e-05,
      "loss": 0.8728,
      "step": 12680
    },
    {
      "epoch": 0.83697462364173,
      "grad_norm": 0.5054008364677429,
      "learning_rate": 3.6081892116971715e-05,
      "loss": 0.836,
      "step": 12690
    },
    {
      "epoch": 0.8376341781126849,
      "grad_norm": 0.5312760472297668,
      "learning_rate": 3.60560880194959e-05,
      "loss": 0.8696,
      "step": 12700
    },
    {
      "epoch": 0.8382937325836397,
      "grad_norm": 0.49542853236198425,
      "learning_rate": 3.6030269271188164e-05,
      "loss": 0.8946,
      "step": 12710
    },
    {
      "epoch": 0.8389532870545946,
      "grad_norm": 0.5444157123565674,
      "learning_rate": 3.600443590626189e-05,
      "loss": 0.8179,
      "step": 12720
    },
    {
      "epoch": 0.8396128415255495,
      "grad_norm": 0.6010988354682922,
      "learning_rate": 3.5978587958949804e-05,
      "loss": 0.8384,
      "step": 12730
    },
    {
      "epoch": 0.8402723959965044,
      "grad_norm": 0.558444082736969,
      "learning_rate": 3.5952725463504005e-05,
      "loss": 0.8369,
      "step": 12740
    },
    {
      "epoch": 0.8409319504674593,
      "grad_norm": 0.5344332456588745,
      "learning_rate": 3.592684845419584e-05,
      "loss": 0.8345,
      "step": 12750
    },
    {
      "epoch": 0.8415915049384141,
      "grad_norm": 0.5073506236076355,
      "learning_rate": 3.590095696531588e-05,
      "loss": 0.8583,
      "step": 12760
    },
    {
      "epoch": 0.842251059409369,
      "grad_norm": 0.5145412683486938,
      "learning_rate": 3.5875051031173907e-05,
      "loss": 0.7981,
      "step": 12770
    },
    {
      "epoch": 0.8429106138803238,
      "grad_norm": 0.48051124811172485,
      "learning_rate": 3.584913068609884e-05,
      "loss": 0.8321,
      "step": 12780
    },
    {
      "epoch": 0.8435701683512787,
      "grad_norm": 0.573727011680603,
      "learning_rate": 3.5823195964438665e-05,
      "loss": 0.8189,
      "step": 12790
    },
    {
      "epoch": 0.8442297228222336,
      "grad_norm": 0.5168202519416809,
      "learning_rate": 3.5797246900560465e-05,
      "loss": 0.8698,
      "step": 12800
    },
    {
      "epoch": 0.8448892772931884,
      "grad_norm": 0.518841028213501,
      "learning_rate": 3.57712835288503e-05,
      "loss": 0.8891,
      "step": 12810
    },
    {
      "epoch": 0.8455488317641433,
      "grad_norm": 0.48710545897483826,
      "learning_rate": 3.574530588371319e-05,
      "loss": 0.8082,
      "step": 12820
    },
    {
      "epoch": 0.8462083862350982,
      "grad_norm": 0.5257467031478882,
      "learning_rate": 3.571931399957309e-05,
      "loss": 0.8382,
      "step": 12830
    },
    {
      "epoch": 0.8468679407060531,
      "grad_norm": 0.5244994759559631,
      "learning_rate": 3.5693307910872796e-05,
      "loss": 0.8125,
      "step": 12840
    },
    {
      "epoch": 0.847527495177008,
      "grad_norm": 0.5413283705711365,
      "learning_rate": 3.566728765207394e-05,
      "loss": 0.8259,
      "step": 12850
    },
    {
      "epoch": 0.8481870496479628,
      "grad_norm": 0.4778168499469757,
      "learning_rate": 3.564125325765694e-05,
      "loss": 0.8147,
      "step": 12860
    },
    {
      "epoch": 0.8488466041189177,
      "grad_norm": 0.5366325974464417,
      "learning_rate": 3.5615204762120925e-05,
      "loss": 0.8248,
      "step": 12870
    },
    {
      "epoch": 0.8495061585898726,
      "grad_norm": 0.460050106048584,
      "learning_rate": 3.558914219998373e-05,
      "loss": 0.857,
      "step": 12880
    },
    {
      "epoch": 0.8501657130608274,
      "grad_norm": 0.5721178650856018,
      "learning_rate": 3.556306560578181e-05,
      "loss": 0.8649,
      "step": 12890
    },
    {
      "epoch": 0.8508252675317822,
      "grad_norm": 0.5086795091629028,
      "learning_rate": 3.553697501407025e-05,
      "loss": 0.8229,
      "step": 12900
    },
    {
      "epoch": 0.8514848220027371,
      "grad_norm": 0.5210456252098083,
      "learning_rate": 3.551087045942263e-05,
      "loss": 0.8378,
      "step": 12910
    },
    {
      "epoch": 0.852144376473692,
      "grad_norm": 0.4959641993045807,
      "learning_rate": 3.548475197643109e-05,
      "loss": 0.8136,
      "step": 12920
    },
    {
      "epoch": 0.8528039309446469,
      "grad_norm": 0.42477983236312866,
      "learning_rate": 3.545861959970618e-05,
      "loss": 0.85,
      "step": 12930
    },
    {
      "epoch": 0.8534634854156018,
      "grad_norm": 0.4822991192340851,
      "learning_rate": 3.54324733638769e-05,
      "loss": 0.8232,
      "step": 12940
    },
    {
      "epoch": 0.8541230398865566,
      "grad_norm": 0.607531726360321,
      "learning_rate": 3.5406313303590577e-05,
      "loss": 0.8402,
      "step": 12950
    },
    {
      "epoch": 0.8547825943575115,
      "grad_norm": 0.4119485020637512,
      "learning_rate": 3.538013945351288e-05,
      "loss": 0.8571,
      "step": 12960
    },
    {
      "epoch": 0.8554421488284664,
      "grad_norm": 0.5254050493240356,
      "learning_rate": 3.5353951848327756e-05,
      "loss": 0.8428,
      "step": 12970
    },
    {
      "epoch": 0.8561017032994213,
      "grad_norm": 0.4991281032562256,
      "learning_rate": 3.532775052273737e-05,
      "loss": 0.8443,
      "step": 12980
    },
    {
      "epoch": 0.8567612577703762,
      "grad_norm": 0.5751249194145203,
      "learning_rate": 3.530153551146206e-05,
      "loss": 0.8199,
      "step": 12990
    },
    {
      "epoch": 0.8574208122413309,
      "grad_norm": 0.5402325987815857,
      "learning_rate": 3.527530684924031e-05,
      "loss": 0.8254,
      "step": 13000
    },
    {
      "epoch": 0.8580803667122858,
      "grad_norm": 0.5223280191421509,
      "learning_rate": 3.5249064570828706e-05,
      "loss": 0.8566,
      "step": 13010
    },
    {
      "epoch": 0.8587399211832407,
      "grad_norm": 0.4691770672798157,
      "learning_rate": 3.5222808711001854e-05,
      "loss": 0.8522,
      "step": 13020
    },
    {
      "epoch": 0.8593994756541956,
      "grad_norm": 0.5179622173309326,
      "learning_rate": 3.5196539304552364e-05,
      "loss": 0.8167,
      "step": 13030
    },
    {
      "epoch": 0.8600590301251505,
      "grad_norm": 0.5165730118751526,
      "learning_rate": 3.51702563862908e-05,
      "loss": 0.8628,
      "step": 13040
    },
    {
      "epoch": 0.8607185845961053,
      "grad_norm": 0.630574107170105,
      "learning_rate": 3.514395999104564e-05,
      "loss": 0.8352,
      "step": 13050
    },
    {
      "epoch": 0.8613781390670602,
      "grad_norm": 0.5269815921783447,
      "learning_rate": 3.51176501536632e-05,
      "loss": 0.8429,
      "step": 13060
    },
    {
      "epoch": 0.8620376935380151,
      "grad_norm": 0.5187968015670776,
      "learning_rate": 3.509132690900763e-05,
      "loss": 0.8003,
      "step": 13070
    },
    {
      "epoch": 0.86269724800897,
      "grad_norm": 0.5297034978866577,
      "learning_rate": 3.5064990291960835e-05,
      "loss": 0.7996,
      "step": 13080
    },
    {
      "epoch": 0.8633568024799249,
      "grad_norm": 0.489135205745697,
      "learning_rate": 3.503864033742244e-05,
      "loss": 0.8399,
      "step": 13090
    },
    {
      "epoch": 0.8640163569508796,
      "grad_norm": 0.6154001355171204,
      "learning_rate": 3.501227708030975e-05,
      "loss": 0.8121,
      "step": 13100
    },
    {
      "epoch": 0.8646759114218345,
      "grad_norm": 0.4901886284351349,
      "learning_rate": 3.498590055555768e-05,
      "loss": 0.8694,
      "step": 13110
    },
    {
      "epoch": 0.8653354658927894,
      "grad_norm": 0.48025843501091003,
      "learning_rate": 3.4959510798118757e-05,
      "loss": 0.852,
      "step": 13120
    },
    {
      "epoch": 0.8659950203637443,
      "grad_norm": 0.4628942012786865,
      "learning_rate": 3.493310784296301e-05,
      "loss": 0.8329,
      "step": 13130
    },
    {
      "epoch": 0.8666545748346992,
      "grad_norm": 0.6147623658180237,
      "learning_rate": 3.4906691725078e-05,
      "loss": 0.7919,
      "step": 13140
    },
    {
      "epoch": 0.867314129305654,
      "grad_norm": 0.5047041177749634,
      "learning_rate": 3.488026247946868e-05,
      "loss": 0.8365,
      "step": 13150
    },
    {
      "epoch": 0.8679736837766089,
      "grad_norm": 0.6306891441345215,
      "learning_rate": 3.4853820141157434e-05,
      "loss": 0.8611,
      "step": 13160
    },
    {
      "epoch": 0.8686332382475638,
      "grad_norm": 0.5087846517562866,
      "learning_rate": 3.482736474518397e-05,
      "loss": 0.8394,
      "step": 13170
    },
    {
      "epoch": 0.8692927927185187,
      "grad_norm": 0.4533502459526062,
      "learning_rate": 3.480089632660533e-05,
      "loss": 0.8554,
      "step": 13180
    },
    {
      "epoch": 0.8699523471894736,
      "grad_norm": 0.54240483045578,
      "learning_rate": 3.477441492049579e-05,
      "loss": 0.817,
      "step": 13190
    },
    {
      "epoch": 0.8706119016604283,
      "grad_norm": 0.5537872910499573,
      "learning_rate": 3.474792056194684e-05,
      "loss": 0.8232,
      "step": 13200
    },
    {
      "epoch": 0.8712714561313832,
      "grad_norm": 0.5160180330276489,
      "learning_rate": 3.472141328606714e-05,
      "loss": 0.8162,
      "step": 13210
    },
    {
      "epoch": 0.8719310106023381,
      "grad_norm": 0.506102442741394,
      "learning_rate": 3.469489312798246e-05,
      "loss": 0.8329,
      "step": 13220
    },
    {
      "epoch": 0.872590565073293,
      "grad_norm": 0.507501482963562,
      "learning_rate": 3.466836012283566e-05,
      "loss": 0.8237,
      "step": 13230
    },
    {
      "epoch": 0.8732501195442479,
      "grad_norm": 0.5725156664848328,
      "learning_rate": 3.464181430578658e-05,
      "loss": 0.847,
      "step": 13240
    },
    {
      "epoch": 0.8739096740152027,
      "grad_norm": 0.5154231190681458,
      "learning_rate": 3.4615255712012084e-05,
      "loss": 0.8328,
      "step": 13250
    },
    {
      "epoch": 0.8745692284861576,
      "grad_norm": 0.4981846511363983,
      "learning_rate": 3.458868437670594e-05,
      "loss": 0.814,
      "step": 13260
    },
    {
      "epoch": 0.8752287829571125,
      "grad_norm": 0.47902771830558777,
      "learning_rate": 3.4562100335078826e-05,
      "loss": 0.8807,
      "step": 13270
    },
    {
      "epoch": 0.8758883374280674,
      "grad_norm": 0.4744430482387543,
      "learning_rate": 3.453550362235822e-05,
      "loss": 0.8027,
      "step": 13280
    },
    {
      "epoch": 0.8765478918990222,
      "grad_norm": 0.5479256510734558,
      "learning_rate": 3.4508894273788426e-05,
      "loss": 0.8511,
      "step": 13290
    },
    {
      "epoch": 0.877207446369977,
      "grad_norm": 0.6024276614189148,
      "learning_rate": 3.448227232463047e-05,
      "loss": 0.8601,
      "step": 13300
    },
    {
      "epoch": 0.8778670008409319,
      "grad_norm": 0.4918282628059387,
      "learning_rate": 3.445563781016209e-05,
      "loss": 0.7992,
      "step": 13310
    },
    {
      "epoch": 0.8785265553118868,
      "grad_norm": 0.4895969033241272,
      "learning_rate": 3.442899076567766e-05,
      "loss": 0.813,
      "step": 13320
    },
    {
      "epoch": 0.8791861097828417,
      "grad_norm": 0.47569799423217773,
      "learning_rate": 3.440233122648817e-05,
      "loss": 0.7922,
      "step": 13330
    },
    {
      "epoch": 0.8798456642537965,
      "grad_norm": 0.5393938422203064,
      "learning_rate": 3.437565922792116e-05,
      "loss": 0.8801,
      "step": 13340
    },
    {
      "epoch": 0.8805052187247514,
      "grad_norm": 0.5467188954353333,
      "learning_rate": 3.4348974805320696e-05,
      "loss": 0.871,
      "step": 13350
    },
    {
      "epoch": 0.8811647731957063,
      "grad_norm": 0.5176929831504822,
      "learning_rate": 3.432227799404728e-05,
      "loss": 0.8133,
      "step": 13360
    },
    {
      "epoch": 0.8818243276666612,
      "grad_norm": 0.5406689047813416,
      "learning_rate": 3.4295568829477854e-05,
      "loss": 0.7797,
      "step": 13370
    },
    {
      "epoch": 0.8824838821376161,
      "grad_norm": 0.481183797121048,
      "learning_rate": 3.426884734700572e-05,
      "loss": 0.8246,
      "step": 13380
    },
    {
      "epoch": 0.8831434366085709,
      "grad_norm": 0.5114526748657227,
      "learning_rate": 3.42421135820405e-05,
      "loss": 0.7745,
      "step": 13390
    },
    {
      "epoch": 0.8838029910795258,
      "grad_norm": 0.5307304263114929,
      "learning_rate": 3.42153675700081e-05,
      "loss": 0.8493,
      "step": 13400
    },
    {
      "epoch": 0.8844625455504806,
      "grad_norm": 0.564628005027771,
      "learning_rate": 3.418860934635065e-05,
      "loss": 0.8158,
      "step": 13410
    },
    {
      "epoch": 0.8851221000214355,
      "grad_norm": 0.4864290952682495,
      "learning_rate": 3.4161838946526473e-05,
      "loss": 0.8065,
      "step": 13420
    },
    {
      "epoch": 0.8857816544923904,
      "grad_norm": 0.5225094556808472,
      "learning_rate": 3.4135056406010014e-05,
      "loss": 0.8429,
      "step": 13430
    },
    {
      "epoch": 0.8864412089633452,
      "grad_norm": 0.5745679140090942,
      "learning_rate": 3.4108261760291806e-05,
      "loss": 0.8411,
      "step": 13440
    },
    {
      "epoch": 0.8871007634343001,
      "grad_norm": 0.5746591687202454,
      "learning_rate": 3.408145504487843e-05,
      "loss": 0.7902,
      "step": 13450
    },
    {
      "epoch": 0.887760317905255,
      "grad_norm": 0.497823566198349,
      "learning_rate": 3.405463629529244e-05,
      "loss": 0.7747,
      "step": 13460
    },
    {
      "epoch": 0.8884198723762099,
      "grad_norm": 0.4694047272205353,
      "learning_rate": 3.402780554707238e-05,
      "loss": 0.8492,
      "step": 13470
    },
    {
      "epoch": 0.8890794268471648,
      "grad_norm": 0.585241436958313,
      "learning_rate": 3.400096283577266e-05,
      "loss": 0.8628,
      "step": 13480
    },
    {
      "epoch": 0.8897389813181196,
      "grad_norm": 0.5171390771865845,
      "learning_rate": 3.397410819696355e-05,
      "loss": 0.8546,
      "step": 13490
    },
    {
      "epoch": 0.8903985357890745,
      "grad_norm": 0.44888442754745483,
      "learning_rate": 3.394724166623113e-05,
      "loss": 0.8302,
      "step": 13500
    },
    {
      "epoch": 0.8910580902600294,
      "grad_norm": 0.47432762384414673,
      "learning_rate": 3.392036327917723e-05,
      "loss": 0.7981,
      "step": 13510
    },
    {
      "epoch": 0.8917176447309842,
      "grad_norm": 0.5711329579353333,
      "learning_rate": 3.38934730714194e-05,
      "loss": 0.7863,
      "step": 13520
    },
    {
      "epoch": 0.8923771992019391,
      "grad_norm": 0.5666019320487976,
      "learning_rate": 3.3866571078590854e-05,
      "loss": 0.831,
      "step": 13530
    },
    {
      "epoch": 0.8930367536728939,
      "grad_norm": 0.4173568785190582,
      "learning_rate": 3.3839657336340415e-05,
      "loss": 0.8111,
      "step": 13540
    },
    {
      "epoch": 0.8936963081438488,
      "grad_norm": 0.48306772112846375,
      "learning_rate": 3.3812731880332476e-05,
      "loss": 0.8548,
      "step": 13550
    },
    {
      "epoch": 0.8943558626148037,
      "grad_norm": 0.5220154523849487,
      "learning_rate": 3.378579474624697e-05,
      "loss": 0.8328,
      "step": 13560
    },
    {
      "epoch": 0.8950154170857586,
      "grad_norm": 0.5229178667068481,
      "learning_rate": 3.375884596977927e-05,
      "loss": 0.8584,
      "step": 13570
    },
    {
      "epoch": 0.8956749715567135,
      "grad_norm": 0.5190449953079224,
      "learning_rate": 3.373188558664022e-05,
      "loss": 0.8653,
      "step": 13580
    },
    {
      "epoch": 0.8963345260276683,
      "grad_norm": 0.4401300549507141,
      "learning_rate": 3.370491363255602e-05,
      "loss": 0.8604,
      "step": 13590
    },
    {
      "epoch": 0.8969940804986232,
      "grad_norm": 0.4707847833633423,
      "learning_rate": 3.367793014326819e-05,
      "loss": 0.8574,
      "step": 13600
    },
    {
      "epoch": 0.8976536349695781,
      "grad_norm": 0.4264417588710785,
      "learning_rate": 3.365093515453357e-05,
      "loss": 0.8218,
      "step": 13610
    },
    {
      "epoch": 0.898313189440533,
      "grad_norm": 0.5680892467498779,
      "learning_rate": 3.362392870212421e-05,
      "loss": 0.7665,
      "step": 13620
    },
    {
      "epoch": 0.8989727439114877,
      "grad_norm": 0.591608464717865,
      "learning_rate": 3.359691082182735e-05,
      "loss": 0.8213,
      "step": 13630
    },
    {
      "epoch": 0.8996322983824426,
      "grad_norm": 0.538254976272583,
      "learning_rate": 3.356988154944541e-05,
      "loss": 0.8613,
      "step": 13640
    },
    {
      "epoch": 0.9002918528533975,
      "grad_norm": 0.5444620847702026,
      "learning_rate": 3.354284092079587e-05,
      "loss": 0.8091,
      "step": 13650
    },
    {
      "epoch": 0.9009514073243524,
      "grad_norm": 0.5183736681938171,
      "learning_rate": 3.351578897171125e-05,
      "loss": 0.8324,
      "step": 13660
    },
    {
      "epoch": 0.9016109617953073,
      "grad_norm": 0.544813871383667,
      "learning_rate": 3.348872573803913e-05,
      "loss": 0.8321,
      "step": 13670
    },
    {
      "epoch": 0.9022705162662621,
      "grad_norm": 0.5314427614212036,
      "learning_rate": 3.3461651255641954e-05,
      "loss": 0.8347,
      "step": 13680
    },
    {
      "epoch": 0.902930070737217,
      "grad_norm": 0.6030846834182739,
      "learning_rate": 3.3434565560397165e-05,
      "loss": 0.8629,
      "step": 13690
    },
    {
      "epoch": 0.9035896252081719,
      "grad_norm": 0.47967347502708435,
      "learning_rate": 3.3407468688197e-05,
      "loss": 0.7961,
      "step": 13700
    },
    {
      "epoch": 0.9042491796791268,
      "grad_norm": 0.4342898726463318,
      "learning_rate": 3.338036067494853e-05,
      "loss": 0.828,
      "step": 13710
    },
    {
      "epoch": 0.9049087341500817,
      "grad_norm": 0.5623383522033691,
      "learning_rate": 3.3353241556573596e-05,
      "loss": 0.8206,
      "step": 13720
    },
    {
      "epoch": 0.9055682886210364,
      "grad_norm": 0.5118647813796997,
      "learning_rate": 3.332611136900874e-05,
      "loss": 0.8407,
      "step": 13730
    },
    {
      "epoch": 0.9062278430919913,
      "grad_norm": 0.4637552797794342,
      "learning_rate": 3.329897014820518e-05,
      "loss": 0.8396,
      "step": 13740
    },
    {
      "epoch": 0.9068873975629462,
      "grad_norm": 0.5614678859710693,
      "learning_rate": 3.327181793012875e-05,
      "loss": 0.851,
      "step": 13750
    },
    {
      "epoch": 0.9075469520339011,
      "grad_norm": 0.4977450370788574,
      "learning_rate": 3.324465475075987e-05,
      "loss": 0.8334,
      "step": 13760
    },
    {
      "epoch": 0.908206506504856,
      "grad_norm": 0.5911426544189453,
      "learning_rate": 3.3217480646093466e-05,
      "loss": 0.7991,
      "step": 13770
    },
    {
      "epoch": 0.9088660609758108,
      "grad_norm": 0.5079803466796875,
      "learning_rate": 3.3190295652138946e-05,
      "loss": 0.8507,
      "step": 13780
    },
    {
      "epoch": 0.9095256154467657,
      "grad_norm": 0.5313887596130371,
      "learning_rate": 3.3163099804920175e-05,
      "loss": 0.8012,
      "step": 13790
    },
    {
      "epoch": 0.9101851699177206,
      "grad_norm": 0.5302901864051819,
      "learning_rate": 3.313589314047536e-05,
      "loss": 0.8105,
      "step": 13800
    },
    {
      "epoch": 0.9108447243886755,
      "grad_norm": 0.47778141498565674,
      "learning_rate": 3.310867569485706e-05,
      "loss": 0.8443,
      "step": 13810
    },
    {
      "epoch": 0.9115042788596304,
      "grad_norm": 0.46718910336494446,
      "learning_rate": 3.3081447504132136e-05,
      "loss": 0.8608,
      "step": 13820
    },
    {
      "epoch": 0.9121638333305851,
      "grad_norm": 0.5107888579368591,
      "learning_rate": 3.305420860438165e-05,
      "loss": 0.8679,
      "step": 13830
    },
    {
      "epoch": 0.91282338780154,
      "grad_norm": 0.5221997499465942,
      "learning_rate": 3.30269590317009e-05,
      "loss": 0.8332,
      "step": 13840
    },
    {
      "epoch": 0.9134829422724949,
      "grad_norm": 0.49340206384658813,
      "learning_rate": 3.299969882219928e-05,
      "loss": 0.8448,
      "step": 13850
    },
    {
      "epoch": 0.9141424967434498,
      "grad_norm": 0.49549180269241333,
      "learning_rate": 3.297242801200033e-05,
      "loss": 0.793,
      "step": 13860
    },
    {
      "epoch": 0.9148020512144047,
      "grad_norm": 0.5069233775138855,
      "learning_rate": 3.29451466372416e-05,
      "loss": 0.8548,
      "step": 13870
    },
    {
      "epoch": 0.9154616056853595,
      "grad_norm": 0.5417354106903076,
      "learning_rate": 3.291785473407464e-05,
      "loss": 0.814,
      "step": 13880
    },
    {
      "epoch": 0.9161211601563144,
      "grad_norm": 0.569395124912262,
      "learning_rate": 3.289055233866497e-05,
      "loss": 0.8133,
      "step": 13890
    },
    {
      "epoch": 0.9167807146272693,
      "grad_norm": 0.5674502849578857,
      "learning_rate": 3.2863239487192014e-05,
      "loss": 0.8036,
      "step": 13900
    },
    {
      "epoch": 0.9174402690982242,
      "grad_norm": 0.4945697784423828,
      "learning_rate": 3.283591621584903e-05,
      "loss": 0.8145,
      "step": 13910
    },
    {
      "epoch": 0.9180998235691791,
      "grad_norm": 0.5142458081245422,
      "learning_rate": 3.280858256084311e-05,
      "loss": 0.8277,
      "step": 13920
    },
    {
      "epoch": 0.9187593780401339,
      "grad_norm": 0.49605944752693176,
      "learning_rate": 3.278123855839508e-05,
      "loss": 0.855,
      "step": 13930
    },
    {
      "epoch": 0.9194189325110887,
      "grad_norm": 0.5406461954116821,
      "learning_rate": 3.275388424473951e-05,
      "loss": 0.818,
      "step": 13940
    },
    {
      "epoch": 0.9200784869820436,
      "grad_norm": 0.5674294829368591,
      "learning_rate": 3.27265196561246e-05,
      "loss": 0.7944,
      "step": 13950
    },
    {
      "epoch": 0.9207380414529985,
      "grad_norm": 0.5565577745437622,
      "learning_rate": 3.269914482881218e-05,
      "loss": 0.8585,
      "step": 13960
    },
    {
      "epoch": 0.9213975959239533,
      "grad_norm": 0.5268924236297607,
      "learning_rate": 3.267175979907765e-05,
      "loss": 0.8569,
      "step": 13970
    },
    {
      "epoch": 0.9220571503949082,
      "grad_norm": 0.494615375995636,
      "learning_rate": 3.264436460320993e-05,
      "loss": 0.8264,
      "step": 13980
    },
    {
      "epoch": 0.9227167048658631,
      "grad_norm": 0.45897209644317627,
      "learning_rate": 3.261695927751141e-05,
      "loss": 0.8705,
      "step": 13990
    },
    {
      "epoch": 0.923376259336818,
      "grad_norm": 0.5044166445732117,
      "learning_rate": 3.25895438582979e-05,
      "loss": 0.8121,
      "step": 14000
    },
    {
      "epoch": 0.9240358138077729,
      "grad_norm": 0.6285238265991211,
      "learning_rate": 3.256211838189859e-05,
      "loss": 0.8331,
      "step": 14010
    },
    {
      "epoch": 0.9246953682787277,
      "grad_norm": 0.519390881061554,
      "learning_rate": 3.2534682884655985e-05,
      "loss": 0.8593,
      "step": 14020
    },
    {
      "epoch": 0.9253549227496826,
      "grad_norm": 0.5182341933250427,
      "learning_rate": 3.2507237402925896e-05,
      "loss": 0.8422,
      "step": 14030
    },
    {
      "epoch": 0.9260144772206375,
      "grad_norm": 0.5825724005699158,
      "learning_rate": 3.247978197307734e-05,
      "loss": 0.8037,
      "step": 14040
    },
    {
      "epoch": 0.9266740316915923,
      "grad_norm": 0.5468394756317139,
      "learning_rate": 3.2452316631492523e-05,
      "loss": 0.8499,
      "step": 14050
    },
    {
      "epoch": 0.9273335861625472,
      "grad_norm": 0.4748375117778778,
      "learning_rate": 3.2424841414566785e-05,
      "loss": 0.8832,
      "step": 14060
    },
    {
      "epoch": 0.927993140633502,
      "grad_norm": 0.5006450414657593,
      "learning_rate": 3.2397356358708555e-05,
      "loss": 0.8283,
      "step": 14070
    },
    {
      "epoch": 0.9286526951044569,
      "grad_norm": 0.5645299553871155,
      "learning_rate": 3.23698615003393e-05,
      "loss": 0.8264,
      "step": 14080
    },
    {
      "epoch": 0.9293122495754118,
      "grad_norm": 0.5048816204071045,
      "learning_rate": 3.2342356875893484e-05,
      "loss": 0.8275,
      "step": 14090
    },
    {
      "epoch": 0.9299718040463667,
      "grad_norm": 0.4915793538093567,
      "learning_rate": 3.2314842521818495e-05,
      "loss": 0.8119,
      "step": 14100
    },
    {
      "epoch": 0.9306313585173216,
      "grad_norm": 0.5013352036476135,
      "learning_rate": 3.2287318474574625e-05,
      "loss": 0.8539,
      "step": 14110
    },
    {
      "epoch": 0.9312909129882764,
      "grad_norm": 0.5083145499229431,
      "learning_rate": 3.225978477063501e-05,
      "loss": 0.846,
      "step": 14120
    },
    {
      "epoch": 0.9319504674592313,
      "grad_norm": 0.5598704814910889,
      "learning_rate": 3.223224144648558e-05,
      "loss": 0.8469,
      "step": 14130
    },
    {
      "epoch": 0.9326100219301862,
      "grad_norm": 0.5258391499519348,
      "learning_rate": 3.2204688538625006e-05,
      "loss": 0.8316,
      "step": 14140
    },
    {
      "epoch": 0.933269576401141,
      "grad_norm": 0.6012523174285889,
      "learning_rate": 3.217712608356468e-05,
      "loss": 0.8153,
      "step": 14150
    },
    {
      "epoch": 0.933929130872096,
      "grad_norm": 0.5268067717552185,
      "learning_rate": 3.2149554117828634e-05,
      "loss": 0.8159,
      "step": 14160
    },
    {
      "epoch": 0.9345886853430507,
      "grad_norm": 0.4189505875110626,
      "learning_rate": 3.212197267795349e-05,
      "loss": 0.842,
      "step": 14170
    },
    {
      "epoch": 0.9352482398140056,
      "grad_norm": 0.502161979675293,
      "learning_rate": 3.2094381800488436e-05,
      "loss": 0.8211,
      "step": 14180
    },
    {
      "epoch": 0.9359077942849605,
      "grad_norm": 0.4730813205242157,
      "learning_rate": 3.2066781521995184e-05,
      "loss": 0.8689,
      "step": 14190
    },
    {
      "epoch": 0.9365673487559154,
      "grad_norm": 0.49195387959480286,
      "learning_rate": 3.203917187904787e-05,
      "loss": 0.8266,
      "step": 14200
    },
    {
      "epoch": 0.9372269032268703,
      "grad_norm": 0.5385797619819641,
      "learning_rate": 3.201155290823306e-05,
      "loss": 0.8066,
      "step": 14210
    },
    {
      "epoch": 0.9378864576978251,
      "grad_norm": 0.537335991859436,
      "learning_rate": 3.198392464614967e-05,
      "loss": 0.8148,
      "step": 14220
    },
    {
      "epoch": 0.93854601216878,
      "grad_norm": 0.524181604385376,
      "learning_rate": 3.1956287129408944e-05,
      "loss": 0.8533,
      "step": 14230
    },
    {
      "epoch": 0.9392055666397349,
      "grad_norm": 0.4703523814678192,
      "learning_rate": 3.1928640394634365e-05,
      "loss": 0.8427,
      "step": 14240
    },
    {
      "epoch": 0.9398651211106898,
      "grad_norm": 0.459078848361969,
      "learning_rate": 3.1900984478461666e-05,
      "loss": 0.8372,
      "step": 14250
    },
    {
      "epoch": 0.9405246755816447,
      "grad_norm": 0.46345698833465576,
      "learning_rate": 3.18733194175387e-05,
      "loss": 0.8589,
      "step": 14260
    },
    {
      "epoch": 0.9411842300525994,
      "grad_norm": 0.48149171471595764,
      "learning_rate": 3.184564524852549e-05,
      "loss": 0.8208,
      "step": 14270
    },
    {
      "epoch": 0.9418437845235543,
      "grad_norm": 0.5642602443695068,
      "learning_rate": 3.181796200809409e-05,
      "loss": 0.8048,
      "step": 14280
    },
    {
      "epoch": 0.9425033389945092,
      "grad_norm": 0.5288432240486145,
      "learning_rate": 3.179026973292859e-05,
      "loss": 0.851,
      "step": 14290
    },
    {
      "epoch": 0.9431628934654641,
      "grad_norm": 0.5397107005119324,
      "learning_rate": 3.176256845972505e-05,
      "loss": 0.8217,
      "step": 14300
    },
    {
      "epoch": 0.9438224479364189,
      "grad_norm": 0.5059977769851685,
      "learning_rate": 3.173485822519145e-05,
      "loss": 0.8279,
      "step": 14310
    },
    {
      "epoch": 0.9444820024073738,
      "grad_norm": 0.5274835824966431,
      "learning_rate": 3.170713906604765e-05,
      "loss": 0.8246,
      "step": 14320
    },
    {
      "epoch": 0.9451415568783287,
      "grad_norm": 0.5186430811882019,
      "learning_rate": 3.1679411019025346e-05,
      "loss": 0.8268,
      "step": 14330
    },
    {
      "epoch": 0.9458011113492836,
      "grad_norm": 0.45758917927742004,
      "learning_rate": 3.165167412086799e-05,
      "loss": 0.8733,
      "step": 14340
    },
    {
      "epoch": 0.9464606658202385,
      "grad_norm": 0.5606436729431152,
      "learning_rate": 3.162392840833077e-05,
      "loss": 0.7949,
      "step": 14350
    },
    {
      "epoch": 0.9471202202911932,
      "grad_norm": 0.4964209198951721,
      "learning_rate": 3.159617391818056e-05,
      "loss": 0.8411,
      "step": 14360
    },
    {
      "epoch": 0.9477797747621481,
      "grad_norm": 0.5006893277168274,
      "learning_rate": 3.156841068719587e-05,
      "loss": 0.8317,
      "step": 14370
    },
    {
      "epoch": 0.948439329233103,
      "grad_norm": 0.47448766231536865,
      "learning_rate": 3.154063875216678e-05,
      "loss": 0.8454,
      "step": 14380
    },
    {
      "epoch": 0.9490988837040579,
      "grad_norm": 0.5972414612770081,
      "learning_rate": 3.151285814989492e-05,
      "loss": 0.8395,
      "step": 14390
    },
    {
      "epoch": 0.9497584381750128,
      "grad_norm": 0.4944043755531311,
      "learning_rate": 3.1485068917193385e-05,
      "loss": 0.8119,
      "step": 14400
    },
    {
      "epoch": 0.9504179926459676,
      "grad_norm": 0.5087505578994751,
      "learning_rate": 3.145727109088672e-05,
      "loss": 0.8401,
      "step": 14410
    },
    {
      "epoch": 0.9510775471169225,
      "grad_norm": 0.5083771347999573,
      "learning_rate": 3.142946470781085e-05,
      "loss": 0.8638,
      "step": 14420
    },
    {
      "epoch": 0.9517371015878774,
      "grad_norm": 0.5725495219230652,
      "learning_rate": 3.140164980481305e-05,
      "loss": 0.8505,
      "step": 14430
    },
    {
      "epoch": 0.9523966560588323,
      "grad_norm": 0.5729572176933289,
      "learning_rate": 3.137382641875187e-05,
      "loss": 0.8358,
      "step": 14440
    },
    {
      "epoch": 0.9530562105297872,
      "grad_norm": 0.6390442252159119,
      "learning_rate": 3.134599458649713e-05,
      "loss": 0.8283,
      "step": 14450
    },
    {
      "epoch": 0.953715765000742,
      "grad_norm": 0.5202282071113586,
      "learning_rate": 3.131815434492981e-05,
      "loss": 0.8617,
      "step": 14460
    },
    {
      "epoch": 0.9543753194716968,
      "grad_norm": 0.5068888068199158,
      "learning_rate": 3.1290305730942035e-05,
      "loss": 0.8538,
      "step": 14470
    },
    {
      "epoch": 0.9550348739426517,
      "grad_norm": 0.46176913380622864,
      "learning_rate": 3.126244878143705e-05,
      "loss": 0.8406,
      "step": 14480
    },
    {
      "epoch": 0.9556944284136066,
      "grad_norm": 0.5682269334793091,
      "learning_rate": 3.123458353332912e-05,
      "loss": 0.8245,
      "step": 14490
    },
    {
      "epoch": 0.9563539828845615,
      "grad_norm": 0.47708868980407715,
      "learning_rate": 3.120671002354353e-05,
      "loss": 0.8426,
      "step": 14500
    },
    {
      "epoch": 0.9570135373555163,
      "grad_norm": 0.5253884196281433,
      "learning_rate": 3.1178828289016495e-05,
      "loss": 0.8186,
      "step": 14510
    },
    {
      "epoch": 0.9576730918264712,
      "grad_norm": 0.5317881107330322,
      "learning_rate": 3.115093836669513e-05,
      "loss": 0.8278,
      "step": 14520
    },
    {
      "epoch": 0.9583326462974261,
      "grad_norm": 0.5529583692550659,
      "learning_rate": 3.1123040293537414e-05,
      "loss": 0.8201,
      "step": 14530
    },
    {
      "epoch": 0.958992200768381,
      "grad_norm": 0.580334484577179,
      "learning_rate": 3.1095134106512116e-05,
      "loss": 0.8389,
      "step": 14540
    },
    {
      "epoch": 0.9596517552393359,
      "grad_norm": 0.507512092590332,
      "learning_rate": 3.106721984259874e-05,
      "loss": 0.8326,
      "step": 14550
    },
    {
      "epoch": 0.9603113097102907,
      "grad_norm": 0.46313729882240295,
      "learning_rate": 3.103929753878754e-05,
      "loss": 0.8121,
      "step": 14560
    },
    {
      "epoch": 0.9609708641812456,
      "grad_norm": 0.5425809025764465,
      "learning_rate": 3.101136723207938e-05,
      "loss": 0.8336,
      "step": 14570
    },
    {
      "epoch": 0.9616304186522004,
      "grad_norm": 0.5370546579360962,
      "learning_rate": 3.098342895948574e-05,
      "loss": 0.8395,
      "step": 14580
    },
    {
      "epoch": 0.9622899731231553,
      "grad_norm": 0.5421480536460876,
      "learning_rate": 3.095548275802867e-05,
      "loss": 0.8375,
      "step": 14590
    },
    {
      "epoch": 0.9629495275941102,
      "grad_norm": 0.5985087752342224,
      "learning_rate": 3.092752866474071e-05,
      "loss": 0.8259,
      "step": 14600
    },
    {
      "epoch": 0.963609082065065,
      "grad_norm": 0.5083304047584534,
      "learning_rate": 3.089956671666488e-05,
      "loss": 0.8183,
      "step": 14610
    },
    {
      "epoch": 0.9642686365360199,
      "grad_norm": 0.5227060317993164,
      "learning_rate": 3.0871596950854575e-05,
      "loss": 0.8691,
      "step": 14620
    },
    {
      "epoch": 0.9649281910069748,
      "grad_norm": 0.47051557898521423,
      "learning_rate": 3.084361940437357e-05,
      "loss": 0.7931,
      "step": 14630
    },
    {
      "epoch": 0.9655877454779297,
      "grad_norm": 0.5483131408691406,
      "learning_rate": 3.081563411429596e-05,
      "loss": 0.8439,
      "step": 14640
    },
    {
      "epoch": 0.9662472999488845,
      "grad_norm": 0.495633989572525,
      "learning_rate": 3.078764111770609e-05,
      "loss": 0.8545,
      "step": 14650
    },
    {
      "epoch": 0.9669068544198394,
      "grad_norm": 0.5021789073944092,
      "learning_rate": 3.075964045169849e-05,
      "loss": 0.8494,
      "step": 14660
    },
    {
      "epoch": 0.9675664088907943,
      "grad_norm": 0.5027854442596436,
      "learning_rate": 3.073163215337792e-05,
      "loss": 0.8507,
      "step": 14670
    },
    {
      "epoch": 0.9682259633617492,
      "grad_norm": 0.5345577597618103,
      "learning_rate": 3.0703616259859184e-05,
      "loss": 0.8456,
      "step": 14680
    },
    {
      "epoch": 0.968885517832704,
      "grad_norm": 0.5073565244674683,
      "learning_rate": 3.0675592808267197e-05,
      "loss": 0.8598,
      "step": 14690
    },
    {
      "epoch": 0.9695450723036588,
      "grad_norm": 0.5810648202896118,
      "learning_rate": 3.064756183573687e-05,
      "loss": 0.785,
      "step": 14700
    },
    {
      "epoch": 0.9702046267746137,
      "grad_norm": 0.5641012787818909,
      "learning_rate": 3.061952337941307e-05,
      "loss": 0.801,
      "step": 14710
    },
    {
      "epoch": 0.9708641812455686,
      "grad_norm": 0.524024486541748,
      "learning_rate": 3.059147747645062e-05,
      "loss": 0.8343,
      "step": 14720
    },
    {
      "epoch": 0.9715237357165235,
      "grad_norm": 0.5879754424095154,
      "learning_rate": 3.0563424164014166e-05,
      "loss": 0.7986,
      "step": 14730
    },
    {
      "epoch": 0.9721832901874784,
      "grad_norm": 0.49657177925109863,
      "learning_rate": 3.053536347927819e-05,
      "loss": 0.8507,
      "step": 14740
    },
    {
      "epoch": 0.9728428446584332,
      "grad_norm": 0.5545039772987366,
      "learning_rate": 3.050729545942695e-05,
      "loss": 0.8096,
      "step": 14750
    },
    {
      "epoch": 0.9735023991293881,
      "grad_norm": 0.482943058013916,
      "learning_rate": 3.0479220141654425e-05,
      "loss": 0.832,
      "step": 14760
    },
    {
      "epoch": 0.974161953600343,
      "grad_norm": 0.5274185538291931,
      "learning_rate": 3.0451137563164263e-05,
      "loss": 0.8248,
      "step": 14770
    },
    {
      "epoch": 0.9748215080712979,
      "grad_norm": 0.50125652551651,
      "learning_rate": 3.042304776116971e-05,
      "loss": 0.7765,
      "step": 14780
    },
    {
      "epoch": 0.9754810625422528,
      "grad_norm": 0.507879376411438,
      "learning_rate": 3.039495077289362e-05,
      "loss": 0.8509,
      "step": 14790
    },
    {
      "epoch": 0.9761406170132075,
      "grad_norm": 0.5121268630027771,
      "learning_rate": 3.0366846635568347e-05,
      "loss": 0.7931,
      "step": 14800
    },
    {
      "epoch": 0.9768001714841624,
      "grad_norm": 0.5252723693847656,
      "learning_rate": 3.0338735386435723e-05,
      "loss": 0.8121,
      "step": 14810
    },
    {
      "epoch": 0.9774597259551173,
      "grad_norm": 0.5790934562683105,
      "learning_rate": 3.031061706274701e-05,
      "loss": 0.8163,
      "step": 14820
    },
    {
      "epoch": 0.9781192804260722,
      "grad_norm": 0.5557512044906616,
      "learning_rate": 3.0282491701762845e-05,
      "loss": 0.8722,
      "step": 14830
    },
    {
      "epoch": 0.9787788348970271,
      "grad_norm": 0.5133851170539856,
      "learning_rate": 3.025435934075317e-05,
      "loss": 0.7813,
      "step": 14840
    },
    {
      "epoch": 0.9794383893679819,
      "grad_norm": 0.5220972299575806,
      "learning_rate": 3.0226220016997237e-05,
      "loss": 0.8075,
      "step": 14850
    },
    {
      "epoch": 0.9800979438389368,
      "grad_norm": 0.5013805627822876,
      "learning_rate": 3.0198073767783495e-05,
      "loss": 0.8391,
      "step": 14860
    },
    {
      "epoch": 0.9807574983098917,
      "grad_norm": 0.5171005129814148,
      "learning_rate": 3.0169920630409586e-05,
      "loss": 0.8054,
      "step": 14870
    },
    {
      "epoch": 0.9814170527808466,
      "grad_norm": 0.5466331243515015,
      "learning_rate": 3.0141760642182275e-05,
      "loss": 0.863,
      "step": 14880
    },
    {
      "epoch": 0.9820766072518015,
      "grad_norm": 0.48877766728401184,
      "learning_rate": 3.011359384041741e-05,
      "loss": 0.8445,
      "step": 14890
    },
    {
      "epoch": 0.9827361617227562,
      "grad_norm": 0.5366662740707397,
      "learning_rate": 3.0085420262439863e-05,
      "loss": 0.8236,
      "step": 14900
    },
    {
      "epoch": 0.9833957161937111,
      "grad_norm": 0.5145524740219116,
      "learning_rate": 3.0057239945583476e-05,
      "loss": 0.85,
      "step": 14910
    },
    {
      "epoch": 0.984055270664666,
      "grad_norm": 0.49465933442115784,
      "learning_rate": 3.0029052927191042e-05,
      "loss": 0.7925,
      "step": 14920
    },
    {
      "epoch": 0.9847148251356209,
      "grad_norm": 0.5308068990707397,
      "learning_rate": 3.0000859244614222e-05,
      "loss": 0.843,
      "step": 14930
    },
    {
      "epoch": 0.9853743796065758,
      "grad_norm": 0.4998134970664978,
      "learning_rate": 2.9972658935213506e-05,
      "loss": 0.8538,
      "step": 14940
    },
    {
      "epoch": 0.9860339340775306,
      "grad_norm": 0.507693350315094,
      "learning_rate": 2.9944452036358167e-05,
      "loss": 0.8029,
      "step": 14950
    },
    {
      "epoch": 0.9866934885484855,
      "grad_norm": 0.6143972873687744,
      "learning_rate": 2.9916238585426216e-05,
      "loss": 0.845,
      "step": 14960
    },
    {
      "epoch": 0.9873530430194404,
      "grad_norm": 0.5952014923095703,
      "learning_rate": 2.9888018619804336e-05,
      "loss": 0.8529,
      "step": 14970
    },
    {
      "epoch": 0.9880125974903953,
      "grad_norm": 0.528600811958313,
      "learning_rate": 2.985979217688786e-05,
      "loss": 0.7844,
      "step": 14980
    },
    {
      "epoch": 0.98867215196135,
      "grad_norm": 0.4936237633228302,
      "learning_rate": 2.983155929408069e-05,
      "loss": 0.7997,
      "step": 14990
    },
    {
      "epoch": 0.989331706432305,
      "grad_norm": 0.5711715221405029,
      "learning_rate": 2.9803320008795266e-05,
      "loss": 0.8221,
      "step": 15000
    },
    {
      "epoch": 0.9899912609032598,
      "grad_norm": 0.48582160472869873,
      "learning_rate": 2.9775074358452503e-05,
      "loss": 0.8028,
      "step": 15010
    },
    {
      "epoch": 0.9906508153742147,
      "grad_norm": 0.48469483852386475,
      "learning_rate": 2.9746822380481763e-05,
      "loss": 0.8425,
      "step": 15020
    },
    {
      "epoch": 0.9913103698451696,
      "grad_norm": 0.5462239384651184,
      "learning_rate": 2.9718564112320796e-05,
      "loss": 0.8491,
      "step": 15030
    },
    {
      "epoch": 0.9919699243161244,
      "grad_norm": 0.5127155184745789,
      "learning_rate": 2.969029959141567e-05,
      "loss": 0.8045,
      "step": 15040
    },
    {
      "epoch": 0.9926294787870793,
      "grad_norm": 0.5171392560005188,
      "learning_rate": 2.966202885522076e-05,
      "loss": 0.8607,
      "step": 15050
    },
    {
      "epoch": 0.9932890332580342,
      "grad_norm": 0.5148513317108154,
      "learning_rate": 2.9633751941198657e-05,
      "loss": 0.803,
      "step": 15060
    },
    {
      "epoch": 0.9939485877289891,
      "grad_norm": 0.5123173594474792,
      "learning_rate": 2.9605468886820147e-05,
      "loss": 0.8307,
      "step": 15070
    },
    {
      "epoch": 0.994608142199944,
      "grad_norm": 0.5569455027580261,
      "learning_rate": 2.9577179729564163e-05,
      "loss": 0.8174,
      "step": 15080
    },
    {
      "epoch": 0.9952676966708988,
      "grad_norm": 0.534589409828186,
      "learning_rate": 2.9548884506917706e-05,
      "loss": 0.8208,
      "step": 15090
    },
    {
      "epoch": 0.9959272511418537,
      "grad_norm": 0.6001644730567932,
      "learning_rate": 2.9520583256375823e-05,
      "loss": 0.8314,
      "step": 15100
    },
    {
      "epoch": 0.9965868056128085,
      "grad_norm": 0.5668461918830872,
      "learning_rate": 2.9492276015441556e-05,
      "loss": 0.8692,
      "step": 15110
    },
    {
      "epoch": 0.9972463600837634,
      "grad_norm": 0.5646265149116516,
      "learning_rate": 2.9463962821625883e-05,
      "loss": 0.8541,
      "step": 15120
    },
    {
      "epoch": 0.9979059145547183,
      "grad_norm": 0.450910359621048,
      "learning_rate": 2.943564371244766e-05,
      "loss": 0.7934,
      "step": 15130
    },
    {
      "epoch": 0.9985654690256731,
      "grad_norm": 0.5016583204269409,
      "learning_rate": 2.9407318725433587e-05,
      "loss": 0.8428,
      "step": 15140
    },
    {
      "epoch": 0.999225023496628,
      "grad_norm": 0.6422037482261658,
      "learning_rate": 2.9378987898118154e-05,
      "loss": 0.792,
      "step": 15150
    },
    {
      "epoch": 0.9998845779675829,
      "grad_norm": 0.4914742112159729,
      "learning_rate": 2.935065126804359e-05,
      "loss": 0.849,
      "step": 15160
    },
    {
      "epoch": 1.0005276435767638,
      "grad_norm": 0.48679330945014954,
      "learning_rate": 2.932230887275982e-05,
      "loss": 0.8407,
      "step": 15170
    },
    {
      "epoch": 1.0011871980477187,
      "grad_norm": 0.582068145275116,
      "learning_rate": 2.929396074982439e-05,
      "loss": 0.8316,
      "step": 15180
    },
    {
      "epoch": 1.0018467525186736,
      "grad_norm": 0.540686845779419,
      "learning_rate": 2.9265606936802457e-05,
      "loss": 0.8298,
      "step": 15190
    },
    {
      "epoch": 1.0025063069896285,
      "grad_norm": 0.5291965007781982,
      "learning_rate": 2.923724747126671e-05,
      "loss": 0.8475,
      "step": 15200
    },
    {
      "epoch": 1.0031658614605834,
      "grad_norm": 0.5179721713066101,
      "learning_rate": 2.9208882390797327e-05,
      "loss": 0.8287,
      "step": 15210
    },
    {
      "epoch": 1.0038254159315383,
      "grad_norm": 0.5664460062980652,
      "learning_rate": 2.918051173298192e-05,
      "loss": 0.7661,
      "step": 15220
    },
    {
      "epoch": 1.0044849704024932,
      "grad_norm": 0.5750847458839417,
      "learning_rate": 2.9152135535415513e-05,
      "loss": 0.8322,
      "step": 15230
    },
    {
      "epoch": 1.005144524873448,
      "grad_norm": 0.5466136932373047,
      "learning_rate": 2.912375383570044e-05,
      "loss": 0.8145,
      "step": 15240
    },
    {
      "epoch": 1.0058040793444027,
      "grad_norm": 0.4758152961730957,
      "learning_rate": 2.9095366671446354e-05,
      "loss": 0.8167,
      "step": 15250
    },
    {
      "epoch": 1.0064636338153576,
      "grad_norm": 0.522501528263092,
      "learning_rate": 2.906697408027013e-05,
      "loss": 0.8623,
      "step": 15260
    },
    {
      "epoch": 1.0071231882863125,
      "grad_norm": 0.5495953559875488,
      "learning_rate": 2.9038576099795854e-05,
      "loss": 0.8087,
      "step": 15270
    },
    {
      "epoch": 1.0077827427572674,
      "grad_norm": 0.5437273979187012,
      "learning_rate": 2.901017276765474e-05,
      "loss": 0.8211,
      "step": 15280
    },
    {
      "epoch": 1.0084422972282223,
      "grad_norm": 0.5142714381217957,
      "learning_rate": 2.898176412148509e-05,
      "loss": 0.774,
      "step": 15290
    },
    {
      "epoch": 1.0091018516991772,
      "grad_norm": 0.6193010210990906,
      "learning_rate": 2.8953350198932257e-05,
      "loss": 0.8004,
      "step": 15300
    },
    {
      "epoch": 1.009761406170132,
      "grad_norm": 0.5164871215820312,
      "learning_rate": 2.8924931037648578e-05,
      "loss": 0.79,
      "step": 15310
    },
    {
      "epoch": 1.010420960641087,
      "grad_norm": 0.4344189763069153,
      "learning_rate": 2.8896506675293344e-05,
      "loss": 0.8296,
      "step": 15320
    },
    {
      "epoch": 1.0110805151120419,
      "grad_norm": 0.5434548854827881,
      "learning_rate": 2.8868077149532723e-05,
      "loss": 0.812,
      "step": 15330
    },
    {
      "epoch": 1.0117400695829968,
      "grad_norm": 0.5497419238090515,
      "learning_rate": 2.8839642498039736e-05,
      "loss": 0.8377,
      "step": 15340
    },
    {
      "epoch": 1.0123996240539515,
      "grad_norm": 0.6154009103775024,
      "learning_rate": 2.8811202758494204e-05,
      "loss": 0.7947,
      "step": 15350
    },
    {
      "epoch": 1.0130591785249063,
      "grad_norm": 0.5419930219650269,
      "learning_rate": 2.8782757968582664e-05,
      "loss": 0.8418,
      "step": 15360
    },
    {
      "epoch": 1.0137187329958612,
      "grad_norm": 0.469460666179657,
      "learning_rate": 2.875430816599837e-05,
      "loss": 0.8352,
      "step": 15370
    },
    {
      "epoch": 1.0143782874668161,
      "grad_norm": 0.5413352251052856,
      "learning_rate": 2.872585338844121e-05,
      "loss": 0.8183,
      "step": 15380
    },
    {
      "epoch": 1.015037841937771,
      "grad_norm": 0.4633040726184845,
      "learning_rate": 2.8697393673617657e-05,
      "loss": 0.7765,
      "step": 15390
    },
    {
      "epoch": 1.015697396408726,
      "grad_norm": 0.565757691860199,
      "learning_rate": 2.866892905924074e-05,
      "loss": 0.8224,
      "step": 15400
    },
    {
      "epoch": 1.0163569508796808,
      "grad_norm": 0.5182371139526367,
      "learning_rate": 2.8640459583029965e-05,
      "loss": 0.8232,
      "step": 15410
    },
    {
      "epoch": 1.0170165053506357,
      "grad_norm": 0.5305317640304565,
      "learning_rate": 2.8611985282711306e-05,
      "loss": 0.801,
      "step": 15420
    },
    {
      "epoch": 1.0176760598215906,
      "grad_norm": 0.5024542212486267,
      "learning_rate": 2.8583506196017107e-05,
      "loss": 0.8176,
      "step": 15430
    },
    {
      "epoch": 1.0183356142925455,
      "grad_norm": 0.5164063572883606,
      "learning_rate": 2.855502236068605e-05,
      "loss": 0.7778,
      "step": 15440
    },
    {
      "epoch": 1.0189951687635002,
      "grad_norm": 0.5681871771812439,
      "learning_rate": 2.8526533814463137e-05,
      "loss": 0.7978,
      "step": 15450
    },
    {
      "epoch": 1.019654723234455,
      "grad_norm": 0.4896976947784424,
      "learning_rate": 2.8498040595099585e-05,
      "loss": 0.8054,
      "step": 15460
    },
    {
      "epoch": 1.02031427770541,
      "grad_norm": 0.5637657046318054,
      "learning_rate": 2.8469542740352818e-05,
      "loss": 0.7965,
      "step": 15470
    },
    {
      "epoch": 1.0209738321763648,
      "grad_norm": 0.5734666585922241,
      "learning_rate": 2.8441040287986398e-05,
      "loss": 0.815,
      "step": 15480
    },
    {
      "epoch": 1.0216333866473197,
      "grad_norm": 0.5376142859458923,
      "learning_rate": 2.8412533275769987e-05,
      "loss": 0.7993,
      "step": 15490
    },
    {
      "epoch": 1.0222929411182746,
      "grad_norm": 0.47688788175582886,
      "learning_rate": 2.8384021741479285e-05,
      "loss": 0.8396,
      "step": 15500
    },
    {
      "epoch": 1.0229524955892295,
      "grad_norm": 0.5853601098060608,
      "learning_rate": 2.8355505722895963e-05,
      "loss": 0.8377,
      "step": 15510
    },
    {
      "epoch": 1.0236120500601844,
      "grad_norm": 0.46929389238357544,
      "learning_rate": 2.8326985257807675e-05,
      "loss": 0.832,
      "step": 15520
    },
    {
      "epoch": 1.0242716045311393,
      "grad_norm": 0.5821115374565125,
      "learning_rate": 2.829846038400794e-05,
      "loss": 0.8017,
      "step": 15530
    },
    {
      "epoch": 1.0249311590020942,
      "grad_norm": 0.5275147557258606,
      "learning_rate": 2.826993113929612e-05,
      "loss": 0.7748,
      "step": 15540
    },
    {
      "epoch": 1.0255907134730489,
      "grad_norm": 0.4697074294090271,
      "learning_rate": 2.8241397561477377e-05,
      "loss": 0.8377,
      "step": 15550
    },
    {
      "epoch": 1.0262502679440038,
      "grad_norm": 0.5319390892982483,
      "learning_rate": 2.8212859688362612e-05,
      "loss": 0.8135,
      "step": 15560
    },
    {
      "epoch": 1.0269098224149587,
      "grad_norm": 0.5258188843727112,
      "learning_rate": 2.818431755776843e-05,
      "loss": 0.8006,
      "step": 15570
    },
    {
      "epoch": 1.0275693768859135,
      "grad_norm": 0.5168179869651794,
      "learning_rate": 2.8155771207517044e-05,
      "loss": 0.8035,
      "step": 15580
    },
    {
      "epoch": 1.0282289313568684,
      "grad_norm": 0.5735469460487366,
      "learning_rate": 2.8127220675436305e-05,
      "loss": 0.7609,
      "step": 15590
    },
    {
      "epoch": 1.0288884858278233,
      "grad_norm": 0.5791534781455994,
      "learning_rate": 2.8098665999359558e-05,
      "loss": 0.8124,
      "step": 15600
    },
    {
      "epoch": 1.0295480402987782,
      "grad_norm": 0.57262122631073,
      "learning_rate": 2.8070107217125674e-05,
      "loss": 0.8088,
      "step": 15610
    },
    {
      "epoch": 1.0302075947697331,
      "grad_norm": 0.5607044100761414,
      "learning_rate": 2.804154436657895e-05,
      "loss": 0.7969,
      "step": 15620
    },
    {
      "epoch": 1.030867149240688,
      "grad_norm": 0.5467885136604309,
      "learning_rate": 2.8012977485569082e-05,
      "loss": 0.8088,
      "step": 15630
    },
    {
      "epoch": 1.0315267037116427,
      "grad_norm": 0.46850356459617615,
      "learning_rate": 2.79844066119511e-05,
      "loss": 0.8295,
      "step": 15640
    },
    {
      "epoch": 1.0321862581825976,
      "grad_norm": 0.5874578952789307,
      "learning_rate": 2.795583178358533e-05,
      "loss": 0.8339,
      "step": 15650
    },
    {
      "epoch": 1.0328458126535525,
      "grad_norm": 0.5937914848327637,
      "learning_rate": 2.7927253038337335e-05,
      "loss": 0.8057,
      "step": 15660
    },
    {
      "epoch": 1.0335053671245074,
      "grad_norm": 0.5366748571395874,
      "learning_rate": 2.7898670414077866e-05,
      "loss": 0.8177,
      "step": 15670
    },
    {
      "epoch": 1.0341649215954622,
      "grad_norm": 0.47821250557899475,
      "learning_rate": 2.787008394868282e-05,
      "loss": 0.8155,
      "step": 15680
    },
    {
      "epoch": 1.0348244760664171,
      "grad_norm": 0.5660244226455688,
      "learning_rate": 2.7841493680033177e-05,
      "loss": 0.7954,
      "step": 15690
    },
    {
      "epoch": 1.035484030537372,
      "grad_norm": 0.6502612233161926,
      "learning_rate": 2.7812899646014966e-05,
      "loss": 0.7861,
      "step": 15700
    },
    {
      "epoch": 1.036143585008327,
      "grad_norm": 0.5137171149253845,
      "learning_rate": 2.7784301884519204e-05,
      "loss": 0.8112,
      "step": 15710
    },
    {
      "epoch": 1.0368031394792818,
      "grad_norm": 0.5020664930343628,
      "learning_rate": 2.775570043344184e-05,
      "loss": 0.8135,
      "step": 15720
    },
    {
      "epoch": 1.0374626939502367,
      "grad_norm": 0.5886096954345703,
      "learning_rate": 2.7727095330683716e-05,
      "loss": 0.8039,
      "step": 15730
    },
    {
      "epoch": 1.0381222484211914,
      "grad_norm": 0.5411577820777893,
      "learning_rate": 2.7698486614150516e-05,
      "loss": 0.8122,
      "step": 15740
    },
    {
      "epoch": 1.0387818028921463,
      "grad_norm": 0.5698654055595398,
      "learning_rate": 2.7669874321752715e-05,
      "loss": 0.8299,
      "step": 15750
    },
    {
      "epoch": 1.0394413573631012,
      "grad_norm": 0.5780178308486938,
      "learning_rate": 2.7641258491405508e-05,
      "loss": 0.7836,
      "step": 15760
    },
    {
      "epoch": 1.040100911834056,
      "grad_norm": 0.4700949490070343,
      "learning_rate": 2.7612639161028804e-05,
      "loss": 0.8376,
      "step": 15770
    },
    {
      "epoch": 1.040760466305011,
      "grad_norm": 0.5386998057365417,
      "learning_rate": 2.7584016368547135e-05,
      "loss": 0.8318,
      "step": 15780
    },
    {
      "epoch": 1.0414200207759658,
      "grad_norm": 0.5147479772567749,
      "learning_rate": 2.7555390151889626e-05,
      "loss": 0.8306,
      "step": 15790
    },
    {
      "epoch": 1.0420795752469207,
      "grad_norm": 0.4795542061328888,
      "learning_rate": 2.7526760548989933e-05,
      "loss": 0.8083,
      "step": 15800
    },
    {
      "epoch": 1.0427391297178756,
      "grad_norm": 0.5005327463150024,
      "learning_rate": 2.7498127597786212e-05,
      "loss": 0.7766,
      "step": 15810
    },
    {
      "epoch": 1.0433986841888305,
      "grad_norm": 0.5824893116950989,
      "learning_rate": 2.746949133622103e-05,
      "loss": 0.8,
      "step": 15820
    },
    {
      "epoch": 1.0440582386597854,
      "grad_norm": 0.5469245314598083,
      "learning_rate": 2.7440851802241373e-05,
      "loss": 0.8154,
      "step": 15830
    },
    {
      "epoch": 1.04471779313074,
      "grad_norm": 0.6703712940216064,
      "learning_rate": 2.7412209033798552e-05,
      "loss": 0.8327,
      "step": 15840
    },
    {
      "epoch": 1.045377347601695,
      "grad_norm": 0.5922505259513855,
      "learning_rate": 2.738356306884815e-05,
      "loss": 0.8196,
      "step": 15850
    },
    {
      "epoch": 1.0460369020726499,
      "grad_norm": 0.5021489262580872,
      "learning_rate": 2.7354913945350006e-05,
      "loss": 0.7733,
      "step": 15860
    },
    {
      "epoch": 1.0466964565436048,
      "grad_norm": 0.5870568156242371,
      "learning_rate": 2.7326261701268124e-05,
      "loss": 0.8064,
      "step": 15870
    },
    {
      "epoch": 1.0473560110145597,
      "grad_norm": 0.5342341661453247,
      "learning_rate": 2.7297606374570665e-05,
      "loss": 0.7878,
      "step": 15880
    },
    {
      "epoch": 1.0480155654855146,
      "grad_norm": 0.5717883706092834,
      "learning_rate": 2.726894800322986e-05,
      "loss": 0.8263,
      "step": 15890
    },
    {
      "epoch": 1.0486751199564694,
      "grad_norm": 0.5930851697921753,
      "learning_rate": 2.7240286625221984e-05,
      "loss": 0.8175,
      "step": 15900
    },
    {
      "epoch": 1.0493346744274243,
      "grad_norm": 0.555741012096405,
      "learning_rate": 2.7211622278527278e-05,
      "loss": 0.8492,
      "step": 15910
    },
    {
      "epoch": 1.0499942288983792,
      "grad_norm": 0.5682945251464844,
      "learning_rate": 2.7182955001129932e-05,
      "loss": 0.8579,
      "step": 15920
    },
    {
      "epoch": 1.050653783369334,
      "grad_norm": 0.6149784922599792,
      "learning_rate": 2.7154284831018036e-05,
      "loss": 0.8437,
      "step": 15930
    },
    {
      "epoch": 1.0513133378402888,
      "grad_norm": 0.5615754723548889,
      "learning_rate": 2.7125611806183472e-05,
      "loss": 0.834,
      "step": 15940
    },
    {
      "epoch": 1.0519728923112437,
      "grad_norm": 0.5518723726272583,
      "learning_rate": 2.709693596462195e-05,
      "loss": 0.7846,
      "step": 15950
    },
    {
      "epoch": 1.0526324467821986,
      "grad_norm": 0.521699070930481,
      "learning_rate": 2.7068257344332864e-05,
      "loss": 0.787,
      "step": 15960
    },
    {
      "epoch": 1.0532920012531535,
      "grad_norm": 0.5240887999534607,
      "learning_rate": 2.703957598331934e-05,
      "loss": 0.7999,
      "step": 15970
    },
    {
      "epoch": 1.0539515557241084,
      "grad_norm": 0.5104392170906067,
      "learning_rate": 2.7010891919588103e-05,
      "loss": 0.8562,
      "step": 15980
    },
    {
      "epoch": 1.0546111101950633,
      "grad_norm": 0.5361871719360352,
      "learning_rate": 2.6982205191149462e-05,
      "loss": 0.8085,
      "step": 15990
    },
    {
      "epoch": 1.0552706646660182,
      "grad_norm": 0.5302392244338989,
      "learning_rate": 2.6953515836017273e-05,
      "loss": 0.8436,
      "step": 16000
    },
    {
      "epoch": 1.055930219136973,
      "grad_norm": 0.5710680484771729,
      "learning_rate": 2.6924823892208866e-05,
      "loss": 0.7765,
      "step": 16010
    },
    {
      "epoch": 1.056589773607928,
      "grad_norm": 0.5633246898651123,
      "learning_rate": 2.689612939774499e-05,
      "loss": 0.8259,
      "step": 16020
    },
    {
      "epoch": 1.0572493280788826,
      "grad_norm": 0.5586814284324646,
      "learning_rate": 2.686743239064979e-05,
      "loss": 0.8446,
      "step": 16030
    },
    {
      "epoch": 1.0579088825498375,
      "grad_norm": 0.6109247207641602,
      "learning_rate": 2.6838732908950732e-05,
      "loss": 0.788,
      "step": 16040
    },
    {
      "epoch": 1.0585684370207924,
      "grad_norm": 0.5148352384567261,
      "learning_rate": 2.6810030990678552e-05,
      "loss": 0.8345,
      "step": 16050
    },
    {
      "epoch": 1.0592279914917473,
      "grad_norm": 0.5439609289169312,
      "learning_rate": 2.678132667386724e-05,
      "loss": 0.8561,
      "step": 16060
    },
    {
      "epoch": 1.0598875459627022,
      "grad_norm": 0.6023163199424744,
      "learning_rate": 2.675261999655394e-05,
      "loss": 0.8097,
      "step": 16070
    },
    {
      "epoch": 1.060547100433657,
      "grad_norm": 0.5256965160369873,
      "learning_rate": 2.672391099677895e-05,
      "loss": 0.775,
      "step": 16080
    },
    {
      "epoch": 1.061206654904612,
      "grad_norm": 0.5441160798072815,
      "learning_rate": 2.6695199712585605e-05,
      "loss": 0.8455,
      "step": 16090
    },
    {
      "epoch": 1.0618662093755669,
      "grad_norm": 0.5260411500930786,
      "learning_rate": 2.6666486182020317e-05,
      "loss": 0.7795,
      "step": 16100
    },
    {
      "epoch": 1.0625257638465218,
      "grad_norm": 0.473296195268631,
      "learning_rate": 2.6637770443132427e-05,
      "loss": 0.8121,
      "step": 16110
    },
    {
      "epoch": 1.0631853183174766,
      "grad_norm": 0.6172069311141968,
      "learning_rate": 2.6609052533974233e-05,
      "loss": 0.828,
      "step": 16120
    },
    {
      "epoch": 1.0638448727884313,
      "grad_norm": 0.5585634708404541,
      "learning_rate": 2.6580332492600905e-05,
      "loss": 0.7759,
      "step": 16130
    },
    {
      "epoch": 1.0645044272593862,
      "grad_norm": 0.5809791088104248,
      "learning_rate": 2.6551610357070434e-05,
      "loss": 0.8133,
      "step": 16140
    },
    {
      "epoch": 1.065163981730341,
      "grad_norm": 0.5252553820610046,
      "learning_rate": 2.6522886165443578e-05,
      "loss": 0.7999,
      "step": 16150
    },
    {
      "epoch": 1.065823536201296,
      "grad_norm": 0.5203016400337219,
      "learning_rate": 2.649415995578383e-05,
      "loss": 0.822,
      "step": 16160
    },
    {
      "epoch": 1.0664830906722509,
      "grad_norm": 0.5954844355583191,
      "learning_rate": 2.646543176615737e-05,
      "loss": 0.8165,
      "step": 16170
    },
    {
      "epoch": 1.0671426451432058,
      "grad_norm": 0.5122602581977844,
      "learning_rate": 2.6436701634632964e-05,
      "loss": 0.7921,
      "step": 16180
    },
    {
      "epoch": 1.0678021996141607,
      "grad_norm": 0.623115599155426,
      "learning_rate": 2.640796959928199e-05,
      "loss": 0.8371,
      "step": 16190
    },
    {
      "epoch": 1.0684617540851156,
      "grad_norm": 0.48406779766082764,
      "learning_rate": 2.6379235698178324e-05,
      "loss": 0.8037,
      "step": 16200
    },
    {
      "epoch": 1.0691213085560705,
      "grad_norm": 0.5848739743232727,
      "learning_rate": 2.6350499969398324e-05,
      "loss": 0.8351,
      "step": 16210
    },
    {
      "epoch": 1.0697808630270251,
      "grad_norm": 0.5309697985649109,
      "learning_rate": 2.632176245102077e-05,
      "loss": 0.8565,
      "step": 16220
    },
    {
      "epoch": 1.07044041749798,
      "grad_norm": 0.5303778648376465,
      "learning_rate": 2.629302318112682e-05,
      "loss": 0.7997,
      "step": 16230
    },
    {
      "epoch": 1.071099971968935,
      "grad_norm": 0.6161729097366333,
      "learning_rate": 2.626428219779994e-05,
      "loss": 0.8235,
      "step": 16240
    },
    {
      "epoch": 1.0717595264398898,
      "grad_norm": 0.5845171213150024,
      "learning_rate": 2.6235539539125862e-05,
      "loss": 0.8098,
      "step": 16250
    },
    {
      "epoch": 1.0724190809108447,
      "grad_norm": 0.5690910220146179,
      "learning_rate": 2.6206795243192565e-05,
      "loss": 0.8089,
      "step": 16260
    },
    {
      "epoch": 1.0730786353817996,
      "grad_norm": 0.571356475353241,
      "learning_rate": 2.6178049348090168e-05,
      "loss": 0.8165,
      "step": 16270
    },
    {
      "epoch": 1.0737381898527545,
      "grad_norm": 0.5281323194503784,
      "learning_rate": 2.6149301891910928e-05,
      "loss": 0.7906,
      "step": 16280
    },
    {
      "epoch": 1.0743977443237094,
      "grad_norm": 0.5516245365142822,
      "learning_rate": 2.612055291274916e-05,
      "loss": 0.8007,
      "step": 16290
    },
    {
      "epoch": 1.0750572987946643,
      "grad_norm": 0.50286465883255,
      "learning_rate": 2.6091802448701215e-05,
      "loss": 0.8348,
      "step": 16300
    },
    {
      "epoch": 1.0757168532656192,
      "grad_norm": 0.604536771774292,
      "learning_rate": 2.6063050537865374e-05,
      "loss": 0.8203,
      "step": 16310
    },
    {
      "epoch": 1.076376407736574,
      "grad_norm": 0.5543378591537476,
      "learning_rate": 2.603429721834188e-05,
      "loss": 0.8479,
      "step": 16320
    },
    {
      "epoch": 1.0770359622075287,
      "grad_norm": 0.5020339488983154,
      "learning_rate": 2.6005542528232818e-05,
      "loss": 0.7891,
      "step": 16330
    },
    {
      "epoch": 1.0776955166784836,
      "grad_norm": 0.5786985158920288,
      "learning_rate": 2.5976786505642088e-05,
      "loss": 0.8337,
      "step": 16340
    },
    {
      "epoch": 1.0783550711494385,
      "grad_norm": 0.5452788472175598,
      "learning_rate": 2.5948029188675365e-05,
      "loss": 0.8378,
      "step": 16350
    },
    {
      "epoch": 1.0790146256203934,
      "grad_norm": 0.610916018486023,
      "learning_rate": 2.5919270615440034e-05,
      "loss": 0.7981,
      "step": 16360
    },
    {
      "epoch": 1.0796741800913483,
      "grad_norm": 0.6328546404838562,
      "learning_rate": 2.5890510824045143e-05,
      "loss": 0.7947,
      "step": 16370
    },
    {
      "epoch": 1.0803337345623032,
      "grad_norm": 0.5229448676109314,
      "learning_rate": 2.5861749852601363e-05,
      "loss": 0.8286,
      "step": 16380
    },
    {
      "epoch": 1.080993289033258,
      "grad_norm": 0.5723561644554138,
      "learning_rate": 2.5832987739220926e-05,
      "loss": 0.8037,
      "step": 16390
    },
    {
      "epoch": 1.081652843504213,
      "grad_norm": 0.5695706009864807,
      "learning_rate": 2.580422452201756e-05,
      "loss": 0.807,
      "step": 16400
    },
    {
      "epoch": 1.0823123979751679,
      "grad_norm": 0.5206248164176941,
      "learning_rate": 2.5775460239106493e-05,
      "loss": 0.7963,
      "step": 16410
    },
    {
      "epoch": 1.0829719524461225,
      "grad_norm": 0.4780265986919403,
      "learning_rate": 2.5746694928604316e-05,
      "loss": 0.8223,
      "step": 16420
    },
    {
      "epoch": 1.0836315069170774,
      "grad_norm": 0.5492380857467651,
      "learning_rate": 2.571792862862902e-05,
      "loss": 0.8205,
      "step": 16430
    },
    {
      "epoch": 1.0842910613880323,
      "grad_norm": 0.5519129633903503,
      "learning_rate": 2.568916137729989e-05,
      "loss": 0.7784,
      "step": 16440
    },
    {
      "epoch": 1.0849506158589872,
      "grad_norm": 0.509762167930603,
      "learning_rate": 2.5660393212737495e-05,
      "loss": 0.803,
      "step": 16450
    },
    {
      "epoch": 1.0856101703299421,
      "grad_norm": 0.5750563740730286,
      "learning_rate": 2.563162417306358e-05,
      "loss": 0.8163,
      "step": 16460
    },
    {
      "epoch": 1.086269724800897,
      "grad_norm": 0.6123660802841187,
      "learning_rate": 2.5602854296401056e-05,
      "loss": 0.7731,
      "step": 16470
    },
    {
      "epoch": 1.086929279271852,
      "grad_norm": 0.5463810563087463,
      "learning_rate": 2.5574083620873978e-05,
      "loss": 0.8075,
      "step": 16480
    },
    {
      "epoch": 1.0875888337428068,
      "grad_norm": 0.5543496012687683,
      "learning_rate": 2.5545312184607416e-05,
      "loss": 0.817,
      "step": 16490
    },
    {
      "epoch": 1.0882483882137617,
      "grad_norm": 0.5343895554542542,
      "learning_rate": 2.551654002572747e-05,
      "loss": 0.8106,
      "step": 16500
    },
    {
      "epoch": 1.0889079426847164,
      "grad_norm": 0.6146511435508728,
      "learning_rate": 2.5487767182361193e-05,
      "loss": 0.8039,
      "step": 16510
    },
    {
      "epoch": 1.0895674971556712,
      "grad_norm": 0.6023949384689331,
      "learning_rate": 2.5458993692636553e-05,
      "loss": 0.8165,
      "step": 16520
    },
    {
      "epoch": 1.0902270516266261,
      "grad_norm": 0.6236518621444702,
      "learning_rate": 2.5430219594682365e-05,
      "loss": 0.8269,
      "step": 16530
    },
    {
      "epoch": 1.090886606097581,
      "grad_norm": 0.4938434064388275,
      "learning_rate": 2.5401444926628248e-05,
      "loss": 0.8249,
      "step": 16540
    },
    {
      "epoch": 1.091546160568536,
      "grad_norm": 0.6218588948249817,
      "learning_rate": 2.5372669726604585e-05,
      "loss": 0.8198,
      "step": 16550
    },
    {
      "epoch": 1.0922057150394908,
      "grad_norm": 0.5304175615310669,
      "learning_rate": 2.5343894032742448e-05,
      "loss": 0.8154,
      "step": 16560
    },
    {
      "epoch": 1.0928652695104457,
      "grad_norm": 0.5762113928794861,
      "learning_rate": 2.53151178831736e-05,
      "loss": 0.8324,
      "step": 16570
    },
    {
      "epoch": 1.0935248239814006,
      "grad_norm": 0.6506617665290833,
      "learning_rate": 2.528634131603036e-05,
      "loss": 0.7989,
      "step": 16580
    },
    {
      "epoch": 1.0941843784523555,
      "grad_norm": 0.5646471977233887,
      "learning_rate": 2.5257564369445626e-05,
      "loss": 0.8396,
      "step": 16590
    },
    {
      "epoch": 1.0948439329233104,
      "grad_norm": 0.5536471009254456,
      "learning_rate": 2.5228787081552806e-05,
      "loss": 0.814,
      "step": 16600
    },
    {
      "epoch": 1.0955034873942653,
      "grad_norm": 0.49897870421409607,
      "learning_rate": 2.520000949048574e-05,
      "loss": 0.7905,
      "step": 16610
    },
    {
      "epoch": 1.09616304186522,
      "grad_norm": 0.4896376430988312,
      "learning_rate": 2.5171231634378707e-05,
      "loss": 0.8267,
      "step": 16620
    },
    {
      "epoch": 1.0968225963361748,
      "grad_norm": 0.5398492217063904,
      "learning_rate": 2.5142453551366275e-05,
      "loss": 0.8251,
      "step": 16630
    },
    {
      "epoch": 1.0974821508071297,
      "grad_norm": 0.5313453078269958,
      "learning_rate": 2.5113675279583365e-05,
      "loss": 0.807,
      "step": 16640
    },
    {
      "epoch": 1.0981417052780846,
      "grad_norm": 0.6137615442276001,
      "learning_rate": 2.508489685716513e-05,
      "loss": 0.777,
      "step": 16650
    },
    {
      "epoch": 1.0988012597490395,
      "grad_norm": 0.522820770740509,
      "learning_rate": 2.5056118322246924e-05,
      "loss": 0.7904,
      "step": 16660
    },
    {
      "epoch": 1.0994608142199944,
      "grad_norm": 0.6334253549575806,
      "learning_rate": 2.5027339712964247e-05,
      "loss": 0.8444,
      "step": 16670
    },
    {
      "epoch": 1.1001203686909493,
      "grad_norm": 0.5604522824287415,
      "learning_rate": 2.499856106745271e-05,
      "loss": 0.8036,
      "step": 16680
    },
    {
      "epoch": 1.1007799231619042,
      "grad_norm": 0.5039092302322388,
      "learning_rate": 2.496978242384795e-05,
      "loss": 0.8408,
      "step": 16690
    },
    {
      "epoch": 1.101439477632859,
      "grad_norm": 0.5355373620986938,
      "learning_rate": 2.494100382028562e-05,
      "loss": 0.8177,
      "step": 16700
    },
    {
      "epoch": 1.1020990321038138,
      "grad_norm": 0.4862341582775116,
      "learning_rate": 2.4912225294901314e-05,
      "loss": 0.7895,
      "step": 16710
    },
    {
      "epoch": 1.1027585865747687,
      "grad_norm": 0.5653737783432007,
      "learning_rate": 2.4883446885830528e-05,
      "loss": 0.9132,
      "step": 16720
    },
    {
      "epoch": 1.1034181410457236,
      "grad_norm": 0.49577951431274414,
      "learning_rate": 2.485466863120859e-05,
      "loss": 0.8172,
      "step": 16730
    },
    {
      "epoch": 1.1040776955166784,
      "grad_norm": 0.550129771232605,
      "learning_rate": 2.4825890569170628e-05,
      "loss": 0.842,
      "step": 16740
    },
    {
      "epoch": 1.1047372499876333,
      "grad_norm": 0.5835773944854736,
      "learning_rate": 2.4797112737851528e-05,
      "loss": 0.8226,
      "step": 16750
    },
    {
      "epoch": 1.1053968044585882,
      "grad_norm": 0.5645239353179932,
      "learning_rate": 2.4768335175385867e-05,
      "loss": 0.8272,
      "step": 16760
    },
    {
      "epoch": 1.1060563589295431,
      "grad_norm": 0.5530699491500854,
      "learning_rate": 2.4739557919907842e-05,
      "loss": 0.8276,
      "step": 16770
    },
    {
      "epoch": 1.106715913400498,
      "grad_norm": 0.5654169321060181,
      "learning_rate": 2.471078100955126e-05,
      "loss": 0.8476,
      "step": 16780
    },
    {
      "epoch": 1.107375467871453,
      "grad_norm": 0.6649032831192017,
      "learning_rate": 2.468200448244949e-05,
      "loss": 0.7893,
      "step": 16790
    },
    {
      "epoch": 1.1080350223424078,
      "grad_norm": 0.5665352940559387,
      "learning_rate": 2.4653228376735362e-05,
      "loss": 0.7691,
      "step": 16800
    },
    {
      "epoch": 1.1086945768133625,
      "grad_norm": 0.5382755398750305,
      "learning_rate": 2.462445273054116e-05,
      "loss": 0.7964,
      "step": 16810
    },
    {
      "epoch": 1.1093541312843174,
      "grad_norm": 0.5529101490974426,
      "learning_rate": 2.4595677581998557e-05,
      "loss": 0.8219,
      "step": 16820
    },
    {
      "epoch": 1.1100136857552723,
      "grad_norm": 0.5242228507995605,
      "learning_rate": 2.4566902969238574e-05,
      "loss": 0.8117,
      "step": 16830
    },
    {
      "epoch": 1.1106732402262272,
      "grad_norm": 0.6301599740982056,
      "learning_rate": 2.4538128930391518e-05,
      "loss": 0.8044,
      "step": 16840
    },
    {
      "epoch": 1.111332794697182,
      "grad_norm": 0.5453548431396484,
      "learning_rate": 2.450935550358694e-05,
      "loss": 0.8587,
      "step": 16850
    },
    {
      "epoch": 1.111992349168137,
      "grad_norm": 0.5987652540206909,
      "learning_rate": 2.4480582726953553e-05,
      "loss": 0.805,
      "step": 16860
    },
    {
      "epoch": 1.1126519036390918,
      "grad_norm": 0.5183432102203369,
      "learning_rate": 2.4451810638619243e-05,
      "loss": 0.7719,
      "step": 16870
    },
    {
      "epoch": 1.1133114581100467,
      "grad_norm": 0.5366532206535339,
      "learning_rate": 2.4423039276710973e-05,
      "loss": 0.7863,
      "step": 16880
    },
    {
      "epoch": 1.1139710125810016,
      "grad_norm": 0.5672691464424133,
      "learning_rate": 2.4394268679354736e-05,
      "loss": 0.8007,
      "step": 16890
    },
    {
      "epoch": 1.1146305670519565,
      "grad_norm": 0.5237433314323425,
      "learning_rate": 2.4365498884675515e-05,
      "loss": 0.7789,
      "step": 16900
    },
    {
      "epoch": 1.1152901215229112,
      "grad_norm": 0.5496178865432739,
      "learning_rate": 2.4336729930797234e-05,
      "loss": 0.8217,
      "step": 16910
    },
    {
      "epoch": 1.115949675993866,
      "grad_norm": 0.5697668790817261,
      "learning_rate": 2.43079618558427e-05,
      "loss": 0.8014,
      "step": 16920
    },
    {
      "epoch": 1.116609230464821,
      "grad_norm": 0.6329261064529419,
      "learning_rate": 2.4279194697933556e-05,
      "loss": 0.8646,
      "step": 16930
    },
    {
      "epoch": 1.1172687849357759,
      "grad_norm": 0.5397175550460815,
      "learning_rate": 2.4250428495190227e-05,
      "loss": 0.855,
      "step": 16940
    },
    {
      "epoch": 1.1179283394067308,
      "grad_norm": 0.5569603443145752,
      "learning_rate": 2.4221663285731875e-05,
      "loss": 0.7462,
      "step": 16950
    },
    {
      "epoch": 1.1185878938776856,
      "grad_norm": 0.6040708422660828,
      "learning_rate": 2.4192899107676343e-05,
      "loss": 0.7782,
      "step": 16960
    },
    {
      "epoch": 1.1192474483486405,
      "grad_norm": 0.5611158013343811,
      "learning_rate": 2.4164135999140112e-05,
      "loss": 0.831,
      "step": 16970
    },
    {
      "epoch": 1.1199070028195954,
      "grad_norm": 0.6430832743644714,
      "learning_rate": 2.4135373998238254e-05,
      "loss": 0.8552,
      "step": 16980
    },
    {
      "epoch": 1.1205665572905503,
      "grad_norm": 0.5347371101379395,
      "learning_rate": 2.4106613143084335e-05,
      "loss": 0.8524,
      "step": 16990
    },
    {
      "epoch": 1.121226111761505,
      "grad_norm": 0.5940733551979065,
      "learning_rate": 2.407785347179045e-05,
      "loss": 0.8171,
      "step": 17000
    },
    {
      "epoch": 1.1218856662324599,
      "grad_norm": 0.6383303999900818,
      "learning_rate": 2.4049095022467094e-05,
      "loss": 0.8148,
      "step": 17010
    },
    {
      "epoch": 1.1225452207034148,
      "grad_norm": 0.5206311345100403,
      "learning_rate": 2.4020337833223165e-05,
      "loss": 0.8517,
      "step": 17020
    },
    {
      "epoch": 1.1232047751743697,
      "grad_norm": 0.5240987539291382,
      "learning_rate": 2.399158194216588e-05,
      "loss": 0.806,
      "step": 17030
    },
    {
      "epoch": 1.1238643296453246,
      "grad_norm": 0.5234224796295166,
      "learning_rate": 2.3962827387400724e-05,
      "loss": 0.8178,
      "step": 17040
    },
    {
      "epoch": 1.1245238841162795,
      "grad_norm": 0.6092124581336975,
      "learning_rate": 2.393407420703143e-05,
      "loss": 0.8218,
      "step": 17050
    },
    {
      "epoch": 1.1251834385872344,
      "grad_norm": 0.5292211771011353,
      "learning_rate": 2.3905322439159905e-05,
      "loss": 0.8425,
      "step": 17060
    },
    {
      "epoch": 1.1258429930581892,
      "grad_norm": 0.5749512314796448,
      "learning_rate": 2.3876572121886186e-05,
      "loss": 0.8278,
      "step": 17070
    },
    {
      "epoch": 1.1265025475291441,
      "grad_norm": 0.4919281005859375,
      "learning_rate": 2.3847823293308372e-05,
      "loss": 0.8333,
      "step": 17080
    },
    {
      "epoch": 1.1271621020000988,
      "grad_norm": 0.5225867033004761,
      "learning_rate": 2.3819075991522604e-05,
      "loss": 0.8135,
      "step": 17090
    },
    {
      "epoch": 1.127821656471054,
      "grad_norm": 0.5524780750274658,
      "learning_rate": 2.3790330254623e-05,
      "loss": 0.7968,
      "step": 17100
    },
    {
      "epoch": 1.1284812109420086,
      "grad_norm": 0.5935143828392029,
      "learning_rate": 2.3761586120701597e-05,
      "loss": 0.7822,
      "step": 17110
    },
    {
      "epoch": 1.1291407654129635,
      "grad_norm": 0.501223623752594,
      "learning_rate": 2.3732843627848327e-05,
      "loss": 0.7745,
      "step": 17120
    },
    {
      "epoch": 1.1298003198839184,
      "grad_norm": 0.6654418706893921,
      "learning_rate": 2.3704102814150912e-05,
      "loss": 0.7994,
      "step": 17130
    },
    {
      "epoch": 1.1304598743548733,
      "grad_norm": 0.5945008993148804,
      "learning_rate": 2.3675363717694878e-05,
      "loss": 0.8149,
      "step": 17140
    },
    {
      "epoch": 1.1311194288258282,
      "grad_norm": 0.48068150877952576,
      "learning_rate": 2.3646626376563468e-05,
      "loss": 0.8087,
      "step": 17150
    },
    {
      "epoch": 1.131778983296783,
      "grad_norm": 0.6289824843406677,
      "learning_rate": 2.36178908288376e-05,
      "loss": 0.8473,
      "step": 17160
    },
    {
      "epoch": 1.132438537767738,
      "grad_norm": 0.5678028464317322,
      "learning_rate": 2.358915711259581e-05,
      "loss": 0.7846,
      "step": 17170
    },
    {
      "epoch": 1.1330980922386928,
      "grad_norm": 0.598111093044281,
      "learning_rate": 2.3560425265914204e-05,
      "loss": 0.803,
      "step": 17180
    },
    {
      "epoch": 1.1337576467096477,
      "grad_norm": 0.6262847185134888,
      "learning_rate": 2.3531695326866423e-05,
      "loss": 0.8234,
      "step": 17190
    },
    {
      "epoch": 1.1344172011806024,
      "grad_norm": 0.510326087474823,
      "learning_rate": 2.3502967333523575e-05,
      "loss": 0.7743,
      "step": 17200
    },
    {
      "epoch": 1.1350767556515573,
      "grad_norm": 0.5270165205001831,
      "learning_rate": 2.3474241323954197e-05,
      "loss": 0.8047,
      "step": 17210
    },
    {
      "epoch": 1.1357363101225122,
      "grad_norm": 0.6041849851608276,
      "learning_rate": 2.3445517336224166e-05,
      "loss": 0.8178,
      "step": 17220
    },
    {
      "epoch": 1.136395864593467,
      "grad_norm": 0.5742443799972534,
      "learning_rate": 2.3416795408396717e-05,
      "loss": 0.8408,
      "step": 17230
    },
    {
      "epoch": 1.137055419064422,
      "grad_norm": 0.5951201319694519,
      "learning_rate": 2.3388075578532338e-05,
      "loss": 0.8082,
      "step": 17240
    },
    {
      "epoch": 1.1377149735353769,
      "grad_norm": 0.5675894021987915,
      "learning_rate": 2.3359357884688747e-05,
      "loss": 0.8107,
      "step": 17250
    },
    {
      "epoch": 1.1383745280063318,
      "grad_norm": 0.5660733580589294,
      "learning_rate": 2.333064236492081e-05,
      "loss": 0.8182,
      "step": 17260
    },
    {
      "epoch": 1.1390340824772867,
      "grad_norm": 0.5753691792488098,
      "learning_rate": 2.3301929057280536e-05,
      "loss": 0.8055,
      "step": 17270
    },
    {
      "epoch": 1.1396936369482416,
      "grad_norm": 0.5021584033966064,
      "learning_rate": 2.327321799981699e-05,
      "loss": 0.8195,
      "step": 17280
    },
    {
      "epoch": 1.1403531914191962,
      "grad_norm": 0.5370116233825684,
      "learning_rate": 2.3244509230576262e-05,
      "loss": 0.7708,
      "step": 17290
    },
    {
      "epoch": 1.1410127458901511,
      "grad_norm": 0.6154167056083679,
      "learning_rate": 2.321580278760141e-05,
      "loss": 0.82,
      "step": 17300
    },
    {
      "epoch": 1.141672300361106,
      "grad_norm": 0.5904589891433716,
      "learning_rate": 2.318709870893239e-05,
      "loss": 0.8274,
      "step": 17310
    },
    {
      "epoch": 1.142331854832061,
      "grad_norm": 0.5207957625389099,
      "learning_rate": 2.315839703260605e-05,
      "loss": 0.7682,
      "step": 17320
    },
    {
      "epoch": 1.1429914093030158,
      "grad_norm": 0.6100203394889832,
      "learning_rate": 2.312969779665604e-05,
      "loss": 0.7972,
      "step": 17330
    },
    {
      "epoch": 1.1436509637739707,
      "grad_norm": 0.5301212072372437,
      "learning_rate": 2.3101001039112794e-05,
      "loss": 0.8483,
      "step": 17340
    },
    {
      "epoch": 1.1443105182449256,
      "grad_norm": 0.5791840553283691,
      "learning_rate": 2.3072306798003433e-05,
      "loss": 0.7987,
      "step": 17350
    },
    {
      "epoch": 1.1449700727158805,
      "grad_norm": 0.5505053400993347,
      "learning_rate": 2.3043615111351766e-05,
      "loss": 0.8104,
      "step": 17360
    },
    {
      "epoch": 1.1456296271868354,
      "grad_norm": 0.5380775332450867,
      "learning_rate": 2.301492601717821e-05,
      "loss": 0.8407,
      "step": 17370
    },
    {
      "epoch": 1.1462891816577903,
      "grad_norm": 0.5672085285186768,
      "learning_rate": 2.298623955349975e-05,
      "loss": 0.7932,
      "step": 17380
    },
    {
      "epoch": 1.1469487361287451,
      "grad_norm": 0.6575245261192322,
      "learning_rate": 2.2957555758329874e-05,
      "loss": 0.8142,
      "step": 17390
    },
    {
      "epoch": 1.1476082905996998,
      "grad_norm": 0.5814910531044006,
      "learning_rate": 2.2928874669678548e-05,
      "loss": 0.8402,
      "step": 17400
    },
    {
      "epoch": 1.1482678450706547,
      "grad_norm": 0.557349681854248,
      "learning_rate": 2.2900196325552143e-05,
      "loss": 0.7956,
      "step": 17410
    },
    {
      "epoch": 1.1489273995416096,
      "grad_norm": 0.6413754820823669,
      "learning_rate": 2.2871520763953395e-05,
      "loss": 0.8004,
      "step": 17420
    },
    {
      "epoch": 1.1495869540125645,
      "grad_norm": 0.5818390846252441,
      "learning_rate": 2.2842848022881367e-05,
      "loss": 0.8082,
      "step": 17430
    },
    {
      "epoch": 1.1502465084835194,
      "grad_norm": 0.5735812783241272,
      "learning_rate": 2.281417814033135e-05,
      "loss": 0.8373,
      "step": 17440
    },
    {
      "epoch": 1.1509060629544743,
      "grad_norm": 0.6169028282165527,
      "learning_rate": 2.2785511154294872e-05,
      "loss": 0.8308,
      "step": 17450
    },
    {
      "epoch": 1.1515656174254292,
      "grad_norm": 0.5504375100135803,
      "learning_rate": 2.2756847102759625e-05,
      "loss": 0.8558,
      "step": 17460
    },
    {
      "epoch": 1.152225171896384,
      "grad_norm": 0.43717899918556213,
      "learning_rate": 2.2728186023709406e-05,
      "loss": 0.7973,
      "step": 17470
    },
    {
      "epoch": 1.152884726367339,
      "grad_norm": 0.5762133002281189,
      "learning_rate": 2.269952795512407e-05,
      "loss": 0.8107,
      "step": 17480
    },
    {
      "epoch": 1.1535442808382936,
      "grad_norm": 0.5894545912742615,
      "learning_rate": 2.267087293497948e-05,
      "loss": 0.8131,
      "step": 17490
    },
    {
      "epoch": 1.1542038353092485,
      "grad_norm": 0.5659452080726624,
      "learning_rate": 2.2642221001247475e-05,
      "loss": 0.8045,
      "step": 17500
    },
    {
      "epoch": 1.1548633897802034,
      "grad_norm": 0.5233924388885498,
      "learning_rate": 2.2613572191895792e-05,
      "loss": 0.7889,
      "step": 17510
    },
    {
      "epoch": 1.1555229442511583,
      "grad_norm": 0.5157888531684875,
      "learning_rate": 2.2584926544888034e-05,
      "loss": 0.8099,
      "step": 17520
    },
    {
      "epoch": 1.1561824987221132,
      "grad_norm": 0.5643179416656494,
      "learning_rate": 2.2556284098183593e-05,
      "loss": 0.7983,
      "step": 17530
    },
    {
      "epoch": 1.156842053193068,
      "grad_norm": 0.6216257214546204,
      "learning_rate": 2.252764488973765e-05,
      "loss": 0.8089,
      "step": 17540
    },
    {
      "epoch": 1.157501607664023,
      "grad_norm": 0.5228362083435059,
      "learning_rate": 2.249900895750107e-05,
      "loss": 0.7835,
      "step": 17550
    },
    {
      "epoch": 1.1581611621349779,
      "grad_norm": 0.5488897562026978,
      "learning_rate": 2.247037633942041e-05,
      "loss": 0.8333,
      "step": 17560
    },
    {
      "epoch": 1.1588207166059328,
      "grad_norm": 0.654600977897644,
      "learning_rate": 2.244174707343779e-05,
      "loss": 0.7936,
      "step": 17570
    },
    {
      "epoch": 1.1594802710768874,
      "grad_norm": 0.620481014251709,
      "learning_rate": 2.2413121197490924e-05,
      "loss": 0.8162,
      "step": 17580
    },
    {
      "epoch": 1.1601398255478423,
      "grad_norm": 0.5952094793319702,
      "learning_rate": 2.2384498749513014e-05,
      "loss": 0.8404,
      "step": 17590
    },
    {
      "epoch": 1.1607993800187972,
      "grad_norm": 0.5122694373130798,
      "learning_rate": 2.235587976743273e-05,
      "loss": 0.8346,
      "step": 17600
    },
    {
      "epoch": 1.1614589344897521,
      "grad_norm": 0.5466095209121704,
      "learning_rate": 2.2327264289174155e-05,
      "loss": 0.812,
      "step": 17610
    },
    {
      "epoch": 1.162118488960707,
      "grad_norm": 0.5331703424453735,
      "learning_rate": 2.2298652352656698e-05,
      "loss": 0.805,
      "step": 17620
    },
    {
      "epoch": 1.162778043431662,
      "grad_norm": 0.5187802314758301,
      "learning_rate": 2.2270043995795114e-05,
      "loss": 0.7978,
      "step": 17630
    },
    {
      "epoch": 1.1634375979026168,
      "grad_norm": 0.5421300530433655,
      "learning_rate": 2.2241439256499388e-05,
      "loss": 0.8455,
      "step": 17640
    },
    {
      "epoch": 1.1640971523735717,
      "grad_norm": 0.5664952993392944,
      "learning_rate": 2.2212838172674736e-05,
      "loss": 0.8071,
      "step": 17650
    },
    {
      "epoch": 1.1647567068445266,
      "grad_norm": 0.5498470067977905,
      "learning_rate": 2.218424078222149e-05,
      "loss": 0.7909,
      "step": 17660
    },
    {
      "epoch": 1.1654162613154815,
      "grad_norm": 0.5885456800460815,
      "learning_rate": 2.2155647123035124e-05,
      "loss": 0.7942,
      "step": 17670
    },
    {
      "epoch": 1.1660758157864364,
      "grad_norm": 0.5358192324638367,
      "learning_rate": 2.2127057233006152e-05,
      "loss": 0.8011,
      "step": 17680
    },
    {
      "epoch": 1.166735370257391,
      "grad_norm": 0.5777158737182617,
      "learning_rate": 2.2098471150020102e-05,
      "loss": 0.8387,
      "step": 17690
    },
    {
      "epoch": 1.167394924728346,
      "grad_norm": 0.6417843699455261,
      "learning_rate": 2.2069888911957447e-05,
      "loss": 0.814,
      "step": 17700
    },
    {
      "epoch": 1.1680544791993008,
      "grad_norm": 0.5849559903144836,
      "learning_rate": 2.2041310556693573e-05,
      "loss": 0.801,
      "step": 17710
    },
    {
      "epoch": 1.1687140336702557,
      "grad_norm": 0.5827060341835022,
      "learning_rate": 2.2012736122098714e-05,
      "loss": 0.7806,
      "step": 17720
    },
    {
      "epoch": 1.1693735881412106,
      "grad_norm": 0.5556344389915466,
      "learning_rate": 2.1984165646037917e-05,
      "loss": 0.8079,
      "step": 17730
    },
    {
      "epoch": 1.1700331426121655,
      "grad_norm": 0.5267961621284485,
      "learning_rate": 2.1955599166370984e-05,
      "loss": 0.8203,
      "step": 17740
    },
    {
      "epoch": 1.1706926970831204,
      "grad_norm": 0.6056419610977173,
      "learning_rate": 2.19270367209524e-05,
      "loss": 0.8116,
      "step": 17750
    },
    {
      "epoch": 1.1713522515540753,
      "grad_norm": 0.575733482837677,
      "learning_rate": 2.189847834763132e-05,
      "loss": 0.792,
      "step": 17760
    },
    {
      "epoch": 1.1720118060250302,
      "grad_norm": 0.5400353670120239,
      "learning_rate": 2.1869924084251516e-05,
      "loss": 0.7937,
      "step": 17770
    },
    {
      "epoch": 1.1726713604959849,
      "grad_norm": 0.5470884442329407,
      "learning_rate": 2.184137396865129e-05,
      "loss": 0.768,
      "step": 17780
    },
    {
      "epoch": 1.1733309149669398,
      "grad_norm": 0.5165489912033081,
      "learning_rate": 2.181282803866348e-05,
      "loss": 0.7874,
      "step": 17790
    },
    {
      "epoch": 1.1739904694378946,
      "grad_norm": 0.5373079776763916,
      "learning_rate": 2.1784286332115323e-05,
      "loss": 0.7938,
      "step": 17800
    },
    {
      "epoch": 1.1746500239088495,
      "grad_norm": 0.5437031984329224,
      "learning_rate": 2.1755748886828507e-05,
      "loss": 0.823,
      "step": 17810
    },
    {
      "epoch": 1.1753095783798044,
      "grad_norm": 0.5490434169769287,
      "learning_rate": 2.172721574061906e-05,
      "loss": 0.7868,
      "step": 17820
    },
    {
      "epoch": 1.1759691328507593,
      "grad_norm": 0.5717520117759705,
      "learning_rate": 2.1698686931297317e-05,
      "loss": 0.8605,
      "step": 17830
    },
    {
      "epoch": 1.1766286873217142,
      "grad_norm": 0.5349151492118835,
      "learning_rate": 2.167016249666785e-05,
      "loss": 0.8,
      "step": 17840
    },
    {
      "epoch": 1.177288241792669,
      "grad_norm": 0.513151228427887,
      "learning_rate": 2.164164247452944e-05,
      "loss": 0.783,
      "step": 17850
    },
    {
      "epoch": 1.177947796263624,
      "grad_norm": 0.5884870290756226,
      "learning_rate": 2.161312690267504e-05,
      "loss": 0.778,
      "step": 17860
    },
    {
      "epoch": 1.1786073507345787,
      "grad_norm": 0.48820847272872925,
      "learning_rate": 2.158461581889169e-05,
      "loss": 0.8613,
      "step": 17870
    },
    {
      "epoch": 1.1792669052055336,
      "grad_norm": 0.5307297706604004,
      "learning_rate": 2.155610926096047e-05,
      "loss": 0.7844,
      "step": 17880
    },
    {
      "epoch": 1.1799264596764885,
      "grad_norm": 0.5557736158370972,
      "learning_rate": 2.152760726665648e-05,
      "loss": 0.7863,
      "step": 17890
    },
    {
      "epoch": 1.1805860141474434,
      "grad_norm": 0.6041250824928284,
      "learning_rate": 2.1499109873748777e-05,
      "loss": 0.8033,
      "step": 17900
    },
    {
      "epoch": 1.1812455686183982,
      "grad_norm": 0.6773850917816162,
      "learning_rate": 2.1470617120000307e-05,
      "loss": 0.8152,
      "step": 17910
    },
    {
      "epoch": 1.1819051230893531,
      "grad_norm": 0.5740440487861633,
      "learning_rate": 2.1442129043167874e-05,
      "loss": 0.793,
      "step": 17920
    },
    {
      "epoch": 1.182564677560308,
      "grad_norm": 0.5717465877532959,
      "learning_rate": 2.1413645681002087e-05,
      "loss": 0.83,
      "step": 17930
    },
    {
      "epoch": 1.183224232031263,
      "grad_norm": 0.6836726069450378,
      "learning_rate": 2.13851670712473e-05,
      "loss": 0.775,
      "step": 17940
    },
    {
      "epoch": 1.1838837865022178,
      "grad_norm": 0.6140259504318237,
      "learning_rate": 2.1356693251641575e-05,
      "loss": 0.8166,
      "step": 17950
    },
    {
      "epoch": 1.1845433409731727,
      "grad_norm": 0.5447438359260559,
      "learning_rate": 2.1328224259916636e-05,
      "loss": 0.7571,
      "step": 17960
    },
    {
      "epoch": 1.1852028954441276,
      "grad_norm": 0.6259800791740417,
      "learning_rate": 2.129976013379778e-05,
      "loss": 0.7944,
      "step": 17970
    },
    {
      "epoch": 1.1858624499150823,
      "grad_norm": 0.5480833649635315,
      "learning_rate": 2.1271300911003888e-05,
      "loss": 0.8111,
      "step": 17980
    },
    {
      "epoch": 1.1865220043860372,
      "grad_norm": 0.5970847606658936,
      "learning_rate": 2.1242846629247327e-05,
      "loss": 0.825,
      "step": 17990
    },
    {
      "epoch": 1.187181558856992,
      "grad_norm": 0.5720780491828918,
      "learning_rate": 2.1214397326233916e-05,
      "loss": 0.8312,
      "step": 18000
    },
    {
      "epoch": 1.187841113327947,
      "grad_norm": 0.6329047083854675,
      "learning_rate": 2.11859530396629e-05,
      "loss": 0.8134,
      "step": 18010
    },
    {
      "epoch": 1.1885006677989018,
      "grad_norm": 0.5721922516822815,
      "learning_rate": 2.115751380722683e-05,
      "loss": 0.7974,
      "step": 18020
    },
    {
      "epoch": 1.1891602222698567,
      "grad_norm": 0.5312685966491699,
      "learning_rate": 2.1129079666611602e-05,
      "loss": 0.7841,
      "step": 18030
    },
    {
      "epoch": 1.1898197767408116,
      "grad_norm": 0.6476850509643555,
      "learning_rate": 2.1100650655496347e-05,
      "loss": 0.8379,
      "step": 18040
    },
    {
      "epoch": 1.1904793312117665,
      "grad_norm": 0.5220320820808411,
      "learning_rate": 2.107222681155341e-05,
      "loss": 0.7927,
      "step": 18050
    },
    {
      "epoch": 1.1911388856827214,
      "grad_norm": 0.6444506645202637,
      "learning_rate": 2.104380817244826e-05,
      "loss": 0.7943,
      "step": 18060
    },
    {
      "epoch": 1.191798440153676,
      "grad_norm": 0.5259702205657959,
      "learning_rate": 2.1015394775839502e-05,
      "loss": 0.7614,
      "step": 18070
    },
    {
      "epoch": 1.192457994624631,
      "grad_norm": 0.5620884895324707,
      "learning_rate": 2.0986986659378777e-05,
      "loss": 0.8022,
      "step": 18080
    },
    {
      "epoch": 1.1931175490955859,
      "grad_norm": 0.5884404182434082,
      "learning_rate": 2.0958583860710733e-05,
      "loss": 0.8566,
      "step": 18090
    },
    {
      "epoch": 1.1937771035665408,
      "grad_norm": 0.5265540480613708,
      "learning_rate": 2.093018641747299e-05,
      "loss": 0.7796,
      "step": 18100
    },
    {
      "epoch": 1.1944366580374957,
      "grad_norm": 0.5781798958778381,
      "learning_rate": 2.0901794367296014e-05,
      "loss": 0.8111,
      "step": 18110
    },
    {
      "epoch": 1.1950962125084506,
      "grad_norm": 0.6055285930633545,
      "learning_rate": 2.0873407747803188e-05,
      "loss": 0.8399,
      "step": 18120
    },
    {
      "epoch": 1.1957557669794054,
      "grad_norm": 0.5819869637489319,
      "learning_rate": 2.084502659661067e-05,
      "loss": 0.8185,
      "step": 18130
    },
    {
      "epoch": 1.1964153214503603,
      "grad_norm": 0.5730735063552856,
      "learning_rate": 2.0816650951327375e-05,
      "loss": 0.827,
      "step": 18140
    },
    {
      "epoch": 1.1970748759213152,
      "grad_norm": 0.5140936374664307,
      "learning_rate": 2.078828084955492e-05,
      "loss": 0.83,
      "step": 18150
    },
    {
      "epoch": 1.19773443039227,
      "grad_norm": 0.6362972855567932,
      "learning_rate": 2.0759916328887574e-05,
      "loss": 0.8249,
      "step": 18160
    },
    {
      "epoch": 1.198393984863225,
      "grad_norm": 0.5397179126739502,
      "learning_rate": 2.0731557426912224e-05,
      "loss": 0.7908,
      "step": 18170
    },
    {
      "epoch": 1.1990535393341797,
      "grad_norm": 0.5912061929702759,
      "learning_rate": 2.0703204181208292e-05,
      "loss": 0.8028,
      "step": 18180
    },
    {
      "epoch": 1.1997130938051346,
      "grad_norm": 0.6152624487876892,
      "learning_rate": 2.0674856629347733e-05,
      "loss": 0.8105,
      "step": 18190
    },
    {
      "epoch": 1.2003726482760895,
      "grad_norm": 0.47451311349868774,
      "learning_rate": 2.0646514808894914e-05,
      "loss": 0.808,
      "step": 18200
    },
    {
      "epoch": 1.2010322027470444,
      "grad_norm": 0.6080948710441589,
      "learning_rate": 2.0618178757406645e-05,
      "loss": 0.7924,
      "step": 18210
    },
    {
      "epoch": 1.2016917572179993,
      "grad_norm": 0.5713484287261963,
      "learning_rate": 2.0589848512432065e-05,
      "loss": 0.8009,
      "step": 18220
    },
    {
      "epoch": 1.2023513116889541,
      "grad_norm": 0.589370608329773,
      "learning_rate": 2.0561524111512662e-05,
      "loss": 0.8316,
      "step": 18230
    },
    {
      "epoch": 1.203010866159909,
      "grad_norm": 0.5649095177650452,
      "learning_rate": 2.0533205592182118e-05,
      "loss": 0.8261,
      "step": 18240
    },
    {
      "epoch": 1.203670420630864,
      "grad_norm": 0.6353856325149536,
      "learning_rate": 2.0504892991966362e-05,
      "loss": 0.8201,
      "step": 18250
    },
    {
      "epoch": 1.2043299751018188,
      "grad_norm": 0.5868256092071533,
      "learning_rate": 2.047658634838347e-05,
      "loss": 0.8213,
      "step": 18260
    },
    {
      "epoch": 1.2049895295727735,
      "grad_norm": 0.6138498783111572,
      "learning_rate": 2.0448285698943627e-05,
      "loss": 0.807,
      "step": 18270
    },
    {
      "epoch": 1.2056490840437284,
      "grad_norm": 0.5619494318962097,
      "learning_rate": 2.0419991081149076e-05,
      "loss": 0.8276,
      "step": 18280
    },
    {
      "epoch": 1.2063086385146833,
      "grad_norm": 0.6716960668563843,
      "learning_rate": 2.0391702532494054e-05,
      "loss": 0.8525,
      "step": 18290
    },
    {
      "epoch": 1.2069681929856382,
      "grad_norm": 0.5778971910476685,
      "learning_rate": 2.0363420090464766e-05,
      "loss": 0.7981,
      "step": 18300
    },
    {
      "epoch": 1.207627747456593,
      "grad_norm": 0.5850443243980408,
      "learning_rate": 2.033514379253933e-05,
      "loss": 0.7916,
      "step": 18310
    },
    {
      "epoch": 1.208287301927548,
      "grad_norm": 0.5164930820465088,
      "learning_rate": 2.030687367618772e-05,
      "loss": 0.8356,
      "step": 18320
    },
    {
      "epoch": 1.2089468563985029,
      "grad_norm": 0.5958451628684998,
      "learning_rate": 2.0278609778871702e-05,
      "loss": 0.7961,
      "step": 18330
    },
    {
      "epoch": 1.2096064108694577,
      "grad_norm": 0.598099946975708,
      "learning_rate": 2.025035213804482e-05,
      "loss": 0.8413,
      "step": 18340
    },
    {
      "epoch": 1.2102659653404126,
      "grad_norm": 0.6404171586036682,
      "learning_rate": 2.0222100791152318e-05,
      "loss": 0.7793,
      "step": 18350
    },
    {
      "epoch": 1.2109255198113673,
      "grad_norm": 0.5784912109375,
      "learning_rate": 2.019385577563111e-05,
      "loss": 0.7858,
      "step": 18360
    },
    {
      "epoch": 1.2115850742823222,
      "grad_norm": 0.525601327419281,
      "learning_rate": 2.0165617128909706e-05,
      "loss": 0.779,
      "step": 18370
    },
    {
      "epoch": 1.212244628753277,
      "grad_norm": 0.5646904706954956,
      "learning_rate": 2.013738488840818e-05,
      "loss": 0.8152,
      "step": 18380
    },
    {
      "epoch": 1.212904183224232,
      "grad_norm": 0.6017470955848694,
      "learning_rate": 2.010915909153813e-05,
      "loss": 0.7865,
      "step": 18390
    },
    {
      "epoch": 1.2135637376951869,
      "grad_norm": 0.5971163511276245,
      "learning_rate": 2.00809397757026e-05,
      "loss": 0.8572,
      "step": 18400
    },
    {
      "epoch": 1.2142232921661418,
      "grad_norm": 0.5196258425712585,
      "learning_rate": 2.005272697829606e-05,
      "loss": 0.8182,
      "step": 18410
    },
    {
      "epoch": 1.2148828466370967,
      "grad_norm": 0.59040766954422,
      "learning_rate": 2.0024520736704317e-05,
      "loss": 0.8146,
      "step": 18420
    },
    {
      "epoch": 1.2155424011080516,
      "grad_norm": 0.6149417757987976,
      "learning_rate": 1.999632108830451e-05,
      "loss": 0.8172,
      "step": 18430
    },
    {
      "epoch": 1.2162019555790065,
      "grad_norm": 0.5045343041419983,
      "learning_rate": 1.9968128070465046e-05,
      "loss": 0.8656,
      "step": 18440
    },
    {
      "epoch": 1.2168615100499613,
      "grad_norm": 0.6588855385780334,
      "learning_rate": 1.9939941720545545e-05,
      "loss": 0.7979,
      "step": 18450
    },
    {
      "epoch": 1.2175210645209162,
      "grad_norm": 0.559449315071106,
      "learning_rate": 1.9911762075896765e-05,
      "loss": 0.7894,
      "step": 18460
    },
    {
      "epoch": 1.218180618991871,
      "grad_norm": 0.5257613062858582,
      "learning_rate": 1.9883589173860604e-05,
      "loss": 0.8484,
      "step": 18470
    },
    {
      "epoch": 1.2188401734628258,
      "grad_norm": 0.5910244584083557,
      "learning_rate": 1.9855423051770023e-05,
      "loss": 0.7977,
      "step": 18480
    },
    {
      "epoch": 1.2194997279337807,
      "grad_norm": 0.6656187176704407,
      "learning_rate": 1.982726374694899e-05,
      "loss": 0.853,
      "step": 18490
    },
    {
      "epoch": 1.2201592824047356,
      "grad_norm": 0.5211578607559204,
      "learning_rate": 1.9799111296712448e-05,
      "loss": 0.8342,
      "step": 18500
    },
    {
      "epoch": 1.2208188368756905,
      "grad_norm": 0.6329109072685242,
      "learning_rate": 1.9770965738366242e-05,
      "loss": 0.7913,
      "step": 18510
    },
    {
      "epoch": 1.2214783913466454,
      "grad_norm": 0.549485981464386,
      "learning_rate": 1.97428271092071e-05,
      "loss": 0.81,
      "step": 18520
    },
    {
      "epoch": 1.2221379458176003,
      "grad_norm": 0.5727259516716003,
      "learning_rate": 1.971469544652257e-05,
      "loss": 0.8258,
      "step": 18530
    },
    {
      "epoch": 1.2227975002885552,
      "grad_norm": 0.5799168944358826,
      "learning_rate": 1.9686570787590957e-05,
      "loss": 0.8108,
      "step": 18540
    },
    {
      "epoch": 1.22345705475951,
      "grad_norm": 0.6752129197120667,
      "learning_rate": 1.965845316968128e-05,
      "loss": 0.8412,
      "step": 18550
    },
    {
      "epoch": 1.2241166092304647,
      "grad_norm": 0.6387220621109009,
      "learning_rate": 1.963034263005324e-05,
      "loss": 0.7992,
      "step": 18560
    },
    {
      "epoch": 1.2247761637014196,
      "grad_norm": 0.6659034490585327,
      "learning_rate": 1.9602239205957156e-05,
      "loss": 0.8107,
      "step": 18570
    },
    {
      "epoch": 1.2254357181723745,
      "grad_norm": 0.5545511245727539,
      "learning_rate": 1.957414293463392e-05,
      "loss": 0.7758,
      "step": 18580
    },
    {
      "epoch": 1.2260952726433294,
      "grad_norm": 0.6981990933418274,
      "learning_rate": 1.9546053853314937e-05,
      "loss": 0.8423,
      "step": 18590
    },
    {
      "epoch": 1.2267548271142843,
      "grad_norm": 0.561258852481842,
      "learning_rate": 1.9517971999222097e-05,
      "loss": 0.8234,
      "step": 18600
    },
    {
      "epoch": 1.2274143815852392,
      "grad_norm": 0.5442118644714355,
      "learning_rate": 1.9489897409567694e-05,
      "loss": 0.8102,
      "step": 18610
    },
    {
      "epoch": 1.228073936056194,
      "grad_norm": 0.5950818657875061,
      "learning_rate": 1.9461830121554414e-05,
      "loss": 0.838,
      "step": 18620
    },
    {
      "epoch": 1.228733490527149,
      "grad_norm": 0.5580155253410339,
      "learning_rate": 1.9433770172375272e-05,
      "loss": 0.8407,
      "step": 18630
    },
    {
      "epoch": 1.2293930449981039,
      "grad_norm": 0.5553641319274902,
      "learning_rate": 1.9405717599213526e-05,
      "loss": 0.8295,
      "step": 18640
    },
    {
      "epoch": 1.2300525994690585,
      "grad_norm": 0.5126636624336243,
      "learning_rate": 1.9377672439242676e-05,
      "loss": 0.8066,
      "step": 18650
    },
    {
      "epoch": 1.2307121539400134,
      "grad_norm": 0.5395209193229675,
      "learning_rate": 1.9349634729626427e-05,
      "loss": 0.8221,
      "step": 18660
    },
    {
      "epoch": 1.2313717084109683,
      "grad_norm": 0.5603570342063904,
      "learning_rate": 1.932160450751857e-05,
      "loss": 0.8099,
      "step": 18670
    },
    {
      "epoch": 1.2320312628819232,
      "grad_norm": 0.5477350950241089,
      "learning_rate": 1.929358181006301e-05,
      "loss": 0.8187,
      "step": 18680
    },
    {
      "epoch": 1.232690817352878,
      "grad_norm": 0.5807533264160156,
      "learning_rate": 1.926556667439364e-05,
      "loss": 0.8133,
      "step": 18690
    },
    {
      "epoch": 1.233350371823833,
      "grad_norm": 0.5522672533988953,
      "learning_rate": 1.9237559137634358e-05,
      "loss": 0.8245,
      "step": 18700
    },
    {
      "epoch": 1.234009926294788,
      "grad_norm": 0.5667465329170227,
      "learning_rate": 1.9209559236898996e-05,
      "loss": 0.7869,
      "step": 18710
    },
    {
      "epoch": 1.2346694807657428,
      "grad_norm": 0.5482111573219299,
      "learning_rate": 1.918156700929126e-05,
      "loss": 0.7505,
      "step": 18720
    },
    {
      "epoch": 1.2353290352366977,
      "grad_norm": 0.5499863028526306,
      "learning_rate": 1.915358249190468e-05,
      "loss": 0.8484,
      "step": 18730
    },
    {
      "epoch": 1.2359885897076526,
      "grad_norm": 0.6239943504333496,
      "learning_rate": 1.9125605721822584e-05,
      "loss": 0.837,
      "step": 18740
    },
    {
      "epoch": 1.2366481441786075,
      "grad_norm": 0.6499251127243042,
      "learning_rate": 1.9097636736118025e-05,
      "loss": 0.8174,
      "step": 18750
    },
    {
      "epoch": 1.2373076986495621,
      "grad_norm": 0.6384209990501404,
      "learning_rate": 1.906967557185374e-05,
      "loss": 0.7907,
      "step": 18760
    },
    {
      "epoch": 1.237967253120517,
      "grad_norm": 0.5927095413208008,
      "learning_rate": 1.904172226608212e-05,
      "loss": 0.7915,
      "step": 18770
    },
    {
      "epoch": 1.238626807591472,
      "grad_norm": 0.6062419414520264,
      "learning_rate": 1.90137768558451e-05,
      "loss": 0.7449,
      "step": 18780
    },
    {
      "epoch": 1.2392863620624268,
      "grad_norm": 0.6219165325164795,
      "learning_rate": 1.898583937817419e-05,
      "loss": 0.7835,
      "step": 18790
    },
    {
      "epoch": 1.2399459165333817,
      "grad_norm": 0.6713700294494629,
      "learning_rate": 1.895790987009038e-05,
      "loss": 0.8172,
      "step": 18800
    },
    {
      "epoch": 1.2406054710043366,
      "grad_norm": 0.5766702890396118,
      "learning_rate": 1.89299883686041e-05,
      "loss": 0.7811,
      "step": 18810
    },
    {
      "epoch": 1.2412650254752915,
      "grad_norm": 0.6060327887535095,
      "learning_rate": 1.8902074910715156e-05,
      "loss": 0.8735,
      "step": 18820
    },
    {
      "epoch": 1.2419245799462464,
      "grad_norm": 0.5441230535507202,
      "learning_rate": 1.8874169533412705e-05,
      "loss": 0.8124,
      "step": 18830
    },
    {
      "epoch": 1.2425841344172013,
      "grad_norm": 0.6223664879798889,
      "learning_rate": 1.8846272273675206e-05,
      "loss": 0.814,
      "step": 18840
    },
    {
      "epoch": 1.243243688888156,
      "grad_norm": 0.5244922041893005,
      "learning_rate": 1.8818383168470353e-05,
      "loss": 0.8526,
      "step": 18850
    },
    {
      "epoch": 1.2439032433591108,
      "grad_norm": 0.6009203791618347,
      "learning_rate": 1.8790502254755023e-05,
      "loss": 0.79,
      "step": 18860
    },
    {
      "epoch": 1.2445627978300657,
      "grad_norm": 0.6063975691795349,
      "learning_rate": 1.876262956947524e-05,
      "loss": 0.8173,
      "step": 18870
    },
    {
      "epoch": 1.2452223523010206,
      "grad_norm": 0.6352492570877075,
      "learning_rate": 1.8734765149566154e-05,
      "loss": 0.8327,
      "step": 18880
    },
    {
      "epoch": 1.2458819067719755,
      "grad_norm": 0.5097781419754028,
      "learning_rate": 1.8706909031951926e-05,
      "loss": 0.8425,
      "step": 18890
    },
    {
      "epoch": 1.2465414612429304,
      "grad_norm": 0.5806178450584412,
      "learning_rate": 1.8679061253545742e-05,
      "loss": 0.8452,
      "step": 18900
    },
    {
      "epoch": 1.2472010157138853,
      "grad_norm": 0.5763145685195923,
      "learning_rate": 1.865122185124971e-05,
      "loss": 0.7994,
      "step": 18910
    },
    {
      "epoch": 1.2478605701848402,
      "grad_norm": 0.558208703994751,
      "learning_rate": 1.862339086195486e-05,
      "loss": 0.8302,
      "step": 18920
    },
    {
      "epoch": 1.248520124655795,
      "grad_norm": 0.5959612727165222,
      "learning_rate": 1.8595568322541067e-05,
      "loss": 0.7906,
      "step": 18930
    },
    {
      "epoch": 1.2491796791267498,
      "grad_norm": 0.5050669312477112,
      "learning_rate": 1.856775426987702e-05,
      "loss": 0.8195,
      "step": 18940
    },
    {
      "epoch": 1.2498392335977047,
      "grad_norm": 0.6480620503425598,
      "learning_rate": 1.8539948740820128e-05,
      "loss": 0.7802,
      "step": 18950
    },
    {
      "epoch": 1.2504987880686596,
      "grad_norm": 0.5807694792747498,
      "learning_rate": 1.8512151772216544e-05,
      "loss": 0.805,
      "step": 18960
    },
    {
      "epoch": 1.2511583425396144,
      "grad_norm": 0.5697056651115417,
      "learning_rate": 1.8484363400901057e-05,
      "loss": 0.7891,
      "step": 18970
    },
    {
      "epoch": 1.2518178970105693,
      "grad_norm": 0.6128772497177124,
      "learning_rate": 1.8456583663697066e-05,
      "loss": 0.8099,
      "step": 18980
    },
    {
      "epoch": 1.2524774514815242,
      "grad_norm": 0.5341264009475708,
      "learning_rate": 1.8428812597416544e-05,
      "loss": 0.8081,
      "step": 18990
    },
    {
      "epoch": 1.2531370059524791,
      "grad_norm": 0.5600847601890564,
      "learning_rate": 1.840105023885994e-05,
      "loss": 0.8213,
      "step": 19000
    },
    {
      "epoch": 1.253796560423434,
      "grad_norm": 0.4721827507019043,
      "learning_rate": 1.837329662481619e-05,
      "loss": 0.7826,
      "step": 19010
    },
    {
      "epoch": 1.254456114894389,
      "grad_norm": 0.5889362692832947,
      "learning_rate": 1.834555179206264e-05,
      "loss": 0.7959,
      "step": 19020
    },
    {
      "epoch": 1.2551156693653436,
      "grad_norm": 0.5458769202232361,
      "learning_rate": 1.8317815777364993e-05,
      "loss": 0.8449,
      "step": 19030
    },
    {
      "epoch": 1.2557752238362987,
      "grad_norm": 0.6288903951644897,
      "learning_rate": 1.8290088617477274e-05,
      "loss": 0.8013,
      "step": 19040
    },
    {
      "epoch": 1.2564347783072534,
      "grad_norm": 0.5463530421257019,
      "learning_rate": 1.8262370349141768e-05,
      "loss": 0.7887,
      "step": 19050
    },
    {
      "epoch": 1.2570943327782083,
      "grad_norm": 0.5617839694023132,
      "learning_rate": 1.8234661009088972e-05,
      "loss": 0.8045,
      "step": 19060
    },
    {
      "epoch": 1.2577538872491631,
      "grad_norm": 0.5814275741577148,
      "learning_rate": 1.8206960634037572e-05,
      "loss": 0.7853,
      "step": 19070
    },
    {
      "epoch": 1.258413441720118,
      "grad_norm": 0.6037193536758423,
      "learning_rate": 1.8179269260694365e-05,
      "loss": 0.8129,
      "step": 19080
    },
    {
      "epoch": 1.259072996191073,
      "grad_norm": 0.5609439611434937,
      "learning_rate": 1.815158692575419e-05,
      "loss": 0.7916,
      "step": 19090
    },
    {
      "epoch": 1.2597325506620278,
      "grad_norm": 0.6056429147720337,
      "learning_rate": 1.8123913665899956e-05,
      "loss": 0.7878,
      "step": 19100
    },
    {
      "epoch": 1.2603921051329827,
      "grad_norm": 0.5459847450256348,
      "learning_rate": 1.8096249517802522e-05,
      "loss": 0.8033,
      "step": 19110
    },
    {
      "epoch": 1.2610516596039376,
      "grad_norm": 0.5837607979774475,
      "learning_rate": 1.8068594518120684e-05,
      "loss": 0.7992,
      "step": 19120
    },
    {
      "epoch": 1.2617112140748925,
      "grad_norm": 0.541070818901062,
      "learning_rate": 1.8040948703501087e-05,
      "loss": 0.8201,
      "step": 19130
    },
    {
      "epoch": 1.2623707685458472,
      "grad_norm": 0.5953345894813538,
      "learning_rate": 1.8013312110578236e-05,
      "loss": 0.8148,
      "step": 19140
    },
    {
      "epoch": 1.263030323016802,
      "grad_norm": 0.593059241771698,
      "learning_rate": 1.7985684775974397e-05,
      "loss": 0.785,
      "step": 19150
    },
    {
      "epoch": 1.263689877487757,
      "grad_norm": 0.5525000691413879,
      "learning_rate": 1.795806673629959e-05,
      "loss": 0.8148,
      "step": 19160
    },
    {
      "epoch": 1.2643494319587119,
      "grad_norm": 0.6783982515335083,
      "learning_rate": 1.793045802815149e-05,
      "loss": 0.8104,
      "step": 19170
    },
    {
      "epoch": 1.2650089864296667,
      "grad_norm": 0.5801666975021362,
      "learning_rate": 1.790285868811542e-05,
      "loss": 0.8218,
      "step": 19180
    },
    {
      "epoch": 1.2656685409006216,
      "grad_norm": 0.5675482749938965,
      "learning_rate": 1.7875268752764296e-05,
      "loss": 0.8307,
      "step": 19190
    },
    {
      "epoch": 1.2663280953715765,
      "grad_norm": 0.5449127554893494,
      "learning_rate": 1.7847688258658552e-05,
      "loss": 0.8229,
      "step": 19200
    },
    {
      "epoch": 1.2669876498425314,
      "grad_norm": 0.6582222580909729,
      "learning_rate": 1.7820117242346144e-05,
      "loss": 0.8356,
      "step": 19210
    },
    {
      "epoch": 1.2676472043134863,
      "grad_norm": 0.5748285055160522,
      "learning_rate": 1.7792555740362422e-05,
      "loss": 0.8247,
      "step": 19220
    },
    {
      "epoch": 1.268306758784441,
      "grad_norm": 0.5172967910766602,
      "learning_rate": 1.776500378923016e-05,
      "loss": 0.7851,
      "step": 19230
    },
    {
      "epoch": 1.268966313255396,
      "grad_norm": 0.5694426894187927,
      "learning_rate": 1.773746142545948e-05,
      "loss": 0.8206,
      "step": 19240
    },
    {
      "epoch": 1.2696258677263508,
      "grad_norm": 0.5869324207305908,
      "learning_rate": 1.7709928685547777e-05,
      "loss": 0.8184,
      "step": 19250
    },
    {
      "epoch": 1.2702854221973057,
      "grad_norm": 0.62547367811203,
      "learning_rate": 1.768240560597972e-05,
      "loss": 0.8131,
      "step": 19260
    },
    {
      "epoch": 1.2709449766682606,
      "grad_norm": 0.5633248090744019,
      "learning_rate": 1.765489222322714e-05,
      "loss": 0.8112,
      "step": 19270
    },
    {
      "epoch": 1.2716045311392155,
      "grad_norm": 0.5703989267349243,
      "learning_rate": 1.762738857374906e-05,
      "loss": 0.8278,
      "step": 19280
    },
    {
      "epoch": 1.2722640856101703,
      "grad_norm": 0.5290929079055786,
      "learning_rate": 1.759989469399157e-05,
      "loss": 0.8018,
      "step": 19290
    },
    {
      "epoch": 1.2729236400811252,
      "grad_norm": 0.6253542900085449,
      "learning_rate": 1.757241062038785e-05,
      "loss": 0.7834,
      "step": 19300
    },
    {
      "epoch": 1.2735831945520801,
      "grad_norm": 0.5754891037940979,
      "learning_rate": 1.7544936389358034e-05,
      "loss": 0.8103,
      "step": 19310
    },
    {
      "epoch": 1.2742427490230348,
      "grad_norm": 0.6191249489784241,
      "learning_rate": 1.7517472037309267e-05,
      "loss": 0.7984,
      "step": 19320
    },
    {
      "epoch": 1.27490230349399,
      "grad_norm": 0.6206302046775818,
      "learning_rate": 1.7490017600635573e-05,
      "loss": 0.7922,
      "step": 19330
    },
    {
      "epoch": 1.2755618579649446,
      "grad_norm": 0.47136321663856506,
      "learning_rate": 1.746257311571785e-05,
      "loss": 0.8027,
      "step": 19340
    },
    {
      "epoch": 1.2762214124358995,
      "grad_norm": 0.6640087962150574,
      "learning_rate": 1.743513861892379e-05,
      "loss": 0.8399,
      "step": 19350
    },
    {
      "epoch": 1.2768809669068544,
      "grad_norm": 0.4717612564563751,
      "learning_rate": 1.7407714146607862e-05,
      "loss": 0.8017,
      "step": 19360
    },
    {
      "epoch": 1.2775405213778093,
      "grad_norm": 0.5602179765701294,
      "learning_rate": 1.7380299735111256e-05,
      "loss": 0.7809,
      "step": 19370
    },
    {
      "epoch": 1.2782000758487642,
      "grad_norm": 0.5533621907234192,
      "learning_rate": 1.735289542076182e-05,
      "loss": 0.7739,
      "step": 19380
    },
    {
      "epoch": 1.278859630319719,
      "grad_norm": 0.5811960697174072,
      "learning_rate": 1.7325501239874037e-05,
      "loss": 0.8069,
      "step": 19390
    },
    {
      "epoch": 1.279519184790674,
      "grad_norm": 0.5109241604804993,
      "learning_rate": 1.7298117228748938e-05,
      "loss": 0.7767,
      "step": 19400
    },
    {
      "epoch": 1.2801787392616288,
      "grad_norm": 0.5800501704216003,
      "learning_rate": 1.72707434236741e-05,
      "loss": 0.8309,
      "step": 19410
    },
    {
      "epoch": 1.2808382937325837,
      "grad_norm": 0.5540037751197815,
      "learning_rate": 1.724337986092356e-05,
      "loss": 0.8158,
      "step": 19420
    },
    {
      "epoch": 1.2814978482035384,
      "grad_norm": 0.6083894968032837,
      "learning_rate": 1.7216026576757805e-05,
      "loss": 0.8392,
      "step": 19430
    },
    {
      "epoch": 1.2821574026744935,
      "grad_norm": 0.5300758481025696,
      "learning_rate": 1.718868360742366e-05,
      "loss": 0.8052,
      "step": 19440
    },
    {
      "epoch": 1.2828169571454482,
      "grad_norm": 0.5607723593711853,
      "learning_rate": 1.716135098915432e-05,
      "loss": 0.825,
      "step": 19450
    },
    {
      "epoch": 1.283476511616403,
      "grad_norm": 0.5369413495063782,
      "learning_rate": 1.7134028758169253e-05,
      "loss": 0.8093,
      "step": 19460
    },
    {
      "epoch": 1.284136066087358,
      "grad_norm": 0.53282630443573,
      "learning_rate": 1.710671695067415e-05,
      "loss": 0.7952,
      "step": 19470
    },
    {
      "epoch": 1.2847956205583129,
      "grad_norm": 0.541825532913208,
      "learning_rate": 1.7079415602860917e-05,
      "loss": 0.7975,
      "step": 19480
    },
    {
      "epoch": 1.2854551750292678,
      "grad_norm": 0.6871068477630615,
      "learning_rate": 1.705212475090756e-05,
      "loss": 0.8133,
      "step": 19490
    },
    {
      "epoch": 1.2861147295002227,
      "grad_norm": 0.5954208374023438,
      "learning_rate": 1.702484443097821e-05,
      "loss": 0.7942,
      "step": 19500
    },
    {
      "epoch": 1.2867742839711775,
      "grad_norm": 0.6732329726219177,
      "learning_rate": 1.699757467922303e-05,
      "loss": 0.7494,
      "step": 19510
    },
    {
      "epoch": 1.2874338384421322,
      "grad_norm": 0.5208237171173096,
      "learning_rate": 1.6970315531778187e-05,
      "loss": 0.8084,
      "step": 19520
    },
    {
      "epoch": 1.2880933929130873,
      "grad_norm": 0.5553262829780579,
      "learning_rate": 1.6943067024765756e-05,
      "loss": 0.8072,
      "step": 19530
    },
    {
      "epoch": 1.288752947384042,
      "grad_norm": 0.6095374226570129,
      "learning_rate": 1.6915829194293777e-05,
      "loss": 0.8065,
      "step": 19540
    },
    {
      "epoch": 1.289412501854997,
      "grad_norm": 0.566206157207489,
      "learning_rate": 1.6888602076456096e-05,
      "loss": 0.8571,
      "step": 19550
    },
    {
      "epoch": 1.2900720563259518,
      "grad_norm": 0.6199831366539001,
      "learning_rate": 1.6861385707332372e-05,
      "loss": 0.795,
      "step": 19560
    },
    {
      "epoch": 1.2907316107969067,
      "grad_norm": 0.5906867980957031,
      "learning_rate": 1.6834180122988043e-05,
      "loss": 0.8285,
      "step": 19570
    },
    {
      "epoch": 1.2913911652678616,
      "grad_norm": 0.5739659070968628,
      "learning_rate": 1.6806985359474206e-05,
      "loss": 0.8048,
      "step": 19580
    },
    {
      "epoch": 1.2920507197388165,
      "grad_norm": 0.5545616745948792,
      "learning_rate": 1.677980145282767e-05,
      "loss": 0.8139,
      "step": 19590
    },
    {
      "epoch": 1.2927102742097714,
      "grad_norm": 0.5878023505210876,
      "learning_rate": 1.675262843907083e-05,
      "loss": 0.8206,
      "step": 19600
    },
    {
      "epoch": 1.2933698286807263,
      "grad_norm": 0.6396400332450867,
      "learning_rate": 1.6725466354211663e-05,
      "loss": 0.8593,
      "step": 19610
    },
    {
      "epoch": 1.2940293831516811,
      "grad_norm": 0.5396580100059509,
      "learning_rate": 1.6698315234243645e-05,
      "loss": 0.7799,
      "step": 19620
    },
    {
      "epoch": 1.2946889376226358,
      "grad_norm": 0.5915452241897583,
      "learning_rate": 1.667117511514573e-05,
      "loss": 0.7915,
      "step": 19630
    },
    {
      "epoch": 1.2953484920935907,
      "grad_norm": 0.5672121644020081,
      "learning_rate": 1.6644046032882298e-05,
      "loss": 0.7882,
      "step": 19640
    },
    {
      "epoch": 1.2960080465645456,
      "grad_norm": 0.7311338186264038,
      "learning_rate": 1.6616928023403107e-05,
      "loss": 0.7996,
      "step": 19650
    },
    {
      "epoch": 1.2966676010355005,
      "grad_norm": 0.5876666307449341,
      "learning_rate": 1.6589821122643238e-05,
      "loss": 0.8324,
      "step": 19660
    },
    {
      "epoch": 1.2973271555064554,
      "grad_norm": 0.6038510799407959,
      "learning_rate": 1.6562725366523037e-05,
      "loss": 0.8012,
      "step": 19670
    },
    {
      "epoch": 1.2979867099774103,
      "grad_norm": 0.5467290282249451,
      "learning_rate": 1.653564079094809e-05,
      "loss": 0.7902,
      "step": 19680
    },
    {
      "epoch": 1.2986462644483652,
      "grad_norm": 0.7234504818916321,
      "learning_rate": 1.6508567431809197e-05,
      "loss": 0.8189,
      "step": 19690
    },
    {
      "epoch": 1.29930581891932,
      "grad_norm": 0.5546754598617554,
      "learning_rate": 1.648150532498225e-05,
      "loss": 0.7958,
      "step": 19700
    },
    {
      "epoch": 1.299965373390275,
      "grad_norm": 0.5073449611663818,
      "learning_rate": 1.6454454506328253e-05,
      "loss": 0.8478,
      "step": 19710
    },
    {
      "epoch": 1.3006249278612296,
      "grad_norm": 0.6308967471122742,
      "learning_rate": 1.6427415011693248e-05,
      "loss": 0.7708,
      "step": 19720
    },
    {
      "epoch": 1.3012844823321847,
      "grad_norm": 0.5615613460540771,
      "learning_rate": 1.6400386876908275e-05,
      "loss": 0.7973,
      "step": 19730
    },
    {
      "epoch": 1.3019440368031394,
      "grad_norm": 0.584670901298523,
      "learning_rate": 1.6373370137789312e-05,
      "loss": 0.8181,
      "step": 19740
    },
    {
      "epoch": 1.3026035912740943,
      "grad_norm": 0.5000048279762268,
      "learning_rate": 1.6346364830137254e-05,
      "loss": 0.8122,
      "step": 19750
    },
    {
      "epoch": 1.3032631457450492,
      "grad_norm": 0.6242878437042236,
      "learning_rate": 1.6319370989737815e-05,
      "loss": 0.7971,
      "step": 19760
    },
    {
      "epoch": 1.303922700216004,
      "grad_norm": 0.5634641647338867,
      "learning_rate": 1.6292388652361544e-05,
      "loss": 0.8355,
      "step": 19770
    },
    {
      "epoch": 1.304582254686959,
      "grad_norm": 0.6160987615585327,
      "learning_rate": 1.6265417853763737e-05,
      "loss": 0.773,
      "step": 19780
    },
    {
      "epoch": 1.3052418091579139,
      "grad_norm": 0.5563802123069763,
      "learning_rate": 1.6238458629684407e-05,
      "loss": 0.842,
      "step": 19790
    },
    {
      "epoch": 1.3059013636288688,
      "grad_norm": 0.5255377292633057,
      "learning_rate": 1.6211511015848195e-05,
      "loss": 0.8243,
      "step": 19800
    },
    {
      "epoch": 1.3065609180998234,
      "grad_norm": 0.4894976317882538,
      "learning_rate": 1.6184575047964405e-05,
      "loss": 0.764,
      "step": 19810
    },
    {
      "epoch": 1.3072204725707786,
      "grad_norm": 0.6574572920799255,
      "learning_rate": 1.615765076172688e-05,
      "loss": 0.776,
      "step": 19820
    },
    {
      "epoch": 1.3078800270417332,
      "grad_norm": 0.5359966158866882,
      "learning_rate": 1.613073819281398e-05,
      "loss": 0.778,
      "step": 19830
    },
    {
      "epoch": 1.3085395815126881,
      "grad_norm": 0.6031800508499146,
      "learning_rate": 1.610383737688857e-05,
      "loss": 0.7966,
      "step": 19840
    },
    {
      "epoch": 1.309199135983643,
      "grad_norm": 0.5762466788291931,
      "learning_rate": 1.6076948349597894e-05,
      "loss": 0.8185,
      "step": 19850
    },
    {
      "epoch": 1.309858690454598,
      "grad_norm": 0.5590838193893433,
      "learning_rate": 1.605007114657361e-05,
      "loss": 0.8083,
      "step": 19860
    },
    {
      "epoch": 1.3105182449255528,
      "grad_norm": 0.6155121922492981,
      "learning_rate": 1.6023205803431696e-05,
      "loss": 0.8529,
      "step": 19870
    },
    {
      "epoch": 1.3111777993965077,
      "grad_norm": 0.6039925813674927,
      "learning_rate": 1.5996352355772422e-05,
      "loss": 0.7866,
      "step": 19880
    },
    {
      "epoch": 1.3118373538674626,
      "grad_norm": 0.6229190230369568,
      "learning_rate": 1.596951083918026e-05,
      "loss": 0.8194,
      "step": 19890
    },
    {
      "epoch": 1.3124969083384175,
      "grad_norm": 0.6370930671691895,
      "learning_rate": 1.5942681289223923e-05,
      "loss": 0.8628,
      "step": 19900
    },
    {
      "epoch": 1.3131564628093724,
      "grad_norm": 0.6257620453834534,
      "learning_rate": 1.5915863741456233e-05,
      "loss": 0.7803,
      "step": 19910
    },
    {
      "epoch": 1.313816017280327,
      "grad_norm": 0.5882974863052368,
      "learning_rate": 1.5889058231414116e-05,
      "loss": 0.7776,
      "step": 19920
    },
    {
      "epoch": 1.314475571751282,
      "grad_norm": 0.5556590557098389,
      "learning_rate": 1.586226479461855e-05,
      "loss": 0.7769,
      "step": 19930
    },
    {
      "epoch": 1.3151351262222368,
      "grad_norm": 0.5639545321464539,
      "learning_rate": 1.5835483466574504e-05,
      "loss": 0.7975,
      "step": 19940
    },
    {
      "epoch": 1.3157946806931917,
      "grad_norm": 0.5980276465415955,
      "learning_rate": 1.580871428277091e-05,
      "loss": 0.8253,
      "step": 19950
    },
    {
      "epoch": 1.3164542351641466,
      "grad_norm": 0.6568490862846375,
      "learning_rate": 1.578195727868061e-05,
      "loss": 0.8241,
      "step": 19960
    },
    {
      "epoch": 1.3171137896351015,
      "grad_norm": 0.5813195109367371,
      "learning_rate": 1.5755212489760305e-05,
      "loss": 0.8222,
      "step": 19970
    },
    {
      "epoch": 1.3177733441060564,
      "grad_norm": 0.626581072807312,
      "learning_rate": 1.5728479951450496e-05,
      "loss": 0.7894,
      "step": 19980
    },
    {
      "epoch": 1.3184328985770113,
      "grad_norm": 0.6403607726097107,
      "learning_rate": 1.5701759699175466e-05,
      "loss": 0.8277,
      "step": 19990
    },
    {
      "epoch": 1.3190924530479662,
      "grad_norm": 0.5415562391281128,
      "learning_rate": 1.567505176834321e-05,
      "loss": 0.809,
      "step": 20000
    },
    {
      "epoch": 1.3197520075189209,
      "grad_norm": 0.5711784958839417,
      "learning_rate": 1.5648356194345408e-05,
      "loss": 0.8135,
      "step": 20010
    },
    {
      "epoch": 1.320411561989876,
      "grad_norm": 0.5610942840576172,
      "learning_rate": 1.562167301255733e-05,
      "loss": 0.778,
      "step": 20020
    },
    {
      "epoch": 1.3210711164608306,
      "grad_norm": 0.6548299789428711,
      "learning_rate": 1.5595002258337864e-05,
      "loss": 0.7675,
      "step": 20030
    },
    {
      "epoch": 1.3217306709317855,
      "grad_norm": 0.6827725768089294,
      "learning_rate": 1.5568343967029414e-05,
      "loss": 0.8149,
      "step": 20040
    },
    {
      "epoch": 1.3223902254027404,
      "grad_norm": 0.6757158637046814,
      "learning_rate": 1.5541698173957865e-05,
      "loss": 0.8296,
      "step": 20050
    },
    {
      "epoch": 1.3230497798736953,
      "grad_norm": 0.5480896830558777,
      "learning_rate": 1.5515064914432553e-05,
      "loss": 0.8166,
      "step": 20060
    },
    {
      "epoch": 1.3237093343446502,
      "grad_norm": 0.5392537117004395,
      "learning_rate": 1.5488444223746192e-05,
      "loss": 0.8068,
      "step": 20070
    },
    {
      "epoch": 1.324368888815605,
      "grad_norm": 0.6353445649147034,
      "learning_rate": 1.546183613717484e-05,
      "loss": 0.7939,
      "step": 20080
    },
    {
      "epoch": 1.32502844328656,
      "grad_norm": 0.6121338605880737,
      "learning_rate": 1.543524068997787e-05,
      "loss": 0.7803,
      "step": 20090
    },
    {
      "epoch": 1.3256879977575147,
      "grad_norm": 0.6256228685379028,
      "learning_rate": 1.54086579173979e-05,
      "loss": 0.8092,
      "step": 20100
    },
    {
      "epoch": 1.3263475522284698,
      "grad_norm": 0.6484335660934448,
      "learning_rate": 1.538208785466072e-05,
      "loss": 0.7915,
      "step": 20110
    },
    {
      "epoch": 1.3270071066994245,
      "grad_norm": 0.6085736751556396,
      "learning_rate": 1.5355530536975325e-05,
      "loss": 0.8024,
      "step": 20120
    },
    {
      "epoch": 1.3276666611703793,
      "grad_norm": 0.5266491174697876,
      "learning_rate": 1.5328985999533797e-05,
      "loss": 0.8293,
      "step": 20130
    },
    {
      "epoch": 1.3283262156413342,
      "grad_norm": 0.6135145425796509,
      "learning_rate": 1.5302454277511286e-05,
      "loss": 0.8145,
      "step": 20140
    },
    {
      "epoch": 1.3289857701122891,
      "grad_norm": 0.6410892605781555,
      "learning_rate": 1.527593540606596e-05,
      "loss": 0.8378,
      "step": 20150
    },
    {
      "epoch": 1.329645324583244,
      "grad_norm": 0.49080774188041687,
      "learning_rate": 1.524942942033896e-05,
      "loss": 0.8334,
      "step": 20160
    },
    {
      "epoch": 1.330304879054199,
      "grad_norm": 0.5814183354377747,
      "learning_rate": 1.5222936355454342e-05,
      "loss": 0.8157,
      "step": 20170
    },
    {
      "epoch": 1.3309644335251538,
      "grad_norm": 0.585276186466217,
      "learning_rate": 1.5196456246519058e-05,
      "loss": 0.7868,
      "step": 20180
    },
    {
      "epoch": 1.3316239879961087,
      "grad_norm": 0.5762225389480591,
      "learning_rate": 1.516998912862288e-05,
      "loss": 0.8178,
      "step": 20190
    },
    {
      "epoch": 1.3322835424670636,
      "grad_norm": 0.5100105404853821,
      "learning_rate": 1.514353503683836e-05,
      "loss": 0.7861,
      "step": 20200
    },
    {
      "epoch": 1.3329430969380183,
      "grad_norm": 0.6492648720741272,
      "learning_rate": 1.51170940062208e-05,
      "loss": 0.7841,
      "step": 20210
    },
    {
      "epoch": 1.3336026514089732,
      "grad_norm": 0.6658373475074768,
      "learning_rate": 1.509066607180819e-05,
      "loss": 0.8538,
      "step": 20220
    },
    {
      "epoch": 1.334262205879928,
      "grad_norm": 0.5783854126930237,
      "learning_rate": 1.506425126862116e-05,
      "loss": 0.7978,
      "step": 20230
    },
    {
      "epoch": 1.334921760350883,
      "grad_norm": 0.5754417777061462,
      "learning_rate": 1.5037849631662961e-05,
      "loss": 0.7594,
      "step": 20240
    },
    {
      "epoch": 1.3355813148218378,
      "grad_norm": 0.5717854499816895,
      "learning_rate": 1.5011461195919358e-05,
      "loss": 0.8123,
      "step": 20250
    },
    {
      "epoch": 1.3362408692927927,
      "grad_norm": 0.6231403946876526,
      "learning_rate": 1.4985085996358656e-05,
      "loss": 0.8193,
      "step": 20260
    },
    {
      "epoch": 1.3369004237637476,
      "grad_norm": 0.6809998154640198,
      "learning_rate": 1.4958724067931607e-05,
      "loss": 0.8199,
      "step": 20270
    },
    {
      "epoch": 1.3375599782347025,
      "grad_norm": 0.5629172921180725,
      "learning_rate": 1.493237544557139e-05,
      "loss": 0.8214,
      "step": 20280
    },
    {
      "epoch": 1.3382195327056574,
      "grad_norm": 0.538102924823761,
      "learning_rate": 1.4906040164193519e-05,
      "loss": 0.8026,
      "step": 20290
    },
    {
      "epoch": 1.338879087176612,
      "grad_norm": 0.5235173106193542,
      "learning_rate": 1.4879718258695868e-05,
      "loss": 0.7747,
      "step": 20300
    },
    {
      "epoch": 1.3395386416475672,
      "grad_norm": 0.5327746868133545,
      "learning_rate": 1.4853409763958565e-05,
      "loss": 0.8357,
      "step": 20310
    },
    {
      "epoch": 1.3401981961185219,
      "grad_norm": 0.6447291970252991,
      "learning_rate": 1.482711471484397e-05,
      "loss": 0.7684,
      "step": 20320
    },
    {
      "epoch": 1.3408577505894768,
      "grad_norm": 0.6309868693351746,
      "learning_rate": 1.480083314619664e-05,
      "loss": 0.7857,
      "step": 20330
    },
    {
      "epoch": 1.3415173050604317,
      "grad_norm": 0.5822016596794128,
      "learning_rate": 1.4774565092843234e-05,
      "loss": 0.7944,
      "step": 20340
    },
    {
      "epoch": 1.3421768595313865,
      "grad_norm": 0.6021371483802795,
      "learning_rate": 1.474831058959253e-05,
      "loss": 0.8116,
      "step": 20350
    },
    {
      "epoch": 1.3428364140023414,
      "grad_norm": 0.6604745984077454,
      "learning_rate": 1.472206967123535e-05,
      "loss": 0.798,
      "step": 20360
    },
    {
      "epoch": 1.3434959684732963,
      "grad_norm": 0.5935235619544983,
      "learning_rate": 1.4695842372544505e-05,
      "loss": 0.7998,
      "step": 20370
    },
    {
      "epoch": 1.3441555229442512,
      "grad_norm": 0.6029685139656067,
      "learning_rate": 1.4669628728274753e-05,
      "loss": 0.8018,
      "step": 20380
    },
    {
      "epoch": 1.344815077415206,
      "grad_norm": 0.6155779361724854,
      "learning_rate": 1.4643428773162771e-05,
      "loss": 0.7862,
      "step": 20390
    },
    {
      "epoch": 1.345474631886161,
      "grad_norm": 0.697088360786438,
      "learning_rate": 1.461724254192709e-05,
      "loss": 0.7783,
      "step": 20400
    },
    {
      "epoch": 1.3461341863571157,
      "grad_norm": 0.620465874671936,
      "learning_rate": 1.4591070069268065e-05,
      "loss": 0.7687,
      "step": 20410
    },
    {
      "epoch": 1.3467937408280706,
      "grad_norm": 0.6178200244903564,
      "learning_rate": 1.4564911389867785e-05,
      "loss": 0.8391,
      "step": 20420
    },
    {
      "epoch": 1.3474532952990255,
      "grad_norm": 0.6703351736068726,
      "learning_rate": 1.4538766538390114e-05,
      "loss": 0.8219,
      "step": 20430
    },
    {
      "epoch": 1.3481128497699804,
      "grad_norm": 0.6984212398529053,
      "learning_rate": 1.4512635549480535e-05,
      "loss": 0.803,
      "step": 20440
    },
    {
      "epoch": 1.3487724042409353,
      "grad_norm": 0.5434896945953369,
      "learning_rate": 1.4486518457766219e-05,
      "loss": 0.7914,
      "step": 20450
    },
    {
      "epoch": 1.3494319587118901,
      "grad_norm": 0.6022245287895203,
      "learning_rate": 1.4460415297855872e-05,
      "loss": 0.7914,
      "step": 20460
    },
    {
      "epoch": 1.350091513182845,
      "grad_norm": 0.624377429485321,
      "learning_rate": 1.443432610433976e-05,
      "loss": 0.8008,
      "step": 20470
    },
    {
      "epoch": 1.3507510676538,
      "grad_norm": 0.5371820330619812,
      "learning_rate": 1.440825091178965e-05,
      "loss": 0.7979,
      "step": 20480
    },
    {
      "epoch": 1.3514106221247548,
      "grad_norm": 0.5285878777503967,
      "learning_rate": 1.4382189754758732e-05,
      "loss": 0.8222,
      "step": 20490
    },
    {
      "epoch": 1.3520701765957095,
      "grad_norm": 0.6214131712913513,
      "learning_rate": 1.4356142667781636e-05,
      "loss": 0.8126,
      "step": 20500
    },
    {
      "epoch": 1.3527297310666644,
      "grad_norm": 0.6240976452827454,
      "learning_rate": 1.4330109685374288e-05,
      "loss": 0.8467,
      "step": 20510
    },
    {
      "epoch": 1.3533892855376193,
      "grad_norm": 0.5760096311569214,
      "learning_rate": 1.4304090842033984e-05,
      "loss": 0.8397,
      "step": 20520
    },
    {
      "epoch": 1.3540488400085742,
      "grad_norm": 0.5903918147087097,
      "learning_rate": 1.4278086172239235e-05,
      "loss": 0.8482,
      "step": 20530
    },
    {
      "epoch": 1.354708394479529,
      "grad_norm": 0.5928918719291687,
      "learning_rate": 1.425209571044982e-05,
      "loss": 0.8182,
      "step": 20540
    },
    {
      "epoch": 1.355367948950484,
      "grad_norm": 0.5865376591682434,
      "learning_rate": 1.4226119491106648e-05,
      "loss": 0.8207,
      "step": 20550
    },
    {
      "epoch": 1.3560275034214389,
      "grad_norm": 0.6136580109596252,
      "learning_rate": 1.420015754863176e-05,
      "loss": 0.8,
      "step": 20560
    },
    {
      "epoch": 1.3566870578923937,
      "grad_norm": 0.5566912293434143,
      "learning_rate": 1.4174209917428306e-05,
      "loss": 0.7866,
      "step": 20570
    },
    {
      "epoch": 1.3573466123633486,
      "grad_norm": 0.6884405016899109,
      "learning_rate": 1.4148276631880437e-05,
      "loss": 0.8446,
      "step": 20580
    },
    {
      "epoch": 1.3580061668343033,
      "grad_norm": 0.5476433038711548,
      "learning_rate": 1.412235772635334e-05,
      "loss": 0.7751,
      "step": 20590
    },
    {
      "epoch": 1.3586657213052584,
      "grad_norm": 0.5802808403968811,
      "learning_rate": 1.4096453235193076e-05,
      "loss": 0.799,
      "step": 20600
    },
    {
      "epoch": 1.359325275776213,
      "grad_norm": 0.5228874683380127,
      "learning_rate": 1.4070563192726671e-05,
      "loss": 0.8439,
      "step": 20610
    },
    {
      "epoch": 1.359984830247168,
      "grad_norm": 0.5136884450912476,
      "learning_rate": 1.4044687633261983e-05,
      "loss": 0.7783,
      "step": 20620
    },
    {
      "epoch": 1.3606443847181229,
      "grad_norm": 0.5910024642944336,
      "learning_rate": 1.4018826591087658e-05,
      "loss": 0.7912,
      "step": 20630
    },
    {
      "epoch": 1.3613039391890778,
      "grad_norm": 0.5563416481018066,
      "learning_rate": 1.3992980100473146e-05,
      "loss": 0.7553,
      "step": 20640
    },
    {
      "epoch": 1.3619634936600327,
      "grad_norm": 0.6473105549812317,
      "learning_rate": 1.3967148195668562e-05,
      "loss": 0.8117,
      "step": 20650
    },
    {
      "epoch": 1.3626230481309876,
      "grad_norm": 0.5593595504760742,
      "learning_rate": 1.3941330910904737e-05,
      "loss": 0.8115,
      "step": 20660
    },
    {
      "epoch": 1.3632826026019425,
      "grad_norm": 0.6227688789367676,
      "learning_rate": 1.3915528280393097e-05,
      "loss": 0.7801,
      "step": 20670
    },
    {
      "epoch": 1.3639421570728971,
      "grad_norm": 0.6196003556251526,
      "learning_rate": 1.3889740338325686e-05,
      "loss": 0.8236,
      "step": 20680
    },
    {
      "epoch": 1.3646017115438522,
      "grad_norm": 0.5288597941398621,
      "learning_rate": 1.3863967118875048e-05,
      "loss": 0.7288,
      "step": 20690
    },
    {
      "epoch": 1.365261266014807,
      "grad_norm": 0.5875284075737,
      "learning_rate": 1.3838208656194223e-05,
      "loss": 0.7876,
      "step": 20700
    },
    {
      "epoch": 1.3659208204857618,
      "grad_norm": 0.5908956527709961,
      "learning_rate": 1.3812464984416726e-05,
      "loss": 0.796,
      "step": 20710
    },
    {
      "epoch": 1.3665803749567167,
      "grad_norm": 0.5640059113502502,
      "learning_rate": 1.3786736137656425e-05,
      "loss": 0.8001,
      "step": 20720
    },
    {
      "epoch": 1.3672399294276716,
      "grad_norm": 0.6048946380615234,
      "learning_rate": 1.3761022150007605e-05,
      "loss": 0.8347,
      "step": 20730
    },
    {
      "epoch": 1.3678994838986265,
      "grad_norm": 0.5536830425262451,
      "learning_rate": 1.373532305554478e-05,
      "loss": 0.8168,
      "step": 20740
    },
    {
      "epoch": 1.3685590383695814,
      "grad_norm": 0.6291602253913879,
      "learning_rate": 1.37096388883228e-05,
      "loss": 0.7887,
      "step": 20750
    },
    {
      "epoch": 1.3692185928405363,
      "grad_norm": 0.6404354572296143,
      "learning_rate": 1.3683969682376693e-05,
      "loss": 0.7914,
      "step": 20760
    },
    {
      "epoch": 1.3698781473114912,
      "grad_norm": 0.5645378232002258,
      "learning_rate": 1.3658315471721688e-05,
      "loss": 0.8385,
      "step": 20770
    },
    {
      "epoch": 1.370537701782446,
      "grad_norm": 0.6588670611381531,
      "learning_rate": 1.363267629035313e-05,
      "loss": 0.8016,
      "step": 20780
    },
    {
      "epoch": 1.3711972562534007,
      "grad_norm": 0.6370968222618103,
      "learning_rate": 1.3607052172246426e-05,
      "loss": 0.7955,
      "step": 20790
    },
    {
      "epoch": 1.3718568107243558,
      "grad_norm": 0.5848293900489807,
      "learning_rate": 1.3581443151357074e-05,
      "loss": 0.8195,
      "step": 20800
    },
    {
      "epoch": 1.3725163651953105,
      "grad_norm": 0.5646858811378479,
      "learning_rate": 1.3555849261620512e-05,
      "loss": 0.7744,
      "step": 20810
    },
    {
      "epoch": 1.3731759196662654,
      "grad_norm": 0.5530049800872803,
      "learning_rate": 1.3530270536952172e-05,
      "loss": 0.791,
      "step": 20820
    },
    {
      "epoch": 1.3738354741372203,
      "grad_norm": 0.5670651793479919,
      "learning_rate": 1.3504707011247365e-05,
      "loss": 0.8341,
      "step": 20830
    },
    {
      "epoch": 1.3744950286081752,
      "grad_norm": 0.5433035492897034,
      "learning_rate": 1.3479158718381246e-05,
      "loss": 0.8089,
      "step": 20840
    },
    {
      "epoch": 1.37515458307913,
      "grad_norm": 0.5983085632324219,
      "learning_rate": 1.3453625692208832e-05,
      "loss": 0.8004,
      "step": 20850
    },
    {
      "epoch": 1.375814137550085,
      "grad_norm": 0.5690645575523376,
      "learning_rate": 1.342810796656487e-05,
      "loss": 0.7438,
      "step": 20860
    },
    {
      "epoch": 1.3764736920210399,
      "grad_norm": 0.5865938663482666,
      "learning_rate": 1.3402605575263845e-05,
      "loss": 0.7654,
      "step": 20870
    },
    {
      "epoch": 1.3771332464919945,
      "grad_norm": 0.5614089369773865,
      "learning_rate": 1.3377118552099912e-05,
      "loss": 0.7979,
      "step": 20880
    },
    {
      "epoch": 1.3777928009629496,
      "grad_norm": 0.6096092462539673,
      "learning_rate": 1.3351646930846886e-05,
      "loss": 0.7826,
      "step": 20890
    },
    {
      "epoch": 1.3784523554339043,
      "grad_norm": 0.5356308221817017,
      "learning_rate": 1.332619074525815e-05,
      "loss": 0.8327,
      "step": 20900
    },
    {
      "epoch": 1.3791119099048592,
      "grad_norm": 0.49735626578330994,
      "learning_rate": 1.3300750029066628e-05,
      "loss": 0.8438,
      "step": 20910
    },
    {
      "epoch": 1.379771464375814,
      "grad_norm": 0.5439593195915222,
      "learning_rate": 1.3275324815984769e-05,
      "loss": 0.8126,
      "step": 20920
    },
    {
      "epoch": 1.380431018846769,
      "grad_norm": 0.5968343019485474,
      "learning_rate": 1.3249915139704455e-05,
      "loss": 0.7955,
      "step": 20930
    },
    {
      "epoch": 1.381090573317724,
      "grad_norm": 0.5874291658401489,
      "learning_rate": 1.3224521033897003e-05,
      "loss": 0.7936,
      "step": 20940
    },
    {
      "epoch": 1.3817501277886788,
      "grad_norm": 0.5604689121246338,
      "learning_rate": 1.3199142532213076e-05,
      "loss": 0.7942,
      "step": 20950
    },
    {
      "epoch": 1.3824096822596337,
      "grad_norm": 0.5957276225090027,
      "learning_rate": 1.3173779668282671e-05,
      "loss": 0.8042,
      "step": 20960
    },
    {
      "epoch": 1.3830692367305886,
      "grad_norm": 0.6727865934371948,
      "learning_rate": 1.3148432475715042e-05,
      "loss": 0.812,
      "step": 20970
    },
    {
      "epoch": 1.3837287912015435,
      "grad_norm": 0.6120534539222717,
      "learning_rate": 1.3123100988098714e-05,
      "loss": 0.8201,
      "step": 20980
    },
    {
      "epoch": 1.3843883456724981,
      "grad_norm": 0.5892711877822876,
      "learning_rate": 1.3097785239001375e-05,
      "loss": 0.7579,
      "step": 20990
    },
    {
      "epoch": 1.385047900143453,
      "grad_norm": 0.5758843421936035,
      "learning_rate": 1.3072485261969846e-05,
      "loss": 0.8137,
      "step": 21000
    },
    {
      "epoch": 1.385707454614408,
      "grad_norm": 0.66371750831604,
      "learning_rate": 1.3047201090530089e-05,
      "loss": 0.8254,
      "step": 21010
    },
    {
      "epoch": 1.3863670090853628,
      "grad_norm": 0.5467724204063416,
      "learning_rate": 1.3021932758187073e-05,
      "loss": 0.7762,
      "step": 21020
    },
    {
      "epoch": 1.3870265635563177,
      "grad_norm": 0.5854601860046387,
      "learning_rate": 1.2996680298424824e-05,
      "loss": 0.8477,
      "step": 21030
    },
    {
      "epoch": 1.3876861180272726,
      "grad_norm": 0.5929408669471741,
      "learning_rate": 1.2971443744706302e-05,
      "loss": 0.8353,
      "step": 21040
    },
    {
      "epoch": 1.3883456724982275,
      "grad_norm": 0.5941386818885803,
      "learning_rate": 1.2946223130473389e-05,
      "loss": 0.7822,
      "step": 21050
    },
    {
      "epoch": 1.3890052269691824,
      "grad_norm": 0.552693247795105,
      "learning_rate": 1.2921018489146875e-05,
      "loss": 0.8083,
      "step": 21060
    },
    {
      "epoch": 1.3896647814401373,
      "grad_norm": 0.5452876091003418,
      "learning_rate": 1.289582985412634e-05,
      "loss": 0.8044,
      "step": 21070
    },
    {
      "epoch": 1.390324335911092,
      "grad_norm": 0.6474664211273193,
      "learning_rate": 1.2870657258790217e-05,
      "loss": 0.801,
      "step": 21080
    },
    {
      "epoch": 1.390983890382047,
      "grad_norm": 0.5309171080589294,
      "learning_rate": 1.2845500736495597e-05,
      "loss": 0.7875,
      "step": 21090
    },
    {
      "epoch": 1.3916434448530017,
      "grad_norm": 0.5736203193664551,
      "learning_rate": 1.2820360320578356e-05,
      "loss": 0.8088,
      "step": 21100
    },
    {
      "epoch": 1.3923029993239566,
      "grad_norm": 0.7004844546318054,
      "learning_rate": 1.2795236044352965e-05,
      "loss": 0.826,
      "step": 21110
    },
    {
      "epoch": 1.3929625537949115,
      "grad_norm": 0.6375095248222351,
      "learning_rate": 1.2770127941112555e-05,
      "loss": 0.8044,
      "step": 21120
    },
    {
      "epoch": 1.3936221082658664,
      "grad_norm": 0.5890763401985168,
      "learning_rate": 1.2745036044128796e-05,
      "loss": 0.8475,
      "step": 21130
    },
    {
      "epoch": 1.3942816627368213,
      "grad_norm": 0.6262532472610474,
      "learning_rate": 1.271996038665188e-05,
      "loss": 0.8054,
      "step": 21140
    },
    {
      "epoch": 1.3949412172077762,
      "grad_norm": 0.6479079723358154,
      "learning_rate": 1.2694901001910509e-05,
      "loss": 0.8084,
      "step": 21150
    },
    {
      "epoch": 1.395600771678731,
      "grad_norm": 0.544293999671936,
      "learning_rate": 1.2669857923111782e-05,
      "loss": 0.8119,
      "step": 21160
    },
    {
      "epoch": 1.3962603261496858,
      "grad_norm": 0.601423442363739,
      "learning_rate": 1.2644831183441244e-05,
      "loss": 0.8142,
      "step": 21170
    },
    {
      "epoch": 1.3969198806206409,
      "grad_norm": 0.6186248660087585,
      "learning_rate": 1.2619820816062711e-05,
      "loss": 0.8067,
      "step": 21180
    },
    {
      "epoch": 1.3975794350915955,
      "grad_norm": 0.5760650634765625,
      "learning_rate": 1.259482685411838e-05,
      "loss": 0.7817,
      "step": 21190
    },
    {
      "epoch": 1.3982389895625504,
      "grad_norm": 0.5994721055030823,
      "learning_rate": 1.256984933072865e-05,
      "loss": 0.8261,
      "step": 21200
    },
    {
      "epoch": 1.3988985440335053,
      "grad_norm": 0.605767548084259,
      "learning_rate": 1.2544888278992195e-05,
      "loss": 0.8046,
      "step": 21210
    },
    {
      "epoch": 1.3995580985044602,
      "grad_norm": 0.5736032128334045,
      "learning_rate": 1.2519943731985806e-05,
      "loss": 0.7868,
      "step": 21220
    },
    {
      "epoch": 1.4002176529754151,
      "grad_norm": 0.595744788646698,
      "learning_rate": 1.2495015722764428e-05,
      "loss": 0.8441,
      "step": 21230
    },
    {
      "epoch": 1.40087720744637,
      "grad_norm": 0.6784983277320862,
      "learning_rate": 1.2470104284361104e-05,
      "loss": 0.7841,
      "step": 21240
    },
    {
      "epoch": 1.401536761917325,
      "grad_norm": 0.5943467617034912,
      "learning_rate": 1.2445209449786884e-05,
      "loss": 0.7988,
      "step": 21250
    },
    {
      "epoch": 1.4021963163882798,
      "grad_norm": 0.5512073636054993,
      "learning_rate": 1.2420331252030855e-05,
      "loss": 0.7966,
      "step": 21260
    },
    {
      "epoch": 1.4028558708592347,
      "grad_norm": 0.6070830225944519,
      "learning_rate": 1.2395469724060035e-05,
      "loss": 0.8414,
      "step": 21270
    },
    {
      "epoch": 1.4035154253301894,
      "grad_norm": 0.5669506192207336,
      "learning_rate": 1.237062489881934e-05,
      "loss": 0.8604,
      "step": 21280
    },
    {
      "epoch": 1.4041749798011443,
      "grad_norm": 0.5757513642311096,
      "learning_rate": 1.2345796809231588e-05,
      "loss": 0.8134,
      "step": 21290
    },
    {
      "epoch": 1.4048345342720991,
      "grad_norm": 0.5894919633865356,
      "learning_rate": 1.2320985488197383e-05,
      "loss": 0.8274,
      "step": 21300
    },
    {
      "epoch": 1.405494088743054,
      "grad_norm": 0.6311694979667664,
      "learning_rate": 1.2296190968595155e-05,
      "loss": 0.7954,
      "step": 21310
    },
    {
      "epoch": 1.406153643214009,
      "grad_norm": 0.6621250510215759,
      "learning_rate": 1.2271413283281006e-05,
      "loss": 0.7888,
      "step": 21320
    },
    {
      "epoch": 1.4068131976849638,
      "grad_norm": 0.6430502533912659,
      "learning_rate": 1.2246652465088784e-05,
      "loss": 0.8383,
      "step": 21330
    },
    {
      "epoch": 1.4074727521559187,
      "grad_norm": 0.6386721134185791,
      "learning_rate": 1.2221908546829957e-05,
      "loss": 0.8225,
      "step": 21340
    },
    {
      "epoch": 1.4081323066268736,
      "grad_norm": 0.5988573431968689,
      "learning_rate": 1.2197181561293622e-05,
      "loss": 0.816,
      "step": 21350
    },
    {
      "epoch": 1.4087918610978285,
      "grad_norm": 0.6191495060920715,
      "learning_rate": 1.2172471541246416e-05,
      "loss": 0.7815,
      "step": 21360
    },
    {
      "epoch": 1.4094514155687832,
      "grad_norm": 0.6011860966682434,
      "learning_rate": 1.2147778519432493e-05,
      "loss": 0.8307,
      "step": 21370
    },
    {
      "epoch": 1.4101109700397383,
      "grad_norm": 0.5985183715820312,
      "learning_rate": 1.2123102528573507e-05,
      "loss": 0.803,
      "step": 21380
    },
    {
      "epoch": 1.410770524510693,
      "grad_norm": 0.5957480072975159,
      "learning_rate": 1.2098443601368526e-05,
      "loss": 0.8124,
      "step": 21390
    },
    {
      "epoch": 1.4114300789816479,
      "grad_norm": 0.6131159067153931,
      "learning_rate": 1.2073801770493997e-05,
      "loss": 0.8531,
      "step": 21400
    },
    {
      "epoch": 1.4120896334526027,
      "grad_norm": 0.5503156781196594,
      "learning_rate": 1.2049177068603723e-05,
      "loss": 0.8175,
      "step": 21410
    },
    {
      "epoch": 1.4127491879235576,
      "grad_norm": 0.5735180974006653,
      "learning_rate": 1.2024569528328824e-05,
      "loss": 0.7863,
      "step": 21420
    },
    {
      "epoch": 1.4134087423945125,
      "grad_norm": 0.6213884949684143,
      "learning_rate": 1.1999979182277646e-05,
      "loss": 0.7929,
      "step": 21430
    },
    {
      "epoch": 1.4140682968654674,
      "grad_norm": 0.531531035900116,
      "learning_rate": 1.1975406063035787e-05,
      "loss": 0.8341,
      "step": 21440
    },
    {
      "epoch": 1.4147278513364223,
      "grad_norm": 0.5258073806762695,
      "learning_rate": 1.1950850203165986e-05,
      "loss": 0.7921,
      "step": 21450
    },
    {
      "epoch": 1.415387405807377,
      "grad_norm": 0.5849917531013489,
      "learning_rate": 1.1926311635208115e-05,
      "loss": 0.8012,
      "step": 21460
    },
    {
      "epoch": 1.416046960278332,
      "grad_norm": 0.6284214854240417,
      "learning_rate": 1.1901790391679161e-05,
      "loss": 0.7843,
      "step": 21470
    },
    {
      "epoch": 1.4167065147492868,
      "grad_norm": 0.5382357239723206,
      "learning_rate": 1.1877286505073114e-05,
      "loss": 0.8127,
      "step": 21480
    },
    {
      "epoch": 1.4173660692202417,
      "grad_norm": 0.5823057293891907,
      "learning_rate": 1.1852800007860984e-05,
      "loss": 0.8127,
      "step": 21490
    },
    {
      "epoch": 1.4180256236911966,
      "grad_norm": 0.5972256064414978,
      "learning_rate": 1.182833093249075e-05,
      "loss": 0.8037,
      "step": 21500
    },
    {
      "epoch": 1.4186851781621515,
      "grad_norm": 0.6301577687263489,
      "learning_rate": 1.180387931138727e-05,
      "loss": 0.8301,
      "step": 21510
    },
    {
      "epoch": 1.4193447326331063,
      "grad_norm": 0.6335862874984741,
      "learning_rate": 1.1779445176952314e-05,
      "loss": 0.8143,
      "step": 21520
    },
    {
      "epoch": 1.4200042871040612,
      "grad_norm": 0.6473941206932068,
      "learning_rate": 1.1755028561564448e-05,
      "loss": 0.8091,
      "step": 21530
    },
    {
      "epoch": 1.4206638415750161,
      "grad_norm": 0.6132245659828186,
      "learning_rate": 1.1730629497579036e-05,
      "loss": 0.8476,
      "step": 21540
    },
    {
      "epoch": 1.421323396045971,
      "grad_norm": 0.6118532419204712,
      "learning_rate": 1.170624801732817e-05,
      "loss": 0.7832,
      "step": 21550
    },
    {
      "epoch": 1.421982950516926,
      "grad_norm": 0.6644175052642822,
      "learning_rate": 1.1681884153120674e-05,
      "loss": 0.8288,
      "step": 21560
    },
    {
      "epoch": 1.4226425049878806,
      "grad_norm": 0.6317498087882996,
      "learning_rate": 1.1657537937241996e-05,
      "loss": 0.8498,
      "step": 21570
    },
    {
      "epoch": 1.4233020594588355,
      "grad_norm": 0.514884352684021,
      "learning_rate": 1.1633209401954202e-05,
      "loss": 0.8146,
      "step": 21580
    },
    {
      "epoch": 1.4239616139297904,
      "grad_norm": 0.5476719737052917,
      "learning_rate": 1.160889857949595e-05,
      "loss": 0.8387,
      "step": 21590
    },
    {
      "epoch": 1.4246211684007453,
      "grad_norm": 0.605689287185669,
      "learning_rate": 1.15846055020824e-05,
      "loss": 0.7732,
      "step": 21600
    },
    {
      "epoch": 1.4252807228717002,
      "grad_norm": 0.6338589787483215,
      "learning_rate": 1.156033020190522e-05,
      "loss": 0.8314,
      "step": 21610
    },
    {
      "epoch": 1.425940277342655,
      "grad_norm": 0.5500427484512329,
      "learning_rate": 1.1536072711132504e-05,
      "loss": 0.7729,
      "step": 21620
    },
    {
      "epoch": 1.42659983181361,
      "grad_norm": 0.5607891082763672,
      "learning_rate": 1.1511833061908753e-05,
      "loss": 0.7731,
      "step": 21630
    },
    {
      "epoch": 1.4272593862845648,
      "grad_norm": 0.6563683152198792,
      "learning_rate": 1.1487611286354812e-05,
      "loss": 0.8208,
      "step": 21640
    },
    {
      "epoch": 1.4279189407555197,
      "grad_norm": 0.5528565049171448,
      "learning_rate": 1.1463407416567876e-05,
      "loss": 0.8296,
      "step": 21650
    },
    {
      "epoch": 1.4285784952264744,
      "grad_norm": 0.6151896119117737,
      "learning_rate": 1.1439221484621376e-05,
      "loss": 0.7722,
      "step": 21660
    },
    {
      "epoch": 1.4292380496974295,
      "grad_norm": 0.5941686630249023,
      "learning_rate": 1.1415053522564981e-05,
      "loss": 0.8092,
      "step": 21670
    },
    {
      "epoch": 1.4298976041683842,
      "grad_norm": 0.6373192667961121,
      "learning_rate": 1.1390903562424567e-05,
      "loss": 0.7926,
      "step": 21680
    },
    {
      "epoch": 1.430557158639339,
      "grad_norm": 0.5912952423095703,
      "learning_rate": 1.136677163620213e-05,
      "loss": 0.8041,
      "step": 21690
    },
    {
      "epoch": 1.431216713110294,
      "grad_norm": 0.5902479290962219,
      "learning_rate": 1.1342657775875793e-05,
      "loss": 0.7994,
      "step": 21700
    },
    {
      "epoch": 1.4318762675812489,
      "grad_norm": 0.616900622844696,
      "learning_rate": 1.1318562013399717e-05,
      "loss": 0.7985,
      "step": 21710
    },
    {
      "epoch": 1.4325358220522038,
      "grad_norm": 0.5874533653259277,
      "learning_rate": 1.1294484380704082e-05,
      "loss": 0.8003,
      "step": 21720
    },
    {
      "epoch": 1.4331953765231586,
      "grad_norm": 0.5619797706604004,
      "learning_rate": 1.1270424909695068e-05,
      "loss": 0.8339,
      "step": 21730
    },
    {
      "epoch": 1.4338549309941135,
      "grad_norm": 0.5925664901733398,
      "learning_rate": 1.1246383632254753e-05,
      "loss": 0.7941,
      "step": 21740
    },
    {
      "epoch": 1.4345144854650682,
      "grad_norm": 0.6364315748214722,
      "learning_rate": 1.1222360580241157e-05,
      "loss": 0.8333,
      "step": 21750
    },
    {
      "epoch": 1.4351740399360233,
      "grad_norm": 0.5474036931991577,
      "learning_rate": 1.1198355785488072e-05,
      "loss": 0.8145,
      "step": 21760
    },
    {
      "epoch": 1.435833594406978,
      "grad_norm": 0.6324875950813293,
      "learning_rate": 1.1174369279805175e-05,
      "loss": 0.8044,
      "step": 21770
    },
    {
      "epoch": 1.436493148877933,
      "grad_norm": 0.5442603826522827,
      "learning_rate": 1.115040109497785e-05,
      "loss": 0.7998,
      "step": 21780
    },
    {
      "epoch": 1.4371527033488878,
      "grad_norm": 0.610609233379364,
      "learning_rate": 1.1126451262767249e-05,
      "loss": 0.7931,
      "step": 21790
    },
    {
      "epoch": 1.4378122578198427,
      "grad_norm": 0.6423842906951904,
      "learning_rate": 1.110251981491017e-05,
      "loss": 0.8003,
      "step": 21800
    },
    {
      "epoch": 1.4384718122907976,
      "grad_norm": 0.6783219575881958,
      "learning_rate": 1.1078606783119055e-05,
      "loss": 0.777,
      "step": 21810
    },
    {
      "epoch": 1.4391313667617525,
      "grad_norm": 0.6192532181739807,
      "learning_rate": 1.1054712199081967e-05,
      "loss": 0.8204,
      "step": 21820
    },
    {
      "epoch": 1.4397909212327074,
      "grad_norm": 0.6282913088798523,
      "learning_rate": 1.1030836094462484e-05,
      "loss": 0.8036,
      "step": 21830
    },
    {
      "epoch": 1.4404504757036622,
      "grad_norm": 0.5335955619812012,
      "learning_rate": 1.1006978500899748e-05,
      "loss": 0.8666,
      "step": 21840
    },
    {
      "epoch": 1.4411100301746171,
      "grad_norm": 0.463339239358902,
      "learning_rate": 1.0983139450008298e-05,
      "loss": 0.7961,
      "step": 21850
    },
    {
      "epoch": 1.4417695846455718,
      "grad_norm": 0.6162631511688232,
      "learning_rate": 1.0959318973378177e-05,
      "loss": 0.837,
      "step": 21860
    },
    {
      "epoch": 1.4424291391165267,
      "grad_norm": 0.6249439716339111,
      "learning_rate": 1.0935517102574757e-05,
      "loss": 0.7661,
      "step": 21870
    },
    {
      "epoch": 1.4430886935874816,
      "grad_norm": 0.5610096454620361,
      "learning_rate": 1.0911733869138796e-05,
      "loss": 0.7936,
      "step": 21880
    },
    {
      "epoch": 1.4437482480584365,
      "grad_norm": 0.5870029926300049,
      "learning_rate": 1.0887969304586348e-05,
      "loss": 0.8329,
      "step": 21890
    },
    {
      "epoch": 1.4444078025293914,
      "grad_norm": 0.6591834425926208,
      "learning_rate": 1.0864223440408689e-05,
      "loss": 0.8199,
      "step": 21900
    },
    {
      "epoch": 1.4450673570003463,
      "grad_norm": 0.5896702408790588,
      "learning_rate": 1.0840496308072364e-05,
      "loss": 0.7952,
      "step": 21910
    },
    {
      "epoch": 1.4457269114713012,
      "grad_norm": 0.6215192079544067,
      "learning_rate": 1.0816787939019062e-05,
      "loss": 0.8418,
      "step": 21920
    },
    {
      "epoch": 1.446386465942256,
      "grad_norm": 0.6321289539337158,
      "learning_rate": 1.0793098364665639e-05,
      "loss": 0.7868,
      "step": 21930
    },
    {
      "epoch": 1.447046020413211,
      "grad_norm": 0.6170499920845032,
      "learning_rate": 1.0769427616404015e-05,
      "loss": 0.7809,
      "step": 21940
    },
    {
      "epoch": 1.4477055748841656,
      "grad_norm": 0.5628238320350647,
      "learning_rate": 1.074577572560117e-05,
      "loss": 0.7912,
      "step": 21950
    },
    {
      "epoch": 1.4483651293551207,
      "grad_norm": 0.6782822608947754,
      "learning_rate": 1.0722142723599119e-05,
      "loss": 0.8704,
      "step": 21960
    },
    {
      "epoch": 1.4490246838260754,
      "grad_norm": 0.5879037976264954,
      "learning_rate": 1.069852864171482e-05,
      "loss": 0.8164,
      "step": 21970
    },
    {
      "epoch": 1.4496842382970303,
      "grad_norm": 0.5934215188026428,
      "learning_rate": 1.0674933511240165e-05,
      "loss": 0.8124,
      "step": 21980
    },
    {
      "epoch": 1.4503437927679852,
      "grad_norm": 0.5908352136611938,
      "learning_rate": 1.0651357363441927e-05,
      "loss": 0.8108,
      "step": 21990
    },
    {
      "epoch": 1.45100334723894,
      "grad_norm": 0.6745395064353943,
      "learning_rate": 1.0627800229561755e-05,
      "loss": 0.781,
      "step": 22000
    },
    {
      "epoch": 1.451662901709895,
      "grad_norm": 0.5967004299163818,
      "learning_rate": 1.060426214081606e-05,
      "loss": 0.8093,
      "step": 22010
    },
    {
      "epoch": 1.4523224561808499,
      "grad_norm": 0.6017720103263855,
      "learning_rate": 1.058074312839605e-05,
      "loss": 0.8131,
      "step": 22020
    },
    {
      "epoch": 1.4529820106518048,
      "grad_norm": 0.5551460385322571,
      "learning_rate": 1.0557243223467636e-05,
      "loss": 0.8141,
      "step": 22030
    },
    {
      "epoch": 1.4536415651227594,
      "grad_norm": 0.5525084733963013,
      "learning_rate": 1.0533762457171403e-05,
      "loss": 0.7938,
      "step": 22040
    },
    {
      "epoch": 1.4543011195937146,
      "grad_norm": 0.602660059928894,
      "learning_rate": 1.0510300860622602e-05,
      "loss": 0.8027,
      "step": 22050
    },
    {
      "epoch": 1.4549606740646692,
      "grad_norm": 0.5001670122146606,
      "learning_rate": 1.0486858464911053e-05,
      "loss": 0.7908,
      "step": 22060
    },
    {
      "epoch": 1.4556202285356241,
      "grad_norm": 0.5953605771064758,
      "learning_rate": 1.0463435301101144e-05,
      "loss": 0.8254,
      "step": 22070
    },
    {
      "epoch": 1.456279783006579,
      "grad_norm": 0.5867521166801453,
      "learning_rate": 1.044003140023177e-05,
      "loss": 0.741,
      "step": 22080
    },
    {
      "epoch": 1.456939337477534,
      "grad_norm": 0.5637813210487366,
      "learning_rate": 1.0416646793316324e-05,
      "loss": 0.7721,
      "step": 22090
    },
    {
      "epoch": 1.4575988919484888,
      "grad_norm": 0.5925215482711792,
      "learning_rate": 1.0393281511342598e-05,
      "loss": 0.815,
      "step": 22100
    },
    {
      "epoch": 1.4582584464194437,
      "grad_norm": 0.6085634827613831,
      "learning_rate": 1.0369935585272808e-05,
      "loss": 0.79,
      "step": 22110
    },
    {
      "epoch": 1.4589180008903986,
      "grad_norm": 0.594815731048584,
      "learning_rate": 1.0346609046043501e-05,
      "loss": 0.8484,
      "step": 22120
    },
    {
      "epoch": 1.4595775553613535,
      "grad_norm": 0.6512571573257446,
      "learning_rate": 1.0323301924565528e-05,
      "loss": 0.805,
      "step": 22130
    },
    {
      "epoch": 1.4602371098323084,
      "grad_norm": 0.6098020672798157,
      "learning_rate": 1.0300014251724036e-05,
      "loss": 0.7833,
      "step": 22140
    },
    {
      "epoch": 1.460896664303263,
      "grad_norm": 0.5692249536514282,
      "learning_rate": 1.0276746058378375e-05,
      "loss": 0.7819,
      "step": 22150
    },
    {
      "epoch": 1.4615562187742182,
      "grad_norm": 0.5564795136451721,
      "learning_rate": 1.0253497375362084e-05,
      "loss": 0.8166,
      "step": 22160
    },
    {
      "epoch": 1.4622157732451728,
      "grad_norm": 0.6502294540405273,
      "learning_rate": 1.0230268233482867e-05,
      "loss": 0.8091,
      "step": 22170
    },
    {
      "epoch": 1.4628753277161277,
      "grad_norm": 0.5914169549942017,
      "learning_rate": 1.0207058663522506e-05,
      "loss": 0.8058,
      "step": 22180
    },
    {
      "epoch": 1.4635348821870826,
      "grad_norm": 0.6946161985397339,
      "learning_rate": 1.0183868696236878e-05,
      "loss": 0.7911,
      "step": 22190
    },
    {
      "epoch": 1.4641944366580375,
      "grad_norm": 0.5570001602172852,
      "learning_rate": 1.0160698362355864e-05,
      "loss": 0.8019,
      "step": 22200
    },
    {
      "epoch": 1.4648539911289924,
      "grad_norm": 0.558890163898468,
      "learning_rate": 1.0137547692583328e-05,
      "loss": 0.8003,
      "step": 22210
    },
    {
      "epoch": 1.4655135455999473,
      "grad_norm": 0.5632757544517517,
      "learning_rate": 1.011441671759707e-05,
      "loss": 0.8181,
      "step": 22220
    },
    {
      "epoch": 1.4661731000709022,
      "grad_norm": 0.6365344524383545,
      "learning_rate": 1.0091305468048822e-05,
      "loss": 0.8286,
      "step": 22230
    },
    {
      "epoch": 1.4668326545418569,
      "grad_norm": 0.5856372117996216,
      "learning_rate": 1.0068213974564147e-05,
      "loss": 0.8314,
      "step": 22240
    },
    {
      "epoch": 1.467492209012812,
      "grad_norm": 0.6081796288490295,
      "learning_rate": 1.0045142267742435e-05,
      "loss": 0.7878,
      "step": 22250
    },
    {
      "epoch": 1.4681517634837666,
      "grad_norm": 0.5490699410438538,
      "learning_rate": 1.0022090378156871e-05,
      "loss": 0.7427,
      "step": 22260
    },
    {
      "epoch": 1.4688113179547215,
      "grad_norm": 0.62840735912323,
      "learning_rate": 9.999058336354353e-06,
      "loss": 0.8142,
      "step": 22270
    },
    {
      "epoch": 1.4694708724256764,
      "grad_norm": 0.5533925294876099,
      "learning_rate": 9.976046172855511e-06,
      "loss": 0.8006,
      "step": 22280
    },
    {
      "epoch": 1.4701304268966313,
      "grad_norm": 0.584366500377655,
      "learning_rate": 9.953053918154607e-06,
      "loss": 0.8489,
      "step": 22290
    },
    {
      "epoch": 1.4707899813675862,
      "grad_norm": 0.6402396559715271,
      "learning_rate": 9.930081602719526e-06,
      "loss": 0.7938,
      "step": 22300
    },
    {
      "epoch": 1.471449535838541,
      "grad_norm": 0.5986910462379456,
      "learning_rate": 9.907129256991732e-06,
      "loss": 0.8286,
      "step": 22310
    },
    {
      "epoch": 1.472109090309496,
      "grad_norm": 0.6115344762802124,
      "learning_rate": 9.884196911386231e-06,
      "loss": 0.8422,
      "step": 22320
    },
    {
      "epoch": 1.4727686447804509,
      "grad_norm": 0.57796710729599,
      "learning_rate": 9.861284596291548e-06,
      "loss": 0.7795,
      "step": 22330
    },
    {
      "epoch": 1.4734281992514058,
      "grad_norm": 0.6050494909286499,
      "learning_rate": 9.838392342069602e-06,
      "loss": 0.8121,
      "step": 22340
    },
    {
      "epoch": 1.4740877537223604,
      "grad_norm": 0.5895455479621887,
      "learning_rate": 9.815520179055793e-06,
      "loss": 0.8373,
      "step": 22350
    },
    {
      "epoch": 1.4747473081933153,
      "grad_norm": 0.5442635416984558,
      "learning_rate": 9.792668137558855e-06,
      "loss": 0.7792,
      "step": 22360
    },
    {
      "epoch": 1.4754068626642702,
      "grad_norm": 0.5683884024620056,
      "learning_rate": 9.769836247860891e-06,
      "loss": 0.7822,
      "step": 22370
    },
    {
      "epoch": 1.4760664171352251,
      "grad_norm": 0.5949370861053467,
      "learning_rate": 9.74702454021728e-06,
      "loss": 0.8117,
      "step": 22380
    },
    {
      "epoch": 1.47672597160618,
      "grad_norm": 0.7127247452735901,
      "learning_rate": 9.72423304485665e-06,
      "loss": 0.7423,
      "step": 22390
    },
    {
      "epoch": 1.477385526077135,
      "grad_norm": 0.5755226016044617,
      "learning_rate": 9.701461791980871e-06,
      "loss": 0.8108,
      "step": 22400
    },
    {
      "epoch": 1.4780450805480898,
      "grad_norm": 0.6091025471687317,
      "learning_rate": 9.67871081176496e-06,
      "loss": 0.7684,
      "step": 22410
    },
    {
      "epoch": 1.4787046350190447,
      "grad_norm": 0.5700154900550842,
      "learning_rate": 9.655980134357107e-06,
      "loss": 0.8291,
      "step": 22420
    },
    {
      "epoch": 1.4793641894899996,
      "grad_norm": 0.5324963331222534,
      "learning_rate": 9.633269789878544e-06,
      "loss": 0.8019,
      "step": 22430
    },
    {
      "epoch": 1.4800237439609543,
      "grad_norm": 0.5627251267433167,
      "learning_rate": 9.610579808423612e-06,
      "loss": 0.8254,
      "step": 22440
    },
    {
      "epoch": 1.4806832984319094,
      "grad_norm": 0.5750599503517151,
      "learning_rate": 9.587910220059631e-06,
      "loss": 0.8373,
      "step": 22450
    },
    {
      "epoch": 1.481342852902864,
      "grad_norm": 0.6064298748970032,
      "learning_rate": 9.56526105482693e-06,
      "loss": 0.8165,
      "step": 22460
    },
    {
      "epoch": 1.482002407373819,
      "grad_norm": 0.6086931824684143,
      "learning_rate": 9.542632342738746e-06,
      "loss": 0.8612,
      "step": 22470
    },
    {
      "epoch": 1.4826619618447738,
      "grad_norm": 0.6298277974128723,
      "learning_rate": 9.52002411378122e-06,
      "loss": 0.8165,
      "step": 22480
    },
    {
      "epoch": 1.4833215163157287,
      "grad_norm": 0.5675653219223022,
      "learning_rate": 9.497436397913368e-06,
      "loss": 0.7859,
      "step": 22490
    },
    {
      "epoch": 1.4839810707866836,
      "grad_norm": 0.6803301572799683,
      "learning_rate": 9.474869225066993e-06,
      "loss": 0.7845,
      "step": 22500
    },
    {
      "epoch": 1.4846406252576385,
      "grad_norm": 0.6293306350708008,
      "learning_rate": 9.452322625146715e-06,
      "loss": 0.7989,
      "step": 22510
    },
    {
      "epoch": 1.4853001797285934,
      "grad_norm": 0.6525318622589111,
      "learning_rate": 9.429796628029838e-06,
      "loss": 0.8538,
      "step": 22520
    },
    {
      "epoch": 1.485959734199548,
      "grad_norm": 0.6509277820587158,
      "learning_rate": 9.407291263566424e-06,
      "loss": 0.7948,
      "step": 22530
    },
    {
      "epoch": 1.4866192886705032,
      "grad_norm": 0.5920518636703491,
      "learning_rate": 9.384806561579141e-06,
      "loss": 0.7992,
      "step": 22540
    },
    {
      "epoch": 1.4872788431414579,
      "grad_norm": 0.5769726634025574,
      "learning_rate": 9.36234255186332e-06,
      "loss": 0.8238,
      "step": 22550
    },
    {
      "epoch": 1.4879383976124128,
      "grad_norm": 0.6404539346694946,
      "learning_rate": 9.339899264186846e-06,
      "loss": 0.8269,
      "step": 22560
    },
    {
      "epoch": 1.4885979520833676,
      "grad_norm": 0.5562175512313843,
      "learning_rate": 9.317476728290146e-06,
      "loss": 0.8011,
      "step": 22570
    },
    {
      "epoch": 1.4892575065543225,
      "grad_norm": 0.7294158935546875,
      "learning_rate": 9.295074973886164e-06,
      "loss": 0.8319,
      "step": 22580
    },
    {
      "epoch": 1.4899170610252774,
      "grad_norm": 0.6766993999481201,
      "learning_rate": 9.272694030660279e-06,
      "loss": 0.7976,
      "step": 22590
    },
    {
      "epoch": 1.4905766154962323,
      "grad_norm": 0.6774852871894836,
      "learning_rate": 9.25033392827033e-06,
      "loss": 0.8286,
      "step": 22600
    },
    {
      "epoch": 1.4912361699671872,
      "grad_norm": 0.6171419024467468,
      "learning_rate": 9.227994696346502e-06,
      "loss": 0.8103,
      "step": 22610
    },
    {
      "epoch": 1.4918957244381421,
      "grad_norm": 0.6780225038528442,
      "learning_rate": 9.205676364491333e-06,
      "loss": 0.8463,
      "step": 22620
    },
    {
      "epoch": 1.492555278909097,
      "grad_norm": 0.560507595539093,
      "learning_rate": 9.183378962279693e-06,
      "loss": 0.813,
      "step": 22630
    },
    {
      "epoch": 1.4932148333800517,
      "grad_norm": 0.5771291851997375,
      "learning_rate": 9.161102519258683e-06,
      "loss": 0.8009,
      "step": 22640
    },
    {
      "epoch": 1.4938743878510066,
      "grad_norm": 0.5692350268363953,
      "learning_rate": 9.138847064947642e-06,
      "loss": 0.8096,
      "step": 22650
    },
    {
      "epoch": 1.4945339423219615,
      "grad_norm": 0.6242731809616089,
      "learning_rate": 9.116612628838091e-06,
      "loss": 0.7972,
      "step": 22660
    },
    {
      "epoch": 1.4951934967929164,
      "grad_norm": 0.5729048848152161,
      "learning_rate": 9.094399240393722e-06,
      "loss": 0.7981,
      "step": 22670
    },
    {
      "epoch": 1.4958530512638712,
      "grad_norm": 0.5291873812675476,
      "learning_rate": 9.072206929050303e-06,
      "loss": 0.7915,
      "step": 22680
    },
    {
      "epoch": 1.4965126057348261,
      "grad_norm": 0.6954770088195801,
      "learning_rate": 9.050035724215703e-06,
      "loss": 0.8027,
      "step": 22690
    },
    {
      "epoch": 1.497172160205781,
      "grad_norm": 0.5207001566886902,
      "learning_rate": 9.0278856552698e-06,
      "loss": 0.8056,
      "step": 22700
    },
    {
      "epoch": 1.497831714676736,
      "grad_norm": 0.6069022417068481,
      "learning_rate": 9.005756751564459e-06,
      "loss": 0.7766,
      "step": 22710
    },
    {
      "epoch": 1.4984912691476908,
      "grad_norm": 0.5734812617301941,
      "learning_rate": 8.983649042423534e-06,
      "loss": 0.7816,
      "step": 22720
    },
    {
      "epoch": 1.4991508236186455,
      "grad_norm": 0.81162428855896,
      "learning_rate": 8.961562557142758e-06,
      "loss": 0.7697,
      "step": 22730
    },
    {
      "epoch": 1.4998103780896006,
      "grad_norm": 0.595378041267395,
      "learning_rate": 8.939497324989748e-06,
      "loss": 0.769,
      "step": 22740
    },
    {
      "epoch": 1.5004699325605553,
      "grad_norm": 0.5689677000045776,
      "learning_rate": 8.917453375203958e-06,
      "loss": 0.8124,
      "step": 22750
    },
    {
      "epoch": 1.5011294870315102,
      "grad_norm": 0.5219847559928894,
      "learning_rate": 8.895430736996645e-06,
      "loss": 0.8031,
      "step": 22760
    },
    {
      "epoch": 1.501789041502465,
      "grad_norm": 0.610152542591095,
      "learning_rate": 8.87342943955084e-06,
      "loss": 0.8004,
      "step": 22770
    },
    {
      "epoch": 1.50244859597342,
      "grad_norm": 0.6241317391395569,
      "learning_rate": 8.85144951202126e-06,
      "loss": 0.7737,
      "step": 22780
    },
    {
      "epoch": 1.5031081504443748,
      "grad_norm": 0.6606419086456299,
      "learning_rate": 8.829490983534324e-06,
      "loss": 0.798,
      "step": 22790
    },
    {
      "epoch": 1.5037677049153297,
      "grad_norm": 0.6633086800575256,
      "learning_rate": 8.807553883188086e-06,
      "loss": 0.8282,
      "step": 22800
    },
    {
      "epoch": 1.5044272593862846,
      "grad_norm": 0.7518271803855896,
      "learning_rate": 8.785638240052227e-06,
      "loss": 0.8244,
      "step": 22810
    },
    {
      "epoch": 1.5050868138572393,
      "grad_norm": 0.6326174139976501,
      "learning_rate": 8.763744083167965e-06,
      "loss": 0.801,
      "step": 22820
    },
    {
      "epoch": 1.5057463683281944,
      "grad_norm": 0.5996015667915344,
      "learning_rate": 8.741871441548056e-06,
      "loss": 0.8471,
      "step": 22830
    },
    {
      "epoch": 1.506405922799149,
      "grad_norm": 0.5443564057350159,
      "learning_rate": 8.720020344176764e-06,
      "loss": 0.7987,
      "step": 22840
    },
    {
      "epoch": 1.5070654772701042,
      "grad_norm": 0.6426756381988525,
      "learning_rate": 8.698190820009764e-06,
      "loss": 0.8308,
      "step": 22850
    },
    {
      "epoch": 1.5077250317410589,
      "grad_norm": 0.6051197052001953,
      "learning_rate": 8.676382897974191e-06,
      "loss": 0.8043,
      "step": 22860
    },
    {
      "epoch": 1.5083845862120138,
      "grad_norm": 0.5542495250701904,
      "learning_rate": 8.654596606968524e-06,
      "loss": 0.8075,
      "step": 22870
    },
    {
      "epoch": 1.5090441406829687,
      "grad_norm": 0.5416606664657593,
      "learning_rate": 8.632831975862579e-06,
      "loss": 0.8075,
      "step": 22880
    },
    {
      "epoch": 1.5097036951539236,
      "grad_norm": 0.5079529285430908,
      "learning_rate": 8.61108903349747e-06,
      "loss": 0.7696,
      "step": 22890
    },
    {
      "epoch": 1.5103632496248784,
      "grad_norm": 0.6498621702194214,
      "learning_rate": 8.589367808685586e-06,
      "loss": 0.7949,
      "step": 22900
    },
    {
      "epoch": 1.5110228040958331,
      "grad_norm": 0.6200118064880371,
      "learning_rate": 8.56766833021053e-06,
      "loss": 0.8284,
      "step": 22910
    },
    {
      "epoch": 1.5116823585667882,
      "grad_norm": 0.5316246747970581,
      "learning_rate": 8.545990626827066e-06,
      "loss": 0.8584,
      "step": 22920
    },
    {
      "epoch": 1.512341913037743,
      "grad_norm": 0.5837723016738892,
      "learning_rate": 8.524334727261145e-06,
      "loss": 0.8238,
      "step": 22930
    },
    {
      "epoch": 1.513001467508698,
      "grad_norm": 0.5774195194244385,
      "learning_rate": 8.502700660209784e-06,
      "loss": 0.8266,
      "step": 22940
    },
    {
      "epoch": 1.5136610219796527,
      "grad_norm": 0.6202636361122131,
      "learning_rate": 8.481088454341114e-06,
      "loss": 0.7763,
      "step": 22950
    },
    {
      "epoch": 1.5143205764506076,
      "grad_norm": 0.599242627620697,
      "learning_rate": 8.459498138294228e-06,
      "loss": 0.7837,
      "step": 22960
    },
    {
      "epoch": 1.5149801309215625,
      "grad_norm": 0.5883403420448303,
      "learning_rate": 8.43792974067929e-06,
      "loss": 0.8001,
      "step": 22970
    },
    {
      "epoch": 1.5156396853925174,
      "grad_norm": 0.6413730978965759,
      "learning_rate": 8.416383290077356e-06,
      "loss": 0.7961,
      "step": 22980
    },
    {
      "epoch": 1.5162992398634723,
      "grad_norm": 0.6023602485656738,
      "learning_rate": 8.394858815040443e-06,
      "loss": 0.7996,
      "step": 22990
    },
    {
      "epoch": 1.516958794334427,
      "grad_norm": 0.5290365219116211,
      "learning_rate": 8.373356344091442e-06,
      "loss": 0.7872,
      "step": 23000
    },
    {
      "epoch": 1.517618348805382,
      "grad_norm": 0.5551934242248535,
      "learning_rate": 8.351875905724044e-06,
      "loss": 0.7933,
      "step": 23010
    },
    {
      "epoch": 1.5182779032763367,
      "grad_norm": 0.5995215177536011,
      "learning_rate": 8.330417528402795e-06,
      "loss": 0.8193,
      "step": 23020
    },
    {
      "epoch": 1.5189374577472918,
      "grad_norm": 0.6809059381484985,
      "learning_rate": 8.308981240562974e-06,
      "loss": 0.8003,
      "step": 23030
    },
    {
      "epoch": 1.5195970122182465,
      "grad_norm": 0.5909712910652161,
      "learning_rate": 8.287567070610613e-06,
      "loss": 0.8035,
      "step": 23040
    },
    {
      "epoch": 1.5202565666892014,
      "grad_norm": 0.5868057608604431,
      "learning_rate": 8.266175046922414e-06,
      "loss": 0.8191,
      "step": 23050
    },
    {
      "epoch": 1.5209161211601563,
      "grad_norm": 0.56301349401474,
      "learning_rate": 8.244805197845734e-06,
      "loss": 0.8133,
      "step": 23060
    },
    {
      "epoch": 1.5215756756311112,
      "grad_norm": 0.6976560950279236,
      "learning_rate": 8.223457551698568e-06,
      "loss": 0.797,
      "step": 23070
    },
    {
      "epoch": 1.522235230102066,
      "grad_norm": 0.5786426067352295,
      "learning_rate": 8.202132136769458e-06,
      "loss": 0.797,
      "step": 23080
    },
    {
      "epoch": 1.522894784573021,
      "grad_norm": 0.6411847472190857,
      "learning_rate": 8.18082898131752e-06,
      "loss": 0.8056,
      "step": 23090
    },
    {
      "epoch": 1.5235543390439759,
      "grad_norm": 0.5769123435020447,
      "learning_rate": 8.159548113572329e-06,
      "loss": 0.7846,
      "step": 23100
    },
    {
      "epoch": 1.5242138935149305,
      "grad_norm": 0.6662389636039734,
      "learning_rate": 8.138289561733973e-06,
      "loss": 0.7897,
      "step": 23110
    },
    {
      "epoch": 1.5248734479858856,
      "grad_norm": 0.5101078748703003,
      "learning_rate": 8.117053353972931e-06,
      "loss": 0.7911,
      "step": 23120
    },
    {
      "epoch": 1.5255330024568403,
      "grad_norm": 0.5832508206367493,
      "learning_rate": 8.095839518430105e-06,
      "loss": 0.8233,
      "step": 23130
    },
    {
      "epoch": 1.5261925569277954,
      "grad_norm": 0.6068770289421082,
      "learning_rate": 8.074648083216723e-06,
      "loss": 0.787,
      "step": 23140
    },
    {
      "epoch": 1.52685211139875,
      "grad_norm": 0.5033300518989563,
      "learning_rate": 8.053479076414338e-06,
      "loss": 0.7904,
      "step": 23150
    },
    {
      "epoch": 1.527511665869705,
      "grad_norm": 0.5479113459587097,
      "learning_rate": 8.032332526074801e-06,
      "loss": 0.8433,
      "step": 23160
    },
    {
      "epoch": 1.5281712203406599,
      "grad_norm": 0.5920894145965576,
      "learning_rate": 8.01120846022017e-06,
      "loss": 0.7985,
      "step": 23170
    },
    {
      "epoch": 1.5288307748116148,
      "grad_norm": 0.7134125232696533,
      "learning_rate": 7.990106906842753e-06,
      "loss": 0.7729,
      "step": 23180
    },
    {
      "epoch": 1.5294903292825697,
      "grad_norm": 0.6525259017944336,
      "learning_rate": 7.96902789390497e-06,
      "loss": 0.792,
      "step": 23190
    },
    {
      "epoch": 1.5301498837535243,
      "grad_norm": 0.6140827536582947,
      "learning_rate": 7.94797144933942e-06,
      "loss": 0.8135,
      "step": 23200
    },
    {
      "epoch": 1.5308094382244795,
      "grad_norm": 0.6362495422363281,
      "learning_rate": 7.926937601048779e-06,
      "loss": 0.769,
      "step": 23210
    },
    {
      "epoch": 1.5314689926954341,
      "grad_norm": 0.6009213924407959,
      "learning_rate": 7.905926376905776e-06,
      "loss": 0.804,
      "step": 23220
    },
    {
      "epoch": 1.5321285471663892,
      "grad_norm": 0.6461767554283142,
      "learning_rate": 7.88493780475316e-06,
      "loss": 0.7764,
      "step": 23230
    },
    {
      "epoch": 1.532788101637344,
      "grad_norm": 0.6046391725540161,
      "learning_rate": 7.863971912403659e-06,
      "loss": 0.7996,
      "step": 23240
    },
    {
      "epoch": 1.5334476561082988,
      "grad_norm": 0.6695650219917297,
      "learning_rate": 7.84302872763997e-06,
      "loss": 0.7954,
      "step": 23250
    },
    {
      "epoch": 1.5341072105792537,
      "grad_norm": 0.6348613500595093,
      "learning_rate": 7.822108278214668e-06,
      "loss": 0.8027,
      "step": 23260
    },
    {
      "epoch": 1.5347667650502086,
      "grad_norm": 0.593072772026062,
      "learning_rate": 7.80121059185023e-06,
      "loss": 0.8247,
      "step": 23270
    },
    {
      "epoch": 1.5354263195211635,
      "grad_norm": 0.5665187835693359,
      "learning_rate": 7.78033569623895e-06,
      "loss": 0.7775,
      "step": 23280
    },
    {
      "epoch": 1.5360858739921184,
      "grad_norm": 0.6689664721488953,
      "learning_rate": 7.759483619042912e-06,
      "loss": 0.8052,
      "step": 23290
    },
    {
      "epoch": 1.5367454284630733,
      "grad_norm": 0.5643017292022705,
      "learning_rate": 7.738654387893995e-06,
      "loss": 0.8193,
      "step": 23300
    },
    {
      "epoch": 1.537404982934028,
      "grad_norm": 0.6332219243049622,
      "learning_rate": 7.717848030393782e-06,
      "loss": 0.8418,
      "step": 23310
    },
    {
      "epoch": 1.538064537404983,
      "grad_norm": 0.5949552655220032,
      "learning_rate": 7.697064574113539e-06,
      "loss": 0.8179,
      "step": 23320
    },
    {
      "epoch": 1.5387240918759377,
      "grad_norm": 0.6400283575057983,
      "learning_rate": 7.67630404659419e-06,
      "loss": 0.8082,
      "step": 23330
    },
    {
      "epoch": 1.5393836463468926,
      "grad_norm": 0.6538004279136658,
      "learning_rate": 7.655566475346298e-06,
      "loss": 0.835,
      "step": 23340
    },
    {
      "epoch": 1.5400432008178475,
      "grad_norm": 0.5514785647392273,
      "learning_rate": 7.63485188784997e-06,
      "loss": 0.8312,
      "step": 23350
    },
    {
      "epoch": 1.5407027552888024,
      "grad_norm": 0.6204501390457153,
      "learning_rate": 7.614160311554888e-06,
      "loss": 0.8094,
      "step": 23360
    },
    {
      "epoch": 1.5413623097597573,
      "grad_norm": 0.6015772819519043,
      "learning_rate": 7.593491773880218e-06,
      "loss": 0.877,
      "step": 23370
    },
    {
      "epoch": 1.5420218642307122,
      "grad_norm": 0.5586115121841431,
      "learning_rate": 7.572846302214601e-06,
      "loss": 0.8028,
      "step": 23380
    },
    {
      "epoch": 1.542681418701667,
      "grad_norm": 0.7056943774223328,
      "learning_rate": 7.552223923916127e-06,
      "loss": 0.7808,
      "step": 23390
    },
    {
      "epoch": 1.5433409731726218,
      "grad_norm": 0.6403433084487915,
      "learning_rate": 7.53162466631227e-06,
      "loss": 0.825,
      "step": 23400
    },
    {
      "epoch": 1.5440005276435769,
      "grad_norm": 0.6124839186668396,
      "learning_rate": 7.511048556699868e-06,
      "loss": 0.8026,
      "step": 23410
    },
    {
      "epoch": 1.5446600821145315,
      "grad_norm": 0.5926290154457092,
      "learning_rate": 7.490495622345075e-06,
      "loss": 0.8174,
      "step": 23420
    },
    {
      "epoch": 1.5453196365854867,
      "grad_norm": 0.6216310262680054,
      "learning_rate": 7.46996589048336e-06,
      "loss": 0.7698,
      "step": 23430
    },
    {
      "epoch": 1.5459791910564413,
      "grad_norm": 0.6308226585388184,
      "learning_rate": 7.449459388319447e-06,
      "loss": 0.7954,
      "step": 23440
    },
    {
      "epoch": 1.5466387455273962,
      "grad_norm": 0.5584129095077515,
      "learning_rate": 7.428976143027224e-06,
      "loss": 0.7876,
      "step": 23450
    },
    {
      "epoch": 1.5472982999983511,
      "grad_norm": 0.6883158087730408,
      "learning_rate": 7.408516181749825e-06,
      "loss": 0.7904,
      "step": 23460
    },
    {
      "epoch": 1.547957854469306,
      "grad_norm": 0.5625373721122742,
      "learning_rate": 7.38807953159949e-06,
      "loss": 0.8231,
      "step": 23470
    },
    {
      "epoch": 1.548617408940261,
      "grad_norm": 0.6092795729637146,
      "learning_rate": 7.367666219657596e-06,
      "loss": 0.8077,
      "step": 23480
    },
    {
      "epoch": 1.5492769634112156,
      "grad_norm": 0.649097740650177,
      "learning_rate": 7.347276272974568e-06,
      "loss": 0.7777,
      "step": 23490
    },
    {
      "epoch": 1.5499365178821707,
      "grad_norm": 0.6178075075149536,
      "learning_rate": 7.3269097185698744e-06,
      "loss": 0.7762,
      "step": 23500
    },
    {
      "epoch": 1.5505960723531254,
      "grad_norm": 0.6393544673919678,
      "learning_rate": 7.306566583432009e-06,
      "loss": 0.8093,
      "step": 23510
    },
    {
      "epoch": 1.5512556268240805,
      "grad_norm": 0.547970712184906,
      "learning_rate": 7.286246894518395e-06,
      "loss": 0.8173,
      "step": 23520
    },
    {
      "epoch": 1.5519151812950351,
      "grad_norm": 0.5908211469650269,
      "learning_rate": 7.2659506787554284e-06,
      "loss": 0.7633,
      "step": 23530
    },
    {
      "epoch": 1.55257473576599,
      "grad_norm": 0.5850244760513306,
      "learning_rate": 7.245677963038347e-06,
      "loss": 0.8296,
      "step": 23540
    },
    {
      "epoch": 1.553234290236945,
      "grad_norm": 0.6321101188659668,
      "learning_rate": 7.225428774231302e-06,
      "loss": 0.7984,
      "step": 23550
    },
    {
      "epoch": 1.5538938447078998,
      "grad_norm": 0.6044578552246094,
      "learning_rate": 7.205203139167227e-06,
      "loss": 0.8495,
      "step": 23560
    },
    {
      "epoch": 1.5545533991788547,
      "grad_norm": 0.649852991104126,
      "learning_rate": 7.185001084647874e-06,
      "loss": 0.7805,
      "step": 23570
    },
    {
      "epoch": 1.5552129536498096,
      "grad_norm": 0.5769763588905334,
      "learning_rate": 7.16482263744373e-06,
      "loss": 0.7945,
      "step": 23580
    },
    {
      "epoch": 1.5558725081207645,
      "grad_norm": 0.617280900478363,
      "learning_rate": 7.144667824293994e-06,
      "loss": 0.8014,
      "step": 23590
    },
    {
      "epoch": 1.5565320625917192,
      "grad_norm": 0.5664177536964417,
      "learning_rate": 7.124536671906571e-06,
      "loss": 0.806,
      "step": 23600
    },
    {
      "epoch": 1.5571916170626743,
      "grad_norm": 0.6676650643348694,
      "learning_rate": 7.104429206957983e-06,
      "loss": 0.7894,
      "step": 23610
    },
    {
      "epoch": 1.557851171533629,
      "grad_norm": 0.613895833492279,
      "learning_rate": 7.0843454560933945e-06,
      "loss": 0.7845,
      "step": 23620
    },
    {
      "epoch": 1.558510726004584,
      "grad_norm": 0.6392267346382141,
      "learning_rate": 7.064285445926505e-06,
      "loss": 0.7939,
      "step": 23630
    },
    {
      "epoch": 1.5591702804755387,
      "grad_norm": 0.6425051093101501,
      "learning_rate": 7.044249203039588e-06,
      "loss": 0.8511,
      "step": 23640
    },
    {
      "epoch": 1.5598298349464936,
      "grad_norm": 0.556755006313324,
      "learning_rate": 7.024236753983418e-06,
      "loss": 0.8516,
      "step": 23650
    },
    {
      "epoch": 1.5604893894174485,
      "grad_norm": 0.6224756836891174,
      "learning_rate": 7.004248125277221e-06,
      "loss": 0.8272,
      "step": 23660
    },
    {
      "epoch": 1.5611489438884034,
      "grad_norm": 0.6352970004081726,
      "learning_rate": 6.9842833434086905e-06,
      "loss": 0.8294,
      "step": 23670
    },
    {
      "epoch": 1.5618084983593583,
      "grad_norm": 0.6254895329475403,
      "learning_rate": 6.96434243483387e-06,
      "loss": 0.7893,
      "step": 23680
    },
    {
      "epoch": 1.562468052830313,
      "grad_norm": 0.6049076318740845,
      "learning_rate": 6.944425425977216e-06,
      "loss": 0.8025,
      "step": 23690
    },
    {
      "epoch": 1.563127607301268,
      "grad_norm": 0.6346916556358337,
      "learning_rate": 6.924532343231485e-06,
      "loss": 0.7668,
      "step": 23700
    },
    {
      "epoch": 1.5637871617722228,
      "grad_norm": 0.6637734770774841,
      "learning_rate": 6.904663212957749e-06,
      "loss": 0.7904,
      "step": 23710
    },
    {
      "epoch": 1.5644467162431779,
      "grad_norm": 0.591850996017456,
      "learning_rate": 6.884818061485321e-06,
      "loss": 0.8569,
      "step": 23720
    },
    {
      "epoch": 1.5651062707141326,
      "grad_norm": 0.5571227073669434,
      "learning_rate": 6.864996915111741e-06,
      "loss": 0.8004,
      "step": 23730
    },
    {
      "epoch": 1.5657658251850874,
      "grad_norm": 0.5591459274291992,
      "learning_rate": 6.8451998001027555e-06,
      "loss": 0.8331,
      "step": 23740
    },
    {
      "epoch": 1.5664253796560423,
      "grad_norm": 0.6120739579200745,
      "learning_rate": 6.825426742692248e-06,
      "loss": 0.8481,
      "step": 23750
    },
    {
      "epoch": 1.5670849341269972,
      "grad_norm": 0.6025487780570984,
      "learning_rate": 6.805677769082247e-06,
      "loss": 0.8433,
      "step": 23760
    },
    {
      "epoch": 1.5677444885979521,
      "grad_norm": 0.5513527393341064,
      "learning_rate": 6.785952905442819e-06,
      "loss": 0.8163,
      "step": 23770
    },
    {
      "epoch": 1.5684040430689068,
      "grad_norm": 0.6942810416221619,
      "learning_rate": 6.766252177912133e-06,
      "loss": 0.7812,
      "step": 23780
    },
    {
      "epoch": 1.569063597539862,
      "grad_norm": 0.6479464769363403,
      "learning_rate": 6.7465756125963425e-06,
      "loss": 0.7861,
      "step": 23790
    },
    {
      "epoch": 1.5697231520108166,
      "grad_norm": 0.6747413277626038,
      "learning_rate": 6.726923235569607e-06,
      "loss": 0.7849,
      "step": 23800
    },
    {
      "epoch": 1.5703827064817717,
      "grad_norm": 0.49388283491134644,
      "learning_rate": 6.707295072874009e-06,
      "loss": 0.7807,
      "step": 23810
    },
    {
      "epoch": 1.5710422609527264,
      "grad_norm": 0.6372382640838623,
      "learning_rate": 6.687691150519551e-06,
      "loss": 0.7993,
      "step": 23820
    },
    {
      "epoch": 1.5717018154236813,
      "grad_norm": 0.6567792296409607,
      "learning_rate": 6.668111494484128e-06,
      "loss": 0.8044,
      "step": 23830
    },
    {
      "epoch": 1.5723613698946362,
      "grad_norm": 0.6088330149650574,
      "learning_rate": 6.648556130713454e-06,
      "loss": 0.8587,
      "step": 23840
    },
    {
      "epoch": 1.573020924365591,
      "grad_norm": 0.5622130036354065,
      "learning_rate": 6.629025085121082e-06,
      "loss": 0.8884,
      "step": 23850
    },
    {
      "epoch": 1.573680478836546,
      "grad_norm": 0.5609302520751953,
      "learning_rate": 6.609518383588317e-06,
      "loss": 0.7804,
      "step": 23860
    },
    {
      "epoch": 1.5743400333075008,
      "grad_norm": 0.5331741571426392,
      "learning_rate": 6.590036051964205e-06,
      "loss": 0.7765,
      "step": 23870
    },
    {
      "epoch": 1.5749995877784557,
      "grad_norm": 0.5762697458267212,
      "learning_rate": 6.570578116065518e-06,
      "loss": 0.8276,
      "step": 23880
    },
    {
      "epoch": 1.5756591422494104,
      "grad_norm": 0.6534186601638794,
      "learning_rate": 6.551144601676681e-06,
      "loss": 0.8163,
      "step": 23890
    },
    {
      "epoch": 1.5763186967203655,
      "grad_norm": 0.6218444108963013,
      "learning_rate": 6.531735534549766e-06,
      "loss": 0.8002,
      "step": 23900
    },
    {
      "epoch": 1.5769782511913202,
      "grad_norm": 0.598573625087738,
      "learning_rate": 6.512350940404438e-06,
      "loss": 0.7777,
      "step": 23910
    },
    {
      "epoch": 1.5776378056622753,
      "grad_norm": 0.5644959211349487,
      "learning_rate": 6.492990844927957e-06,
      "loss": 0.811,
      "step": 23920
    },
    {
      "epoch": 1.57829736013323,
      "grad_norm": 0.5573763251304626,
      "learning_rate": 6.473655273775087e-06,
      "loss": 0.8174,
      "step": 23930
    },
    {
      "epoch": 1.5789569146041849,
      "grad_norm": 0.6255615949630737,
      "learning_rate": 6.4543442525681255e-06,
      "loss": 0.7781,
      "step": 23940
    },
    {
      "epoch": 1.5796164690751398,
      "grad_norm": 0.6131150126457214,
      "learning_rate": 6.435057806896813e-06,
      "loss": 0.8318,
      "step": 23950
    },
    {
      "epoch": 1.5802760235460946,
      "grad_norm": 0.5963195562362671,
      "learning_rate": 6.41579596231833e-06,
      "loss": 0.8412,
      "step": 23960
    },
    {
      "epoch": 1.5809355780170495,
      "grad_norm": 0.6147268414497375,
      "learning_rate": 6.39655874435727e-06,
      "loss": 0.8097,
      "step": 23970
    },
    {
      "epoch": 1.5815951324880042,
      "grad_norm": 0.6898005604743958,
      "learning_rate": 6.377346178505581e-06,
      "loss": 0.7921,
      "step": 23980
    },
    {
      "epoch": 1.5822546869589593,
      "grad_norm": 0.5684288144111633,
      "learning_rate": 6.358158290222541e-06,
      "loss": 0.8226,
      "step": 23990
    },
    {
      "epoch": 1.582914241429914,
      "grad_norm": 0.5642861127853394,
      "learning_rate": 6.338995104934728e-06,
      "loss": 0.8034,
      "step": 24000
    },
    {
      "epoch": 1.583573795900869,
      "grad_norm": 0.5649825930595398,
      "learning_rate": 6.3198566480360045e-06,
      "loss": 0.7769,
      "step": 24010
    },
    {
      "epoch": 1.5842333503718238,
      "grad_norm": 0.583200216293335,
      "learning_rate": 6.3007429448874385e-06,
      "loss": 0.7858,
      "step": 24020
    },
    {
      "epoch": 1.5848929048427787,
      "grad_norm": 0.5720763802528381,
      "learning_rate": 6.281654020817299e-06,
      "loss": 0.7903,
      "step": 24030
    },
    {
      "epoch": 1.5855524593137336,
      "grad_norm": 0.5873232483863831,
      "learning_rate": 6.262589901121044e-06,
      "loss": 0.8214,
      "step": 24040
    },
    {
      "epoch": 1.5862120137846885,
      "grad_norm": 0.5910772681236267,
      "learning_rate": 6.24355061106123e-06,
      "loss": 0.8017,
      "step": 24050
    },
    {
      "epoch": 1.5868715682556433,
      "grad_norm": 0.6636596918106079,
      "learning_rate": 6.224536175867538e-06,
      "loss": 0.8145,
      "step": 24060
    },
    {
      "epoch": 1.587531122726598,
      "grad_norm": 0.6654101014137268,
      "learning_rate": 6.205546620736694e-06,
      "loss": 0.7943,
      "step": 24070
    },
    {
      "epoch": 1.5881906771975531,
      "grad_norm": 0.6048585176467896,
      "learning_rate": 6.186581970832453e-06,
      "loss": 0.8313,
      "step": 24080
    },
    {
      "epoch": 1.5888502316685078,
      "grad_norm": 0.6448466777801514,
      "learning_rate": 6.167642251285593e-06,
      "loss": 0.7644,
      "step": 24090
    },
    {
      "epoch": 1.589509786139463,
      "grad_norm": 0.6216910481452942,
      "learning_rate": 6.148727487193823e-06,
      "loss": 0.8317,
      "step": 24100
    },
    {
      "epoch": 1.5901693406104176,
      "grad_norm": 0.6383108496665955,
      "learning_rate": 6.129837703621818e-06,
      "loss": 0.8389,
      "step": 24110
    },
    {
      "epoch": 1.5908288950813725,
      "grad_norm": 0.63627028465271,
      "learning_rate": 6.110972925601102e-06,
      "loss": 0.8508,
      "step": 24120
    },
    {
      "epoch": 1.5914884495523274,
      "grad_norm": 0.593573808670044,
      "learning_rate": 6.0921331781301106e-06,
      "loss": 0.8329,
      "step": 24130
    },
    {
      "epoch": 1.5921480040232823,
      "grad_norm": 0.5346869826316833,
      "learning_rate": 6.073318486174082e-06,
      "loss": 0.7971,
      "step": 24140
    },
    {
      "epoch": 1.5928075584942372,
      "grad_norm": 0.4986163377761841,
      "learning_rate": 6.054528874665072e-06,
      "loss": 0.8252,
      "step": 24150
    },
    {
      "epoch": 1.593467112965192,
      "grad_norm": 0.5987104773521423,
      "learning_rate": 6.035764368501884e-06,
      "loss": 0.8334,
      "step": 24160
    },
    {
      "epoch": 1.594126667436147,
      "grad_norm": 0.6206068396568298,
      "learning_rate": 6.0170249925500575e-06,
      "loss": 0.7933,
      "step": 24170
    },
    {
      "epoch": 1.5947862219071016,
      "grad_norm": 0.5876137018203735,
      "learning_rate": 5.9983107716418435e-06,
      "loss": 0.8059,
      "step": 24180
    },
    {
      "epoch": 1.5954457763780567,
      "grad_norm": 0.581540584564209,
      "learning_rate": 5.979621730576135e-06,
      "loss": 0.8066,
      "step": 24190
    },
    {
      "epoch": 1.5961053308490114,
      "grad_norm": 0.593189001083374,
      "learning_rate": 5.960957894118499e-06,
      "loss": 0.8295,
      "step": 24200
    },
    {
      "epoch": 1.5967648853199665,
      "grad_norm": 0.5631604790687561,
      "learning_rate": 5.942319287001044e-06,
      "loss": 0.7807,
      "step": 24210
    },
    {
      "epoch": 1.5974244397909212,
      "grad_norm": 0.6551171541213989,
      "learning_rate": 5.923705933922502e-06,
      "loss": 0.7679,
      "step": 24220
    },
    {
      "epoch": 1.598083994261876,
      "grad_norm": 0.5785092115402222,
      "learning_rate": 5.905117859548104e-06,
      "loss": 0.8023,
      "step": 24230
    },
    {
      "epoch": 1.598743548732831,
      "grad_norm": 0.5633843541145325,
      "learning_rate": 5.8865550885096074e-06,
      "loss": 0.7987,
      "step": 24240
    },
    {
      "epoch": 1.5994031032037859,
      "grad_norm": 0.6311590075492859,
      "learning_rate": 5.868017645405221e-06,
      "loss": 0.7853,
      "step": 24250
    },
    {
      "epoch": 1.6000626576747408,
      "grad_norm": 0.656356155872345,
      "learning_rate": 5.849505554799595e-06,
      "loss": 0.7754,
      "step": 24260
    },
    {
      "epoch": 1.6007222121456954,
      "grad_norm": 0.7022365927696228,
      "learning_rate": 5.831018841223795e-06,
      "loss": 0.8057,
      "step": 24270
    },
    {
      "epoch": 1.6013817666166505,
      "grad_norm": 0.6113805174827576,
      "learning_rate": 5.812557529175239e-06,
      "loss": 0.7862,
      "step": 24280
    },
    {
      "epoch": 1.6020413210876052,
      "grad_norm": 0.5850211381912231,
      "learning_rate": 5.794121643117711e-06,
      "loss": 0.7823,
      "step": 24290
    },
    {
      "epoch": 1.6027008755585603,
      "grad_norm": 0.5751693248748779,
      "learning_rate": 5.775711207481277e-06,
      "loss": 0.8013,
      "step": 24300
    },
    {
      "epoch": 1.603360430029515,
      "grad_norm": 0.612724781036377,
      "learning_rate": 5.757326246662281e-06,
      "loss": 0.8005,
      "step": 24310
    },
    {
      "epoch": 1.60401998450047,
      "grad_norm": 0.6902143955230713,
      "learning_rate": 5.738966785023331e-06,
      "loss": 0.8164,
      "step": 24320
    },
    {
      "epoch": 1.6046795389714248,
      "grad_norm": 0.5919621586799622,
      "learning_rate": 5.720632846893217e-06,
      "loss": 0.7999,
      "step": 24330
    },
    {
      "epoch": 1.6053390934423797,
      "grad_norm": 0.6050026416778564,
      "learning_rate": 5.702324456566938e-06,
      "loss": 0.8025,
      "step": 24340
    },
    {
      "epoch": 1.6059986479133346,
      "grad_norm": 0.6934067606925964,
      "learning_rate": 5.684041638305593e-06,
      "loss": 0.816,
      "step": 24350
    },
    {
      "epoch": 1.6066582023842892,
      "grad_norm": 0.6082236170768738,
      "learning_rate": 5.665784416336445e-06,
      "loss": 0.8474,
      "step": 24360
    },
    {
      "epoch": 1.6073177568552444,
      "grad_norm": 0.6929888129234314,
      "learning_rate": 5.6475528148528e-06,
      "loss": 0.8254,
      "step": 24370
    },
    {
      "epoch": 1.607977311326199,
      "grad_norm": 0.6141098141670227,
      "learning_rate": 5.6293468580140395e-06,
      "loss": 0.8182,
      "step": 24380
    },
    {
      "epoch": 1.6086368657971541,
      "grad_norm": 0.5887795090675354,
      "learning_rate": 5.61116656994555e-06,
      "loss": 0.8084,
      "step": 24390
    },
    {
      "epoch": 1.6092964202681088,
      "grad_norm": 0.5878868699073792,
      "learning_rate": 5.593011974738696e-06,
      "loss": 0.7687,
      "step": 24400
    },
    {
      "epoch": 1.6099559747390637,
      "grad_norm": 0.6481924653053284,
      "learning_rate": 5.57488309645082e-06,
      "loss": 0.841,
      "step": 24410
    },
    {
      "epoch": 1.6106155292100186,
      "grad_norm": 0.6095759272575378,
      "learning_rate": 5.556779959105152e-06,
      "loss": 0.7907,
      "step": 24420
    },
    {
      "epoch": 1.6112750836809735,
      "grad_norm": 0.5441837310791016,
      "learning_rate": 5.538702586690853e-06,
      "loss": 0.8129,
      "step": 24430
    },
    {
      "epoch": 1.6119346381519284,
      "grad_norm": 0.5181625485420227,
      "learning_rate": 5.520651003162891e-06,
      "loss": 0.7661,
      "step": 24440
    },
    {
      "epoch": 1.6125941926228833,
      "grad_norm": 0.6456611156463623,
      "learning_rate": 5.50262523244211e-06,
      "loss": 0.8086,
      "step": 24450
    },
    {
      "epoch": 1.6132537470938382,
      "grad_norm": 0.5870319604873657,
      "learning_rate": 5.484625298415114e-06,
      "loss": 0.8152,
      "step": 24460
    },
    {
      "epoch": 1.6139133015647928,
      "grad_norm": 0.6919241547584534,
      "learning_rate": 5.466651224934294e-06,
      "loss": 0.7845,
      "step": 24470
    },
    {
      "epoch": 1.614572856035748,
      "grad_norm": 0.6638587117195129,
      "learning_rate": 5.448703035817756e-06,
      "loss": 0.8435,
      "step": 24480
    },
    {
      "epoch": 1.6152324105067026,
      "grad_norm": 0.5628007650375366,
      "learning_rate": 5.430780754849301e-06,
      "loss": 0.8099,
      "step": 24490
    },
    {
      "epoch": 1.6158919649776577,
      "grad_norm": 0.5648647546768188,
      "learning_rate": 5.412884405778424e-06,
      "loss": 0.8164,
      "step": 24500
    },
    {
      "epoch": 1.6165515194486124,
      "grad_norm": 0.6228564977645874,
      "learning_rate": 5.395014012320232e-06,
      "loss": 0.811,
      "step": 24510
    },
    {
      "epoch": 1.6172110739195673,
      "grad_norm": 0.5722883939743042,
      "learning_rate": 5.377169598155441e-06,
      "loss": 0.8016,
      "step": 24520
    },
    {
      "epoch": 1.6178706283905222,
      "grad_norm": 0.5846453309059143,
      "learning_rate": 5.3593511869303535e-06,
      "loss": 0.7895,
      "step": 24530
    },
    {
      "epoch": 1.618530182861477,
      "grad_norm": 0.5563955903053284,
      "learning_rate": 5.341558802256799e-06,
      "loss": 0.853,
      "step": 24540
    },
    {
      "epoch": 1.619189737332432,
      "grad_norm": 0.589167594909668,
      "learning_rate": 5.323792467712135e-06,
      "loss": 0.785,
      "step": 24550
    },
    {
      "epoch": 1.6198492918033867,
      "grad_norm": 0.6283437013626099,
      "learning_rate": 5.306052206839188e-06,
      "loss": 0.8389,
      "step": 24560
    },
    {
      "epoch": 1.6205088462743418,
      "grad_norm": 0.5898224115371704,
      "learning_rate": 5.288338043146229e-06,
      "loss": 0.8151,
      "step": 24570
    },
    {
      "epoch": 1.6211684007452964,
      "grad_norm": 0.6777439117431641,
      "learning_rate": 5.27065000010695e-06,
      "loss": 0.818,
      "step": 24580
    },
    {
      "epoch": 1.6218279552162516,
      "grad_norm": 0.5743446946144104,
      "learning_rate": 5.252988101160442e-06,
      "loss": 0.8047,
      "step": 24590
    },
    {
      "epoch": 1.6224875096872062,
      "grad_norm": 0.6315315365791321,
      "learning_rate": 5.235352369711141e-06,
      "loss": 0.8442,
      "step": 24600
    },
    {
      "epoch": 1.6231470641581611,
      "grad_norm": 0.5853379368782043,
      "learning_rate": 5.217742829128797e-06,
      "loss": 0.8024,
      "step": 24610
    },
    {
      "epoch": 1.623806618629116,
      "grad_norm": 0.6441596746444702,
      "learning_rate": 5.200159502748478e-06,
      "loss": 0.7766,
      "step": 24620
    },
    {
      "epoch": 1.624466173100071,
      "grad_norm": 0.5711107850074768,
      "learning_rate": 5.182602413870491e-06,
      "loss": 0.8099,
      "step": 24630
    },
    {
      "epoch": 1.6251257275710258,
      "grad_norm": 0.5894733667373657,
      "learning_rate": 5.165071585760398e-06,
      "loss": 0.8102,
      "step": 24640
    },
    {
      "epoch": 1.6257852820419807,
      "grad_norm": 0.6057029962539673,
      "learning_rate": 5.147567041648937e-06,
      "loss": 0.7781,
      "step": 24650
    },
    {
      "epoch": 1.6264448365129356,
      "grad_norm": 0.7251249551773071,
      "learning_rate": 5.130088804732036e-06,
      "loss": 0.7977,
      "step": 24660
    },
    {
      "epoch": 1.6271043909838903,
      "grad_norm": 0.6940408945083618,
      "learning_rate": 5.112636898170742e-06,
      "loss": 0.8161,
      "step": 24670
    },
    {
      "epoch": 1.6277639454548454,
      "grad_norm": 0.5553697347640991,
      "learning_rate": 5.0952113450912355e-06,
      "loss": 0.7842,
      "step": 24680
    },
    {
      "epoch": 1.6284234999258,
      "grad_norm": 0.5917607545852661,
      "learning_rate": 5.077812168584761e-06,
      "loss": 0.7922,
      "step": 24690
    },
    {
      "epoch": 1.629083054396755,
      "grad_norm": 0.6597845554351807,
      "learning_rate": 5.0604393917076055e-06,
      "loss": 0.8443,
      "step": 24700
    },
    {
      "epoch": 1.6297426088677098,
      "grad_norm": 0.6031802296638489,
      "learning_rate": 5.043093037481086e-06,
      "loss": 0.783,
      "step": 24710
    },
    {
      "epoch": 1.6304021633386647,
      "grad_norm": 0.5948882699012756,
      "learning_rate": 5.025773128891492e-06,
      "loss": 0.8233,
      "step": 24720
    },
    {
      "epoch": 1.6310617178096196,
      "grad_norm": 0.6422739028930664,
      "learning_rate": 5.008479688890089e-06,
      "loss": 0.8248,
      "step": 24730
    },
    {
      "epoch": 1.6317212722805745,
      "grad_norm": 0.6182518005371094,
      "learning_rate": 4.991212740393045e-06,
      "loss": 0.8165,
      "step": 24740
    },
    {
      "epoch": 1.6323808267515294,
      "grad_norm": 0.6159992814064026,
      "learning_rate": 4.9739723062814295e-06,
      "loss": 0.8042,
      "step": 24750
    },
    {
      "epoch": 1.633040381222484,
      "grad_norm": 0.5987078547477722,
      "learning_rate": 4.956758409401194e-06,
      "loss": 0.7775,
      "step": 24760
    },
    {
      "epoch": 1.6336999356934392,
      "grad_norm": 0.6602343916893005,
      "learning_rate": 4.939571072563096e-06,
      "loss": 0.8174,
      "step": 24770
    },
    {
      "epoch": 1.6343594901643939,
      "grad_norm": 0.6065938472747803,
      "learning_rate": 4.922410318542733e-06,
      "loss": 0.7932,
      "step": 24780
    },
    {
      "epoch": 1.635019044635349,
      "grad_norm": 0.6987325549125671,
      "learning_rate": 4.905276170080431e-06,
      "loss": 0.7737,
      "step": 24790
    },
    {
      "epoch": 1.6356785991063036,
      "grad_norm": 0.5951793193817139,
      "learning_rate": 4.8881686498813e-06,
      "loss": 0.7763,
      "step": 24800
    },
    {
      "epoch": 1.6363381535772585,
      "grad_norm": 0.6437860727310181,
      "learning_rate": 4.871087780615144e-06,
      "loss": 0.8006,
      "step": 24810
    },
    {
      "epoch": 1.6369977080482134,
      "grad_norm": 0.5470296144485474,
      "learning_rate": 4.854033584916462e-06,
      "loss": 0.8303,
      "step": 24820
    },
    {
      "epoch": 1.6376572625191683,
      "grad_norm": 0.5686923265457153,
      "learning_rate": 4.837006085384399e-06,
      "loss": 0.8304,
      "step": 24830
    },
    {
      "epoch": 1.6383168169901232,
      "grad_norm": 0.6447607278823853,
      "learning_rate": 4.820005304582714e-06,
      "loss": 0.7821,
      "step": 24840
    },
    {
      "epoch": 1.6389763714610779,
      "grad_norm": 0.6381845474243164,
      "learning_rate": 4.80303126503979e-06,
      "loss": 0.8364,
      "step": 24850
    },
    {
      "epoch": 1.639635925932033,
      "grad_norm": 0.4987076222896576,
      "learning_rate": 4.786083989248538e-06,
      "loss": 0.8175,
      "step": 24860
    },
    {
      "epoch": 1.6402954804029877,
      "grad_norm": 0.6408962607383728,
      "learning_rate": 4.769163499666443e-06,
      "loss": 0.8169,
      "step": 24870
    },
    {
      "epoch": 1.6409550348739428,
      "grad_norm": 0.5356785655021667,
      "learning_rate": 4.752269818715446e-06,
      "loss": 0.7907,
      "step": 24880
    },
    {
      "epoch": 1.6416145893448975,
      "grad_norm": 0.5934824347496033,
      "learning_rate": 4.73540296878201e-06,
      "loss": 0.8481,
      "step": 24890
    },
    {
      "epoch": 1.6422741438158523,
      "grad_norm": 0.6478430032730103,
      "learning_rate": 4.718562972217011e-06,
      "loss": 0.7804,
      "step": 24900
    },
    {
      "epoch": 1.6429336982868072,
      "grad_norm": 0.5883824229240417,
      "learning_rate": 4.7017498513357545e-06,
      "loss": 0.788,
      "step": 24910
    },
    {
      "epoch": 1.6435932527577621,
      "grad_norm": 0.681681215763092,
      "learning_rate": 4.684963628417949e-06,
      "loss": 0.7992,
      "step": 24920
    },
    {
      "epoch": 1.644252807228717,
      "grad_norm": 0.6568331718444824,
      "learning_rate": 4.668204325707609e-06,
      "loss": 0.7977,
      "step": 24930
    },
    {
      "epoch": 1.644912361699672,
      "grad_norm": 0.5915022492408752,
      "learning_rate": 4.6514719654131286e-06,
      "loss": 0.7947,
      "step": 24940
    },
    {
      "epoch": 1.6455719161706268,
      "grad_norm": 0.619634211063385,
      "learning_rate": 4.634766569707169e-06,
      "loss": 0.8562,
      "step": 24950
    },
    {
      "epoch": 1.6462314706415815,
      "grad_norm": 0.6007986068725586,
      "learning_rate": 4.618088160726677e-06,
      "loss": 0.7471,
      "step": 24960
    },
    {
      "epoch": 1.6468910251125366,
      "grad_norm": 0.6730979681015015,
      "learning_rate": 4.601436760572831e-06,
      "loss": 0.8011,
      "step": 24970
    },
    {
      "epoch": 1.6475505795834913,
      "grad_norm": 0.5380508899688721,
      "learning_rate": 4.584812391311003e-06,
      "loss": 0.8208,
      "step": 24980
    },
    {
      "epoch": 1.6482101340544464,
      "grad_norm": 0.6031770706176758,
      "learning_rate": 4.568215074970778e-06,
      "loss": 0.791,
      "step": 24990
    },
    {
      "epoch": 1.648869688525401,
      "grad_norm": 0.615338146686554,
      "learning_rate": 4.551644833545868e-06,
      "loss": 0.8092,
      "step": 25000
    },
    {
      "epoch": 1.649529242996356,
      "grad_norm": 0.6194599866867065,
      "learning_rate": 4.5351016889941145e-06,
      "loss": 0.8308,
      "step": 25010
    },
    {
      "epoch": 1.6501887974673108,
      "grad_norm": 0.6836225390434265,
      "learning_rate": 4.518585663237446e-06,
      "loss": 0.825,
      "step": 25020
    },
    {
      "epoch": 1.6508483519382657,
      "grad_norm": 0.6452698111534119,
      "learning_rate": 4.502096778161871e-06,
      "loss": 0.7912,
      "step": 25030
    },
    {
      "epoch": 1.6515079064092206,
      "grad_norm": 0.636465311050415,
      "learning_rate": 4.48563505561741e-06,
      "loss": 0.8178,
      "step": 25040
    },
    {
      "epoch": 1.6521674608801753,
      "grad_norm": 0.5928276181221008,
      "learning_rate": 4.469200517418115e-06,
      "loss": 0.7823,
      "step": 25050
    },
    {
      "epoch": 1.6528270153511304,
      "grad_norm": 0.6056523323059082,
      "learning_rate": 4.452793185341994e-06,
      "loss": 0.8313,
      "step": 25060
    },
    {
      "epoch": 1.653486569822085,
      "grad_norm": 0.5615975260734558,
      "learning_rate": 4.436413081131005e-06,
      "loss": 0.8348,
      "step": 25070
    },
    {
      "epoch": 1.6541461242930402,
      "grad_norm": 0.638456404209137,
      "learning_rate": 4.420060226491041e-06,
      "loss": 0.8151,
      "step": 25080
    },
    {
      "epoch": 1.6548056787639949,
      "grad_norm": 0.6386964917182922,
      "learning_rate": 4.403734643091875e-06,
      "loss": 0.8095,
      "step": 25090
    },
    {
      "epoch": 1.6554652332349498,
      "grad_norm": 0.6466909646987915,
      "learning_rate": 4.38743635256714e-06,
      "loss": 0.797,
      "step": 25100
    },
    {
      "epoch": 1.6561247877059047,
      "grad_norm": 0.6175558567047119,
      "learning_rate": 4.371165376514297e-06,
      "loss": 0.8249,
      "step": 25110
    },
    {
      "epoch": 1.6567843421768595,
      "grad_norm": 0.6386156678199768,
      "learning_rate": 4.35492173649463e-06,
      "loss": 0.7941,
      "step": 25120
    },
    {
      "epoch": 1.6574438966478144,
      "grad_norm": 0.6918299198150635,
      "learning_rate": 4.338705454033182e-06,
      "loss": 0.8136,
      "step": 25130
    },
    {
      "epoch": 1.6581034511187691,
      "grad_norm": 0.7022925615310669,
      "learning_rate": 4.322516550618757e-06,
      "loss": 0.8031,
      "step": 25140
    },
    {
      "epoch": 1.6587630055897242,
      "grad_norm": 0.6506940126419067,
      "learning_rate": 4.306355047703867e-06,
      "loss": 0.7453,
      "step": 25150
    },
    {
      "epoch": 1.659422560060679,
      "grad_norm": 0.6400471329689026,
      "learning_rate": 4.290220966704711e-06,
      "loss": 0.7688,
      "step": 25160
    },
    {
      "epoch": 1.660082114531634,
      "grad_norm": 0.6685969829559326,
      "learning_rate": 4.274114329001172e-06,
      "loss": 0.7827,
      "step": 25170
    },
    {
      "epoch": 1.6607416690025887,
      "grad_norm": 0.5573239326477051,
      "learning_rate": 4.258035155936743e-06,
      "loss": 0.8142,
      "step": 25180
    },
    {
      "epoch": 1.6614012234735436,
      "grad_norm": 0.5721292495727539,
      "learning_rate": 4.24198346881853e-06,
      "loss": 0.8086,
      "step": 25190
    },
    {
      "epoch": 1.6620607779444985,
      "grad_norm": 0.6257614493370056,
      "learning_rate": 4.225959288917228e-06,
      "loss": 0.8277,
      "step": 25200
    },
    {
      "epoch": 1.6627203324154534,
      "grad_norm": 0.6866161823272705,
      "learning_rate": 4.209962637467058e-06,
      "loss": 0.8228,
      "step": 25210
    },
    {
      "epoch": 1.6633798868864083,
      "grad_norm": 0.7327253818511963,
      "learning_rate": 4.193993535665791e-06,
      "loss": 0.7469,
      "step": 25220
    },
    {
      "epoch": 1.6640394413573631,
      "grad_norm": 0.6483702063560486,
      "learning_rate": 4.178052004674665e-06,
      "loss": 0.7842,
      "step": 25230
    },
    {
      "epoch": 1.664698995828318,
      "grad_norm": 0.634652316570282,
      "learning_rate": 4.162138065618401e-06,
      "loss": 0.8013,
      "step": 25240
    },
    {
      "epoch": 1.6653585502992727,
      "grad_norm": 0.6225734353065491,
      "learning_rate": 4.146251739585135e-06,
      "loss": 0.8043,
      "step": 25250
    },
    {
      "epoch": 1.6660181047702278,
      "grad_norm": 0.6606937646865845,
      "learning_rate": 4.130393047626444e-06,
      "loss": 0.8123,
      "step": 25260
    },
    {
      "epoch": 1.6666776592411825,
      "grad_norm": 0.6344702243804932,
      "learning_rate": 4.1145620107572614e-06,
      "loss": 0.7918,
      "step": 25270
    },
    {
      "epoch": 1.6673372137121376,
      "grad_norm": 0.6618313789367676,
      "learning_rate": 4.098758649955872e-06,
      "loss": 0.8129,
      "step": 25280
    },
    {
      "epoch": 1.6679967681830923,
      "grad_norm": 0.7696662545204163,
      "learning_rate": 4.082982986163908e-06,
      "loss": 0.7915,
      "step": 25290
    },
    {
      "epoch": 1.6686563226540472,
      "grad_norm": 0.5740074515342712,
      "learning_rate": 4.067235040286277e-06,
      "loss": 0.8715,
      "step": 25300
    },
    {
      "epoch": 1.669315877125002,
      "grad_norm": 0.680923581123352,
      "learning_rate": 4.051514833191172e-06,
      "loss": 0.787,
      "step": 25310
    },
    {
      "epoch": 1.669975431595957,
      "grad_norm": 0.6620702743530273,
      "learning_rate": 4.0358223857100226e-06,
      "loss": 0.8038,
      "step": 25320
    },
    {
      "epoch": 1.6706349860669119,
      "grad_norm": 0.5888247489929199,
      "learning_rate": 4.020157718637469e-06,
      "loss": 0.7963,
      "step": 25330
    },
    {
      "epoch": 1.6712945405378665,
      "grad_norm": 0.5931589603424072,
      "learning_rate": 4.0045208527313335e-06,
      "loss": 0.8,
      "step": 25340
    },
    {
      "epoch": 1.6719540950088216,
      "grad_norm": 0.6023751497268677,
      "learning_rate": 3.988911808712617e-06,
      "loss": 0.7914,
      "step": 25350
    },
    {
      "epoch": 1.6726136494797763,
      "grad_norm": 0.6235506534576416,
      "learning_rate": 3.973330607265452e-06,
      "loss": 0.7857,
      "step": 25360
    },
    {
      "epoch": 1.6732732039507314,
      "grad_norm": 0.5739419460296631,
      "learning_rate": 3.957777269037042e-06,
      "loss": 0.7726,
      "step": 25370
    },
    {
      "epoch": 1.673932758421686,
      "grad_norm": 0.6558060050010681,
      "learning_rate": 3.942251814637709e-06,
      "loss": 0.8036,
      "step": 25380
    },
    {
      "epoch": 1.674592312892641,
      "grad_norm": 0.6836333274841309,
      "learning_rate": 3.926754264640798e-06,
      "loss": 0.7994,
      "step": 25390
    },
    {
      "epoch": 1.6752518673635959,
      "grad_norm": 0.625515341758728,
      "learning_rate": 3.911284639582699e-06,
      "loss": 0.7524,
      "step": 25400
    },
    {
      "epoch": 1.6759114218345508,
      "grad_norm": 0.5542259216308594,
      "learning_rate": 3.8958429599627755e-06,
      "loss": 0.8588,
      "step": 25410
    },
    {
      "epoch": 1.6765709763055057,
      "grad_norm": 0.6386640071868896,
      "learning_rate": 3.880429246243367e-06,
      "loss": 0.8087,
      "step": 25420
    },
    {
      "epoch": 1.6772305307764603,
      "grad_norm": 0.5683929920196533,
      "learning_rate": 3.865043518849767e-06,
      "loss": 0.7986,
      "step": 25430
    },
    {
      "epoch": 1.6778900852474155,
      "grad_norm": 0.6959534883499146,
      "learning_rate": 3.849685798170163e-06,
      "loss": 0.8188,
      "step": 25440
    },
    {
      "epoch": 1.6785496397183701,
      "grad_norm": 0.5607390999794006,
      "learning_rate": 3.834356104555659e-06,
      "loss": 0.8532,
      "step": 25450
    },
    {
      "epoch": 1.6792091941893252,
      "grad_norm": 0.599860668182373,
      "learning_rate": 3.81905445832017e-06,
      "loss": 0.8207,
      "step": 25460
    },
    {
      "epoch": 1.67986874866028,
      "grad_norm": 0.5843977332115173,
      "learning_rate": 3.8037808797404954e-06,
      "loss": 0.7927,
      "step": 25470
    },
    {
      "epoch": 1.6805283031312348,
      "grad_norm": 0.5897167921066284,
      "learning_rate": 3.788535389056211e-06,
      "loss": 0.7981,
      "step": 25480
    },
    {
      "epoch": 1.6811878576021897,
      "grad_norm": 0.6084034442901611,
      "learning_rate": 3.7733180064696893e-06,
      "loss": 0.815,
      "step": 25490
    },
    {
      "epoch": 1.6818474120731446,
      "grad_norm": 0.6056513786315918,
      "learning_rate": 3.758128752146045e-06,
      "loss": 0.8144,
      "step": 25500
    },
    {
      "epoch": 1.6825069665440995,
      "grad_norm": 0.5559295415878296,
      "learning_rate": 3.7429676462131136e-06,
      "loss": 0.766,
      "step": 25510
    },
    {
      "epoch": 1.6831665210150544,
      "grad_norm": 0.5697713494300842,
      "learning_rate": 3.727834708761452e-06,
      "loss": 0.7816,
      "step": 25520
    },
    {
      "epoch": 1.6838260754860093,
      "grad_norm": 0.6037575006484985,
      "learning_rate": 3.712729959844263e-06,
      "loss": 0.8125,
      "step": 25530
    },
    {
      "epoch": 1.684485629956964,
      "grad_norm": 0.5951526761054993,
      "learning_rate": 3.6976534194774287e-06,
      "loss": 0.8355,
      "step": 25540
    },
    {
      "epoch": 1.685145184427919,
      "grad_norm": 0.635144054889679,
      "learning_rate": 3.682605107639406e-06,
      "loss": 0.7882,
      "step": 25550
    },
    {
      "epoch": 1.6858047388988737,
      "grad_norm": 0.6611377000808716,
      "learning_rate": 3.667585044271288e-06,
      "loss": 0.7948,
      "step": 25560
    },
    {
      "epoch": 1.6864642933698288,
      "grad_norm": 0.5945231318473816,
      "learning_rate": 3.652593249276709e-06,
      "loss": 0.7924,
      "step": 25570
    },
    {
      "epoch": 1.6871238478407835,
      "grad_norm": 0.5711615085601807,
      "learning_rate": 3.6376297425218585e-06,
      "loss": 0.7651,
      "step": 25580
    },
    {
      "epoch": 1.6877834023117384,
      "grad_norm": 0.6354359984397888,
      "learning_rate": 3.622694543835431e-06,
      "loss": 0.8049,
      "step": 25590
    },
    {
      "epoch": 1.6884429567826933,
      "grad_norm": 0.6226228475570679,
      "learning_rate": 3.607787673008606e-06,
      "loss": 0.7954,
      "step": 25600
    },
    {
      "epoch": 1.6891025112536482,
      "grad_norm": 0.6653103232383728,
      "learning_rate": 3.592909149795043e-06,
      "loss": 0.7887,
      "step": 25610
    },
    {
      "epoch": 1.689762065724603,
      "grad_norm": 0.6024723649024963,
      "learning_rate": 3.57805899391081e-06,
      "loss": 0.7584,
      "step": 25620
    },
    {
      "epoch": 1.6904216201955578,
      "grad_norm": 0.5744298696517944,
      "learning_rate": 3.563237225034413e-06,
      "loss": 0.7976,
      "step": 25630
    },
    {
      "epoch": 1.6910811746665129,
      "grad_norm": 0.5463761687278748,
      "learning_rate": 3.548443862806719e-06,
      "loss": 0.7559,
      "step": 25640
    },
    {
      "epoch": 1.6917407291374675,
      "grad_norm": 0.6243197321891785,
      "learning_rate": 3.533678926830955e-06,
      "loss": 0.7711,
      "step": 25650
    },
    {
      "epoch": 1.6924002836084227,
      "grad_norm": 0.6674763560295105,
      "learning_rate": 3.518942436672701e-06,
      "loss": 0.8286,
      "step": 25660
    },
    {
      "epoch": 1.6930598380793773,
      "grad_norm": 0.6699982285499573,
      "learning_rate": 3.5042344118598137e-06,
      "loss": 0.8206,
      "step": 25670
    },
    {
      "epoch": 1.6937193925503322,
      "grad_norm": 0.5957103371620178,
      "learning_rate": 3.4895548718824443e-06,
      "loss": 0.7995,
      "step": 25680
    },
    {
      "epoch": 1.694378947021287,
      "grad_norm": 0.7029277682304382,
      "learning_rate": 3.4749038361929913e-06,
      "loss": 0.7842,
      "step": 25690
    },
    {
      "epoch": 1.695038501492242,
      "grad_norm": 0.5681647062301636,
      "learning_rate": 3.460281324206094e-06,
      "loss": 0.7992,
      "step": 25700
    },
    {
      "epoch": 1.695698055963197,
      "grad_norm": 0.7148775458335876,
      "learning_rate": 3.4456873552985743e-06,
      "loss": 0.7899,
      "step": 25710
    },
    {
      "epoch": 1.6963576104341518,
      "grad_norm": 0.5896807312965393,
      "learning_rate": 3.4311219488094544e-06,
      "loss": 0.7733,
      "step": 25720
    },
    {
      "epoch": 1.6970171649051067,
      "grad_norm": 0.5539336204528809,
      "learning_rate": 3.4165851240398866e-06,
      "loss": 0.764,
      "step": 25730
    },
    {
      "epoch": 1.6976767193760613,
      "grad_norm": 0.6221079230308533,
      "learning_rate": 3.402076900253148e-06,
      "loss": 0.8234,
      "step": 25740
    },
    {
      "epoch": 1.6983362738470165,
      "grad_norm": 0.6422823667526245,
      "learning_rate": 3.387597296674641e-06,
      "loss": 0.8512,
      "step": 25750
    },
    {
      "epoch": 1.6989958283179711,
      "grad_norm": 0.6487817168235779,
      "learning_rate": 3.373146332491814e-06,
      "loss": 0.7879,
      "step": 25760
    },
    {
      "epoch": 1.699655382788926,
      "grad_norm": 0.6562302708625793,
      "learning_rate": 3.3587240268541762e-06,
      "loss": 0.7887,
      "step": 25770
    },
    {
      "epoch": 1.700314937259881,
      "grad_norm": 0.5367435812950134,
      "learning_rate": 3.3443303988732578e-06,
      "loss": 0.7913,
      "step": 25780
    },
    {
      "epoch": 1.7009744917308358,
      "grad_norm": 0.6663900017738342,
      "learning_rate": 3.3299654676225867e-06,
      "loss": 0.795,
      "step": 25790
    },
    {
      "epoch": 1.7016340462017907,
      "grad_norm": 0.6704986095428467,
      "learning_rate": 3.315629252137678e-06,
      "loss": 0.8316,
      "step": 25800
    },
    {
      "epoch": 1.7022936006727456,
      "grad_norm": 0.5732877254486084,
      "learning_rate": 3.3013217714159715e-06,
      "loss": 0.8099,
      "step": 25810
    },
    {
      "epoch": 1.7029531551437005,
      "grad_norm": 0.587375819683075,
      "learning_rate": 3.287043044416843e-06,
      "loss": 0.7608,
      "step": 25820
    },
    {
      "epoch": 1.7036127096146552,
      "grad_norm": 0.552054762840271,
      "learning_rate": 3.2727930900615572e-06,
      "loss": 0.8175,
      "step": 25830
    },
    {
      "epoch": 1.7042722640856103,
      "grad_norm": 0.6114928126335144,
      "learning_rate": 3.2585719272332643e-06,
      "loss": 0.7723,
      "step": 25840
    },
    {
      "epoch": 1.704931818556565,
      "grad_norm": 0.6393599510192871,
      "learning_rate": 3.244379574776954e-06,
      "loss": 0.7694,
      "step": 25850
    },
    {
      "epoch": 1.70559137302752,
      "grad_norm": 0.5469321012496948,
      "learning_rate": 3.2302160514994296e-06,
      "loss": 0.8425,
      "step": 25860
    },
    {
      "epoch": 1.7062509274984747,
      "grad_norm": 0.760794460773468,
      "learning_rate": 3.216081376169311e-06,
      "loss": 0.8194,
      "step": 25870
    },
    {
      "epoch": 1.7069104819694296,
      "grad_norm": 0.6197438836097717,
      "learning_rate": 3.201975567516974e-06,
      "loss": 0.816,
      "step": 25880
    },
    {
      "epoch": 1.7075700364403845,
      "grad_norm": 0.6593007445335388,
      "learning_rate": 3.187898644234552e-06,
      "loss": 0.8373,
      "step": 25890
    },
    {
      "epoch": 1.7082295909113394,
      "grad_norm": 0.596043050289154,
      "learning_rate": 3.1738506249758966e-06,
      "loss": 0.7892,
      "step": 25900
    },
    {
      "epoch": 1.7088891453822943,
      "grad_norm": 0.6729983687400818,
      "learning_rate": 3.1598315283565605e-06,
      "loss": 0.8246,
      "step": 25910
    },
    {
      "epoch": 1.709548699853249,
      "grad_norm": 0.592121422290802,
      "learning_rate": 3.145841372953759e-06,
      "loss": 0.808,
      "step": 25920
    },
    {
      "epoch": 1.710208254324204,
      "grad_norm": 0.5267177820205688,
      "learning_rate": 3.131880177306376e-06,
      "loss": 0.7904,
      "step": 25930
    },
    {
      "epoch": 1.7108678087951588,
      "grad_norm": 0.682929277420044,
      "learning_rate": 3.1179479599149077e-06,
      "loss": 0.7997,
      "step": 25940
    },
    {
      "epoch": 1.7115273632661139,
      "grad_norm": 0.6856069564819336,
      "learning_rate": 3.1040447392414434e-06,
      "loss": 0.7882,
      "step": 25950
    },
    {
      "epoch": 1.7121869177370685,
      "grad_norm": 0.5914722681045532,
      "learning_rate": 3.090170533709666e-06,
      "loss": 0.8497,
      "step": 25960
    },
    {
      "epoch": 1.7128464722080234,
      "grad_norm": 0.5459190011024475,
      "learning_rate": 3.0763253617047904e-06,
      "loss": 0.8171,
      "step": 25970
    },
    {
      "epoch": 1.7135060266789783,
      "grad_norm": 0.6196326017379761,
      "learning_rate": 3.0625092415735775e-06,
      "loss": 0.7907,
      "step": 25980
    },
    {
      "epoch": 1.7141655811499332,
      "grad_norm": 0.5496552586555481,
      "learning_rate": 3.0487221916242733e-06,
      "loss": 0.7832,
      "step": 25990
    },
    {
      "epoch": 1.7148251356208881,
      "grad_norm": 0.5956296920776367,
      "learning_rate": 3.034964230126611e-06,
      "loss": 0.8034,
      "step": 26000
    },
    {
      "epoch": 1.715484690091843,
      "grad_norm": 0.5823291540145874,
      "learning_rate": 3.021235375311768e-06,
      "loss": 0.7987,
      "step": 26010
    },
    {
      "epoch": 1.716144244562798,
      "grad_norm": 0.6842591166496277,
      "learning_rate": 3.0075356453723667e-06,
      "loss": 0.808,
      "step": 26020
    },
    {
      "epoch": 1.7168037990337526,
      "grad_norm": 0.7035054564476013,
      "learning_rate": 2.9938650584624333e-06,
      "loss": 0.7812,
      "step": 26030
    },
    {
      "epoch": 1.7174633535047077,
      "grad_norm": 0.5906645059585571,
      "learning_rate": 2.9802236326973487e-06,
      "loss": 0.7919,
      "step": 26040
    },
    {
      "epoch": 1.7181229079756624,
      "grad_norm": 0.5931512117385864,
      "learning_rate": 2.966611386153892e-06,
      "loss": 0.8114,
      "step": 26050
    },
    {
      "epoch": 1.7187824624466175,
      "grad_norm": 0.6326853632926941,
      "learning_rate": 2.9530283368701394e-06,
      "loss": 0.7874,
      "step": 26060
    },
    {
      "epoch": 1.7194420169175721,
      "grad_norm": 0.5746793150901794,
      "learning_rate": 2.9394745028455045e-06,
      "loss": 0.786,
      "step": 26070
    },
    {
      "epoch": 1.720101571388527,
      "grad_norm": 0.6155476570129395,
      "learning_rate": 2.9259499020406726e-06,
      "loss": 0.8124,
      "step": 26080
    },
    {
      "epoch": 1.720761125859482,
      "grad_norm": 0.6925846338272095,
      "learning_rate": 2.912454552377586e-06,
      "loss": 0.8228,
      "step": 26090
    },
    {
      "epoch": 1.7214206803304368,
      "grad_norm": 0.6041793823242188,
      "learning_rate": 2.8989884717394444e-06,
      "loss": 0.8044,
      "step": 26100
    },
    {
      "epoch": 1.7220802348013917,
      "grad_norm": 0.5684294700622559,
      "learning_rate": 2.8855516779706382e-06,
      "loss": 0.787,
      "step": 26110
    },
    {
      "epoch": 1.7227397892723464,
      "grad_norm": 0.5601034164428711,
      "learning_rate": 2.872144188876777e-06,
      "loss": 0.8011,
      "step": 26120
    },
    {
      "epoch": 1.7233993437433015,
      "grad_norm": 0.5804260969161987,
      "learning_rate": 2.8587660222246016e-06,
      "loss": 0.8106,
      "step": 26130
    },
    {
      "epoch": 1.7240588982142562,
      "grad_norm": 0.6315391063690186,
      "learning_rate": 2.8454171957420285e-06,
      "loss": 0.829,
      "step": 26140
    },
    {
      "epoch": 1.7247184526852113,
      "grad_norm": 0.5435738563537598,
      "learning_rate": 2.832097727118077e-06,
      "loss": 0.7895,
      "step": 26150
    },
    {
      "epoch": 1.725378007156166,
      "grad_norm": 0.6043064594268799,
      "learning_rate": 2.8188076340028735e-06,
      "loss": 0.7951,
      "step": 26160
    },
    {
      "epoch": 1.7260375616271209,
      "grad_norm": 0.593038022518158,
      "learning_rate": 2.8055469340076134e-06,
      "loss": 0.7747,
      "step": 26170
    },
    {
      "epoch": 1.7266971160980757,
      "grad_norm": 0.5918024778366089,
      "learning_rate": 2.7923156447045323e-06,
      "loss": 0.8139,
      "step": 26180
    },
    {
      "epoch": 1.7273566705690306,
      "grad_norm": 0.6056832671165466,
      "learning_rate": 2.779113783626916e-06,
      "loss": 0.7975,
      "step": 26190
    },
    {
      "epoch": 1.7280162250399855,
      "grad_norm": 0.6603617668151855,
      "learning_rate": 2.7659413682690234e-06,
      "loss": 0.8294,
      "step": 26200
    },
    {
      "epoch": 1.7286757795109402,
      "grad_norm": 0.6331311464309692,
      "learning_rate": 2.752798416086133e-06,
      "loss": 0.7895,
      "step": 26210
    },
    {
      "epoch": 1.7293353339818953,
      "grad_norm": 0.5651764869689941,
      "learning_rate": 2.7396849444944307e-06,
      "loss": 0.8386,
      "step": 26220
    },
    {
      "epoch": 1.72999488845285,
      "grad_norm": 0.6461435556411743,
      "learning_rate": 2.7266009708710766e-06,
      "loss": 0.8364,
      "step": 26230
    },
    {
      "epoch": 1.730654442923805,
      "grad_norm": 0.565617561340332,
      "learning_rate": 2.7135465125541354e-06,
      "loss": 0.8264,
      "step": 26240
    },
    {
      "epoch": 1.7313139973947598,
      "grad_norm": 0.6211186647415161,
      "learning_rate": 2.700521586842547e-06,
      "loss": 0.796,
      "step": 26250
    },
    {
      "epoch": 1.7319735518657147,
      "grad_norm": 0.6177099943161011,
      "learning_rate": 2.687526210996125e-06,
      "loss": 0.85,
      "step": 26260
    },
    {
      "epoch": 1.7326331063366696,
      "grad_norm": 0.6954526305198669,
      "learning_rate": 2.6745604022355153e-06,
      "loss": 0.8372,
      "step": 26270
    },
    {
      "epoch": 1.7332926608076245,
      "grad_norm": 0.6717360019683838,
      "learning_rate": 2.661624177742203e-06,
      "loss": 0.8176,
      "step": 26280
    },
    {
      "epoch": 1.7339522152785793,
      "grad_norm": 0.5314779281616211,
      "learning_rate": 2.6487175546584505e-06,
      "loss": 0.7842,
      "step": 26290
    },
    {
      "epoch": 1.7346117697495342,
      "grad_norm": 0.6311419606208801,
      "learning_rate": 2.635840550087307e-06,
      "loss": 0.8061,
      "step": 26300
    },
    {
      "epoch": 1.7352713242204891,
      "grad_norm": 0.6443468928337097,
      "learning_rate": 2.62299318109257e-06,
      "loss": 0.8335,
      "step": 26310
    },
    {
      "epoch": 1.7359308786914438,
      "grad_norm": 0.5492504835128784,
      "learning_rate": 2.610175464698755e-06,
      "loss": 0.8256,
      "step": 26320
    },
    {
      "epoch": 1.736590433162399,
      "grad_norm": 0.6499750018119812,
      "learning_rate": 2.5973874178911035e-06,
      "loss": 0.792,
      "step": 26330
    },
    {
      "epoch": 1.7372499876333536,
      "grad_norm": 0.5789830088615417,
      "learning_rate": 2.5846290576155258e-06,
      "loss": 0.7956,
      "step": 26340
    },
    {
      "epoch": 1.7379095421043087,
      "grad_norm": 0.646263062953949,
      "learning_rate": 2.5719004007785987e-06,
      "loss": 0.8111,
      "step": 26350
    },
    {
      "epoch": 1.7385690965752634,
      "grad_norm": 0.5217257738113403,
      "learning_rate": 2.559201464247532e-06,
      "loss": 0.8302,
      "step": 26360
    },
    {
      "epoch": 1.7392286510462183,
      "grad_norm": 0.5562120676040649,
      "learning_rate": 2.5465322648501655e-06,
      "loss": 0.82,
      "step": 26370
    },
    {
      "epoch": 1.7398882055171732,
      "grad_norm": 0.587908148765564,
      "learning_rate": 2.533892819374914e-06,
      "loss": 0.773,
      "step": 26380
    },
    {
      "epoch": 1.740547759988128,
      "grad_norm": 0.6600366234779358,
      "learning_rate": 2.521283144570788e-06,
      "loss": 0.7903,
      "step": 26390
    },
    {
      "epoch": 1.741207314459083,
      "grad_norm": 0.6248031258583069,
      "learning_rate": 2.5087032571473266e-06,
      "loss": 0.7825,
      "step": 26400
    },
    {
      "epoch": 1.7418668689300376,
      "grad_norm": 0.615217924118042,
      "learning_rate": 2.496153173774604e-06,
      "loss": 0.8362,
      "step": 26410
    },
    {
      "epoch": 1.7425264234009927,
      "grad_norm": 0.6468397378921509,
      "learning_rate": 2.4836329110832086e-06,
      "loss": 0.7988,
      "step": 26420
    },
    {
      "epoch": 1.7431859778719474,
      "grad_norm": 0.6291540861129761,
      "learning_rate": 2.4711424856641964e-06,
      "loss": 0.8245,
      "step": 26430
    },
    {
      "epoch": 1.7438455323429025,
      "grad_norm": 0.5626056790351868,
      "learning_rate": 2.4586819140690943e-06,
      "loss": 0.7779,
      "step": 26440
    },
    {
      "epoch": 1.7445050868138572,
      "grad_norm": 0.6262189149856567,
      "learning_rate": 2.4462512128098717e-06,
      "loss": 0.7705,
      "step": 26450
    },
    {
      "epoch": 1.745164641284812,
      "grad_norm": 0.60690838098526,
      "learning_rate": 2.4338503983589026e-06,
      "loss": 0.769,
      "step": 26460
    },
    {
      "epoch": 1.745824195755767,
      "grad_norm": 0.5185971856117249,
      "learning_rate": 2.421479487148978e-06,
      "loss": 0.7812,
      "step": 26470
    },
    {
      "epoch": 1.7464837502267219,
      "grad_norm": 0.6107097268104553,
      "learning_rate": 2.409138495573246e-06,
      "loss": 0.7936,
      "step": 26480
    },
    {
      "epoch": 1.7471433046976768,
      "grad_norm": 0.624666690826416,
      "learning_rate": 2.396827439985208e-06,
      "loss": 0.8287,
      "step": 26490
    },
    {
      "epoch": 1.7478028591686314,
      "grad_norm": 0.573738694190979,
      "learning_rate": 2.3845463366987007e-06,
      "loss": 0.8372,
      "step": 26500
    },
    {
      "epoch": 1.7484624136395865,
      "grad_norm": 0.643282949924469,
      "learning_rate": 2.372295201987873e-06,
      "loss": 0.8416,
      "step": 26510
    },
    {
      "epoch": 1.7491219681105412,
      "grad_norm": 0.6036027073860168,
      "learning_rate": 2.360074052087155e-06,
      "loss": 0.7965,
      "step": 26520
    },
    {
      "epoch": 1.7497815225814963,
      "grad_norm": 0.668479323387146,
      "learning_rate": 2.3478829031912426e-06,
      "loss": 0.8128,
      "step": 26530
    },
    {
      "epoch": 1.750441077052451,
      "grad_norm": 0.560921847820282,
      "learning_rate": 2.3357217714550848e-06,
      "loss": 0.7554,
      "step": 26540
    },
    {
      "epoch": 1.751100631523406,
      "grad_norm": 0.5599755048751831,
      "learning_rate": 2.3235906729938417e-06,
      "loss": 0.8108,
      "step": 26550
    },
    {
      "epoch": 1.7517601859943608,
      "grad_norm": 0.6397618055343628,
      "learning_rate": 2.3114896238828925e-06,
      "loss": 0.8349,
      "step": 26560
    },
    {
      "epoch": 1.7524197404653157,
      "grad_norm": 0.6178655624389648,
      "learning_rate": 2.2994186401577716e-06,
      "loss": 0.791,
      "step": 26570
    },
    {
      "epoch": 1.7530792949362706,
      "grad_norm": 0.7424609661102295,
      "learning_rate": 2.2873777378141947e-06,
      "loss": 0.7657,
      "step": 26580
    },
    {
      "epoch": 1.7537388494072255,
      "grad_norm": 0.5651452541351318,
      "learning_rate": 2.275366932808004e-06,
      "loss": 0.7932,
      "step": 26590
    },
    {
      "epoch": 1.7543984038781804,
      "grad_norm": 0.6681116223335266,
      "learning_rate": 2.2633862410551683e-06,
      "loss": 0.8513,
      "step": 26600
    },
    {
      "epoch": 1.755057958349135,
      "grad_norm": 0.6477376818656921,
      "learning_rate": 2.251435678431743e-06,
      "loss": 0.7905,
      "step": 26610
    },
    {
      "epoch": 1.7557175128200901,
      "grad_norm": 0.5751684308052063,
      "learning_rate": 2.239515260773853e-06,
      "loss": 0.831,
      "step": 26620
    },
    {
      "epoch": 1.7563770672910448,
      "grad_norm": 0.5635110139846802,
      "learning_rate": 2.227625003877701e-06,
      "loss": 0.7815,
      "step": 26630
    },
    {
      "epoch": 1.757036621762,
      "grad_norm": 0.6373135447502136,
      "learning_rate": 2.2157649234994944e-06,
      "loss": 0.8066,
      "step": 26640
    },
    {
      "epoch": 1.7576961762329546,
      "grad_norm": 0.6144300103187561,
      "learning_rate": 2.2039350353554785e-06,
      "loss": 0.8136,
      "step": 26650
    },
    {
      "epoch": 1.7583557307039095,
      "grad_norm": 0.6329615712165833,
      "learning_rate": 2.1921353551218603e-06,
      "loss": 0.7747,
      "step": 26660
    },
    {
      "epoch": 1.7590152851748644,
      "grad_norm": 0.6645671725273132,
      "learning_rate": 2.1803658984348414e-06,
      "loss": 0.801,
      "step": 26670
    },
    {
      "epoch": 1.7596748396458193,
      "grad_norm": 0.6029779314994812,
      "learning_rate": 2.1686266808905676e-06,
      "loss": 0.8115,
      "step": 26680
    },
    {
      "epoch": 1.7603343941167742,
      "grad_norm": 0.7071924209594727,
      "learning_rate": 2.156917718045104e-06,
      "loss": 0.7652,
      "step": 26690
    },
    {
      "epoch": 1.7609939485877288,
      "grad_norm": 0.5901879668235779,
      "learning_rate": 2.1452390254144446e-06,
      "loss": 0.8238,
      "step": 26700
    },
    {
      "epoch": 1.761653503058684,
      "grad_norm": 0.6033055186271667,
      "learning_rate": 2.1335906184744374e-06,
      "loss": 0.7974,
      "step": 26710
    },
    {
      "epoch": 1.7623130575296386,
      "grad_norm": 0.6236467957496643,
      "learning_rate": 2.121972512660836e-06,
      "loss": 0.7641,
      "step": 26720
    },
    {
      "epoch": 1.7629726120005937,
      "grad_norm": 0.6290685534477234,
      "learning_rate": 2.11038472336921e-06,
      "loss": 0.7739,
      "step": 26730
    },
    {
      "epoch": 1.7636321664715484,
      "grad_norm": 0.6045517921447754,
      "learning_rate": 2.0988272659549808e-06,
      "loss": 0.8081,
      "step": 26740
    },
    {
      "epoch": 1.7642917209425033,
      "grad_norm": 0.6901437640190125,
      "learning_rate": 2.087300155733357e-06,
      "loss": 0.811,
      "step": 26750
    },
    {
      "epoch": 1.7649512754134582,
      "grad_norm": 0.5917714238166809,
      "learning_rate": 2.0758034079793347e-06,
      "loss": 0.8048,
      "step": 26760
    },
    {
      "epoch": 1.765610829884413,
      "grad_norm": 0.5725721716880798,
      "learning_rate": 2.064337037927688e-06,
      "loss": 0.7878,
      "step": 26770
    },
    {
      "epoch": 1.766270384355368,
      "grad_norm": 0.6000796556472778,
      "learning_rate": 2.0529010607729242e-06,
      "loss": 0.8137,
      "step": 26780
    },
    {
      "epoch": 1.7669299388263227,
      "grad_norm": 0.6986436247825623,
      "learning_rate": 2.0414954916692904e-06,
      "loss": 0.8074,
      "step": 26790
    },
    {
      "epoch": 1.7675894932972778,
      "grad_norm": 0.5333727598190308,
      "learning_rate": 2.03012034573071e-06,
      "loss": 0.7536,
      "step": 26800
    },
    {
      "epoch": 1.7682490477682324,
      "grad_norm": 0.6346567273139954,
      "learning_rate": 2.018775638030826e-06,
      "loss": 0.7852,
      "step": 26810
    },
    {
      "epoch": 1.7689086022391876,
      "grad_norm": 0.5741221308708191,
      "learning_rate": 2.0074613836029236e-06,
      "loss": 0.7825,
      "step": 26820
    },
    {
      "epoch": 1.7695681567101422,
      "grad_norm": 0.5961235761642456,
      "learning_rate": 1.9961775974399468e-06,
      "loss": 0.8353,
      "step": 26830
    },
    {
      "epoch": 1.7702277111810971,
      "grad_norm": 0.6274139285087585,
      "learning_rate": 1.984924294494453e-06,
      "loss": 0.7683,
      "step": 26840
    },
    {
      "epoch": 1.770887265652052,
      "grad_norm": 0.6172131896018982,
      "learning_rate": 1.9737014896786115e-06,
      "loss": 0.8453,
      "step": 26850
    },
    {
      "epoch": 1.771546820123007,
      "grad_norm": 0.7044884562492371,
      "learning_rate": 1.962509197864182e-06,
      "loss": 0.8254,
      "step": 26860
    },
    {
      "epoch": 1.7722063745939618,
      "grad_norm": 0.5985585451126099,
      "learning_rate": 1.9513474338824748e-06,
      "loss": 0.8301,
      "step": 26870
    },
    {
      "epoch": 1.7728659290649167,
      "grad_norm": 0.6298454403877258,
      "learning_rate": 1.9402162125243706e-06,
      "loss": 0.8077,
      "step": 26880
    },
    {
      "epoch": 1.7735254835358716,
      "grad_norm": 0.6074488162994385,
      "learning_rate": 1.929115548540253e-06,
      "loss": 0.7929,
      "step": 26890
    },
    {
      "epoch": 1.7741850380068263,
      "grad_norm": 0.682905912399292,
      "learning_rate": 1.9180454566400247e-06,
      "loss": 0.8093,
      "step": 26900
    },
    {
      "epoch": 1.7748445924777814,
      "grad_norm": 0.6187024712562561,
      "learning_rate": 1.9070059514930822e-06,
      "loss": 0.7925,
      "step": 26910
    },
    {
      "epoch": 1.775504146948736,
      "grad_norm": 0.7148305773735046,
      "learning_rate": 1.8959970477282735e-06,
      "loss": 0.7884,
      "step": 26920
    },
    {
      "epoch": 1.7761637014196912,
      "grad_norm": 0.6598269939422607,
      "learning_rate": 1.8850187599339096e-06,
      "loss": 0.8044,
      "step": 26930
    },
    {
      "epoch": 1.7768232558906458,
      "grad_norm": 0.6087578535079956,
      "learning_rate": 1.8740711026577206e-06,
      "loss": 0.8131,
      "step": 26940
    },
    {
      "epoch": 1.7774828103616007,
      "grad_norm": 0.6049370169639587,
      "learning_rate": 1.8631540904068629e-06,
      "loss": 0.7979,
      "step": 26950
    },
    {
      "epoch": 1.7781423648325556,
      "grad_norm": 0.5292810201644897,
      "learning_rate": 1.8522677376478642e-06,
      "loss": 0.8086,
      "step": 26960
    },
    {
      "epoch": 1.7788019193035105,
      "grad_norm": 0.5981762409210205,
      "learning_rate": 1.8414120588066436e-06,
      "loss": 0.7902,
      "step": 26970
    },
    {
      "epoch": 1.7794614737744654,
      "grad_norm": 0.6287865042686462,
      "learning_rate": 1.8305870682684578e-06,
      "loss": 0.8154,
      "step": 26980
    },
    {
      "epoch": 1.78012102824542,
      "grad_norm": 0.6655606627464294,
      "learning_rate": 1.819792780377899e-06,
      "loss": 0.7849,
      "step": 26990
    },
    {
      "epoch": 1.7807805827163752,
      "grad_norm": 0.6155992746353149,
      "learning_rate": 1.809029209438884e-06,
      "loss": 0.7921,
      "step": 27000
    },
    {
      "epoch": 1.7814401371873299,
      "grad_norm": 0.7183905243873596,
      "learning_rate": 1.798296369714622e-06,
      "loss": 0.8068,
      "step": 27010
    },
    {
      "epoch": 1.782099691658285,
      "grad_norm": 0.6222487092018127,
      "learning_rate": 1.7875942754275898e-06,
      "loss": 0.8344,
      "step": 27020
    },
    {
      "epoch": 1.7827592461292396,
      "grad_norm": 0.5872424244880676,
      "learning_rate": 1.776922940759529e-06,
      "loss": 0.8019,
      "step": 27030
    },
    {
      "epoch": 1.7834188006001945,
      "grad_norm": 0.48592233657836914,
      "learning_rate": 1.7662823798514249e-06,
      "loss": 0.8084,
      "step": 27040
    },
    {
      "epoch": 1.7840783550711494,
      "grad_norm": 0.6957306265830994,
      "learning_rate": 1.7556726068034795e-06,
      "loss": 0.7976,
      "step": 27050
    },
    {
      "epoch": 1.7847379095421043,
      "grad_norm": 0.5849895477294922,
      "learning_rate": 1.7450936356750907e-06,
      "loss": 0.7798,
      "step": 27060
    },
    {
      "epoch": 1.7853974640130592,
      "grad_norm": 0.6629789471626282,
      "learning_rate": 1.7345454804848504e-06,
      "loss": 0.7876,
      "step": 27070
    },
    {
      "epoch": 1.786057018484014,
      "grad_norm": 0.7732569575309753,
      "learning_rate": 1.7240281552105081e-06,
      "loss": 0.7914,
      "step": 27080
    },
    {
      "epoch": 1.786716572954969,
      "grad_norm": 0.6583324670791626,
      "learning_rate": 1.7135416737889625e-06,
      "loss": 0.8115,
      "step": 27090
    },
    {
      "epoch": 1.7873761274259237,
      "grad_norm": 0.6596052050590515,
      "learning_rate": 1.7030860501162371e-06,
      "loss": 0.8074,
      "step": 27100
    },
    {
      "epoch": 1.7880356818968788,
      "grad_norm": 0.5933216214179993,
      "learning_rate": 1.6926612980474655e-06,
      "loss": 0.8023,
      "step": 27110
    },
    {
      "epoch": 1.7886952363678335,
      "grad_norm": 0.6136610507965088,
      "learning_rate": 1.682267431396875e-06,
      "loss": 0.795,
      "step": 27120
    },
    {
      "epoch": 1.7893547908387883,
      "grad_norm": 0.6610358953475952,
      "learning_rate": 1.6719044639377595e-06,
      "loss": 0.7924,
      "step": 27130
    },
    {
      "epoch": 1.7900143453097432,
      "grad_norm": 0.607771098613739,
      "learning_rate": 1.6615724094024843e-06,
      "loss": 0.7755,
      "step": 27140
    },
    {
      "epoch": 1.7906738997806981,
      "grad_norm": 0.6132540106773376,
      "learning_rate": 1.6512712814824167e-06,
      "loss": 0.7878,
      "step": 27150
    },
    {
      "epoch": 1.791333454251653,
      "grad_norm": 0.6049851775169373,
      "learning_rate": 1.6410010938279818e-06,
      "loss": 0.7968,
      "step": 27160
    },
    {
      "epoch": 1.791993008722608,
      "grad_norm": 0.5582095980644226,
      "learning_rate": 1.6307618600485712e-06,
      "loss": 0.7967,
      "step": 27170
    },
    {
      "epoch": 1.7926525631935628,
      "grad_norm": 0.5795011520385742,
      "learning_rate": 1.6205535937125844e-06,
      "loss": 0.7944,
      "step": 27180
    },
    {
      "epoch": 1.7933121176645175,
      "grad_norm": 0.6117453575134277,
      "learning_rate": 1.6103763083473694e-06,
      "loss": 0.7785,
      "step": 27190
    },
    {
      "epoch": 1.7939716721354726,
      "grad_norm": 0.6477991938591003,
      "learning_rate": 1.6002300174392193e-06,
      "loss": 0.8261,
      "step": 27200
    },
    {
      "epoch": 1.7946312266064273,
      "grad_norm": 0.6463562846183777,
      "learning_rate": 1.5901147344333705e-06,
      "loss": 0.7661,
      "step": 27210
    },
    {
      "epoch": 1.7952907810773824,
      "grad_norm": 0.587986409664154,
      "learning_rate": 1.5800304727339482e-06,
      "loss": 0.8229,
      "step": 27220
    },
    {
      "epoch": 1.795950335548337,
      "grad_norm": 0.56048184633255,
      "learning_rate": 1.5699772457039936e-06,
      "loss": 0.8257,
      "step": 27230
    },
    {
      "epoch": 1.796609890019292,
      "grad_norm": 0.5849249958992004,
      "learning_rate": 1.5599550666653945e-06,
      "loss": 0.8239,
      "step": 27240
    },
    {
      "epoch": 1.7972694444902468,
      "grad_norm": 0.5611611008644104,
      "learning_rate": 1.5499639488989248e-06,
      "loss": 0.8773,
      "step": 27250
    },
    {
      "epoch": 1.7979289989612017,
      "grad_norm": 0.6698682904243469,
      "learning_rate": 1.5400039056441772e-06,
      "loss": 0.8124,
      "step": 27260
    },
    {
      "epoch": 1.7985885534321566,
      "grad_norm": 0.6456633806228638,
      "learning_rate": 1.530074950099583e-06,
      "loss": 0.7882,
      "step": 27270
    },
    {
      "epoch": 1.7992481079031113,
      "grad_norm": 0.642745852470398,
      "learning_rate": 1.520177095422362e-06,
      "loss": 0.8279,
      "step": 27280
    },
    {
      "epoch": 1.7999076623740664,
      "grad_norm": 0.5709918141365051,
      "learning_rate": 1.5103103547285252e-06,
      "loss": 0.7625,
      "step": 27290
    },
    {
      "epoch": 1.800567216845021,
      "grad_norm": 0.5563238859176636,
      "learning_rate": 1.5004747410928639e-06,
      "loss": 0.7995,
      "step": 27300
    },
    {
      "epoch": 1.8012267713159762,
      "grad_norm": 0.6466971039772034,
      "learning_rate": 1.4906702675489076e-06,
      "loss": 0.81,
      "step": 27310
    },
    {
      "epoch": 1.8018863257869309,
      "grad_norm": 0.6266707181930542,
      "learning_rate": 1.480896947088936e-06,
      "loss": 0.7501,
      "step": 27320
    },
    {
      "epoch": 1.8025458802578858,
      "grad_norm": 0.5325832962989807,
      "learning_rate": 1.4711547926639342e-06,
      "loss": 0.7601,
      "step": 27330
    },
    {
      "epoch": 1.8032054347288407,
      "grad_norm": 0.5419653654098511,
      "learning_rate": 1.4614438171835858e-06,
      "loss": 0.8051,
      "step": 27340
    },
    {
      "epoch": 1.8038649891997955,
      "grad_norm": 0.5886746048927307,
      "learning_rate": 1.451764033516273e-06,
      "loss": 0.8162,
      "step": 27350
    },
    {
      "epoch": 1.8045245436707504,
      "grad_norm": 0.6023467779159546,
      "learning_rate": 1.4421154544890293e-06,
      "loss": 0.7919,
      "step": 27360
    },
    {
      "epoch": 1.8051840981417053,
      "grad_norm": 0.6059507727622986,
      "learning_rate": 1.4324980928875547e-06,
      "loss": 0.7967,
      "step": 27370
    },
    {
      "epoch": 1.8058436526126602,
      "grad_norm": 0.5828139185905457,
      "learning_rate": 1.4229119614561555e-06,
      "loss": 0.8134,
      "step": 27380
    },
    {
      "epoch": 1.806503207083615,
      "grad_norm": 0.6271647810935974,
      "learning_rate": 1.4133570728977836e-06,
      "loss": 0.7957,
      "step": 27390
    },
    {
      "epoch": 1.80716276155457,
      "grad_norm": 0.5952603220939636,
      "learning_rate": 1.4038334398739666e-06,
      "loss": 0.7744,
      "step": 27400
    },
    {
      "epoch": 1.8078223160255247,
      "grad_norm": 0.5841818451881409,
      "learning_rate": 1.3943410750048291e-06,
      "loss": 0.8163,
      "step": 27410
    },
    {
      "epoch": 1.8084818704964798,
      "grad_norm": 0.583544135093689,
      "learning_rate": 1.3848799908690563e-06,
      "loss": 0.7842,
      "step": 27420
    },
    {
      "epoch": 1.8091414249674345,
      "grad_norm": 0.6535754203796387,
      "learning_rate": 1.375450200003875e-06,
      "loss": 0.8113,
      "step": 27430
    },
    {
      "epoch": 1.8098009794383894,
      "grad_norm": 0.5610626935958862,
      "learning_rate": 1.366051714905056e-06,
      "loss": 0.7657,
      "step": 27440
    },
    {
      "epoch": 1.8104605339093442,
      "grad_norm": 0.5700557827949524,
      "learning_rate": 1.3566845480268765e-06,
      "loss": 0.7881,
      "step": 27450
    },
    {
      "epoch": 1.8111200883802991,
      "grad_norm": 0.5087742805480957,
      "learning_rate": 1.3473487117821259e-06,
      "loss": 0.7937,
      "step": 27460
    },
    {
      "epoch": 1.811779642851254,
      "grad_norm": 0.5830112099647522,
      "learning_rate": 1.3380442185420505e-06,
      "loss": 0.8397,
      "step": 27470
    },
    {
      "epoch": 1.8124391973222087,
      "grad_norm": 0.5971534848213196,
      "learning_rate": 1.3287710806363958e-06,
      "loss": 0.8171,
      "step": 27480
    },
    {
      "epoch": 1.8130987517931638,
      "grad_norm": 0.6187090277671814,
      "learning_rate": 1.3195293103533263e-06,
      "loss": 0.7883,
      "step": 27490
    },
    {
      "epoch": 1.8137583062641185,
      "grad_norm": 0.5818140506744385,
      "learning_rate": 1.3103189199394628e-06,
      "loss": 0.8735,
      "step": 27500
    },
    {
      "epoch": 1.8144178607350736,
      "grad_norm": 0.6792356967926025,
      "learning_rate": 1.3011399215998338e-06,
      "loss": 0.7959,
      "step": 27510
    },
    {
      "epoch": 1.8150774152060283,
      "grad_norm": 0.5853057503700256,
      "learning_rate": 1.291992327497865e-06,
      "loss": 0.8166,
      "step": 27520
    },
    {
      "epoch": 1.8157369696769832,
      "grad_norm": 0.528980553150177,
      "learning_rate": 1.2828761497553825e-06,
      "loss": 0.7907,
      "step": 27530
    },
    {
      "epoch": 1.816396524147938,
      "grad_norm": 0.6840847730636597,
      "learning_rate": 1.273791400452562e-06,
      "loss": 0.7674,
      "step": 27540
    },
    {
      "epoch": 1.817056078618893,
      "grad_norm": 0.6091808676719666,
      "learning_rate": 1.2647380916279427e-06,
      "loss": 0.8216,
      "step": 27550
    },
    {
      "epoch": 1.8177156330898478,
      "grad_norm": 0.6639935374259949,
      "learning_rate": 1.2557162352784064e-06,
      "loss": 0.7533,
      "step": 27560
    },
    {
      "epoch": 1.8183751875608025,
      "grad_norm": 0.6252728700637817,
      "learning_rate": 1.2467258433591395e-06,
      "loss": 0.8153,
      "step": 27570
    },
    {
      "epoch": 1.8190347420317576,
      "grad_norm": 0.6612904071807861,
      "learning_rate": 1.2377669277836512e-06,
      "loss": 0.8006,
      "step": 27580
    },
    {
      "epoch": 1.8196942965027123,
      "grad_norm": 0.6393854022026062,
      "learning_rate": 1.2288395004237286e-06,
      "loss": 0.8181,
      "step": 27590
    },
    {
      "epoch": 1.8203538509736674,
      "grad_norm": 0.6946678757667542,
      "learning_rate": 1.219943573109439e-06,
      "loss": 0.8324,
      "step": 27600
    },
    {
      "epoch": 1.821013405444622,
      "grad_norm": 0.5817633867263794,
      "learning_rate": 1.2110791576290997e-06,
      "loss": 0.8268,
      "step": 27610
    },
    {
      "epoch": 1.821672959915577,
      "grad_norm": 0.4907081127166748,
      "learning_rate": 1.2022462657292843e-06,
      "loss": 0.7987,
      "step": 27620
    },
    {
      "epoch": 1.8223325143865319,
      "grad_norm": 0.6870168447494507,
      "learning_rate": 1.1934449091147825e-06,
      "loss": 0.8311,
      "step": 27630
    },
    {
      "epoch": 1.8229920688574868,
      "grad_norm": 0.6277094483375549,
      "learning_rate": 1.18467509944859e-06,
      "loss": 0.8072,
      "step": 27640
    },
    {
      "epoch": 1.8236516233284417,
      "grad_norm": 0.6111871004104614,
      "learning_rate": 1.175936848351919e-06,
      "loss": 0.8094,
      "step": 27650
    },
    {
      "epoch": 1.8243111777993966,
      "grad_norm": 0.6213318705558777,
      "learning_rate": 1.1672301674041403e-06,
      "loss": 0.7951,
      "step": 27660
    },
    {
      "epoch": 1.8249707322703514,
      "grad_norm": 0.5851269960403442,
      "learning_rate": 1.1585550681428054e-06,
      "loss": 0.7909,
      "step": 27670
    },
    {
      "epoch": 1.8256302867413061,
      "grad_norm": 0.6120691895484924,
      "learning_rate": 1.1499115620636103e-06,
      "loss": 0.7837,
      "step": 27680
    },
    {
      "epoch": 1.8262898412122612,
      "grad_norm": 0.6957132816314697,
      "learning_rate": 1.1412996606203818e-06,
      "loss": 0.8056,
      "step": 27690
    },
    {
      "epoch": 1.826949395683216,
      "grad_norm": 0.6566064953804016,
      "learning_rate": 1.1327193752250665e-06,
      "loss": 0.8448,
      "step": 27700
    },
    {
      "epoch": 1.827608950154171,
      "grad_norm": 0.6958504319190979,
      "learning_rate": 1.1241707172477245e-06,
      "loss": 0.818,
      "step": 27710
    },
    {
      "epoch": 1.8282685046251257,
      "grad_norm": 0.6409597396850586,
      "learning_rate": 1.1156536980164972e-06,
      "loss": 0.8035,
      "step": 27720
    },
    {
      "epoch": 1.8289280590960806,
      "grad_norm": 0.5831571817398071,
      "learning_rate": 1.107168328817601e-06,
      "loss": 0.8198,
      "step": 27730
    },
    {
      "epoch": 1.8295876135670355,
      "grad_norm": 0.6030648946762085,
      "learning_rate": 1.0987146208953164e-06,
      "loss": 0.7906,
      "step": 27740
    },
    {
      "epoch": 1.8302471680379904,
      "grad_norm": 0.6270460486412048,
      "learning_rate": 1.0902925854519603e-06,
      "loss": 0.816,
      "step": 27750
    },
    {
      "epoch": 1.8309067225089453,
      "grad_norm": 0.6238800287246704,
      "learning_rate": 1.0819022336478884e-06,
      "loss": 0.8364,
      "step": 27760
    },
    {
      "epoch": 1.8315662769799,
      "grad_norm": 0.7053459882736206,
      "learning_rate": 1.0735435766014657e-06,
      "loss": 0.8146,
      "step": 27770
    },
    {
      "epoch": 1.832225831450855,
      "grad_norm": 0.6827749609947205,
      "learning_rate": 1.0652166253890567e-06,
      "loss": 0.7585,
      "step": 27780
    },
    {
      "epoch": 1.8328853859218097,
      "grad_norm": 0.6241556406021118,
      "learning_rate": 1.0569213910450153e-06,
      "loss": 0.7673,
      "step": 27790
    },
    {
      "epoch": 1.8335449403927648,
      "grad_norm": 0.6750614643096924,
      "learning_rate": 1.04865788456166e-06,
      "loss": 0.8058,
      "step": 27800
    },
    {
      "epoch": 1.8342044948637195,
      "grad_norm": 0.6766597628593445,
      "learning_rate": 1.0404261168892786e-06,
      "loss": 0.8014,
      "step": 27810
    },
    {
      "epoch": 1.8348640493346744,
      "grad_norm": 0.6603577733039856,
      "learning_rate": 1.0322260989360788e-06,
      "loss": 0.7803,
      "step": 27820
    },
    {
      "epoch": 1.8355236038056293,
      "grad_norm": 0.5669465661048889,
      "learning_rate": 1.0240578415682157e-06,
      "loss": 0.7876,
      "step": 27830
    },
    {
      "epoch": 1.8361831582765842,
      "grad_norm": 0.6441722512245178,
      "learning_rate": 1.0159213556097474e-06,
      "loss": 0.7779,
      "step": 27840
    },
    {
      "epoch": 1.836842712747539,
      "grad_norm": 0.5952627658843994,
      "learning_rate": 1.0078166518426356e-06,
      "loss": 0.8614,
      "step": 27850
    },
    {
      "epoch": 1.8375022672184937,
      "grad_norm": 0.6190049052238464,
      "learning_rate": 9.997437410067228e-07,
      "loss": 0.7864,
      "step": 27860
    },
    {
      "epoch": 1.8381618216894489,
      "grad_norm": 0.6481174230575562,
      "learning_rate": 9.91702633799721e-07,
      "loss": 0.7902,
      "step": 27870
    },
    {
      "epoch": 1.8388213761604035,
      "grad_norm": 0.5373259782791138,
      "learning_rate": 9.836933408772009e-07,
      "loss": 0.7392,
      "step": 27880
    },
    {
      "epoch": 1.8394809306313586,
      "grad_norm": 0.600075364112854,
      "learning_rate": 9.757158728525673e-07,
      "loss": 0.7742,
      "step": 27890
    },
    {
      "epoch": 1.8401404851023133,
      "grad_norm": 0.5878193974494934,
      "learning_rate": 9.677702402970724e-07,
      "loss": 0.7823,
      "step": 27900
    },
    {
      "epoch": 1.8408000395732682,
      "grad_norm": 0.5557110905647278,
      "learning_rate": 9.598564537397548e-07,
      "loss": 0.8337,
      "step": 27910
    },
    {
      "epoch": 1.841459594044223,
      "grad_norm": 0.6779967546463013,
      "learning_rate": 9.519745236674704e-07,
      "loss": 0.7992,
      "step": 27920
    },
    {
      "epoch": 1.842119148515178,
      "grad_norm": 0.5710042119026184,
      "learning_rate": 9.441244605248561e-07,
      "loss": 0.7914,
      "step": 27930
    },
    {
      "epoch": 1.8427787029861329,
      "grad_norm": 0.5980215668678284,
      "learning_rate": 9.363062747143242e-07,
      "loss": 0.8313,
      "step": 27940
    },
    {
      "epoch": 1.8434382574570878,
      "grad_norm": 0.6596368551254272,
      "learning_rate": 9.285199765960456e-07,
      "loss": 0.7727,
      "step": 27950
    },
    {
      "epoch": 1.8440978119280427,
      "grad_norm": 0.6728708744049072,
      "learning_rate": 9.207655764879225e-07,
      "loss": 0.8028,
      "step": 27960
    },
    {
      "epoch": 1.8447573663989973,
      "grad_norm": 0.5726572871208191,
      "learning_rate": 9.130430846656074e-07,
      "loss": 0.7922,
      "step": 27970
    },
    {
      "epoch": 1.8454169208699525,
      "grad_norm": 0.5376827716827393,
      "learning_rate": 9.053525113624534e-07,
      "loss": 0.7349,
      "step": 27980
    },
    {
      "epoch": 1.8460764753409071,
      "grad_norm": 0.5951704978942871,
      "learning_rate": 8.976938667695278e-07,
      "loss": 0.7845,
      "step": 27990
    },
    {
      "epoch": 1.8467360298118622,
      "grad_norm": 0.5681121349334717,
      "learning_rate": 8.900671610355876e-07,
      "loss": 0.7981,
      "step": 28000
    },
    {
      "epoch": 1.847395584282817,
      "grad_norm": 0.5649076104164124,
      "learning_rate": 8.82472404267054e-07,
      "loss": 0.7924,
      "step": 28010
    },
    {
      "epoch": 1.8480551387537718,
      "grad_norm": 0.6413310766220093,
      "learning_rate": 8.749096065280266e-07,
      "loss": 0.7917,
      "step": 28020
    },
    {
      "epoch": 1.8487146932247267,
      "grad_norm": 0.6765559315681458,
      "learning_rate": 8.673787778402498e-07,
      "loss": 0.8659,
      "step": 28030
    },
    {
      "epoch": 1.8493742476956816,
      "grad_norm": 0.6600339412689209,
      "learning_rate": 8.59879928183105e-07,
      "loss": 0.8127,
      "step": 28040
    },
    {
      "epoch": 1.8500338021666365,
      "grad_norm": 0.569463312625885,
      "learning_rate": 8.524130674935876e-07,
      "loss": 0.808,
      "step": 28050
    },
    {
      "epoch": 1.8506933566375912,
      "grad_norm": 0.61843341588974,
      "learning_rate": 8.449782056663191e-07,
      "loss": 0.8086,
      "step": 28060
    },
    {
      "epoch": 1.8513529111085463,
      "grad_norm": 0.5859857201576233,
      "learning_rate": 8.375753525535046e-07,
      "loss": 0.8057,
      "step": 28070
    },
    {
      "epoch": 1.852012465579501,
      "grad_norm": 0.5519652366638184,
      "learning_rate": 8.30204517964947e-07,
      "loss": 0.7882,
      "step": 28080
    },
    {
      "epoch": 1.852672020050456,
      "grad_norm": 0.6738079190254211,
      "learning_rate": 8.228657116680056e-07,
      "loss": 0.8171,
      "step": 28090
    },
    {
      "epoch": 1.8533315745214107,
      "grad_norm": 0.6553736329078674,
      "learning_rate": 8.155589433876065e-07,
      "loss": 0.793,
      "step": 28100
    },
    {
      "epoch": 1.8539911289923656,
      "grad_norm": 0.6549698710441589,
      "learning_rate": 8.082842228062215e-07,
      "loss": 0.815,
      "step": 28110
    },
    {
      "epoch": 1.8546506834633205,
      "grad_norm": 0.5773028135299683,
      "learning_rate": 8.010415595638476e-07,
      "loss": 0.7916,
      "step": 28120
    },
    {
      "epoch": 1.8553102379342754,
      "grad_norm": 0.6732603311538696,
      "learning_rate": 7.938309632580132e-07,
      "loss": 0.8104,
      "step": 28130
    },
    {
      "epoch": 1.8559697924052303,
      "grad_norm": 0.5733674168586731,
      "learning_rate": 7.866524434437361e-07,
      "loss": 0.801,
      "step": 28140
    },
    {
      "epoch": 1.856629346876185,
      "grad_norm": 0.6804933547973633,
      "learning_rate": 7.795060096335488e-07,
      "loss": 0.809,
      "step": 28150
    },
    {
      "epoch": 1.85728890134714,
      "grad_norm": 0.7227495312690735,
      "learning_rate": 7.72391671297451e-07,
      "loss": 0.8407,
      "step": 28160
    },
    {
      "epoch": 1.8579484558180948,
      "grad_norm": 0.5979621410369873,
      "learning_rate": 7.653094378629183e-07,
      "loss": 0.7738,
      "step": 28170
    },
    {
      "epoch": 1.8586080102890499,
      "grad_norm": 0.5470470786094666,
      "learning_rate": 7.582593187148795e-07,
      "loss": 0.771,
      "step": 28180
    },
    {
      "epoch": 1.8592675647600045,
      "grad_norm": 0.647771954536438,
      "learning_rate": 7.512413231957061e-07,
      "loss": 0.8138,
      "step": 28190
    },
    {
      "epoch": 1.8599271192309594,
      "grad_norm": 0.6679745316505432,
      "learning_rate": 7.442554606052088e-07,
      "loss": 0.7772,
      "step": 28200
    },
    {
      "epoch": 1.8605866737019143,
      "grad_norm": 0.6812871098518372,
      "learning_rate": 7.373017402006133e-07,
      "loss": 0.8325,
      "step": 28210
    },
    {
      "epoch": 1.8612462281728692,
      "grad_norm": 0.5566513538360596,
      "learning_rate": 7.30380171196543e-07,
      "loss": 0.7712,
      "step": 28220
    },
    {
      "epoch": 1.8619057826438241,
      "grad_norm": 0.6013004779815674,
      "learning_rate": 7.234907627650361e-07,
      "loss": 0.8472,
      "step": 28230
    },
    {
      "epoch": 1.862565337114779,
      "grad_norm": 0.577357292175293,
      "learning_rate": 7.166335240354982e-07,
      "loss": 0.7835,
      "step": 28240
    },
    {
      "epoch": 1.863224891585734,
      "grad_norm": 0.6533622741699219,
      "learning_rate": 7.098084640947105e-07,
      "loss": 0.8029,
      "step": 28250
    },
    {
      "epoch": 1.8638844460566886,
      "grad_norm": 0.6401820778846741,
      "learning_rate": 7.030155919868165e-07,
      "loss": 0.7857,
      "step": 28260
    },
    {
      "epoch": 1.8645440005276437,
      "grad_norm": 0.5723109245300293,
      "learning_rate": 6.962549167132987e-07,
      "loss": 0.7685,
      "step": 28270
    },
    {
      "epoch": 1.8652035549985984,
      "grad_norm": 0.5786372423171997,
      "learning_rate": 6.895264472329771e-07,
      "loss": 0.7816,
      "step": 28280
    },
    {
      "epoch": 1.8658631094695535,
      "grad_norm": 0.6384173631668091,
      "learning_rate": 6.828301924620001e-07,
      "loss": 0.7928,
      "step": 28290
    },
    {
      "epoch": 1.8665226639405081,
      "grad_norm": 0.631347119808197,
      "learning_rate": 6.761661612738224e-07,
      "loss": 0.802,
      "step": 28300
    },
    {
      "epoch": 1.867182218411463,
      "grad_norm": 0.744253933429718,
      "learning_rate": 6.69534362499194e-07,
      "loss": 0.7876,
      "step": 28310
    },
    {
      "epoch": 1.867841772882418,
      "grad_norm": 0.6918014883995056,
      "learning_rate": 6.62934804926163e-07,
      "loss": 0.7935,
      "step": 28320
    },
    {
      "epoch": 1.8685013273533728,
      "grad_norm": 0.604925811290741,
      "learning_rate": 6.563674973000422e-07,
      "loss": 0.7913,
      "step": 28330
    },
    {
      "epoch": 1.8691608818243277,
      "grad_norm": 0.585300624370575,
      "learning_rate": 6.498324483234231e-07,
      "loss": 0.8186,
      "step": 28340
    },
    {
      "epoch": 1.8698204362952824,
      "grad_norm": 0.6663399934768677,
      "learning_rate": 6.433296666561339e-07,
      "loss": 0.8498,
      "step": 28350
    },
    {
      "epoch": 1.8704799907662375,
      "grad_norm": 0.5360284447669983,
      "learning_rate": 6.368591609152514e-07,
      "loss": 0.7987,
      "step": 28360
    },
    {
      "epoch": 1.8711395452371922,
      "grad_norm": 0.6080397367477417,
      "learning_rate": 6.30420939675086e-07,
      "loss": 0.8008,
      "step": 28370
    },
    {
      "epoch": 1.8717990997081473,
      "grad_norm": 0.5916748046875,
      "learning_rate": 6.240150114671578e-07,
      "loss": 0.7896,
      "step": 28380
    },
    {
      "epoch": 1.872458654179102,
      "grad_norm": 0.6351580023765564,
      "learning_rate": 6.176413847802098e-07,
      "loss": 0.8017,
      "step": 28390
    },
    {
      "epoch": 1.8731182086500568,
      "grad_norm": 0.6313516497612,
      "learning_rate": 6.113000680601611e-07,
      "loss": 0.8033,
      "step": 28400
    },
    {
      "epoch": 1.8737777631210117,
      "grad_norm": 0.5805239677429199,
      "learning_rate": 6.049910697101313e-07,
      "loss": 0.7932,
      "step": 28410
    },
    {
      "epoch": 1.8744373175919666,
      "grad_norm": 0.5967887043952942,
      "learning_rate": 5.987143980904026e-07,
      "loss": 0.833,
      "step": 28420
    },
    {
      "epoch": 1.8750968720629215,
      "grad_norm": 0.5833791494369507,
      "learning_rate": 5.924700615184298e-07,
      "loss": 0.7703,
      "step": 28430
    },
    {
      "epoch": 1.8757564265338764,
      "grad_norm": 0.6744887232780457,
      "learning_rate": 5.862580682688135e-07,
      "loss": 0.8229,
      "step": 28440
    },
    {
      "epoch": 1.8764159810048313,
      "grad_norm": 0.5459935069084167,
      "learning_rate": 5.80078426573294e-07,
      "loss": 0.795,
      "step": 28450
    },
    {
      "epoch": 1.877075535475786,
      "grad_norm": 0.6006172299385071,
      "learning_rate": 5.739311446207457e-07,
      "loss": 0.7936,
      "step": 28460
    },
    {
      "epoch": 1.877735089946741,
      "grad_norm": 0.6047661304473877,
      "learning_rate": 5.678162305571583e-07,
      "loss": 0.8336,
      "step": 28470
    },
    {
      "epoch": 1.8783946444176958,
      "grad_norm": 0.5724004507064819,
      "learning_rate": 5.61733692485636e-07,
      "loss": 0.7871,
      "step": 28480
    },
    {
      "epoch": 1.8790541988886507,
      "grad_norm": 0.658601701259613,
      "learning_rate": 5.556835384663673e-07,
      "loss": 0.7798,
      "step": 28490
    },
    {
      "epoch": 1.8797137533596056,
      "grad_norm": 0.6241977214813232,
      "learning_rate": 5.496657765166419e-07,
      "loss": 0.7971,
      "step": 28500
    },
    {
      "epoch": 1.8803733078305604,
      "grad_norm": 0.633460521697998,
      "learning_rate": 5.436804146108143e-07,
      "loss": 0.8022,
      "step": 28510
    },
    {
      "epoch": 1.8810328623015153,
      "grad_norm": 0.6179897785186768,
      "learning_rate": 5.377274606803123e-07,
      "loss": 0.8115,
      "step": 28520
    },
    {
      "epoch": 1.8816924167724702,
      "grad_norm": 0.5969645977020264,
      "learning_rate": 5.318069226136174e-07,
      "loss": 0.8075,
      "step": 28530
    },
    {
      "epoch": 1.8823519712434251,
      "grad_norm": 0.5832700133323669,
      "learning_rate": 5.259188082562483e-07,
      "loss": 0.8137,
      "step": 28540
    },
    {
      "epoch": 1.8830115257143798,
      "grad_norm": 0.6285104751586914,
      "learning_rate": 5.20063125410769e-07,
      "loss": 0.7557,
      "step": 28550
    },
    {
      "epoch": 1.883671080185335,
      "grad_norm": 0.5918856859207153,
      "learning_rate": 5.142398818367589e-07,
      "loss": 0.7935,
      "step": 28560
    },
    {
      "epoch": 1.8843306346562896,
      "grad_norm": 0.6324704885482788,
      "learning_rate": 5.084490852508172e-07,
      "loss": 0.7444,
      "step": 28570
    },
    {
      "epoch": 1.8849901891272447,
      "grad_norm": 0.5948597192764282,
      "learning_rate": 5.026907433265393e-07,
      "loss": 0.8408,
      "step": 28580
    },
    {
      "epoch": 1.8856497435981994,
      "grad_norm": 0.5209887623786926,
      "learning_rate": 4.969648636945184e-07,
      "loss": 0.8001,
      "step": 28590
    },
    {
      "epoch": 1.8863092980691543,
      "grad_norm": 0.6631389856338501,
      "learning_rate": 4.912714539423296e-07,
      "loss": 0.7824,
      "step": 28600
    },
    {
      "epoch": 1.8869688525401092,
      "grad_norm": 0.5967811942100525,
      "learning_rate": 4.856105216145212e-07,
      "loss": 0.7951,
      "step": 28610
    },
    {
      "epoch": 1.887628407011064,
      "grad_norm": 0.5898823738098145,
      "learning_rate": 4.799820742126066e-07,
      "loss": 0.7931,
      "step": 28620
    },
    {
      "epoch": 1.888287961482019,
      "grad_norm": 0.717066764831543,
      "learning_rate": 4.743861191950472e-07,
      "loss": 0.8245,
      "step": 28630
    },
    {
      "epoch": 1.8889475159529736,
      "grad_norm": 0.585137665271759,
      "learning_rate": 4.6882266397725305e-07,
      "loss": 0.8382,
      "step": 28640
    },
    {
      "epoch": 1.8896070704239287,
      "grad_norm": 0.6448143720626831,
      "learning_rate": 4.632917159315603e-07,
      "loss": 0.7635,
      "step": 28650
    },
    {
      "epoch": 1.8902666248948834,
      "grad_norm": 0.6555895209312439,
      "learning_rate": 4.577932823872394e-07,
      "loss": 0.8097,
      "step": 28660
    },
    {
      "epoch": 1.8909261793658385,
      "grad_norm": 0.5938135981559753,
      "learning_rate": 4.523273706304648e-07,
      "loss": 0.8214,
      "step": 28670
    },
    {
      "epoch": 1.8915857338367932,
      "grad_norm": 0.6311590075492859,
      "learning_rate": 4.4689398790431494e-07,
      "loss": 0.7846,
      "step": 28680
    },
    {
      "epoch": 1.892245288307748,
      "grad_norm": 0.6728568077087402,
      "learning_rate": 4.4149314140877474e-07,
      "loss": 0.7662,
      "step": 28690
    },
    {
      "epoch": 1.892904842778703,
      "grad_norm": 0.5750419497489929,
      "learning_rate": 4.361248383007027e-07,
      "loss": 0.7597,
      "step": 28700
    },
    {
      "epoch": 1.8935643972496579,
      "grad_norm": 0.6412707567214966,
      "learning_rate": 4.3078908569383346e-07,
      "loss": 0.7943,
      "step": 28710
    },
    {
      "epoch": 1.8942239517206128,
      "grad_norm": 0.6414803266525269,
      "learning_rate": 4.2548589065876945e-07,
      "loss": 0.8001,
      "step": 28720
    },
    {
      "epoch": 1.8948835061915676,
      "grad_norm": 0.6530488133430481,
      "learning_rate": 4.202152602229781e-07,
      "loss": 0.8507,
      "step": 28730
    },
    {
      "epoch": 1.8955430606625225,
      "grad_norm": 0.679801344871521,
      "learning_rate": 4.1497720137076145e-07,
      "loss": 0.7956,
      "step": 28740
    },
    {
      "epoch": 1.8962026151334772,
      "grad_norm": 0.5268911719322205,
      "learning_rate": 4.0977172104326723e-07,
      "loss": 0.7891,
      "step": 28750
    },
    {
      "epoch": 1.8968621696044323,
      "grad_norm": 0.5881314277648926,
      "learning_rate": 4.04598826138472e-07,
      "loss": 0.7669,
      "step": 28760
    },
    {
      "epoch": 1.897521724075387,
      "grad_norm": 0.6459960341453552,
      "learning_rate": 3.9945852351116474e-07,
      "loss": 0.8379,
      "step": 28770
    },
    {
      "epoch": 1.898181278546342,
      "grad_norm": 0.5741415023803711,
      "learning_rate": 3.943508199729579e-07,
      "loss": 0.8246,
      "step": 28780
    },
    {
      "epoch": 1.8988408330172968,
      "grad_norm": 0.6332618594169617,
      "learning_rate": 3.8927572229225384e-07,
      "loss": 0.7727,
      "step": 28790
    },
    {
      "epoch": 1.8995003874882517,
      "grad_norm": 0.5780948996543884,
      "learning_rate": 3.8423323719425365e-07,
      "loss": 0.7893,
      "step": 28800
    },
    {
      "epoch": 1.9001599419592066,
      "grad_norm": 0.5841256976127625,
      "learning_rate": 3.792233713609428e-07,
      "loss": 0.779,
      "step": 28810
    },
    {
      "epoch": 1.9008194964301615,
      "grad_norm": 0.5591938495635986,
      "learning_rate": 3.7424613143107746e-07,
      "loss": 0.7964,
      "step": 28820
    },
    {
      "epoch": 1.9014790509011164,
      "grad_norm": 0.5748834609985352,
      "learning_rate": 3.693015240001846e-07,
      "loss": 0.8756,
      "step": 28830
    },
    {
      "epoch": 1.902138605372071,
      "grad_norm": 0.5563206672668457,
      "learning_rate": 3.6438955562055065e-07,
      "loss": 0.7879,
      "step": 28840
    },
    {
      "epoch": 1.9027981598430261,
      "grad_norm": 0.6695645451545715,
      "learning_rate": 3.5951023280119953e-07,
      "loss": 0.8213,
      "step": 28850
    },
    {
      "epoch": 1.9034577143139808,
      "grad_norm": 0.6426726579666138,
      "learning_rate": 3.5466356200790364e-07,
      "loss": 0.7981,
      "step": 28860
    },
    {
      "epoch": 1.904117268784936,
      "grad_norm": 0.6324830651283264,
      "learning_rate": 3.498495496631726e-07,
      "loss": 0.7583,
      "step": 28870
    },
    {
      "epoch": 1.9047768232558906,
      "grad_norm": 0.70176100730896,
      "learning_rate": 3.4506820214622594e-07,
      "loss": 0.7816,
      "step": 28880
    },
    {
      "epoch": 1.9054363777268455,
      "grad_norm": 0.6347261667251587,
      "learning_rate": 3.403195257930064e-07,
      "loss": 0.8076,
      "step": 28890
    },
    {
      "epoch": 1.9060959321978004,
      "grad_norm": 0.6823022365570068,
      "learning_rate": 3.35603526896161e-07,
      "loss": 0.7579,
      "step": 28900
    },
    {
      "epoch": 1.9067554866687553,
      "grad_norm": 0.5782615542411804,
      "learning_rate": 3.309202117050353e-07,
      "loss": 0.8361,
      "step": 28910
    },
    {
      "epoch": 1.9074150411397102,
      "grad_norm": 0.6254626512527466,
      "learning_rate": 3.2626958642566487e-07,
      "loss": 0.8308,
      "step": 28920
    },
    {
      "epoch": 1.9080745956106648,
      "grad_norm": 0.5811460018157959,
      "learning_rate": 3.2165165722076453e-07,
      "loss": 0.84,
      "step": 28930
    },
    {
      "epoch": 1.90873415008162,
      "grad_norm": 0.51547771692276,
      "learning_rate": 3.170664302097226e-07,
      "loss": 0.8014,
      "step": 28940
    },
    {
      "epoch": 1.9093937045525746,
      "grad_norm": 0.5484378933906555,
      "learning_rate": 3.1251391146859264e-07,
      "loss": 0.7637,
      "step": 28950
    },
    {
      "epoch": 1.9100532590235297,
      "grad_norm": 0.5861138105392456,
      "learning_rate": 3.079941070300879e-07,
      "loss": 0.7827,
      "step": 28960
    },
    {
      "epoch": 1.9107128134944844,
      "grad_norm": 0.6178954243659973,
      "learning_rate": 3.035070228835674e-07,
      "loss": 0.7855,
      "step": 28970
    },
    {
      "epoch": 1.9113723679654393,
      "grad_norm": 0.5996459722518921,
      "learning_rate": 2.99052664975033e-07,
      "loss": 0.7935,
      "step": 28980
    },
    {
      "epoch": 1.9120319224363942,
      "grad_norm": 0.7222851514816284,
      "learning_rate": 2.9463103920712163e-07,
      "loss": 0.7818,
      "step": 28990
    },
    {
      "epoch": 1.912691476907349,
      "grad_norm": 0.6564216017723083,
      "learning_rate": 2.9024215143908796e-07,
      "loss": 0.8046,
      "step": 29000
    },
    {
      "epoch": 1.913351031378304,
      "grad_norm": 0.5750806927680969,
      "learning_rate": 2.858860074868158e-07,
      "loss": 0.854,
      "step": 29010
    },
    {
      "epoch": 1.9140105858492589,
      "grad_norm": 0.5890952348709106,
      "learning_rate": 2.8156261312279055e-07,
      "loss": 0.78,
      "step": 29020
    },
    {
      "epoch": 1.9146701403202138,
      "grad_norm": 0.5353766083717346,
      "learning_rate": 2.772719740760987e-07,
      "loss": 0.787,
      "step": 29030
    },
    {
      "epoch": 1.9153296947911684,
      "grad_norm": 0.6585463881492615,
      "learning_rate": 2.730140960324312e-07,
      "loss": 0.8043,
      "step": 29040
    },
    {
      "epoch": 1.9159892492621236,
      "grad_norm": 0.7087804079055786,
      "learning_rate": 2.6878898463405523e-07,
      "loss": 0.8154,
      "step": 29050
    },
    {
      "epoch": 1.9166488037330782,
      "grad_norm": 0.6948942542076111,
      "learning_rate": 2.645966454798282e-07,
      "loss": 0.8357,
      "step": 29060
    },
    {
      "epoch": 1.9173083582040333,
      "grad_norm": 0.6298587918281555,
      "learning_rate": 2.604370841251674e-07,
      "loss": 0.8252,
      "step": 29070
    },
    {
      "epoch": 1.917967912674988,
      "grad_norm": 0.5921646356582642,
      "learning_rate": 2.5631030608206653e-07,
      "loss": 0.7834,
      "step": 29080
    },
    {
      "epoch": 1.918627467145943,
      "grad_norm": 0.6113035678863525,
      "learning_rate": 2.522163168190678e-07,
      "loss": 0.8191,
      "step": 29090
    },
    {
      "epoch": 1.9192870216168978,
      "grad_norm": 0.6097885370254517,
      "learning_rate": 2.4815512176127044e-07,
      "loss": 0.8086,
      "step": 29100
    },
    {
      "epoch": 1.9199465760878527,
      "grad_norm": 0.5496877431869507,
      "learning_rate": 2.441267262903169e-07,
      "loss": 0.8146,
      "step": 29110
    },
    {
      "epoch": 1.9206061305588076,
      "grad_norm": 0.6302570104598999,
      "learning_rate": 2.401311357443786e-07,
      "loss": 0.7706,
      "step": 29120
    },
    {
      "epoch": 1.9212656850297622,
      "grad_norm": 0.5880969762802124,
      "learning_rate": 2.3616835541816483e-07,
      "loss": 0.833,
      "step": 29130
    },
    {
      "epoch": 1.9219252395007174,
      "grad_norm": 0.7146965861320496,
      "learning_rate": 2.322383905628972e-07,
      "loss": 0.814,
      "step": 29140
    },
    {
      "epoch": 1.922584793971672,
      "grad_norm": 0.6757823824882507,
      "learning_rate": 2.2834124638632658e-07,
      "loss": 0.834,
      "step": 29150
    },
    {
      "epoch": 1.9232443484426271,
      "grad_norm": 0.6999145746231079,
      "learning_rate": 2.2447692805269427e-07,
      "loss": 0.7789,
      "step": 29160
    },
    {
      "epoch": 1.9239039029135818,
      "grad_norm": 0.5567458271980286,
      "learning_rate": 2.2064544068275682e-07,
      "loss": 0.7861,
      "step": 29170
    },
    {
      "epoch": 1.9245634573845367,
      "grad_norm": 0.6425588726997375,
      "learning_rate": 2.1684678935375845e-07,
      "loss": 0.7737,
      "step": 29180
    },
    {
      "epoch": 1.9252230118554916,
      "grad_norm": 0.6440041661262512,
      "learning_rate": 2.130809790994337e-07,
      "loss": 0.7504,
      "step": 29190
    },
    {
      "epoch": 1.9258825663264465,
      "grad_norm": 0.6200813055038452,
      "learning_rate": 2.0934801490999635e-07,
      "loss": 0.821,
      "step": 29200
    },
    {
      "epoch": 1.9265421207974014,
      "grad_norm": 0.6447944045066833,
      "learning_rate": 2.0564790173213388e-07,
      "loss": 0.7804,
      "step": 29210
    },
    {
      "epoch": 1.927201675268356,
      "grad_norm": 0.5899184942245483,
      "learning_rate": 2.0198064446900756e-07,
      "loss": 0.814,
      "step": 29220
    },
    {
      "epoch": 1.9278612297393112,
      "grad_norm": 0.6010910272598267,
      "learning_rate": 1.9834624798023006e-07,
      "loss": 0.8052,
      "step": 29230
    },
    {
      "epoch": 1.9285207842102658,
      "grad_norm": 0.6138426065444946,
      "learning_rate": 1.9474471708188225e-07,
      "loss": 0.8303,
      "step": 29240
    },
    {
      "epoch": 1.929180338681221,
      "grad_norm": 0.5905739068984985,
      "learning_rate": 1.9117605654648264e-07,
      "loss": 0.8096,
      "step": 29250
    },
    {
      "epoch": 1.9298398931521756,
      "grad_norm": 0.5698313117027283,
      "learning_rate": 1.8764027110299008e-07,
      "loss": 0.8557,
      "step": 29260
    },
    {
      "epoch": 1.9304994476231305,
      "grad_norm": 0.6312277913093567,
      "learning_rate": 1.8413736543681503e-07,
      "loss": 0.7648,
      "step": 29270
    },
    {
      "epoch": 1.9311590020940854,
      "grad_norm": 0.7178078293800354,
      "learning_rate": 1.8066734418978047e-07,
      "loss": 0.7871,
      "step": 29280
    },
    {
      "epoch": 1.9318185565650403,
      "grad_norm": 0.5866417288780212,
      "learning_rate": 1.7723021196014433e-07,
      "loss": 0.7638,
      "step": 29290
    },
    {
      "epoch": 1.9324781110359952,
      "grad_norm": 0.5741631388664246,
      "learning_rate": 1.7382597330257432e-07,
      "loss": 0.8046,
      "step": 29300
    },
    {
      "epoch": 1.93313766550695,
      "grad_norm": 0.5900196433067322,
      "learning_rate": 1.7045463272815366e-07,
      "loss": 0.7694,
      "step": 29310
    },
    {
      "epoch": 1.933797219977905,
      "grad_norm": 0.6223501563072205,
      "learning_rate": 1.6711619470437258e-07,
      "loss": 0.8038,
      "step": 29320
    },
    {
      "epoch": 1.9344567744488597,
      "grad_norm": 0.5457992553710938,
      "learning_rate": 1.638106636551201e-07,
      "loss": 0.8517,
      "step": 29330
    },
    {
      "epoch": 1.9351163289198148,
      "grad_norm": 0.5838644504547119,
      "learning_rate": 1.6053804396067295e-07,
      "loss": 0.7684,
      "step": 29340
    },
    {
      "epoch": 1.9357758833907694,
      "grad_norm": 0.6603797078132629,
      "learning_rate": 1.5729833995770383e-07,
      "loss": 0.797,
      "step": 29350
    },
    {
      "epoch": 1.9364354378617246,
      "grad_norm": 0.6109535694122314,
      "learning_rate": 1.540915559392675e-07,
      "loss": 0.8132,
      "step": 29360
    },
    {
      "epoch": 1.9370949923326792,
      "grad_norm": 0.6293129324913025,
      "learning_rate": 1.5091769615478702e-07,
      "loss": 0.8362,
      "step": 29370
    },
    {
      "epoch": 1.9377545468036341,
      "grad_norm": 0.6027792692184448,
      "learning_rate": 1.4777676481006475e-07,
      "loss": 0.7832,
      "step": 29380
    },
    {
      "epoch": 1.938414101274589,
      "grad_norm": 0.6462914943695068,
      "learning_rate": 1.4466876606726297e-07,
      "loss": 0.7979,
      "step": 29390
    },
    {
      "epoch": 1.939073655745544,
      "grad_norm": 0.5773576498031616,
      "learning_rate": 1.4159370404490945e-07,
      "loss": 0.7953,
      "step": 29400
    },
    {
      "epoch": 1.9397332102164988,
      "grad_norm": 0.5936535000801086,
      "learning_rate": 1.3855158281787794e-07,
      "loss": 0.7555,
      "step": 29410
    },
    {
      "epoch": 1.9403927646874535,
      "grad_norm": 0.5910260081291199,
      "learning_rate": 1.3554240641740211e-07,
      "loss": 0.7682,
      "step": 29420
    },
    {
      "epoch": 1.9410523191584086,
      "grad_norm": 0.6144441366195679,
      "learning_rate": 1.3256617883104783e-07,
      "loss": 0.7915,
      "step": 29430
    },
    {
      "epoch": 1.9417118736293633,
      "grad_norm": 0.6245831251144409,
      "learning_rate": 1.296229040027269e-07,
      "loss": 0.8063,
      "step": 29440
    },
    {
      "epoch": 1.9423714281003184,
      "grad_norm": 0.5887926816940308,
      "learning_rate": 1.2671258583268064e-07,
      "loss": 0.7825,
      "step": 29450
    },
    {
      "epoch": 1.943030982571273,
      "grad_norm": 0.7208525538444519,
      "learning_rate": 1.238352281774824e-07,
      "loss": 0.8249,
      "step": 29460
    },
    {
      "epoch": 1.943690537042228,
      "grad_norm": 0.585756778717041,
      "learning_rate": 1.2099083485002105e-07,
      "loss": 0.7675,
      "step": 29470
    },
    {
      "epoch": 1.9443500915131828,
      "grad_norm": 0.6151425242424011,
      "learning_rate": 1.1817940961950935e-07,
      "loss": 0.7904,
      "step": 29480
    },
    {
      "epoch": 1.9450096459841377,
      "grad_norm": 0.6283534169197083,
      "learning_rate": 1.1540095621147274e-07,
      "loss": 0.8098,
      "step": 29490
    },
    {
      "epoch": 1.9456692004550926,
      "grad_norm": 0.6851007342338562,
      "learning_rate": 1.1265547830774381e-07,
      "loss": 0.8109,
      "step": 29500
    },
    {
      "epoch": 1.9463287549260473,
      "grad_norm": 0.5581753849983215,
      "learning_rate": 1.0994297954645127e-07,
      "loss": 0.7712,
      "step": 29510
    },
    {
      "epoch": 1.9469883093970024,
      "grad_norm": 0.6845225691795349,
      "learning_rate": 1.0726346352203376e-07,
      "loss": 0.8152,
      "step": 29520
    },
    {
      "epoch": 1.947647863867957,
      "grad_norm": 0.605595588684082,
      "learning_rate": 1.0461693378521487e-07,
      "loss": 0.8162,
      "step": 29530
    },
    {
      "epoch": 1.9483074183389122,
      "grad_norm": 0.6354922652244568,
      "learning_rate": 1.0200339384300872e-07,
      "loss": 0.8063,
      "step": 29540
    },
    {
      "epoch": 1.9489669728098669,
      "grad_norm": 0.6054199934005737,
      "learning_rate": 9.942284715871442e-08,
      "loss": 0.7971,
      "step": 29550
    },
    {
      "epoch": 1.9496265272808218,
      "grad_norm": 0.6663618683815002,
      "learning_rate": 9.687529715191047e-08,
      "loss": 0.7953,
      "step": 29560
    },
    {
      "epoch": 1.9502860817517766,
      "grad_norm": 0.5727202296257019,
      "learning_rate": 9.436074719844923e-08,
      "loss": 0.7885,
      "step": 29570
    },
    {
      "epoch": 1.9509456362227315,
      "grad_norm": 0.6474148035049438,
      "learning_rate": 9.187920063045141e-08,
      "loss": 0.832,
      "step": 29580
    },
    {
      "epoch": 1.9516051906936864,
      "grad_norm": 0.6238139867782593,
      "learning_rate": 8.943066073631435e-08,
      "loss": 0.8041,
      "step": 29590
    },
    {
      "epoch": 1.9522647451646413,
      "grad_norm": 0.5499809980392456,
      "learning_rate": 8.701513076067869e-08,
      "loss": 0.7788,
      "step": 29600
    },
    {
      "epoch": 1.9529242996355962,
      "grad_norm": 0.6836000084877014,
      "learning_rate": 8.463261390446175e-08,
      "loss": 0.7739,
      "step": 29610
    },
    {
      "epoch": 1.9535838541065509,
      "grad_norm": 0.6283951997756958,
      "learning_rate": 8.228311332481864e-08,
      "loss": 0.772,
      "step": 29620
    },
    {
      "epoch": 1.954243408577506,
      "grad_norm": 0.5444614291191101,
      "learning_rate": 7.996663213516165e-08,
      "loss": 0.7846,
      "step": 29630
    },
    {
      "epoch": 1.9549029630484607,
      "grad_norm": 0.5624345541000366,
      "learning_rate": 7.768317340514642e-08,
      "loss": 0.7867,
      "step": 29640
    },
    {
      "epoch": 1.9555625175194158,
      "grad_norm": 0.5255836844444275,
      "learning_rate": 7.543274016066915e-08,
      "loss": 0.8102,
      "step": 29650
    },
    {
      "epoch": 1.9562220719903705,
      "grad_norm": 0.574588418006897,
      "learning_rate": 7.321533538386105e-08,
      "loss": 0.8184,
      "step": 29660
    },
    {
      "epoch": 1.9568816264613254,
      "grad_norm": 0.5666007995605469,
      "learning_rate": 7.103096201308556e-08,
      "loss": 0.765,
      "step": 29670
    },
    {
      "epoch": 1.9575411809322802,
      "grad_norm": 0.5659819841384888,
      "learning_rate": 6.887962294294393e-08,
      "loss": 0.806,
      "step": 29680
    },
    {
      "epoch": 1.9582007354032351,
      "grad_norm": 0.5693690180778503,
      "learning_rate": 6.676132102424737e-08,
      "loss": 0.8027,
      "step": 29690
    },
    {
      "epoch": 1.95886028987419,
      "grad_norm": 0.5768720507621765,
      "learning_rate": 6.46760590640394e-08,
      "loss": 0.779,
      "step": 29700
    },
    {
      "epoch": 1.9595198443451447,
      "grad_norm": 0.606207013130188,
      "learning_rate": 6.26238398255763e-08,
      "loss": 0.7832,
      "step": 29710
    },
    {
      "epoch": 1.9601793988160998,
      "grad_norm": 0.6650753021240234,
      "learning_rate": 6.060466602833271e-08,
      "loss": 0.8116,
      "step": 29720
    },
    {
      "epoch": 1.9608389532870545,
      "grad_norm": 0.6604939103126526,
      "learning_rate": 5.8618540347990525e-08,
      "loss": 0.8095,
      "step": 29730
    },
    {
      "epoch": 1.9614985077580096,
      "grad_norm": 0.5078802108764648,
      "learning_rate": 5.6665465416436136e-08,
      "loss": 0.7879,
      "step": 29740
    },
    {
      "epoch": 1.9621580622289643,
      "grad_norm": 0.6700655221939087,
      "learning_rate": 5.4745443821763166e-08,
      "loss": 0.8114,
      "step": 29750
    },
    {
      "epoch": 1.9628176166999192,
      "grad_norm": 0.6484373807907104,
      "learning_rate": 5.285847810826139e-08,
      "loss": 0.8497,
      "step": 29760
    },
    {
      "epoch": 1.963477171170874,
      "grad_norm": 0.6642829775810242,
      "learning_rate": 5.100457077642229e-08,
      "loss": 0.8363,
      "step": 29770
    },
    {
      "epoch": 1.964136725641829,
      "grad_norm": 0.5557263493537903,
      "learning_rate": 4.918372428292517e-08,
      "loss": 0.8005,
      "step": 29780
    },
    {
      "epoch": 1.9647962801127838,
      "grad_norm": 0.6279803514480591,
      "learning_rate": 4.7395941040639933e-08,
      "loss": 0.8498,
      "step": 29790
    },
    {
      "epoch": 1.9654558345837387,
      "grad_norm": 0.5895106196403503,
      "learning_rate": 4.564122341862709e-08,
      "loss": 0.8074,
      "step": 29800
    },
    {
      "epoch": 1.9661153890546936,
      "grad_norm": 0.6393919587135315,
      "learning_rate": 4.391957374212385e-08,
      "loss": 0.8016,
      "step": 29810
    },
    {
      "epoch": 1.9667749435256483,
      "grad_norm": 0.6000730395317078,
      "learning_rate": 4.2230994292558054e-08,
      "loss": 0.7748,
      "step": 29820
    },
    {
      "epoch": 1.9674344979966034,
      "grad_norm": 0.587570309638977,
      "learning_rate": 4.057548730752592e-08,
      "loss": 0.7784,
      "step": 29830
    },
    {
      "epoch": 1.968094052467558,
      "grad_norm": 0.6112786531448364,
      "learning_rate": 3.8953054980800376e-08,
      "loss": 0.8131,
      "step": 29840
    },
    {
      "epoch": 1.9687536069385132,
      "grad_norm": 0.6138783693313599,
      "learning_rate": 3.736369946232554e-08,
      "loss": 0.8195,
      "step": 29850
    },
    {
      "epoch": 1.9694131614094679,
      "grad_norm": 0.728312611579895,
      "learning_rate": 3.58074228582167e-08,
      "loss": 0.8106,
      "step": 29860
    },
    {
      "epoch": 1.9700727158804228,
      "grad_norm": 0.6900039911270142,
      "learning_rate": 3.428422723075475e-08,
      "loss": 0.7994,
      "step": 29870
    },
    {
      "epoch": 1.9707322703513777,
      "grad_norm": 0.5480231046676636,
      "learning_rate": 3.279411459838067e-08,
      "loss": 0.7692,
      "step": 29880
    },
    {
      "epoch": 1.9713918248223326,
      "grad_norm": 0.6333634853363037,
      "learning_rate": 3.133708693570103e-08,
      "loss": 0.7794,
      "step": 29890
    },
    {
      "epoch": 1.9720513792932874,
      "grad_norm": 0.6649506688117981,
      "learning_rate": 2.9913146173474185e-08,
      "loss": 0.787,
      "step": 29900
    },
    {
      "epoch": 1.9727109337642421,
      "grad_norm": 0.4889037311077118,
      "learning_rate": 2.8522294198618516e-08,
      "loss": 0.7896,
      "step": 29910
    },
    {
      "epoch": 1.9733704882351972,
      "grad_norm": 0.6194847226142883,
      "learning_rate": 2.7164532854201395e-08,
      "loss": 0.808,
      "step": 29920
    },
    {
      "epoch": 1.974030042706152,
      "grad_norm": 0.6829218864440918,
      "learning_rate": 2.583986393944471e-08,
      "loss": 0.8069,
      "step": 29930
    },
    {
      "epoch": 1.974689597177107,
      "grad_norm": 0.6882838606834412,
      "learning_rate": 2.454828920971375e-08,
      "loss": 0.788,
      "step": 29940
    },
    {
      "epoch": 1.9753491516480617,
      "grad_norm": 0.6496096849441528,
      "learning_rate": 2.328981037652833e-08,
      "loss": 0.7632,
      "step": 29950
    },
    {
      "epoch": 1.9760087061190166,
      "grad_norm": 0.5673264861106873,
      "learning_rate": 2.206442910753781e-08,
      "loss": 0.7861,
      "step": 29960
    },
    {
      "epoch": 1.9766682605899715,
      "grad_norm": 0.6036912798881531,
      "learning_rate": 2.0872147026543277e-08,
      "loss": 0.806,
      "step": 29970
    },
    {
      "epoch": 1.9773278150609264,
      "grad_norm": 0.5649986863136292,
      "learning_rate": 1.971296571348369e-08,
      "loss": 0.8245,
      "step": 29980
    },
    {
      "epoch": 1.9779873695318813,
      "grad_norm": 0.5428677201271057,
      "learning_rate": 1.8586886704430318e-08,
      "loss": 0.8154,
      "step": 29990
    },
    {
      "epoch": 1.978646924002836,
      "grad_norm": 0.602078378200531,
      "learning_rate": 1.7493911491595072e-08,
      "loss": 0.774,
      "step": 30000
    },
    {
      "epoch": 1.979306478473791,
      "grad_norm": 0.6510676741600037,
      "learning_rate": 1.6434041523316622e-08,
      "loss": 0.793,
      "step": 30010
    },
    {
      "epoch": 1.9799660329447457,
      "grad_norm": 0.6513907313346863,
      "learning_rate": 1.5407278204068732e-08,
      "loss": 0.7836,
      "step": 30020
    },
    {
      "epoch": 1.9806255874157008,
      "grad_norm": 0.6857253313064575,
      "learning_rate": 1.4413622894457468e-08,
      "loss": 0.7807,
      "step": 30030
    },
    {
      "epoch": 1.9812851418866555,
      "grad_norm": 0.6612505316734314,
      "learning_rate": 1.345307691120734e-08,
      "loss": 0.7765,
      "step": 30040
    },
    {
      "epoch": 1.9819446963576104,
      "grad_norm": 0.6458035111427307,
      "learning_rate": 1.2525641527175168e-08,
      "loss": 0.815,
      "step": 30050
    },
    {
      "epoch": 1.9826042508285653,
      "grad_norm": 0.5740944743156433,
      "learning_rate": 1.1631317971341759e-08,
      "loss": 0.8259,
      "step": 30060
    },
    {
      "epoch": 1.9832638052995202,
      "grad_norm": 0.6129186749458313,
      "learning_rate": 1.0770107428806352e-08,
      "loss": 0.755,
      "step": 30070
    },
    {
      "epoch": 1.983923359770475,
      "grad_norm": 0.6856721043586731,
      "learning_rate": 9.9420110407894e-09,
      "loss": 0.83,
      "step": 30080
    },
    {
      "epoch": 1.98458291424143,
      "grad_norm": 0.5338472127914429,
      "learning_rate": 9.14702990463534e-09,
      "loss": 0.7711,
      "step": 30090
    },
    {
      "epoch": 1.9852424687123849,
      "grad_norm": 0.5985947251319885,
      "learning_rate": 8.385165073801493e-09,
      "loss": 0.7747,
      "step": 30100
    },
    {
      "epoch": 1.9859020231833395,
      "grad_norm": 0.6340281367301941,
      "learning_rate": 7.656417557863615e-09,
      "loss": 0.7788,
      "step": 30110
    },
    {
      "epoch": 1.9865615776542946,
      "grad_norm": 0.6839861869812012,
      "learning_rate": 6.960788322513123e-09,
      "loss": 0.8268,
      "step": 30120
    },
    {
      "epoch": 1.9872211321252493,
      "grad_norm": 0.6112245321273804,
      "learning_rate": 6.29827828955154e-09,
      "loss": 0.7713,
      "step": 30130
    },
    {
      "epoch": 1.9878806865962044,
      "grad_norm": 0.6058304309844971,
      "learning_rate": 5.668888336898826e-09,
      "loss": 0.8565,
      "step": 30140
    },
    {
      "epoch": 1.988540241067159,
      "grad_norm": 0.5842820405960083,
      "learning_rate": 5.0726192985794955e-09,
      "loss": 0.788,
      "step": 30150
    },
    {
      "epoch": 1.989199795538114,
      "grad_norm": 0.6127521395683289,
      "learning_rate": 4.509471964730949e-09,
      "loss": 0.8039,
      "step": 30160
    },
    {
      "epoch": 1.9898593500090689,
      "grad_norm": 0.5821914076805115,
      "learning_rate": 3.979447081606247e-09,
      "loss": 0.8399,
      "step": 30170
    },
    {
      "epoch": 1.9905189044800238,
      "grad_norm": 0.5806571245193481,
      "learning_rate": 3.4825453515574536e-09,
      "loss": 0.8254,
      "step": 30180
    },
    {
      "epoch": 1.9911784589509787,
      "grad_norm": 0.6134341359138489,
      "learning_rate": 3.018767433046743e-09,
      "loss": 0.7945,
      "step": 30190
    },
    {
      "epoch": 1.9918380134219333,
      "grad_norm": 0.5684574842453003,
      "learning_rate": 2.5881139406436216e-09,
      "loss": 0.7578,
      "step": 30200
    },
    {
      "epoch": 1.9924975678928885,
      "grad_norm": 0.575101912021637,
      "learning_rate": 2.190585445022153e-09,
      "loss": 0.8154,
      "step": 30210
    },
    {
      "epoch": 1.9931571223638431,
      "grad_norm": 0.5798953771591187,
      "learning_rate": 1.826182472966509e-09,
      "loss": 0.7824,
      "step": 30220
    },
    {
      "epoch": 1.9938166768347982,
      "grad_norm": 0.5315937399864197,
      "learning_rate": 1.4949055073570916e-09,
      "loss": 0.7947,
      "step": 30230
    },
    {
      "epoch": 1.994476231305753,
      "grad_norm": 0.5395314693450928,
      "learning_rate": 1.1967549871816363e-09,
      "loss": 0.7722,
      "step": 30240
    },
    {
      "epoch": 1.9951357857767078,
      "grad_norm": 0.5870974659919739,
      "learning_rate": 9.317313075296597e-10,
      "loss": 0.8507,
      "step": 30250
    },
    {
      "epoch": 1.9957953402476627,
      "grad_norm": 0.5886339545249939,
      "learning_rate": 6.998348195980109e-10,
      "loss": 0.8084,
      "step": 30260
    },
    {
      "epoch": 1.9964548947186176,
      "grad_norm": 0.619418740272522,
      "learning_rate": 5.010658306742188e-10,
      "loss": 0.7772,
      "step": 30270
    },
    {
      "epoch": 1.9971144491895725,
      "grad_norm": 0.6857271790504456,
      "learning_rate": 3.354246041614717e-10,
      "loss": 0.8017,
      "step": 30280
    },
    {
      "epoch": 1.9977740036605272,
      "grad_norm": 0.6156572699546814,
      "learning_rate": 2.0291135955086138e-10,
      "loss": 0.7994,
      "step": 30290
    },
    {
      "epoch": 1.9984335581314823,
      "grad_norm": 0.590694010257721,
      "learning_rate": 1.035262724463637e-10,
      "loss": 0.7933,
      "step": 30300
    },
    {
      "epoch": 1.999093112602437,
      "grad_norm": 0.5420209169387817,
      "learning_rate": 3.726947454263385e-11,
      "loss": 0.8108,
      "step": 30310
    },
    {
      "epoch": 1.999752667073392,
      "grad_norm": 0.6134430170059204,
      "learning_rate": 4.1410536416597314e-12,
      "loss": 0.7825,
      "step": 30320
    },
    {
      "epoch": 2.0,
      "step": 30324,
      "total_flos": 1.277030840488113e+19,
      "train_loss": 0.8469410036818174,
      "train_runtime": 209948.1178,
      "train_samples_per_second": 2.311,
      "train_steps_per_second": 0.144
    }
  ],
  "logging_steps": 10,
  "max_steps": 30324,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.277030840488113e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}